PS 4-50 - The present, the absent, and the barely hanging on: Modelling habitat suitability in a dynamic successional system

Abstract: Habitat suitability models (HSMs) are a powerful tool for managing and restoring plant species of concern. But accurately predicting suitable habitat is challenging for species that occupy temporal niches in systems with rapid geomorphic and successional dynamics. Muhlenbergia sericea (sweetgrass) is a dominant species in the early-successional maritime grasslands that occur behind dunes on barrier islands along the Atlantic and Gulf coasts of the Southeastern U.S. This long-lived, perennial bunchgrass acts as an ecosystem engineer and a refuge to biota in this xeric habitat, and has strong cultural and artistic ties to the Gullah Geechee people of South Carolina and Georgia. These grasslands emerge, persist, and transition through frequent geomorphic changes from accretion, erosion, and storm overwash, and can undergo slow or rapid succession to more mesic vegetation. Today, climate change and human development are intensifying rates of grassland habitat conversion and loss.

Understanding the current distribution of sweetgrass along the Georgia coast, and identifying suitable sites for restoration or creation of maritime grasslands, are critical first steps to facilitate future management. To address those needs, we conducted field surveys to locate sweetgrass populations on 10 barrier islands on the Georgia Coast and developed an HSM for sweetgrass using an ensemble model. During the 2022 Fall/Winter flowering season, we collected 175 occurrence locations and 100 true absence points. The ensemble HSM included 15 environmental predictor variables, including soil and geomorphological attributes, bioclimatic variables, and vegetative communities, sampled to 2m resolution.

The field surveys identified sweetgrass populations in maritime grasslands and also in other vegetation types, including Pinus elliottii forest stands and Morella cerifera shrub thickets. While most sweetgrass plants we saw were robust and flowering, we observed a few areas in which plants were stunted or senescent, suggesting those sites may no longer be “suitable” habitat. To explore the implications of senescent population habitats being unsuitable for conservation or restoration, we reran the HSM with only robust individuals as occurrences, and again with senescent individuals included as absence points. Through comparative analysis of the different model output predictions, we concluded that environmental predictor variables do not adequately capture habitat suitability for a species with such fine-scale spatial and temporal dynamism. Still, the model explores the potential benefits of including qualified judgements of population health within traditional HSMs and is highly useful for targeting further assessments of the species’ distribution, population, and community dynamics using empirical studies.