Causal impact of fire on a globally rare wetland plant: a 40-year Bayesian time series analysis

Background Canby’s dropwort ( Oxypolis canbyi (J.M. Coult. & Rose) Fernald) was listed as federally endangered in 1986, yet the species has continued to decline and is no longer found in 11 counties throughout its former range. The seasonal wetlands in which this forb occurs are disappearing from the landscape, often closing in and transitioning to wet forest or are drained and converted to agriculture. We document the effects of reintroducing fire to the only population of O. canbyi in Maryland and examine the resulting population increase using Bayesian interrupted time series analysis with a counterfactual. Results After cutting woody vegetation, 3.74 times more stems of dropwort per year were produced over the baseline of no intervention. Subsequently, after fire was reintroduced, 10.80 times more dropwort stems per year were produced in comparison to the time period after cutting woody vegetation. The counterfactual prediction showed that in the absence of intervention with fire, dropwort stem production would likely have declined. Conclusions Cutting woody vegetation set the stage for the growth of fine fuels by increasing sun exposure on wetland grasses. It was only after fire was reintroduced that the population of plants expanded significantly in extent and flower production. A process for model selection with a directed acyclic graph followed by Bayesian interrupted time series analysis and a counterfactual was useful for causal inference. Application of fire is an important step in the recovery of the federally endangered dropwort.