Investigating pollen dispersal of Amaranthus tuberculatus within rooftop and ground level habitats in the New York metropolitan area
Urbanization causes habitat loss and fragmentation that increases non-native species colonization and decreases native biodiversity. However, integrating urban green space, such as parks and gardens, into the urban matrix can reduce these impacts. A limitation on these green spaces is that, as fragments, they may be associated with reduced gene flow among populations, resulting in decreased fitness. In 2014, we investigated pollen dispersal, a direct measure of gene flow, among Amaranthus tuberculatus, a wind-pollinated, dioecious, annual plant, within an urban, rooftop habitat in Bronx, NY, and within a ground-level semi-urban habitat in Westchester County, NY, using seed production as a proxy for pollen dispersal. The external transcribed spacer nuclear region was used for genotyping of parents and seeds to confirm pollen dispersed from experimental male plants. We hypothesized that: 1) seed production would differ between sites, 2) seed production by female plants would decrease with decreasing distance from the pollen source (i.e. male plants), and 3) seed production would vary with distance to pollen source, biomass, height, number of inflorescences, days to first collection, and flowering duration.
Seed production differed between the Bronx (mean = 1072 seeds per plant) and Westchester (mean = 187 seeds per plant) sites (t = 3.79, df = 31, P = 0.001). Further, at the Westchester site, seed production decreased with distance from the pollen source, while seed production did not clearly vary with distance from pollen source at the Bronx site. A backward stepwise elimination linear regression analysis for the Westchester site yielded a model with two significant predictors: distance to pollen source (F = 12.49, P = 0.005) and flowering duration (F = 11.33, P = 0.007). Similarly, the best model using AICc selection for the Westchester site included distance to pollen source, biomass, number of inflorescences, days to first collection, and flowering duration, but only distance to pollen source and flowering duration were significant. For the Bronx site, the backward stepwise analysis did not yield any significant independent variables, and AICc selection produced a best model including only flowering duration, which was not significant. Results from the Bronx site may be attributed to local, rooftop wind patterns, which is currently being investigated.