63690 CLIMATE RISK AND BUSINESS AGRIBUSINESS Ghana Oil Palm Development Company Full Report Acknowledgments © 2011, International Finance Corporation Authored by Vladimir Stenek, International Finance Corporation Richenda Connell, Acclimatise The authors wish to thank the following institutions for their valuable contributions to the study: Ghana Oil Palm Development Company (GOPDC); Agriculture- engineering Department, Faculty of Engineering Science, University of Ghana, Legon; CSIR Water Research Institute, Ghana; Department of Crop Science, University of Ghana, Legon; Department of Geology, University of Ghana, Legon; Food and Agriculture Organization, Ghana; Ghana Environmental Protection Agency, Ghana Meteorological Services, GOPDC Outgrowers’ Association; Kade Agricultural Research Center, University of Ghana, Kwaebibirem District, Ghana; Oil Palm Research Institute, Ghana; Presidential Special Initiative, Ghana; St. Dominic’s Hospital, Akwatia; University of Ghana, Legon; World Health Organization, Ghana; Centre de cooperation international en recherché agronomique pour le developpement (CIRAD); and SIAT. CLIMATE RISK AND BUSINESS AGRIBUSINESS Ghana Oil PalmDevelopment Company Full Report Table of Contents Chapter 1: Observed climate and future climate change at GOPDC ........................................................... 9 1.1 Observed climatic conditions at GOPDC................................................................................................10 1.2 Future climatic conditions at GOPDC.....................................................................................................18 Chapter 1 References.........................................................................................................................................20 Chapter 2: Climate risk analysis for oil palm yield at GOPDC Kwae Nucleus Estate................................. 21 2.1 Literature review: factors influencing oil palm yield ................................................................................22 2.2 Analysis of current climatic conditions at Kwae Nucleus Estate and their suitability for oil palm production ...........................................................................................................................................................32 2.3 Analyses of current and potential future fresh fruit bunch yields at Kwae Nucleus Estate .....................32 2.4 Potential future impacts of climate change on financial performance.....................................................61 2.5 Concluding remarks and suggested adaptation actions .........................................................................63 Chapter 2 References.........................................................................................................................................64 Chapter 3: Climate risk analysis for oil palm pests and diseases at GOPDC ............................................ 66 3.1 Pests and diseases relevant to oil palm at GOPDC and potential for losses of FFB..............................67 3.2 Pest and disease control at GOPDC and costs......................................................................................68 3.3 Climate sensitivities of pests and diseases and possible impacts of climate change.............................69 3.4 Recommended adaptation actions .........................................................................................................71 Chapter 3 References.........................................................................................................................................72 Chapter 4: Climate risk analysis for ecosystem services............................................................................ 74 4.1 Pollination...............................................................................................................................................75 4.2 Biodiversity plots ....................................................................................................................................78 4.3 Adaptation options and conclusions .......................................................................................................80 Chapter 4 References.........................................................................................................................................81 Chapter 5: Climate risk analysis for GOPDC industrial operations............................................................. 83 5.1 Electrical equipment ...............................................................................................................................84 5.2 Mill..........................................................................................................................................................84 5.3 Cooling water system .............................................................................................................................85 5 5.4 Refinery..................................................................................................................................................87 5.5 Fractionation plant..................................................................................................................................95 5.6 Power production ...................................................................................................................................98 Chapter 5 References.......................................................................................................................................101 Chapter 6: Climate risk analysis for water and wastewater at GOPDC.................................................... 102 6.1 Groundwater ........................................................................................................................................103 6.2 Wastewater ..........................................................................................................................................111 6.3 Conclusion ...........................................................................................................................................113 Chapter 6 References.......................................................................................................................................113 Chapter 7: Climate risk analysis for community and social issues ........................................................... 116 7.1 Climate impacts on the community around GOPDC ............................................................................117 7.2 General social issues in the community around GOPDC .....................................................................121 7.3 Direct and indirect impacts of community issues on GOPDC...............................................................121 7.4 Possible GOPDC actions .....................................................................................................................122 Chapter 7 References.......................................................................................................................................125 Chapter 8: Climate risk analysis for malaria at GOPDC ........................................................................... 127 8.1 Climate and malaria .............................................................................................................................130 8.2 Economic impacts of malaria for GOPDC ............................................................................................136 8.3 Impacts of climate change....................................................................................................................138 8.4 Existing and proposed adaptation actions............................................................................................138 Chapter 8 References.......................................................................................................................................140 Annexes .................................................................................................................................................... 142 Annex A: Greenhouse gas emissions scenarios...............................................................................................142 Annex B: Results of statistical analyses of FFB yields and climate variables ...................................................144 Annex C: Methods used to perturb observed yield models...............................................................................154 Annex D: Statistical tests ..................................................................................................................................155 Annex E: Results of Pearson’s correlation analysis ..........................................................................................156 6 Overview of Ghana Oil Palm Development Company Ghana Oil Palm Development Company (GOPDC) Limited is an integrated agroindustrial company specialized in the organic cultivation of oil palm (OP), extraction of crude palm oil (CPO) and palm kernel oil (PKO) and in the refining and fractionation of CPO. The company was set up in 1975 as a state enterprise with the main objective of diversifying agricultural production in Ghana and was later transformed into a limited liability company in 1995. Located in the Eastern Region of Ghana in Kwaebibirem District, GOPDC owns and manages approximately 20,500ha of OP plantation divided between Kwae and Okumaning estates. Some 6,500ha is directly run by GOPDC staff (of which there are approximately 280), while 14,000ha is farmed by a body of 7,000 outgrowers who own land located within 30km of the oil palm mill at Kwae estate. GOPDC assists outgrowers in the development of their plantations and they sell fruits to the company. Approximately 300ha within the concession area is farmed by smallholders – farmers who are permitted to develop a temporary plantation within the estate and sell their OP fruits to GOPDC. The catchment area, as described, is shown in the figure below. Three main operations are undertaken by GOPDC: OP plantation development; processing of fresh fruit bunches (FFBs) in the mill to produce CPO and PKO; and refining/fractionation of CPO into higher value products. Plantation development activities include seed germination, OP cultivation in the pre-nursery, nursery and main plantation, non-mechanized harvesting and collection of FFBs and transportation of FFBs to the mill. GOPDC operates a 60 tons/hour FFB mill (which is currently being increased to provide 80 tons/hour capacity) and a 60 tons/day palm kernel mill, where all FFBs are processed to produce CPO, PKO and palm kernel cake (PKC). Some 98% of CPO and is processed further at the refinery/fractionation plant (which has a capacity of 100 tons/day) to produce higher value products. Olein and stearin are the main refined products from CPO, with refined bleached deodorized oil (RBDO) and palm fatty acid distillate (PFAD) as additional, smaller volume products. To give an indication of the order of magnitude of their operation, GOPDC was expected to produce approximately 26,000 tons of olein and stearin, 1,300 tons of PKO and 1,300 tons of RPKO in 2009. This equates to approximately $20.9m in sales. Products are stored in tanks at Kwae Nucleus Estate and at Tema Tank Farm, located at Tema Harbor (approximately 120km southeast of the estate), for distribution to market. The storage tanks provide a total capacity of 15,000 tons. Further to its core business, GOPDC is committed to maintaining high environmental standards. GOPDC minimizes its use of chemicals on the land (e.g. by following the rules of Integrated Pest Management wherever possible), manages biodiversity by planting and/or maintaining riparian and upland biodiversity plots on its estates (intended both to conserve species and to offer soil protection), manages water use and treatment (e.g. GOPDC treats effluent from the mill and refinery plants), and manages energy use as well as gaseous and solid waste streams. GOPDC is also committed to promoting development and minimizing adverse impacts in the areas within which it operates as well as building good relations with its local communities and other stakeholders. As part of this commitment, GOPDC has appointed a Community Relations Officer and has developed a Social Action Plan addressing issues such as facilitation of smallholder and outgrower schemes, provision of infrastructure in the surrounding communities and health care provision. Health care provision, for example, includes spraying against mosquitoes in Kwae and the introduction of malaria prevention and sensitization programs for community groups and schools. Most of GOPDC’s sales are to the domestic market in Ghana, driven by a deficit of palm oil in the country. Globally, Malaysian and Indonesian exports dominate the international market. This Southeast Asian dominance stems from the region consistently producing higher yields than West African countries, due in part to more favorable climatic conditions. 7 Catchment area of GOPDC operations (30km radius from Kwae mill) Kwae Concession Okumaning Concession 8 Chapter 1: Observed climate and future climate change at GOPDC 9 1.1 Observed climatic conditions at GOPDC This chapter summarizes data on observed and future climate conditions at GOPDC’s plantations for the factors identified in Chapter 2 as being important to oil palm fresh fruit bunch (FFB) yield, namely: x rainfall (mm), x ° temperature (mean, minimum and maximum) ( C), x sunshine hours (hours), x relative humidity (RH; %), x potential evapotranspiration (PET; mm), and x soil moisture (mm). General climatic conditions in Ghana The climate of Ghana is tropical, and is strongly influenced by the West African Monsoon. The rainfall seasons of Ghana are controlled by the movement of the tropical rain belt (also known as the Inter- Tropical Conversion Zone, ITCZ), which oscillates between the northern and southern tropics over the course of a year. The dominant wind direction in regions south of the ITCZ is south-westerly, blowing moist air from the Atlantic onto the continent, but north of the ITCZ the prevailing winds come from the north east, bringing hot and dusty air from the Sahara desert, known as the Harmattan. As the ITCZ migrates between its north and south positions over the course of the year, the regions between these northern and southernmost positions of the ITCZ experience a shift between the two opposing prevailing wind directions. This pattern is referred to as the West African Monsoon. The southern regions of Ghana, where GOPDC’s plantations are located, have two wet seasons, one in March to July, and a shorter wet season in September to November, corresponding to the northern and southern passages of the ITCZ across the region. Seasonal rainfall varies considerably on inter-annual and inter-decadal timescales, due in part to variations in the movements and intensity of the ITCZ and consequent variations in timing and intensity of the West African Monsoon. This means that long term trends in rainfall are difficult to identify. Rainfall over Ghana was particularly high in the 1960s, and decreased to particularly low levels in the late 1970s and early 1980s, which caused an overall decreasing trend in the period 1960 to 2006, of an average 2.3mm per month or 2.4% per decade. The most well documented cause of these variations is the El Niño Southern Oscillation (ENSO). El Niño events are associated with drier than average conditions in West Africa. Local climatic conditions The climate data local to GOPDC’s plantations held by Acclimatise are summarized in Table 1. Data for Kwae Agric Office were provided by GOPDC. Data on rainfall and hydrological deficit at Kwae, provided by GOPDC, were produced by the Centre de Coopération Internationale en Recherche Agronomique pour le Développement (CIRAD). Data for Akim Oda Met Station, which, at approximately 30km south of Kwae, is the nearest met station to Kwae, were obtained from Ghana Meteorological Services. These data are presented in Figures Figure 1 to Figure 13 below. 10 Table 1: Monthly climate data Climate variable Kwae Agric Office Akim Oda Met Station Rainfall (mm) 1977–2007 1961–2007 ° Tmean ( C) Calculated by averaging Tmin 1967–2007 and Tmax ° Tmin ( C) 1999, 2002–2007 1967–2007 ° Tmax ( C) 1999, 2002–2007 1967–2007 (a) (b) RH (%) 1997–1998, 2002–2007 1969–2006 Sunshine hours (hours) 2001–2005, 2007 1976–1993, 1996–1997, 2000–2006 PET (mm) Not available 1991–2006 Hydrological deficit (mm) 1977–2008 Not available (a) Provided as “RH minimum day” and “RH maximum night” (b) Provided as “RH at 06:00 hours” and “RH at 15:00 hours” Rainfall The monthly average rainfall data (Figure 1) clearly show the two wet seasons described above, and the dry seasons in August and December to February. The agreement between average monthly rainfall recorded at Akim Oda and Kwae is good, though Akim Oda exhibits stronger year-to-year variability (shown by the error bars in Figure 1). Observed trends in rainfall The annual average of monthly precipitation recorded at Akim Oda Met Station for 1970 to 2007 is shown in Figure 2 for the dry season and Figure 3 for the rainy season. Both data sets exhibit a very small decreasing trend over this time period. The year 2007 saw particularly low rainfall in the dry season, with approximately half the normal rainfall. Figure 1: Monthly average daily rainfall (mm) recorded at Kwae (provided by CIRAD, 1988–2007) and Akim Oda (1961–2007). The bars show the high and low ranges of recorded monthly values. 11 Figure 2: Trend in average of monthly precipitation (mm) at Akim Oda,1970–2007, for the dry season (Nov–Mar) Figure 3: Trend in average of monthly precipitation (mm) at Akim Oda, 1970–2007, for the rainy season (Apr–Oct, excluding Jul–Aug) Temperature Monthly average temperature data are shown in Figures Figure 4 to Figure 6. Temperatures at Kwae and Akim Oda are generally similar, though temperatures at Kwae are consistently higher, particularly minimum temperatures. It is clear that temperature exhibits much less year-to-year variability than rainfall (shown by the smaller error bars). 12 Figure 4: Monthly average daily mean temperature (°C) recorded at Kwae Agric Office (1999–2007) and Akim Oda (1961–2007). The bars show the high and low ranges of recorded monthly values. Figure 5: Monthly average daily minimum temperature (°C) recorded at Kwae Agric Office (1999– 2007) and Akim Oda (1961–2007). The bars show the high and low ranges of recorded monthly values. 13 Figure 6: Monthly average daily maximum temperature (°C) recorded at Kwae Agric Office (1999– 2007) and Akim Oda (1961–2007). The bars show the high and low ranges of recorded monthly values. Observed trends in temperature The annual average of monthly mean temperatures recorded at Akim Oda Met Station from 1970 to 2007 is shown in Figure 7. There is an increasing trend in annual average mean temperature, with an increase of 1.5°C occurring over that period. This represents an increase of approximately 0.04°C per year between 1970 and 2007 and is an indication that the effects of climate change are already underway. Figure 7: Trend in annual average of monthly mean temperature (°C) at Akim Oda, 1970–2007 The annual average of monthly maximum temperatures recorded at Akim Oda Met Station from 1970 to 2007 is shown in Figure 8. As can be seen, there is an increasing trend in annual average maximum 14 temperature, with an increase of 1.9°C occurring over that period. This represents an increase of approximately 0.05°C per year between 1970 and 2007. Figure 8: Trend in annual average of monthly max temperature (°C) at Akim Oda, 1970–2007 Relative humidity Monthly average relative humidity (RH) data are shown in Figures Figure 9 and Figure 10. RH at Kwae and Akim Oda are similar, though RH at Akim Oda is generally higher, particularly at night. Figure 9: Monthly average daily relative humidity (%) recorded at Kwae Agric Office (1997–2007) and Akim Oda (1967–2007). The bars show the high and low ranges of recorded monthly values. 15 Figure 10: Monthly average nightly relative humidity (%) recorded at Kwae Agric Office (1997– 2007) and Akim Oda (1967–2007). The bars show the high and low ranges of recorded monthly values. Sunshine hours Monthly average daily sunshine hours data are shown in Figure 11. Sunshine hours at Kwae tend to be slightly higher than at Akim Oda. Figure 11: Monthly average daily sunshine (hrs) recorded at Kwae Agric Office (2001–2008) and Akim Oda (1976–2007). The bars show the high and low ranges of recorded monthly values. Potential evapotranspiration Monthly average daily potential evapotranspiration data recorded at Akim Oda are shown in Figure 12. 16 Figure 12: Monthly average daily potential evapotranspiration (mm) recorded at Akim Oda (1991– 2007). The bars show the high and low ranges of recorded monthly values. Soil moisture deficit Monthly records of soil moisture deficit from Kwae Agricultural Office, produced by CIRAD, are shown in Figure 13. Figure 13: Monthly soil moisture deficit (mm, Kwae Agric. Office, 1997–2008, from CIRAD) 17 1.2 Future climatic conditions at GOPDC Projected changes in rainfall Projections of future changes in rainfall for Ghana are presented in Figure 14 for the A2 greenhouse gas emissions scenario (McSweeney et al., 2008). (For further discussion on emissions scenarios, see Annex A: Greenhouse gas emissions scenarios). Unfortunately, rainfall projections from different climate models show a wide range for Ghana, with around half the models projecting rainfall increases and half projecting decreases. This is in part due to disagreements in the climate models about future changes in the amplitude of El Niño Southern Oscillation (ENSO) events, which strongly influence West African climate. This uncertainty is significant for GOPDC, as many of its activities – most notably oil palm productivity – are dependent on rainfall. As shown in Figure 14, over the area of GOPDC’s plantations, the median trend for Jan/Feb/Mar and Apr/May/Jun rainfall is a zero change, and for Jul/Aug/Sep and Oct/Nov/Dec the median trend is for a slight increase (on the order of a few mm per month). However, the maximum and minimum climate model outputs show significantly larger changes – reductions of up to 19mm/month in Jul/Aug/Sep and increases of up to 38mm/month in Apr/May/Jun. Taking the largest projected reductions in rainfall from Figure 14 indicates that climate change could lead to average rainfall changes of about 10–15% at GOPDC plantations by the 2030s, with the years between now and the 2030s seeing extreme low and high seasonal rainfall occurring increasingly often. Combined with higher temperatures, reduced rainfall would lead to significant reductions in soil moisture. The climate models are in better agreement that the proportion of total annual rainfall that falls in “heavy” 1 events will tend to increase in the future. Seasonally, this varies between tendencies to decreases in Jan/Feb/Mar and to increases in Jul/Aug/Sep and Oct/Nov/Dec. Projected changes in the maximum intensity of 1-day and 5-day rainfall events also tend towards increases. 1 A “heavy” event is defined as a daily rainfall total which exceeds the threshold that is exceeded on 5% of rainy days in the current climate of that region and season. 18 Figure 14: Projected changes in monthly average precipitation (mm) in Ghana by the 2030s (relative to 1970–1999 baseline) for the A2 greenhouse gas emissions scenario. In each grid box, the central value (large number) shows the median of the climate models and the values in the upper and lower corners are the maximum and minimum across all the models. GOPDC is located in the bottom grid box shown for Ghana. (Source: McSweeney et al., 2008) Jan/Feb/Mar Jul/Aug/Sep Apr/May/Jun Oct/Nov/Dec Projected changes in temperature Projections of future changes in temperature for Ghana are presented in Figure 15 for the A2 greenhouse gas emissions scenario (McSweeney et al., 2008). Here, the picture is much clearer than for rainfall. Annual temperature increases of 1.2°C (low to high range of 0.8–1.5°C) are projected in the 2030s and 2.3°C (1.7–2.7°C) in the 2060s. Seasonal temperature increases are similar to the annual average. The climate models also indicate that extremely hot periods will occur more frequently and cold periods will become increasingly rare. As noted in Section 1.1, temperatures at Kwae have already risen considerably over recent decades. These future increases will act to reduce soil moisture. 19 Figure 15: Projected increases in annual average temperatures (°C) in Ghana by the 2030s and 2060s (relative to 1970–1999 baseline) for the A2 greenhouse gas emissions scenario Chapter 1 References McSweeney, C., New, M., Lizcano, G. (2008). UNDP Climate Change Country Profiles – Ghana. 1DNLüHQRYLü1DQG56ZDUW (GV Special Report on Emissions Scenarios. A Special Report of Working Group III of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, 599 pp. 20 Chapter 2: Climate risk analysis for oil palm yield at GOPDC Kwae Nucleus Estate 21 Overview This chapter provides an analysis of current and potential future climatic influences on yields of oil palm fresh fruit bunches (FFB) at GOPDC’s Kwae Nucleus Estate, based on data on FFB yields and climate held by Acclimatise. The following factors are known to influence FFB yield: x Climate, x Endogenous annual cycling of yield independent of climatic factors, x Age of palm (maturity), x Soil type, x Seed type (planting material), x Management practices, x Pollinators, and x Pests and diseases. Therefore, to attempt to understand the relationships between climate and yield, it is important to analyze separately data sets which are homogeneous with respect to palm age, soil type and seed type. Where we have data on yield as a function of soil type or seed type, we have performed some tests to understand the influences of these factors. However, due to data gaps, most of the analyses described in this chapter have investigated yields for mature palms (age between 11 to 21 years), across all available data sets, without disaggregating for soil type or seed type. Sections in this chapter This chapter comprises the following sections: x Section 2.1: Literature review: factors influencing oil palm yield x Section 2.2: Analysis of current climatic conditions at Kwae Nucleus Estate and their suitability for oil palm production x Section 2.3: Analyses of current and potential future fresh fruit bunch yields at Kwae Nucleus Estate x Section 2.4: Potential future impacts of climate change on financial performance x Section 2.5: Concluding remarks and suggested adaptation actions 2.1 Literature review: factors influencing oil palm yield Oil palm (OP) growth and yield depend to a large extent on the physical and climatic characteristics of the environment in which the palm is grown. This section first discusses the cyclical nature of yields from oil palms that is independent of climatic factors, followed by climate-related and other factors that are known 22 to be important, including palm age (maturity), soil type, seed type, management practices, pollinators, and pests and diseases. The review focuses on information presented in wider published literature but in some instances we have presented information on Kwae within this section. Note that Sections 2.2 to 2.5 form the main focus of the discussions and analyses specific to Kwae. Cyclical nature of yields Large seasonal variations in OP yield are expected in regions such as West Africa where regular and severe dry periods are common. However, researchers have reported that similar, though less extreme seasonal variations are also evident in regions with more uniform climates such as Malaysia. This annual cycling is reported to have a large influence on yield even in regions that lack marked seasonal changes in climatic factors such as radiation or rainfall, and are also said to persist in irrigated conditions (Henson, 2005). Climatic factors can exacerbate this inherent (endogenous) yield cycle, resulting in markedly different yield profiles across different environments. To begin understanding which climatic factors could affect yield cycles at Kwae, published seasonal yield cycles from different countries are presented in Figure 16a, compared with Kwae (Figure 16b). It can be seen that the annual yield cycle for Benin displays a profile similar to that experienced at Kwae. However, Benin sees a greater proportion (85%) of its annual total yield in January to April, compared to Kwae (62%). The yield cycle for Nigeria, a neighbor of Benin, shows a suppressed peak, with lower percentage contribution to yields in January to May compared to Benin and Kwae. Malaysia, which has a uniform climate, is also shown to have an annual (more modest) yield cycle. As noted earlier, the cycle in Malaysia is reported to be independent of climate (Henson, 2005). Figure 16: Seasonal yield cycles in different countries compared to seasonal yields at Kwae for 1988–1997 years of harvest (YOH) (a) Source: Corley & Tinker (2003) (b) Standard published climate charts for Benin and Nigeria (Cotonou and Lagos respectively, which are comparable in latitude to Kwae) are presented in Figure 17. A similar climate chart was also developed for Kwae using the available data from Kwae Agric. Office/CIRAD. (Note that we do not know the time 23 periods covered by the Cotonou and Lagos climate charts. The chart for Kwae has been compiled from the same data sets presented in Chapter 1.) Figure 17: Climate charts for Kwae, Cotonou and Lagos Chart compiled from Kwae Agric. Office/CIRAD Source: worldtravelguide.net data Note: Humidity (%) data is the average of the maximum night and minimum day relative humidity Source: worldtravelguide.net These figures indicate the following: 24 x Rainfall: o annual average rainfall for Kwae and Cotonou are comparable at approximately 1300mm whilst Lagos receives approximately 1600mm per year, o all three locations exhibit bimodal rainfall patterns, however the bimodal pattern for Kwae is more even. x Humidity: o Cotonou and Lagos show comparable levels of humidity throughout the year, whereas relative humidity for Kwae is lower. (Note that this may be due to differences in the method of calculation used in the standard climate charts and the chart developed specifically for Kwae.) x Temperature: o Cotonou and Lagos exhibit a similar mean temperature profile, although maximum temperatures are higher at Lagos during the earlier and later parts of the year, o Kwae shows higher mean temperatures throughout the year compared to the other two locations. Corley and Tinker (2003) in their definitive book The Oil Palm cautioned that much of the published research on correlations between climatic factors and yield components may be flawed, because the existence of a correlation does not necessarily indicate cause and effect. Corley suggested that “it is almost inevitable that correlations can be found between monthly averages of two factors that both vary seasonally”. This demonstrates the difficulties in evaluating climatic influences on yield at monthly time steps. In conclusion, the presence and contribution of the inherent yield cycle that is independent of climatic factors affects our ability to “disentangle” the effects that climatic factors alone have on yield at monthly time intervals. This makes it difficult to undertake sensible analyses at monthly timescales. Therefore the analyses of climate influences on yield presented in Section 2.3 are based on annual data. Important climatic factors and their influences on yield The ideal climatic requirements for oil palm production, according to Hartley (1988) and Paramananthan et al. (2000, see Table 2) are: x Total annual rainfall of 2000–2500mm, evenly distributed without a marked dry season, and preferably at least 100mm each month. (This accounts for high yields in Southeast Asia.) x A mean maximum temperature of about 29–30°C (according to Hartley 1988) and a mean annual temperature of about 22–24°C (according to Hartley 1988), or 26–29°C (according to Paramananthan et al., 2000). If the temperature falls below 19°C, particularly at night, fresh fruit bunch (FFB) development is affected and yield reduced. Growth in young seedlings stops at temperatures below 15°C. x 2 Sunshine of 5–7 hours/day in all months and solar radiation of at least 15 MJ/m per day. x Mean wind speed of less than 10m/s. 25 Table 2: Classification of climatic conditions and their suitability for oil palm (Paramananthan et al., 2000) For comparison with the Paramananthan et al. (2000) classification presented in Table 2, Table 3 shows climatic conditions close to Kwae, based on data from Akim Oda Met Station, approximately 30km to the south. According to the Paramananthan et al. (2000) classification, the climate in the region of Kwae is moderately suitable/currently unsuitable in terms of annual rainfall and duration of the dry season. Although these classifications of ideal conditions can act as useful guidelines, such assessments can only be approximate, since precise, direct relationships between individual climatic factors and OP yield cannot be readily deduced, due to the wide range of other factors that affect yield. Table 3: Local climatic conditions at Kwae (based on Akim Oda Met Station data) Climate variable Akim Oda Met Station data Average annual rainfall (mm/yr) (1970–2007) 1409 Duration of dry season (months) 3–5 0 Average annual mean temperature ( C) (1970– 26.8 2007) Mean wind speed (1989–2006) 1.5 knots = 0.8 m/s In tropical regions, it is reported that rainfall is often one of the main limiting factors for OP production (Asamoah et al. 2008), and the West African OP growing region is characterized by spells of low or no rainfall (Hartley, 1988 In Corley & Tinker, 2003). Rainfall exerts its main effect on OP growth through its influence on soil moisture content. Solar radiation has also been reported as a factor causing major fluctuations in the yield of OP (Sparnaaij et al., 1963). Generally, dry matter production is directly proportional to the amount of photosynthetically active radiation (PAR) intercepted by the crop canopy (Monteith, 1977). Related to this, leaf area index (LAI) is an important parameter in crop growth (Noor & Harun, 2004). Early studies by 26 Hardon et al. (1969) showed a positive correlation between leaf area and yield of different palms within the same family. OP has a large leaf area and converts a relatively high proportion of the available solar energy to harvestable dry matter. Corley and Gray (1976) showed that LAI increases with palm age and reaches a stable maximum after about 10 years. Hartley (1988) also noted that in certain areas with very low radiation but well distributed and adequate rainfall, yields could be higher than those in regions with much higher radiation but with seasonal dry periods. In addition, according to Henson (1991), relative humidity strongly influences the photosynthetic capacity of OP. Low relative humidity restricts stomatal opening and hence CO2 intake. According to GOPDC, low relative humidity during the Harmattan winds (see Chapter 1) leads to more desiccated OP fruit and lower oil content. The optimum requirements for OP are high atmospheric humidity of more than 80% and high CO2 concentrations. Three successive months with less than 50% relative humidity are not suitable for OP. Oboh and Fekorede (1999) have also identified correlations between potential evapotranspiration (PET) and OP yield. Further information on rainfall and yield In a study, Goh et al. (1994) compared data on rainfall and FFB yield from a number of countries (Table 4 and Figure 18). The relationship was moderately strong, and FFB yields were relatively poor for locations with rainfall less than 2000mm/year or several dry months. Table 4: Typical rainfall and FFB yield in various countries where OP is cultivated For comparison, Figure 18 also shows: x Average annual FFB yield at Kwae Nucleus Estate for 1979–1982 years of planting (YOP) and 1990–1995 years of harvesting (YOH), x Average annual rainfall recorded at Kwae Agric Office from 1987 to 1995 (to allow for the fact that rainfall up to 3 years prior to YOH can influence OP yield (Oboh & Fekorede, 1999). From this figure, it can be clearly seen that FFB yields at Kwae over the period 1990–1995 were lower than those typically recorded in other countries with higher rainfall. 27 Figure 18: Plot of rainfall vs. FFB yield for various countries where OP is cultivated Factors influencing the lifecycle of oil palm The development stages of fruit components There is much variation in the lengths of OP development stages reported across different studies and locations. Table 5 summarizes the stages which determine the final inflorescence and bunch characteristics reported in different studies. Table 5: Development stages of fruit components (from various studies) Approximate months before harvest Breure and Oboh and Menendez Fakorede (1990), (1990), Corley and Tinker(2003), Development stage Malaysia Nigeria Various studies/locations Inflorescence 38 — 44 (Ivory Coast). Corley also found a initiation range of 26–37 months for different clones Sex determination 18 30 21–29 Inflorescence 11 11 9–10 abortion Flowers per spikelet 19 — 12–15 Spikelet number 24 17–24 Within 9 months Frame weight (stalk 7–9 — No clear response plus spikelet) Anthesis and Fruit 6 — 5 set 28 Variation in several developmental processes is reported to contribute to cycles in fruit bunch production in OP. These include the rate of frond emergence (Chang et al., 1988), the inflorescence development rate both before and after frond emergence, and the rate of bunch development after anthesis, together with factors such as pollination efficiency that affect bunch weight (Henson, 2004). Several studies have indicated the likely importance of rates of inflorescence and bunch development on seasonal yield patterns (Corley, 1977; Chang et al., 1993, 1995; Lamade et al., 1998 In Henson, 2005). Corley (1976) assembled data that indicated that specific stages of inflorescence development were reached progressively earlier in older palms. Thus, the stage of initiation of the first bract (considered to precede sex determination) occurred at frond –24 months in 1.5-year-old palms and frond –29 months in 27.5-year-old palms. The first FFBs ripen about 2–3 years after planting. In terms of the timings of negative impacts on the lifecycle, Oboh and Fakorede (1999) report that floral abortion can occur 11 months before harvest and bunch failure can occur 1–3 months before harvest. Sex ratios and FFB yield OP produces both male and female flowers separately on the same palm. The proportion of female inflorescences to total inflorescences (sex ratio) is an important determinant of FFB yield, and high FFB yields tend to be associated with high sex ratios. It should be noted that very high sex ratios can lead to inadequate pollination, as a lack of male flowers has a negative effect on the population of the pollinator Elaeidobius Kamerunicus, which lays its eggs in male flowers. (For further information on potential climatic influences on E. Kamerunicus, see Chapter 4, Ecosystem services.) Hardon and Corley (1976) found that low sex ratios can occur between 19 and 21 months after a severe dry season. Soil moisture influences on development and yield Soil moisture deficit (SMD) is an important factor affecting OP yield. Soil mediates the effects of precipitation and temperature (through evapotranspiration) and SMD varies depending on soil type. In OP, both rates of spear leaf opening (Hartley, 1977) and spear leaf extension (Henson, 1991; Henson et al., 2005) are reported to decline during drought while palms growing on dry sites have lower leaf area, intercept less radiation and show reduced vegetative biomass production and yield compared with those on wetter sites (Henson, 1991; Henson & Chang, 2000). Inflorescence sex determination and inflorescence abortion both respond directly to soil moisture whereas drought favors maleness and higher inflorescence abortion (Turner, 1977; Corley & Tinker, 2003). However, the quantitative relationships between such responses and soil moisture still remain poorly defined. Maillard et al. (In Corley & Tinker, 2003) identified the effects of severe droughts on OP. They noticed numerous closed spears, broken green leaves, dried out leaves, toppled spears and, at times, death of palms. Water deficit also affected the oil content of the fruit bunches with lower oil in normally ripe bunches, preventing complete ripening. In more serious cases, numerous fruits dried up and extraction rates were reduced by 30–40% for several weeks. At GOPDC, in an effort to avoid negative responses to a soil moisture deficit, both the pre-nursery and main nursery are irrigated (by means of a sprinkler and drip irrigation tubes, respectively). It is clear, therefore, that the analysis of yields against climate data needs to consider a range of time periods before harvest, to capture these different effects. 29 Overview of other factors influencing fresh fruit bunch yield In addition to climate, the three main additional factors that determine yield (Griffiths et al., 2002) are: x Age of palm (maturity), x Soil type (and related to this, soil moisture), and x Seed type (planting material). These are discussed in the following sections, together with the influences of management practices, pollinators, pests and diseases. Age of palm (maturity) According to Griffiths et al. (2002), in most environments in South East Asia, FFB yields peak between 7 and 10 years after planting. The decline in yield that often occurs in subsequent years is related to reduced palm stand due to pest and disease infestations, and poor fruit bunch recovery due to difficulties with harvesting tall palms. A study of OP in Nigeria (Oboh & Fakorede, 1999) states that OP reaches maturity after about 6 years. According to Abdullah (2003) in a study of Malaysian OP, fruit production th th reaches a maximum in the 12 or 13 year, after which it starts to decline. GOPDC’s financial model (provided to Acclimatise by IFC) indicates that yield is expected to be stable for those palms that are age 11–21 years. (See Section 2.3 for further details). Soil type OP can be grown on a wide range of soils, the best being coastal alluvial clay, riverine and coastal alluvial soils and soils of volcanic origin (RMRDC, 2004). The best soils for planting are those that are flat or gently undulating and deep (minimum of 90–100 cm), well drained, with no compacted or impermeable horizons close to the surface, permitting root proliferation. Soils capable of retaining sufficient available moisture, not less than 100mm/100cm soil, are best. High yields are sustained in soils with high and balanced nutrient status. Soil pH is important as this can influence nutrient availability. A pH range of 5.6–6.0 is optimal. Areas with pH <4 or >7 are unfavorable. The OP can tolerate temporary flooding, if the water is not stagnant. According to GOPDC, the soil series at Kwae Nucleus Estate include Bekawi, Nzima (upper slopes), Kokofu and Kakum (middle slopes), Temang and Oda (lower slopes). A study of soil properties for a semi-deciduous forest area in Kade describes the basic physical and chemical properties of these six soil profiles (Figure 19). According to the study, the soil sequence shows longitudinal gradients in textures, iron content and drainage conditions, and marked vertical gradients in carbon, nitrogen and phosphorus contents, soil reaction and base saturation, with highest values in the topsoil. Upper slope soils are clayey and show distinct enrichment of clay in the subsoil. They are well drained, rich in iron oxides, and strongly leached, with low electrical conductivity values and base saturation. Drainage becomes poorer towards the valley bottom, where soils generally show loamy textures. 30 Figure 19: Sequence of soils in a semi-deciduous forest area, Kade Source: University of Ghana Agricultural Research Station, Kade. Seed type Planting materials are constantly under development, aimed at improving yields of OP. GOPDC has used a range of different seed types since the Kwae Nucleus Estate was established (see Section 2.3). GOPDC reports that the different seed types at the plantations have differing sensitivities to rainfall: x Pobe (Benin) tolerates low rainfall best of all, x IHRO and Lame (RCI) also perform well in low rainfall conditions, and x Unipalm (Zaire) requires higher rainfall and has not performed well at Kwae. Management practices Plantation management practices, such as use of fertilizers, cover crops and pesticides, all contribute to changes in yield. These factors may in turn be susceptible to average or extreme climatic conditions, ultimately representing additional indirect effects of climate on yield outputs. Fertilizers, for example, are used to obtain an appropriate nutrient balance for the growth of OP throughout its lifecycle (Corley & Tinker, 2003). This in turn is intended to maximize yield. At GOPDC, inorganic fertilizers (e.g. NPK) are applied in the pre-nursery and main nursery, while organic fertilizers are used in the main plantation (GOPDC, 2008). Depending on the timing of application, these fertilizers may be susceptible to leaching by heavy rains or may evaporate in hot, dry conditions. This in turn may lead to negative impacts on yield. Cover crops are also grown to maximize yield. Soil cover protects the soil surface from erosion by raindrops or running water, improves soil structure and infiltration rate (reducing runoff), helps retain soil moisture, moderates soil temperature and protects young palms from Oryctes beetle damage. Legume cover crops also augment nitrogen fixation (Corley & Tinker, 2003). These factors may all contribute to increasing yield. At GOPDC, Pueraria phaseoloides and Mucana bracteata are grown as cover crops for these purposes. Like OP, however, these crops may also be affected by climate change. If growth is negatively affected, this may negatively impact on yield. Additionally, cover plants may compete against the palm crop for nutrients and water, especially at the seedling stage (Corley & Tinker, 2003). Unless managed appropriately, this may be a threat to OP yield under future climate conditions. 31 Pollinators, pest and diseases Clearly, pollinators are an essential component of a successful OP plantation (see Chapter 4, Ecosystem services, for further details). Pest and diseases can have significant episodic impacts on yield, which may affect yields for several years after the initial infestation (see Chapter 3, Pests and diseases). All of these can be affected by climate, though the influences are not well understood. 2.2 Analysis of current climatic conditions at Kwae Nucleus Estate and their suitability for oil palm production In this section we comment on observed climate data at GOPDC (as shown in Chapter 1) in light of knowledge about factors influencing oil palm (Section 2.1). In terms of rainfall, it is clear that total annual rainfall at Kwae is below the 2000mm threshold for optimal OP growth (Chapter 1, Figure 5). Furthermore, average monthly rainfall in Jan, Feb, Aug and Dec is below the 100mm monthly threshold described by Hartley (1988) (Figure 5), and this extended dry period will have an important influence on OP production. Monthly average daily mean temperatures (Figure 4) are in the range 27–31°C, thus occasionally exceeding the “highly suitable” 26–29°C range described by Paramananthan et al. (2000) and well above the ideal mean range of 22–24°C described by Hartley (1988). Monthly average daily minimum temperatures are well above the 19°C threshold provided by Hartley (Figure 5). However, monthly average daily maximum temperatures are in the range 30–35°C, thus exceeding the 30°C threshold given by Hartley, and in the “suitable” or “moderately suitable” Paramananthan classification (Figure 6). With regard to relative humidity (RH), values during the day at Kwae are consistently below the “80% or higher” ideal conditions described by Henson (1991) though rarely do they fall below the 50% “not suitable” threshold he provides (see Section 2.1). Values for RH at night are above the 80% threshold in all months except November and December. (We believe the two zero values for RH in November and December – see Figure 10 – may be an error in the Kwae record). Low RH restricts stomatal opening and hence CO2 intake. Sunshine hours at Kwae, however, fall below the 5–7 hours/day threshold given by Hartley (1988) for June to September, and average only 2.5 hours/day in August (Figure 11). In terms of potential evapotranspiration, while no thresholds are provided in the literature, Oboh and Fakorede (1999) observed a correlation between PET and FFB yields. Finally, while soil moisture deficit (SMD) is an important factor affecting OP yield, the quantitative relationships remain poorly defined. 2.3 Analyses of current and potential future fresh fruit bunch yields at Kwae Nucleus Estate Complexities in analyzing influences on yield Section 2.1 demonstrated that there are a wide range of factors influencing yield, many of which vary considerably across different studies and locations. In addition to the direct influence of climate, other factors such as soil type, seed type, use of fertilizers, plantation management techniques and control measures, pollinators, pest and diseases can all contribute to changes in yield. These factors may in turn be susceptible to incremental or average changes in climatic conditions, ultimately representing additional indirect effects of climate on yield outputs. Examples of episodic events and management practices that can affect yield and add to the complexities of determining the influence of climatic factors alone are noted (GOPDC, Pers. Comm.): 32 x A serious leaf miner outbreak that occurred in 1987. This affected plots planted in 1977 to 1981. The outbreak was reported to have affected yield for up to 5 years after the initial infestation. x Fertilizers are applied in the rainy season (April–mid July) and climatic factors can have indirect impacts on the effects of fertilizers on yield. It was reported that heavy rainfall can wash away fertilizer and dry conditions can lead to its evaporation. Yield analyses related to climate should ideally be undertaken using data from a controlled environment to help isolate the influences of factors on the data set being analyzed. Clearly, since it is based on field data, the yield analysis in the following sections is susceptible to the influences of non-climate factors that are difficult to disentangle. Data available for analyses FFB yield data GOPDC has records of monthly FFB yields at Kwae Nucleus Estate for the 1988–2008 years of harvest (YOH), for the 1979–2005 years of planting (YOP). The available data are summarized in Table 6; Acclimatise has the data shown in the green and yellow cells. Table 6: FFB yield data held by Acclimatise for Kwae Nucleus Estate 33 Figure 20 provides a sample of these data, to show variations in monthly FFB yields across the year. It can be seen that the peak harvesting season is from January to May. Figure 20: Monthly FFB yields for 1994, 1996 and 2005 YOH Soil type and seed type data The data we hold on soil types and seed types at Kwae Nucleus Estate are summarized in Table 7. Table 7: Summary of data held on soil types and seed types for Kwae Nucleus Estate Soil type (%) Year of planting Seed type % Lowland Upland 1979 IHRO (RCI) 100 43.56 56.44 1980 IHRO (RCI) 100 28.81 71.19 1981 IHRO (RCI) 100 9.41 90.59 1982 IHRO (RCI) 100 23.68 76.32 1985 IHRO (RCI) 100 35.77 64.23 1988 IHRO (RCI) 100 100.00 0.00 1989 IHRO (RCI)/ 50/50 100.00 0.00 KUSI (Ghana) 1990 IHRO IN VITRO 100 48.19 51.81 (RCI) 1997 UNIPALM (Zaire) 100 100.00 0.00 1999 UNIPALM 56/44 100.00 0.00 (Zaire)/POBE (Benin) 2000 POBE (Benin) 100 65.41 34.59 34 Soil type (%) Year of planting Seed type % Lowland Upland Oda/Temang Kokofu Nzima/Bekwai (lowland) (upland) (upland) 2004 LAME (RCI) 100 20.50 16.75 62.75 2005 POBE (Benin) 100 10.50 35.43 54.07 Analysis of observed influence of soil type on yield To begin to understand the influence of soil type on FFB yield, some tests have been performed on monthly FFB yields associated with one seed type (IHRO) for mature palms. Two tests were performed (see Table 8): x Test 1: Plots planted between 1979 and 1982 and harvested between 1991 and 1995, with varying percentages of lowland and upland soils. Palm ages are between 8 and 16 years at harvest. x Test 2: Plots planted in 1985 and 1988 and harvested in 2007–2008, with varying percentages of lowland and upland soils. Palm ages are between 19 and 23 years at harvest. Table 8: Soil type (%) for 1979–1982 and 1985 & 1988 YOP Soil type % YOH YOP Seed type Lowland Upland Test 1: 1979 IHRO (RCI) 43.56 56.44 1991–1995 1980 IHRO (RCI) 28.81 71.19 YOH 1981 IHRO (RCI) 9.41 90.59 1982 IHRO (RCI) 23.68 76.32 Test 2: 1985 IHRO (RCI) 35.77 64.23 2007–2008 1988 IHRO (RCI) 100.00 0.00 YOH The results of these tests are shown in Figures Figure 21 and Figure 22. According to interviews with GOPDC staff, we expect yields from lowland soils to exceed those on upland soils. However, examining Test 1 results (Figure 21) reveals that the plot with highest FFB yields in the peak harvest period (January to May) is also the plot with the highest percentage of upland soils (green bars in Figure 21). Yields from June to December, however, were better for the plot with the highest percentage of lowland soils (blue bars in Figure 22). 35 Figure 21: Average monthly FFB yields (tons/ha) for 1979–1982 YOP and 1991–1995 YOH, demonstrating the effects of soil type on yield For Test 2, average monthly yields are plotted in Figure 22. Here, the effect on soil type is more mixed, with some months showing higher yields for the 1988 YOP (100% lowland) and others showing higher yields for 1985 YOP (36% lowland). There is no obvious pattern to these data across the year. Figure 22: Average monthly FFB yields (tons/ha) for 1985 & 1988 YOP and 2007–2008 YOH, demonstrating the effects of soil type on yield Analysis of observed influence of seed type on yield To study the effects of seed type on FFB yields, two tests were performed (see Table 9): 36 x Test 1: Plots planted in 1988–1989 with different seed types, and harvested in 2007–2008. Both plots have 100% lowland soils. Palm ages are between 18 and 20 years at harvest. x Test 2: Plots planted in 1997 and 1999 with different seed types, and harvested in 2007–2008. Both plots have 100% lowland soils. Palm ages are between 8 and 11 years at harvest. Table 9: Seed type (%) for 1988, 1989, 1997 and 1999 YOP YOH YOP Seed type % Soil type (% lowland) Test 1: 1988 IHRO (RCI) 100 100 2007–2008 1989 IHRO (RCI)/ 50/50 100 YOH KUSI (Ghana) Test 2: 1997 UNIPALM (Zaire) 100 100 2007–2008 1999 UNIPALM (Zaire)/ 56/44 100 YOH POBE (Benin) The results (see Figures Figure 23 and Figure 24) demonstrate differences in yield according to seed type. For 1988 YOP (100% IHRO), yield was greater than for 1989 YOP (50% IHRO, 50% Kusi), particularly during the peak production period (Figure 23). Average annual yield was about 60% higher for 1988 YOP. In Test 2 (Figure 24), the yields for plots planted in 1997 (100% Unipalm) show a large peak in February, but are similar to 1999 YOP (56% Unipalm/44% Pobe) for the rest of the year. It may be that the February yield data are an anomaly. There is no obvious indication from these data that the introduction of Pobe (reported by GOPDC to be more tolerant of low rainfall conditions; see Section 2.1) improved yields. Figure 23: Average monthly FFB yields (tons/ha) for 1988 & 1989 YOP and 2007–2008 YOH, demonstrating the effects of seed type on yield 37 Figure 24: Average monthly FFB yields (tons/ha) for 1997 & 1999 YOP and 2007–2008 YOH, demonstrating the effects of seed type on yield Analysis of observed influence of climate on yield In order to assess the potential effects of climate change on future yield for GOPDC, first, the relationships between climate variables and observed FFB yield per hectare need to be investigated and modeled. The following sections present the results of analyses undertaken on annual average data for yield (tons/ha) and climate variables. (As discussed earlier in Section 2.1 it is not possible to undertake the analysis at monthly time intervals because of the cyclical effect on yields which is not climatically driven). The analyses were undertaken on data for Kwae Nucleus Estate only and not for Okumaning or smallholders/outgrowers. This was due to the Kwae yield data being available over a long time period, together with the hectares associated with each YOP. This had the following advantages: x data sets recorded over a long time period allow for more robust analyses of the influences of climate on yield, x the yield data in tons can be normalized into tons per hectare and linked directly back to each YOP and its associated soil type(s) and seed type(s). Determining a constant yield phase in the lifecycle of the OP In order to begin isolating and determining the relationships between climate data and yield, an analysis was undertaken to assess the common ages between which palms are considered to be in a constant (stable) growth phase. This allows an assessment of the effects of climate variables in isolation of age factors such as annual increases in yield due to younger plants maturing and decreases in annual yield due to older plants declining. Estimates of annual FFB yield relative to age (red line on Figure 25) for 1979–2000 YOP were obtained from GOPDC’s financial model, provided by IFC. The data indicate that yield is expected to be in a stable phase for palms between the ages of 11 and 21 years. Prior to this, palms are in a growth phase, and following this period, yields are expected to decline. The analyses described in the following sections are all based on palms in this age range. 38 Figure 25 also presents actual yield data from Kwae Nucleus Estate for 1979–2000 YOP. In general it appears that observed yields are approximately 50% below those estimated in the financial model. However, in the 2008 harvest, the 1990 YOP yielded more than the financial model estimate. Figure 25: Estimated annual FFB yield (tons/ha) from GOPDC financial model compared to actual yield at Kwae Nucleus Estate Pair wise analysis of “good” and “bad” yield years against climatic factors Monthly FFB yield (tons) data were aggregated and normalized to annual FFB yield (tons/ha) data for each individual YOP and only for those palms ages 11–21 years. Initially, a preliminary assessment of the yield data was undertaken by calculating and plotting the percentage difference between each YOH and the average yield across all YOH. This allowed a visual assessment to be made of “good” (i.e. above-average) versus “bad” (i.e. below average) YOH, in turn allowing pairs of years to be isolated based on significantly noticeable switches from below- to above- average (and vice versa) yield. In addition, it also allowed incremental increases or decreases across a number of years to be picked up for further investigation. Figure 26 presents the average plantation yield for each YOH for palms ages 11–21 years, as a percentage of the average plantation yield across all YOH under analysis. The same analysis method was adopted for annualized climate variables in order to compare these to variations in annual yield. The following set of climate variables was initially investigated, and the outputs are presented in Figures Figure 26 to Figure 29: x annual average temperature (mean, maximum, and minimum) (the upward trends described in Section 1.1 can be clearly seen), x annual average rainfall, and x annual soil moisture deficit (produced by CIRAD and calculated using the IHRO method). 39 Figure 26: Changes in FFB yield (tons/ha) compared to average FFB yield across all YOH (palms ages 11–21 years) Figure 27: Changes in observed mean, minimum and maximum temperature compared to the average across all YOH 40 41 Figure 28: Changes in observed annual average rainfall (mm) compared to the average across all YOH Figure 29: Changes in observed annual hydrological deficit (mm) compared to the average across all YOH Table 10 presents a summary of observations from the above analyses. In essence, the key observations are: x 1990–1992: three consecutive years of below-average yield, x 1991: exceptionally low yield for all YOP across all YOH, 42 x 1993–1995: two years of consecutive decline from above-average yield in 1993 to below-average yield in 1995, x 1997–1998: large swing from below-average yield to exceptionally high average yield, x 1998–2000: two consecutive years of decline from the exceptionally high yield in 1998 to marginally above-average yield in 2000, x 2002–2004: switch from above-average yield in 2002 to below-average yield in 2003 and switch back to above-average yield in 2004, x 2004–2006: three consecutive years of above-average yield, x 2006–2007: switch from above-average yield in 2006 to below-average yield in 2007. Exceptionally low and exceptionally high yields were noted for years 1991 and 1998, respectively: x 1991 exceptionally below-average yield: o mean (Tmean), minimum (Tmin) and maximum temperature (Tmax) were consecutively below average for years 1988–1991 (and through to 1994), o hydrological deficit showed a consecutive increase (i.e. a worsening) over four years from 1987 to 1990, and a decrease in 1991, o rainfall, which contributes to hydrological deficit, showed a consecutive decrease over four years from 1987 to 1990, switching back to above average in 1991. x 1997–1998 saw a large swing from below-average to exceptionally above-average yield: o rapid increase in Tmax between 1996 and 1998, o rapid increase (i.e. a worsening) in hydrological deficit between 1995 and 1997, o slight increase in rainfall from 1997 to 1998. As can be seen from Table 10 and the summary above, there are preliminary indications of the effects on yield of climate variables in the years leading up to the YOH. From the 1991 and 1998 data, it appears that yield is positively correlated with temperature. Hydrological deficit and decreasing rainfall may have contributed to poor yields in 1991. However, the high yields in 1998 were associated with a worsening in hydrological deficit. Not surprisingly, there is not an immediately obvious and consistent pattern for all YOH. 43 Table 10: Observations related to changes in annual average FFB yield (tons/ha) in relation to changes in annual average climate variables Rainfall and hydrological Yield (tons/ha) Temperature (°C) deficit (mm) 1990– consecutively below- 1987– Tmax: consecutively 1987– Hydrodeficit: 1992 average yield across 1994 below average across all 1991 consecutive increase all years years from below average in Tmin: switch from above 1987 to above average average in 1987 to below in 1991 average in 1988, Rainfall: consecutive declining further to the increase from below second highest below average in 1987 to average in 1989, above average in 1991 followed by an incline to marginally below average in 1991 1991 exceptionally below- 1991 Tmax: exceptionally average yield in 1991 below average in 1991 compared to all other compared to all other years years Tmin: marginally below average 1991– incline in yield from 1991– Tmax: rapid incline from 1991– Rainfall: switch from 1992 1991 to 1992, 1992 exceptionally below 1992 above average in 1991 although both years average in 1991 to to below average in are below average marginally below average 1992 compared to other in 1992 Hydrodeficit: switch years Tmin: rapid decline from from below average in marginally below average 1991 to above average in 1991 to below average in 1992 in 1992 1992– switch from below- 1992– Tmax: marginal incline 1992– Rainfall: switch from 1993 average yield in 1992 1993 from 1992 to 1993 1993 below average in 1992 to marginally above- Tmin: decline from below to above average in average yield in 1993 average in 1992 to 1993 exceptionally below Hydrodeficit: switch average in 1993 from above average in 1992 to below average in 1993 44 Rainfall and hydrological Yield (tons/ha) Temperature (°C) deficit (mm) 1993– consecutive decline 1993– Tmax: below average 1993– Rainfall: decline from 1996 from marginally 1996 with a decline from 1993 1996 above average in 1993 above-average yield in to 1994, followed by to below average in 1993 to below- switch from below 1996 average yield in 1995, average in 1994 to above Hydrodeficit: followed by a switch to average in 1995, consecutively below marginally above- switching back to below average from 1993 to average yield in 1996 average in 1996 1995 with a swing to Tmin: consecutive incline marginally above from exceptionally average in 1996 below average in 1993 to above average in 1996 1997– large swing from 1997– Tmax: rapid incline from 1996– Rainfall: incline from 1998 marginally below- 1998 above-average in 1997 to 1997– below average in 1997 average yield in 1997 exceptionally above 1999 to marginally below- to an exceptionally average in 1998 average in 1998 and above-average yield Tmin: rapid incline from switch to above average in 1998 above average in 1997 to in 1999 exceptionally above Hydrodeficit: rapid average in 1998 incline from marginally above average in 1996 to second highest above average in 1997, followed by consecutive decline to exceptionally below average in 1999 1998– consecutive decline 1998– Tmax: decline from 1998 1999– Rainfall: switch from 2000 from exceptionally 2000 to 1999 followed by 2001 above average with above-average yield in incline to second highest consecutive decline to 1998 to marginally above average in 2000 second highest below above-average yield in Tmin: switch from average in 2001 2000 exceptionally above Hydrodeficit: average in 1998 to consecutive incline from below average in 1999, exceptionally below followed by incline to average in 1999 to below average in 2000 above average in 2001 2000– consecutive incline 2000– Tmax: consecutive 2001– Rainfall: switch from 2002 from marginally 2002 decline from second 2002 second highest below above-average yield highest above average in average in 2001 to to above-average 2000 to above average in second highest above yield 2002 average in 2002 Tmin: incline from below Hydrodeficit: switch average in 2000 to above from above average in average in 2001, 2001 to second highest declining back to below average in 2002 marginally above average in 2002 45 Rainfall and hydrological Yield (tons/ha) Temperature (°C) deficit (mm) 2002– switch from above- 2002– Tmax: incline in above 2002– Rainfall: decline from 2004 average yield in 2002 2004 average from 2002 to 2003 second highest above to below-average in 2003 followed by rapid average in 2002 to 2003 and switching decline to marginally above average in 2003 back to above- above average in 2004 Hydrodeficit: switch average yield in 2004 Tmin: consecutive incline from second highest from marginally above below average in 2002 average in 2002 to above to average in 2003 average in 2004 (with continued incline through to 2006) 2004– all years have above- 2004– Tmax: switch from 2004– Rainfall: switch from 2006 average yield, the 2006 marginally above 2006 above average in 2004 highest yield for this average in 2004 to to exceptionally below period is in 2006 marginally below average average in 2005 in 2005, declining further compared to all other in 2006 years Tmin: consecutive incline Hydrodeficit: switch to second highest from below average in average Tmin in 2006 2004 to exceptionally above average in 2005, followed by decline to above average in 2006 2006– switch from above- 2006– Tmax: switch from below 2006– Rainfall: switch from 2007 average yield in 2006 2007 average in 2006 to above 2007 below average in 2006 to below-average yield average in 2007 to exceptionally above in 2007 Tmin: decline from average in 2007 second highest in 2006 Hydrodeficit: switch to above average in 2007 from above average in 2006 to average in 2007 Linear regression and multivariate analyses of climate variables and annual FFB yield data for Kwae Statistical correlations were undertaken using the statistical analysis program SPSS Statistics 17.0 to determine the significance of correlations between yield and climate variables for further regression analysis. Based on the literature review presented in Section 2.1 and the pair wise analysis described in the previous section, the predictor variables for correlation were defined as follows: x Annual average temperature for the current YOH (Year 0) and lagged by up to three years before YOH (Year -1, Year -2 and Year -3): o Tmean (°C), o Tmin (°C), o Tmax (°C). x Annual average rainfall for current YOH (Year 0) and lagged by up to three years before YOH (Year -1, Year -2 and Year -3). 46 x Hydrological deficit (IHRO method): o Annual hydrological deficit for current YOH (Year 0) and lagged by one year before YOH (Year -1), o 10-year annual average hydrological deficit at YOH, to evaluate whether the effects of long term deficit can be determined, o Relative humidity at 1500 hrs. The full set of correlation coefficients for the above predictor variables are presented in Annex B: Results of statistical analyses of FFB yields and climate variables. It should be noted that all of the above variables, except annual average rainfall, returned some statistically significant correlations for some YOP/YOH combinations, i.e. all variables did not consistently show statistically significant correlations for all YOP/YOH. Surprisingly, annual average rainfall for the current YOH and lagged by up to three years before YOH did not return any statistically significant correlations for any YOP, and at this stage it was therefore discounted from inclusion in the subsequent regression analyses. All the variables defined above were entered into the statistical modeling program using an ordered sequence of preferred methods for undertaking multivariate regression analysis (see Box 1 for further details of the methodology). Box 1: Description of statistical analysis method A sequence of regressions were undertaken where Tmax and hydrological deficit predictor variables were regressed individually and in combination with each other in SPSS using its Stepwise Backward method for regression. This method places all of the predictor variables into a model and the contribution of each predictor to the dependent variable (i.e. yield) is calculated based on a significance test. In SPSS, if predictor variables are not found to make a statistically significant contribution to the model they are removed and the program re- estimates the model using the remaining predictors. In some instances the preferred Stepwise Backward method resulted in exclusion of all of the predictor variables for certain YOP—i.e. none of the variables entered into the model were calculated to have a statistical significance to the predictor variable. In these cases, SPSS’s Stepwise Forward method was the next preferred method, whereby a model is defined that only contains the constant for the linear regression equation. SPSS then searches for predictors that best predict the dependent variable by selecting those that have the highest simple correlation with the outcome. If a predictor significantly improves the ability of the model to predict the outcome, it is retained and the program then searches for other predictors. The final, least preferred method is the Enter method which forces the program to include all of the predictors in the model simultaneously. This method is generally not recommended unless there are good theoretical reasons for forcing the inclusion of the chosen predictors. SPSS can generate a number of different models for the same set of predictors and independent variable. Each modeled outcome is provided with a model summary together with an analysis of variance (ANOVA) which presents the results of an F-test and significance test. The F-test is a ratio which is large (greater than at least 1) for a good model, indicating that the difference between the modeled data and observed data are small. The associated significance value for the F-test indicates the likelihood that the F-ratio would happen by chance alone. Therefore, if more than one model is suggested by SPSS, the best model is the one with a high F-test ratio and low F-test significance value. Table 11 presents the SPSS linear regression model outputs for the Kwae yield and climate 47 data sets where the F-test was >1 and the significance of the F-test ratio was lowest. All other models where the F-test was <1 are not detailed further. For the Kwae data sets, the statistics program accepted and returned statistically significant models for relationships between Tmax and hydrological deficit variables, either singularly or in combination, depending on the YOP being regressed. The program rejected other variables or generated models that were not statistically significant and were therefore considered to be a poor representation of observed data. Table 11 presents the SPSS linear regression model outputs for the Kwae yield and climate data sets. It can be seen from Table 11 column 2 that for some YOP, the modeling showed hydrological deficit variables were the best predictors of yield. For other YOP, Tmax variables were better. In some cases, both hydrological deficit and Tmax were good predictors. It should be noted of course that hydrological deficit and temperature are themselves correlated, i.e. high temperatures will lead to increased hydrological deficit. Furthermore, higher temperatures can help with fruit ripening, so higher temperatures can have both positive and negative effects on yields. Finally, temperature is positively correlated with radiation, which is needed for photosynthesis (see Section 2.1), and so statistical relationships between increased yields and higher Tmax may be reflecting the influence of higher levels of radiation. Comparing the observed yields (column 5) and the modeled yields (column 6) for Kwae generated from the linear regression calculations indicates that the models seem to be accurate at predicting observed long-term yields. Table 11: SPSS regression outputs for FFB yield (tons/ha) versus maximum temperature and hydrological deficit predictor variables SPSS regression mode and predictors Significance Observed Modeled YOP and YOH included in model F-test* of F-test** yield yield YOP:1979 SPSS Stepwise 14.9 0.005 136.2 136.7 YOH:1990– Backward mode 2000 Annual average Tmax Year 0, Year -1 Annual hydrological deficit Year 0, Year -1 10 year annual average hydrological deficit YOP: 1980 SPSS Stepwise 8.5 0.01 120.6 119.1 YOH:1991– Backward mode 2001 Annual average Tmax Year 0, Year -1 Annual hydrological deficit Year -1 YOP: 1981 SPSS Stepwise 6.5 0.034 129.0 130.0 YOH:1992– Backward mode 2001 Annual hydrological deficit Year -1 48 SPSS regression mode and predictors Significance Observed Modeled YOP and YOH included in model F-test* of F-test** yield yield YOP: 1982 SPSS Stepwise 9.6 0.01 134.9 132.6 YOH:1993– Backward mode 2002 Annual average Tmax Year -2 10 year annual average hydrological deficit YOP: 1985 SPSS Stepwise 6.6 0.029 158.7 158.3 YOH:1996– Backward mode 2006 Annual hydrological deficit Year -1 YOP: 1988–89 SPSS Stepwise 9.7 0.021 136.0 137.9 YOH:1999– Forward mode 2006 Annual hydrological deficit Year 0 YOP: OPRI SPSS Stepwise 22.1 0.003 60.5 59.0 1989 Forward mode YOH: 2000– Annual average Tmax 2006 Year -1 YOP: 1990 SPSS Stepwise 21.5 0.006 88.4 88.5 YOH: 2001– Forward mode 2006 Annual average Tmax Year 0 * Must be large (i.e. greater than 1) for a model to be considered “good” ** Likelihood that F-test ratio can be explained by chance alone. The lower the number, the better. Figure 30 presents charts showing the observed versus modeled yields. The charts allow visual comparison of year-on-year observed versus modeled yield to determine whether the models are following the general trend on a year-by-year basis, together with their ability to capture the peaks and troughs. It can be seen that the models generally appear to perform well, i.e. differences between observed and modeled yields in any one year are generally less than 20%. This gives us some confidence in the models. 49 Figure 30: Observed versus modeled FFB yield (tons/ha) for each YOP and across all YOH 50 51 Analysis of influence of climate change on yield Method To demonstrate the possible effects of climate change on yields in the future, the modeled yields outlined in the previous section were all recalculated as if they occurred 20 years later and 30 years later, under conditions of climate change, e.g.: x 1990–2000 yields from 1979 YOP were simulated as if harvesting were done in 2010–2020 (20 years later) and in 2020–2030 (30 years later), and x 1991–2001 yields from 1980 YOP were simulated as if harvesting were done in 2011–2021 (20 years later) and in 2021–2031 (30 years later). In this way, we are simulating two future programs of harvesting (starting in 2010 and 2020, respectively) which are identical to that which was undertaken in the past. We are then able to compare yields from the future simulated harvests with the historic (actual) harvest, to evaluate the differences that climate change could make. 52 Clearly, we recognize that, in reality, palms planted in the 1970s and 1980s will not still be in place by the 2020s–2030s, but the purpose of our analysis is to demonstrate the potential impacts of climate change on yields, by comparing historic yields with modeled future yields. This analysis uses the following projections provided in the UNDP climate change data set for Ghana, under the A2 greenhouse gas emissions scenario (see Section 1.2): x Mean, minimum and maximum modeled increases in temperature, x Mean, minimum and maximum modeled changes in precipitation. The modeled changes in precipitation were used to calculate changes in hydrological deficit, using the IRHO method. (Further details are provided in Annex C: Methods used to perturb observed yield models). Results Table 12 presents a summary of the observed and projected FFB yields in tons/ha. These are also presented graphically in Figures Figure 31 and Figure 32. Finally, Table 13 presents projected changes in FFB yields in tons. Examining the projected changes in yields in Table 12 and Figures Figure 31–Figure 32 shows that some of the models are projecting large decreases (as much as -75%), whereas some project very large increases (up to +166%) under the climate change scenarios. These yield changes are being driven by the projected changes in temperature and hydrological deficit. As outlined in Section 1.2, the climate models are not in good agreement for Ghana about whether rainfall in future will increase or decrease. (Reduced rainfall, combined with higher temperatures, would increase hydrological deficit.) This uncertainty may account for the wide variation in our future yield projections. Table 121: Summary of observed versus modeled min, mean and max FFB yield projected to 2010 and 2020 first years of harvest Total observed yield across Total projected yield, Projected Total projected yield, Projected all YOH first YOH beginning % change first YOH beginning % change YOP (tons/ha) 2010 (tons/ha) in yield 2020 (tons/ha) in yield 1979 1990–2000 2010–2020 YOH: -31 to 146% 2020–2030 YOH: -13 to YOH: 136.2 min: 94.6 min: 119 166% mean: 210 mean: 244 max: 335 max: 363 1980 1991–2001 2011–2021 YOH: 0 to 83% 2021–2031 YOH: 14 to 99% YOH: 120.6 min: 120 min: 138 mean: 174 mean: 197 max: 220 max: 240 1981 1992–2001 2012–2021 YOH: -14 to 12% 2022–2031 YOH: -14 to 14% YOH: 129 min: 145 min: 147 mean: 129 mean: 131 max: 110 max: 110 53 Total observed yield across Total projected yield, Projected Total projected yield, Projected all YOH first YOH beginning % change first YOH beginning % change YOP (tons/ha) 2010 (tons/ha) in yield 2020 (tons/ha) in yield 1982 1993–2002 2013–2022 YOH: -34 to 81% 2023–2032 YOH: -36 to 82% YOH: 134.9 min: 244 min: 246 mean: 88 mean: 87 max: negative yield/ max: negative yield/ model out of bounds model out of bounds 1985 1996–2006 2016–2026 YOH: -10 to 8% 2026–2036 YOH: -12 to 10% YOH: 158.7 min: 171 min: 175 mean: 157 mean: 159 max: 142 max: 140 1988/89 1999–2006 2019–2026 YOH: -10 to 16% 2029–2036 YOH: -14 to 17% YOH: 136 min: 122 min: 117 mean: 138 mean: 138 max: 158 max: 160 OPRI 2000–2006 2020–2026 YOH: -4 to 112% 2030–2036 YOH: 17 to 133% 1989 YOH: 60.5 min: 58 min: 71 mean: 95 mean: 106 max: 129 max: 141 1990 2001–2006 2021–2026 YOH: -15 to -68% 2031–2036 YOH: -49 to -75% YOH: 88.4 min: 85 min: 75 mean: 45 mean: 28 max: 7 max: negative yield/ model out of bounds Figure 31: Observed versus modeled FFB yield (tons/ha) for each YOP and across all YOH projected forward by 20 years 54 55 56 Figure 32: Observed versus modeled FFB yield (tons/ha) for each YOP and across all YOH projected forward by 30 years 57 58 Table 13 presents a summary of the observed versus modeled projected FFB yields in tons for each YOP. Overall, the models generally project increases in yield in future under climate change scenarios. Under the minimum climate change scenario, which is the scenario where rainfall decreases (see Section 1.2), the total yield across all YOP (shown in the bottom row of Table 13) is projected to show a slight decrease for the YOH beginning in 2010. However, for the YOH beginning in 2020, the total yield across 59 all YOP is projected to increase under the minimum scenario. This is surprising, given that the minimum scenario shows worse hydrological deficit. One explanation is that the projected increases in temperature are driving the projected yield increases in our models, through their influence on bunch ripening. Alternatively, as outlined in the previous section, models indicating increased yields under higher temperatures may, in fact, be reflecting the positive influence of higher radiation levels on yield. We sound a note of caution here about the postulated positive influences of temperature on yield. While our regression models suggest that yield is positively correlated with temperature, the published literature (Section 2.1) suggests that when temperatures become too high, they are detrimental to oil palm. These effects may not be captured in our models, as they were developed based on the observed temperature conditions, which are lower than temperatures that will be experienced in the future. Hence, it is possible that the projected yield increases shown in Table 13 are over-optimistic. Overall, despite the efforts made in our analysis, we find it difficult to draw strong conclusions about the future impacts of climate change on yields at Kwae. Table 13: Summary of observed versus modeled min, mean and max FFB yields (tons) projected to 2010 and 2020 first YOH Total observed yield across all Total projected change in yield, Total projected change in yield, YOP YOH (tons) first YOH beginning 2010 (tons) first YOH beginning 2020 (tons) 1979 1990–2000 2010–2020 YOH: 2020–2030 YOH: YOH: 102,295 min: -31,260 min: -12,887 mean: 55,200 mean: 80,779 max: 149,571 max: 170,232 1980 1991–2001 2011–2021 YOH: 2021–2031 YOH: YOH: 136,400 min: -488 min: 19,558 mean: 60,408 mean: 85,814 max: 112,884 max: 134,881 1981 1992–2001 2012–2021 YOH: 2022–2031 YOH: YOH: 50,777 min: 6,142 min: 7,219 mean: -28 mean: 758 max: -7,333 max: -7,335 1982 1993–2002 2013–2022 YOH: 2023–2032 YOH: YOH: 15,504 min: 12,565 min: 12,790 mean: -5,340 mean: -5,518 max: negative yield/model out of max: negative yield/model out of bounds bounds 1985 1996–2006 2016–2026 YOH: 2026–2036 YOH: YOH: 43,259 min: 3,301 min: 4,291 mean: -355 mean: -34 max: -4,509 max: -5,140 60 Total observed yield across all Total projected change in yield, Total projected change in yield, YOP YOH (tons) first YOH beginning 2010 (tons) first YOH beginning 2020 (tons) 1988/89 1999–2006 2019–2026 YOH: 2029–2036 YOH: YOH: 816 min: -83 min: -113 mean: 14 mean: 11 max: 134 max: 142 OPRI 2000–2006 2020–2026 YOH: 2030–2036 YOH: 1989 YOH: 496 min: -17 min: 84 mean: 286 mean: 374 max: 1, 557 max: 662 1990 2001–2006 2021–2026 YOH: 2031–2036 YOH: YOH: 1,723 min: -67 min: -253 mean: -842 mean: -1,174 max: -1,580 max: negative yield/model out of bounds Total 1990–2006 2010–2026 YOH: 2020–2036 YOH: across YOH: 351,270 min: -9,907 min: 30,689 all YOP mean: 109,343 mean: 162,184 max: 250,724 max: 293,442 2.4 Potential future impacts of climate change on financial performance Based on the wide ranges of future projected yields summarized in the previous section, Table 14 below provides ranges of future projected changes in revenue from oil palm products (assuming all CPO is converted to olein and stearin). As noted above, the yield projections range from negative to strongly positive, on account of the uncertainties in the modeling. x The projected total changes in revenue across the 2010–2026 YOH range from -$1.3m to +$35m (undiscounted). o Taking a mid-point year of 2018 and applying a discount rate of 12% gives annual projected changes in revenue of -$27,000 to + $720,000. x Across the 2020–2036 YOH, the projected total revenue changes range from +$4.2m to +$41m (undiscounted). o Taking a mid-point year of 2028 and applying a discount rate of 12% gives annual projected changes in revenue of +$29,000 to +$280,000m. 61 Table 14: Summary of modeled min, mean and max FFB (tons) projected to 2010 and 2020 first YOH and the associated projected changes in revenue (undiscounted). Revenues are calculated assuming that all FFBs are converted to olein and stearin products, using GOPDC average sale prices for olein and stearin in 2009 provided in GOPDC’s financial model. Total projected Total projected change in yield change in yield (compared with Projected change (compared with Projected change observed yield), first in revenue, first observed yield), first in revenue, first YOH beginning 2010 YOH beginning YOH beginning 2020 YOH beginning YOP (tons) 2010 ($×000) (tons) 2020 ($x000) 1979 2010–2020 YOH: 2010–2020 YOH: 2020–2030 YOH: 2020–2030 YOH: min: -31,260 min: -4,331 min: -12,887 min: -1,785 mean: 55,200 mean: 7,648 mean: 80,779 mean: 11,192 max: 149,571 max: 20,723 max: 170,232 max: 23,585 1980 2011–2021 YOH: 2011–2021 YOH: 2021–2031 YOH: 2021–2031 YOH: min: -488 min: -68 min: 19,558 min: 2,710 mean: 60,408 mean: 8, 369 mean: 85,814 mean: 11,889 max: 112,884 max: 15, 640 max: 134,881 max: 18,687 1981 2012–2021 YOH: 2012–2021 YOH: 2022–2031 YOH: 2022–2031 YOH: min: 6,142 min: 851 min: 7,219 min: 1,000 mean: -28 mean:-4 mean: 758 mean: 105 max: -7,333 max:-1,016 max: -7,335 max: -1,016 1982 2013–2022 YOH: 2013–2022 YOH: 2023–2032 YOH: 2023–2032 YOH: min: 12,565 min: 1,741 min: 12,790 min: 1,772 mean: -5,340 mean: -740 mean: -5,518 mean: -765 max: negative values/ max: NA max: negative values/ max: NA model out of bounds model out of bounds 1985 2016–2026 YOH: 2016–2026 YOH: 2026–2036 YOH: 2026–2036 YOH: min: 3,301 min: 457 min: 4,291 min: 595 mean: -355 mean: -49 mean: -34 mean: -5 max: -4,509 max: -625 max: -5,140 max: -712 1988/89 2019–2026 YOH: 2019–2026 YOH: 2029–2036 YOH: 2029–2036 YOH: min: -83 min: -11 min: -113 min: -16 mean: 14 mean: 1 mean: 11 mean: 2 max: 134 max: 19 max: 142 max: 20 OPRI 2020–2026 YOH: 2020–2026 YOH: 2030–2036 YOH: 2030–2036 YOH: 1989 min: -17 min: -2 min: 84 min: 12 mean: 286 mean: 40 mean: 374 mean: 52 max: 1,557 max: 216 max: 662 max: 92 1990 2021–2026 YOH: 2021–2026 YOH: 2031–2036 YOH: 2031–2036 YOH: min: -67 min: -9 min: -253 min: -35 mean: -842 mean: -117 mean: -1,174 mean: -163 max: -1,580 max: -219 max: negative values/ max: NA 62 Total projected Total projected change in yield change in yield (compared with Projected change (compared with Projected change observed yield), first in revenue, first observed yield), first in revenue, first YOH beginning 2010 YOH beginning YOH beginning 2020 YOH beginning YOP (tons) 2010 ($×000) (tons) 2020 ($x000) model out of bounds Total 2010–2026 YOH: 2010–2026 YOH: 2020–2036 YOH: 2020–2036 YOH: across all min: -9,907 min: -1,372 min: 30,689 min: 4,253 modeled mean: 109,343 mean: 15,148 mean: 161,010 mean: 22,307 YOP max: 250,724 max: 34,738 max: 293,442 max: 40,656 2.5 Concluding remarks and suggested adaptation actions As discussed in Section 2.3, the yield modeling is fraught with uncertainties and confounding factors, so these projections can only be considered as indicative of what might happen. The models we have developed in this analysis were good at reproducing the observed yields at Kwae. However, when we project into the future to take account of climate change, the models are operating outside the temperature ranges in the observed record, and may therefore not perform well. Furthermore, there are uncertainties about whether rainfall will increase or decrease in future (see Chapter 1). Finally, as outlined in Section 2.1, climatic factors have differing effects on various aspects of the lifecycle of oil palm. All in all, we are not able to state with confidence whether climate change is likely to lead to increases or decreases in oil palm yield at Kwae. Existing and proposed adaptation actions GOPDC is already undertaking a number of actions which help build climate resilience on its plantations: x Mulching/composting of empty fruit bunches (EFBs) and palm fronds, and applying the mulch to the plantations, which helps preserve soil moisture. x Planting cover crops, and sub-soiling across slopes (i.e. planting palms in 1m deep trenches), both of which help to reduce soil erosion. (GOPDC does not consider terracing or bunding to be necessary to prevent soil erosion and preserve soil moisture, as the slopes on GOPDC’s plantations are shallow, unlike in Malaysia where terracing is used). x Testing the benefits of applying Palm Oil Mill Effluent (POME) as a source of nutrients on a small area of palms (48ha, 1979 YOP) at Kwae, to see if this increases yield. It should be noted that the main reason for doing this is not for nutrient/irrigation purposes, but to make use of the effluent, rather than disposing of it via the wastewater treatment system. There will not be enough POME to apply to the whole Kwae Estate (5,000 ha). GOPDC does not consider that it is useful to test the yield benefits of irrigating the palms using groundwater, as there will not be sufficient groundwater resources to irrigate the whole estate and irrigation would be unsustainable and expensive. Experts at CIRAD have recommended that GOPDC improve soil structure in upland areas using EFBs. Finally, GOPDC recognizes that research is needed into new OP seeds, better able to tolerate the relatively low rainfall conditions experienced in Ghana. 63 Chapter 2 References Abdullah, R. (2003). Short-Term and Long-Term Projection of Malaysian Palm Oil Production. Oil Palm Industry Economic Journal, 3(1), pp 32–26. Asamoah T.E.O., Dwarko, D.A., Nuertey, B.N., Okyere-Boateng, G., Baidoo-Addo, H., Marfo-Ahenkora, E. and Opoku, A. (2008). Effect of Water Stress on the Yield Performance of Seven Dura x Pisifera Oil Palm Progenies. Journal of Ghana Science Association, 10(1). Breure and Menendez. (1990). In: Muhamad Rizal: Rainfall impact on Oil Palm Production and OER at FELDA Triang 2. Chang, K.C., Foster, H.l. & Abas, Z. (1988). Monthly frond production of oil palm in Malaysia. Oleagineaux, 43, pp 439–444. Chang, K. C., Rao, V., Hasnuddin, M. Y. and Zakaria, A. (1993). The role of developmental times of inflorescences in the seasonal flowering behavior of oil palm. Unpublished. 14 pp. Corley, R.H.V. and Gray, B.S. (1976). Yield and yield components. Oil Palm Research (Corley, R.H.V., Hardon, J.J. and Wood, B.J. eds.). Elsevier Scientific Pub. Co., Amsterdam, p. 78–86. Corley, R.H.V. (1977). Oil palm yield components and yield cycles. In: International developments in oil palm (Ed. By D.A. Earp & W. Newell), pp. 116–129, Incorp. Soc. Planters, Kuala Lumpur. Corley, R.H.V. and Tinker, P.B. (2003). The Oil Palm. Fourth Edition. Wiley-Blackwell. Goh, K.J., Chew, P.S. and Teo, C.B. (1994). Commercial yield performance of oil palm in Sabah, Malaysia. Planter, Kuala Lumpur, 70, pp 497–507. Griffiths, W., Fairhurst, T. Rankine, I., Gfroerer Kerstan, A., and Taylor, C. (2002). Identification and elimination of yield gaps in oil palm. Use of OMP7 and GIS1. Hardon, J.J., Williams, C.N. and Watson, I. (1969). Leaf area and yield in the oil palm in Malaya. Expl. Agric., 5, pp 25–32. Hardon, J.J. and Corley, R.H.V. Pollination. Oil Palm Research (Corley, R.H.V., Hardon, J.J. and Wood, B.J. eds.). Elsevier Scientific Pub. Co., Amsterdam, pp 300–305. nd Hartley, C.W.S. (1977). The Oil Palm. 2 edition. Longmans, London. rd Hartley, C.W.S. (1988). The Oil Palm. 3 edition. Longmans, London. Henson, I.E. (1991). Limitations to gas exchange, growth and yield of oil palm by soil water supply and atmospheric humidity. Trans. Malay. Soc. Pl. Physiol., 2, pp 39–45 Henson, I.E. and Chai, S.H. (1998). Analysis of oil palm productivity. III. Seasonal variation in assimilate requirements, assimilation capacity, assimilate storage and apparent photosynthetic conversion efficiency. J. Oil Palm Research, 10(1) pp 35–51. Henson I.E., & Chang. (2000). Oil palm productivity and its component processes. In: Advances in oil palm research, (Eds. Basiron, Y., Jalani, B.S. and Chan K.W.), 1, pp 97–145, Malaysian Palm Oil Board, Kuala Lumpur. 64 Henson, I.E. and Dolmat, M.D. (2004). Seasonal Variation in Yield and Developmental Processes in an Oil Palm Density Trial on a Peat Soil: 2. Bunch Weight Components. Journal of Oil Palm Research, 16 (2) pp 106–120. Henson, I.E. (2005). Modelling seasonal variation in oil palm bunch production using a spreadsheet programme. J. Oil Palm Research, 17 pp 27–40. Henson, I.E, Roslan Md Noor, M., Haniff Harun, M., Yahya, Z. and Nor Aishah Mustakim, S. (2005). Stress development and its detection in young oil palms in north Kedah, Malaysia. J. Oil Palm Research, 17(1) pp 11–26. Lamade, E., Bonnot, F., Pamin, K. and Setyo, I.E. (1998). Quantitative approach of oil palm phenology in different environments for La Me Deli and Yangambi Deli materials. Investigations in the inflorescence cycles process. Proc. of the 1998 International Oil Palm Conference. Nusa Dua, Bali, Indonesia. pp 287–301. McSweeney, C., New, M. and Lizcano, G. (2008). UNDP Climate Change Country Profiles – Ghana. Maillard, G., Daniel C. & Ochs R. (1974). Analyse des effects de la secheresse sur le palmier a huille. Oleagineux, 29, pp 397–404. Monteith, J.L. (1977). Climate and the efficiency of crop production in Britain. Philosophical Transactions of the Royal Society, B., 281, pp 277–294. Noor, M.R. and Haniff Harun, M. (2004) Water deficit and Irrigation in Oil Palm: A Review of Recent Studies and Findings. Oil Palm Bulletin 49, p.16. Oboh, B.O. and Fakorede, M.A.B. (1999). Effects of weather on yield components of the oil palm in a forest location in Nigeria. Journal of Oil Palm Research, 11(1), pp 79–89. Paramananthan, S., Chew, P.S., and Goh, K.J. (2000). Towards a practical framework for land cultivation st for oil palm in the 21 century. In: Proc. Int. Planters Conf. ‘Plantation tree crops in the new millennium: the way ahead’ (Ed. E. Pushparajah), Incorp. Soc. Planters, Kuala Lumpur, pp. 869– 885, Raw Materials Research and Development Council (RMRDC), Abuja. (2004). Report On Survey Of Selected Agricultural Raw Materials In Nigeria. Roslan Md Noor, M. and Haniff Harun, M. The Role of Leaf Area Index (LAI) in Oil Palm. Oil Palm Bulletin. Malaysian Palm Oil Board (MPOB). Sparnaaij, L. D., Rees, A.R., and Chapas, L.C. (1963). Annual yield variation in the oil palm. J. W. Afr. Inst. Oil Palm Res., 4 pp110–125. Squire, G.R. (1985). A physiological analysis for oil palm trials. PORIM Bulletin No. 12, pp 12–31. Turner, P.D. (1977). The effects of drought on oil palm yields in South-east Asia and the south Pacific region. In: International developments in oil palm (Eds. D.A. Earp & W. Newall), Incorp. Soc Planters, Kuala Lumpur, pp 673–694. 65 Chapter 3: Climate risk analysis for oil palm pests and diseases at GOPDC 66 Overview While we know that pests and diseases are in general affected by climate (Iglesias & Rosenzweig, 2007; Corley & Tinker, 2003), little is known about their specific climate sensitivities. However, as pests and diseases can be such a major cause of damage to oil palm plantations, this chapter presents the available information to begin exploring connections between climate change, pest and disease infestation and oil palm production success. 3.1 Pests and diseases relevant to oil palm at GOPDC and potential for losses of FFB Until World War II, it seems that oil palm was largely free from serious pests and diseases. However, as the area under the crop expanded, serious outbreaks have occurred around the world. In Africa, the diseases Fusarium wilt and dry basal rot have caused considerable damage, as have a variety of arthropod and mammal pests (Corley & Tinker, 2003). With regard to West Africa, noted oil palm diseases include brown germ (which affects germinating seeds), seedling leaf diseases such as Cercospora leaf spot, the seedling root disease, blast, and stem and root diseases dry basal rot and Fusarium wilt. Pests include nursery pests such as red spider mite, several species of Oryctes beetle which cause stem damage to young palms, several species of Coelaenomenodera leaf miner which is a leaf pest of mature palms, as well as rats and birds (Corley & Tinker, 2003). At GOPDC, pests and diseases that disturb or attack the oil palm during the germination and nursery stages include caterpillars, rodents, blast, antrachnose, Cercospora elaeidis leaf spot and bud rot; while those that are effective in the plantation include grass cutters, rodents, Oryctes rhinoceros beetles, Rhynchophorus ferrugineus (red palm weevils), the Coelaenomenodera lameensis leaf miner and Fusarium wilt disease (GOPDC, 2008). The most notable infestations at GOPDC have been by the leaf miner (C. lameensis), which causes widespread defoliation. “Very extensive” defoliation by the leaf miner was observed on approximately 2,000ha of GOPDC plantation in early 1987, when action was not being taken to control it, with a total of 13,000 tons FFB estimated lost that year (Asamoah & Appiah, 1998). This defoliation thus affected 75% of 2 3 the plantation land , and led to a 54% loss of tons FFB on that land . Losses are also known to have continued through 1988 and 1989 (GOPDC, Pers. Comm.). According to Appiah et al. (2007), during a heavy leaf miner attack on oil palm, with about 90% of fronds defoliated, there can be a 50% loss in FFB yields over a two year period. If in 2009, 2,000ha were affected again by the leaf miner, leading to 54% losses in yield for that area (as in 1987), this would equate to revenues not earned of $1.6m (if FFB were converted to CPO), and to revenues not earned of $1.8m (if FFB were converted to olein and stearin) (Table 15). If in 2009, in a more 4 extreme scenario, 75% of the current plantation was affected by the leaf miner (i.e. 3,135ha ) and there were losses of 50% in this area for two years (as suggested by Appiah et al., 2007), this loss of FFB would equate to revenues not earned of $4.6m (if FFB were converted to CPO), and a total of $5.3m (if FFB 2 In 1987, 2,663ha were planted with oil palm at Kwae. 3 The average yield per hectare is 12.12 tons, calculated from GOPDC monthly yield data for 1990 to 2007 (i.e. years after the 1987 leaf miner impacts). Expected yield for the 2,000ha affected would have been 24,240 tons in 1987 (i.e. 2,000 ha × 12.12 tons/ha). Thus a 13,000 tons FFB loss (across the 2,000ha) equates to 54% of what would be expected in a “normal” year. 4 There were 4,180ha of oil palm plantation in Kwae in 2007 (GOPDC, “Germinated seed_2008” workbook). 67 were converted to olein and stearin, Table 16). (Note that losses of FFB for 2009 and 2010 were estimated using the average yield per hectare figure of 12.12 tons/ha.) Table 15: Estimated revenues not earned from oil palm products as a result of a 54% loss of FFB across 2,000ha in the year 2009 (i.e. 13,000 tons), as was lost to a leaf miner invasion at GOPDC in 1987 Revenues not earned GOPDC average sale ($x000), resulting Oil palm product Tons of product lost price ($/ton) for 2009 from FFB losses CPO* 2,730 580 1,583 Olein** 1,739 755 1,313 Stearin*** 745 655 488 *CPO (tons) equates to 21% of FFB (tons) (see Chapter 2, Yield). **Olein (tons) equates to 70% of RBDPO (which in turn is 91% of CPO). ***Stearin (tons) equates to 30% of RBDPO (which in turn is 91% of CPO). Table 16: Revenues not earned from oil palm products as a result of a 50% loss in FFB yield across two years (e.g. 2009 & 2010) for 75% of the plantation at Kwae* Revenues not earned Oil palm product (as GOPDC average sale ($x000), resulting in Table 15) Tons of product lost price ($/ton) for 2009 from FFB losses CPO 7,979 580 4,628 Olein 5,083 755 3,837 Stearin 2,178 655 1,427 *Total projected FFB yield from Kwae for 2009 & 2010 is 101,320 tons FFB according to IFC’s Financial Model for GOPDC. 75% of this is 75,990 tons. Thus, a 50% reduction in this two-year yield equates to 37,995 tons. 3.2 Pest and disease control at GOPDC and costs In order to avoid losses of FFB, various disease and pest control methods are employed at GOPDC. Insecticides, fungicides and rodenticides are used in the germination and nursery stages as well as elimination (to treat bud rot fungal disease) (GOPDC, 2008). In the plantation, in an effort to maintain 5 organic status and adopt an Integrated Pest Management system , non-insecticide methods are used where possible. Examples include the use of wire nets to keep grass cutters and rodents away from palms, the handpicking of Oryctes rhinoceros beetles, the manual removal of red palm weevils and weaver bird nests, and the use of resistant hybrids to avoid a Fusarium wilt infestation. 5 Integrated Pest Management (IPM) systems involve the encouragement of biological control of pests, the adoption of agronomic methods that minimize the risk of pest outbreaks and the selective use of chemicals where pesticide application is unavoidable (Corley & Tinker, 2003). 68 In the case of leaf miner infestation however, insecticides are often employed in a process known as hot fogging. Phytosanitary surveillance is conducted on an ongoing basis and insecticide is only applied after infestation levels have exceeded the leaf miner adult index level of 10. GOPDC reports that parts of the plantation need to be treated with insecticide every 1 to 2 years to control the leaf miner. In 2007, about 25% of the plantation at Kwae needed treatment, and some palms were damaged by the pest (GOPDC, Pers. Comm.). In order to calculate the total costs associated with oil palm pests and diseases, it is important to consider the direct losses of FFB associated with pest and disease damage, as discussed above, as well as the costs of pest/disease control (in terms of labor and resources). The costs of pest/disease control to GOPDC are not known to us. It is also worth accounting for reductions in bunch yield which may result from the death of pollinators due to the application of pesticides to treat leaf miner infestation (Asamoah & Appiah, 1998) (see Chapter 4, Ecosystem services). 3.3 Climate sensitivities of pests and diseases and possible impacts of climate change In order to assess climate risks to GOPDC, it is important to consider how climate change might affect pest and disease occurrence, which in turn may affect yield and revenue. Those pests and diseases that are found at GOPDC and are also known to be affected by climate are listed in Table 17 (with information about their specific climate sensitivities). Table 17: Pests and diseases observed at GOPDC which have climatic sensitivities noted in the literature Pest/disease Climate sensitivity Cercospora leaf spot “Adequate water supplies” (Corley & Tinker, 2003, p393) are required for (seedling leaf disease) nurseries to remain free/largely free of this disease. Blast (seedling root Too little shade and over/under-irrigation of polybags during the short dry disease) season can lead to the development of blast (Corley & Tinker, 2003). It is also worth noting that the vector for blast is a leafhopper, Recilia mica, which is primarily associated with grasses (Paspalum & Pennisetum) (Howard et al., 2001). While information on the climate sensitivities of this vector and grass habitat are presently unavailable, it is worth noting that it is not just the climate sensitivity of the disease/pest, but also its vectors and habitats that are important in considering climate change impacts. Oryctes rhinoceros (stem Rain has been noted to reduce the immigration rate of O. rhinoceros into pest of young palms) blocks of 3–5 year old coconuts in New Britain (Bedford, 1980), although a rainfall threshold is not known. Leaf miner Female leaf miners lay zero or few eggs under “very dry conditions” in (Coelaenomenodera Ghana (Asamoah & Appiah, 1998, p2). However, the leaf miner is known lameensis, leaf pest of to have invaded Ghana 20 years ago during a drought (GOPDC, Pers. mature palms) Comm.). Appiah (Pers. Comm.) also notes that leaf miner incidence reduces during the wet season. According to GOPDC (Pers. Comm.) leaf miner larvae development inside palm leaves is adversely affected by high temperatures. In a laboratory study of Dialectica scalariella, a leaf-mining lepidopteron, on its host plant Paterson’s Curse, Echium plantagineum L. 69 Pest/disease Climate sensitivity (Boraginaceae), Johns and Hughes (2002) found that a doubling of CO2 concentration (from ambient levels of 360 ppmv to approximately 700 SSPY DQGWHPSHUDWXUHLQFUHDVHRIÛ& IURPÛ&WRÛ&GXULQJWKHGD\ DQGIURPÛ&WRÛ&DWQLJKW KDGDVLJQLILFDQWQHJDWLYHLPSDFWRQWKH pest. (Note that according to the Intergovernmental Panel on Climate Change (IPCC), CO2 concentrations are projected to be about 400ppmv by (a) 2020 and in the range 450 to 500ppmv by 2040 ). Although the CO2 increase slowed insect development, the net result from the additional temperature increase was that larval development was accelerated by 14 days. Rather than leading to the production of more generations per year, this led to reduced emergence success of larvae and a decline in moth weight which is linked to lowered fecundity. Reduced emergence success is suggested to have resulted from the inability of larvae to consume adequate nitrogen to support their temperature-driven development rate (perhaps partly caused by the CO2 and temperature-driven hastening of leaf development, leading to more mature leaves with lower nitrogen concentration) (Johns & Hughes, 2002). As well as leaf miner responses, it is also possible that some ant species which predate on them (Corley & Tinker, 2003; Appiah, publication date unknown) are affected by changes in climate. Grasshoppers More are present during droughts (GOPDC, Pers. Comm.) Caterpillars While we are not aware of the climate sensitivities of caterpillars themselves, Siburat and Mojiun “observed outbreaks of leaf-eating caterpillars after floods, which might have eliminated natural enemies whose adults lived on ground vegetation” (Corley & Tinker, 2003, p426). They also suggested that drought could have a similar effect. (a) Source: ISAM model (reference) in Appendix II SRES Tables, IPCC, 2007. As is clear from Table 17, information about the relationship between oil palm pests and diseases and climate in West Africa is lacking. With regard to temperature, although the D. scalariella leaf miner shows DQHJDWLYHUHVSRQVHWRDÛ&LQFUHDVHLWLVGLIILFXOWWRUHODWHWKLVWRUHVSRQVHVRIWKHRLOSDOPC. lameensis leaf miner. First, both the host plant and insect pest are different. Second the increase in daytime temperature for D. scalariella ZDVIURPÛ&WRÛ&ZKLOHWKHLQFUHDVHVSUHGLFWHGIRU*KDQDE\WKH UNDP are 1.2°C (low to high range of 0.8–Û& E\WKHVDQGƒ& Û&WRÛ& E\WKHV IURPDPHDQGDLO\WHPSHUDWXUHEDVHOLQHRIÛ&DW$NLP2GD –2007) (UNDP, 2008; Chapter 1, Climate). As shown in Table 17, with regard to rainfall and water supply, some pests and diseases respond positively to increases while others respond negatively. As no thresholds are known with respect to rain and water availability, it is difficult to speculate how pests and diseases will respond under future precipitation scenarios taking account of climate change. However, in the case of the Ghanaian oil palm leaf miner, there is evidence to suggest that extremes of wet and dry conditions have a negative impact on the pest. As the UNDP climate projections for Ghana indicate that the proportion of total annual rainfall that falls in “heavy” events will tend to increase in the future, it is possible that the leaf miners will be adversely affected, which may in turn reduce the threat of FFB loss. It is important to note however that there is significant uncertainty about this expectation. According to GOPDC, the leaf miner larvae are also negatively affected by high temperatures, so it is possible that higher temperatures due to climate change might reduce their prevalence. 70 It is also important to be aware of the complex structure of ecosystems in which oil palm pests and diseases operate. While pests and diseases may be affected by future climate change and may thus affect yield, their natural predators may also be affected by climate, which in turn may affect yield. Additionally, of course, other factors such as the use of pesticides on a plantation can affect the presence of pest/disease predators and thus may influence yields (Corley & Tinker, 2003). Finally, it is worth considering the risks posed by the desert locust (Schistocerca gregaria) whose distribution area extends from West Africa to India (see Box 2), and which had a devastating outbreak in West Africa in 2004 (NASA, 2009). While the most up-to-date information from the Food and Agricultural Organization (FAO) shows that Ghana is not in the immediate area of threat of current Desert Locust Infestation, the FAO is presently conducting research on the potential impacts of climate change on desert locust range (FAO, 2009). It may be useful for GOPDC to follow the outcomes of the FAO research, to ascertain whether locust ranges are projected to move across Ghana. 3.4 Recommended adaptation actions Due to the complex nature of the relationship between oil palm pests and diseases and climatic conditions, a sensible adaptation action would be to invest in understanding the risks better. By working together with in-country and international experts (e.g. the Oil Palm Research Institute (OPRI), University of Ghana Kade Agricultural Research Centre and the Centre de coopération internationale en recherche agronomique pour le développement (CIRAD)), GOPDC could increase its understanding of the impacts of climate change on pests and diseases, and so on oil palm. According to the Ghana Environmental Protection Agency (EPA, Pers. Comm.), once Ghana’s Climate Change Adaptation Strategy (CCAS) is published, GOPDC can submit proposals for research on climate change impacts on pests and diseases of oil palm. It is also recommended that GOPDC correlate observed climate data and leaf miner egg/larvae monitoring data from its plantations, to understand if and how they are related. Similar correlations could be developed for other pests and diseases affecting the plantations. These actions may also facilitate GOPDC in developing an early detection system for pest and disease outbreaks. Box 2: The desert locust Plagues of desert locust, Schistocerca gregaria, have been recognized as a threat to agricultural production in Africa and western Asia for thousands of years. Normally, the desert locust is a solitary insect that occurs in desert and scrub regions of northern Africa, the Sahel (Burkina Faso, Chad, Mali, Mauritania, and Niger), the Arabian Peninsula and parts of Asia to western India. During the solitary phase (green area on map below), locust populations are low and present no economic threat. However, after periods of drought, when vegetation flushes occur in major desert locust breeding areas, rapid population build-ups and competition for food occasionally result in a transformation from solitary behavior to gregarious behavior. Following this transformation, which can occur over two or three generations (durations of locust life cycles are variable, depending on species and environmental conditions; in Africa, there are generally 3-5 generations per year) locusts often form dense bands of flightless nymphs and swarms of winged adults that can devastate agricultural areas. Under these conditions, the locusts' recession area can expand to envelop the sub-Sahel from Guinea to Tanzania, the Middle East, western Asia to Bangladesh and parts of southern Europe (yellow areas on map) (Showler 1995a,b). 71 Map from http://earthobservatory.nasa.gov/Features/Locusts/locusts2.php adapted from Showler, 1996 A single swarm of locusts can be small (hundreds of square meters) or huge, composed of billions of locusts, with up to 80 million per square kilometer over an area of more than 1,000 square kilometers. In one day, a swarm of locusts can fly 100km in the general direction of prevailing winds. Bands of nymphs can march about 1.5 km per day. Plagues often involve hundreds of swarms. Chapter 3 References Appiah, S. O. (publication date unknown) Incidence and intensity of the oil palm leaf miner Coelaenomenodera minuta uhmann (Coleoptera: Chrysomelidae; Hispinae) in oil palm plantations in Ghana. Oil Palm Research Institute, Kade. Appiah, S. O., Dimkpa, S. O. N., Afreh-Nuamah, K., Yawson, G. K. (2007). The Effect of Some Oil Palm Elaeis guineensis Jacq. Progenies on the Development of the Oil Palm Leaf Miner, Coelaenomenodera lameensis Berti and Mariua (Coleoptera: Chrysomelidae) in Ghana. African Journal of Science and Technology, Science and Engineering Series, 8 (2): 92–96. Asamoah, T. E. O., Appiah, S. O. (1998). Agro-management strategies towards mitigating the effects of damage caused by the leaf miner, coelaenomenodera minuta uhmann (Coleoptera: Chrysomelidae: Hispinae) in Ghana. Oil Palm Research Institute, C. S. I. R, Kade. Bedford, G. O. (1980). Biology, Ecology, and Control of Palm Rhinoceros Beetles. Ann. Rev. Entoml., 25: 309–339. th Corley, R. H. V., Tinker, P. B. (2003). The Oil Palm (4 ed.) Wiley, Hoboken, NJ, USA. GOPDC. (2008). Environmental Management Plan. FAO. (2009) Desert Locust situation update, http://www.fao.org/ag/locusts/en/info/info/index.html, Date accessed 01/05/09. 72 Howard, F.W., Giblin-Davis, R., Moore, D., Abad, R. (2001). Insects on Palms. CABI Publishing, UK. Iglesias, A., Rosenzweig, C. (2007). Climate and Pest Outbreaks. In: D. Pimentel (Ed.) Encyclopaedia of Pest Management, Volume II (pp. 87–89). Taylor & Francis, UK. Johns, C. V., Hughes, L. (2002). Interactive effects of elevated CO2 and temperature on the leaf-miner Dialectica scalariella Zeller (Lepidoptera: Gracillariidae) in Paterson's Curse, Echium plantagineum (Boraginaceae). Global Change Biology, 8: 142–152. McSweeney, C., New, M., Lizcano, G. (2008). UNDP Climate Change Country Profiles – Ghana. NASA. (2009). Locusts plague North and Western Africa, http://earthobservatory.nasa.gov/IOTD/view.php?id=4905, Date accessed 01/05/09. Showler, A.T. (1995a.) Desert locust control, public health, and environmental sustainability in North Africa, pp. 217–239. In W.D. Swearingen & A. Bencherifa [eds.], The North African environment at risk. Westview Press, Boulder, CO. Showler, A.T. (1995b.) Locust (Orthoptera: Acrididae) outbreak in Africa and Asia, 1992–1994: an overview. Amer. Entomol. 41: 179–185. Showler, A.T. (1996). The Desert Locust in Africa and Western Asia: Complexities of War, Politics, Perilous Terrain, and Development. Kika de la Garza Subtropical Agricultural Research Center, USDA-ARS, Texas, USA. 73 Chapter 4: Climate risk analysis for ecosystem services 74 Overview Ecosystem services are “the benefits people obtain from ecosystems” (Millennium Ecosystem Assessment, 2005, p. v). They can be organized into four types of service: provisioning services (e.g. food), regulating services (e.g. climate regulation), cultural services (e.g. recreation) and supporting services (e.g. nutrient cycling) (Millennium Ecosystem Assessment, 2005). Trivedi et al. (2008) also characterize them as either having direct benefits to people (e.g. provisioning services and some cultural services) or indirect benefits (e.g. regulating and supporting services). The Millennium Ecosystem Assessment (2005) reminds us that the various species within an ecosystem will respond differently to environmental change, so that established equilibria between species in an ecosystem may not be maintained: “Within functional groups, species respond differently to environmental fluctuations. This response diversity derives from variation in the response of species to environmental drivers, heterogeneity in species distributions, differences in ways that species use seasonal cycles or disturbance patterns, or other mechanisms.” This means, for instance, that balances between predators and their prey may change, as climate changes, with potentially unforeseen consequences. In this chapter, we identify ecosystem services that are relevant to GOPDC and consider how they might positively or negatively affect oil palm yield and, in turn, GOPDC’s financial return. The two most important issues in this respect are pollination of the oil palm and the ecosystem services provided by GOPDC’s Biodiversity Plots (BDPs). In terms of pollination, which is a regulating service, we consider climate risks to pollinators and propose consequent potential impacts on yield and income. In terms of the BDPs, we consider any ecosystem services they may provide and what role they might play in building resilience to the impacts of climate change. We stress that understanding of how climate change will interact with ecosystem services remains uncertain. The aim of this chapter is to demonstrate potential issues that could be important to GOPDC, so that the company can consider whether it would be beneficial to understand more about ecosystem services and their interplay with climate risks, through becoming involved in research that is underway exploring these issues. 4.1 Pollination Pollinators relevant to oil palm globally, in West Africa and at GOPDC While it was originally thought that oil palm was largely wind pollinated, Syed, in his study of Malaysia and Cameroon in 1979 and 1982, found that the palm was mainly insect pollinated (Corley & Tinker, 2003; CABI, 2003). Oil palm insect pollinators vary across the globe, with the weevil Elaeidobius kamerunicus the key pollinator for West Africa where it is native (Corley & Tinker, 2003). As would be expected, E. kamerunicus is also the key pollinator for GOPDC (GOPDC, Pers. Comm.). Interestingly, while pollination has always been acceptable in West Africa, it has not been in other regions such as Southeast Asia (where different insect pollinators are more prevalent). Thus, in the same way that oil palm was originally translocated to Southeast Asia, so too was the E. kamerunicus, in an effort to increase yields (Chinchilla- López & Richardson, 1991). Although E. kamerunicus is known for supporting good oil palm yields, there are instances in which inadequate fruit sets sometimes occur. In an effort to calculate a minimum weevil population to ensure a good fruit set, Donough et al. (In Corley & Tinker, 2003) estimated that approximately 20,000 weevils/ha was sufficient, and Syed and Saleh (In Corley & Tinker, 2003) considered 700 weevils/female inflorescence the minimum amount. Bulgarelli et al. (2002) also noted that, when the weevil population drops below a certain threshold, it needs at least a month to recover and be identified again in sampling. Thus, while E. kamerunicus is a robust pollinator, its presence may vary. This can partly be attributed to its 75 climatic sensitivities. It is noted, however, that GOPDC has never had any problems with low pollinator numbers. Before discussing the climatic sensitivities of E. kamerunicus, it is first worth noting that, in addition to climate, both oil palm pests and the application of pesticide and weevil parasites can threaten these oil palm pollinators. In terms of pests, the main predators of the E. kamerunicus are rats, which have been observed to eat up to 80% of the weevil larvae in Southeast Asia (Corley & Tinker, 2003), though with no clear impact on fruit set. The use of chemicals to treat leaf miner infestation can also result in pollinator death and indeed reduce bunch yields (Asamoah & Appiah, 1998) and parasite infection of the weevil can affect their reproductive rate and so also oil palm productivity (Corley & Tinker, 2003). Climate sensitivities of pollinators and possible impacts of climate change The oil palm insect pollinators E. subvittatis and E. kamerunicus, both native to Africa, have been observed to reduce in number under high rainfall conditions. Gentry et al. (In Moura et al., 2008) found E. subvittatis to reduce in numbers in high precipitation regions of Columbia (where there is 4,000mm annual rainfall), and Mariau & Gentry (In Moura et al., 2008) found E. subvittatis to decline in Pará, Brazil, during the rainy season (perhaps because of fungus on the male flower after anthesis). Sugih et al. (In Corley & Tinker, 2003) also found the population of E. kamerunicus weevils declined in Indonesia during periods of heavy rainfall. Interestingly, the population of E. kamerunicus weevils was also observed to increase in Sul da Bahia, Brazil, as average monthly rainfall increased from 93.7mm to 160.8mm from the period August 2004– February 2005 to June 2005–March 2006. The total number of weevils observed increased from 91,343 to 262,855 between these periods (Moura et al., 2008). While this trend is different from those reported above, it does not suggest a contradiction. This is because the minimum and maximum rainfall thresholds are not known. With regard to E. kamerunicus, this climate sensitivity information does however suggest that monthly rainfall figures below 160mm are not ideal and that annual rainfall above 4,000mm is too high. However, as E. kamerunicus was specifically selected for introduction to Malaysia because “its numbers were less reduced than those of other species during the wet season in Cameroon” (Corley & Tinker, 2003, p126), this species may in fact tolerate higher rainfall thresholds than E. subvittatis. With regard to temperature, little is known about pollinator sensitivity. Moura et al. (2008), in their study of E. subvittatis and E. kamerunicus in Brazil, did however find E. subvittatis to be more active in hot months (withiQWKHPRQWKO\UDQJHRIÛ&WRÛ& DQGE. kamerunicus to have no relationship with monthly temperature variations (r= -0.15, p<0.05) (Moura et al., 2008, p293–4). When considering the impact of future climate change on pollinators at GOPDC, it is thus useful to consider current and projected changes in rainfall. First, average monthly rainfall at Kwae between 1988 and 2007 has ranged from 13mm to 178mm (see Chapter 2), indicating that rainfall is often not ideal for E. kamerunicus (Moura et al., 2008), though, as noted above, GOPDC has never experienced any problems due to low pollinator numbers. Secondly, future rainfall projections for Ghana are wide ranging. As stated in Chapter 1 (Climate), approximately half of the climate models project increases and the other half project decreases. The maximum and minimum model outputs show reductions of up to 19mm/month in Jul/Aug/Sept and increases of up to 38mm/month in Apr/May/June. While we cannot estimate the impact of increased monthly rainfall on E. kamerunicus due to a lack of information about the upper threshold, we can suggest that any rainfall reductions may lead to a reduction in the number of E. kamerunicus present, with potential consequent impacts on oil palm yield. 76 Valuing ecosystem services Potential costs and revenues not earned if natural pollination services are lost While we cannot predict whether or by how much the pollinator E. kamerunicus will increase or decrease in occurrence at GOPDC plantations due to climate change, here we suggest how financial losses could be calculated, in the case that E. kamerunicus does go into decline. The calculations presented here address the extreme scenario that there is a total loss of the naturally occurring E. kamerunicus (with no option of replacing it with another natural pollinator). Financial impacts (or the value of the ecosystem service) are determined by: 1. Calculating any reduction in yield (and thus revenues not earned) associated with the replacement of natural by assisted pollination. 2. Calculating the cost of assisted pollination (as a replacement). Reductions in yields associated with replacement of natural by assisted pollination Harun and Noor (2002) found that the introduction of E. kamerunicus in Malaysia increased fruit set by 20% (from 50% to 70%), compared with assisted hand pollination. Basri (In Corley & Tinker, 2003) also concluded that introductions of E. kamerunicus in Malaysia increased the fruit set by 19% from 52% to 71%. Assuming that moving from natural pollination by E. kamerunicus to assisted pollination would reduce fruit set by 20%, we are able to estimate financial impacts for GOPDC. With the knowledge that fruit set and bunch weight are significantly correlated (Corley & Tinker, 2003), we can assume bunch weight and hence revenues reduce by 20% after the switch from natural to assisted pollination. Here we have estimated this level of loss for the year 2010, as an example of the potential impacts of loss of natural pollinators (Table 18). As with the potential impact of pests (Chapter 3), the loss of pollinators could lead to revenues not earned exceeding $1 million. If all FFBs were converted to CPO, we estimate a loss of $1.2m; if all FFBs were converted to olein and stearin, we estimate a loss of $1.4m. Note: Clearly it seems unlikely that natural pollinators would disappear completely (i.e. the scenario of 100% assisted pollination is an extreme case). Table 18: Projected loss in yield and revenue for 2010 under a scenario where GOPDC switched from natural to assisted pollination and lost 20% of fruit set Revenues not earned GOPDC average sale ($×000), resulting Oil palm product Tons of product lost price ($/ton) for 2010 from FFB losses CPO* 2,150 580 1,247 Olein** 1,370 755 1,034 Stearin*** 587 655 384 Note: The GOPDC financial model (provided by IFC) projects an FFB yield of 51,195 tons for 2010 from Kwae and Okumaning, and thus a 20% loss would equate to 10,239 tons of FFB. * CPO (tons) equates to 21% of FFB (tons) **Olein (tons) equates to 70% of RBDPO (i.e. 91% of CPO). ***Stearin (tons) equates to 30% of RBDPO (i.e. 91% of CPO). 77 Calculating costs of assisted pollination To calculate costs of assisted pollination for GOPDC would require data on costs of materials and labor, which we do not hold. However, the financial benefits of natural pollination for oil palm in Malaysia were demonstrated by Leslie Davidson, a former deputy chairman of Unilever Plantations. He introduced E. 6 kamerunicus to plantations in Pamol Sabah, Malaysia. According to an internet blog , prior to its introduction, Pamol Sabah employed over 500 workers for manual assisted pollination and within two years, the work force decreased from 2,134 to 1,557. At the same time the plantation reported an increase of 29.4% in production of palm oil, and 42.9% in palm kernels. 4.2 Biodiversity plots GOPDC has planted and/or maintained Biodiversity Plots (BDPs) at both Okumaning and Kwae (ranging 7 from approximately 2 to 18ha in size ) in line with the company’s commitment to balance maximum oil palm production with conservation of biodiversity (GOPDC, 2008). The plots are positioned in upland sections of the plantation as well as along the banks of streams, where they are designed to provide a 30m wide “buffer zone” (where no clearing or erection of any structure is permitted). The plots, which in Okumaning are about 1.5km away from each other, are “remnant fragments of the forest that once existed in the area...part of the South-East subtype of Moist Semi-deciduous (MSSE) forest type” (Ghana Wildlife Society, 2007). As well as containing characteristic species of this forest type, they also include, in higher abundance, species typical of Dry Semi-deciduous forest (Ghana Wildlife Society, 2007). GOPDC’s BDPs are designed to be refuges for wildlife as well as to serve as filters for surface runoff from the plantation areas, thus reducing stream bank erosion and pollution of the waterways (GOPDC, 2008). It is possible, however, that they provide other benefits. Here we consider a range of ecosystem services provided by forests/forest patches which may positively or negatively affect oil palm yield. Attempts are made, where possible, to draw comparisons between forest services and the ecosystem service potential of biodiversity plots. Finally, we consider how the BDPs may help to build resilience to climate change. Ecosystem services with positive impacts on oil palm yield Climate regulation An important and well known ecosystem service provided by forests, which is relevant to the production of oil palm, is climate regulation. It is commonly thought that forests generate rainfall. According to Bonan (2008), climate model simulations show tropical forests “maintain high rates of evapotranspiration, decrease surface air temperature, and increase precipitation compared with pastureland.” In the case of the Amazon basin, research conducted by Oxford’s Global Canopy Program has shown that between 25% and 50% of rainfall is recycled by the forest (Trivedi et al., 2008). Supporting this, studies by Werth and Avissar (2002) and da Silva et al. (2008), which both simulated deforestation of the Amazon, found that forest losses led to reduced precipitation in the Amazon as well as remotely. Similarly, it is thought that the high levels of deforestation that Ghana and several other West African countries have experienced in the last century have led to declining rainfall levels (Appiah et al., 2007; Butler, 2006). While it is therefore tempting to suggest that the forested BDPs at GOPDC’s plantations may have a positive impact on rainfall levels locally (and thus might be a helpful mechanism for improving resilience to drought), we cannot make this assertion. Rainfall recycling by forests has only been identified as deriving from regional scale forests, not from the much smaller areas typical of the BDPs (Trivedi et al., 2008). Anecdotally, however, Xavier Bonneau of CIRAD (Pers. Comm.) has reported that rain gauges near forest 6 http://semalu.blogspot.com/2008/11/datuk-leslie-davidson-planter.html 7 As of August 2008 (GOPDC, 2008). 78 on the edge of plantations record higher rainfall. It is also apparent that GOPDC’s BDPs provide a local temperature regulation effect; temperatures within the BDPs are noticeably lower than in surrounding oil palm habitat (Richenda Connell, Pers. Comm.). Due to the importance of forest climate regulation services, it would be useful for GOPDC to follow research on this topic to see if any local-scale connections can be made between forest presence and rainfall. Carbon storage Tropical forests are also known to provide the important regulating service of carbon sequestration because they are either carbon neutral or carbon sinks (Bonan, 2008). Carbon sequestration is an important way of mitigating climate change as it ensures less carbon dioxide enters the atmosphere. Small, local-scale forests are part of the effort to mitigate climate change (as is shown through the generation of forestry offset projects), though it should be noted that there is no direct link between forest patches at GOPDC, carbon sequestered and local climate conditions. Despite this, GOPDC’s BDPs, if managed appropriately now and in the future, are a potential direct source of income. By demonstrating “additional” forest protection and carbon sequestration through these BDPs, GOPDC could either apply for carbon credits for sale on the regulated carbon market, or generate credits to sell on the voluntary market (on which significantly more forestry projects are currently accepted). Interestingly, this represents an indirect (policy-driven) climate change impact on GOPDC’s financial performance. Pest control In an oil palm plantation context, the presence of natural forest is associated with the presence of high bird and butterfly diversity (Koh, 2008a). Koh (2008b) found insectivorous birds play an important role in protecting oil palm from insect pests (i.e. performing a regulating ecosystem service). Maintaining forests may therefore have a positive impact on oil palm yield. In fact, Koh (2008b), who conducted a bird- enclosure experiment on oil palm seedlings in East Sabah, Malaysia, found that removing bird predators from oil palm plantations could lead to ~28% foliar damage over the effective life span of an oil palm leaf, and in turn lead to fruit yield losses. It is not clear how pests or their predators will respond to climate change. However, as pest control is essential under present climatic conditions, the preservation of predators is important anyway. This is supported by anecdotal evidence from GOPDC: outgrowers who grow oil palm between forest patches experience fewer leaf miner attacks (Emmanuel Wiafe, GOPDC, Pers. Comm.), and fewer rodents have been observed in plantations (perhaps as result of an increased number of birds of prey related to BDPs) (Bonneau, X., Pers. Comm.). As BDPs may prove useful under future climate change (if pest abundance increases; see Chapter 3), it may be sensible for GOPDC to monitor progress in understanding the relationships between pests, predators and forests. Currently the Oil Palm Research Institute (OPRI) is researching which bird and insect species control the leaf miner, while the academic community internationally continues to investigate how forest patches in agricultural plantations can reduce pest numbers. Reducing soil erosion Tropical forests reduce surface erosion as they protect the soil from rain (Bruijnzeel, 2004), and thus perform a “soil conservation” function (Trivedi et al., 2008, p31). In the case of oil palm, erosion is a serious problem as it leads to the loss of important nutrients for growth (Corley & Tinker, 2003). Additionally, erosion under oil palm leads to the removal of soil between the tertiary and quaternary feeding roots, which can cause them to dry up and die, in turn leading to a reduction in water and nutrient uptake capacity in the root system (Hartemink, 2003). 79 While a mature oil palm canopy can protect against erosion itself (Corley & Tinker, 2003), BDPs may add extra value. They have more layers of vegetation, which intercept rainfall more successfully than oil palms, thus reducing the erosive effects of rainfall on the slopes where they are situated. The planting of cover crops at both GOPDC plantations and the practice of contour planting at Okumaning also help to reduce soil erosion and gulley formation. One of the more certain impacts of climate change is that it will lead to increased rainfall intensity, so it would appear that the BDPs can help to build resilience against this particular climate change risk. Wind-breaks Wind-breaks are a barrier of trees and shrubs that help slow down wind speeds. Trees have been used as wind-breaks in both the semi-arid tropics and temperate regions to protect crops and soil against wind damage, with some studies finding increased crop yields behind wind-breaks (Vandenbeldt, 1990). As the GOPDC BDPs are relatively small patches, not in straight lines or covering the whole length of the plantation, their potential to be effective wind-breaks is likely to be low. However, while oil palms can be damaged by high winds, in general winds at GOPDC’s plantations are not strong (Emmanuel Wiafe, GOPDC, Pers. Comm.). Furthermore, there is little evidence to suggest that climate change will lead to increases in wind speeds in the future. As well as regulating wind speed, wind-breaks are also known to regulate relative humidity. In sheltered areas, relative humidity is typically 2 to 4% higher than in open areas, depending on the density of the windbreak (Brandle, 2009). As higher humidity is related to reduced rates of evapotranspiration, which in turn is related to increased soil moisture (assuming precipitation and runoff are constant), we suggest that wind-breaks may help to preserve soil moisture. As sufficient soil moisture is important for oil palm development, wind-breaks may therefore be a valuable climate change adaptation option. As GOPDC’s BDPs are relatively small and do not reach across the plantation, this contribution is currently likely to be minimal. Ecosystem services with negative impacts on oil palm yield It is worth noting that forests and biodiversity patches do not always provide beneficial services to people (Trivedi et al., 2008). At GOPDC, it seems that BDPs may provide a habitat for certain oil palm pests such as the grass cutter and weaver bird (Emmanuel Wiafe, GOPDC, Pers. Comm.). Thus, while BDPs may provide habitats for pest predators, they may also provide them for pests. Additionally, while there are various ecosystem services performed by forest patches which may positively influence yield (as outlined in the previous section), the BDPs occupy upland areas which could otherwise be used for growing oil palm, though upland soils are less productive for oil palm (Emmanuel Wiafe, GOPDC, Pers. Comm.). There is currently a lively discussion among the academic community about the benefits of biodiversity patches in plantations (Bhagwat & Willis, 2009). 4.3 Adaptation options and conclusions Ecosystem services are important to GOPDC. The natural pollinator E. kamerunicus is enormously valuable and if it were severely affected, the financial consequences for GOPDC would be significant. However, in relation to understanding of climate change impacts on plant-pollinator interactions, it is acknowledged in the scientific literature that “we know much less about potential biological effects of climate warming on ecological interactions in the tropics than in temperate areas” (Hegland et al., 2009). Uncertainties and lack of knowledge make it difficult to evaluate how the species might be affected by climate change and further research would be useful. There may be potential for OPRI or CIRAD to 80 undertake research in this area, for instance, by investigating observed relationships between densities of the oil palm pollinator and climatic conditions. Most of the literature on biodiversity and oil palm is focused on nature conservation objectives and highlights the negative impacts of plantations on biodiversity. GOPDC’s commitment to developing and maintaining BDPs is driven by the company’s objective to conserve biodiversity. However, it seems that the BDPs also provide positive ecosystem services which help to build GOPDC’s climate resilience, particularly in relation to pest control and reduction in soil erosion. Clearly, however, further work could be done to improve understanding of the benefits of BDPs in oil palm plantations, and there is an active research community investigating these linkages, which GOPDC could engage with. For instance, GOPDC has confirmed that it would be useful to link up to the research being undertaken by the Oil Palm Research Institute on the bird and insect species that control the leaf miner. Similarly, researchers at the University of Oxford are investigating the benefits of forest patches in oil palm plantations. If it is not already doing so, GOPDC could ensure that it is fully exploiting the data it collects at the plantations to better understand the correlations between climatic conditions and ecosystem services. For instance, GOPDC considers that it could install some additional rain gauges, in an experiment to compare rainfall amounts near BDPs with those in the middle of oil palm plantations, further from BDPs, to ascertain whether the BDPs have any influence on local rainfall. Furthermore, if GOPDC installed a full weather station at its Okumaning plantation (which has forest nearby), it could ascertain whether relative humidity there is higher than at Kwae. Finally, GOPDC can discuss with Ghana Environmental Protection Agency (EPA) the possibilities of gaining carbon credits for its BDPs. They may be eligible, depending on their size. Chapter 4 References Appiah, M., Blay, D., Damnyag, L., Dwomoh, F. K., Pappinen, A., and Luukkanen, O. (2007). Dependence on forest resources and tropical deforestation in Ghana. Environment, Development and Sustainability, 11 (3): 471–487. Asamoah, T. E. O., Appiah, S. O. (1998). Agro-management strategies towards mitigating the effects of damage caused by the leaf miner, coelaenomenodera minuta uhmann (Coleoptera: Chrysomelidae: Hispinae) in Ghana. Oil Palm Research Institute, C. S. I. R, Kade. Bonan, G. B. (2008). Forests and Climate Change: Forcings, Feedbacks, and the Climate Benefits of Forests. Science, 320: 1444–1449. Brandle, R. (2009). ‘How Windbreaks Work’ Retrieved 22 May 2009, from http://plasticulture.cas.psu.edu/WindBreaks.html Bruijnzeel, L. A. (2004). Hydrological functions of tropical forests: not seeing the soil for the trees? Agriculture, Ecosystems & Environment, 104 (1): 185–228. Bulgarelli, J., Chincillia, C., Rodríguez, R. (2002). Male inflorescences, population of Elaeidobious kamerunicus and pollination in a young commercial oil palm plantation in a dry area of Costa Rica. ASD Oil Palm Papers, No 24: 32–37. Butler, R. A. (2006). ‘Local and National Consequences: Loss of Local Climate Regulation’ Retrieved 15 May 2009, from Mongabay.com / A Place Out of Time: Tropical Rainforests and the Perils They Face. Web site: http://rainforests.mongabay.com/0902.htm CABI. (2003). The oil palm mystery. Far Eastern Agriculture, January/February 2003. 81 Chinchilla-López, C. M., Richardson, D. L. (1991) Pollinating insects and the pollination of oil palms in Central America. ASD Technical Bulletin, no. 2, p1–18. Corley, R. H. V., Tinker, P. B. (2003). The Oil Palm (4th ed.) Wiley, Hoboken, NJ, USA. Da Silva, R. R., Werth, D., Avissar, R. (2008). Regional Impacts of Future Land-Cover Changes on the Amazon Basin Wet-Season Climate. Journal of Climate, 21: 1153–1170. Ghana Wildlife Society. (2007). Ecological Monitoring of the Biodiversity Plots of the Okumaning Oil Palm Estate of the Ghana Oil Palm Development Company. Ghana Wildlife Society GOPDC. (2008). Environmental Management Plan. Hartemink, A. E. (2003). Soil fertility decline in the Tropics: with case studies on plantations. CABI, UK. Harun, M. H., Noor, M. R. (2002). Fruit Set and Oil Palm Bunch Components. Journal of Oil Palm Research,14 (2): 24–33. Hegland, S.J., Nielsen, A., Lázaro, A., Bjerknes, A.-L.1 and Totland, Ø. (2009). How does climate warming affect plant-pollinator interactions? Ecology Letters, 12: 184–195. Koh, L. P. (2008a) Can Oil Palm Plantations be made more Hospitable for Forest Butterflies and Birds? Journal of Applied Ecology, 45: 1002–1009. Koh, L. P. (2008b). Birds Defend Oil Palm from Herbivorous Insects. Ecological Applications, 18: 821– 825. Millennium Ecosystem Assessment. (2005). Ecosystems and Human Well-being: Synthesis. Island Press, Washington, DC. Moura, J. I. L., Cividanes, F. J., dos Santos Filho, L. P., Valle, R. R. (2008). Polinização do dendezeiro por besouros no Sul da Bahia. Pesq. Agropec. Bras., Brasília, 43 (3): 289–294. Trivedi, M., Pagageorgiou, S., Moran, D. (2008). What are Rainforests Worth? And why it makes economic sense to keep them standing. Forest Foresight Report 4. Global Canopy Programme, Oxford. Vandenbeldt, R. J. (1990). Agroforestry in the semiarid tropics. In: K.G. MacDicken and N.T. Vergara (eds). Agroforestry: classification and management. New York, John Wiley & Sons. Werth, D., Avissar, R. (2002). The local and global effects of Amazon deforestation. Journal of Geophysical Research, 107 (D20): 55-1–55-8. 82 Chapter 5: Climate risk analysis for GOPDC industrial operations 83 Overview This chapter explores the impacts of climate change on the industrial operations at GOPDC, including the: x electrical equipment, x mill (including worker safety and productivity), x cooling water system, x refinery, x fractionation plant, and x power plant. The depth of analysis for each of these issues is variable, depending on the availability of data and information. Climate change impacts on the water supply and wastewater treatment systems are addressed separately in Chapter 6, though this chapter reviews the financial impacts of a lack of water at the refinery, as experienced in early 2010. 5.1 Electrical equipment Sensitivity of electrical systems to temperature In general, the operation and efficiency of electrical equipment is negatively affected by higher temperatures. For instance, electrical breakers may start to trip at lower amps than their design ratings, and larger breakers may be needed to counteract this effect. The efficiency of each electrical component is generally reduced as temperatures rise (e.g. cables, breakers, transformers etc.) and the effect on each component in the chain is cumulative across the system as a whole. The impacts may be significant; for ° instance, a 5 C increase in ambient temperature can lead to a reduction in cable performance of 5% or more. Recommended adaptation actions A detailed research study would be needed to establish the overall impact on efficiency of electrical systems at GOPDC associated with higher temperatures due to climate change, and to identify whether any electrical equipment needs upgrading. Some potential solutions that may be applicable include: larger cables, aerial cabling, wider cable trays, extra cable trays and air conditioning for breaker panels. 5.2 Mill Sensitivity of mill operations to temperature The milling process itself is not sensitive to the small temperature increases associated with climate change, as the major unit operations are mechanical in nature. However, due to the dependence of the mill on groundwater, it is vulnerable to impacts on groundwater availability (see Chapter 6). Steam is widely used in the mill to distribute heat. The steam condenses more during the night, when it is cooler, so more steam is needed to counteract this effect. Higher temperatures due to climate change will lead to less steam being needed. 84 Staff performance decreases with increases in temperature and relative humidity. Staff are less alert, less productive, more likely to have accidents, more easily fatigued and more dehydrated. This issue is most prominent in the mill (though it also affects the plantation workers), where hard physical labor is involved, for instance in moving cages. With higher temperatures due to climate change, workers may need to take more, or longer, breaks. More workers are already needed during the day shift at the mill than at night, due to higher daytime temperatures: there are five workers in the day and four at night. Similarly, at present, three workers are needed to operate the crane during the day, but only two at night. Recommended adaptation actions As ambient temperatures increase, it may be necessary to consider increasing the number of night workers. It may also be necessary to consider building an enclosure for workers close to the turbine (as are found elsewhere) to shield them from high temperatures and/or to try to reduce the heat being emitted by the turbine. 5.3 Cooling water system Users of cooling water The GODPC industrial operations use cooling water extensively. Three uses have been identified as sensitive to cooling water temperature, and these are evaluated further in Sections 5.4–5.6 below: 1. Ejector system condensers at the refinery: The refining process is sensitive to small temperature increases, primarily because of the effects of higher temperatures on the cooling water system. Increases in cooling water temperature reduce the efficiency of condensation and hence the effectiveness of vacuum producing systems, resulting in lower CPO throughput. 2. Chiller (refrigeration) system condensers for crystallization of stearin in the fractionation plant: All mechanical refrigeration systems work on the principle of evaporating the refrigerant at low pressure (low boiling temperature) and condensing at high pressure (high temperature). If the available condensing heat removal media (cooling water) temperature increases, the compressor has to deliver at a higher pressure. The crystallization stage in the refinery, which produces solid crystals of stearin, uses chilled water provided by a conventional refrigeration system. Cooling water is required to condense the refrigerant after compression and increases in cooling water temperature lead to higher condensation pressures and increased compression loads (i.e. the compressor needs to run longer, and therefore consumes more electrical power). In addition, the crystallization cycle time will increase, so fewer batches will be processed. 3. Condensing steam turbine condensers in the power plant: Electrical power is generated by expanding high pressure steam through a turbine connected to a generator. The low pressure exhaust steam must be condensed to recover condensate for recycling to the boiler. If the available condensing heat removal media (cooling water) temperature increases, the turbine exhaust condenser will operate at a higher pressure, and will reduce turbine efficiency. More steam will therefore need to be consumed to maintain constant power output. However as long as the power plant can supply adequate power, the milling and refinery operations are not impacted. Sensitivity of the cooling water system to climatic conditions The GODPC cooling water system is a classical circulating closed loop system, using a cooling tower in which hot cooling water flows down a packed tower against an ambient air flow provided by a fan system. 85 The theoretical achievable cooling water temperature exiting the cooling tower is limited by the ambient air wet bulb temperature. The design of the cooling tower will specify the difference in temperature between the wet bulb temperature and the cooled water exit temperature from the cooling tower. This difference is the driving force in the heat transfer and will determine the size of the tower and the required air flow. ° Practically, it is uneconomic to achieve a difference of less than 5 C. Table 19 gives typical ambient air conditions at GODPC, based on climate data measured in 2007 for the hottest and coldest months of the year. Table 19: Typical ambient air conditions at GOPDC in 2007 ° ° Dry bulb temperature ( C) Relative Humidity (%) Wet bulb temperature ( C) Month Day Night Day Night Day Night 35 50 26 February 24 90 23 29 70 25 August 22 90 21 As can be seen from Table 19, the wet bulb temperature has a typical diurnal (day to night) variation of 3 ° ° to 4 C and a seasonal variation of 1 to 2 C. The impact of increased air temperature and increased wet bulb temperature arising from climate change will be to reduce the capacity of the cooling tower to reject heat to the atmosphere. In the absence of cooler and condenser data sheets listing the current temperature difference between hot and cold streams, calculated as the log mean temperature difference (LMTD), it is difficult to predict the impact that a small inlet water temperature rise will have. For those exchangers with very small LMTD, the heat that a ° cooling tower can reject drops significantly with higher temperature/higher humidity. The effect of a 2 C temperature increase, for instance, could be large (i.e. up to 15–20% reduction in cooling capacity) whereas for other exchangers the effect would be negligible. In an extreme example, where the temperature rise between inlet and outlet cooling water temperatures is ° 6 C, a second cooling tower would need to be built, and cooling water flows would increase by 100%. The additional cooling tower would remove the heat, but the cooling water system would not be able to function correctly, as flow rates would increase by 100% and system pressure would drop by 400%. The vacuum system in the refinery, the crystallizers in the fractionation plant and the turbine for the power ° plant were all designed for a maximum cooling water temperature of 32 C. As an approximation, for the ° analyses described in Sections 5.4–5.6 it has been assumed that a 1 C increase in air temperature due to climate change would result in the same increase in cooling water temperature. Recommended adaptation actions GOPDC has recently installed an additional cooling tower and does not have any plans to install another. However, if any upgrades to the cooling system are undertaken, GOPDC has confirmed that they will take account of rising temperatures due to climate change. 86 5.4 Refinery Vacuum generating systems The GOPDC refinery processes CPO into refined, bleached and deodorized oil (RBDO) which is further processed in the fractionation plant into olein and stearin products (see Section 5.5). The refinery is equipped with a number of vacuum generating systems. Vacuum conditions are required in the drying tank, the bleacher and the deodorizer. The deodorizing process is a vacuum stripping process at elevated temperatures in which free fatty acids and odiferous compounds are removed to obtain an odorless on-specification oil. This process operates at very high vacuum conditions, as indicated by the test data in Figures Figure 33 and Figure 34. The configuration of the ejector systems is not known, but the process description states that it is a combination of ejectors and booster compressors, utilizing barometric condensers (AY & A Consult, 2005). The design of the combined ejectors and compressors will be based on achieving the design vacuum at a specified mass and volumetric flow rate. The function of the condensers is to reduce the mass and volume flow by cooling and condensing water vapor, and other condensable vapor, between stages. The cooling water temperature is a critical factor as it controls the partial pressure of water vapor and other components and therefore the ability to condense. The standard data in Table 20 indicate saturated steam conditions for a range of cooling water temperatures. Table 20: Relationship between cooling water temperatures and saturated steam pressure ° Temperature ( C) Pressure (bar) 20 0.0234 22 0.0264 24 0.0298 26 0.0336 28 0.0378 30 0.0425 32 0.0476 34 0.0532 ° Table 20 demonstrates the impact of higher cooling water temperatures, as, for instance, a rise of 2 C will increase the interstage mass flow rates by more than 10%. If the ejector or compressor cannot compress the additional flow, the system pressure will rise to reduce volume. As a result, it would be expected that ° the typical diurnal (day to night) variations in wet bulb temperature at Kwae (of 3 to 4 C, see Table 19), would cause variations in the vacuum system interstage mass flow rates of up to 20% when the design cooling water temperature for the vacuum systems is exceeded. 8 This is borne out by GOPDC’s experience: the company advises that, when cooling water temperatures ° exceed 32 C, the vacuum strength reduces and it is necessary to add fresh water to the basin coming 8 Gopi Kumar, GOPDC Refinery Production Manager, Pers. Comm. 87 from the water treatment plant, and reduce the flow of product. Water consumption therefore increases and the flow of product decreases. GOPDC reports that under these conditions CPO processing rates can vary from 4 tons/hour to 3.2 tons/hour between night and day, which indicates a de-rating factor of about 20%. This effect is demonstrated by the data on ambient air temperature vs. process vacuum, presented in Figures Figure 33 and Figure 34 below, recorded during December 2008 by Gopi Kumar, GOPDC Refinery Production Manager. Relationship between observed atmospheric temperature and vacuum strength This section presents the results of the following analyses: 1. First, the relationship between atmospheric temperatures and vacuum strength at the plant is calculated, using data on temperature and vacuum strength recorded at the refinery during December 2008. 2. Using projections of increases in atmospheric temperature from climate models, potential annual impacts on vacuum strength, and hence throughput of CPO and production of olein and stearin are calculated for the period 2010 to 2030. 3. Assuming that the reduced throughput of CPO at the refinery would mean that more CPO is sold and less olein and stearin, the consequent impacts in financial performance are calculated for the period 2010 to 2030. ° Hourly atmospheric temperature ( C) and vacuum strength (mm Hg) data were recorded for five consecutive days, from December 12 to 16, 2008. Figure 33 illustrates hourly co-variations in atmospheric temperature and vacuum strength over this period. Significant degrees of correspondence between temperature and vacuum data can be clearly seen. Figure 33: Co-variations in atmospheric temperature and vacuum strength at GOPDC refinery observed from December 12 to 16, 2008 88 9 Examination of these data (Figure 34) indicates that: x ° For air temperatures below 32 C, the vacuum strength is generally constant at about 5mm of mercury (pink squares). x ° For temperatures of 32 C and above, vacuum strength reduces (i.e. the vacuum reading in mm Hg increases) as temperatures increase (blue diamonds). At these temperatures, the relationship between atmospheric temperature and vacuum strength (Figure 34, black line) has the following equation: ° Vacuum strength (mm Hg) = 1.0 x Temperature ( C) – 25.7 (Equation 1) Statistical tests on these data are provided in Annex D: Statistical tests, and indicate that the relationship is statistically significant. Figure 34: Correlation between atmospheric temperature and vacuum strength, December 12 to 16, ° 2008, for temperatures above 32 C As noted in Chapter 1, climate change is projected with a high degree of confidence to cause temperature ° increases. This will lead to cooling water temperatures in excess of 32 C occurring more frequently, and a consequent reduction in the strength of the vacuum that can be achieved, causing a reduction in olein and stearin output. 9 Note that there are 65 data points in Figure 34 but many are identical, which is why only 20 points can be seen in the figure. 89 GOPDC has recorded data on cooling water temperature at the refinery. This demonstrates that, at ° present (in 2009), cooling water temperatures exceed the 32 C design threshold for just 2% of the time. The median climate change projection indicates that by 2020, air temperatures are expected to increase ° by a further 0.5 C. This is calculated to lead to a five-fold increase in the incidence of temperatures in ° excess of 32 C, i.e. 11% of the time (using the approximation that cooling water temperatures rise in line with air temperatures). It is also interesting to note at this point the criticality of the maximum design temperature for refinery ° ° output. Had the design temperature been set just 1 C lower, at 31 C, it would be exceeded 20% of the time today, and 37% of the time by 2020. Projected impacts of rising temperatures on olein and stearin production in the refinery Assumptions regarding plant future capacity As per the GOPDC financial model provided by IFC, the annual refinery/fractionation plant output (metric tons) from 2007 to 2017 shows an increase year-on-year. The current maximum annual capacity of the plant is 27,000 metric tons. With an injection of capital in 2011 (as per the industrial investment program outlined in the GOPDC financial model), the annual installed capacity at the plant is intended to increase, reaching 38,696 metric tons in 2017 (Figure 35). In the absence of information on changes in plant capacity beyond 2017, it has been assumed that capacity remains constant at 38,696 metric tons from 2018 to 2030. Figure 35: Assumed annual output of olein and stearin (metric tons) Using Equation 1 above, and taking account of the percentage of the time now and in the future when cooling water temperatures exceed the 32°C design threshold, we have calculated the effects of rising temperatures on vacuum strength, and hence annual production of olein and stearin. This analysis uses the median projection of annual average temperature increases for the period 2010 to 2030 (McSweeney et al., 2008; see Chapter 1, Climate). It is worth recalling at this point that the median climate model projection gives an increase in annual average temperatures of 1.2°C by the 2030s compared to the 90 1970–1999 baseline period, whereas the low-to-high climate model range is 0.8 to 1.5°C by the 2030s. Hence, the projected reduction in olein and stearin production, taking account of climate change, could be somewhat lower or higher than presented here. The results of the analysis are presented in Tables Table 21 and Table 22. The analysis uses the following assumptions: 1. CPO not processed into olein and stearin is sold to market instead, at a price of US$580 per metric ton 2. Sale price of US$755 per metric ton for olein 10 3. Sale price of US$655 per metric ton for stearin 4. Discount rates in the range 4% to 16% are examined, with 12% being the discount rate that GOPDC considers most appropriate. Note: The climate change data incorporate some variability, and therefore do not show smooth 11 temperature increases year-on-year . Table 21: Annual output of olein and stearin as per GOPDC financial model; projected future temperature changes at GOPDC; and projected resultant reduction in vacuum strength, CPO throughput and olein and stearin output Projected Annual output of median olein and stearin as annual Projected average % of time Projected per GOPDC reduction in temperature when Projected annual financial model annual output increase cooling reduction amount of (assumed constant of olein and relative to water in CPO not from 2017 to 2030) stearin (metric (metric tons) 1970–1999 temperature vacuum processed tons) baseline exceeds strength (metric ° ° Year Olein Stearin ( C) 32 C (%) (%) tons) Olein Stearin 2010 18,857 8,081 0.65 5.4% 2% 30 19 8 2011 18,699 8,014 0.70 6.3% 3% 50 32 14 2012 19,397 8,313 0.70 6.3% 3% 52 33 14 2013 20,960 8,983 0.57 4.0% 1% 7 4 2 2014 22,816 9,778 0.69 6.1% 3% 56 35 15 2015 24,582 10,535 0.75 7.2% 4% 98 62 27 2016 26,093 11,183 0.74 7.0% 3% 96 61 26 2017 27,087 11,609 0.72 6.7% 3% 86 55 23 10 The sale prices for CPO, olein and stearin are the “GOPDC sale prices” shown in the spreadsheet financial model provided by IFC. The spreadsheet also gives “IFC sale prices”, which are somewhat higher. 11 It should be noted that the projected annual average temperature increases shown in Table 19 are unlikely to be correct for a given year, though the average trend over the period 2010 to 2030 is projected with high confidence, and is the median across a number of climate change models. 91 Projected Annual output of median annual olein and stearin as Projected average % of time Projected per GOPDC reduction in temperature when Projected annual financial model annual output increase cooling reduction amount of (assumed constant of olein and from 2017 to 2030) relative to water in CPO not stearin (metric (metric tons) 1970–1999 temperature vacuum processed tons) baseline exceeds strength (metric ° ° Year Olein Stearin ( C) 32 C (%) (%) tons) Olein Stearin 2018 27,087 11,609 0.80 8.1% 4% 149 95 41 2019 27,087 11,609 0.71 6.5% 3% 79 50 22 2020 27,087 11,609 0.96 11.0% 7% 317 202 87 2021 27,087 11,609 0.94 10.6% 6% 293 187 80 2022 27,087 11,609 0.86 9.2% 5% 205 131 56 2023 27,087 11,609 1.01 11.9% 8% 381 242 104 2024 27,087 11,609 1.00 11.7% 7% 367 234 100 2025 27,087 11,609 1.01 11.9% 8% 381 242 104 2026 27,087 11,609 0.77 7.6% 4% 123 79 34 2027 27,087 11,609 0.74 7.0% 3% 100 64 27 2028 27,087 11,609 1.09 13.3% 9% 493 314 135 2029 27,087 11,609 1.11 13.7% 9% 523 333 143 2030 27,087 11,609 1.05 12.6% 8% 435 277 119 Table 21 demonstrates that, in the 2020s, an average of 3 days of production may be lost each year in the refinery due to the effects of rising temperatures. Based on the reductions in olein and stearin output, and assuming the excess unprocessed CPO is sold to market, projections of net impacts on financial performance are shown in Table 22. A variety of discount 12 rates have been used to calculate future impacts . Figure 36 presents the financial impacts graphically, for a discount rate of 12%, as recommended by GOPDC. The annual average net reduction in income over the period 2010–2030 is US$3,600 (12% discount rate). 12 The discount rate of 4% used was based on a recent study by Noormahayu et al. (2009). According to Noormahayu, previous cost-benefit analyses of oil palm cultivation have assumed a long-term interest rate of 10%, which may be a pessimistic reflection of the real situation in recent decades. 92 Table 22: Projected reduction in income (US$) due to impacts of rising temperatures on olein and stearin throughput in the refinery Increase in Net reduction in income income from Reduction in selling un- income from processed olein and CPO (un- stearin (un- Un- Discount Discount Discount Discount Year discounted) discounted) discounted rate 4% rate 8% rate 12% rate 16% 2010 17,380 -19,770 -2,390 -2,298 -2,213 -2,134 -2,060 2011 28,987 -32,972 -3,986 -3,685 -3,417 -3,177 -2,962 2012 30,069 -34,203 -4,134 -3,676 -3,282 -2,943 -2,649 2013 3,922 -4,462 -539 -461 -396 -343 -298 2014 32,268 -36,705 -4,437 -3,647 -3,020 -2,518 -2,112 2015 56,659 -64,449 -7,791 -6,157 -4,909 -3,947 -3,198 2016 55,943 -63,636 -7,692 -5,845 -4,488 -3,480 -2,722 2017 49,761 -56,603 -6,842 -5,000 -3,697 -2,763 -2,087 2018 86,214 -98,069 -11,854 -8,329 -5,930 -4,275 -3,117 2019 45,807 -52,106 -6,298 -4,255 -2,917 -2,028 -1,428 2020 183,796 -209,068 -25,272 -16,416 -10,839 -7,265 -4,939 2021 169,863 -193,220 -23,356 -14,588 -9,275 -5,995 -3,935 2022 119,041 -135,409 -16,368 -9,830 -6,019 -3,751 -2,377 2023 220,725 -251,075 -30,350 -17,526 -10,333 -6,210 -3,800 2024 213,102 -242,403 -29,302 -16,270 -9,237 -5,353 -3,162 2025 220,725 -251,075 -30,350 -16,204 -8,859 -4,951 -2,824 2026 71,550 -81,388 -9,838 -5,051 -2,659 -1,433 -789 2027 58,075 -66,060 -7,985 -3,942 -1,998 -1,038 -552 2028 285,891 -325,201 -39,310 -18,658 -9,109 -4,564 -2,343 2029 303,325 -345,032 -41,707 -19,035 -8,948 -4,324 -2,143 2030 252,386 -287,089 -34,703 -15,229 -6,894 -3,212 -1,537 Total -344,505 -196,101 -118,439 -75,704 -51,033 Annual average -16,405 -9,338 -5,640 -3,605 -2,430 93 Figure 36: Projected net reduction in income due to impacts of rising temperatures on refinery vacuum strength and hence olein and stearin production, at 12% discount rate Water availability at the refinery As already noted, water is critical to the operation of the refinery. GOPDC has stated that from January to May 2010, 3–4 days of production were lost in the refinery, due to a lack of water. Each day without production involves a loss of 70 tons of olein and 30 tons of stearin, though the financial impact is somewhat counteracted by selling CPO instead. The net financial loss for each day of foregone olein and stearin production is US$8,760, so in the first 5 months of 2010, GOPDC lost US$26,000– US$35,000. It is not possible to provide quantified estimates of how climate change may exacerbate risks to water resource availability at the refinery, nor anywhere else on the Estate. However, as we emphasize in Chapter 6, future water resource availability has emerged as a critical area of uncertainty for GOPDC to manage. It is noted that rising temperatures lead to increased water demand at the mill, refinery and power plant. At the same time (as discussed in Chapter 6) climate change and competition from other water users may increase pressures on water supply. Existing and recommended adaptation actions for the refinery GOPDC has recently installed a second cooling tower, for the refinery vacuum system, to reduce water consumption. This was done because the first cooling tower is so far from the refinery that it was proving difficult to manage water levels in the water reservoir/basin. With the addition of the new cooling tower, GOPDC estimates that about 75% of the temperature effect on the vacuum shown above will remain. 94 Upgrading the current vacuum producing equipment would help to alleviate the effects of rising temperatures and hence maintain current fractionation rates. GODPC has stated that when it increases the refinery capacity, it will take account of projected temperature increases, to ensure that the design temperatures chosen are appropriate over the facility lifetime. As highlighted above, small changes in the design temperature threshold can have significant impacts on the percentage of the time when this temperature is exceeded. Adaptation actions to manage risks to water resource availability are discussed in Chapter 6. 5.5 Fractionation plant Relationship between temperature and crystallization time GODPC utilizes a selective crystallizing and filtration process to separate olein and stearin products in the fractionation plant. The crystallizers utilize cooling provided by chilled water from a chiller system. The chilled water is provided by a conventional refrigeration system, so cooling water is required to condense the refrigerant after compression. Higher cooling water temperatures (as described in Section 5.3) will result in higher condensation pressures and increased compression loads. If the compressor design compression ratio is exceeded, the suction pressure will increase and result in warmer chilled water. As noted above, the crystallization system was designed for a maximum cooling water temperature of ° 32 C. Above this temperature, the crystallization cycle time extends. Hence, the number of batches of olein and stearin that can be processed is reduced (1 batch is 24 metric tons of olein and stearin combined). Projected impacts of rising temperatures on the crystallization process ° GOPDC estimates that a temperature increase of 0.5 C (the projected increase from 2009 to 2020 for the median climate change projection) would extend the crystallization time for each batch from 11 hours to 11.5 hours. They have three crystallizers and can currently process five batches per day. Therefore, a ° temperature increase of 0.5 C would result in the loss of 1 batch every 4.4 days, if cooling water ° temperatures were above 32 C for all of that time. Based on the knowledge that the fractionation plant runs for 25 days per month, 12 months per year, and taking account of the percentage of time when cooling water temperatures exceed the design threshold, the estimated impacts of climate change on olein and stearin production in the fractionation plant are summarized in Table 23. On average, in the 2020s, it is estimated that 2 days of production are lost each year in the fractionation plant. Projections of net impacts on financial performance are shown in Table 24. Figure 37 presents the financial impacts graphically, for a discount rate of 12%. The annual average net reduction in income over the period 2010–2030 is US$3,400 (12% discount rate). The other consequence of higher temperatures is an increase in power requirements to deliver chilled water in the fractionation plant. While biomass fuel for the power plant is free, increased power demand leads to higher water consumption in the power plant, with associated costs as discussed in the next section. 95 Table 23: Projected future temperature changes at GOPDC and estimated resultant reduction in CPO throughput and olein and stearin output due to slower crystallization rates Projected median annual Projected average annual Projected reduction in annual temperature % of time when amount of output of olein and stearin increase relative cooling water CPO not (metric tons) to 1970–1999 temperature processed ° ° Year baseline ( C) exceeds 32 C (%) (metric tons) Olein Stearin 2010 0.65 5.4% 68 0 0 2011 0.70 6.3% 78 43 19 2012 0.70 6.3% 81 50 21 2013 0.57 4.0% 55 52 22 2014 0.69 6.1% 93 35 15 2015 0.75 7.2% 118 59 25 2016 0.74 7.0% 122 75 32 2017 0.72 6.7% 120 78 33 2018 0.80 8.1% 146 77 33 2019 0.71 6.5% 117 93 40 2020 0.96 11.0% 198 74 32 2021 0.94 10.6% 191 126 54 2022 0.86 9.2% 165 122 52 2023 1.01 11.9% 214 105 45 2024 1.00 11.7% 211 136 58 2025 1.01 11.9% 214 134 58 2026 0.77 7.6% 136 136 58 2027 0.74 7.0% 127 87 37 2028 1.09 13.3% 240 81 35 2029 1.11 13.7% 246 153 65 2030 1.05 12.6% 227 157 67 96 Table 24: Projected reduction in income (US$) due to impacts of rising temperatures on olein and stearin throughput in the fractionation plant Increase in Reduction Net reduction in income income from in income selling un- from olein processed and stearin CPO (un- (un- Un- Discount Discount Discount Discount Year discounted) discounted) discounted rate 4% rate 8% rate 12% rate 16% 2010 39,352 -44,762 -5,411 -5,203 -5,010 -4,831 -4,665 2011 45,503 -51,760 -6,257 -5,785 -5,364 -4,988 -4,650 2012 47,202 -53,692 -6,490 -5,770 -5,152 -4,620 -4,158 2013 32,120 -36,536 -4,416 -3,775 -3,246 -2,807 -2,439 2014 53,940 -61,357 -7,417 -6,096 -5,048 -4,208 -3,531 2015 68,337 -77,733 -9,396 -7,426 -5,921 -4,760 -3,857 2016 70,729 -80,455 -9,725 -7,390 -5,675 -4,399 -3,441 2017 69,670 -79,249 -9,580 -7,000 -5,176 -3,869 -2,922 2018 84,688 -96,333 -11,645 -8,181 -5,825 -4,199 -3,062 2019 67,792 -77,114 -9,321 -6,297 -4,318 -3,001 -2,113 2020 114,725 -130,500 -15,775 -10,247 -6,766 -4,535 -3,083 2021 110,971 -126,229 -15,258 -9,530 -6,059 -3,916 -2,570 2022 95,952 -109,145 -13,193 -7,924 -4,851 -3,024 -1,916 2023 124,112 -141,177 -17,065 -9,855 -5,810 -3,492 -2,137 2024 122,235 -139,042 -16,807 -9,332 -5,298 -3,071 -1,814 2025 124,112 -141,177 -17,065 -9,111 -4,981 -2,784 -1,588 2026 79,056 -89,926 -10,870 -5,580 -2,938 -1,583 -872 2027 73,424 -83,520 -10,096 -4,984 -2,526 -1,313 -698 2028 139,130 -158,261 -19,130 -9,080 -4,433 -2,221 -1,140 2029 142,885 -162,532 -19,647 -8,966 -4,215 -2,037 -1,010 2030 131,621 -149,719 -18,098 -7,942 -3,595 -1,675 -802 Total -252,664 -155,475 -102,208 -71,333 -52,466 Annual average -12,032 -7,404 -4,867 -3,397 -2,498 97 Figure 37: Projected net reduction in income due to impacts of rising temperatures on crystallization rates in the fractionation plant and hence olein and stearin production, at 12% discount rate Recommended adaptation actions If any upgrades to the refrigeration and chiller system in the fractionation plant are undertaken, it is recommended to check design thresholds for resilience to climate change. 5.6 Power production Sensitivity of power production to temperature GODPC has installed a 2.5 MW power station to provide all power needs. The 30 ton/hr boiler operates on biomass waste (empty fruit bunches, fibers and kernel shells) from the mill. Warmer air as a result of climate change will be beneficial to the boiler, as less pre-heating will be needed. High pressure steam from the boiler is passed through a back pressure turbine to generate power and provide all process steam needed for the processing operations. The balance of electrical power is generated from feeding high pressure steam to a condensing turbine. The condenser utilizes cooling water and so its performance will be affected by rising cooling water temperatures. The temperature of the cooling medium has a significant effect on the efficiency of a condensing steam turbine. The lower this temperature, the higher the efficiency that can be attained because the pressure in the condenser is lower, producing a greater useful enthalpy drop in the steam turbine output. This is illustrated in Figures Figure 38 and Figure 39. (GOPDC has a wet cooling tower). Note that Figures Figure 98 38 and Figure 39 are generic figures; they are not specific to GOPDC’s power plant. Figure 40 shows the relationship between cooling water inlet temperature and condenser pressure based on data recorded by GOPDC in May–June 2010. While there is a lot of scatter in the data, the increase in condenser pressure associated with higher cooling water temperatures can be clearly seen. Figure 38: Typical condenser pressure values as a function of the temperature of the cooling medium for water cooling with a wet cooling tower; direct water cooling (using water drawn from a river, lake or the sea); and direct air cooling 400 350 Condenser Pressure (mbar) 300 250 Cooling Tower 200 Direct Cooling Air Cooled Condenser 150 100 50 0 -15 -10 -5 0 5 10 15 20 25 30 35 40 45 o Temperature of Cooling Medium ( C) Figure 39: Typical effect of condenser pressure on steam turbine efficiency 110 % Steam Turbine Ouput (MW) 105 100 95 Single Pressure Dual Pressure 90 85 80 20 40 60 80 100 120 140 160 180 Condenser Pressure (mbar) 99 Figure 40: Relationship between cooling water temperature and condenser pressure for GOPDC’s turbine ° The design of the GOPDC condensing turbine and the condenser is not known but typically a 1 C increase in temperature will result in about a 1% decrease in turbine efficiency, so steam consumption will rise by 1% to maintain a constant power output. As long as the turbine has the incremental capacity to accept more steam feed and sufficient biomass waste is available, the increase in fuel cost for GOPDC is marginal. However, if purchased fuel is used (if biomass is not available) then the cost of fuel purchase will increase. We do not have data on how much fuel is needed (if any) to supplement biomass, so below we have provided some assumptions and associated estimates, drawing on information on fuel use in GOPDC’s Environmental Management Plan (EMP, 2008): 1. We assume that all fuel used at GODPC is oil based. 2. The term “diesel” is used freely in the EMP and we do not know whether this is purchased as transport diesel or light distillate fuel oil, and whether these requirements are separated because of differences in taxation between transport and heating uses. 3. The EMP says that average annual fuel usage is 1.13m liters for the “Factory”. We do not know what this includes, but it could cover: o diesel for transport o fuel for hot air production for drying in the mill (if air is not heated by steam) o emergency power generation from diesel generators o emergency steam production from the package boiler o supplementary firing for the power plant if insufficient biomass is available. 4. Excluding any tax costs, the current spot price for diesel in Europe is in the region of US$0.8 per liter. Therefore the total fuel cost to GODPC at spot prices would be approximately US$900,000 per annum. 100 5. As a worst case assumption, we could assume that all of the 1.13m liters was used in power production. ° 6. Cooling water temperature increases of 1 C (in line with the projected median air temperature increase by the 2030s) would result in an approximate 1% loss of efficiency and hence 1% more fuel. The additional purchase cost for fuel would be US$9,000 per annum (undiscounted) (or US$500 per annum at a discount rate of 12%) by the 2030s, i.e. small, even under this worst case scenario. Rising temperatures will have other cost implications for running the power plant. Water use will increase, bringing extra costs for water pumping and treatment. The numbers of fans running may need to increase, with associated increased power consumption. There may also be increased maintenance costs, if higher temperatures lead to increased dust levels and the air filters and fans need to be cleaned more frequently. Recommended adaptation actions As biomass waste is free and GODPC reports that the power plant can supply the required power, there would appear to be no need for action to combat the impacts of temperature increases on power output. However, we recommend that the margins of design for the power plant in relation to temperature be confirmed. Chapter 5 References AY & A Consult Ltd. (2005). Ghana Oil Palm Development Company Ltd Revised Environmental Impact Statement for the proposed vegetable oil refinery / fractionation plant at Kwae. Field, A. (2005). Discovering statistics using SPSS. SAGE Publications Ltd, London, UK. McSweeney, C., New, M. and Lizcano, G. (2008). UNDP Climate Change Country Profiles – Ghana. Noormahayu, M.N., Khalid, A.R., and Elsadig, M.A. (2009). Financial assessment of oil palm cultivation on peatland in Selangor, Malaysia. Mires and Peat, 5, pp. 1–18. SAL Consult. (2008). Environmental Management Plan. 101 Chapter 6: Climate risk analysis for water and wastewater at GOPDC 102 Introduction Water is a central issue at GOPDC, as the oil palm nursery and the industrial operations depend on groundwater and because the industrial operations generate wastewater (primarily palm oil mill effluent, or POME). Groundwater is used for irrigating seedlings in the nursery; as steam for electricity generation; for boiling and sterilizing palm fruits; for mill, refinery and fractionation operations and as potable water for offices and residential areas (GOPDC, 2008). For the last two to three years, there has been some competition for groundwater resources between the nursery and the industrial operations, and in 2010, three to four days of production at the refinery were lost due to lack of water. Concerns over groundwater availability are therefore gaining prominence. POME is the liquid waste from the milling process which requires treatment before it can be discharged into water courses. In this chapter water input issues are addressed, including the factors that control groundwater quantity and quality; the use, quantity and quality of groundwater at GOPDC; how climate change and other pressures might affect groundwater reserves; and possible climate change adaptation measures. Similarly, in the case of POME, the wastewater management system at GOPDC is considered, including how this system and surrounding water courses might be affected by climate change, and possible adaptation options. Overall, the issue of ongoing groundwater availability emerges as a key area of uncertainty for GOPDC. The resource is not well studied; Newmont Ghana is developing the nearby Akyem gold mine, which will have significant impacts on water resources in the area; and climate change will also affect water resources. Given the criticality of water for operations at GOPDC, it is a high priority for the company to develop a better understanding of future water availability and to work together with other stakeholders to devise management strategies. 6.1 Groundwater The key issues In general, three main factors influence the quantities of groundwater available in a given location: rock porosity, rock permeability and the rate of replenishment. Replenishment of groundwater resources refers to the degree to which water abstracted from the aquifer is replaced, and is dependent on a variety of factors such as the nature of the rocks, soil and vegetation covering them, and also on the climate of the region (Price, 1985). The main climatic factors influencing groundwater recharge are precipitation and temperature (and hence evaporation and evapotranspiration). Groundwater recharge is mediated by a number of interlinked factors, which makes it very complex to model. Groundwater resources are impacted by various factors such as surface water supplies, catchment area, climate, soil, and the amount of water abstracted. Figure 41 illustrates where groundwater sits within the overall water cycle, and how it interacts with other factors. Many groundwaters both change into, and are recharged by, surface waters (Bates et al., 2008). Soil plays a crucial role in regulating groundwater recharge, since water has to percolate through the soil in order to reach the aquifer. Individual intense, heavy rainfall events risk exceeding the infiltration capacity of the soil (i.e. the maximum rate at which water can enter the soil), which means less water is able to soak into the aquifer through the ground and more of it escapes as runoff into surface water resources. Also, if a heavy downpour follows an extended dry period, the rainfall will tend to run rapidly off the soil surface, rather than percolating into it. In Ghana as in many other places, climate change is projected to lead to an increased proportion of total annual rainfall falling in heavy rainfall events (McSweeney et al., 2008), which may reduce groundwater recharge even if total rainfall increases. Abstraction of groundwater, whether it be by communities or industrial operations, will also influence groundwater levels and recharge. 103 Figure 41: The water cycle (USGS, 2009) Changes in climatic conditions, mainly temperature and precipitation, can have important influences on groundwater recharge and levels. Groundwater levels tend to be more strongly correlated with precipitation, with the majority of recharge taking place when precipitation percolates through the ground down to aquifers. Prolonged dry periods can alter an aquifer’s ability to transport groundwater, reducing the groundwater supply as the aquifer ultimately becomes less conductive (Chen, 2004). Such dry periods may also dry out the soil to the point that when rain eventually falls, the soil is unable to absorb it fast enough to prevent the water from running away along the ground surface, also reducing recharge (Holman, 2006). Increased temperatures also lead to increased evaporation and evapotranspiration, particularly for shallow aquifers closer to the Earth’s surface, which in turn may lead to a decrease in surface runoff, water levels and thus in groundwater recharge (Döll & Frölke, 2005). It should be noted that climate impacts on groundwater levels generally display a time delay, adding another dimension of complexity. For example, Chen (2004) found that the upper carbonate aquifer in Manitoba, Canada had a time delay of up to 1.5 to 2 years following a precipitation or temperature event before the response could be seen in groundwater levels. Recent research by Ng (2008) has revealed that climate impacts on groundwater recharge can be dramatic. According to this research, just a 20% decrease in rainfall can in some cases lead to a 70% decrease in aquifer recharge. To begin to understand climate risks to groundwater resources, hydrogeological surveys can be conducted and models can be developed relating climate (and climate change) to resource availability. However, no such information exists for the aquifers on which GOPDC depends. 104 Groundwater at GOPDC – usage, quantity and quality Groundwater usage Groundwater pumped from boreholes constitutes the sole water supply (apart from rainfall) for GOPDC’s Kwae and Okumaning estates. At Kwae, a total of nine boreholes are currently operational. At Okumaning there are four boreholes and seven hand-dug wells, of which two boreholes and four wells are currently functional. These boreholes tap into an aquifer shared with the local community (Atsrim, Pers. Comm., 2008). Of all the groundwater uses mentioned in the introduction to this chapter, the mill, refinery and nursery use the largest amounts. Most water is used for mill operations, though the amount of water used annually can 3 3 13 vary greatly. It ranged between 92,000m in 2007 and 242,000m in 2002 (GOPDC, 2008) . At the mill, water use is proportional to the production rate of crude palm oil (CPO): processing 1 ton of fresh fruit bunches (FFBs) uses approximately 1 ton of water. Annual trends of water use in mill operations show higher usage in the first half of the year, with a peak around March–April associated with the peak in FFB processing (GOPDC, Pers. Comm., 2008). The refinery uses considerably less water than the mill. In 2008, the first year of refinery operations, it was 3 estimated that the refinery used between 10,000 and 14,000 m of water per month (GOPDC, 2008). Peak water use months are also March and April, corresponding to the peak in FFB processing and consequent refinery operations. The water consumption for nursery irrigation is much lower again, with annual 3 consumption estimated at 17,000 m in 2006 (GOPDC, 2008). Prior to its use in the industrial processes, it must be filtered and softened. Some parts of the operation require demineralized water; the largest quantity being used in the boiler. Groundwater quantity Kwae estate lies on the Lower Birimian hydrogeological grouping, where aquifers occur mostly in the weathered zones below the bedrock. The main weathered zone aquifer materials are dark graphite with traces of quartz and some sandstone (GOPDC, 2005). Intense fracturing and moderate weathering of rock formations in the Lower Birimian have resulted in high yielding aquifers, and as a result the area has good groundwater resources. Borehole yields within the Birimian formation have been found to range from less 3 3 than 1.0 up to 27.0 m /hour, with a mean estimated yield of 3 m /hour (Dapaah-Saikwan, 2000). GOPDC has recently commissioned a rehabilitation exercise on its boreholes (Starco Ventures, 2009), which involved measuring borehole yields during blowing (development) and pumping tests. Blowing 3 yields at Kwae were in the range 7.2–30 m /hour, while pumping test yields were in the range 5.4–20 3 3 m /hour. For one borehole at Okumaning, the blowing yield was low, at 1.5 m /hour. With the exception of one borehole at Kwae (Estate Pump 2) and the borehole at Okumaning, the boreholes were found to be high yielding and have fast recovery rates after pumping (within an hour). The exercise provided some recommended actions to improve the quantity and quality of borehole water, namely: drilling two fresh boreholes; extending the depths of some pumps; adding valves to control some of the pumps; and increasing the size of one of the pumps serving the mill. It is understood that the aquifer in the Birimian formation, which GOPDC is tapping into, is fairly shallow (Atsrim, Pers. Comm., 2008). According to Dapaah-Saikwan (2000), the aquifer depths in this formation range from 13 to 60m. It is therefore possible that GOPDC is vulnerable both to increasing temperatures 13 Note that the lack of adequate meters for measuring water consumption and abstraction means that these figures are not precise. Nevertheless, these estimates, listed in GOPDC’s Environmental Management Plan (2008), provide a good insight into the breakdown of water consumption from different operations. 105 leading to increased desaturation of the aquifer, and to overexploitation by other users (Nyame, Pers. Comm., 2008). The tests by Starco Ventures show that the boreholes at Kwae range in depth from 42 to 66m. GOPDC has also recorded some monthly measurements of the “Evolution Dynamic Water Level” at Borehole 3 (which has been used to create Figures Figure 42, Figure 43 and Figure 44 in the following section). Groundwater quality Measurements of groundwater quality from two boreholes drilled in 2004 close to the mill indicated that the water was acceptable for domestic use and drinking (GOPDC, 2005). Only manganese levels were higher than the World Health Organization (WHO) recommended limit, and one borehole showed some bacterial growth. More recent measurements of levels of bacteria in GOPDC’s boreholes indicated that seven out of nine of them were safe to drink, conforming to WHO and Ghana standards (Water Research Institute, 2008 & 2009). Climate, climate change and groundwater As noted in the previous section, there are multiple factors affecting groundwater levels, with precipitation and temperature being the key climate variables. Here, first the data provided by GOPDC are examined to identify any existing trends, and then possible future impacts of climate change are discussed. 14 Figures Figure 42 and Figure 43 illustrate the Evolution Dynamic Water Level in GOPDC’s Borehole 3 . Figure 42 shows that there are no clear seasonal fluctuations discernable in the groundwater levels in Borehole 3, nor are there clear trends in the evolution of the groundwater level over a 10 year period. It was reported by Mr. Atsrim, GOPDC Deputy Managing Director (Pers. Comm. 2008) that the borehole yield had decreased somewhat in one of GOPDC’s boreholes that was first used 20 years ago. Figure 43 15 supports this claim, with groundwater levels showing a decrease since 2002 . Mr. Atsrim also reported that the water level in boreholes generally drops over the course of the year, associated with the rates of water use for processing FFBs. This is also supported by the slight downward trend that can be seen between the average monthly water levels during the first half of the year and the second (see Figure 44). 14 It is not clear whether the data have been recorded as spot measurements or monthly average dynamic water levels within Borehole 3. 15 It should be noted that there are missing data points for the years 2005-08. 106 Figure 42: Annual changes in the dynamic water level in Borehole 3 for 1996–2008 (excluding 2006) Figure 43: Annual average water level in Borehole 3 from 1996 to 2008 107 Figure 44: Downward trend in monthly average water level in Borehole 3 over the course of a year Without knowledge of the other factors influencing recharge of the aquifer used by GOPDC (as discussed in the introduction) – for instance how long it takes for groundwater levels to respond to precipitation or temperature events – it is difficult to begin assessing the relationship between climatic conditions and 16 groundwater levels . It is therefore not possible to draw any firm conclusions about how temperature and precipitation affect groundwater levels. Some global-scale research on groundwater and climate change impacts has been undertaken, which includes Ghana. Figure 45 presents a map of average country values of computed groundwater recharge (calibrated to arid and semi-arid regions) (Döll and Frölke, 2005). This shows Ghana as having groundwater recharge rates of 75–100mm/year on average, for the period 1961–1990. Figure 45: Long-term (1961–90) average groundwater recharge (Döll & Frölke, 2005) 16 It is worth noting that barometric pressure fluctuations, however, can have a discernible impact on well water levels and may influence them, which may have some impact on readings. If atmospheric pressure is suspected to change with climate change, it may be useful to monitor this variable as well (Spane, 2002). 108 Figure 46 shows global maps of projected percentage change in long-term average annual groundwater recharge by the 2050s, compared to the 1961–90 average. Projections for the A2 (medium-high) greenhouse gas emissions scenario are shown in the middle row, and projections for the B2 (medium-low) emissions scenario are shown in the bottom row. For each emissions scenario, the outputs of two global climate models (HadCM3, the UK Hadley Centre model, and the ECHAM4/OPYC3 German Max Planck Institute for Meteorology model) are shown. (For further information on greenhouse gas emissions scenarios, see Annex A: Greenhouse gas emissions scenarios). Figure 46: Percentage change in 30-year average groundwater recharge between 1961–90 and 2041–70 (the 2050s), as projected by two global climate models under two different greenhouse gas emission scenarios (Döll & Frölke, 2005) In southern Ghana, the model outputs indicate a wide range of potential changes in groundwater recharge – ranging from -70% to +10% under the A2 emissions scenario, and from -10% to +30% under the B2 emissions scenario, relative to the 1961–90 baseline. This uncertainty is underlined by work conducted by Dapaah-Saikwan (2000) on groundwater recharge 17 rates for the Pra Basin in Ghana. It is understood that this research used the output of two climate models, which indicated overall annual decreases in precipitation and increases in temperature due to 17 Kwae is located in the eastern part of the Pra Basin, which covers 23,000 km2 and is located in the south western part of the country. The Pra River, from which it gets its name, is composed of four major tributaries: the Ofin, Oda, Anum, and Birim. 109 18 climate change. One approach adopted in this research, the “Continuity Equation” , indicated an annual deficit in groundwater recharge under all scenarios investigated, with a decreasing trend in groundwater recharge associated with greater amounts of climate change. An alternate approach applied, the Water 19 Balance model , indicated positive recharge values for all scenarios through to 2050, although they too 20 showed a decreasing trend with greater amounts of climate change, for both wet and dry seasons . The studies mentioned above used a small number of climate models which indicate that Ghana could experience decreased precipitation in the future. However, the UNDP Climate Change Country Profile for Ghana used in our study draws on a much a wider set of climate models (see Chapter 1). Across this broad set of models, rainfall projections are wide ranging, with approximately half of the models projecting increases in annual precipitation in future, and half projecting decreases. This makes it difficult to state with any confidence how local groundwater recharge at GOPDC may change. In terms of groundwater quality, again projections are unclear. Projected increases in precipitation intensity due to climate change can increase the rate of transport of pathogens and dissolved pollutants into groundwater, leading to a deterioration in quality (Bates et al., 2008). If they are realized, decreases in recharge rates pose an increased risk of contaminated groundwater, due to decreased dilution of pollutants (Eckhardt et al., 2003). Existing and recommended adaptation actions Concerns over groundwater availability have recently gained prominence at GOPDC, since the company has experienced some shortfalls in supply, as outlined above. As a result, water meters have recently been purchased, so that groundwater use can be closely monitored. GOPDC also has various plans to improve water efficiency including: stopping leaks; investigating opportunities for water recycling and reuse in the mill; and harvesting rainwater from the mill roof. However, it is recommended that GOPDC go further than these demand-side management actions. Given that water is essential to so many elements of its operations, GOPDC needs to develop a better understanding of future water availability, working together with other stakeholders to assess the risks and develop management strategies. The uncertainties GOPDC faces are not uncommon. The Intergovernmental Panel on Climate Change has identified lack of data and knowledge about groundwater as a key research gap (Bates et al., 2008). In line with this finding, GOPDC’s information about long-term borehole levels and yields is limited, and the company does not understand the factors affecting supply and demand from the aquifer it uses, presenting difficulties in determining the sustainability of current and future abstraction rates. As well as considering the effects of climate change, it is essential that GOPDC is aware of other groundwater users and how their needs are changing. Significantly, Newmont Ghana is developing the Akyem gold mine near the town of New Abirim, approximately 10 km from the Kwae estate. Ghana Environmental Protection Agency (EPA) highlighted that over its lifetime the mine will create a large pit (approximately 2.6km long, 900m wide and 450m deep) which could potentially draw down water from the aquifer used by GOPDC, if it is lower (Ransford Sekyi, Ghana EPA, Pers. Comm.). Many millions of tons of water will be pumped out by Newmont over the coming 20 years. 18 Note that this approach used the mean annual values of precipitation, evapotranspiration, runoff and abstraction rates to calculate a baseline scenario for change in groundwater storage which was applied to low, medium, and high climatic sensitivity scenarios to estimate groundwater recharge in the 2020s and 2050s. 19 See Dapaah-Siakwan (2000), Section 5.3.1. 20 It should be noted that the Continuity Equation approach is considered to be limited, and that the climatic scenarios used in the work by Dapaah-Siakwan are somewhat out of date. 110 According to Ghana EPA, while Newmont Ghana has done extensive investigation of groundwater in the mine area, it is not clear that they have investigated and ruled out potential impacts on the aquifer GOPDC uses. It is therefore recommended that GODPC engage with Ghana EPA and Newmont Ghana, and request that Newmont Ghana undertake a hydrogeological assessment that encompasses GODPC’s aquifer, to better understand the factors that influence its recharge. As well as investigating the mine’s impact, this assessment should also develop understanding of the relationships between GOPDC’s aquifer and climatic variables, and it is recommended that GOPDC request that the assessment also incorporate the impacts of climate change. Depending on the results of these assessments, further investigation into alternative groundwater sources may be deemed sensible. Both the exploration of deeper resources in existing boreholes and alternative borehole sites may be useful. With regard to deeper boreholes, it is understood that some promising findings have been made in the Affram Plains, where drilling for community boreholes has yielded very 3 high amounts of groundwater (18–30 m /hour) from boreholes about 100–160m in depth (Armah, Pers. Comm., 2008). It should be noted however that the Affram Plains are located in the Voltaian Formation approximately 100km east of Kwae, on a different geological structure to GOPDC, and that deeper drilling may be costly. In addition, given the uncertainties about future climate change impacts, it is recommended that GOPDC undertake long-term monitoring of groundwater yields (both of its own boreholes, and perhaps also those of the community), and analyze these alongside observed climatic conditions, in order to provide an early warning mechanism about potential resource constraints. 6.2 Wastewater Wastewater management at GOPDC At GOPDC, effluent from the mill is treated in a number of different ways before it is discharged into the Aberewa Stream. Initially, raw effluent is passed through a tricanter facility (centrifuge) that recovers excess oil and stores this in an oil recovery tank. From this point the effluent is known as Palm Oil Mill Effluent (POME) which then passes through a 5m deep four-chamber fat pit that acts as both sludge pit (desilting and desludging) and oil trap. Following this, the POME travels into two stabilization ponds – first an anaerobic pond (with a hydraulic retention time of 30–80 days), and secondly a facultative pond, bringing the total hydraulic retention time to 75–120 days. These ponds are designed to remove organic matter and solids, and to reduce both Biochemical Oxygen Demand (BOD) and Chemical Oxygen Demand (COD). The treated effluent then passes through an open drain into Aberewa Stream leading to Bobri Stream (GOPDC, 2008; GOPDC, 2003). A map showing the relative position of the mill and streams is shown in Figure 47. Owing to a doubling in the capacity of the mill since the wastewater treatment system was constructed, generally, wastewater quality parameters for conductivity, total dissolved solids (TDS), BOD, COD, oil and turbidity do not meet Ghana EPA permissible levels (GOPDC, 2008). In response to this, a third stabilization pond has recently been constructed. This will “enable the existing ponds to be desludged and also take some of the effluent during the peak production periods thereby reducing the load on the existing ponds and enable(ing) them to perform better” (GOPDC, 2008, p. ix). GOPDC reports that the new pond has dramatically reduced levels of pollutants, though they are not yet reaching Ghana EPA standards. While it is understood that local communities do not use water from Bobri stream for drinking (GOPDC, 2008), interviews conducted at GOPDC in December 2008 suggest that some farmers occasionally use river water for drinking when they are out working in the fields, and borehole water is not accessible. (It should be stressed that it is not known if this applies to Bobri Stream). 111 Figure 47: Map of GOPDC water courses, mill site and effluent discharge route Wastewater, climate and climate change In considering the relationships between wastewater management and climate, it is useful to separate the relationship between climate and the treatment system (the tricanter, fat pit and stabilization pond system) from the relationship between climate and treated effluent in streams. In terms of the relationship between climate and the treatment system, two key issues can be identified. First, flooding of the fat pit or ponds: according to the IPCC (2007), water quality impacts may occur more frequently in future as a result of overloading the capacity of wastewater treatment plants during extreme rainfall (IPCC, 2007, p189). While climate change projections for Ghana regarding seasonal rainfall amounts are uncertain, there is greater confidence in an increasing number of “heavy” rainfall events. Either increases in seasonal rainfall or these heavier events could therefore potentially lead to overflows from the treatment process. However, stormwater drainage does not enter the treatment ponds (with the exception of the planned redirection of wastewater from the tricanter section of the mill), so increases in stormwater runoff will not affect the treatment ponds. In relation to flooding of the ponds from direct rainfall falling on to them, it is understood from GOPDC that overflows have not occurred in the past, even under conditions of extreme rainfall, and that the added capacity provided by the third stabilization pond will reduce the risk of this happening in the future. Secondly, the relationship between temperature and the bacteria in the anaerobic pond is of interest. Digesters using mesophilic bacteria are usually heated to 30–Û&ZKLOHWKRVHXVLQJWKHUPRSKLOLF EDFWHULDDUHKHDWHGWRDSSUR[LPDWHO\Û& )$&( It is understood that GOPDC’s treatment ponds do not require heating, because ambient air temperatures are sufficient to maintain bacterial activity (and it is therefore assumed that the bacteria are mesophilic). Due to climate change, air temperatures at *23'&DUHSURMHFWHGWRLQFUHDVHE\DSSUR[LPDWHO\Û&E\WKHV UHODWLYHWRWKHEDVHOLQHSHULRG 1970–1999; see Chapter 1, Climate). This may improve the efficiency of the anaerobic digestion process and hence the quality of the effluent entering the stream. With regard to the relationship between climate and treated effluent in streams, rainfall and temperature are the key influencing factors. In terms of rainfall, if seasonal rainfall amounts were to increase (which is uncertain), then higher stream flow rates would be more effective in diluting the effluent discharge (Whitehead et al., 2008). (Clearly, in contrast, any reductions in seasonal rainfall driven by climate change 112 would reduce the capacity of the streams to dilute effluents). Furthermore, the baseline stream water quality could itself be affected by changes in rainfall – for instance, higher rainfall can lead to increased erosion of stream banks and hence higher suspended solid concentrations, as well as changes in BOD and COD. Additionally, if stream water use patterns changed in the future as a result of climate change (e.g. if increased volumes of water were used for irrigation), stream flow rates could again be affected and in turn water quality. Rising ambient air temperatures due to climate change will lead to higher stream temperatures, which can detrimentally affect water quality. According to the IPCC, “water quality generally would be degraded by higher water temperatures (high confidence)” (IPCC, 2001, Technical Summary, Section 4.1). Other things being equal (e.g. stream flow) “increasing water temperature alters the rate of operation of biogeochemical processes (some degrading, some cleaning) and, most important, lowers the dissolved oxygen concentration of water” (IPCC, 2001, Technical Summary, Section 4.1). These changes could in turn affect the health of freshwater ecosystems. Possible adaptation strategies It has not been possible to establish quantitative links between climate or climate change and wastewater management based on the limited information held about the processes at GOPDC and on the streams. Additionally, uncertainty about whether future seasonal rainfall will increase or decrease makes it difficult to undertake a robust analysis of climate change impacts and to recommend firm adaptation actions. The addition of the third treatment pond has reduced pollutant concentrations, though they do not yet meet Ghana EPA standards. Given that BOD has been a concern, it may be worth GOPDC further investigating the potential impacts of higher temperatures (which are projected with high confidence) on dissolved oxygen concentrations in the streams. As noted in Chapter 2 (Yield), GOPDC is testing POME on a small area of palms behind the mill. The main reason for doing this is to make use of the effluent, rather than disposing of it via the wastewater treatment system. It is also understood that GOPDC is investigating a pilot bio-methanation project. 6.3 Conclusion Water is a critical issue to consider at GOPDC, both at present and in light of possible climate change impacts and demand from other users. As discussed, further investigation into the relationships between climate variables and both groundwater and wastewater is needed before GOPDC can properly understand its vulnerabilities and establish what actions it should take. Chapter 6 References Armah. (2008). Discussion on groundwater in Kwae. (Personal communication, December, 2008). Atsrim. (2008). Discussion on groundwater in Kwae. (Personal communication, December, 2008). Bates, B.C., Kundzewicz. Z.W., Wu, S., Palutikof, J.P. (Eds.). (2008). Climate Change and Water. Technical Paper of the Intergovernmental Panel on Climate Change, IPCC Secretariat, Geneva, 210 pp. Brouyere, S.,G. Carabin, Dassargues, A. (2004). Climate change impacts on groundwater resources: modelled deficits in a chalky aquifer, Geer basin, Belgium. Hydrogeological Journal, 12: 123–134. Chen, Z., Grasby, S., Osadetz, K. (2004). Relation between climate variability and groundwater levels in the upper carbonate aquifer, southern Manitoba, Canada. Journal of Hydrology, 290: 43–62. 113 Dapaah-Siakwan, S. (2000). Chapter 7: Impact of Potential Climate Change on Groundwater Recharge. In: Climate Change Vulnerability and Adaptation Assessment of Water Resources in Ghana (2000) Ghana Environmental Protection Agency, Republic of Ghana. Döll, P., Flörke, M. (2005). Global-scale estimation of diffuse groundwater recharge. Eckhardt, K., U. Ulbrich. (2003). Potential impacts of climate change on groundwater recharge and streamflow in a central European low mountain range. Journal of Hydrology, 284: 244–252. FACE. (2009). Website: http://www.face-online.org.uk/ Last accessed 22/07/09. Freympong, D.G., Agyekum,W.A., Larmie, S.A. (1996). An Assessment of Groundwater Resources of the Eastern Region of Ghana. Water Resources Institute (CSIR), Accra, Ghana. 53pp. GOPDC (2003) Final Environmental Impact Statement for the Proposed Oil Palm Development Project at Okumaning. GOPDC. (2005). Revised Environmental Impact Statement for the Proposed Vegetable Oil Refinery/ Fractionation Plant at Kwae. Accra, Ghana. 61pp. GOPDC. (2006). Report on Monthly Effluent Quality Monitoring (August 2006–October 2006) GOPDC. (2008). Environmental Management Plan Holman, I.P. (2006). Climate change impacts on groundwater recharge-uncertainty, shortcomings, and the way forward? Hydrogeology Journal, 14: 637–647. IPCC. (2001). Climate Change 2001: Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Third Assessment Report of the Intergovernmental Panel on Climate Change, J.J. McCarthy, O. F. Canziani, N.A. Leary, D. J. Dokken, K. S. White, Eds., Cambridge University Press, Cambridge, UK. IPCC. (2007). Climate Change 2007: Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, M.L. Parry, O.F. Canziani, J.P. Palutikof, P.J. van der Linden and C.E. Hanson, Eds., Cambridge University Press, Cambridge, UK, 976pp. McSweeney, C., New, M., Lizcano, G. (2008). UNDP Country Change Profiles: Ghana. Available online: http://country-profiles.geog.ox.ac.uk/UNDP_reports/Ghana/Ghana.hires.report.pdf Ng, Gene-Hua Crystal. (2008). Water supplies could be strongly affected by climate change. Website: http://web.mit.edu/newsoffice/2008/agu-groundwater-1218.html Last accessed 24/07/2009. Nyame (2008) Discussion on groundwater in Kwae. (Personal communication, December, 2008). Opoku-Ankomah, Y. (2000). Chapter 5: Impacts of Potential Climate Change on River Discharge. In: Climate Change Vulnerability and Adaptation Assessment of Water Resources in Ghana (2000) Ghana Environmental Protection Agency, Republic of Ghana. Price, M. (1985). Introducing Groundwater. George Allen & Unwin, London, UK. 195pp. Whitehead, P., Butterfield, D., Wade, A. (2008). Potential impacts of climate change on river water quality. Environment Agency, Bristol, UK. 114 Spane, F. A. (2002), Considering barometric pressure in groundwater flow investigations, Water Resour. Res., 38(6), 1078, doi:10.1029/2001WR000701. Starco Ventures Ltd. (2009). Borehole rehabilitation report for the Director of Operations, Ghana Oil Palm Development Company. Koforidua, Ghana. USGS [United States Geological Survey]. (2009). Website: http://ga.water.usgs.gov/edu/watercycle.html. Last accessed 27/07/2009. Water Research Institute, Council for Scientific and Industrial Research. (2008). Analytical Report (on bacterial quality of GOPDC boreholes), November 2008. Achimota, Ghana. Water Research Institute, Council for Scientific and Industrial Research. (2009). Analytical Report (on bacterial quality of GOPDC boreholes), March 2009. Achimota, Ghana. 115 Chapter 7: Climate risk analysis for community and social issues 116 Overview Ghana has for some time been regarded as a success story in a region that has been scarred by decades of civil war and conflict. A relatively peaceful history and stable economy have helped place it within the “Medium Human Development” category of the latest Human Development Index (HDI), higher than any other West African nation. Despite this, Ghana remains a poor country relative to developed nations. It was ranked 135 out of 177 countries in the United Nations Human Development Report 2008, with adult literacy standing at 59%, 50% of the population enrolled in education beyond primary level, and 25% of people without access to clean water (UNDP, 2008). Within Ghana, the Eastern Region, where GOPDC is located, is also relatively poor: the incidence of poverty and extreme poverty in this region are above the national average. The government considers this to result from (i) the inability of farmers to process agricultural produce during the immediate post harvest period (e.g. because small scale farmers cannot afford to buy processing equipment), (ii) the poor quality of feeder roads in the food crop growing areas of the region, (iii) poor health among farmers (associated with high levels of malaria and diarrhea and low levels of immunization), and (iv) low levels of basic education, poor provision of school infrastructure and learning materials and an inability to attract and retain teachers (Government of Ghana, 2003). In Kwaebibirem District (the district where GOPDC is located), agriculture is the major economic activity, with food crop production the most common activity, followed by cash crop production (Ghana Districts website, 2009). Climate change impacts on GOPDC (e.g. positive/negative impacts on oil palm yield or reduced output from the refinery due to higher temperatures) are also likely to affect the local community in the area. The local community is defined here as those people living or working in and around GOPDC’s plantations and facilities, including outgrowers, smallholders, farmers, local villagers, GOPDC workers and their families. In this chapter, we address (i) how the community surrounding GOPDC may be affected by climate change, (ii) what other social issues are present in the community, (iii) how this combination of climate driven and non-climate driven social and community issues may impact upon GOPDC and (iv) what possible actions GOPDC might take to address these impacts. Given the wide-ranging set of issues discussed in this chapter and because we have less data on these issues than we do for other activities managed by GOPDC, this chapter is more qualitative than others. 7.1 Climate impacts on the community around GOPDC The climate change scenarios that have been discussed with respect to their impacts on GOPDC (see Chapter 1, Climate) are the same scenarios that may affect the local community. Here we consider what aspects of community members’ lives may be affected (agricultural/pastoral yields, transport to market, price of agricultural produce, health, water access) as well as how they may be affected (increasing/reducing their income, food security and/or well-being). It is important to note that while the community may be affected similarly to GOPDC, their experience of climate change may differ, due to poverty which in turn increases vulnerability and makes adaptation to climate risks more difficult. GOPDC’s Community Relations Officer notes that climate impacts already affect the community and have knock-on consequences for GOPDC: low rains lead to low incomes, and consequent increased delinquency, prostitution, HIV/AIDS and child labor. Kwaebibirem District has the second highest rate of HIV in Ghana’s Eastern Region and this has a negative impact on GOPDC (Alhaj Bashir Manu, Pers. Comm.) Low rains have also led to farmers invading GOPDC’s Biodiversity Plots for bushmeat. 117 Climate impacts on agriculture and livestock This section presents a broad overview of climate-related risks to the agricultural crops and livestock that support community livelihoods. Agriculture is the main occupation for rural workers in Ghana, involving some 75% of the rural workforce nationally (Ghana Statistical Service, 2008) and 77% of the workforce in Kwaebibirem District (Government of Ghana, 2003). Oil palm Oil palm yield at GOPDC could be affected by climate change. As described in Chapter 2, there is uncertainty in modeling the impacts of climate change on yield – some models project decreases in fresh fruit bunch (FFB) yield (as much as -75%) while others project increases (up to +166%). Community members that grow oil palm are likely also to experience changes in yield driven by climate change. Depending on the nature of change, this may in turn affect their income security. In Ghana’s forest zone, the annual value of all agricultural crops to households is GHC 32.32 million, with oil palm harvest providing GHC 2.68 million, i.e. about 8% of the total. Oil palm is the third most important harvested crop, behind plantains (contributing GHC 7.12 million) and tomatoes (GHC 3.46 million). Some of the harvested oil palm is used by the households themselves, but they also sell their produce. The annual value of sales of all agricultural crops to households in forest zones is GHC 13.57 million. Tomatoes make the highest contribution to sales (GHC 3.0 million), followed by cassava (GHC 2.75 million) and oil palm (GHC 1.17 million, or 9% of the total value of sales) (Ghana Statistical Service, 2008). Across rural populations in Ghana, average annual household spending on food and beverages is GHC 2,721 of which palm oil takes GHC 21.6 (i.e. less than 1% of the food and beverage budget) and palm kernel oil takes GHC 10.0. So, while palm oil is not a large component of household expenditure, as indicated above, it is an important source of income. Other food crops According to Mr. E. K. Ametepe, the Kwaebibirem District Director of Agriculture, many in the local community grow other food crops in addition to or instead of oil palm. These include cocoa (the most common), citrus, root and tuber crops (cocoyam, cassava, yam), grains (maize, rice), plantain, kola, beans and cowpea. Like oil palm, these crops may be influenced by future changes in climate. Here we note some of the ways that these crops are affected by climate, and thus some of the ways that they may be affected by climate change. Cocoa & Citrus: Neither cocoa nor citrus plants can be grown on lowland as they respond negatively to flooding (Outgrowers’ meeting, GOPDC, 2008). Cocoa is highly susceptible to drought, and can only be profitably grown under temperatures varying between 30–Û&PHDQPD[LPXP–Û&PHDQPLQLPXP DQGDEVROXWHPLQLPXPRIÛ& *KDQD(QYLURQPHQWDO3URWHFWLRQ$JHQF\ $GGLWLRQDOO\FLWUXVLVQRW drought resistant and needs well drained soils (E. K. Ametepe, Pers. Comm.). Cocoyam, Cassava, Yam: The main stress factors that affect these crops are drought, water logging, temperature extremes, solar radiation extremes and nutrient imbalances. For example, Ghanaian root and tuber crops suffered in 1990 under drought conditions but recovered in 1991–1992 when rains returned to “normal” (Ghana Environmental Protection Agency, 2008). Maize: Cereals such as maize are projected to experience negative yield impacts in low latitudes even under moderate temperature increases of WRÛ& DVare projected for Ghana by the 2030s) (IPCC, 2007; McSweeney et al., 2008). Even if potential production increases due to increases in atmospheric concentrations of CO2, higher temperatures combined with possible increased frequencies in drought 118 could reduce grain and other crop yields substantially (IPCC, 2007). Mr. Ametepe noted that maize yield could decrease by as much as 50% following a week of drought at a critical time of year (E. K. Ametepe, Pers. Comm.). Additionally, “recent studies in Ghana indicate that an increased number of extreme weather events will worsen food security, decreasing, for example, the maize yield by seven per cent by 2020” (Ghana Environmental Protection Agency, 2008, p6–8). Plantain, Kola nuts & Cowpea: Limited information is available about climate impacts on crops such as plantain and cowpea (West Africa Centre for Crop Improvement website, 2009). As with oil palm, currently these crops are all rain fed (Bashir Manu, Community Relations Officer, GOPDC, Pers. Comm.). Depending on the need and capacity to adapt to changes in climate, the local community may be affected by changes in income derived from these crops as well as by changing food security. Livestock According to Mr. E. K. Ametepe, sheep and goats – which are the most common pastoral animals kept in the district – may be negatively affected by increases in rainfall, which lead to foot rot, skin diseases and increased incidence of parasites. Under a future with more extreme rainfall conditions (as indicated by UNDP climate projections for Ghana) these diseases may become more common. Climate impacts on transport infrastructure As most roads in and around GOPDC are not paved, they can become difficult to use or unusable by vehicles during the rainy season, and some locations are occasionally cut off by river flooding after heavy rains (E. K. Ametepe, Pers. Comm.). This can lead to spoiling of farmers’ produce, and GOPDC’s outgrowers provided a rough estimate that about 2% of their FFB harvest is left to rot, because they cannot get it to the GOPDC collection points (Outgrowers’ meeting, GOPDC, Dec 2008). If, as projected, intense rainfall events become more common in future, it may become harder for farmers to transport oil palm, other cash crops or livestock to market on existing roads. Climate impacts on sale prices for agricultural goods In recent years the demand for palm oil has increased due to a range of factors: increased prosperity in India and China, escalating fossil fuel prices and increasing concerns over energy security. In turn the price of palm oil has generally increased, though it sees large fluctuations (Wahid et al., 2007). Climatic factors may have also played a role in palm oil prices: world production of palm oil experienced a shortfall from 17.84 million tons in 1997 to 16.68 million tons in 1998 in response to wide-scale droughts brought about by an El Niño climatic event which coincided with a time when palms were facing cyclical stress (Basiron, 2002). Any climate-related impacts on world palm oil prices will in turn affect outgrowers, who receive a sale price from GOPDC which is linked to the world market price. Similarly, the farming community may be affected by price fluctuations in the markets of other cash crops due to changes in climate and world production levels. Climate impacts on health In its Third Assessment Report, the IPCC concluded that: “overall climate change is projected to increase threats to human health, particularly in lower income populations, predominantly within the tropical/sub-tropical countries” (IPCC, 2001, p 44). 119 The IPCC Working Group II Technical Report (2001) states that climate change can affect human health directly (e.g. impacts of thermal stress, death/injury in floods and storms) and indirectly through changes in the ranges of disease vectors (e.g. mosquitoes), water-borne pathogens, water quality, air quality and food availability and quality. Actual health impacts will however be strongly influenced by local environmental and socio-economic conditions, and by the range of social, institutional, technological and behavioral adaptations taken to reduce the full range of threats to health (NCAP, 2006). Economically, negative health impacts can lead to “a loss of manpower, decreased productivity and stress to the National Health Insurance Scheme” (Ghana Environmental Protection Agency, 2008, p6–8). Around GOPDC, malaria is the most common illness: 56% of worker attendance at the GOPDC clinic was due to malaria in 2007. Some 35% of total reported cases at St. Dominic’s Hospital, Akwatia (the nearest hospital to GOPDC) between 1996 and 2008 were malaria cases. As detailed in Chapter 8, malaria incidence will likely be affected by climate change, and conditions favorable to malaria may become more common. This in turn may lead to more disease among the community and thus increased sick days taken and even, in some cases, death. While this directly affects GOPDC in terms of diminished workforce productivity and sick leave and healthcare costs, it also affects the wider community. The second most frequently reported illness at St. Dominic’s Hospital is acute respiratory infection (ARI) (10% of total reported hospital cases from 1996–2008). However, respiratory tract infections accounted for only 5% of worker attendance at the GOPDC clinic, behind malaria (56%), lacerated wounds (11%) and hypertension (7%). ARI is a common and serious complaint among many rural populations in developing countries and is generally linked to poorly ventilated cooking conditions where communities are burning traditional biomass fuels such as wood, agricultural wastes and animal dung (Kilabuko & Nakai, 2007). We have briefly investigated possible relationships between ARI cases and climatic factors: as shown in Figure 48, there seems to be an increase in cases of ARI at St Dominic’s Hospital one month after increased rainfall. This may suggest that ARI incidence rises because people spend more time indoors during rainy periods. Figure 48: Trends in average monthly reported cases of ARI against average monthly rainfall, 2004–2007 (St Dominic’s Hospital & Akim Oda rainfall data) 120 Communicable diseases such as typhoid fever and cholera (water-borne) as well as dengue and yellow fever (vector-borne) can also be exacerbated by floods (WHO, 2006). In the tropics, diarrheal diseases typically peak during the rainy season, which is most likely linked to the spreading of bacteria. As Kwaebibirem District has the lowest coverage of potable water supply to rural areas in Ghana and with about 40% of houses not having a toilet facility (Ghana Districts website, 2009), the district may be more vulnerable to these diseases. Climate impacts on water availability and quality Hand-pumped boreholes are the main source of water for the neighboring communities in the GOPDC area (GOPDC Social Impact Assessment, 2005). It is understood that the aquifer supplying these boreholes is shared with GOPDC, which is also reliant on groundwater for a range of operations (see Chapter 6, Water). Analyses of borehole water quality conducted by the Water Research Institute (WRI) did not find any evidence of impacts of GOPDC’s liquid effluent on borehole water quality. Climate change impacts on rainfall and temperature may have a direct impact on groundwater recharge and quality, although the relationship between these factors is not well understood (Bates et al., 2008; Chen, 2004). Increasing temperatures will however increase evaporation rates, which may in turn reduce groundwater recharge and limit groundwater supply (Döll & Flörke, 2005). Without a more developed understanding of the aquifer’s characteristics, it is not possible to assess how it might be affected by climate change (see Chapter 6, Water, for further details). 7.2 General social issues in the community around GOPDC The main social and community issues around GOPDC are related to land, as agriculture is the primary occupation in Kwaebibirem District. Land availability is impacted by the high population density in the area and by oil palm and citrus plantations. Consequently, farmers work small, dispersed plots of land with poor transport infrastructure between them. This system makes it uneconomical to introduce agricultural innovations such as mechanization and irrigation and also means that economies of scale are difficult, as is rapid delivery of goods to market. This competition for land has resulted in high rates and land disputes (Ghana Districts website, 2009). Land ownership has also been a contentious issue in the area, particularly related to GOPDC’s Okumaning concession. This is because, while the concession was previously government owned and then sold to GOPDC, the government never developed the land, so locals continued to farm it. When GOPDC acquired the land and began developing, settlers on the concession were obliged to leave yet felt a sense of ownership, as they had been living and working on it for over 25 years. There were 2,344 affected farmers on the concession, with about 50% of these already relocated and paid compensation. Farmers are paid cash to compensate for crops and structures lost as a result of resettlement, they are offered employment and outgrower opportunities at GOPDC to compensate for loss of livelihood and are not compensated for land. Resulting from the resettlement process, some local people felt frustrated and resentful towards GOPDC. 7.3 Direct and indirect impacts of community issues on GOPDC In this section we consider how both climate impacts on the local community as well as their general social situation may affect GOPDC directly (notably financially) and indirectly. Direct effects refer to issues such as impacts on outgrowers’ FFB yields and consequent impacts on the quantities of palm oil produced by GOPDC, whereas indirect effects refer to more uncertain impacts such as community dissatisfaction. 121 Direct impacts on GOPDC Any changes in the FFB yields of outgrowers driven by climate change will directly affect GOPDC as it will lead to changes in FFBs sold to GOPDC. GOPDC has approximately 7,000 outgrower farmers on long- term contract with the company, with an arrangement that GOPDC will supply oil palm seedlings, fertilizer, pest control, etc. and that the outgrowers will sell a certain portion of their oil palm crop to the company. Climatic disruptions to transport can also directly affect GOPDC if outgrowers are not able to deliver their FFBs to the 31 collection centers in its catchment area. Changes in the market price for palm oil (potentially linked to climate impacts on worldwide production) again may affect GOPDC via the community. This is because the price outgrowers receive for their FFBs from GOPDC is tied to the global market price. If the price becomes unfavorable to the outgrower, he/she may attempt to sell his/her FFBs elsewhere (even if this is not legal), thus leading to a reduction in GOPDC’s overall palm oil output. Finally, health impacts in the community associated with climatic conditions can also have a direct impact on GOPDC as outlined in Chapter 8 (Malaria). If disease incidence increases among outgrowers due to climate change, while GOPDC is not responsible for their sick pay or healthcare costs, the company may suffer from revenues-not-earned if illness leads to lower FFB quantities delivered to GOPDC. Indirect impacts on GOPDC In addition to direct financial impacts, GOPDC may be affected by community dissatisfaction. The combination of existing community pressures (as discussed in Section 7.2) and the new pressures that climate change may exert, together, could lead to increased hardship and potential tension. GOPDC notes that the community has high expectations of the company. If the community experiences difficulties, there are expectations that the company will address them. It is important to note here that this type of indirect risk is uncertain and that GOPDC is already engaging with and supporting the local community via job creation and investment in education and health. GOPDC provides employment and income (directly and indirectly) to over 50,000 people, a GOPDC Community Relations Officer works on improving communication between the company and the community, and GOPDC runs a school from kindergarten to Junior Secondary level for approximately 500 children on the Kwae Estate. Additionally, GOPDC assists local communities with the rehabilitation/construction of primary schools, village markets, water boreholes, power lines and sanitary facilities, supports communities during installation of chiefs and funeral ceremonies and also invests 1% of its turnover every year in community development projects. 7.4 Possible GOPDC actions Here we note that GOPDC is not responsible for climate change impacts on the local community or directly responsible for their social situation. Nevertheless, we now explore how it may be beneficial for GOPDC to act on these issues even though they are not its legal responsibility. With regard to the direct community impacts on GOPDC, the following may be of interest: x Conducting a cost benefit analysis (CBA) on whether it is worthwhile to work in collaboration with public authorities to improve transport infrastructure, which would in turn improve the ability of outgrowers (or others) to deliver FFBs to GOPDC collection centers. 122 x Investigating whether it is economically worthwhile to further invest in malaria control (as discussed in Chapter 8). With regard to the indirect community impacts on GOPDC, the following measures should be considered: x Involving GOPDC’s Community Relations Manager in discussing climate change issues with the community. x Supporting community awareness and education about crop diversification in the face of climate change. x Partnering with other organizations (NGOs, government and/or companies) locally, regionally, nationally and/or internationally to work on community issues. (Examples of collaboration on community issues in Ghana are presented in Boxes Box 3 and Box 4, for AngloGold Ashanti’s Malaria Campaign and the President’s Special Initiative on Oil Palm). Box 3: Case study—AngloGold Ashanti Malaria Campaign In 2006 the AngloGold Ashanti Obuasi gold mine launched its malaria campaign and managed to cut the local malaria infection rate by 73% in less than three years. The success of the initiative has been held as an example for the mining industry and for Africa’s fight against malaria. The campaign comprised a range of elements including: x Surveillance, monitoring and research using global experts in malaria and parasite prevalence x Community interaction: information, education, communication x Prevention: o Indoor residual spraying o Preventing mosquitoes from biting (nets, screening, repellents) o Controlling mosquito breeding (larvicide, environmental management) x Treatment (early, effective diagnosis and treatment; anti-malarial drugs) Key to its success was an innovative move to educate the wider community about infectious disease prevention through a radio program which reached well beyond the immediate area surrounding its Obuasi operation in Ghana. The Obuasi Malaria Control Centre was also set up as the headquarters for the Obuasi program, and as a training center for malaria control interventions at other AngloGold Ashanti operations, as well as a satellite research center for use by academic and government agencies. (Sources: Upton, M, World Malaria Day: Private Sector Must Do Its Share in Combating Infectious Disease, April 21, 2009, AngloGold Ashanti Report to Society 2007 Case Studies – Regional Health Threats, Obuasi malaria control programme: a model for Africa). 123 Box 4: Case study—The President’s Special Initiative on Oil Palm In 2003 the President of Ghana launched the President’s Special Initiatives on a range of different economic activities including Oil Palm. The aims of the PSI on Oil Palm are to achieve sustainable economic growth; create new pillars of growth; and generate wealth and employment, especially in rural areas. The initiative aims to achieve the following: x Each project will focus on a privately owned mill processing FFBs. x Each project will have a nursery, that will provide seedlings to the farming community that supplies FFB in return to the mill. x Outgrowers will be supported in the set-up of associations, management of members, planning, business advice and credit management. Together with the mill they should be given the opportunity to form a company to legalize contracting arrangements and initiate shareholder options for farmers. x Land will be owned or leased by the company and not by the individual farmers, so that the fruits from the land will be guaranteed for the mill, even if individual farmers decide to step out. x The company will allocate the land to the smallholders, provide them with seedlings and other inputs, and let them produce and sell their palm fruits to the mill. If a farmer decides to step out, he/she will forfeit his right to the land and the trees, and will be replaced by somebody else. Thus, the PSI supports the following activities: x organizing and training farmers, x establishing and registering the companies, x reallocation of land, x ensuring that the mills are fully operational, x ensuring that the nurseries are operational, x undertaking further studies on the best approach to implement the PSI more widely. (Source: Dako, P.K. (2008). FAO Support to the President’s Special Initiative on Oil Palm – Quality Management Consultancy Report). 124 Chapter 7 References AngloGold Ashanti (2007). Report to Society 2007 Case Studies – Regional Health Threats, Obuasi malaria control programme: a model for Africa. Basiron, Y. (2002). Palm Oil and Its Global Supply and Demand Prospects. Oil Palm Industry Economic Journal, 2 (1). Bates, B.C., Kundzewicz, Z.W., Wu, S., Palutikof, J.P. (Eds.) (2008). Climate Change and Water. Technical Paper of the Intergovernmental Panel on Climate Change, IPCC Secretariat, Geneva. Chen, Z., Grasby, S., Osadetz, K. (2004). Relation between climate variability and groundwater levels in the upper carbonate aquifer, southern Manitoba, Canada. Journal of Hydrology, 290 pp 43–62. Döll, P., Flörke, M. (2005). Global-scale estimation of diffuse groundwater recharge. Frankfurt Hydrology Paper 03. Institute of Physical Geography, Frankfurt University. Dako, P.K. (2008). FAO Support to the President’s Special Initiative on Oil Palm – Quality Management Consultancy Report. Ghana Districts website www.ghanadistricts.com Date accessed 19/05/2009 Ghana Environmental Protection Agency (2008). Ghana Climate Change Impacts, Vulnerability and Adaptation Assessments (funded under The Netherlands Climate Assistance Program). Ghana Statistical Service (2008). Ghana Living Standards Survey: Report of the Fifth Round (GLSS 5). GOPDC. (2005).The Okumaning Oil Palm Plantation Development Programme: Social Impact Assessment. Government of Ghana. (2003). Ghana Poverty Reduction Strategy 2003–2005: An Agenda for Growth and Prosperity. IPCC. (2001). Technical Summary, Climate Change 2001: Impacts, Adaptation and Vulnerability, A Report of Working Group II of the Intergovernmental Panel on Climate Change. IPCC. (2007). Climate Change 2007: Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Kilabuko, J.H., Nakai, S. (2007). Effects of Cooking Fuels on Acute Respiratory Infections in Children in Tanzania, International Journal of Environmental Research and Public Health, 4(4) pp 283–288. McSweeney, C., New, M., Lizcano, G. (2008). UNDP Climate Change Country Profiles – Ghana. Netherlands Climate Assistance Program (NCAP). (2006). Ghana Workplan. UNDP. (2008). Human Development Report 2007/8. Fighting climate change: Human solidarity in a divided world. Upton, M. (2009). World Malaria Day: Private Sector Must Do Its Share in Combating Infectious Disease. Wahid M.B. et al., (2007). Recent Development in the World Palm Oil Prices: An Overview, Oil Palm Industry Economic Journal, 7(2). 125 West Africa Centre for Crop Improvement website http://www.wacci.edu.gh/ Date accessed 10/06/09. Additional sources of information Business Action for Africa www.baa.org Date accessed 19/05/2009 GOPDC. (2005). Environmental Management Plan. GOPDC. (2007). Updated Resettlement Action Plan for the Oil Palm Plantation Development Project at Okumaning International Finance Corporation. (2007). Environmental and Social Review Summary 25988. The Open Society Institute for Southern Africa www.osisa.org Date accessed 19/05/2009 World Health Organization (WHO). (2009). Climate Change and Human Health – risks and responses. Health impacts of climate extremes. www.who.int Date accessed 10/06/2009. 126 Chapter 8: Climate risk analysis for malaria at GOPDC 127 Overview Malaria, one of the major causes of mortality and morbidity in Africa, is hyper-endemic in Ghana. 21 Plasmodium falciparum malaria is endemic across the whole country (Figure 49). Figure 49: The spatial distribution of Plasmodium falciparum malaria endemicity in Ghana Source: Hays et al., 2009 Malaria affects both adults and children and contributes to health problems and death in various ways: it causes frequent acute infections, anemia as a result of repeated or chronic malaria infection and increased susceptability to other diseases such as respiratory infections and diarrhea. In Ghana, malaria accounts for a significant portion of the disease burden, causing about 10.6% of lost Disability Adjusted Life Years and costing an equivalent of about 3% of GDP annually in economic burden (National Development Planning Commission, 2005). Across Ghana, there was no evidence of a reduction in malaria cases between 2001 and 2007, and reported deaths increased in 2007 (WHO, 2008). Companies operating in locations with malaria are affected by its impacts on their workers and their workers’ dependents: if a worker or his dependent is infected, he may have to remain absent from work. The total number of reported cases of malaria at GOPDC’s clinic over the period 2005 to 2009 is shown in Table 25. 21 Of malaria infections detected by blood slide examination, Plasmodium falciparum accounts for about 90% of malaria cases in Ghana (WHO, 2005). 128 Table 25: Total number of malaria cases reported at GOPDC’s clinic, 2005–2009 Month 2005 2006 2007 2008 2009 January 159 181 224 221 159 February 142 113 180 178 116 March 136 109 158 135 149 April 105 128 159 177 172 May Not 118 350 202 171 available June 153 203 329 296 190 July 181 232 216 282 274 August 205 257 292 182 184 September 109 205 277 161 134 October 148 273 322 217 138 November 223 285 294 211 126 December 202 251 190 163 175 Total 1,763 2,355 2,991 2,425 1,988 In 2007, 56% of worker attendance at the GOPDC clinic was due to malaria (Figure 50). While malaria is clearly the most common ailment reported, the GOPDC clinic Medical Assistant suggests this proportion will be an overestimate. This is because malaria diagnoses are currently undertaken without laboratory testing, and occasionally cases of stress are misdiagnosed as malaria, because they can have similar symptoms (A.B. Arthur, Pers. Comm.). Figure 50: Breakdown of reported workers’ diseases at GOPDC clinic, 2007 129 The clinic’s records show that around 30% of malaria cases reported in 2007 were attributable to workers (Figure 51), the remainder being their dependents and private patients. Malaria transmission involves a complex interaction between Plasmodium parasites, mosquito vectors and humans. In sub-Saharan Africa, climate has been found to play an important role in influencing the spatial distribution, intensity of transmission and seasonality of malaria (IPCC, 2007). In this chapter, we explore the relationship between climatic variables and malaria, using a Pearson’s correlation analysis to examine the empirical relationship between these variables in the GOPDC context. We also investigate the current cost of malaria for GOPDC, the implications of climate change on malaria incidence, and possible adaptation measures. Figure 51: Total reported cases of malaria at GOPDC clinic vs. cases attributable to GOPDC workers, 2007 (Note: Data are not available on malaria in workers for April and December) 8.1 Climate and malaria Climatic drivers of malaria 22 Climate variability and its impact on the breeding activity of Anopheles is considered one of the key environmental contributors to malaria transmission (McMichael & Martens, 1995). Climate impact on parasite development also has an effect on rates of malaria transmission. Rainfall affects malaria incidence in terms of the total amount of rainfall and the number of rainy days/degree of wetness after a rainfall event (Bhattacharya et al., 2006). Mosquitoes breed in standing water (usually freshwater pools or marshes) and therefore mosquito abundance is affected by rainfall and the availability of surface water (van Lieshout et al., 2004). While an increase in rainfall can have a positive effect on mosquito breeding and density, a decrease in rainfall can limit mosquito populations and reduce rates of malaria transmission. Interestingly, although malaria cases usually occur after periods of 22 Anopheles is a genus of mosquito. Human malaria is transmitted only by females of the genus Anopheles which are infected with a Plasmodium parasite. 130 heavy rainfall, excessive rainfall does not always trigger an epidemic (Zulueta et al., 1980). On the contrary, a negative correlation was observed between rainfall and malaria incidence in a nine year study on the Colombian Pacific coast (Gonzalez et al., 1997). Additionally, in a study of the Amazon Basin, both positive and negative correlations between rainfall and malaria incidence were found, with negative associations found in areas dominated by wetland and large rivers. This is understood to be linked to mosquito habitats being “wash(ed) out or becom(ing) too deep during months with high precipitation” (Olson et al., 2009, p660). This finding ties in with experiences at GOPDC where heavy rainfall has been noted to flush out breeding sites and thus lead to a reduction in malaria incidence (Emmanuel Wiafe & Anke Massart, Pers. Comm.). Relative humidity also affects malaria transmission through its effect on the longevity of the mosquito population. An increase in relative humidity has been recorded to lead to an increase in the number of mosquitoes biting an individual per unit time, the “human biting rate” (van der Hoek et al., 1997). However, if the average monthly relative humidity is below 55% and above 80% the scope of malaria transmission diminishes due to a shortening of the life span of the mosquito. Temperature has also been found to be associated with malaria: a number of studies have reported associations between interannual variability in temperature and malaria transmission. In highland areas of Kenya, malaria admissions have been associated with unusually high maximum temperatures 3–4 months previously, as well as rainfall (Githeko & Ndegwa, 2001). Additionally, an analysis of malaria morbidity data from Ethiopia found that epidemics were associated with high minimum temperatures in the preceding months (Abeku et al., 2003). GOPDC also reports that malaria cases increase when temperatures are higher (Emmanuel Wiafe, Pers. Comm.). Despite these observations, recent research questions the validity of widespread claims that rising temperatures have already led to increases in worldwide malaria morbidity and mortality. In fact, according to this research, economic development, greater funding for malaria control and the efficacy of affordable treatment have led to declines in the range and intensity of malaria over the last hundred years or so, even though temperatures have been rising (Gething et al., 2010). The link between temperature and malaria is often attributed to the growth rate of the vector population (i.e. mosquitoes) being dependent on temperature. Higher temperatures shorten the mosquito generation time, and thus may result in higher vector densities that increase the likelihood of transmission. Temperature also affects the development of the parasite in the mosquito vector. The duration of “sporogony”, the time required to complete the sexual stage of the parasite in the mosquito, is inversely related to ambient temperature. (Bouma et al., 1996). In forecasting a malaria epidemic, Onori and Grab (1980) identified temperature and humidity as the most significant factors. For most Anopheles vector species of malaria, the optimal temperature range for their development lies ° ° between 20 C and 30 C (McMichael & Martens, 1995). Also, the minimum temperature required for the transmission of Plasmodium falciparum, the parasite species that causes malaria in most parts of Ghana, ° is 19 C (WHO, 2005; Bhattacharya et al., 2006). Climate and malaria cases at St. Dominic’s Hospital, Akwatia Cases of malaria are reported throughout the year at St. Dominic’s Hospital, Akwatia (the nearest hospital to GOPDC). To examine the conditions contributing to malaria transmission in the region around GOPDC, monthly climatic data from Akim Oda Met Station on mean rainfall, number of rainy days, mean nightly (06.00h) and daily (15.00h) relative humidity, and maximum, minimum and mean temperature for the period 2004 to 2007 were analyzed against the monthly malaria cases obtained from St. Dominic’s Hospital for the same period. 131 As shown in Figure 52, there are two rainy seasons, in April–June and September–October, which one might expect to be positively related to malaria incidence. Mean monthly relative humidity varies between 53% and 74% during the day (15:00h) and 93% and 95% during the night (06:00h), thus exceeding optimal levels for malaria transmission during the night. Mean monthly temperatures averaged across the ° ° period 2004–2007 ranged from 25.6 C to 29.0 C, falling within the optimal temperature range for ° Anopheles development. Monthly maximum temperatures reach 34.4 C, slightly exceeding this range, and ° monthly minimum temperatures remain within the optimal range, at 21.4 C. Anecdotal evidence also ° indicates that temperature can drop to 16 C in the mornings (Emmanuel Wiafe & Anke Massart, Pers. Comm.), though we do not know whether these short, periodic lows affect mosquito development. (Further information on climatic data from Akim Oda can be found in Chapter 1). Figure 52: Time series of monthly total rainfall (mm) recorded at Akim Oda (2004–2007) At St. Dominic’s Hospital, the annual number of malaria cases ranged between 8,217 and 13,855 in 2004 23 to 2007 (Table 26) . Table 26: Total number of malaria cases reported at St Dominic’s Hospital, 2004–2007 Year Total number of malaria cases 2004 9,196 2005 8,217 2006 13,855 2007 13,289 The distribution of malaria incidence recorded at St. Dominic’s Hospital across these years is shown in Figure 53. Although malaria cases are prevalent throughout the year, it can be seen that a greater percentage of malaria cases was found between May and July and November and December, coinciding with rainy and post-rainy seasons. 23 Note that Table 26 is not intended to indicate a change in malaria incidence recorded over time. While it shows how many malaria cases were recorded per year, the four year timescale is not long enough to observe clear trends. 132 Figure 53: Average percentage of malaria cases in each month, 2004–2007, St. Dominic’s Hospital Correlation between climatic variables and malaria incidence recorded at St. Dominic’s Hospital A Pearson’s correlation analysis was conducted, relating monthly incidence (%) of malaria cases at St. Dominic’s Hospital and monthly climatic conditions for the period 2004–2007 at Akim Oda. Different time lags between incidence of malaria cases and climatic conditions were examined, to allow for the complex interactions between parasites, mosquitoes and humans which lead to malaria transmission. Malaria is only transmitted once mosquitoes have developed, become infected with a Plasmodium parasite, the parasite has developed and later been injected into a human. Estimates for the duration of the lag period between mosquito development and malaria transmission include 15 days for the pre-imaginal 24 development of the vector Anopheles, 4 to 7 days for the gonadotrophic cycle for female mosquitoes and 12 days for the sporogonic cycle for the Plasmodium falciparum parasites in the vector mosquitoes (Devi & Jauhari, 2006). Effectively, around 30 days are needed for the development of a new generation of infective female insect vectors, after which a human must be bitten before a case of malaria occurs. Results The highest correlation coefficients were found between the monthly incidence (%) of malaria cases and the number of rainy days two months before malaria cases were reported (r=0.79, p<0.01; Figures Figure 54, Figure 55) and total rainfall two months before cases were reported (r=0.74, p<0.01; Figures Figure 56, Figure 57). GOPDC agrees that this supports its experience and understanding of climate and malaria incidence (Emmanuel Wiafe & Anke Massart, Pers. Comm.). High correlation coefficients were also found between incidence of malaria and the number of rainy days one month before, and total rainfall one month before cases were reported (r=0.71, p<0.05; r=0.70, p<0.05, respectively). 24 One complete round of ovarian development in the insect vector from the time when the blood meal is taken to the time when the fully developed eggs are laid. 133 Figure 54: Correlation between malaria cases (%) and number of rainy days (two months lagged), 2004–2007, St. Dominic’s Hospital Figure 55: Trends in monthly incidence of malaria cases (%) and number of rainy days with two month lag, 2004–2007, St. Dominic’s Hospital 134 Figure 56: Correlation between malaria cases (%) and total monthly rainfall (two months lagged), 2004–2007, St. Dominic’s Hospital Figure 57: Trends in monthly incidence (%) of malaria cases and rainfall with two month lag, 2004– 2007, St. Dominic’s Hospital Relatively high correlation coefficients were also found between the incidence of malaria cases and daily (15:00h), nightly (06:00h) and mean relative humidity with a one month lag (see Annex E). Low correlations were found with minimum, maximum and mean temperature (both lagged and not-lagged), and with rainfall and rainy days with no lag period (see Annex E). These findings indicate that the antecedent number of rainy days and antecedent total rainfall have the strongest correlations with malaria incidence in the area around GOPDC. 135 8.2 Economic impacts of malaria for GOPDC As shown earlier, malaria affects workers at GOPDC (Figures Figure 50, Figure 51), and consequently has a negative economic impact on the company. It is the leading cause of workdays lost due to illness. Malaria can affect GOPDC economically in three ways: x First, a worker who takes sick days as a result of malaria infection may be entitled to paid sick leave, x Secondly, a worker who requires healthcare as a result of infection may have his/her healthcare costs covered by GOPDC, and the costs of treating malaria are GHC 4–5 (approximately $0.50), if there are no further medical complications. x Thirdly, any worker who is unable to work (because either he/she or his/her dependents are 25 infected with malaria ) will have an economic impact on GOPDC through revenues not earned. This is because these workers are not replaced (Emmanuel Wiafe & Anke Massart, Pers. Comm.). To address the first financial impact, the cost to GOPDC of paying for sick leave, it is important to distinguish between different types of worker. Permanent workers can claim up to three months of continuous paid sick leave (equal to his/her monthly salary), one-year contract workers can claim a “certain” number of sick days (receiving their daily salary) with a doctor’s permission and six-month contract workers can only claim this sick leave with a doctor’s permission and if their disease/injury is work-related (Emmanuel Wiafe & Anke Massart, Pers. Comm.). Thus, GOPDC is affected economically because it is obliged to pay a worker’s salary without receiving any work in return. This cost could be calculated, given the average salary at GOPDC (taking into account these different types of worker) and depending on whether malaria is considered a work-related disease. To address the second financial impact, the cost of healthcare to GOPDC, again it is important to distinguish between different types of worker. Both permanent workers and one-year contract workers can attend the clinic and receive free medication for any disease or injury. Six-month contract workers, on the other hand, must pay GHC 6–10 to attend the clinic and pay for their medicine, unless their disease or injury is work-related (Emmanuel Wiafe & Anke Massart, Pers. Comm.). For the purposes of this assessment, since it is not known how many GOPDC workers affected by malaria were on six- month contracts, it has been assumed that GOPDC pays the GHC 4–5 (US$0.5) treatment costs for all affected workers. Here we also partially assess the third impact – revenues not earned – as a result of GOPDC workers at Kwae and Okumaning taking sick leave due to having malaria. Note that we have not considered the impact of smallholder and outgrower malaria cases, as their contribution to revenue is different from that of nucleus estate workers and we do not have data on their work days lost to malaria. In order to calculate revenues not earned, we have attributed a financial value to a man-day (based on revenue derived from oil palm production), and combined this with a calculation of the number of man- days missed as a result of malaria. 25 While it is possible that workers are absent in order to care for sick dependents, GOPDC comments that it is unusual that they do this. GOPDC does note however that it is difficult to keep track of the reasons for absenteeism and that if in future this was useful, they could keep appropriate records (Emmanuel Wiafe & Anke Massart, Pers. Comm.). 136 First, the revenue associated with a man-day was calculated by dividing the 2007 sales resulting from GOPDC own production (i.e. 38.5% of the total) by the number of man-days worked (Table 27). Table 27: Data used for estimating revenues not earned due to malaria, 2007 CPO own (b) Total sales from production Total sales own production, (10,667 tons)/ ($x000) smallholders CPO total attributable to Total number of (c) and outgrowers production GOPDC own man-days Value of a man- (a) (d) ($x000) (27,682 tons) (%) production worked day ($) 23,017 38.5 8,862 218,062 40.64 Sources: (a) GOPDC Financial Model – P&S$ worksheet – cell I316 (b) GOPDC Financial Model – P&S$ worksheet – cell I64 (c) GOPDC Financial Model – P&S$ worksheet – cell I66 (d) GOPDC Agric Report 2007 – OTHERS_ Statement worksheet Secondly, the number of man-days missed as a result of malaria was calculated using GOPDC Clinic 26 Quarterly reports . The reports indicate that the average malaria episode lasts approximately three to five days – a duration over which most patients cease to work. Using the knowledge that in 2007, 739 cases of 27 malaria were recorded for GOPDC workers , Table 28 presents estimates of man-days lost (providing both a low estimate and high estimate using the assumptions of 3 days lost per episode and 5 days lost per episode, respectively). Table 28 also presents estimates of revenues not earned in 2007 associated with these man-days lost to malaria. Note that revenues not earned range from approximately $90,000 to $150,000. The cost of treating malaria in these workers (approximately $0.50 per person, unless there are complications) adds a modest $370 onto these figures. Referring back to Table 25 indicates that the total number of malaria cases in 2007 for all attendees at the clinic (i.e. workers and non-workers) were higher than the average over the period 2005–2009. For 2005, the year with the lowest number of reported cases (1,763; see Table 25) and assuming the same proportion of worker cases (i.e. about 30%) and the same value of a man-day, revenues not earned in 2005 would have been in the range $50,000 to $90,000. Table 28: Estimates of revenues not earned resulting from man-days lost to malaria for GOPDC own production, 2007 Low estimate (3 days leave High estimate (5 days leave per malaria case) per malaria case) Man-days lost based on 739 2,217 3,695 malaria cases in 2007 Revenues not earned ($) 90,099 150,165 26 GOPDC Clinic – Quarterly Reports – 2003-2005 27 GOPDC Clinic Monthly Report, 2007 – data missing for April & December 137 Note: Revenues not earned were calculated assuming that all man-days have an equal impact on oil palm product production and that loss of a man-day has a direct impact on production (e.g. FFBs not harvested). We acknowledge that while a loss of man-days may lead to production losses (Emmanuel Wiafe & Anke Massart, Pers. Comm.), they may not always, due to the different nature of workers’ jobs. Furthermore, different workers generate different revenues: if a senior staff member is off sick, this affects the efficiency of other workers. Similarly, man-days lost in the peak harvest season are more costly than in the lean season (GOPDC, Pers. Comm.). We also acknowledge that the calculations of revenues not earned are limited, by only including nucleus estate worker absenteeism. However we also note that GOPDC considers the loss of smallholder and outgrower man-days to malaria to have less financial impact (e.g. because they only harvest FFB every two weeks and can alter harvest time to suit their illness; Emmanuel Wiafe & Anke Massert, Pers. Comm.). 8.3 Impacts of climate change The UNDP climate change projections for Ghana show a wide range of change in rainfall from different climate models (McSweeney et al., 2008). Over the area of GOPDC’s plantations, the median trend for Jan/Feb/Mar and Apr/May/Jun rainfall is a zero change, and for Jul/Aug/Sep and Oct/Nov/Dec the median trend is for a slight increase. The maximum and minimum model outputs show significantly larger changes – ranging from decreases of 19mm/month to increases of 33 mm/month in Jul/Aug/Sep, and decreases of 17mm/month to increases of 38mm/month for Apr/May/Jun (see Chapter 1, Climate). The largest projected changes in rainfall indicate that climate change could lead to average rainfall decreases or increases of about 10–15% at GOPDC plantations by the 2030s, with the years between now and the 2030s seeing extreme low and high rainfall occurring increasingly often. The UNDP climate projections also indicate that the proportion of total annual rainfall that falls in “heavy” events tends to increase in the future. Seasonally, this varies between tendencies to decrease in Jan/Feb/Mar and to increase in Jul/Aug/Sep and Oct/Nov/Dec. Projected changes in 1-day and 5-day rainfall maxima also tend towards increases. This may lead to more standing water and thus we speculate that conditions favorable to malaria may be more common. The UNDP projections do not provide data on future changes in relative humidity. For temperature, the projections indicate increases of 1.2°C (low to high range of 0.8–1.5°C) by the 2030s and 2.3°C (1.7– 2.7°C) in tKHV:LWKPHDQGDLO\WHPSHUDWXUHVLQ$NLP2GDDOUHDG\UHDFKLQJÛ&DQGPD[LPXP GDLO\WHPSHUDWXUHVUHDFKLQJÛ& VHHChapter 1, Climate), these changes may limit malaria transmission, as they increase temperature above the optimal temperature range for Anopheles development (which is 20ÛC to 30ÛC). Nevertheless, increases in daily minimum temperature might be beneficial for malaria transmission. Malaria is a complex disease and all published malaria models have limited parameterization of some of the key factors that influence intensity of malaria transmission (Reiter et al., 2004; IPCC, 2007). Furthermore, due to the lack of climate model agreement on the direction of future change in rainfall in Ghana, it is not clear whether or not future climate conditions will be more or less favorable for malaria transmission. 8.4 Existing and proposed adaptation actions GOPDC undertook a malaria spraying exercise at the Kwae Estate and in surrounding villages in 2007. However, spraying is only effective for 2 to 3 months and it is understood that it has not been undertaken since. The data in Table 25 highlight that malaria cases in 2007 and 2008 were the highest over the period 2005–2009, which suggests that the spraying did not have a large impact. Weeding and cleaning of the estate is, however, done every quarter, and helps to keep mosquito numbers down. 138 GOPDC intends soon to have a laboratory at the GOPDC clinic for the diagnosis of malaria. More accurate diagnosis should reduce the costs of malaria treatment. It would also help GOPDC to undertake more accurate analyses correlating malaria incidence with rainfall (Emmanuel Wiafe and Anke Massart, Pers. Comm.). The company will also be providing treated bednets to pregnant women in Kwae and Anweam. The GOPDC clinic has made efforts to raise awareness of how to prevent malaria transmission, and recognizes the need for collective action: “It is of no use keeping ones surrounding clean while that of the neighbour is a breeding place for Mosquitoes – The Mosquito knows no barrier.”—“Mosquito News”, Dr Osafo Yao Sei, GOPDC Clinic Under conditions of uncertainty about the impacts of climate change, decision-makers should seek to identify “no-regrets” adaptation measures. These are measures that have benefits that exceed their costs, whatever the extent of climate change (Willows & Connell, 2003). Given that malaria costs GOPDC around $50,000 to $150,000 per year, potential “no-regrets” adaptation measures to reduce the incidence of malaria include: x Stepping up actions to control malaria, for instance increasing spraying in the area around GOPDC operations. As noted above, spraying was last done in 2007, and did not appear to have had a dramatic effect on the number of malaria cases. x Working in partnership with local, regional and/or national authorities implementing malaria control 28 programs – for instance, the WHO/UNDP/World Bank/UNICEF Roll Back Malaria Initiative . Partnership working between the private and public sectors to tackle malaria has proved effective in Mozambique (see Box 5). GOPDC is already part of the Global Business Coalition, which covers malaria, HIV/AIDS and tuberculosis, and which encourages innovative financing and coordination between business and donor agencies. x Connecting with malaria early warning systems in the region. Box 5: Case study—BHP Billiton fighting malaria with communities and governments BHP Billiton, which supplies natural resources including fossil fuels, metals and diamonds, recognized that the success of its Mozal aluminum smelter project in southern Mozambique was threatened by endemic malaria. Through its impacts on workers and their families, the disease was leading to absenteeism and low staff morale, and affecting the facility’s productivity. It also meant Mozal was a less attractive destination for skilled employees. BHP Billiton became a partner in and funder of the Lubombo Spatial Development Initiative (LSDI), a program set up by the governments of Mozambique, Swaziland and South Africa to develop the Lubombo region. The company realized that limiting malaria programs to the Mozal site itself would have little effect, given the prevalence of the disease. Working with the LSDI team, BHP Billiton implemented a malaria control program in the area around its Mozal operations. The program involved spraying buildings and homes in the area and controlling mosquito 28 www.rollbackmalaria.org 139 breeding sites, as well as establishing a malaria laboratory for early diagnosis and treatment, distributing mosquito nets and raising awareness among the local community. In the six years since the program began, absenteeism from school due to the disease was reduced from 20% to 1%, the number of recorded cases of malaria was reduced by more than 70%, and fatalities were reduced by 97%. In 1999, 85% of children who lived near the smelter were infected. By June 2005, this figure had fallen to about 20%. Source: BHP Billiton Community Programs. (2006). Yesterday Today Tomorrow. BHP Billiton Chapter 8 References Abeku, T., van Oortmarssen, G., Borsboom, G., de Vlas, S., Habbema, J. (2003). Spatial and temporal variations of malaria epidemic risk in Ethiopia: factors involved and implications. Acta Trop., 87, pp 331–340. Bhattacharya, S., Sharma, C., Dhiman, R. C., Mitra, A. (2006). Climate change and malaria in India. Current Science, 90(3). BHP Billiton Community Programs. (2006). Yesterday Today Tomorrow. BHP Billiton. Bouma, M.J., Dye, C., van der Kaay, H. (1996). Falciparum Malaria and Climate Change in the Northwest Frontier Province of Pakistan. Am. J. Trop. Med. Hyg., 55(2), pp 131–137. Devi, N.P., Jauhari, R.K. (2006). Climatic variables and malaria incidence in Dehradun, Uttaranchal, India. J Vect Borne Dis, 43, pp 21–28. Gething, P.W., Smith, D.L., Patil, A.P., Tatem, A.J., Snow, R.W. and Hay, S.I. (2010). Climate change and the global malaria recession. Nature, 465, pp 342–345. Githeko, A.K., Ndegwa, W. (2001). Predicting malaria epidemics in the Kenyan Highlands using climate data: a tool for decision makers. Global Change Human Health, 2, pp 54–63. Gonzalez, J.M., Olano, V., Vergara, J., Arevalo-Herrera, M., Carrasquilla, G., Herrera, S., Lopez, J.A. (1997). Unstable, low-level transmission of malaria on the Colombian Pacific coast. Ann Trop Med Parasitol, 91(4), pp 349–358. Hay, S.I. et al. (2009). A world malaria map: Plasmodium falciparum endemicity in 2007. PLoS Medicine 6(3), e1000048. IPCC. (2007). Climate Change 2007. Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, M.L. Parry, O.F. Canziani, J.P. Palutikof, P.J. van der Linden and C.E. Hanson, Eds., Cambridge University Press, Cambridge, UK. Leighton C., Foster R. (1993). Economic Impacts of Malaria in Kenya and Nigeria. Major Applied Research Paper No. 6, Health Financing and Sustainability (HFS) Project, Maryland, USA. McMichael, A.J., Martens, W.J.M. (1995). The health impact of global climate changes: grappling with scenarios, predictive models and multiple uncertainties. 140 McSweeney, C., New, M., Lizcano, G. (2008). UNDP Climate Change Country Profiles – Ghana. National Development Planning Commission. (2005). Growth and Poverty Reduction Strategy (GPRS II), 2006–2009. Republic of Ghana. Olson, S. H., Gangnon, R., Elguero, E., Durieux, L., Guégan, J-F, Foley, J. A. And Patz, J. A. (2009). Links between Climate, Malaria, and Wetlands in the Amazon Basin. Emerging Infectious Diseases, 15(4), pp 659–662. Onori, E., Grab B. (1980). Indicators for the forecasting of malaria epidemics. Bulletin of the World Health Organisation, 58(1), pp 91–98. Reiter, P., Thomas, C.J., Atkinson, P., Randolph, S.E., Rogers, D.J., Shanks, G.D., Snow R.W., and Spielman, A. (2004). Global warming and malaria: a call for accuracy. The Lancet Infectious Diseases, 4(6), pp 323–324. van Lieshout, M., Kovats, R.S., Livermore, M.T.J., Martens, P. (2004). Climate change and malaria: analysis of the SRES climate and socio-economic scenarios. Global Environmental Change 14, pp 87–99. van der Hoek, W., Konradsen, F., Perera, D., Amerasinghe, P.H., Amerasinghe, F.P. (1997). Correlation between rainfall and malaria in the dry zone of Sri Lanka. Annals of tropical medicine and parasitology, 91(8), pp 945–949. Willows, R.I. and Connell, R.K. (Eds.) (2003). Climate adaptation: Risk, uncertainty and decision-making. UKCIP Technical Report. UKCIP, Oxford, UK. World Health Organisation. (2005). Roll Back Malaria Monitoring and Evaluation Report, Ghana Profile, WHO Press, Switzerland. World Health Organisation. (2008). World Malaria Report, WHO Press, Switzerland. Zulueta, J.D.E., Mujtaba, S.M., Shah, I.H. (1980). Malaria control and long-term periodicity of the disease in Pakistan. Trans R Soc Trop Med Hyg, 74, pp 624–32. 141 Annexes Annex A: Greenhouse gas emissions scenarios To provide a basis for estimating future climate change, the Intergovernmental Panel on Climate Change (IPCC) prepared the Special Report on Emissions Scenarios, detailing 40 greenhouse gas and sulphate aerosol emission scenarios that combine a variety of assumptions about demographic, economic and WHFKQRORJLFDOIDFWRUVOLNHO\WRLQIOXHQFHIXWXUHHPLVVLRQV 1DNLüHQRYLüDQG6ZDUW (DFKVFHQDULR represents a variation within one of four “storylines”: A1, A2, B1 and B2. Further details of the storylines are provided in Box A1. Projected carbon dioxide, methane, nitrous oxide and sulphate aerosol emissions based on these scenarios are shown in Figure A1 below for six “marker scenarios”. All the scenarios are considered equally sound by the IPCC and no probabilities are attached. The climate maps shown in Figure 14 above are based on the A2 emissions scenario, which is commonly termed a “medium-high” emissions scenario. Higher and lower scenarios are also possible, depending on the course of international action to tackle greenhouse gas emissions. At present, emissions are rising fast, at or above the upper end of the scenarios shown in Figure A1. Because of the long lags in the Earth’s climate system, unless strong action is taken in the next decade or two, the extent of climate change over the remainder of the century and beyond will become increasingly severe. Box A1: Storylines for scenarios of greenhouse gas emissions A1: The A1 storyline describes a future world of very rapid economic growth, a global population that peaks in mid-century and declines thereafter, and the rapid introduction of new and more efficient technologies. Major underlying themes are convergence among regions, capacity building and increased cultural and social interactions, with a substantial reduction in regional differences in per capita income. The A1 storyline develops into three scenario groups that describe alternative directions of technological change in the energy system. They are distinguished by their technological emphasis: fossil intensive (A1FI), non-fossil energy sources and technologies (A1T), or a balance across all sources (A1B) (where balanced is defined as not relying too heavily on one particular energy source, on the assumption that similar improvement rates apply to all energy supply and end use technologies). A2: The A2 storyline describes a very heterogeneous world. The underlying theme is self reliance and preservation of local identities. Fertility patterns across regions converge very slowly, which results in continuously increasing population. Economic development is primarily regionally oriented and per capita economic growth and technological change more fragmented and slower than other storylines. B1: The B1 storyline describes a convergent world with the same global population as in the A1 storyline (one that peaks in mid-century and declines thereafter) but with rapid change in economic structures toward a service and information economy, with reductions in material intensity and the introduction of clean and resource efficient technologies. The emphasis is on global solutions to economic, social and environmental sustainability, including improved equity, but without additional climate initiatives, i.e. it does not include implementation of the United Nations Framework Convention on Climate Change or the Kyoto Protocol. B2: The B2 storyline describes a world in which the emphasis is on local solutions to economic, social and environmental sustainability. It is a world with continuously increasing global population, at a rate lower than A2, intermediate levels of economic development, and 142 less rapid and more diverse technological change than in B1 and A1. While the scenario is also oriented towards environmental protection and social equity, it focuses on local and regional levels. Figure A1: Man-made emissions of carbon dioxide (CO2), methane (CH4), nitrous oxide (N2O) and sulphur dioxide (SO2 ) for six SRES scenarios (see Box A1) and the IS92a scenario from the IPCC Second Assessment Report in 1996 for comparison 143 Annex B: Results of statistical analyses of FFB yields and climate variables Table A1: FFB yield data correlated with temperature variables (annual averages) TMax TMin TMean Year 0 1 2 3 0 1 2 3 0 1 2 3 * * ** Planted Pearson .558 .719 .014 .506 .491 .528 .438 .027 .650 .752 .336 .299 1979 Correlation Sig. (2- .074 .013 .967 .112 .125 .095 .178 .936 .030 .008 .312 .373 tailed) Sum of 7.066 8.607 .160 4.037 4.642 5.327 4.384 .192 6.259 7.523 3.379 2.031 Squares and Cross- products Covariance .707 .861 .016 .404 .464 .533 .438 .019 .626 .752 .338 .203 N 11 11 11 11 11 11 11 11 11 11 11 11 * * ** * Planted Pearson .713 .626 .170 .478 .595 .476 .541 .335 .788 .714 .439 .536 1980 Correlation Sig. (2- .014 .040 .617 .137 .053 .139 .086 .314 .004 .014 .177 .089 tailed) Sum of 8.361 7.554 1.942 5.140 5.421 4.299 5.213 3.199 7.245 6.563 4.189 5.142 Squares and Cross- products Covariance .836 .755 .194 .514 .542 .430 .521 .320 .724 .656 .419 .514 N 11 11 11 11 11 11 11 11 11 11 11 11 * Planted Pearson .360 .308 -.239 .349 .743 .303 .194 .060 .618 .430 .006 .221 1981 Correlation Sig. (2- .308 .386 .505 .323 .014 .394 .591 .869 .057 .215 .986 .539 tailed) 144 TMax TMin TMean Year 0 1 2 3 0 1 2 3 0 1 2 3 Sum of 2.562 2.523 -1.835 2.633 4.765 1.925 1.232 .405 3.717 2.723 .040 1.462 Squares and Cross- products Covariance .285 .280 -.204 .293 .529 .214 .137 .045 .413 .303 .004 .162 N 10 10 10 10 10 10 10 10 10 10 10 10 Planted Pearson .387 .303 -.381 .393 .517 .348 .245 .327 .501 .374 -.041 .407 1982 Correlation Sig. (2- .270 .395 .278 .262 .126 .325 .494 .357 .140 .287 .911 .243 tailed) Sum of 3.489 2.798 -4.036 3.901 4.075 2.891 2.018 2.687 3.709 2.915 -.335 3.305 Squares and Cross- products Covariance .388 .311 -.448 .433 .453 .321 .224 .299 .412 .324 -.037 .367 N 10 10 10 10 10 10 10 10 10 10 10 10 Planted Pearson .181 .093 -.400 .019 .385 .500 .200 .115 .550 .470 -.175 .108 1985 Correlation Sig. (2- .595 .785 .223 .955 .243 .117 .556 .736 .080 .145 .607 .752 tailed) Sum of 1.325 .639 -2.985 .148 2.172 2.709 1.162 .763 2.201 1.900 -.925 .662 Squares and Cross- products Covariance .133 .064 -.299 .015 .217 .271 .116 .076 .220 .190 -.092 .066 N 11 11 11 11 11 11 11 11 11 11 11 11 * * * Planted Pearson .152 .591 .126 -.257 -.719 .187 -.360 -.144 -.748 .684 -.136 -.181 145 TMax TMin TMean Year 0 1 2 3 0 1 2 3 0 1 2 3 1988/ Correlation 1989 Sig. (2- .696 .094 .746 .505 .029 .629 .341 .712 .020 .042 .728 .641 tailed) Sum of .980 4.511 .816 -2.150 -4.613 1.370 -2.378 -.883 -1.404 2.956 -.587 -.947 Squares and Cross- products Covariance .123 .564 .102 -.269 -.577 .171 -.297 -.110 -.175 .370 -.073 -.118 N 9 9 9 9 9 9 9 9 9 9 9 9 ** * * Planted Pearson .211 .887 .603 .330 -.696 -.710 -.313 .131 -.749 -.003 .191 .435 OPRI Correlation 1989 Sig. (2- .617 .003 .113 .425 .055 .049 .451 .757 .033 .994 .650 .281 tailed) Sum of .640 2.759 1.831 .852 -1.689 -2.047 -.996 .387 -.610 -.002 .399 .909 Squares and Cross- products Covariance .091 .394 .262 .122 -.241 -.292 -.142 .055 -.087 .000 .057 .130 N 8 8 8 8 8 8 8 8 8 8 8 8 ** Planted Pearson -.901 -.192 -.012 -.310 .471 .458 .101 .232 -.024 .280 -.423 .000 1990 Correlation Sig. (2- .006 .680 .979 .499 .287 .301 .830 .616 .959 .543 .345 .999 tailed) Sum of -2.369 -.624 -.033 -.796 1.078 1.109 .244 .733 -.021 .173 -.285 -.001 Squares and Cross- products 146 TMax TMin TMean Year 0 1 2 3 0 1 2 3 0 1 2 3 Covariance -.395 -.104 -.005 -.133 .180 .185 .041 .122 -.004 .029 -.048 .000 N 7 7 7 7 7 7 7 7 7 7 7 7 ** * ** ** All Pearson .462 .641 .245 .461 .462 .495 .376 .237 .636 .719 .403 .419 Planted Correlation Years Average Sig. (2- .054 .004 .327 .054 .053 .037 .124 .344 .005 .001 .097 .083 tailed) Sum of 7.517 10.887 4.167 8.010 6.758 7.452 5.202 3.057 8.022 9.613 5.522 5.673 Squares and Cross- products Covariance .442 .640 .245 .471 .398 .438 .306 .180 .472 .565 .325 .334 N 18 18 18 18 18 18 18 18 18 18 18 18 ** Correlation is significant at the 0.01 level (2-tailed). * Correlation is significant at the 0.05 level (2-tailed). 147 Table A2: FFB yield data correlated with annual hydrological deficit (calculated by IHRO method) 10-Year Annual Cumulative Year 0 Year 1 Average 3 Year Planted Pearson Correlation -.136 .279 .061 .297 1979 Sig. (2-tailed) .691 .406 .859 .375 Sum of Squares and -621.373 1278.585 49.759 1486.075 Cross-products Covariance -62.137 127.858 4.976 148.607 N 11 11 11 11 Planted Pearson Correlation .014 .388 .376 .358 1980 Sig. (2-tailed) .968 .238 .255 .280 Sum of Squares and 62.295 1698.692 312.481 1709.720 Cross-products Covariance 6.229 169.869 31.248 170.972 N 11 11 11 11 * Planted Pearson Correlation -.069 .670 .417 .535 1981 Sig. (2-tailed) .850 .034 .231 .111 Sum of Squares and -209.010 1969.910 236.845 1801.485 Cross-products Covariance -23.223 218.879 26.316 200.165 N 10 10 10 10 * Planted Pearson Correlation .065 .571 .561 .660 1982 Sig. (2-tailed) .858 .085 .091 .038 Sum of Squares and 272.771 2238.760 412.971 2798.359 Cross-products Covariance 30.308 248.751 45.886 310.929 N 10 10 10 10 * Planted Pearson Correlation -.162 .655 .404 .427 1985 Sig. (2-tailed) .634 .029 .217 .190 Sum of Squares and -678.361 2791.029 212.135 1920.845 Cross-products Covariance -67.836 279.103 21.213 192.084 N 11 11 11 11 * Planted Pearson Correlation -.693 .474 -.410 -.081 1988/1989 Sig. (2-tailed) .038 .197 .313 .849 Sum of Squares and -3518.607 2402.680 -198.654 -369.094 148 10-Year Annual Cumulative Year 0 Year 1 Average 3 Year Cross-products Covariance -439.826 300.335 -28.379 -52.728 N 9 9 8 8 ** Planted Pearson Correlation -.220 -.162 -.690 -.895 OPRI 1989 Sig. (2-tailed) .601 .702 .086 .007 Sum of Squares and -471.499 -395.209 -111.793 -1385.773 Cross-products Covariance -67.357 -56.458 -18.632 -230.962 N 8 8 7 7 Planted Pearson Correlation .415 .407 .538 .285 1990 Sig. (2-tailed) .355 .365 .271 .585 Sum of Squares and 954.370 922.816 131.160 679.510 Cross-products Covariance 159.062 153.803 26.232 135.902 N 7 7 6 6 All Planted Pearson Correlation -.040 .387 .297 .290 Years Sig. (2-tailed) .874 .113 .247 .259 Average Sum of Squares and -313.167 2994.367 375.482 2361.659 Cross-products Covariance -18.422 176.139 23.468 147.604 N 18 18 17 17 ** Correlation is significant at the 0.01 level (2-tailed). * Correlation is significant at the 0.05 level (2-tailed). Table A3: FFB yield data correlated with annual average rainfall Rainfall Year 0 Year 1 Year 2 Year 3 Planted 1979 Pearson .089 -.314 -.232 -.342 Correlation Sig. (2-tailed) .794 .348 .493 .304 Sum of Squares 545.082 -1818.367 -1385.666 -2438.053 and Cross- products Covariance 54.508 -181.837 -138.567 -243.805 149 Rainfall Year 0 Year 1 Year 2 Year 3 N 11 11 11 11 Planted 1980 Pearson -.044 -.333 -.085 -.245 Correlation Sig. (2-tailed) .899 .317 .804 .468 Sum of Squares -281.768 -1941.605 -468.362 -1399.559 and Cross- products Covariance -28.177 -194.160 -46.836 -139.956 N 11 11 11 11 Planted 1981 Pearson .142 -.524 -.115 .321 Correlation Sig. (2-tailed) .695 .120 .752 .365 Sum of Squares 621.535 -2068.050 -447.465 1109.345 and Cross- products Covariance 69.059 -229.783 -49.718 123.261 N 10 10 10 10 Planted 1982 Pearson -.074 -.455 -.247 .030 Correlation Sig. (2-tailed) .838 .187 .492 .935 Sum of Squares -464.001 -2570.123 -1261.620 150.567 and Cross- products Covariance -51.556 -285.569 -140.180 16.730 N 10 10 10 10 Planted 1985 Pearson .033 -.542 .067 .177 Correlation Sig. (2-tailed) .924 .085 .846 .603 Sum of Squares 181.355 -3029.905 339.632 913.361 and Cross- products Covariance 18.136 -302.990 33.963 91.336 N 11 11 11 11 Planted Pearson .338 -.566 -.026 -.114 1988/1989 Correlation Sig. (2-tailed) .374 .112 .946 .771 150 Rainfall Year 0 Year 1 Year 2 Year 3 Sum of Squares 2759.230 -3972.970 -189.637 -730.147 and Cross- products Covariance 344.904 -496.621 -23.705 -91.268 N 9 9 9 9 Planted OPRI Pearson -.304 .128 .559 -.661 1989 Correlation Sig. (2-tailed) .464 .762 .150 .075 Sum of Squares -1157.328 434.836 1882.746 -2011.723 and Cross- products Covariance -165.333 62.119 268.964 -287.389 N 8 8 8 8 Planted 1990 Pearson -.467 -.372 .699 .388 Correlation Sig. (2-tailed) .291 .411 .081 .390 Sum of Squares -1835.553 -1244.519 2516.500 1190.454 and Cross- products Covariance -305.925 -207.420 419.417 198.409 N 7 7 7 7 All Planted Pearson -.026 -.359 .045 -.200 Years Correlation Average Sig. (2-tailed) .918 .143 .858 .427 Sum of Squares -313.417 -3820.000 508.167 -2228.317 and Cross- products Covariance -18.436 -224.706 29.892 -131.077 N 18 18 18 18 ** Correlation is significant at the 0.01 level (2-tailed). * Correlation is significant at the 0.05 level (2-tailed). 151 Table A4: FFB yield data correlated with annual average relative humidity at 1500 hours Relative Humidity Year 0 Year 1 Year 2 Year 3 Planted 1979 Pearson Correlation .152 -.360 .385 -.299 Sig. (2-tailed) .656 .277 .243 .371 Sum of Squares and 8.504 -19.138 20.920 -16.562 Cross-products Covariance .850 -1.914 2.092 -1.656 N 11 11 11 11 Planted 1980 Pearson Correlation -.033 -.341 .479 -.278 Sig. (2-tailed) .922 .306 .136 .408 Sum of Squares and -1.789 -18.209 24.335 -14.410 Cross-products Covariance -.179 -1.821 2.434 -1.441 N 11 11 11 11 Planted 1981 Pearson Correlation .369 -.340 .469 -.308 Sig. (2-tailed) .294 .336 .171 .387 Sum of Squares and 12.520 -12.810 16.470 -11.015 Cross-products Covariance 1.391 -1.423 1.830 -1.224 N 10 10 10 10 * Planted 1982 Pearson Correlation .254 -.321 .691 -.374 Sig. (2-tailed) .478 .366 .027 .287 Sum of Squares and 11.099 -14.098 33.686 -17.030 Cross-products Covariance 1.233 -1.566 3.743 -1.892 N 10 10 10 10 Planted 1985 Pearson Correlation .099 -.136 .571 -.026 Sig. (2-tailed) .772 .689 .067 .939 Sum of Squares and 3.244 -5.181 21.128 -.943 Cross-products Covariance .324 -.518 2.113 -.094 N 11 11 11 11 Planted Pearson Correlation -.097 -.278 -.697 .644 1988/1989 Sig. (2-tailed) .820 .506 .055 .085 Sum of Squares and -2.743 -6.270 -15.078 22.230 152 Relative Humidity Year 0 Year 1 Year 2 Year 3 Cross-products Covariance -.392 -.896 -2.154 3.176 N 8 8 8 8 Planted OPRI Pearson Correlation -.485 -.565 -.170 .068 1989 Sig. (2-tailed) .270 .187 .715 .886 Sum of Squares and -4.676 -4.340 -1.177 .409 Cross-products Covariance -.779 -.723 -.196 .068 N 7 7 7 7 * Planted 1990 Pearson Correlation .685 .273 .310 .874 Sig. (2-tailed) .133 .600 .550 .023 Sum of Squares and 6.440 3.450 3.510 8.220 Cross-products Covariance 1.288 .690 .702 1.644 N 6 6 6 6 ** Correlation is significant at the 0.01 level (2-tailed). * Correlation is significant at the 0.05 level (2-tailed). 153 Annex C: Methods used to perturb observed yield models x Maximum temperature: o UNDP ensemble min, mean and maximum data files for mean annual temperature were used to represent the range of uncertainty across the climate projection models. o As there are no data sets for maximum temperature, it is assumed that the probability distribution for future changes in annual mean temperature are also the same for other temperature variables, in this case, annual mean maximum temperature. o Future changes in temperature from the observed maximum temperature baseline year used to develop the models were applied to the observed maximum temperatures to project future temperatures beginning 2010 and 2020 respectively. x Rainfall (for the IHRO hydrological deficit calculation): o The IHRO method for calculating hydrological deficit was used to recalculate observed monthly hydrological deficit data from Jan 1977 onwards (the date from which monthly rainfall and rain days data are available from Kwae Agric. Office/CIRAD). o The recalculated monthly observed hydrological deficits were in turn aggregated up to annual values and a 10 year annual average hydrological deficit data set was calculated. o UNDP ensemble min, mean and maximum data files for mean monthly rainfall were used to represent the range of uncertainty across the climate projection models. o Future changes in mean monthly rainfall from the observed baseline year used to develop the models were applied to the observed monthly rainfall data set and future projections for hydrological deficit were calculated – note that future rain days data were not available and therefore future projections have assumed that the number of rain days remain unchanged. x The projected future changes in temperature and hydrological deficit were applied to the baseline data in the models to return changes in future FFB yields. 154 Annex D: Statistical tests Table A5 illustrates results of statistical tests on the model described in Equation 1, Section 5.4, which gives the best fit (highest value of R-squared) to the observed temperature and vacuum data. The value ° of the coefficient estimated by the model implies that a 1 C change in temperature would lead to a 1% change in vacuum. Table A5: Model results and statistical tests using the Simple Linear Regression method. (The dependent variable is vacuum strength and the sample size (number of observations) is 64). Independent variable Atmospheric temperature Coefficient 1.02 Standard error .28 t-statistics 3.7 Probability (t-statistics) .001 R-correlation 0.59 R-squared 0.34 F-statistics 13.5 Probability (F-statistics) .001 The regression coefficient was assessed through a t-statistics test. The value of the t-statistics for the independent variable (temperature) indicates that the coefficient (1.0) is statistically valid. Thus, temperature is a statistically significant determinant of vacuum strength. The F-ratio, which tests the overall fit of the regression model, is significant at p=0.001. This result tells us that there is a 0.1% chance that an F-ratio this large would happen by chance alone (Field, 2005). The R-squared value is a measure of the goodness of fit of the regression, indicating how well temperature explains the vacuum. The value of 0.34 shown in Table A5 means that 34% of the variation in vacuum is explained by variations in temperature. However, it is noted that temperatures were measured only to the nearest degree Celsius, and more precise readings would likely have improved the R-squared value. 155 Annex E: Results of Pearson’s correlation analysis Table A6: Results of Pearson’s correlation analysis, testing relationship between percentage of average monthly malaria cases and climatic variables (N = 12) Months Pearson lag Correlation Sig. (2-tailed) Average monthly malaria cases 2004–2007 1 Rainfall 0 -.235 .463 1 .702* .011 2 .737** .006 Tmin 0 -.410 .186 1 .203 .526 2 .204 .524 Tmax 0 -.437 .155 1 -.248 .436 2 -.165 .609 Tmean 0 -.496 .101 1 -.145 .653 2 -.090 .781 RH_0600 0 .506 .093 1 .626* .029 2 .041 .899 RH_1500 0 .347 .269 1 .551 .064 2 .386 .215 RHmean 0 .389 .212 1 .623* .030 2 .420 .175 Rainy days 0 .064 .843 1 .708* .010 2 .793** .002 * Correlation is significant at the 0.05 level (2-tailed) ** Correlation is significant at the 0.01 level (2-tailed) 156 International Finance Corporation 2121 Pennsylvania Ave. NW Washington, DC 20433 Tel. 1-202-473-1000 www.ifc.org/climatechange The material in this publication is copyrighted. IFC encourages the dissemination of the content for educational purposes. Content from this publication may be used freely without prior permission, provided that clear attribution is given to IFC and that content is not used for commercial purposes. The findings, interpretations, views, and conclusions expressed herein are those of the authors and do not necessarily reflect the views of the Executive Directors of the International Finance Corporation or of the International Bank for Reconstruction and Development (the World Bank) or the governments they represent, or those of Ghana Oil Palm Development Company and the individuals and institutions that contributed to this study.