Inferring community assembly processes from mangrove species–area relationships

The increasing species–area relationship (SAR) is a nearly universal ecological law. But recent theory has predicted that in systems with low large-scale diversity the law should be violated and the SAR should be nearly flat at intermediate scales, with species richness roughly constant at some value typically greater than one. We tested this prediction using a global dataset of mangrove trees—a species-poor group. We used a published global dataset of mangrove tree distributions to construct an SAR spanning local to global scales. We found that over a large range of scales (≈10^-4 to 10^6 km^2) the SAR was close to flat, in stark contrast to a classical power-law SAR, which would predict roughly a 300-fold change in species richness over these scales. Importantly, species richness was not simply equal to the minimum value of one or the maximum value of global mangrove richness over these scales, either of which possibilities would be reconcilable with the classical theory, but instead was maintained at an average value of between two and three species. Our theoretical interpretation of the results is that there are two to three stabilising niches (i.e., niches that would allow two to three species to stably coexist without substantial immigration) for mangrove trees in typical coastal settings and that immigrant propagule diversity is in most cases too low (because of low mangrove tree metacommunity diversity) for there to be more species than niches. Only at scales greater than ≈10^6 km^2 is the diversity of immigrants typically sufficient to yield more species than niches. We speculate that in other systems, local niche diversity may be similarly low but that the nearly flat SAR phase is hidden by immigration from diverse source pools.

Methods
We based our dataset on a recently compiled large field-based dataset on mangrove forest structure and biomass (Rovai et al. 2021, Rovai 2023). In its raw form, the dataset is a compilation of summary data for 3,497 mangrove plots distributed globally across 67 countries, synthesised from approximately 500 different sources (both published and unpublished). We removed 1,433 plots for which mangrove species identities and species richness were not available, leaving 2,064 plots. A total of 54 mangrove and mangrove-associated tree taxa were represented at these plots (i.e., roughly 74% of the global total).

Our goal was to construct SARs that are, as far as possible, based on accurate taxonomy, based on true mangrove tree species, and representative of natural conditions without anthropogenic influence. To this end we performed several data curation steps. We updated the name of Avicennia lanata to Avicennia rumphiana. We removed from the dataset species that are not considered true mangroves and we merged synonymous taxa (as per the World Register of Marine Species). This resulted in the purging of three species considered mangrove associates (Cerbera odollam, Intsia bijuga, and Pterocarpus officinalis) and the merger of five taxa with synonymous taxa (Avicennia africana was absorbed into Avicennia germinans; Bruguiera decandra into Ceriops decandra; Excoecaria ovalis into Excoecaria agallocha; Rhizophora brevistyla into Rhizophora harrisonii; and Xylocarpus mekongensis into Xylocarpus moluccensis). After these changes, the number of species in our dataset was reduced to 46. In addition, the dataset contained errors in which the wrong mangrove species was recorded in a small number of plots. This was determined by checking for distributional outliers against known mangrove distribution patterns by biogeographical region (Spalding, 2010). There were 20 plots with such errors (Supplemental Information S1). These errors were corrected to account for likely misidentifications where possible, but two plots with particularly dubious occurrences (Laguncularia racemosa in a plot in the Philippines and Avicennia marina in a plot in Venezuela) were deleted.

We removed from the database a further seven mangrove plots that had no remaining species after deletions. Plots containing only introduced mangrove species were also removed from the dataset. There were ten such plots, all in Hawaii. Our final dataset thus comprised 2,045 plots.

Rovai, A. S. 2023. Mangrove forest structure dataset associated to Rovai et al. 2021 in GEB. figshare: https://doi.org/10.6084/m9.figshare.13570601.v1

Rovai, A. S., R. R. Twilley, E. Castañeda-Moya, S. R. Midway, D. A. Friess, C. C. Trettin, J. J. Bukoski, A. E. L. Stovall, P. R. Pagliosa, A. L. Fonseca, R. A. Mackenzie, A. Aslan, S. D. Sasmito, M. Sillanpää, T. G. Cole, J. Purbopuspito, M. W. Warren, D. Murdiyarso, W. Mofu, S. Sharma, P. H. Tinh, and P. Riul. 2021. Macroecological patterns of forest structure and allometric scaling in mangrove forests. Global Ecology and Biogeography 30:1000-1013.