How ancient forest fragmentation and riparian connectivity generate high levels of genetic diversity in a microendemic Malagasy tree

Abstract

Understanding landscape changes is central to predicting evolutionary trajectories and defining conservation practices. While human-driven deforestation is intense throughout Madagascar, exceptions in areas such as the Loky-Manambato region (north) raise questions regarding the causes and age of forest fragmentation. The Loky-Manambato region also harbours a rich and endemic flora, whose evolutionary origin remains poorly understood. We assessed the genetic diversity of an endangered microendemic Malagasy olive species (Noronhia spinifolia Hong-Wa) to better understand the vegetation dynamics in the Loky-Manambato region and its influence on past evolutionary processes. We characterized 72 individuals sampled across eight forests through nuclear and mitochondrial restriction-associated DNA sequencing data and chloroplast microsatellites. Combined population and landscape genetics analyses indicate that N. spinifolia diversity is largely explained by the current forest cover, highlighting a long-standing habitat mosaic in the region. This sustains a major and long-term role of riparian corridors in maintaining connectivity across these antique mosaic habitats, calling for the study of organismal interactions that promote gene flow.

Keywords: Madagascar; Malagasy olive; RADseq; connectivity; cpSSR; gene flow; habitat loss and fragmentation; habitat mosaic; landscape genetics; mitochondrial DNA.

FIGURE 1

Map of Noronhia spinifolia sampling in the Loky‐Manambato (LM) region. The small black points represent samples collected for all Noronhia species (~30 distinct taxa) and illustrate the survey effort conducted in the region. The yellow and red dots represent N. spinifolia samples, with the red dots corresponding to samples included in our genomic analyses. The forest cover is adapted from Hansen et al. (2013). Pixels with <30% tree cover are represented in white to illustrate the presence of riparian forests along streams of the LM region

FIGURE 2

Organellar DNA haplotype network of Noronhia spinifolia. Pie chart size is proportional to the occurrence number of a given haplotype. All edges of equal weight are represented. Distances among haplotypes are represented both through longer edges and the grey scale. The network highlights the huge organellar DNA diversity in N. spinifolia, with only one haplotype shared by individuals from at least two forests. It further shows a limited spatial structure, with, for instance, haplotypes from Solaniampilana and Benanofy grouping together at the bottom of the network

FIGURE 3

Spatial genetic structure of Noronhia spinifolia in the Loky‐Manambato region. ngsadmix ancestry proportions (for K = 3 genetic clusters) are represented either (a) spatially by sampling site, or (b) per individual. Size of pie charts (in a) is proportional to the number of samples per site. Pie shares represent the sums of individual ancestry proportions that are shown in (b). Results are arbitrarily represented for K = 3, according to the likelihood and deltaK results in Figure S8, because this K value best illustrates the continuous pattern of structure inferred using ngsadmix and other approaches

FIGURE 4

Landscape contribution to nuclear gene flow in Noronhia spinifolia. The figure shows the four landscape variables (a–d) probably contributing to N. spinifolia nuclear gene flow (pollen and seeds), the confidence interval of their fitted composite conductance (e, f), the model fit relationship between the optimized resistance distance and the nuclear genetic (g), and the relative effects of the four retained variables (h). This illustrates a major conducting effect of the current forest cover (ptc: Percentage tree cover) on the connectivity of N. spinifolia, and it further shows a positive effect of wind speed in November (wind11) and a composite effect of the topography (positive effect of slope and negative effect of altitude [alt])