Can species distribution models using remotely sensed variables inform reintroductions? Trialling methods with Carterocephalus palaemon the Chequered Skipper Butterfly

Abstract

Despite large numbers of reintroduction projects taking place and the high cost involved, there is a generally low success rate. Insects in particular are understudied within reintroduction ecology, with guidelines focusing on more iconic vertebrate taxa. Species distribution models (SDMs) examine the associations between species observations and environmental variables to find the conditions in which populations could survive. This study utilises two frequently used SDM approaches, a regression model (general linear model (GLM)) and a machine learning method (MaxEnt) to model habitat suitability for Chequered Skipper, Carterocephalus palaemon, butterflies, which are being reintroduced to Northamptonshire following extinction in England. We look at how SDMs using widespread remotely sensed variables could be used to inform the reintroduction process by finding areas of suitable habitat that were previously overlooked. These remotely sensed variables have the potential to inform reintroductions without extensive on the ground research as they cover huge areas and are widely available. We found that both models are successful in discriminating between presences and absences, using only a limited number of explanatory variables. We conclude that these wide-scale SDMs are useful as a first step in the decision-making process in determining appropriate sites for reintroductions, but that they are less accurate when establishing precisely where species should be placed.