Genetic diversity and population divergences of an indigenous tree (Coffea mauritiana) in Reunion Island: role of climatic and geographical factors

Abstract

Oceanic islands are commonly considered as natural laboratories for studies on evolution and speciation. The evolutionary specificities of islands associated with species biology provide unique scenarios to study the role of geography and climate in driving population divergence. However, few studies have addressed this subject in small oceanic islands with heterogeneous climates. Being widely distributed in Reunion Island forest, Coffea mauritiana represents an interesting model case for investigating patterns of within-island differentiation at small spatial scale. In this study, we examined the genetic diversity and population divergences of C. mauritiana using SNP markers obtained from 323 individuals across 34 locations in Reunion Island. Using redundancy analysis, we further evaluated the contribution of geographic and climatic factors to shaping genetic divergence among populations. Genetic diversity analyses revealed that accessions clustered according to the source population, with further grouping in regional clusters. Genetic relationships among the regional clusters underlined a recent process of expansion in the form of step-by-step colonization on both sides of the island. Divergence among source populations was mostly driven by the joint effect of geographic distance and climatic heterogeneity. The pattern of isolation-by-geography was in accordance with the dispersal characteristics of the species, while isolation-by-environment was mostly explained by the heterogeneous rainfall patterns, probably associated with an asynchronous flowering among populations. These findings advance our knowledge on the patterns of genetic diversity and factors of population differentiation of species native to Reunion Island, and will also usefully guide forest management for conservation.

Fig. 1

Sampling localities and climates on Reunion Island. a Provenance of source populations of C. mauritiana on Reunion Island. PdN Piton des Neiges, PdF Piton de la Fournaise. b Spatial rainfall distribution over Reunion Island

Fig. 2

Unweighted NJ tree based on simple matching distances estimated from 2251 SNP markers, among 24 coffee accessions. Clades supported by a bootstrap value of 100% are indicated by a blue dash

Fig. 3

Spatial genetic structure of 323 C. mauritiana trees across 34 sampling sites in Reunion Island, estimated from 3953 SNP markers. a Unweighted NJ tree based on simple matching distances. Correspondence between the population number and the name of the population are listed in Table 1. Colors distinguish the main geographical sampling regions. Red: east and north-east; Blue: north-west; Yellow and brown: west; Green: south. b Group assignment from sNMF at K = 7. Biogeographic regions associated with the clusters are represented by seven colors. Each individual tree is represented by a bar and coefficient ancestry relative to each cluster is indicated by colored segments. W western genetic group, NE north-eastern genetic group, NW north-western genetic group, E eastern genetic group, SW south-western genetic group, SE south-eastern genetic group, C central genetic group. c Spatial representation of the genetic structure of C. mauritiana populations obtained from sNMF analysis (K = 7). Each sampling site is represented by a dot. Colors distinguish the genetic groups inferred from sNMF. Yellow stars indicate source populations showing admixed individuals

Fig. 4

Results of TreeMix analysis of the phylogenic relationships among the seven clusters defined by sNMF analysis. C. myrtifolia (MYR) and C. bernardiniana (BER) served as outgroup populations. E eastern cluster, NE north-eastern cluster, SE south-eastern cluster, C central cluster, SW south-western cluster, W western cluster, NW north-western cluster. a Maximum-likelihood tree of 323 C. mauritiana divided into the seven population groups included in the analysis. Horizontal branch lengths are proportional to the amount of genetic drift that has occurred in each branch. b Residual heatmap from the tree in a. Pairs with high residual values are candidates for migration events. c Maximum- likelihood tree that best fits the data and its associated residual heatmap d. The migration weight indicates the proportion of ancestry deriving from the migration edge

Fig. 5

Variance partitioning results of dbRDA analyses. IBD comprises four geographic vectors (geo1, geo2, geo3, geo4). IBE comprises three environmental variables: Tmean mean of average daily temperature, CDD mean number of consecutive dry days, PET average monthly potential evapotranspiration. a Variance partitioning between IBD and IBE. The overlapping zone represents the intersection between IBD and IBE (IBD ∩ IBE). b Variance partitioning between IBD, IBE excluding CDD and CDD. Values below zero are not shown