One sixth of Amazonian tree diversity is dependent on river floodplains

Abstract

Amazonia's floodplain system is the largest and most biodiverse on Earth. Although forests are crucial to the ecological integrity of floodplains, our understanding of their species composition and how this may differ from surrounding forest types is still far too limited, particularly as changing inundation regimes begin to reshape floodplain tree communities and the critical ecosystem functions they underpin. Here we address this gap by taking a spatially explicit look at Amazonia-wide patterns of tree-species turnover and ecological specialization of the region's floodplain forests. We show that the majority of Amazonian tree species can inhabit floodplains, and about a sixth of Amazonian tree diversity is ecologically specialized on floodplains. The degree of specialization in floodplain communities is driven by regional flood patterns, with the most compositionally differentiated floodplain forests located centrally within the fluvial network and contingent on the most extraordinary flood magnitudes regionally. Our results provide a spatially explicit view of ecological specialization of floodplain forest communities and expose the need for whole-basin hydrological integrity to protect the Amazon's tree diversity and its function.

Fig. 1. Broad-scale geographic and environmental patterning of species turnover across floodplain and adjacent terra firme forest habitats, for várzea–terra firme and igapó–terra firme comparisons.

a, Spatial patterns of species turnover for várzea and igapó, showing a concentration of high species turnover located centrally within the fluvial network. Grey rivers are masked out because they either correspond to a different floodplain habitat or did not meet minimum sampling criteria for analysis. b, Regional differences in seasonal flooding are described as an annual flood wave that originates in Andean headwaters, peaks in central Amazonia and dissipates near the Amazon mouth. Floodplains positioned at the peak of this flood wave are seasonally inundated by the highest-amplitude and longest-lasting floods. LWT, land water thickness. c, Patterning of species turnover of várzea and igapó with surrounding terra firme along the flood wave. The black dashed line shows the lower bound of species turnover with flooding, assessed with quantile regression at τ = 0.1. d, Mapped residuals from quantile regression modelling for várzea and igapó. Throughout much of western Amazonia, species turnover is relatively higher than expected given the lower flooding implied by its headwater position on the flood wave.

Fig. 2. Relationships between species turnover and the relative abundance and richness of floodplain specialists, habitat generalists and spillover from terra firme (terra firme specialists) in várzea and igapó.

With increasing levels of species turnover, floodplain specialists become more dominant, while spillover from terra firme species decreases. The proportions are derived from interpolated compositional grids of várzea and igapó after cross-referencing with the names of the 1,666 species tested for habitat association. The relationships with species turnover are derived from simple least squares models. The coloured boxes indicate the proportion of explained variance (r²) and P values. The trend lines (black) are bounded by coloured bands showing the 95% CIs. Density plots for the relative abundance and richness of each species group are shown in the right margins.

Extended Data Fig. 1. Distribution of inventory data used to create habitat-specific compositional grids.

Sampled sites include 1,705 mostly 1-ha tree inventory plots with full information on species composition and abundances. Plots were classified as terra firme (n = 1,250, 73%), várzea (n = 271, 16%), or igapó (n = 184, 11%), following habitat designations of ATDN contributors.

Extended Data Fig. 2. Schematic of the methods used to compare floodplain and terra firme tree compositions, illustrated for two grid cells.

(a) Forest plot inventories (colored dots) were separated into várzea, igapó and terra firme categories, and species abundance information for separate várzea, igapó and terra firme grids was calculated at each 1-degree cell (only two shown), using distance-weighted interpolations of inventory plot data from an approximately 300 km circular window (red lines). (b) Floodplain and terra firme grids were overlaid and species turnover computed at analogous (vertically overlapping) cells. (c) Spatially-continuous grids of species turnover for várzea-terra firme and igapó-terra firme comparisons. The number of cells where species turnover is calculated depends on the spatial distribution of floodplain inventories and how it overlaps with terra firme inventories. This included 301 cells and 347 cells for várzea-terra firme and igapó-terra firme comparisons, respectively. In an alternative procedure of calculating species turnover, the interpolation step was excluded and cell compositional data was pooled only from plots located inside cells. For this second approach, the resulting number of cells where species turnover was calculated was 25 and 22 for várzea-terra firme and igapó-terra firme comparisons, respectively.

Extended Data Fig. 3. Comparison of flooding relationships with species turnover using two alternative procedures for populating cell compositional data.

While interpolating species abundances maximizes the number of cells where species turnover can be calculated, it introduces spatial autocorrelation. On the other hand, pooling inventories within grid cells reduces the number of cells where species turnover can be calculated, but it maintains spatial independence among cells. We compared both methods to assess the robustness of our results to spatial dependencies. For the approach based on pooling, species cell abundance information was pooled only from inventories located inside individual grid cells, rather than interpolated from inventories from a larger 300 km circular window, in order to avoid residual spatial autocorrelation. Quantile regression slopes (at tau = 0.1) and their 95% confidences intervals are shown for várzea- and igapó-terra firme. The lower bounds of várzea-terra firme species turnover with flooding are statistically equivalent between pooled compositional data (slope ± 95% CI = 1.29 × 10⁻² ± 1.21 × 10⁻², t = 2.20, n = 25, p = 0.038) and interpolated data (slope ± 95% CI = 1.21 × 10⁻² ± 2.48 × 10⁻³, t = 9.57, n = 301, p < 0.001). The lower bounds for igapó-terra firme are likewise similar between pooled (slope ± 95% CI = 1.54 × 10⁻² ± 1.19 × 10⁻², t = 2.66, n = 22, p = 0.015) and interpolated methods (slope ± 95% CI = 1.05 × 10⁻² ± 2.60 × 10⁻³, t = 7.86, n = 347, p < 0.001). Slopes from all comparisons were significant (p < 0.05) and had overlapping 95% confidence intervals.