The rescue effect and inference from isolation-extinction relationships

The rescue effect in metapopulations hypothesizes that less isolated patches are unlikely to go extinct because recolonization may occur between breeding seasons (“recolonization rescue”), or immigrants may sufficiently bolster population size to prevent extinction altogether (“demographic rescue”). These mechanisms have rarely been demonstrated directly, and most evidence of the rescue effect is from relationships between isolation and extinction. We determined the frequency of recolonization rescue for metapopulations of black rails (Laterallus jamaicensis) and Virginia rails (Rallus limicola) from occupancy surveys conducted during and between breeding seasons, and assessed the reliability of inferences about the occurrence of rescue drawn from isolation-extinction relationships, including autologistic isolation measures that corrected for unsurveyed patches and imperfect detection. Recolonization rescue occurred at expected rates, but was elevated during periods of disturbance that resulted in nonequilibrium metapopulation dynamics. Inferences from extinction-isolation relationships were unreliable, particularly for autologistic measures and for the more vagile Virginia rail.

Methods
This dataset contains files for (1) estimating the frequency of recolonization rescues corrected for imperfect detection of species’ occupancy state and (2) conducting the Bayesian explicit recolonization rescue model published in Van Schmidt & Beissinger (2020) "The rescue effect and inference from isolation-extinction relationships." Black rail and Virginia rail occupancy data was gathered by surveying 273 wetlands in the foothills of the California Sierra Nevada mountains annually. The study area was the zone III Sierra Nevada Foothills ecoregion (US Environmental Protection Agency 2013) in Nevada, Yuba, and southern Butte counties, plus a 1 km buffer to quantify isolation for sites near the study area boundary.

Surveys were conducted from 2002–2016 during the breeding season (late May to early August). Wetlands were surveyed with call broadcast methods, with up to 3 visits each summer (5 in 2002) to estimate detection probability. Surveyors entered the wetland and conducted playback at points spaced every 50 m throughout the wetland. The playback sequence included two sets of kic-kic-kerr calls and two sets of grr calls for black rails, and two sets of a mix of grunt and tick-it calls for Virginia rails. Each call sequence was 30 seconds and followed by 30 seconds of silence for listening for responses. Once a species was detected in that wetland, calls were stopped for that species for the remainder of the visit. If both species were detected at a wetland, we did not revisit it for the remainder of that season. We used the same methodology to resurvey 125 wetlands during the non-breeding season (January 8th–29th) of 2014–2016 to determine the frequency of recolonization rescue.

Wetland habitat covariates were gathered using a mix of field data collection and manual interpretation of summer 2013 GeoEye-1 0.4 m imagery. We mapped all emergent wetlands > 5×5 m within the study area in in Google Earth 7.1.5). Areas covered by hydrophytes (Typha spp., Scirpus spp., Juncus effusus, Leersia oryzoides, or various sedges) were considered wetland, including those that appeared seasonally dried. Any green vegetation inside a 5-m buffer along these hydrophytes was included as a wetland-upland transition zone, but open water and rice fields were excluded. Each wetland’s geomorphology was classified as slope (shallow hillside flow), pond fringe, fluvial, rice fringe, irrigation ditch, or waterfowl impoundment. We combined historic imagery, field data, and talks with local landowners to determine the water sources of wetlands. Occupancy was estimated using an autogressive model as described in detail in the paper.