Climate change and bird extinctions in the Amazon

current environmental laws, and a tendency for deforestation across the Amazon basin based on the historical deforestation rate and its variations from 1997 to 2002. On the other hand, GOV is an optimistic future deforestation scenario for 2050. It depicts a context where there would be more investments in sustainable development, greater compliance with environmen- tal legislation, expansion of protected areas in the Amazon basin, besides taking satellite infor- mation into account in environmental licensing [26]. As our target taxa are dependent on forests, we expect that deforestation may affect the col- onization potential of climatically suitable areas in the future since species tend to behave in different ways in the face of climate change and deforestation [24, 73]. Thus, we created differ- ent dispersal contexts to compare the effect of climate change on species distribution in the future, considering unlimited, limited, and no dispersal events. In the first framework, we con- sider unlimited dispersal, which allows species to colonize all new suitable environments. On the second one, dispersal is limited, thus reducing species ability to colonize the newly suitable areas. For this, we applied the MSDM method aforementioned. In the third scenario, we con- sider that the species cannot disperse into new suitable habitat, consequently being restricted to their current distribution. These three dispersal schemes combine either ‘optimistic’ or ‘pes- simistic’ projections of climate change and deforestation. Therefore, our optimistic projections

combine the three dispersal contexts with the rcp4.5 greenhouse gas emission model and the GOV deforestation model (Mitigation Scenario). In contrast, the pessimistic predictions com- bine them with the rcp8.5 greenhouse gas emission and the BAU deforestation models (Busi- ness-as-usual Scenario). Stacked species distribution models To obtain species richness maps in the different environmental scenarios, we performed stacked-species distribution models (S-SDM) of all the species for both current and future pro- jections. This method has already shown to be effective in several different situations [74, 75, 76]. For all stackings, we used the “raster calculator” tool available in QuantumGis 2.8 software. Results We obtained a total of 596 unique occurrence records for the 16 target taxa with Dendrexe- tastes rufigula paraensis having the least number of occurrence records (7) and Thamnophilus aethiops incertus with the highest number of records (80) (S2 Table). The mean Sorensen index value was 0.71, ranging between 0.66 and 1 (S2 Table). The species Threnetes leucurus medianus was removed from the analysis because it presented results that may have been biased by the mathematical models in the dispersion calculation, due to the distribution of its occurrence records. By overlapping the taxon distribution models with the current and future forest remnant predictions for each climatic scenario, we observed an average loss of 73.80% of suitable areas in either in the mitigation and the business-as-usual scenarios (Table 2).

Among the most threatened taxa according to the list of Brazilian endangered species [35], Piculus paraensis lost its total suitable area in five combinations of dispersal-climate- Fig 2. Concept map of the methods used in this work. (A) the study area and the taxa information used to construct the models. (B) The production of species distribution models from the RCPs, selection of climate variables, dispersal frameworks, and the algorithms used to produce the models. (C) the ensemble approach to generate a consensus model for both present and future scenarios. (D) the overlap of species distribution models with local current forest cover (2017) and two future deforestation scenarios (2050). (E) The endemic taxa final model for the Bele´m area of endemism (BAE). https://doi.org/10.1371/journal.pone.0236103.g002 PLOS ONE Bird extinctions in the Amazon PLOS ONE | https://doi.org/10.1371/journal.pone.0236103 July 17, 2020 7 / 18