Aim: To (a) assess the environmental suitability for rainforest tree species of Moraceae and Urticaceae across Amazonia during the Mid-Late Holocene and (b) determine the extent to which their distributions increased in response to long-term climate change over this period.
Location: Amazonia.
Methods: We used MaxEnt and inverse distance weighting interpolation to produce environmental suitability and relative abundance models at 0.5-degree resolution for tree species of Moraceae and Urticaceae, based on natural history collections and a large plot dataset. To test the response of the Amazon rainforest to long-term climate change, we quantified the increase in environmental suitability and modelled species richness for both families since the Mid-Holocene (past 6,000 years). To test the correlation between the relative abundance of these species in modern vegetation versus modern pollen assemblages, we analysed the surface pollen spectra from 46 previously published paleoecological sites.
Results: We found that the mean environmental suitability in Amazonia for species of Moraceae and Urticaceae showed a slight increase (6.5%) over the past 6,000 years, although southern ecotonal Amazonia and the Guiana Shield showed much higher in- creases (up to 68%). The accompanied modelled mean species richness increased by as much as 120% throughout Amazonia. The mean relative abundance of Moraceae and Urticaceae correlated significantly with the modern pollen assemblages for these families.
Main Conclusions: Increasing precipitation between the Mid- and Late Holocene ex- panded suitable environmental conditions for Amazonian humid rainforest tree spe- cies of Moraceae and Urticaceae, leading to rainforest expansion in ecotonal areas of Amazonia, consistent with previously published fossil pollen data.

The occurrence dataset of Moraceae and Urticaceae species was downloaded from the Global Biodiversity Information Facility (GBIF, http://www.gbif.org/, January 2018), using the function ‘gbif’ from the R package ‘dismo’ (Hijmans, Phillips, Leathwick, & Elith, 2017), based on the most recent list of Amazonian tree species (ter Steege et al., 2019). We assigned all individuals to species level. All species occurrence were checked for inconsistencies using a cleaning pipeline (Gomes et al., 2018), removing all records with: imprecise georeferences, duplicates at 0.5-degree spatial resolution, records assigned to capital cities or centroids of provinces, latitude equal to longitude or equal to exactly zero. We used occurrence data from the entire Neotropics to prevent SMD deficiencies that are associated with models based on a species' partial geographical range (Raes, 2012). Species with a small number of records (<6) were not used as they may develop inaccurate predictions (van Proosdij, Sosef, Wieringa, & Raes, 2016).