Synopsis: Roughly half of the world’s peatlands are thought to be dominated by Spahgnum peatmosses, and up to 40 species of peatmoss can be found in a single peatland. They coexist by partitioning the habitat into very narrow, species-specific realized niches. Peat moss species tend to be found in a narrow range along two general gradients: a hummock-hollow gradient, describing how high above the water table a species is usually found, and a micronutrient/pH gradient.
Our main question was whether species found in similar microhabitats tend to be phylogenetically related. We used ecological survey data taken from five major studies across the world (Canada and Scandinavia), along with a database of DNA sequences for each species, to estimate whether Sphangum preferences contain phylogenetic signal: do related species have similar microhabitat preferences?
Using a variety of comparative phylogenetic models, we found that the hummock-hollow gradient does show significant phylogenetic signal. We also found that subgenera that mostly contain hummock-preferring species have faster-evolving microhabiat preferences than subgenera that contain mostly hollow-preferring species. The pH gradient has a more complex phylogenetic signal, that may be masked by a sudden change in the evolution of pH preference in one Sphagnum subgenus.
Our results suggest that the classic microtopography and niche partitioning seen in Sphagnum-dominated peatlands is generated by assemblages of related species. This will instruct future studies on the evolution of functional traits in Sphagnum, such as decomposition rate, which will now need to account for phylogenetic history.
My role: This research was part of my PhD dissertation. Several peatland ecologists did much of the hard work for this study, meticulously documenting the microhabitat conditions in thousands of peatland plots all over the world! I was fortunate to use their data to help analyze the microhabitat preferences in an evolutionary context. At publication, the tree I assembled is the most comprehensive in both taxon sampling (41 species) and gene sampling (18 genes, including 7 nuclear genes), though the Shaw lab will surely eclipse that soon with phylogenomic data. I also developed a pipeline for analyzing the 1000 trees sampled from the BEAST analysis using the comparative methods developed in R.