All species have limits to their geographic distributions and in some cases, these limits can be accurately estimated, whereas in other cases, there is a continuous change in these limits, be it expansion or contraction, making it harder to accurately map species ranges. Although climatic and anthropogenic factors play a major role in defining the range of a particular species, it is imperative to determine how species react to such changes and this is not possible without an accurate knowledge of its distribution.
The International Union for Conservation of Nature has maintained a Red List of all species since 1994, and their aim is to provide a robust threat assessment for all recognized taxa on the planet, which includes spatial data such as the Extent of Occurrence (EOO) and the Area of Occupancy (AOO). However, data concerning EOO’s and AOO’s has remained largely inaccurate due to insufficient data regarding the distribution of a species in the first place. In this study, we considered the 25 diverse and endemic avifauna of the Western Ghats to test the hypothesis that majority of their spatial extent has been overestimated and are inaccurate. With advances in species distribution modeling, we used machine-learning techniques such as RandomForests and Boosted Regression Trees to predict suitable habitat for these birds. We used occurrence data from a citizen science repository (eBird) and environmental data that consisted of temperature, precipitation, elevation, aspect, slope and a proxy for greenness (Enhanced Vegetation Index). In addition, we utilized a fine-scale land cover map to obtain knowledge regarding the habitat classes that the species occur in and the ones they are known to occur in from published literature. By applying a threshold that provides us a range of suitable habitat (low, medium and high), we compared our maps generated to the EOO and AOO predicted by IUCN for each species.
Results/Conclusions
Our results suggest that (a) species distribution maps utilized by IUCN for threat assessment are largely overestimated and inaccurate; (b) suitable habitat exists outside these areas that warrant field surveys and (c) many species have shown significant expansion outside these areas. We suggest immediate extinction risk assessment using the data that was generated from this study and accordingly, take necessary conservation action. We insist that any GIS map data that is used needs to be validated before threat assessment. Lastly, our study has implications for conservation of avifauna around the globe.