Northeastern Naturalist 56 A. Cherukuri, A. Strong, and T.M. Donovan 22001188 NORTHEASTERN NATURALIST V2o5l.( 12)5:,5 N6–o7. 11 Monitoring Least Bitterns (Ixobrychis exilis) in Vermont: Detection Probability and Occupancy Modeling Aswini Cherukuri1,*, Allan Strong1, and Therese M. Donovan2 Abstract- Ixobrychus exillis (Least Bittern) is listed as a species of high concern in the North American Waterbird Conservation Plan and is a US Fish and Wildlife Service migratory bird species of conservation concern in the Northeast. Little is known about the population of Least Bitterns in the Northeast because of their low population density, tendency to nest in dense wetland vegetation, and secretive behavior. Urban and agricultural development is expected to encroach on and degrade suitable wetland habitat; however, we cannot predict the effects on Least Bittern populations without more accurate information on their abundance and distribution. We conducted surveys of wetlands in Vermont to assess the efficacy of a monitoring protocol and to establish baseline Least Bittern abundance and distribution data at a sample of 29 wetland sites. Surveys yielded detections of 31 individuals at 15 of 29 sites across 3 biophysical regions and at 5 sites where occupancy had not been previously reported. Probability of occupancy was positively related to wetland size and number of patches, though the relationships were not strong enough to conclude if these were true determinants of occupancy. Call–response broadcast surveys yielded 30 detections, while passive surveys yielded 13. Call–response broadcasts (P = 0.897) increased the rate of detection by 55% compared to passive surveys (P = 0.577). Our results suggest that call–response broadcast surveys are an effective means of assessing Least Bittern occupancy and may reduce bias in long-term monitoring programs. Introduction Almost one-quarter of all US bird species rely on freshwater wetlands for some portion of their annual cycle (Corrigan 2014). Wetlands provide birds with food, cover, and critical breeding habitat. Since 2004, wetland loss has accelerated by 140% as a result of direct loss through filling and drainage, and as a result of indirect degradation through climate change, introduction of exotic species, eutrophication, and altered hydrology (North American Bird Conservation Initiative 2016). Ixobrychus exilis Gmelin (Least Bittern) is an uncommon obligate wetland species. Although not of conservation concern globally, it is a species of greatest conservation need in all northeastern states, and is listed as endangered in Massachusetts (MDFW 2015) and Maine (MDIFW 2015), threatened in New York (NYS-DECFW 2015), and special concern in Vermont (VWAPT 2015) and New Hampshire (NHWAPT 2015). Like many wetland species, Least Bittern populations are limited by availability of high-quality wetland habitat (VWAPT 2015). 1Rubenstein School of Environment and Natural Resources, University of Vermont, Burlington, VT 05405. 2US Geological Survey, Vermont Cooperative Fish and Wildlife Research Unit, Rubenstein School of Environment and Natural Resources, University of Vermont, Burlington, VT 05405. *Corresponding author - aswinicherukuri1@gmail.com. Manuscript Editor: Jean-Pierre Savard Northeastern Naturalist Vol. 25, No. 1 A. Cherukuri, A. Strong, and T.M. Donovan 2018 57 Invasive species, such as Lythrum salicaria L. (Purple Loosestrife) and Phragmites australis L. (Phragmites), and agricultural runoff threaten Least Bittern populations by degrading habitat quality, water quality, and prey populations (VWAPT 2015). Marsh birds can serve as indicator species for evaluating wetland-ecosystem health and restoration efforts (COSEWIC 2009). However, there is a lack of basic information on population numbers, trends, and specific factors affecting marshbird distribution because wetland habitat is often difficult to access (Gibbs and Melvin 1993). Further, because marsh birds are secretive and vocalize infrequently, detection rates during passive roadside surveys such as the Breeding Bird Survey (BBS) are often low (Steidl et al. 2013). A long-term monitoring and management plan is needed to learn more about changes in the Northeast’s Least Bittern population. Documented changes in abundance and distribution over time can then be used to establish management practices and assess whether these practices are maintaining the population. Successful conservation efforts depend on the development of efficient tools to monitor wildlife populations, which allows managers to quantify population changes and minimize uncertainty (Boutin et al. 2009). Therefore, any effective monitoring protocol must allow researchers to estimate detection probability—the probability that a species is detected if it is present (Pollock et al. 2002). Accurate abundance and density estimates depend on detection probability because surveys typically count less than 100% of the individuals within a sampling area (Conway 2011); passive detection of secretive wetland birds is notoriously inefficient (Conway and Gibbs 2011, Steidl et al. 2013). By contrast, conspecific-call broadcasting is an effective tool for eliciting vocal responses of many bird species. This technique would be particularly useful for detecting Least Bitterns because they are difficult to observe visually in dense emergent vegetation (Swift et al. 1988). However, the lack of an estimate of detection probability confounds assessments of population and distribution patterns and obscures ecological relationships that may be causing population changes (Fiske and Chandler 2011). Thus, it is necessary to determine which monitoring strategies maximize detection and minimize false absences (Conway 2011). We surveyed wetlands throughout Vermont to assess factors affecting Least Bittern detection and occupancy. Although Least Bittern presence has been confirmed in several of the state’s biophysical regions (Renfrew 2013), their statewide distribution is poorly studied (VWAPT 2015). We used a single-season occupancy-analysis framework (MacKenzie et al. 2002) to estimate detection probability and site-occupancy probability. We assessed how time of year, temperature, and the use of conspecific playbacks affected Least Bittern detection probability, and assessed how wetland size, configuration, and composition affected Least Bittern occupancy probability. Methods Study species The Least Bittern breeding range extends from southeastern Canada through the Atlantic Coast, the Caribbean, and parts of South America (Poole et al. 2009). Northeastern Naturalist 58 A. Cherukuri, A. Strong, and T.M. Donovan 2018 Vol. 25, No. 1 Populations in the western US are scattered and are concentrated in wetlands associated with rivers, lakes, and estuaries. The global Least Bittern population estimate is 128,100 based on BBS data (Wetlands International 2006). Though this is the most authoritative estimate of the population, the siting of BBS routes along roadsides and away from the Least Bittern’s wetland habitat makes the accuracy of this estimate uncertain. Pre-survey field procedures Site selection. We conducted surveys at 29 sites in Vermont (Table 1, Fig. 1), where each site was a 19.6-ha circular area with a radius of 250 m centered on a survey station. We selected roughly an equal number of sites with confirmed Least Table 1. Vermont survey sites sampled between 18 May and 2 July 2015 and whether Least Bitterns had previously been reported in eBird or Vermont Important Bird Areas species lists. The number of Least Bitterns detected is noted for each site. %EV = percent emergent vegetation, CB = call–response broadcast survey, P = passive survey, and LEBI = Least Bittern. # LEBI detected Wetland # of # Site name Town size patches %EV Reported? CB P 1 Shelburne Pond Shelburne 1020.22 1 50 Yes 0 0 2 Carse Wetlands Hinesburg 155.86 2 25 No 0 0 3 Mud Creek Alburgh 1151.00 5 55 Yes 1 0 4 Dillenbeck Bay Alburgh 111.90 1 55 No 0 0 5 McCuen Slang Addison 101.90 2 15 No 0 0 6 Dead Creek Addison 339.47 3 40 Yes 0 1 7 Delta Park Colchester 51.40 1 45 Yes 0 0 8 Halfmoon Cove Colchester 320.60 1 35 Yes 0 0 9 Intervale Management Area Burlington 285.58 0 0 No 0 0 10 Muddy Brook Pullout Shelburne 37.80 4 15 Yes 1 0 11 West Rutland Marsh West Rutland 827.23 5 60 Yes 3 1 12 Tinmouth Channel Rutland 379.76 4 80 Yes 4 1 13 South Slang Ferrisburgh 330.72 3 30 Yes 1 0 14 East Slang Ferrisburgh 1786.20 2 45 No 2 1 15 Waits River Bradford 20.71 4 20 Yes 1 1 16 Herrick’s Cove Bellows Falls 546.83 3 15 No 0 0 17 Keeler Bay South Hero 265.57 5 20 No 4 2 18 Sandbar State Park Milton 678.18 3 85 Yes 0 0 19 South Stream Bennington 172.40 4 70 Yes 0 0 20 Rake Branch Searsburg 48.57 4 35 No 0 0 21 Sanford Brook Orwell 21.68 4 35 No 0 0 22 South Fork East Creek Benson 24.82 2 10 No 1 0 23 South Bay WMA- Black River Coventry 343.98 5 80 No 0 0 24 South Bay WMA Barton River Coventry 40.32 1 5 No 0 0 25 Lake Bomoseen/Hubbardton Castleton 2543.69 2 20 Yes 1 0 26 Pond Woods WMA Benson 57.60 2 60 No 1 0 27 St. Albans Bay/ Black Creek St. Albans 301.12 3 50 No 4 2 28 Missisquoi- Jeep Trail Swanton 586.19 2 30 Yes 3 1 29 Missisquoi- Burton's pothole Swanton 409.40 2 55 Yes 3 3 Total: 30 13 Northeastern Naturalist Vol. 25, No. 1 A. Cherukuri, A. Strong, and T.M. Donovan 2018 59 Bittern occupancy and sites with suitable habitat but unknown occupancy status. We chose sites with documented Least Bittern presence in previous years (n = 15) from eBird (http://ebird.org/ebird/map/) and Vermont Important Bird Areas species lists (Audubon Vermont 2015). To select the remaining sites (n = 14), we searched Google Earth aerial imagery to identify wetlands with suitable habitat but with no documented Least Bittern detections. Missisquoi National Wildlife Refuge was split into 2 sites, with plot centers 2.90 km apart, because it contained large discontinuous tracts of suitable habitat. Dispersal between tracts was considered unlikely because the home-range area of Least Bitterns is estimated to be 9.7 ha (Bogner and Baldassarre 2002a). Figure 1. Map of the study area and study sites for monitoring Least Bitterns in Vermont. Northeastern Naturalist 60 A. Cherukuri, A. Strong, and T.M. Donovan 2018 Vol. 25, No. 1 Least Bittern breeding habitat varies throughout its range, but dense emergent vegetation is uniformly preferred (COSEWIC 2009). In the northern portion of their range, Least Bitterns are strongly associated with Typha spp. (cattails), which provide excellent concealment above deep water (Nadeau et al. 2008). Hemi-marshes dominated by Scirpus spp. and Schoemoplectus spp. (bulrushes), Sparganium spp. (bur-reeds), and phragmites were also considered suitable habitat, though generally lower quality (Jobin et al. 2011b). We established sites in pairs based on proximity between the sites (less than 1 hour driving) so that 2 surveys could be completed each day. All sites were accessible by foot or kayak with minimal disturbance to the habitat. Where possible, we followed established trails to minimize disturbance to vegetation. Field-survey methods We positioned survey stations in the largest accessible patch of suitable habitat within each wetland; they were initially remotely identified using Google Earth. We finalized the exact station location during the first site visit to best represent the largest tract of suitable habitat. In the Northeast, Least Bitterns return to breeding grounds between April and May, with pair formation beginning soon after and sometimes extending into June (Bogner and Baldassarre 2002a). First-clutch initiation peaks in early June and second broods may occur through July (Graber et al. 1978). We therefore timed data collection between 18 May and 2 July 2015, throughout the nest initiation, incubation, nestling, and early fledging phases. We conducted surveys from 30 min before sunrise to 10:00, when Least Bitterns tend to be most vocal (Jobin et al. 2011b). We did not conduct surveys during rain, extreme heat (>30 °C) or extreme wind (>19 km/h). In 5 cases, we carried out surveys later in the day when inclement weather cleared and travel constraints prevented a return visit to the site. We conducted 2 surveys at least 10 days apart at each site: a call–response broadcast and a passive survey. The order of survey methods was randomized. Call-response broadcast. Call–response broadcasts used the Least Bittern’s “coo-coo-coo” call to stimulate activity. The call–response broadcast surveys followed the methods outlined by Jobin et al. (2011a): 5 min of passive listening followed by 5 min of broadcasts and concluded with 3 mins of passive listening. The broadcast portion consisted of five 30-second intervals of the Least Bittern call (“coo-coo-coo”) with 30-second intervals of silence after each sequence for a total of 2.5 min of broadcast sounds. We obtained calls from Stokes Field Guide to Bird Songs and edited them to fit the aforementioned format. We played the call– response broadcasts on a wireless speaker placed in an accessible location near the survey station. The speaker was directed towards the portion of the wetland with the greatest amount of suitable habitat. The surveyor remained stationary for the duration of the 13-min survey. We assumed that surveyors could confidently hear and identify any Least Bittern call within 250 m. Passive survey. Passive surveys lasted 2 h and did not involve any intentional stimulation of Least Bitterns. During passive surveys, observers quietly walked or Northeastern Naturalist Vol. 25, No. 1 A. Cherukuri, A. Strong, and T.M. Donovan 2018 61 paddled through suitable habitat within the wetland, minimizing disturbance and not venturing farther than 250 m from the survey station. We recorded the time of all Least Bittern detections (visual or aural), as well as sex, behavioral observations, and distance to the bird from the survey station at time of first detection. We tracked subsequent movements to minimize the chance of double counting. Covariates To understand factors affecting Least Bittern detection and occupancy, we included methodological, temporal, and wetland characteristics in our analysis. Potential covariates affecting detection probability (P) included Julian date (julian), temperature °C (temp), and method used. Potential covariates affecting occupancy probability (ψ) included wetland size (size), configuration (patches), and composition (%EV). We determined wetland size from the National Wetlands Inventory Map Data (Vermont Center for Geographic Information 2016), and included open water and vegetation. We quantified wetland configuration as the number of patches totally or partially within a 9.7-ha circular area surrounding the survey station—the size of the average Least Bittern home range (Bogner and Baldassarre 2002a). We assumed that the home range encompassed the total area that individuals used on a regular basis. We considered a patch to be a distinct area of emergent vegetation that was completely surrounded by either land or water. If a patch of emergent vegetation was connected to another patch, regardless of the length or width of the connection, we defined it as 1 patch, assuming that a Least Bittern would travel through the emergent vegetation to minimize risk of predation. We represented wetland composition as the percent of emergent vegetation within the home-range areas. We used Google Earth aerial imagery to estimate percent of emergent vegetation within each 9.7-ha circular area centered on each survey station. Analytical methods Each survey received a score of 1 if we detected ≥1 Least Bittern and a score of 0 if no Least Bitterns were detected. We created encounter histories for each site based on detection or non-detection of Least Bitterns for each visit. Single-season occupancy models (MacKenzie et al. 2002) for the Least Bittern were fit using the “occu” function in the R package, unmarked (Fiske and Chandler 2011). Model set We conducted an initial exploratory analysis of covariates affecting the probability of detection by analyzing 5 alternative models (Table 2). In 4 of the 5 models, occupancy was an additive function of all 3 covariates hypothesized to influence Least Bittern occupancy (%EV, size, and patches). In the 5th model, we set occupancy as a function of no covariates (i.e., only the intercept was estimated). The 5 models varied in how detection probability was modeled. Each detection covariate (julian, temp, method) was analyzed as a separate model; our sample size was Northeastern Naturalist 62 A. Cherukuri, A. Strong, and T.M. Donovan 2018 Vol. 25, No. 1 Table 3. AIC model selection results for Least Bittern occupancy from data collected in Vermont wetlands between May and July 2015. Models are listed in order of support. K = number of parameters in the model, ΔAIC is the difference between a model’s AIC score and the top model; AICwt is the model’s weight. The detection and occupancy coefficients are provided for each model, with standard errors in parentheses. In all models, detection was modeled as a function of sampling method. See Methods for covariate descriptions. Detection coefficients Occupancy coefficients Model K AIC ΔAIC AICwt Intercept Method Interecpt Patches %EV Size Wetland size.Patches 5 70.46 0.00 0.28 0.30 (0.54) 1.83 (1.09) -2.07 (1.14) 0.60 (0.36) - 0.00 (0) Patches 4 71.37 0.91 0.18 0.27 (0.55) 1.75 (1.08) -1.31 (0.96) 0.55 (0.34) - - Wetland size.Patches.%EV 6 71.56 1.10 0.16 0.31 (0.53) 1.85 (1.08) -1.67 (1.18) 0.70 (0.38) -0.02 (0.02) 0.00 (0) Wetland size 4 71.94 1.48 0.13 0.26 (0.54) 1.69 (1.02) -0.36 (0.55) - - 0.00 (0) Null 3 72.64 2.18 0.09 0.29 (0.54) 1.79 (1.08) 0.17 (0.41) - - - Patches.%EV 5 73.00 2.54 0.08 0.28 (0.54) 1.78 (1.08) -1.00 (1.06) 0.61 (0.35) -0.01 (0.02) - Wetland size.%EV 5 73.81 3.35 0.05 0.30 (0.54) 1.83 (1.09) -0.24 (0.84) - -0.01 (0.02) 0.00 (0) %EV 4 74.64 4.18 0.03 0.29 (0.54) 1.79 (1.08) 0.17 (0.80) - 0.00 (0.02) - Table 2. AIC model-selection results for Least Bittern detection models from data collected in Vermont wetlands between May and July 2015. All models are listed in order of support. K = number of parameters in the model, ΔAIC is the difference between a model’s AIC score and the top model, and AICwt is the model’s weight. The detection and occupancy coefficients are provided for each model, with standard errors in parentheses. In all models, ψ was modeled as a function of percent emergent vegetation*wetland size*number of patches. See Methods for covariate descriptions. Detection Coefficients Occupancy Coefficients Model K AIC ΔAIC AICwt Intercept Julian Method Temp Intercept Patches %EV Size P (method); psi (full) 6 71.56 0.00 0.53 0.31 (0.53) - 1.85 (1.08) - -1.67 (1.18) 0,70 (0.38) -0.02 (o.02) 0.00 (0) P (null); psi (full) 5 73.76 2.20 0.18 0.90 (0.49) - - - -1.72 (1.25) 0.73 (0.40) -0.02 (0.02) 0.00 (0) P (null); psi (null) 2 74.60 3.04 0.12 0.83 (0.52) - - - 0.28 (0.45) - - - P (temp); psi (full) 6 74.64 3.07 0.11 2.33 (1.55) - - -0.09 (0.09) -1.69 (1.24) 0.80 (0.44) -0.02 (0.02) 0.00 (0) P (julian); psi (full) 6 75.75 4.19 0.07 0.30 (4.82) 0.00 (0.03) - - -1.64 (1.24) 0.72 (0.40) -0.02 (0.02) 0.00 (0) Northeastern Naturalist Vol. 25, No. 1 A. Cherukuri, A. Strong, and T.M. Donovan 2018 63 insufficient to analyze additive or interactive detection models. A null model for P was also fit. We ranked the 5 exploratory models according to Akaike’s information criterion (AIC) and selected covariates for detection probability in the model with the lowest AIC score as the most appropriate for use in occupancy models. Model support was based on the difference between a model’s AIC score and the best-supported model (ΔAIC). As a rule of thumb, a model with ΔAIC < 2 suggests substantial evidence for the model, values between 3 and 7 indicate that the model has considerably less support, and values >10 indicate that the model is very unlikely (Burnham and Anderson 2002). After the exploratory analysis was completed, we developed a model set (n = 8 models) in which we used the most important detection covariate from the exploratory analysis in all 8 models. The model set included all univariate models and all additive combinations of covariates (Table 3). Sample sizes were insufficient to run interactive occupancy models. The model with the lowest AIC score was selected as the best supported model. Model averaging. We expected high uncertainty in the occupancy-model selection process due to the relatively small number of sites surveyed. Therefore, we used model averaging to draw conclusions from all models. Here, the model coefficients (betas) were developed by multiplying them by their AIC weights and summing across all models. Results We detected Least Bitterns at 15 of 29 survey sites between 18 May and 2 July 2015 for an uncorrected occupancy rate of 0.52. We detected Least Bitterns at 5 sites where they had not been reported previously. All occupied sites were within biophysical regions in which occupancy had been confirmed previously. We detected 18 individuals with the call–response broadcast method only, 1 individual from the passive listening method only (Dead Creek Wildlife Management Area), and 12 individuals in both surveys. The total maximum number of individuals encountered was 31. All vocal detections were of males because females do not produce the “coo-coo-coo” call. Detections of male Least Bitterns do not necessarily equate to breeding pairs, so female abundance cannot be extrapolated from the number of males. Only 6 of the 43 detections were visual; 3 were males and 3 were unknown. All visual detections occurred during passive surveys and occurred ≥49 min into the survey. No visual detections occurred during call–response broadcasts because we conducted these surveys at a stationary point lasting just 13 minutes versus the 2-h duration of passive surveys. Of the 30 individuals that responded to broadcasts, 18 (60%) started to respond during the last 30-second interval of the broadcast (minute 10) or during the 2nd passive listening period (minutes 10–13). In the exploratory analysis of covariates that affected probability, the highestranked probability-of-detection model included the detection covariate method and all occupancy covariates (wetland size, number of patches, and %EV; AIC weight = 0.53; Table 2). Call–response broadcasts showed significantly greater detection rates than passive listening surveys (Fig. 2); probability of detection using a Northeastern Naturalist 64 A. Cherukuri, A. Strong, and T.M. Donovan 2018 Vol. 25, No. 1 call–response broadcast survey was 0.897 (SE = 0.096), whereas the probability of detection using a passive survey was 0.577 (SE = 0.13). Thus, use of broadcasts increased detection rate by 55%. Although all models were within the range of support (ΔAIC < 5), models including temperature and Julian date had less support than the null model. In the final model set (n = 8 models), all models included only method as a detection covariate. The top model for estimating Least Bittern occupancy probability was the additive model of wetland size and number of patches (“P ~ method; y ~ wetlandsize + patches”, Table 3). All 8 models had substantial support based on ΔAIC values less than 5 (Table 3). Model-averaged betas indicated that probability of occupancy may be positively affected by wetland size and the number of patches. Probability of occupancy for wetlands < 200 ha varied from 0.3 to 0.75, whereas wetlands ≥ 200 ha had occupancy probabilities varying from 0.5 to 0.8 (Fig. 3). The probability of occupancy for wetlands with 1 patch varied from 0.3 to 0.55, wetlands with 3 patches vaired from 0.5 to 0.7, and wetlands with 5 patches varied from 0.7 to 0.8 (Fig. 4). Although these patterns are not statistically significant due to small sample size, they may be biologically relevant. There was no relationship between probability of occupancy and percent of emergent vegetation within the home-range area. Discussion Call–response broadcast surveys significantly increased detection probability of Least Bitterns. The low probability of visual detection (14%) emphasizes the importance of detecting this species by vocalization for an accurate estimate of probability of occupancy. We heard Least Bitterns vocalizing for a total of 13.5 min out of 1800 min of passive surveying at sites where presence was confirmed (n = 15). Thus, the probability of hearing a Least Bittern in a 13-min timeframe with no intentional stimulation was 9.75%. We conducted call–response broadcast Figure 2. Probability of detection (p) of Least Bitterns in Vermont using call–response broadcast or passive–survey methods (+ 95% confidence intervals) . Northeastern Naturalist Vol. 25, No. 1 A. Cherukuri, A. Strong, and T.M. Donovan 2018 65 Figure 3. Model-averaged probabilities of occupancy (ψ) of Least Bitterns from data collected in Vermont during the period May–July 2015 based on wetland size. Figure 4. Model-averaged probabilities of occupancy (ψ) of Least Bitterns from data collected in Vermont during the period May–July 2015 based on number of patches within the survey plot. Northeastern Naturalist 66 A. Cherukuri, A. Strong, and T.M. Donovan 2018 Vol. 25, No. 1 surveys from a stationary point for 13 min; passive surveys involved listening and thoroughly scanning through all suitable habitat within the 250-m radius for 2 h. Despite the significant difference in survey time and effort, call–response broadcast surveys increased detection rates by 55%. Although there were relatively few sites (n = 29) and we detected relatively few birds (n = 31), these data clearly show that call–response broadcast surveys improve detection rates, and thus, provide more accurate estimates for calculation of population trends. Our results demonstrating that call–response broadcast surveys increase detection rates are consistent with those from several other studies. By surveying radio-marked Least Bitterns to determine actual response rate, Bogner and Baldassarre (2002b) also concluded that conspecific broadcasts effectively elicited vocal responses and increased detection. Similarly, Swift et al. (1988) found that only 16% of all Least Bitterns seen or heard were observed before the broadcast had begun, so the majority of detections occurred in response to broadcasts. Gibbs and Melvin (1993) saw a 750% increase in detection using call–response broadcasts over passive surveys. However, Lor and Malecki (2002) found that detection decreased by 11% when using call–response broadcasts. They noted that call–response broadcasts caused some individuals to move towards the speaker, perhaps as a territorial behavior; however, use of vocal cues ultimately did not increase detection because the birds remained silent. Other studies showed mixed results for use of call–response broadcasts. Manci and Rusch (1988) found a slight increase in detection, from 3 individuals detected using passive surveys to 4 detected using broadcasts; their results were not conclusive because overall detection rates were low. Detection models that included Julian date and temperature were within the range of support but ranked below the null model. We conducted all surveys between May and July; this time-range is when previous studies have found that detectability is greatest (Conway and Gibbs 2011, Tozer et al. 2016). Therefore, the model that included Julian date may not have ranked highly because there were no surveys done outside of the time range where detectability would change significantly. Based on their surveys conducted at temperatures ranging from 0 °C to 30 °C, Tozer et al. (2016) found that temperature had little impact on detectability; temperatures during our surveys ranged from 3.9 °C to 23.9°C. Consequently, that Julian date and temperature did not have significant impacts on detectability was not surprising because the protocol restricted the time frame and temperature range of surveys. Our data indicated multiple (≥3) patches of large (>200 ha) wetlands increased the probability of Least Bittern occupancy. Large patches may increase occupancy because they allow for a greater number of territories, greater distance between territories, and thus decreased conspecific aggression (Bogner and Baldassarre 2002a). Other studies have found that Least Bitterns are present only in large marshes and that probability of occupancy increases with wetland size (Craig 2008). However, wetland size did not rank highly against models that included percent emergent vegetation, distance to water openings, relative density of cattails, and percent horizontal cover (Lor and Malecki 2006). The smallest wetland in Northeastern Naturalist Vol. 25, No. 1 A. Cherukuri, A. Strong, and T.M. Donovan 2018 67 which we documented Least Bitterns was 22 ha, while Jobin et al. (2011a) reported Least Bitterns in 14 wetlands less than 5 ha. Moore et al. (2009) found limited support for wetland size alone as a determinant of occupancy. Our data indicate that a greater number of patches may increase the probability of occupancy. Least Bitterns are strongly associated with a hemi-marsh structure (Jobin et al. 2011b), which is characterized by patches of emergent vegetation in open water. Previous studies have shown that Least Bitterns nest most commonly toward the edge of emergent-vegetation islands (Bogner and Baldassarre 2002a, Post 1998,). In a study of Least Bittern nesting sites, Winstead and King (2006) found that all nests were within 1 m of open-water areas that were at least 3 m in diameter. Therefore, several patches of emergent vegetation would provide increased edge habitat for nesting. More patches may increase the probability of more than 1 breeding pair existing within the sampled area. Models with percent emergent vegetation had the least support in our analyses. Several studies have found that emergent vegetation is positively related with occupancy because it is used for nesting and foraging habitat (Budd and Krementz 2010, Fairbairn and Dinsmore 2001). We measured the percent of emergent vegetation using aerial imagery. We made the measurements after visits to each site; measurements could be inaccurate if we classified other features as emergent vegetation as a result of low-resolution imagery. To quantify percent emergent vegetation, Budd and Krementz (2010) identified vegetation to at least the genus level. Additionally, Least Bitterns more specifically prefer emergent vegetation that is flooded during the nesting and breeding season, so this attribute should be a requisite when measuring percent emergent vegetation (Moore et al. 2009). Therefore, inaccurate data may have had an impact on the strength of support for models that included percent emergent vegetation. Although water level was not included in our study, previous studies have shown that Least Bittern occupancy is highly sensitive to fluctuations in water level (Jobin et al. 2009; Weller 1961). Moore et al. (2009) suggested that suitable Least Bittern habitat should contain 1–4 ha of semi-permanent or permanently flooded emergent vegetation. Areas with low water levels may dry later in the season and cause individuals to disperse if they cannot forage locally (Budd and Krementz 2010). Higher water levels may also increase reproductive success by making it difficult for ground-based predators to access nests (Post 1998). To better understand Least Bittern abundance, distribution, and occupancy in New England, more extensive and standardized data are needed. Expanding a standardized call–response broadcast study, such as the Great Lakes Marsh Monitoring Program (Tozer 2013), could lend itself to more robust data collection. The Great Lakes Monitoring Program uses citizen scientists to conduct surveys of marsh birds in the Great Lakes Basin using broadcasts of focal species, including the Least Bittern (Tozer 2013). These data allow biologists to establish Least Bittern population estimates with greater accuracy and monitor changes in abundance and distribution, which inform wetland management and conservation practices. Data from Vermont Breeding Bird Atlases (Laughlin and Kibbe 1985, Renfrew 2013) were insufficient to show whether there has been a significant change in the Northeastern Naturalist 68 A. Cherukuri, A. Strong, and T.M. Donovan 2018 Vol. 25, No. 1 species’ status because Least Bitterns are rarely found by most observers, and the 25-year time interval may be too long to assess trends. Least Bitterns are also detected infrequently through the Breeding Bird Survey, and most regions with Least Bittern occupancy show important data deficiencies (Sauer et al. 2014). This study provides a more thorough baseline assessment of the distribution of Least Bitterns in Vermont as called for by the Vermont Wildlife Action Plan (VWAPT 2015). Long-term information on abundance and distribution will inform management practices and demonstrate whether or not management practices are effectively maintaining the population. Our data suggested that large wetlands with patchy configuration increase the probability of occupancy. Fortunately, the majority of sites where Least Bittern presence was confirmed have some degree of protection. Many sites are within designated Important Bird Areas, state wildlife management areas, state parks, or national wildlife refuges (NWR). Only 2 sites, Muddy Brook Pullout and Keeler Bay, are not specifically protected, and are of conservation concern given that they are both adjacent to roads and are subject to habitat degradation from agricultural and urban runoff. Although this study was relatively small in scale, the substantial increase in detection using broadcasts is a significant finding that can inform monitoring practices. The elusive behavior of this species makes visual surveys impractical and ineffective. Monitoring Least Bitterns using call–response broadcast surveys will increase detection and provide abundance estimates with greater accuracy so we can confidently document changes in the population. Importantly, a silent listening period following the broadcast segment must be included in the protocol because Least Bitterns do not immediately respond to conspecific calls (Jobin et al. 2011a). Surveys should be conducted where Least Bittern presence has already been documented to track changes in occupancy, in addition to wetlands with no previous record of Least Bitterns. All hemi-marshes with dense cattail stands and patches of open water in Vermont should be surveyed. Although these findings suggest that Least Bittern occupancy is greater in large wetlands (>200 ha), the probability of occupancy in small wetlands with suitable habitat is >30%. Thus, these sites may be of significant value to the population. Due to the relative rarity and widespread decrease of large wetlands, it is necessary to also survey small wetlands with suitable habitat. Acknowledgments The Least Bittern is protected by the Migratory Bird Treaty Act. Therefore, we obtained a scientific collection permit in order to conduct call–response broadcast surveys. The scientific collection permit was issued under the authority of 10 VSA §4152. Thanks to Jon Kart of the Vermont Fish and Wildlife Department for processing our request. Permission is required for research conducted in National Wildlife Refuges and University of Vermont Natural Areas. We presented the scientific collection permit to the managers of Dead Creek NWR, Missisquoi NWR, and University of Vermont Natural Areas. Thanks to Amy Alfieri, Ken Sturm, and Rick Paradis for allowing us to conduct research at these sites and providing valuable advice. Use of trade names or products does not constitute endorsement by the Northeastern Naturalist Vol. 25, No. 1 A. Cherukuri, A. Strong, and T.M. Donovan 2018 69 US Government. The Vermont Cooperative Fish and Wildlife Research Unit is jointly supported by the US Geological Survey, University of Vermont, Vermont Department of Fish and Wildlife, and Wildlife Management Institute. We thank friends, family, and classmates for their support and encouragement. Literature Cited Audubon Vermont. 2015. Important Bird Areas in Vermont. Available online at http:// vt.audubon.org/conservation/important-bird-areas-vermont. Accessed March 2015. Bogner, H.E., and G. A. Baldassarre. 2002a. Home range, movement, and nesting of Least Bitterns in Western New York. The Wilson Bulletin 114:297-308. Bogner, H.E., and G.A. Baldassarre. 2002b. The effectiveness of call–response surveys for detecting Least Bitterns. Journal of Wildlife Management 66:967–984. Boutin, S., D.L. Haughland, J. Schieck, J. Herbers, and E. Bayne. 2009. A new approach to forest-biodiversity monitoring in Canada. Forest Ecology and Management 258:S168– S175. Budd, M.J., and D.G. Krementz. 2010. Habitat use by Least Bitterns in the Arkansas Delta. 2010. The International Journal of Waterbird Biology. 33:140–147. Burnham K.P., and D.R. Anderson. 2002. Model Selection and Multimedia Inference: A Practical Information-theoretic Approach. Springer-Verlag, New York, NY 488 pp. Committee on the Status of Endangered Wildlife in Canada (COSEWIC). 2009. COSEWIC assessment and update status report of the Least Bittern Ixobrychus exilis in Canada. Ottawa, ON, Canada. 36 pp. Conway, C.J. 2011. Standardized North American marsh bird monitoring protocols. Waterbirds 34:319–346. Conway, C.J., and J.P. Gibbs. 2011. Summary of intrinsic and extrinsic factors affecting detection probability of marsh birds. Wetlands 31:403–411. Corrigan, D. 2014. Environmental Missouri: Issues and Sustainability: What You Need to Know. Reedy Press LLC, St. Louis, MO. 240 pp. Craig, R.J. 2008. Determinants of species–area relationships for marsh-nesting birds. Journal of Field Ornithology 79:269–279. Fairbairn, S.E., and J.J. Dinsmore. 2001. Local and landscape-level influences on wetland bird communities of the prairie pothole region of Iowa, USA. Wetlands 21:41–47. Fiske, I.J., and R.B. Chandler. 2011. Unmarked: An R package for fitting hierarchical models of wildlife occurrence and abundance. Journal of Statistical Software 43:1–23. Gibbs, J.P., and S.M. Melvin. 1993. Call–response surveys for monitoring breeding waterbirds. Journal of Wildlife Management 57:27–34. Graber, J.W., R.R. Graber, and E.L. Kirk. 1978. Illinois birds: Ciconiiformes. Illinois Natural History Survey Biological Notes 109:1–80. Jobin B., L. Robillard, and C. Latendresse. 2009. Response of a Least Bittern (Ixobrychus exilis) population to interannual water-level fluctuations. Waterbirds: The International Journal of Waterbird Biology 32:73–80. Jobin, B., R. Bazin, L. Maynard, A. McConnell, and J. Stewart. 2011a. National Least Bittern Survey Protocol. Technical Report Series No. 519. Environment Canada, Canadian Wildlife Service, Quebec Region, QC, Canada. 26 pp. Jobin B., P. Fradette, and S. Labrecque. 2011b. Habitat Use by Least Bitterns (Ixobrychus exilis) in Quebec. Waterbirds: The International Journal of Waterbird Biology 3:143–150. Laughlin, S.B., and D.R. Kibbe (Eds.). 1985. Atlas of Breeding Birds of Vermont. University Press of New England, Hanover, NH. 456 pp. Northeastern Naturalist 70 A. Cherukuri, A. Strong, and T.M. Donovan 2018 Vol. 25, No. 1 Lor, S.K., and R.A. Malecki. 2002. Call–response surveys to monitor marsh-bird population trends. Wildlife Society Bulletin 30:1195–1201. Lor, S.K., and R.A. Malecki. 2006. Breeding ecology and nesting habitat associations of 5 marsh-bird species in Western New York. The International Journal of Waterbird Biology 29:427–436. MacKenzie, D.I., J.D. Nichols, G.B., Lachman, S. Droege, J.A. Royle, and C.A. Langtimm. 2002. Estimating site-occupancy rates when detection probabilities are less than one. Ecology 83:2248–2255. Maine Department of Inland Fisheries and Wildlife (MDIFW). 2015. Maine’s wildlife action plan. Maine Department of Inland Fisheries and Wildlife, Augusta, ME. 382 pp. Manci, K.M., and D.H. Rusch. 1988. Indices to distribution and abundance of some inconspicuous waterbirds on Horicon Marsh. Journal of Field Ornithology 59:67–75. Massachusetts Division of Fisheries and Wildlife (MDFW). 2015. Massachusetts State wildlife action plan 2015. Department of Fish and Game, Westborough, MA. 500 pp. Moore, S., J.R. Nawrot, and J.P. Severson. 2009. Wetland-scale habitat determinants influencing Least Bittern use of created wetlands. Waterbirds 32:16–24. Nadeau, C., C. Conway, B. Smith, and T.E Lewis. 2008. Maximizing detection probability of wetland-dependent birds during point-count surveys in Northwestern Florida. Wilson Journal of Ornithology 120:513–518. New Hampshire Wildlife Action Plan Team (NHWAPT). 2015. New Hampshire wildlife action plan 2015, Revised Edition. New Hampshire Fish and Game Department, Concord, NH. 200 pp. New York State Department of Environmental Conservation, Fish, and Wildlife (NYSDECFW). 2015. New York State action plan. Albany, NY. 105 pp. North American Bird Conservation Initiative, US Committee. 2009. The state of the birds, United States of America. US Department of Interior, Washington, DC. 36 pp. North American Bird Conservation Initiative, US Committee. 2016. The state of the birds, United States of America. US Department of Interior, Washington, DC. 20 pp. Pollock, K.H., J.D. Nichols, T.R. Simons, G.L. Farnsworth, L.L. Bailey, and J.R. Sauer. 2002. Large-scale wildlife monitoring studies: Statistical methods for design and analysis. Environmetrics 13:105–119. Poole, A.F., P.E. Lowther, J.P. Gibbs, F.A. Reid, and S.M. Melvin. 2009. Least Bittern (Ixobrychus exilis), version 2.0. In P.G. Rodewald (Ed.). The Birds of North America. Cornell Lab of Ornithology, Ithaca, NY. Available online at https://doi.org/10.2173/ bna.17. Accessed March 2016. Post, W. 1998. Reproduction of Least Bitterns in a managed wetland. 1998. Colonial Waterbirds 21:268–273. Renfrew, R.B. 2013. The 2nd Atlas of Breeding Birds of Vermont. University Press of New England, Hanover, NH. 572 pp. Sauer, J. R., J.E. Hines, J. E. Fallon, K.L. Pardiek, D. J. Ziolkowski Jr., and W. A. Link. 2014. The North American Breeding Bird Survey, Results and Analysis 1966–2013. US Geological Survey, Patuxent Wildlife Research Center, Laurel, MD, USA. Steidl, R.J., C.J. Conway, and A.R. Litt. 2013. Power to detect trends in abundance of secretive marsh birds: Effects of species traits and sampling effort. The Journal of Wildlife Management 77:445–453. Swift, B.L., S.R. Orman, and J.W. Ozard. 1988. Response of Least Bitterns to tape-recorded calls. Wilson Bulletin 100:496–499. Northeastern Naturalist Vol. 25, No. 1 A. Cherukuri, A. Strong, and T.M. Donovan 2018 71 Tozer, D.C. 2013. The Great Lakes Marsh Monitoring Program 1995–2012: 18 years of surveying birds and frogs as indicators of ecosystem health. Bird Studies Canada, Port Rowan, ON, Canada. 10 pp. Available online at https://www.birdscanada.org/download/ GLMMPreport.pdf. Tozer, D.C., K.L. Drake, and C.M. Falconer. 2016. Modeling detection probability to improve marsh bird surveys in southern Canada and the Great Lakes states. Avian Conservation and Ecology 11(2):3. Vermont Center for Geographic Information. 2016. National Wetlands Inventory Map Data. National Wetlands Inventory. Agency of Commerce and Community Development, Montpelier, VT. Vermont Wildlife Action Plan Team (VWAPT). 2015. Vermont wildlife action plan 2015. Vermont Fish and Wildlife Department, Montpelier, VT. 1177 pp. Weller, M.W. 1961. Breeding Biology of the Least Bittern. The Wilson Bulletin. 73:11–35. Wetlands International. 2006. Waterbird population estimates, 4th Edition. Wetlands International, Wageningen, The Netherlands. 8 pp. Winstead, N.A., and S.L. King. 2006. Least Bittern nesting sites at Reelfoot Lake, Tennessee. Southeastern Naturalist. 5:317–320.