Dados do Trabalho

Título

Exploring the Intersection of Landscape, Biodiversity, and Trypanosoma cruzi Infection in Fragmented Atlantic Rainforest Environments: An Analysis Utilizing Machine Learning

Introdução

<p>The landscape transformation can decrease niche-specific mammals diversity, which tend to be replaced by generalist species. If these species are potential reservoirs of <em>T. cruzi</em>, an increase in the transmission of this parasite is expected. Marsupials and rodents serve as models for understanding this effect.</p>

Objetivo (s)

<p>This study aims to investigate the contributions of landscape features, morphological characteristics and the composition of marsupials and rodents fauna to <em>T. cruzi</em> infection (based on parasitological diagnosis and/or serological) in the Atlantic Rainforest, using a machine learning approach.</p>

Material e Métodos

<p>For this purpose, a database with 3,268 occurrences of marsupials and rodents distributed in 21 municipalities was compiled (CEUA LW-39/14 and 13373 MMA/ICMBio/SISBIO). Two spatial delimitation scales were evaluated, 3 km and 10 km buffers, generating 31 and 19 communities, respectively. Fauna diversity was assessed using six metrics, while seven landscape metrics derived from the geometry of land use classes obtained from the MapBiomas platform were utilized.</p>

Resultados e Conclusão

<p>For both buffers higher body mass strongly indicates the likelihood of <em>T. cruzi </em>infection. Tail/Body ratio also plays a significant role, although lower and higher values can impact <em>T. cruzi</em> infection within the 3km, while only lower values have influence within a 10km. In the 3km, both Fisher's Alpha and the Menhinick index suggest that lower diversity and species richness are associated with a risk of <em>T. cruzi</em> infection, meaning areas with fewer diverse species are more susceptible to positivity. Conversely, In the 10km range a higher proportion of rodents is a crucial predictor, with higher values indicating an increased likelihood of infection. For 3km, higher values of Urban fractal dimension, reflecting increased complexity in urban landscapes, favor the identification of infection for <em>T. cruzi</em>. Similarly, a higher percentage of similar adjacencies in urban environments is associated with an increased likelihood of <em>T. cruzi</em> positivity for 10km. In concordance, lower values of fractal dimension in forest environments are linked to an increased <em>T. cruzi</em> positivity for 10km. These results highlight the complex interactions between host ecology, landscape, and <em>T. cruzi</em> infection dynamics in fragmented environments. Thus, these findings can help prioritize intervention areas in the enzootic transmission cycle of <em>T. cruzi</em>, especially where increased transmission is expected due to habitat fragmentation.</p>

Palavras Chave

Trypanosoma cruzi; Landscape; Biodiversity; Machine Learning.