Life history induces markedly divergent insect responses to habitat loss

Colares, Lucas F., Peres, Carlos A. and Dambros, Cristian S. (2026) Life history induces markedly divergent insect responses to habitat loss. Journal of Animal Ecology, 95 (1). pp. 54-64. ISSN 0021-8790

Microsoft Word (rba08-Journal_of_Animal_Ecology) - Published Version Available under License Creative Commons Attribution. Download (1MB)

Abstract

Habitat loss poses a major threat to tropical biodiversity, but its effects on distinct taxa remain unclear. Furthermore, most studies have failed to investigate the effects of habitat loss for taxa with contrasting life histories, potentially underestimating those impacts. Here, using an unprecedented sampling effort, we investigated the effects of forest amount on the diversity, composition and size structure of Amazonian terrestrial and aquatic insects. We sampled the insect fauna across Earth's largest man-made forest archipelago 36 years after impoundment (Balbina reservoir, Central Amazon, Brazil) using 236 sticky traps placed on forest islands, the open-water matrix and adjacent continuous forests. Using fivefold cross-validated computer vision models, we identified and measured 22,471 individual insects. To consider sampling bias on diversity estimation, we used individual-based rarefaction to partition diversity into components that explained community evenness and regional species pool size. We also applied coverage-based rarefaction to estimate changes in community composition, reducing potential bias. Low forest amount led to low dominance of terrestrial insects; conversely, it boosted populations of aquatic insects. We report similar effects of forest cover on regional species pool size of aquatic and terrestrial insects, highlighting the importance of large tracts of forest within the landscape to foster diverse communities. Large terrestrial insects were most likely to disperse across the inhospitable floodwater matrix compared to their smaller counterparts. Future studies should consider multi-taxa approaches to properly quantify impact estimates of land-use change on biodiversity, which can diverge widely depending on species life history traits. Generalizations and any target conservation action cannot be made without explicitly considering how forest cover can affect species depending on their life history traits.

Item Type:	Article
Additional Information:	DATA AVAILABILITY STATEMENT: Processed data available from the Zenodo repository (https://doi.org/10.5281/zenodo.15238078; Colares, 2025c). We also provide the raw images of the sticky traps (https://doi.org/10.6084/m9.figshare.23823591; Colares, 2025d), the fivefold dataset with insect images used for deep learning training (https://doi.org/10.6084/m9.figshare.28688198; Colares, 2025a) and the five trained models to conduct inference on insect taxonomy using deep learning (https://doi.org/10.6084/m9.figshare.28820993; Colares, 2025b).
Uncontrolled Keywords:	amazon,artificial intelligence,balbina,body size,deep learning,fragmentation,ecology, evolution, behavior and systematics,animal science and zoology ,/dk/atira/pure/subjectarea/asjc/1100/1105
Faculty \ School:	Faculty of Science > School of Environmental Sciences University of East Anglia Research Groups/Centres > Theme - ClimateUEA
UEA Research Groups:	Faculty of Science > Research Groups > Environmental Biology Faculty of Science > Research Centres > Centre for Ecology, Evolution and Conservation
Related URLs:	http://www.scopus.com/inward/record.url?...
Depositing User:	LivePure Connector
Date Deposited:	31 Mar 2026 15:30
Last Modified:	31 Mar 2026 15:30
URI:	https://ueaeprints.uea.ac.uk/id/eprint/102684
DOI:	10.1111/1365-2656.70117

Downloads

Actions (login required)

View Item