Plant-soil feedbacks have recently been implicated in producing profound community consequences, and a growing number of studies have measured significant positive and negative feedbacks in greenhouse and field transplant experiments.  However, these experiments typically measure individual (not population) characteristics under controlled conditions; therefore, it is not known whether these measured feedbacks actually drive community dynamics.  In this study, we take the opposite perspective and infer feedbacks from community dynamics derived from survey data from a permanent transect. We are studying feedbacks in a dry grassland in the Jack Pine barrens of Northern Michigan; the transect measures 0.75 by 4.5 m, and has been censused for three years by overlaying a 3 by 3 cm grid and counting stems in each cell.  Previous research shows that species are aggregated with conspecifics, but not heterospecifics, at small spatial scales. Species are also associated with different environmental conditions over these small scales, including differences in nitrogen, moisture, soil organic matter, and light.  Markov models of yearly transition probabilities suggest that plants are fairly static, with high probabilities of self-replacement over time, but also show some mobility in colonizing bare and occupied cells.  The spatial aggregation and the static temporal pattern suggest that positive feedbacks may be occurring, i.e., intraspecific interactions are more positive than interspecific interactions.  In this study, we test whether positive feedbacks occur by fitting population growth models to the data using maximum likelihood and determining whether intraspecific interaction coefficients are more positive than interspecific interaction coefficients.

Results/Conclusions

Both negative and positive interaction coefficients were found among the four dominant plant species (5 coefficients were negative, 3 were positive, 8 were not significant), indicating both competition and facilitation occur in this grassland.  Reindeer lichen cover also affected the community, negatively affecting three species and positively affecting one.  Three species produced negative feedbacks because the intraspecific interaction coefficient was the most negative or the second most negative compared to interspecific coefficients.  However, one species produced a positive feedback because the intraspecific interaction coefficient was the second most positive compared to interspecific coefficients.  Spatial simulations of the population models confirm that the fit parameters produce negative/positive feedbacks over time.   These results suggest that negative feedbacks may be more common than positive feedbacks in this system, and that feedbacks can influence the dynamics of communities. Transplant experiments are underway to test these feedbacks experimentally.