2011 Annual Report to U.S. Fish and Wildlife Service: California Clapper Rail ( Rallus longirostris obsoletus ) TE-807078-12 Submitted to U.S. Fish and Wildlife Service, Sacramento January 31, 2012 Submitted by PRBO Conservation Science Leonard Liu, Julian Wood, Leo Salas, and Nadav Nur PRBO Conservation Science, 3820 Cypress Drive #11, Petaluma, CA 94954 Contact: [email protected] INTRODUCTION The California Clapper Rail ( Rallus longirostris obsoletus ) is one of the most endangered species in California. The species is dependent on tidal wetlands, which have decreased over 75% from the historical extent in San Francisco Bay. A complete survey of its population and distribution within the San Francisco Bay Estuary was begun in 2005. In 2011, PRBO Conservation Science (PRBO) completed the seventh year of field work designed to provide an Estuary-wide abundance estimate and examine the temporal and spatial patterns in California Clapper Rail populations. Field work was performed in collaboration with partners conducting call-count surveys at complementary wetlands (Avocet Research Associates [ARA], California Department of Fish and Game [CDFG], California Coastal Conservancy’s Invasive Spartina Project [ISP], and U.S. Fish and Wildlife Service [USFWS]). This report details PRBO’s California Clapper Rail surveys in 2011 under U.S. Fish and Wildlife service permit TE-807078-12. We also present the results of a preliminary analysis of California Clapper population trends 2005-2011, and results of an analysis performed for a poster presentation at the State of the Estuary conference into detectability of Clapper Rails during the surveys (Appendix 1). A final report for the Section 6-funded project, Temporal and spatial patterns in population trends of the California clapper rail (Rallus longirostris obsoletus) in the San Francisco Estuary: population models and development of methodology and protocol for long-term monitoring , will be presented to CDFG July 1, 2012. The final report will contain: an estimate of population levels for the entire San Francisco Estuary using 2005 through 2011 survey data; an examination and description of the annual and spatial variability in abundance estimates; analyses of habitat and landscape predictors of rail presence and abundance; and recommendations for implementation of California Clapper Rail long-term monitoring, including level of effort (and therefore costs). METHODS Call-count surveys were initiated January 15 and continued until April 15. PRBO surveyed 50 marshes in the Estuary; 33 sites were in San Pablo Bay, 6 sites in Central San Francisco Bay, and 11 sites in South San Francisco Bay (Table 1, Figures 1 and 2). Sites were surveyed 3 to 5 times by experienced permitted biologists using a point transect method, with 10 minutes per listening station. Listening stations were located primarily at marsh edges, levees bordering and within marshes, boardwalks, and boat-accessible channels within the marsh. Stations were placed 200-400 meters apart. Locations of surveyed marshes are presented in Figures 1 and 2. All Clapper Rails (CLRA), as well as California Black Rail (Laterallus jamaicencis coturniculus ), Virginia Rail (Rallus limicola ), and Sora (Porzana carolina ), detected from a listening station were recorded along with the time, direction and distance from the listening station. The actual number of rails detected was recorded, or if the detection was not heard clearly because of confounding circumstances (e.g., distance from observer or environmental conditions) a range in number of rails (e.g., 1 to 2, 2 to 4) was recorded. If no CLRA were detected within 200 meters of a listening station after 2 passive surveys, playback (up to 1 minute) of Clapper Rail vocalizations was used in an attempt to stimulate a response at the end of the third survey. Playback 2 surveys consisted of 5 minutes of passive listening (with no CLRA detected), then 1 minute of playback followed by 4 minutes of passive listening. Clapper Rails detected during transit between listening stations as well as before or after the 10-minute listening period were also recorded. Survey Summary We summarized our 2011 data by selecting the survey visit with the greatest number of detections, including detections outside of the official 10-minute survey period, at each area and taking the mean based on the minimum and maximum number of Clapper Rails detected for each survey site. Trend Analysis We analyzed Clapper Rail survey data from years 2005-2011. We divided the CLRA data from the San Francisco Bay Estuary into two regions for abundance estimation: North Bay (north of the San Francisco- Oakland Bay Bridge) and South Bay (south of the Bay Bridge), and excluded Suisun Bay because of the low number of detections. We only considered detections during the 10-minute survey period and within 200m of the survey point, as the ability of surveyors to detect rails diminishes beyond 200m (PRBO unpublished data). To further increase the chances of detection per point per season, we considered data only from points surveyed at least three times within each survey year. Zeroes in the dataset are both true (surveys of marshes where the species was not present) and false (the species was present but not detected). To improve our ability to estimate detection probability, we removed data from marshes with no CLRA detected in any of the survey years, as we would not be able to estimate detection probability at a marsh that never had any CLRA. Furthermore, the goal of this analysis was to describe trends in abundance which is not possible at marshes where rails were never detected. Filtered this way, the dataset still included a large number of true and false zeroes (53%). Therefore, we used a mixture model consisting of a negative binomial abundance model (generalized linear model with negative binomial link) and an intercept-only detection model (logistic regression) (i.e., zero-inflated model, Zuur et al. 2009). Negative binomial modeling accounts for increased variance in the data caused by true zeroes (species is not present). The zero-inflated model allows us to estimate the effect of false zeroes (the species is present but not detected), and thus we can estimate detection probability. For this zero-inflated negative binomial model we calculated an index of abundance at each point: the maximum count from all visits to the point within each year. Our statistical model therefore assumes that the variation in counts at each point is caused by year, Bay region, survey method (whether call playback was used), and probability of detection effects. Because the abundance equation of the mixture uses a negative binomial link, it also accounts for over-dispersion in the data. We found that detection probability varies with year; in some years individuals are more readily detected. Year effects were included as a discrete factor, thus not forcing the data to conform to a linear trend in time. This resulted in individual estimates of the abundance index for each year and Bay region independent from any trend effects and allowing the estimates to vary freely between years. For the purpose of illustrating the resulting pattern over the period 2005-2011, we fit a generalized additive model (GAM) to the predicted estimates of the abundance index vs. year for each Bay region separately, using a cubic spline with four degrees of freedom as the smoother. We did not include autoregressive effects for annual estimates (i.e., the effect of prior years’ abundance on each year’s estimate), because the smoother naturally accounts for it. Abundance index estimates for each year and their standard errors were plotted along with the GAM curve showing the smoothed pattern. All analyses reported here were performed with the R statistical software (R Development Core Team 2011) and package “pscl” (Zeileis et al. 2008). Finalized results of the trend analysis will be in the July 1, 2012 report. Calling Rate Analysis In a separate analysis, we looked at the following factors as possible contributors to the variability among surveys: time of day, time of year, temperature, wind speed, moon phase, and tide height. We used surveys conducted 1 to 5 times per year between 19 December and 26 May from 2005 through 2010. We included 3 surveys where playbacks of Clapper Rail vocalizations were used on the third visit. Only surveys in San Pablo and San Francisco Bays were included in this analysis, and the survey effort was divided at the San Francisco- Oakland Bay Bridge into North Bay and South Bay regions. All Clapper Rail detections, including those outside the official listening period, were included in this analysis. In total, 1155 listening stations were used from 2005-10. The analyses all included a random effect for each survey station, and included as fixed effects: year and bay (North vs. South) and the interaction between year and Bay. The assumption is that detections differ by year in North Bay independently of how they differ by year in South Bay. Having controlled for survey station, bay, year, and bay x year, we are thus comparing surveys at the same station within and among years. The analysis was conducted on log-transformed counts. RESULTS 2011 Survey Summary PRBO detected 647 (range: 585 to 708) Clapper Rails at the 50 survey sites in 2011 (Table 1). In San Pablo Bay, we detected 307 (range: 281 to 333) CLRA. In Central San Francisco Bay, we detected 122 (range: 113 to 130) CLRA. In South San Francisco Bay, we detected 218 (range: 191 to 245) CLRA. These summaries do not represent our population estimates for these sites, as we did not attempt to calculate abundances or densities based on the detections. Nor did we apply the densities to non-surveyed suitable habitat at the sites. These summaries do not represent the Bay-wide population estimate as not all suitable habitat was surveyed.