DISTRIBUTION OF SAG0 PALM (Metroxylon spp) & HABITAT DISTRIBUTION PREDICTION AT PULAU SERAM, MALUKU PROVINCE
Lilik B.Prasetyo(1), Agus P. Kartono(1), Haris Syahbuddin(2), Samin Botanri (3), and Agus Jacob(3), Beni Okarda (4)
(1)Bogor Agricultural University, Bogor (2) Ministry of Agriculture, (3) Darusalam University-Ambon, (4) Master Degree Program of Information Technology for Natural Resources Management-IPB
Keywords : Spatial Modelling, Metroxylon spp, Distribution, Sago palm, GIS & Remote Sensing
Area of Sago palms in Indonesia is the largest among Sago starch production countries such as Papua New Guinea, Malaysia and Thailand. It was estimated around 50% of Sago palms of the world is situated in Indonesia, which is distributed in Sumatra, Maluku & Papua. Sago palm area in Maluku is estimated of about 30.1 thousand hectare (Mulyono & Suward,2000), meanwhile Ruhendi (2000) has estimated with higher figure (800 thousand hectare). In other publication Louhenapessy (1992) has annouced lower figure (26 thousand hectare).
Sago palms (Metroxylon spp) is multi-function trees. Its starch have been utilized by rural community at Papua, Maluku and other eastern part of Indonesia as staple food. It can be utilized also for raw material of food industries, plywood and also for bioenergy. Since two decades ago, food habit/preference of community has shifted to rice, which lead to decrease sago starch consumption, and its habitat conversion.
Based on the above situation, it is an urgent need to estimate actual Sago palm distribution and its potential habitat for management purposes. Up to now, estimation of Sago palm distribution was based on terestrial survei & sampling, and therefore its accuracy is questioned. Remote Sensing and Geographical Information System (GIS) can be utilized to enhance the accuracy of invetory purposes. The research objectives are as follows (1) develop distribution map of sago palms in Seram Island based on satellite imagery, (2) develop spatial model of sago palms distribution by GIS. Classification of Landsat TM was conducted by Supervised classification with maximum likelihood technique. Meanwhile, its habitat prediction was modeled by logistic regression.
It was estimated that area of sago in Seram island was about 17 902 hectare.It distributed at low altitude (250 m dpl)(left figure, green colour indicated Sago habitat distribution prediction).
Logistic regression model of its habitat distribution is presented in equation below :
P =1/1 + e -(-3.698+(3.384x[elevation])-( 0.639x[slope])+(0.398*[river distance])+(0.359* [coastline distance)-(0.150* [soil type]))
Based on Hosmer and Lemeshow Test model was valid. However, Nagelkerke R Square of the model was very low ( 35%). It means that there are other factors contributed to the sago distribution. Regarding role of each variable, variable of elevation,slope, distance from river and coastline had significance impact on its habitat distribution. further analysis showed that utilizing sago distribution derived from landsat as validation, accuracy of the model was about 76 %.
Paper was presented in Plant Ecology and Diversity Observation Network and Capacity Building in Indonesia (16 – 17 July 2010, Univ.Udayana- Bali)