Recovering species demographic history from multi-model inference: the case of a Neotropical savanna tree species

Abstract

Background: Glaciations were recurrent throughout the Quaternary and potentially shaped species genetic structure worldwide by affecting population dynamics. Here, we implemented a multi-model inference approach to recover the distribution dynamics and demographic history of a Neotropical savanna tree, Tabebuia aurea (Bignoniaceae). Exploring different algorithms and paleoclimatic simulations, we used ecological niche modelling to generate alternative hypotheses of potential demographic changes through the last glacial cycle and estimated genetic parameters using coalescent modelling.

Results: Comparing predictions from demographic hypotheses with genetic parameters of modern populations, our findings revealed a likely scenario of population decline, with spatial displacement towards Northeast Brazil from the last glacial maximum to the mid-Holocene. Subsequently, populations expanded in response to the return of the climatically suitable conditions in Central-West Brazil. Nevertheless, a wide historical refugium across Central Brazil likely maintained large populations connected throughout time. The expected genetic signatures from such predicted distribution dynamics are also corroborated by spatial genetic structure observed in modern populations.

Conclusion: By exploring uncertainties inherent in multiple working hypotheses, we have shown that multi-model inference is a fruitful and efficient approach to recover the nature, timing and geographical context of the Tabebuia aurea population dynamic in response to the Quaternary climate changes.

Figure 1

Current geographical distribution of Tabebuia aurea across the Neotropics. (a) 237 occurrence records used in ecological niche modelling; (b) 20 populations sampled for genetic analyses. Area in grey represents the Brazilian savannah in Central Brazil. Details on the sampled populations are provided in Additional file 1, Table S13.

Figure 2

Ecological space of climatic conditions in Neotropics during the LGM (21 ka, blue squares), mid-Holocene (6 ka, red triangles) and the present day (0 ka, green circles). The climatic preferences from current occurrence records of Tabebuia aurea are represented by black dots. Note that the climatic conditions in Neotropics matching the T. aurea's preferences were less available during the LGM than Holocene and present-day, mainly due to temperature decrease, consequently allowing a general scenario of range expansion through the time. The bioclimatic variables were obtained from AOGCM CCSM4.

Figure 3

Climatic suitability for Tabebuia aurea . Maps of consensus representing the potential distribution of species across the Neotropics during the (a) LGM (21 ka), (b) mid-Holocene (6 ka), and (c) present day. Historical refugium (d) shows areas climatically suitable throughout the time.

Figure 4

Uncertainty from the components of ecological niche modelling. (a) Time, (b) Atmosphere–ocean Global Circulation Models (AOGCMs), and (c) algorithms.

Figure 5

Demographic history scenarios simulated for Tabebuia aurea and their palaeodistributional representations. Circles represent hypothetical demes and indicate population stability or shrinkage through the time. LGM, last glacial maximum; Hol, mid-Holocene; Pres, present day; N0, N500 and N1750, effective population size at time t0 (present day), time t500 (500 generations ago, matching mid-Holocene), and time t1750 (1750 generations ago, matching LGM), respectively; Nt, logarithm function for effective population size variation in coalescent simulation. The migration rate was 0.01/generation.