Drivers and spatial patterns of avian defaunation in tropical forests

Aim: Wildlife overexploitation, either for food consumption or for the pet trade, is one of the main threats to bird species in tropical forests. Yet, the spatial distribution and intensity of harvesting pressure on tropical birds remain challenging to quantify. Here, we identify the drivers of hunting-induced declines in bird abundance and quantify the magnitude and the spatial extent of avian defaunation at a pantropical scale.

Location: Pantropical

Methods: We compiled 2968 abundance estimates in hunted and non-hunted sites across the tropics spanning 518 bird species. Using a Bayesian modelling framework, we fitted species’ abundance response ratios to a set of drivers of hunting pressure and species traits. Subsequently, we applied our model to quantify the spatial patterns of avian defaunation across tropical forests and to assess avian defaunation across biogeographic realms, and for species captured for the pet trade or for food consumption.

Results: Body mass and its interactions with hunters accessibility and proximity  to urban markets were the most important drivers of hunting-induced bird abundance declines. We estimated a mean abundance reduction of 12% across the tropics, and that 50% of all tropical forests harbour defaunated avian communities. Large-bodied species and the Indomalayan realm displayed the greatest abundance declines. Further, hotspots of high defaunation extended over 11.1% of the pantropical forest area, with distinct spatial patterns for species captured for the pet trade (Brazil, China and Indonesia) and for food consumption (SE Asia and West Africa).  

Conclusions: Our study emphasizes the role of hunters’ accessibility and the proximity to urban markets as major drivers of bird abundance declines due to hunting and trapping. We further identified hotspots where overexploitation has detrimental effects on tropical birds, encompassing local extinction events, thus underscoring the urgent need for conservation efforts to address unsustainable exploitation for both subsistence and trade.
Methods
We expanded the dataset of hunting impacts on bird populations from Benítez-López et al. (2017) by supplementing additional bird abundance data from local hunting studies through a systematic search of the literature (see details in Supplementary Methods 1). Our final dataset comprises 2968 abundance estimates for 518 tropical bird species at both hunted and non-hunted sites (control) based on 60 local hunting studies (Figure S1, Table SX). Studies that report potential confounding effects, such as habitat loss and logging were not included in our analysis. Changes in abundance due to hunting pressure were subsequently expressed as the response ratio (RR) between the abundance of each bird species (s) in hunted (Xsh) and non-hunted (Xsc) sites within each study (RR =Xsh/Xsc) (Peres & Palacios, 2007; Benítez-López et al., 2017, 2019) RR = 0 then indicates local extinction; 0 < RR < 1, reduction in abundance; RR ≈  1, no changes in abundance and RR > 1, increase in abundance.

We compiled the following information from each study: the geographic coordinates of hunted and unhunted sites in each study, the hunter’s access point to the hunted site (i.e. roads, settlements or rivers), and the motivation for hunting (i.e. subsistence, commercial or both). We further compiled information on different predictors often recognised as drivers of hunting pressure, including the distance to access points, human population density, poverty level, and travel time to major cities, as well as information on factors that modulate species responses to hunting pressure, such as net primary productivity or whether hunting activities took place inside or outside protected areas (Peres, 2000; Brashares et al., 2010; Benítez-López et al., 2017, 2019; Whytock et al., 2018; Bogoni et al., 2020; Scabin & Peres, 2021) (Supplementary Table 2).

Please, note that all continuous predictors were scaled and centered around zero with a SD equal to 1 before model fitting.