Revisiting the historical distribution of Seasonally Dry Tropical Forests: new insights based on palaeodistribution modelling and palynological evidencegeb

Abstract

Aim To investigate the potential distribution of Seasonally Dry Tropical Forests (SDTFs) during the Quaternary climatic fluctuations; to reassess the formerly proposed ‘Pleistocenic arc hypothesis’ (PAH); and to identify historically stable and unstable areas of SDTF distributions in the light of palaeodistribution modelling. Location SDTFs in lowland cis-Andean eastern-central South America. Methods We first developed georeferenced maps depicting the current distributional extent of SDTFs under two distinct definitions (narrow and broad). We then generated occurrence datasets, which were used with current and past bioclimatic variables to predict SDTF occurrence by implementing the maximum entropy machine-learning algorithm. We obtained historical stability maps by overlapping the presence/absence projections of each of three climatic scenarios [current, 6 kyr bp during the Holocene, and 21 kyr bp during the Last Glacial Maximum (LGM)]. Finally, we checked the consistencies of the model prediction with qualitative comparisons of vegetation types inferred from available fossil pollen records. Results The present-day SDTF distribution is disjunct, but we provide evidence that it was even more disjunct during the LGM. Reconstructions support a progressive southward and eastward expansion of SDTFs on a continental scale since the LGM. No significant expansion of SDTFs into the Amazon Basin was detected. Areas of presumed long-term stability are identified and confirmed (the three nuclear regions, Caatinga, Misiones and Piedmont, plus the Chiquitano region), and these possibly acted as current and historical refugial areas. Main conclusions The LGM climate was probably too dry and cold to support large tracts of SDTF, which were restricted to climatically favourable areas relative to the present day (in contrast with the PAH, as it was originally conceived). Expansions of SDTFs are proposed to have occupied the southern portion of Caatinga nucleus more recently during the early–middle Holocene transition. We propose an alternative scenario amenable to further testing of an earlier SDTF expansion (either at the Lower Pleistocene or the Tertiary), followed by fragmentation in the LGM and secondary expansion in the Holocene. The stability maps were used to generate specific genetic predictions at both continental and regional scales (stable areas are expected to have higher genetic diversity and endemism levels than adjacent unstable areas) that can be used to direct field sampling to cover both stable (predicted refugia) and unstable (recently colonized) areas. Lastly, we discuss the possibility that SDTFs may experience future expansion under changing climate scenarios and that both stable and unstable areas should be prioritized by conservation initiatives.