Nest failure for coastal marsh bird species is primarily caused by predation and nest flooding. As sea level rise makes nest flooding more likely, the threat of nest predation will constrain the potential adaptive responses of marsh nesting species. Thus, understanding the predictors of nest predation is important for the conservation of salt marsh-dwelling bird species, such as Seaside Sparrows (Ammospiza maritima; SESP). Predator activity may be influenced by landscape features (particularly habitat edges), potentially making nest predation predictable. We aimed to understand the predictability of SESP nest predation relative to two major landscape features: roads and tidal rivers, as both of these edges may be entryways or attractants for predators in  marshes. In coastal Georgia, USA, we assessed mammalian predator activity relative to the two features of interest, and hypothesized that mammalian predator activity would be greater close to roads and tidal rivers. We also recorded SESP nest locations and nest predation events and hypothesized that nest predation events would increase with increasing predator activity. Consistent with our first hypothesis, mammalian predator activity increased close to roads and tidal rivers, but mammalian predator distribution did not explain the spatial variation in SESP nest predation thus not supporting our second hypothesis. SESP also placed their nests in locations with high mammalian predator activity, indicating that the ability to avoid nesting in high-risk areas may be constrained by habitat or resource limitations. Additionally, mammals may not be the primary nest predators, as we found that one bird species (Marsh Wren, Cistothorus palustris) contributed substantially to nest predation rates. Understanding the predictability of mammalian predator distribution can allow for focused predator management efforts, such as exclusion, to habitat edges where we found the highest mammalian predator activity, which could relax the constraint of nest predation on SESP’s ability to respond to the intensifying threat of sea level rise.

Camera trapping for mammalian predators: Camera traps units were a modified version of the Hunt trap design outlined in a study by McCleery et al. (2014). We modified the Hunt trap design to be pointed toward the ground, but positioned higher above the ground to prevent damage from inundation during high tides and to avoid filling up camera trap memory cards with images of moving grass. We affixed the cameras to the inside of upside-down 26.5 L buckets, and each bucket was held above the ground by four 3 m PVC poles (Figure 2). Because of the small range of vision of each camera due to their orientation, we attached cotton balls soaked in sardine juice to each PVC pole to encourage any mammalian predators near the camera trap to walk under the bucket. The cotton balls were refreshed with sardine juice every two weeks.  When mammalian predators appeared on the camera trap, we recorded the species, time, and date. Two consecutive camera trap appearances by the same species were marked separately only when there were 2 or more minutes between camera triggers (only 4% of detections were < 10 minutes apart, and only 11% were < 30 minutes apart). To account for differences in the number of days that each camera trap recorded data, we standardized counts by dividing the number of recorded predator occurrences on each camera trap by the number of days that the camera trap was active, and multiplied that value by the number of recording days for the longest-running camera trap: S = [X/D]* 92; where S is the standardized number of detections at a given camera, X is the number of predator detections at that camera, D is the number of days the camera trap was active, and 92 is the number of days the longest running camera was active.

Nest monitoring: We searched for nests every 4-5 days in May-July using the protocol outlined by Hunter et al. (2016). We placed iButtons (Thermochron iButton DS1921G, Maxim Integrated Products, San Diego, CA, U.S.A.)  inside the nests to record nest temperature every 20 minutes for as long as the nests were active, using the protocol from Hunter et al. (2016). This enabled us to determine nest fate and estimate the date and time of nest failure, because nest temperature destabilized upon nest failure due to the lack of incubation (Hunter et al. 2016). We visually checked active nests every 4-5 days for predation activity. In the absence of a nest video camera (see below), we determined the cause of nest failure by comparing the time of destabilized nest temperature (failure) with the timing of extremely high tides, such as spring tides that occur every 14 days. We used tidal gauge data from the Fort Pulaski NOAA station (http://tidesandcurrents.noaa.gov/), approximately 100 km from our sites, to assign the maximum tidal height in meters above mean sea level to each nest during that nest's period of observation (from date found to date of fledging or failure). If there was no spring tide (or other extremely high tide due to a storm) within 1 hour of nest failure, we attributed the nest failure to predation. During April-July 2019 and 2020, we equipped a subset of nests with a video monitoring system (Hart et al. 2021) to continuously record nest activity until the nest fledged or failed. Video recording enabled identification of nest predator species and validation of nest fate assignment using iButton data. For each video monitoring unit, we used a security camera connected to a digital video recorder (DVR), which was located inside of a waterproof case along with a 25-volt battery to power both the camera and DVR. Each camera was affixed to a garden pole and placed directly above the nest. The wires connecting the camera to the DVR were waterproofed using electrical tape and marine epoxy and were approximately 6 m in length, allowing the DVR to be kept away from the nest. This recording system has also been used successfully for monitoring Seaside Sparrows in Gulf coast marshes, and did not increase the risk of nest predation for video-monitored nests (Hart et al. 2021). 

.csv files can be opened with any text editor.