Discussion
Birds are thought to perform the widest range of ecological functions in the ecosystem (Sekercioglu, 2006). Birds benefit humans by providing a variety of important ecosystem services, including provisioning (food, feathers), regulating (seed dispersal, pollination), cultural (art, religion), and supporting (soil formation, nutrient cycling) services (Millennium Ecosystem Assessment, 2005). Hornbills are frugivore, which means they eat fruit. Frugivores play a critical role in the ecosystem’s structure and function (Nathan and Muller-Landau, 2000). However, due to the gradually increasing human population following anthropogenic disturbance, there has been a severe decline in the population of these birds.
Species distribution models (SDMs) are widely used in biogeography, biodiversity, and macro-ecology research to model species’ geographic distribution based on the correlation between known occurrence records and environmental conditions at occurrence locations (Elith et al., 2011). SDM creates geographic maps of environmental suitability for a species(Raes, 2012) . We chose the MaxEnt model to predict the distribution of the Great Hornbill and the Rufous-necked hornbill because its distribution prediction provides a powerful new tool that uses only presence data for species distribution modeling. Since its inception, MaxEnt has grown in popularity (Renner and Warton, 2013). MaxEnt is a presence-only model that allows scientists to tap into the wealth of natural history data while avoiding the high cost of sampling species across their entire range (Phillips et al., 2006). Data on presence is plentiful, but data on absence is difficult to come by and frequently unreliable due to a lack of survey effort (Liu et al., 2013). MaxEnt compares the distribution of presences along environmental gradients to the distribution of background points, which are drawn at random from the study area (Renner and Warton, 2013). Furthermore, it considers interactions between environmental variables and appears to perform reasonably well with small amounts of occurrence data when valid occurrence data and appropriate predictor variables are chosen(Phillips et al., 2006).
The distribution model for the great hornbill and the rufous-necked hornbill was created using 21 environmental variables and three topographical variables (elevation, aspect, and slope). However, after performing a multi-collinearity test in the ENM tool, only thirteen variables were used to prepare a habitat suitability model and conduct further analysis. The results revealed that the most important environmental variables for predicting the species presence location for GH and RNH were elevation (30.4%), precipitation of wettest month (27.2%), Aridity Index (19%), and precipitation of wettest month (44.7%), aspect (22.3%), and aridity index (11.5%), respectively.
The probability of GH occurrence and elevation has a negative relationship, meaning that the likelihood of seeing great hornbills decreases as elevation rises (fig 6& 7). According to the current study, the probability of GH occurrence is high below 3000 masl. The great hornbill is known to frequent wet evergreen and mixed deciduous forests, ranging out into open deciduous areas to visit fruits and ascend slopes to at least 1,560 masl in south India (Mudappa and Raman, 2009) and up to 2,000 masl in Thailand(Poonswad et al., 2013). The RNH is found in primary subtropical evergreen and deciduous forests between 600 and 2000 masl worldwide but have been seen as high as 2900 masl (Chimchome et al., 1997). RNH presence was predicted in the species occurrences predicted map at Lhuentse, Yangtse, Punakha, and even up to Gasa, which is over 3000 masl. Habitat suitability analysis shows that GH and RNH are highly suitable for 2 and 3% of the total country’s area, respectively. Their occurrence and distribution are affected by factors such as vegetation type, habitat, temperature, availability of food, forest size, and tall old trees with holes(Wagh et al., 2015).
Figure 6: Predicted GH occurrence map from low to high probability values
Figure 7: Predicted RNH occurrence map from low to high probability values

Conclusion

The distribution potential of great hornbill and Rufous-necked Hornbill was delineated using species distribution modeling. This model is a very effective tool for mapping the suitable habitats of different species under the influence of topographic and climatic factors. In comparison, Great hornbill has 2 % of the area of Bhutan suitable for habitat whereas Rufous necked hornbill has total of 3% of most suitable habitat in Bhutan. However Rufous-necked hornbill habitat range is high in central, south-east and north-east region of Bhutan. The probability of occurrence of the great hornbill is high in the southern region and elevation below 3000 msl. Rufous-necked hornbill prefers semi-arid and arid region whereas great hornbill prefers humid condition. The model also revealed that the highest contributing variables for predicting this threatened species were aspect and precipitation of wettest month. The baseline information generated from the present study can be useful facilitating fieldwork, planning and future scientific management of these species in the country.