Methodology comparison using occupancy modeling for six species of rails native to the Texas Gulf Coast

Abstract

Rails are a group of marsh birds which are traditionally difficult to monitor due to low and variable detection estimates. Many species of rail in the United States are experiencing population declines and are species of special conservation concern with added state and federal protection. However, the secrecy of rails creates inefficient and expensive survey conditions that hinder the ability to conserve and monitor populations. Black Rails, for example, are federally listed as Threatened but have very low detection probabilities that require extensive revisits to locations. Determining survey methodology that would maximize the detection probabilities for rails would minimize long-term monitoring costs by reducing the amount of field hours required to revisit survey areas.From February-July of 2022, I utilized four survey methods (autonomous recording units, call-playback, forward-looking-infrared camera equipped drone, and trail cameras) to monitor six species of rail (King Rail [Rallus elegans], Clapper Rail [R. crepitans], Virginia Rail [R. limicola], Sora [Porzana carolina], Eastern Black Rail [Laterallus jamaicensis] and Yellow Rail [Coturnicops noveboracensis]) that winter and/or breed on the Texas Gulf Coast. Vegetation sampling conducted from June-July was used to determine the influence of micro-habitat features on the occupancy of the six target rail species to calculate occupancy estimates for each survey site. Using occupancy modeling I calculated and compared detection estimates generated for each survey method in order to determine the methods which maximized detection probability at a species level. Advanced cluster analyses were trained, developed, and then subsequently used to sort out known vocalizations of each study species to reduce the required time spent sorting acoustic data. Additionally, I conducted a general cost analysis of the three most common survey techniques (call-playback, passive acoustic monitoring, and passive visual monitoring survey methods) to assess the expenses generated for theoretical long-term survey implementation. Average above-ground biomass was a significant factor influencing occupancy for both Black Rail as well as Yellow Rail. Black rail detection probability (p ̂) was highest in the spring/summer using autonomous recording units (p ̂=0.279) and was low in the winter using call-playback (p ̂=0.041). Yellow Rail detection probability was highest with winter call-playback (p ̂=0.043). King/Clapper rail detection probability was influenced strongly by wind speed in the winter and water levels in the spring/summer and standing water levels influenced both winter and spring/summer occupancy. Call-playback had the highest detection probability in winter for King/Clapper Rail (p ̂=0.733) and Virginia Rail (p ̂=0.326) and autonomous recording units were best in the spring/summer for King/Clapper Rail (p ̂=0.706) and Sora (p ̂=0.474). Drone thermal imaging work revealed the possibility of finding rails as small as Black Rail and Yellow Rail, as 27 total rail detections included 12 Black/Yellow Rails, 12 Sora, 1 Virginia, and 2 King/Clappers. Behavioral response to the drone of targeted rails was minimal, with preliminary data showing 81% of birds exhibiting no visual response to the drone although behavioral intensity would increase with increased duration of hovering over individuals. Our Gaussian kernel regression results revealed that wind speed (p=0.04) and percent cloud cover (p=0.01) influenced the number of rails detected on a given survey. By establishing the methods that display the highest detection probabilities for each species, we were able to weigh the monetary cost of the methods against the potential loss/gain of detection probability. Passive acoustic monitoring reduced costs for multiple species including Black Rail, Sora, and King/Clapper Rail, as this method is far cheaper annually than traditional call-playback methods and maintains adequate detection probabilities for these species. Monitoring for Black Rail, Sora, and King/Clapper Rail may be more attainable for agencies with limited funding or resources. The potential of using drone thermal imagery also may further reduce costs while maintaining adequate detection probability for these secretive species. Future research should focus on estimating the detection probability using drone thermal imaging for each of these species, especially the smallest and most difficult to detect such as Black and Yellow Rail.