Restoring and maintaining biodiversity in a changing world is increasingly challenging due to the competing needs of species for suitable space and resources. One ecosystem that has seen considerable anthropogenic changes in extent and structure is the longleaf pine (Pinus palustris) ecosystem. Understanding how wildlife responds to restoration is important to informing forest restoration and conservation. We monitored game birds and mid-large-sized mammal occupancy in and around hardwood patches embedded within a longleaf pine woodland at The Jones Center at Ichauway in Newton, GA. We found that 11 species use the transition zone between the longleaf pine and hardwood hammocks. Gray squirrels (Sciurus carolinensis), Virginia opossums (Didelphis virginiana), nine-banded armadillos (Dasypus novemcinctus) occupancy increased along the gradient while and fox squirrel (Sciurus niger) declined. Our results suggest that oak patches and transitional zones are important to maintaining biodiversity within the longleaf pine ecosystem.

Study Area

We monitored wildlife in and around five hardwood patches at The Jones Center at Ichauway in Newton, GA (Figure 1). The Jones Center at Ichauway is a 12,000-ha property managed for conservation and scientific research. At the time of our study, Ichauway was comprised of a range of ecological communities which included stands of longleaf, slash (P. elliottil), and loblolly pine (P. taeda), as well as mixed pine hardwoods, riparian hardwood forests, depressional wetlands, and shrub-scrub uplands. Over 7,000 ha of the property was open canopy upland pine-grassland vegetation comprised of second-generation longleaf pine and managed with frequent prescribed fire and silviculture, which included removal of ‘off site’ hardwoods. Less common on the property were closed canopy hardwood patches composed of a diverse array of oak species including Quercus incana, Q. falcata, Q. laevis, Q. stellata, Q. virginiana, and Q. hemisphaerica (Jacqmain et al. 1999; Loudermilk et al. 2013). The hard transitions from open canopy pine to closed canopy hardwood hammocks occurring at The Jones Center provide excellent conditions to investigate the role of oak hammocks in other longleaf and open pine systems.

Animal activity

We randomly selected five 4–10-ha hardwood patches in the northern sector of the Jones Center to monitor with camera traps (Fig 1A). Within each patch, we then randomly selected four points along the edge to center a perpendicular transect of cameras; we placed transects at least 25 m apart to improve independence. To investigate how animals respond to the edge at a fine scale we placed a camera 10 m into the longleaf pine, one at the patch edge, and one 10 m into the hardwood patch (Fig 1B). At each camera location, we assessed visual obstruction using a modified Robel pole (Robel et al. 1970; Sovie et al. 2016). We installed each camera 50cm above the ground and angled it towards a bait pile of pecans and cracked corn (Greene et al. 2016). We deployed cameras in each patch five times with each deployment lasting 10–15 days and replaced bait every 5 days. We stopped collecting community composition data in two of the patches after two deployments because we manipulated gray squirrels for a related study (Sovie et al. 2021). We programmed cameras to take three photos every time the camera was tripped using the camera’s normal sensitivity setting and rest for 3 minutes between bursts. We considered photos of animals of the same species taken >20mins apart as independent observations (Greene et al. 2016). We treated camera failure as missing data (Foster & Harmsen 2012) and assumed failures were randomly distributed and do not affect our analysis (Little & Rubin 2014). We followed the American Society of Mammologists guidelines (Sikes et al. 2016) for studying mammals. The University of Florida Institutional Animal Care and Use Committee (IACUC) approved our study (Protocol #: 201709855).                                                    

Data Analysis

We visually identified species in pictures and extracted metadata (date/time) from each photograph. We produced detection history matrices for each camera based on a sampling occasion of five days (0 = species not detected; 1 = species detected; NA = inactive sampling unit or occasion). We collapsed our detection data to 5-day sessions to reduce the complexity of the detection matrix and improve model performance (Nichols et al., 2008).