Eric Bandou

Fine‐scale genetic structure and gene dispersal inferences in 10 Neotropical tree species

The extent of gene dispersal is a fundamental factor of the population and evolutionary dynamics of tropical tree species, but directly monitoring seed and pollen movement is a difficult task. However, indirect estimates of historical gene dispersal can be obtained from the fine‐scale spatial genetic structure of populations at drift–dispersal equilibrium. Using an approach that is based on the slope of the regression of pairwise kinship coefficients on spatial distance and estimates of the effective population density, we compare indirect gene dispersal estimates of sympatric populations of 10 tropical tree species. We re‐analysed 26 data sets consisting of mapped allozyme, SSR (simple sequence repeat), RAPD (random amplified polymorphic DNA) or AFLP (amplified fragment length polymorphism) genotypes from two rainforest sites in French Guiana. Gene dispersal estimates were obtained for at least one marker in each species, although the estimation procedure failed under insufficient marker polymorphism, limited sample size, or inappropriate sampling area. Estimates generally suffered low precision and were affected by assumptions regarding the effective population density. Averaging estimates over data sets, the extent of gene dispersal ranged from 150 m to 1200 m according to species. Smaller gene dispersal estimates were obtained in species with heavy diaspores, which are presumably not well dispersed, and in populations with high local adult density. We suggest that limited seed dispersal could indirectly limit effective pollen dispersal by creating higher local tree densities, thereby increasing the positive correlation between pollen and seed dispersal distances. We discuss the potential and limitations of our indirect estimation procedure and suggest guidelines for future studies.

Limited pollen dispersal and biparental inbreeding in Symphonia globulifera in French Guiana

In this paper, we report a study of the mating system and gene flow of Symphonia globulifera, a hermaphroditic, mainly bird-pollinated tree species with a large geographic distribution in the tropical Americas and Africa. Using three microsatellites, we analysed 534 seeds of 28 open pollinated families and 164 adults at the experimental site ‘Paracou’ in French Guiana. We observed, compared to other tropical tree species, relatively high values for the effective number of alleles. Significant spatial genetic structure was detected, with trees at distances up to 150 m more genetically similar than expected at random. We estimated parameters of the mating system and gene flow by using the mixed mating model and the TwoGener approach. The estimated multilocus outcrossing rate, tm, was 0.920. A significant level of biparental inbreeding and a high proportion of full-sibs were estimated for the 28 seed arrays. We estimated mean pollen dispersal distances between 27 and 53 m according to the dispersal models used. Although the adult population density of S. globulifera in Paracou was relatively high, the joint estimation of pollen dispersal and density of reproductive trees gave effective density estimates of 1.6 and 1.3 trees/ha. The parameters of the mating system and gene flow are discussed in the context of spatial genetic and demographic structures, flowering phenology and pollinator composition and behaviour.

Fine-scale spatial genetic structure of eight tropical tree species as analysed by RAPDs

The fine-scale spatial genetic structure of eight tropical tree species (Chrysophyllum sanguinolentum, Carapa procera, Dicorynia guianensis, Eperua grandiflora, Moronobea coccinea, Symphonia globulifera, Virola michelii, Vouacapoua americana) was studied in populations that were part of a silvicultural trial in French Guiana. The species analysed have different spatial distribution, sexual system, pollen and seed dispersal agents, flowering phenology and environmental demands. The spatial position of trees and a RAPD data set for each species were combined using a multivariate genetic distance method to estimate spatial genetic structure. A significant spatial genetic structure was found for four of the eight species. In contrast to most observations in temperate forests, where spatial structure is not usually detected at distances greater than 50 m, significant genetic structure was found at distances up to 300 m. The relationships between spatial genetic structure and life history characteristics are discussed.

Long‐distance pollen flow and tolerance to selfing in a neotropical tree species

Outcrossing rates, pollen dispersal and male mating success were assessed in Dicorynia guianensis Amshoff, a neotropical tree endemic to the Guiana shield. All adult trees within a continuous area of 40 ha (n = 157) were mapped, and were genotyped with six microsatellite loci. In addition, progenies were genotyped from 22 mature trees. At the population level, the species was mostly outcrossing (tm = 0.89) but there was marked variation among individuals. One tree exhibited mixed mating, confirming earlier results obtained with isozymes that D. guianensis can tolerate selfing. A Bayesian extension of the fractional paternity method was used for paternity analysis, and was compared with the neighbourhood method used widely for forest trees. Both methods indicated that pollen dispersal was only weakly related to distance between trees within the study area, and that the majority (62%) of pollen came from outside the study stand. Using maximum likelihood, male potential population size was estimated to be 1119, corresponding to a neighbourhood size of 560 hectares. Male mating success was, however, related to the diameter of the stem and to flowering intensity assessed visually. The mating behaviour of D. guianensis is a combination of long‐distance pollen flow and occasional selfing. The species can still reproduce when it is extremely rare, either by selfing or by dispersing pollen at long distances. These results, together with the observation that male mating success was correlated with the size of the trees, could be implemented in management procedures aiming at regenerating the species.

Nuclear and chloroplast genetic structure indicate fine-scale spatial dynamics in a neotropical tree population

Dicorynia guianensis is a canopy tree, endemic to the tropical rain forest of French Guiana. We compared generational and spatial genetic structure for maternally and biparentally inherited markers in two cohorts (adult and seedling) in order to infer processes shaping the distribution of genetic diversity. The study was conducted on a 40 ha study plot located at Paracou near Kourou, where 172 adults trees and 375 saplings were sampled. Aggregation of trees was therefore suggested at different distances, ranging from 100 to 400 m. There was a strong link between demographic and genetic spatial structures at small distances (less than 100 m) that is likely to be the consequence of restricted seed dispersal. Genetic differentiation was more pronounced between spatial aggregates than between cohorts. Despite the spatial differentiation, the species was able to maintain high levels of diversity for maternal genomes, suggesting rapid turnover of aggregates. Spatial autocorrelation was larger for chloroplast than nuclear markers indicating a strong asymmetry between pollen and seed flow. Fixation indices indicated a lower heterozygote deficiency for the adults, maybe because of gradual elimination of selfed trees. Genetic relatedness at lower distances was higher in adult trees than in saplings, as a result of generation overlapping in the adult cohort. Overall, our results confirm earlier biological knowledge about the dispersion mechanisms of the species, and lead to an enhanced role of spatial processes in the dynamics of genetic diversity of D. guianensis.

Spatial and temporal distribution of chloroplast DNA polymorphism in a tropical tree species

The level and the spatial organization of chloroplast DNA polymorphism were investigated in Dicorynia guianensis Hamshoff (Caesalpiniaceae) at different spatial and temporal scales. D. guianensis is a canopy tree of the rain forest that is distributed throughout the Guiana plateau in small aggregates. Twelve different haplotypes were identified using restriction analysis of polymerase chain reaction (PCR) amplified fragments of the chloroplast genome. When populations from different areas of French Guiana were compared, a clear geographical pattern of haplotype frequencies was identified along the Atlantic coast. This pattern is most likely the result of the restriction–expansion dynamics of the tropical forest during the Quaternary. At the local level, D. guianensis was characterized by a high level of within population diversity. Maintenance of within population diversity results from the dynamics of the aggregates; stochastic demography associated with the turnover of aggregates generates genetic differentiation among them. At the stand level, a strong spatial aggregation of haplotypes persisted from the adult to the seedling cohort indicating limited seed flow. There was also a strong difference in levels of diversity between the cohorts which suggested that recruitment over several years is needed in order to maintain genetic diversity during regeneration.

Small scale spatial genetic structure of six tropical tree species in French Guiana

The small scale spatial genetic structure of six tropical tree species (Carapa procera, Chrysophyllum sanguinolentum, Dicorynia guianensis, Eperua grandiflora, Virola michelii, Vouacapoua americana) was studied for populations located within a sylvicultural trial in French Guiana. The analysed species have different biological and ecological characteristics with respect to spatial distribution, sexual system, pollen and seed dispersal agents, flowering phenology, and they differ in their environmental demands. The spatial position of trees and at least one data set on the genetic information for allozymes, cp-DNA marker, RAPDs and AFLPs served as basis for the analysis. As estimator for spatial genetic structure, Moran’s index was computed for allozymes and cp-DNA markers, while for RAPDs and AFLPs mean Tanimoto distances between pairs of trees in different spatial distance classes were calculated. Each species showed significant spatial structure for at least one type of gene marker. The expansion of significant spatial genetic structures was relatively large and varied between 60m up to 800m. The strongest spatial autocorrelation was found for maternally inherited cp-DNA marker. Due to distinctly higher level of genetic variation, AFLPs and RAPDs were more sensitive to detect spatial structure than allozymes. Relations between the spatial genetic structures and the species’ life characteristics are discussed.