O futuro das espécies de plantas invasoras na Caatinga: projeções climáticas e visualização interativa

Abstract

The increase in invasive plant species, which can thrive in new environments despite the challenges posed by climate change, represents a significant ecological concern. In this context, the Caatinga, a seasonally dry tropical forest spanning 833,000 km², emerges as a particularly vulnerable ecosystem. Climatic fluctuations may eventually render it even more susceptible to biological invasions, opening up space for restructuring, such as the formation of communities without current analogs. To study the complexities of this vulnerability, we utilized Species Distribution Models (SDMs), projecting the future distribution of 19 plant species already recognized as invasive in the Caatinga under two contrasting climate scenarios, divided into four time intervals up to the year 2100. The modeling procedure was carried out using the Maximum Entropy algorithm (MaxEnt), which is a statistical technique used to model species distributions based on presence data and environmental variables. Presence data and variables were obtained from online databases. These data underwent treatment steps, including the removal of duplicate points and reprojection, clipping, and creating a bias surface. After creating individual models for each species, we overlaid them, resulting in consensus models that offer a more integrated view of the potentially suitable invasion areas in the Caatinga. To further explore the models, we grouped the species into trees, shrubs, herbaceous plants, and grasses. Additionally, we explored the possibility of forming communities without current analogs by the end of the century. The analyses reveal that the same areas, with average annual temperatures between 24 and 27°C and higher precipitation, are more prone to invasions in all time intervals. Trees and grasses were the life forms with a greater amount of invasion-suitable areas. Megathyrsus maximus Jacq., Cenchrus ciliaris L., Cenchrus echinatus L., Dodonaea viscosa Jacq., and Prosopis juliflora (Sw.) DC. are the species with the highest invasion potential in both scenarios. Despite the significant projection of loss of suitable invasion areas in the Caatinga (-46.3% optimistic, -30.1% pessimistic, between the first and last time intervals), specific hotspots, such as higher altitude and more humid areas, could still form communities without analogs, particularly under ecosystemic stress induced by climate change. This raises significant concerns about the survival of native species, which are pressured by their physiological and resource limitations. In this context, the importance of relevant measures to protect the Caatinga ecosystem and halt its biodiversity loss is highlighted. This is in line with the achievement of UN Sustainable Development Goal 15 (Life on Land) of the 2030 Agenda, by adding new knowledge that can support the protection of this unique ecosystem. As a means to facilitate the dissemination of the generated results and their application in public policies, an interactive R/Shiny application was also developed to compose the second chapter of this dissertation. The R programming language was used, implementing essential libraries and functions, such as the Shiny package and complementary web technologies. SDMs for each species, by life form, and the resulting modeling consensuses were stored and organized in specific directories to facilitate script reading and promote visualization on the internet. Dynamic data (tables) were structured in separate directories to ensure efficiency in management. The hosting service identification was based on publications made between 2014 and December 2023. The initial screen design was defined to ensure a user-friendly experience. This application can be observed at https://llle.shinyapps.io/SDM_App_2024/. This tool serves as a connection between academic research and its practical application. It allows users to visualize all figures generated in species distribution modeling, understand the application development process, and thus apply this knowledge in their own research.