Urbanization influences ecological communities through habitat conversion and fragmentation, changes to environmental conditions, and additions and removals of species by humans. Each of these effects acts as a "filter", limiting (and potentially adding to) the pool of species that can inhabit urban areas. Ecologists have hypothesized how each of these filters will influence the taxonomic, functional, and phylogenetic diversity of city floras and plant communities. I tested these hypotheses using data on plant species composition from 42 riparian forest sites spanning an urbanization gradient in the Research Triangle area of North Carolina, from which I calculated taxonomic, functional, and phylogenetic diversity of each community. I used publicly available spatial data as proxies for habitat conversion (the amount of forest remaining within a 1km radius), fragmentation (the landscape division index, calculated using land cover data), and human preferences (population density from the 2010 census). I also collected data on environmental variables associated with urbanization (soil phosphorus content, air temperature, and stream incision) at each site. Because these filters are not independent (e.g. decreased forest cover leads to warmer temperatures), I used structural equation modeling to assess direct and indirect effects of each filter on diversity.
I found that some environmental variables influenced by urbanization acted as "filters" of biodiversity, while other environmental changes increased plant biodiversity. Specifically, sites with higher soil phosphorus content had lower taxonomic and phylogenetic richness, while sites with more incised streams and warmer temperatures had higher phylogenetic richness and functional evenness. Sites with lower forest cover (i.e. greater habitat transformation) had higher phylogenetic evenness, possibly due to an increase in the number of exotic plant species at these sites. In addition, forest cover had indirect effects on taxonomic and phylogenetic richness via effects on site-level environmental conditions. I saw no effects of habitat fragmentation on diversity, perhaps because these effects were accounted for by changes in diversity with forest cover. I expected that sites with higher population density would have higher taxonomic and phylogenetic diversity because of the influx of ornamental species from plantings in residential neighborhoods, but found the opposite trend. These results provide insight into the complex ways that urbanization influences plant communities and can be used to inform strategies for mitigating the effects of urbanization on biodiversity. This analytical framework could be used to compare the effects of the four filters of urbanization in different cities.