Temperature and dispersal play a role in plant community composition along an urban-to-rural gradient

Background/Question/Methods

Understanding the ecological processes operating within cities is becoming increasingly important.  While many studies have investigated the effects of habitat loss and fragmentation on natural populations and communities, fewer have addressed the impacts of altered environmental conditions associated with urbanization, or the relative importance of spatial and environmental factors in structuring communities.  Urban areas tend to have higher temperatures (the urban heat island effect) and altered hydrology due to increased runoff, which can cause stream erosion and water table lowering.  These two conditions may interact to cause water stress for plants in riparian areas.  In order to determine whether urban environments influence plant community composition, I sampled vegetation in 26 sites of floodplain forest along an urban-to-rural gradient within the Research Triangle area of North Carolina, including the cities of Raleigh, Durham, and Chapel Hill.  Data on plant species identity and cover were collected within a 10X50m plot at each site, and environmental variables such as canopy openness and stream channel depth were measured.  One HOBO data logger was installed at each site to measure continuous air temperature throughout the year.

Results/Conclusions

Mantel tests showed that differences in community composition between sites were correlated with spatial distances between sites (Mantel’s R: 0.22, p-value: 0.001), and with differences in minimum temperatures between sites (Mantels’ R: 0.34, p-value: 0.001).  While these results suggest that spatial processes such as dispersal may be important for structuring these communities, a strong correlation between spatial distance and differences in temperature (Mantel’s R:0.47, p-value:0.001) indicates that this pattern may be due to the spatial autocorrelation of temperature within this landscape.  None of the other measured environmental variables was shown to be correlated with species composition based on Mantel’s test.  A Nonmetric Multidimensional Scaling ordination showed that the amount of impervious surface within a 1km radius was also important in explaining some of the variation in community composition across sites.  As expected, more urban sites have higher abundances of many non-native species, suggesting that dispersal from developed areas contributes propagules of introduced species into forest patches, while sites with higher mean and minimum temperatures have higher abundances of evergreen and upland species.  These results suggest that different aspects of the urbanization gradient may have differential effects on plant species composition.