From inselberg to inselberg: Floristic patterns across scales in French Guiana (South America)

- Inselberg vegetation was compared in 22 inselbergs of French Guiana, South America.
- Floristic composition is mostly driven by a spatially-structured environmental gradient.
- Inselbergs show a southward increasing contribution of drought-adapted plant traits.
- Biogeography, environment, history contributes to the change in inselberg vegetation.
- Two groups of inselbergs can be identified, with a northern group of high endemism.

Granitic outcrop vegetation was compared in 22 inselbergs of French Guiana, South America, using RLQ and fourth-corner analyses to identify the main relationships between environmental gradients and plant traits. At the scale of the whole territory the distribution of species and species traits was mostly driven by a spatially-structured gradient embracing regional climate (annual rainfall), forest matrix (canopy openness), and inselberg features (altitude, shape, habitats, summit forest, degree of epiphytism, fire events). Biogeographic, environmental and past historical factors contribute to explain the variation observed at coarse scale and two groups of inselbergs are identified. A first group occupies the southern peneplain in a semi-open forest matrix and exhibits a higher representation of suffrutescent species and climbers, a lower representation of upright shrubs, a lower degree of Guiana Shield endemism, and a higher incidence of human use and autochory. All these features suggest an adaptation to more disturbed environments linked to past climate changes and savannization and to human influences. A second group, characterized by opposite plant traits, occupies the northern part of French Guiana and the far south within a closed forest matrix. Within archipelagos (inselbergs at less than 7 km distance), C-score and Mantel tests revealed a random co-occurrence of plant species and an increase of floristic dissimilarity with distance without any concomitant change in plant traits, respectively, suggesting that spatially-structured stochastic factors (limitation by dispersal) were the driving force of vegetation change at fine scale.