The Use of Remote Sensing and Geographic Information System to Determine the Spatial Distribution of Melaleuca cajuputi as a Major Bee Plant in Marang, Terengganu

Abstract

In Malaysia, honey is chiefly obtained from species of honeybees known as Apis dorsata and to a lesser extent Apis cerana. Honey from Apis dorsata is a supplementary source of income to many rural poor in the district of Marang, Terengganu. The colonies of A. dorsata are found to nest in aggregates on tall bee trees (tree emergent) in the open, as well as, nesting singly in concealed locations when nesting low, especially in the submerged forest of Melaleuca cajuputi as in the vast hectare (> 200,000 hectares) of Melalueca forest along the coastal areas of Terengganu. So, Melaleuca forest mapping and flower mapping can be reliable methods for determining this species distribution as the main source of nectar and pollen for these aforementioned honey bees. In ecology, biomass can be defined as accumulation of living matter which is useful as a biophysical index for mapping of flower in forest. In this study, we used SPOT-5 and RADARSAT-1 for inventory of Melaleuca forest in Marang and developed Above Ground Biomass (AGB) estimation model as indirect index for obtaining and producing distribution of Melaleuca cajuputi flowers. Also, Apis dorsata colonies distribution and motorbike parking points of honey hunters were collected using GPS in field survey to determine distribution of colonies and improve searching ability in Apis dorsata colonies harvesting by honey hunters in the study area. The Melaleuca forest, located in Marang, Terengganu, Malaysia which is lying in upper left latitude 5°17'15.473"N, and longitude 103°05'25.021"E and lower right latitude 4°37'55.236" N, longitude 103°45'47.568"E was chosen for this research. SPOT-5 was enhanced, classified and vectorized using image processing software for the purpose of Melaleuca forest mapping. Based on the image analysis of the SPOT-5 image the Melaleuca forest were classified as five classes Melaleuca Cajuputi, Acacia auriculiformis, non-vegetation, water bodies and Cloud/haze/Shadow. The analysis showed that Melaleuca cajuputi covered 76,061.73ha (61.72%), Acacia auriculiformis 24,484.32ha (19.88%), non-vegetation 9,991.76ha (8.11%), water bodies 2,203.47ha (1.79%) and Cloud/Haze/Shadow 10,491.86ha (8.51%) with an overall classification accuracy of 91.79% while the statistics value obtained from kappa coefficient was more than 0.86 which is relatively quite good results for image processing. Based on Melaleuca forest inventory, 10 plots of 10m 􀂯10m were established to measure the AGB along with two stand parameters namely the DBH and height. The measurements were analyzed to determine the descriptive statistics of each plot using SPSS software. AGB was later regressed against the radar backscattering coefficient, derived earlier from the RADARSAT-1 images, to establish the regression fit. Once the regression fit model was obtained, the radar backscattering coefficient data were converted to the AGB mapping. The accuracy of AGB was evaluated using the accuracy assessment in ERDAS IMAGINE v 9.1. The relationship between AGB and radar backscattering is described with the linear regression equation of AGB=238.73+ (15.4) × σ◦ (R2 = 0.89) for flowering season and AGB=170.843+ (13.82) × σ◦ (R2 = 0.81) for non-flowering season. The range of measured AGB was similar to the overall range in AGB from radar backscattering. Through the model of AGB value estimation, AGB mapping on Melaleuca forest can be done with each pixel holding its own AGB value. This study found that remote sensing can be used to estimate AGB in forest stands and also to produce AGB map in Melaleuca forest. Then, flowering map was derived from AGB classified map in two different seasons of flowering and non-flowering. Finally, all maps and information were transferred and overlaid in Arc/GIS v 9.1 to find probability and searching ability of Apis dorsata colonies by honey hunters.