GIS-BASED ECOCROP MODELLING TO ASSESS POTENTIAL CLIMATE CHANGE EFFECTS ON SAGO PALM SUITABILITY DISTRIBUTION Meriam Makinano-Santillan, Jojene R. Santillan Caraga Center for Geoinformatics, College of Engineering and Information Technology Caraga State University, Ampayon, Butuan City, Philippines Email: santillan.jr2@gmail.com, meriam.makinano@gmail.com KEY WORDS: Sago palm, EcoCrop, climate change, impact analysis, suitability, cultivation ABSTRACT: This paper presents an analysis of the potential effects of changing climate on the geographical distribution of suitable areas in the Philippines for the cultivation of Sago palms. The Sago palm is considered to be the highest starch producer at 25 tons per hectare per year. Sago palms are being cultivated in Visayas and Mindanao although its utilization as source of starch is not yet maximized. Its mass cultivation and commercial utilization has recently gained interest from the government in order to develop and sustain a large-scale Sago starch industry. We used the FAO EcoCrop model to predict climatically-suitable areas of Sago palms for the current and future climate scenarios. EcoCrop is a simple mechanistic model which uses climate datasets and expert-derived temperature and rainfall ranges as inputs to determine the main niche of a crop and then produces a suitability score as output. However, it requires validation to determine accuracy of its predictions. For the current climate, we utilized WorldClim version 1.4, a dataset of global climate surfaces representative of the years 1950 to 2000. The predicted distribution for the current climate was validated using 472 random samples of confirmed Sago palm locations obtained through satellite image analysis and field surveys. Results of the EcoCrop model validation revealed a high model prediction rate of 90.04%. We then predicted the Sago palm suitable areas for future climate scenarios where we utilized future climate datasets projected by the Community Climate System Model (CCSM) 4 for the year 2050 under a Representative Concentration Pathway (RCP) 2.6 emission scenario. The comparison of the current and future distributions of Sago palm suitable areas revealed an increase of 6% in total suitable areas from current to the future scenarios which may indicate that the projected year 2050 climate may have positive effect on Sago palm suitable areas. The most interesting result, however, is the 40% increase in areas with "excellent" suitability. These results may indicate that the future climate scenario is favorable for the mass cultivation of Sago palm in the Philippines. 1. INTRODUCTION The Sago palm (Figure 1) is considered to be the highest starch producer at 25 tons per hectare per year (Bujang, 2008). Over a growing period of 10 years or more, the starch is accumulated in the palm s trunk, and approximately 200 kg of dry starch can be harvested per trunk (The Society of Sago Palm Studies, 2015). Sago palms are being cultivated in Visayas and Mindanao although its utilization as source of starch is not yet maximized. Its mass cultivation and commercial utilization has recently gained interest from the government in order to develop and sustain a large-scale Sago starch industry. This is due to its advantages of being economically acceptable, relatively sustainable, environmentally friendly, Figure 1. Sago palms in Balite, Aklan, Philippines. uniquely versatile, vigorous, and promotes socially stable agroforestry systems (Flach, 1997; Stanton, 1991). Sago palms are now being grown commercially in Malaysia, Indonesia and Papua New Guinea for production of Sago starch and/or conversion to animal food or fuel ethanol (McClatchey et al., 2006). Some studies have been conducted to locate areas in the Philippines that are suitable for the mass propagation of Sago palms (UP TCAGP, 2013; Santillan and Santillan, 2013). In these studies, suitable areas were identified using habitat suitability models, with biophysical (e.g., elevation, slope, soil texture) and bioclimatic datasets (e.g., rainfall, and 91
temperature) as major inputs. From these works, thousands of hectares of lands in the Visayas and Mindanao were found to be biophysically and bioclimatically suitable for growing Sago palms (UP TCAGP, 2013). What was not considered in these suitability studies is the effect of changing climate on the Sago palm suitability distribution. The biophysical and bioclimatic environmental requirements of Sago palms are restricted to areas near sea level up to 700 m above sea level whose annual rainfall ranging from 2100-5800 mm, and with temperature above 18 o C (FAO, 2007). In the advent of changing climate which has a direct impact on changes in rainfall and temperature, it is important to investigate how the future climate scenario could affect the distribution of areas currently considered as suitable for mass propagation of Sago palms. Such investigation can help us understand and answer some questions related to Sago palm suitability distribution such as: Will the present locations of Sago palm-suitable areas meet its rainfall and temperature requirements in a future climate scenario? Will there be an increase or decrease in suitable areas? In this work, we present a geospatial analysis of the potential effects of climate change on Sago palm suitability distribution in the Philippines. The study aims to (i.) generate climatically-suitable areas of Sago palms for the current and future climate scenarios through the use of a Geographic Information System (GIS)-based Food and Agriculture Organization (FAO) EcoCrop model, and (ii) provide comparisons on the differences in geographic distributions of suitable areas for the two climate scenarios. The suitability modelling and analysis was complemented by information derived from satellite image analysis and field surveys that made possible the validation of the FAO EcoCrop modelling results. 2. METHODS AND DATASETS 2.1 Overview A flowchart of the approach employed in this study is shown in Figure 2. Basically, we used current and future climate datasets and information on the bioclimatic requirements of Sago palms as inputs into the FAO EcoCrop model to generate maps showing the location of Sago palm suitable areas for the two climate scenarios. These maps were then compared in terms of change in suitable areas, and further assessed taking into account the changes in climate. Current Climate (WorldClim ver. 1.4) Sago Palm Temperature and Rainfall Requirements EcoCrop Model Future Climate (CCSM4 RCP 2.6-2050) 2.2 The FAO EcoCrop Model The FAO EcoCrop is a simple mechanistic model that has been largely used to predict suitability of various crops under different climatic conditions. It was originally developed by Hijmans et al. (2001) based on the FAO-EcoCrop database (hence, the name EcoCrop ). The model uses expert-based temperature and rainfall ranges which are reported in the FAO-EcoCrop database (http://ecocrop.fao.org) as inputs to determine the main niche of a crop and then produces a suitability score as output (Ramirez-Villegas et al., 2013). Each of these ecological ranges is defined by a pair of parameters for each variable (i.e. temperature and rainfall). The temperature and rainfall ranges are defined by the absolute range (minimum and maximum absolute temperatures and rainfall at which the crop can grow) and by the Current Climate Suitable Areas Map Comparison Future Climate Suitable Areas Map Assessment of Change in Suitable Areas with Respect to Change in Climate Comparison Figure 2. Flowchart of the approach employed to analyze the potential effects of climate change on Sago palm suitability distribution. optimum range (minimum and maximum optimum temperatures and rainfall). Using a gridded data of temperature and rainfall, the model s algorithm determines the conditions over the growing season at a particular place. When the conditions are beyond the absolute thresholds, the suitability index is zero (not suitable); when they are between absolute and optimum thresholds, the suitability score ranges from 1 to 99, and when they are within the optimum 92
conditions the suitability score is 100% (highly suitable). The model performs two different calculations separately, one for rainfall and the other for temperatures and then calculates the interaction by multiplying them. The final suitability scores are then grouped to indicate suitability types: very marginal (1-20%), marginal (20-40%), suitable (40-60%), very suitable (60-80%), and excellent (80-100%). 2.3 Climate Datasets For the current climate, we used WorldClim version 1.4, available at http://www.worldclim.org. WorldClim is a set of global climate layers (climate grids) with a 30 arc-second spatial resolution (approximately 1 km x 1 km) depicting monthly climatology (maximum, minimum and mean temperatures, and total monthly rainfall) representative of the years 1950 2000 (Figure 3). This dataset is well documented in Hijmans et al. (2005) and has been used in climate change-related suitability studies such as those by Lane and Jarvis (2007), and Ramirez-Villegas et al. (2013), among others. For the future climate, we use 1 x 1 km climate grids depicting monthly climatology projections by the Community Climate System Model (CCSM) 4 global climate model (GCM) for the year 2050 under a Representative Concentration Pathway 2.6 emission scenario (Figure 3). These datasets, downloaded from http://worldclim.org/cmip5_30s, were downscaled versions of the CCSM4 GCM projections, and calibrated (bias corrected) using WorldClim 1.4 as baseline 'current' climate. Details about the downscaling and calibration are available at http://worldclim.org/downscaling. The CCSM4 is one of GCMs whose climate projections were used in the Fifth Assessment IPCC report. The RCP 2.6 scenario projects an average global warming increase of 1.0 0 C with likely range of increase from 0.4 to 1.6 0 C for the years 2045 2065. The CCSM4 RCP 2.6 dataset was selected for the reason that this was the most accessible dataset during the time when this study was conducted in 2014. Some of the recent studies utilizing CCSM4 RCP 2.6 for assessing climate change impacts on crop suitability include those of Palazzoli et al (2015), Liu et al (2015), and Bocchiola (2015). Figure 3. Example current and future climate datasets used in EcoCrop modeling. Also included are computed changes in Annual Mean Temperature and Annual Precipitation. 93
2.4 Generation of Suitable Areas for Current and Future Climate Scenarios DIVA GIS Version 7.5 (Hijmans et al., 2012) was used to implement the EcoCrop model and to generate the maps of suitable areas with a resolution of 1 x 1 km. This software has a built-in function for the EcoCrop model, including the FAO database of crop temperature and rainfall ranges. Based on the database, the Sago palm s absolute temperature range is from 18 40 o C, with 25 36 o C as the optimum range. The absolute total annual rainfall range is from 2100 5800 mm, with 3000 4500 mm as the optimum range. The growing season is 1 year (365 days). The current and future climate layers depicting minimum and maximum monthly temperatures and minimum and maximum monthly rainfall were used in the EcoCrop modeling. 2.5 Accuracy of the EcoCrop Model and Validation of Current Suitable Areas Map This work may be considered as the first climate-based suitability modeling of Sago palms using the EcoCrop model. As such, there is a need to verify the accuracy of model-generated suitable areas map for the current climate scenario. To do this, we utilized 472 random points representing actual locations of Sago palms in Visayas and Mindanao (Figure 4). These points were extracted from a database of mapped Sago Figure 4. Map showing the 472 randomly-selected data points representing actual locations of Sago palms in Visayas and Mindanao that were used to assess the accuracy of EcoCrop model-derived map of suitable areas for the current climate scenario. Sago palm data from UP TCAGP (2013). palms in Visayas and Mindanao that was generated from a study conducted by the University of the Philippines Training Center for Applied Geodesy and Photogrammetry from 2012 to 2013. This database was generated by mapping Sago palm locations using medium resolution (Landsat ETM+, ALOS AVNIR-2) and high resolution (Worldview-2, Google Earth) satellite images acquired between the years 2008-2012, and verified through field surveys conducted in 2012. Details about the methods used can be found in UP TCAGP (2013). The accuracy of the current climate suitable areas map was assessed by checking how many of the 472 randomly selected Sago palm locations were correctly classified as suitable in the current suitable areas map (i.e., suitability score is > 0%). 3. RESULTS AND DISCUSSION 3.1 Changes in Temperature and Rainfall under the Current and Future (2050) Climate Scenarios The Philippines average annual mean temperature for the current and future climates were computed from the given climate datasets as 25.46 ± 2.10 o C and 26.43± 2.09 o C, respectively; while for the annual rainfall, the computed values were 2,548 ± 586 mm (current climate), and 2,635 ± 614 mm (future climate). The maps showing the changes in mean annual temperature and annual rainfall were previously shown in Figure 3. The change in annual mean temperature (from current to future) ranges from -4 to +5 o C. Based on the climate datasets, the average change in mean annual temperature was computed at 0.97 ± 0.58 o C. On the other hand, the change in annual rainfall ranges from -1,344 to +1,041 mm, with an average change of 88 ± 95 mm. From these computed values, it is very apparent that the Philippines is projected to become warmer and to receive more rainfall in the year 2050 based on the CCSM4 GCM model projections. How these changes could impact the geographic distribution of Sago palm suitable areas are explained in the next sub-sections. 94
27,577 25,602 31,740 31,452 36,434 33,557 31,828 26,568 Area, in sq. km. 50,158 70,328 115,619 105,848 36th Asian Conference on Remote Sensing 2015 (ACRS 2015), Quezon City, Philippines, Volume 1, pp. 91-97, 2015 Figure 5. Maps of Sago palm suitable areas in the Philippines for the current and future climate scenarios. 3.2 Sago Palm Suitable Areas for the Current and Future Climate Scenario and their Changes in Geographic Distribution The maps of Sago palm suitable areas in the Philippines for the current and future climate scenarios are shown in Figure 5. The accuracy of the EcoCrop model s prediction of the current suitable areas was found to be high, with a prediction rate of 90.04%, i.e., 425 out of the 472 random samples of confirmed Sago palm locations were correctly predicted as suitable by the EcoCrop model. For the current climate, about 177,377 km 2 were found to be generally suitable, while about 187,509 km 2 were found to be generally suitable for the future climate. There was an increase of about 6% in suitable areas from the current to the future climate scenario. 140,000 120,000 Current Climate Future Climate 100,000 80,000 60,000 40,000 20,000 0 Excellent Very suitable Suitable Marginal Very marginal Not suitable Figure 6. Statistics of Sago palm suitable areas for the current and future climate scenarios. 95
On the other hand, there was a general decrease of about 8% of areas not suitable for Sago palm. Looking into the specific suitability classes (Figure 6), increase in excellent suitable areas was the most significant effect of climate change, with about 40% increase in area from current to future scenario. Although slight decreases in area for other suitability classes are also noticeable, majority of these decrease in areas were found to be due to their conversion into excellent suitable areas. 4. CONCLUSIONS In this paper, we presented an analysis of the potential effects of changing climate on the geographical distribution of suitable areas in the Philippines for the starch-rich Sago palm. Using the EcoCrop model and climate datasets, we were able to derive maps of suitable areas of Sago palm for the current and future climate scenarios. The comparison of the current and future distributions of Sago palm suitable areas revealed an increase of 6% in total suitable areas from current to the future scenarios which may indicate that the projected year 2050 climate may have positive effect on Sago palm suitable areas albeit minimal. The most interesting result, however, is the 40% increase in areas with excellent suitability. These results may indicate that the future climate scenario is favorable for the mass propagation of Sago palm in the Philippines. This has practical implication in the appropriate planning and development of Sago palm plantations wherein the effects of climate change can be considered throughout the process. 5. RECOMMENDATIONS While the study was able to identify specific locations of Sago palm suitable areas, the locations identified as suitable should be interpreted as areas that have the climatic conditions suited for growing Sago palm. Factors such as existing land-uses/land-cover in the suitable areas, and socioeconomic conditions are not yet accounted. These factors must be accounted in order to narrow down the suitable areas only to those locations where it is indeed possible to grow sago palms. An example of this would be determining whether a suitable area in the map has other land-uses (e.g., forest, protected area, cropland, grassland, etc.), and determining whether these suitable areas have favorable conditions for sago palms to be propagated at plantation scale (i.e., appropriateness of soil type, nearness to water supply, roads, processing plants, etc.). All of these are subject to further study and analysis. ACKNOWLEDGEMENTS The authors would like to acknowledge and thank the Research Laboratory for Applied Geodesy & Space Technology (AGST Lab) of the UP Training Center for Applied Geodesy & Photogrammetry, University of the Philippines-Diliman for providing access to the Sago Palm locations database. REFERENCES Abd-Azis, S., 2002. Sago starch and its utilization. Journal of Bioscience and Bioengineering, 94(6), pp. 526 529. Bocchiola, D., 2015. Impact of potential climate change on crop yield and water footprint of rice in the Po valley of Italy. Agricultural Systems, Vol. 139, pp. 223-237. Bujang, K.B., 2008. Potentials of Bioenergy from the Sago Industries in Malaysia. In: Doelle, H.W., J.S. Rokem, and M. Berovic (eds.). Biotechnology, Vol XIV, Encyclopedia of Life Support Systems. FAO, 2007. Ecocrop Dataset for Metroxylon sagu. Retrieved August 1, 2015 from http://ecocrop.fao.org/ecocrop/srv/en/datasheet?id=1466 Flach, M., 1997. Metroxylon sagu Rottb. - Promoting the Conservation and Use of Underutilized and Neglected Crops. Institute of Plant Genetics and Crop Plant Research, Gatersleben/International Plant Genetic Resources, International Plant Genetic Resources, Rome, Italy. Hijmans, R.J., Guarino, L. Mathur, P., 2012. DIVA GIS Version 7.5. http://www.diva-gis.org/ Hijmans, R.J., Guarino, L., Cruz, M., Rojas, E., 2001. Computer tools for spatial analysis of plant genetic resources data: 1. DIVA-GIS. Plant Genetic Resources Newsletter, 127, pp. 15-19. 96
Lane, A., Jarvis, A., 2007. Changes in climate will modify the geography of crop suitability: agricultural biodiversity can help with adaptation. SAT ejournal, 4(1), pp. 1-12. Liu, C., Hofstra, N, Leemans, R., 2015. Preparing suitable climate scenario data to assess impacts on local food safety. Food Research International, 68, pp. 31-40. McClatchey, W, H.I. Manner, Elevitch, C.R., 2006. Metroxylon amicarum, M. paulcoxii, M. sagu, M. salomonense, M. vitiense, and M. warburgii (sago palm). In: Elevitch, C.R. (ed.) Species Profiles for Pacific Island Agroforestry. Permanent Agriculture Resources, Hōlualoa, Hawaii. Palazzoli, I., Maskey, S., Uhlenbrook, S., Nana, E., Bocchiola, D., 2015. Impact of prospective climate change on water resources and crop yields in the Indrawati basin, Nepal. Agricultural Systems, 133, pp. 143-157. Ramirez-Villegas, J., Jarvis, A., Läderach, P., 2013. Empirical approaches for assessing impacts of climate change on agriculture: the EcoCrop model and a case study with grain sorghum. Agricultural and Forest Meteorology, 170, pp. 67-78. Santillan, M., Santillan, J., 2013. Habitat suitability analysis of the starch-rich Sago palm using satellite-derived data and a species distribution model. In: Proceedings of the 34th Asian Conference on Remote Sensing, ACRS 2013 Bridging Sustainable Asia, October 20-24, Bali, Indonesia. Stanton, W.R., 1991. Long term and ancillary environmental benefits from sago agroforestry systems. In: Ng Thai-Tsiung, Tie Yiu-Liong & Kueh Hong-Siong (eds). Proceedings of the Fourth International Sago Symposium. Kuching, Sarawak, Malaysia, pp. 24-35. The Society of Sago Palm Studies, 2015. The Sago Palm: The Food and Environmental Challenges of the 21st Century, Kyoto University Press, Kyoto Japan; Trans Pacific Press, Victoria, Australia. UP TCAGP, 2013. Mapping Sago Habitats and Sago Suitable Sites using Optical and Radar Image Analysis and Suitability Relationships. Terminal Report. Training Center for Applied Geodesy and Photogrammetry (TCAGP), University of the Philippines, Diliman, Quezon City. Available online: http://dx.doi.org/10.13140/rg.2.1.3190.2886 97