Modeling fifteen-year patterns in the growth, mortality, and regeneration of trees in an old-growth, mixed-mesophytic forest

Background/Question/Methods

Community dynamics are driven by mortality, growth, and regeneration rates of populations. Understanding how these rates differ among species is necessary to document and predict changes in composition over time. Here, we use a combination of spatial statistics and multiple logistic regression to model the mortality and regeneration of individual trees in an old-growth, mixed mesophytic forest (Dysart Woods) in southeastern Ohio. In 1996, two permanent 50×70 m plots were established at Dysart Woods, and all stems ≥2.5 cm DBH were tagged, mapped, and measured. In 2011, these individuals were remeasured, and all new stems ≥1 cm DBH were mapped. The O-ring statistic was used to analyze recruitment and mortality patterns.  Logistic regression was used to determine the factors influencing the probability of individual survivorship. Predictor variables in these models include the size, decline index, and plant neighborhood of each individual. Neighborhoods were defined as the number, size, and identity of surrounding stems within a specified radius. The decline index is a semi-quantitative assessment of overall health that incorporates crown transparency, leaf die-back, and epicormic branching. We also used an eigenvector mapping procedure to reduce the influence of spatial autocorrelation.

Results/Conclusions

Mortality and regeneration rates both differed significantly from random simulations as indicated by the O-ring statistic. Random labeling of survival status showed that dead stems were found closer to other individuals than expected at distances of less than 5 meters, and new stems were significantly farther away from individuals present in 1996 than expected by chance. Results from the logistic regression analyses showed that tree DBH and decline index were highly significant at predicting mortality. Decline index was negatively correlated with survival, while increased plant size increased survivorship. Plant neighborhood was significantly associated with mortality probability at radii from five to ten meters from the target individual, although this differed between species. Plant neighborhood showed mostly negative impacts on survival, although the number of heterospecific neighbors significantly increased the survival probability of Beech. The impact of conspecific versus heterospecific individuals varied with the size of the radius used. These results show that demographic rates differ significantly from random at Dysart Woods. Furthermore, incorporating decline index as a control variable and accounting for spatial autocorrelation in mortality modelling is important and largely overlooked in most studies. This information allows a better understanding of population and community dynamics of old-growth systems.