Duane R. Diefenbach

VARIABILITY IN GRASSLAND BIRD COUNTS RELATED TO OBSERVER DIFFERENCES AND SPECIES DETECTION RATES

Abstract Differences among observers in ability to detect and identify birds has been long recognized as a potential source of error when surveying terrestrial birds. However, few published studies address that issue in their methods or study design. We used distance sampling with line transects to investigate differences in detection probabilities among observers and among three species of grassland songbirds: Henslows Sparrow (Ammodramus henslowii), Grasshopper Sparrow (A. savannarum), and Savannah Sparrow (Passerculus sandwichensis). Our review of 75 papers published in 1985–2001 found that the most commonly used methods were fixed-width transects (31%, 23 papers) and fixed-radius point counts (20%, 15 papers). The median half-width of fixed-width strip transects used by researchers was 50 m, but our results indicated detection probabilities were <1.0 at distances >25 m for most observers and species. Beyond 50 m from the transect line, we found that as many as 60% of birds were missed by observers and that the proportion missed differed among observers and species. Detection probabilities among observers ranged from 0.43 to 1.00 for Henslows Sparrow, from 0.44 to 0.66 for Grasshopper Sparrow, and from 0.60 to 0.72 for Grasshopper Sparrow for birds detected within 58–100 m of the transect line. Using our estimates of detection probabilities for Henslows Sparrows among six observers in a computer simulation of a monitoring program, we found that bird counts from fixed-width transects required an additional 2–3 years of monitoring to detect a given decline in abundance compared to density estimates that used a method to correct for missed birds. We recommend that researchers employ survey methods that correct for detection probabilities <1.0.

Survival rates, mortality causes, and habitats of Pennsylvania white-tailed deer fawns

Abstract Estimates of survival and cause-specific mortality of white-tailed deer (Odocoileus virginianus) fawns are important to population management. We quantified cause-specific mortality, survival rates, and habitat characteristics related to fawn survival in a forested landscape and an agricultural landscape in central Pennsylvania. We captured and radiocol-lared neonatal (<3 weeks) fawns in 2000–2001 and monitored fawns from capture until death, transmitter failure or collar release, or the end of the study. We estimated survivor-ship functions and assessed influence on fawn survival of road density, habitat edge density, habitat patch diversity, and proportion of herbaceous habitat. We captured 110 fawns in the agricultural landscape and 108 fawns in the forested landscape. At 9 weeks after capture, fawn survival was 72.4% (95% CI=63.3–80.0%) in the agricultural landscape and 57.2% (95% CI=47.5–66.3%) in the forested landscape. Thirty-four-week survival was 52.9% (95% CI = 42.7–62.8%) in the agricultural landscape and 37.9% (95% CI = 27.7–49.3%) in the forested landscape. We detected no relationship between fawn survival and road density, percent herbaceous cover, habitat edge density, or habitat patch diversity (all P>0.05). Predation accounted for 46.2% (95% CI=37.6–56.7%) of 106 mortalities through 34 weeks. We attributed 32.7% (95% CI=21.9–48.6%) and 36.7% (95% CI=25.5–52.9%) of 49 predation events to black bears (Ursus americanus) and coyotes (Canis latrans), respectively. Natural causes, excluding predation, accounted for 27.4% (95% CI=20.1–37.3) of mortalities. Fawn survival in Pennsylvania was comparable to reported survival in forested and agricultural regions in northern portions of the white-tailed deer range. We have no evidence to suggest that the fawn survival rates we observed were preventing population growth. Because white-tailed deer are habitat generalists, home-range-scale habitat characteristics may be unrelated to fawn survival; therefore, future studies should consider landscape-related characteristics on fawn survival.

FOREST COVER INFLUENCES DISPERSAL DISTANCE OF WHITE-TAILED DEER

Abstract Animal dispersal patterns influence gene flow, disease spread, population dynamics, spread of invasive species, and establishment of rare or endangered species. Although differences in dispersal distances among taxa have been reported, few studies have described plasticity of dispersal distance among populations of a single species. In 2002–2003, we radiomarked 308 juvenile (7- to 10-month-old), male white-tailed deer (Odocoileus virginianus) in 2 study areas in Pennsylvania. By using a meta-analysis approach, we compared dispersal rates and distances from these populations together with published reports of 10 other nonmigratory populations of white-tailed deer. Population density did not influence dispersal rate or dispersal distance, nor did forest cover influence dispersal rate. However, average (r2 = 0.94, P < 0.001, d.f. = 9) and maximum (r2 = 0.86, P = 0.001, d.f. = 7) dispersal distances of juvenile male deer were greater in habitats with less forest cover. Hence, dispersal behavior of this habitat generalist varies, and use of landscape data to predict population-specific dispersal distances may aid efforts to model population spread, gene flow, or disease transmission.

INCORPORATING AVAILABILITY FOR DETECTION IN ESTIMATES OF BIRD ABUNDANCE

Abstract Several bird-survey methods have been proposed that provide an estimated detection probability so that bird-count statistics can be used to estimate bird abundance. However, some of these estimators adjust counts of birds observed by the probability that a bird is detected and assume that all birds are available to be detected at the time of the survey. We marked male Henslows Sparrows (Ammodramus henslowii) and Grasshopper Sparrows (A. savannarum) and monitored their behavior during May-July 2002 and 2003 to estimate the proportion of time they were available for detection. We found that the availability of Henslows Sparrows declined in late June to <10% for 5- or 10-min point counts when a male had to sing and be visible to the observer; but during 20 May-19 June, males were available for detection 39.1% (SD = 27.3) of the time for 5-min point counts and 43.9% (SD = 28.9) of the time for 10-min point counts (n = 54). We detected no temporal changes in availability for Grasshopper Sparrows, but estimated availability to be much lower for 5-min point counts (10.3%, SD = 12.2) than for 10-min point counts (19.2%, SD = 22.3) when males had to be visible and sing during the sampling period (n = 80). For distance sampling, we estimated the availability of Henslows Sparrows to be 44.2% (SD = 29.0) and the availability of Grasshopper Sparrows to be 20.6% (SD = 23.5). We show how our estimates of availability can be incorporated in the abundance and variance estimators for distance sampling and modify the abundance and variance estimators for the double-observer method. Methods that directly estimate availability from bird counts but also incorporate detection probabilities need further development and will be important for obtaining unbiased estimates of abundance for these species. Incorporación de la Disponibilidad para la Detección en las Estimaciones de Abundancia de Aves

INTEGRATING WILDLIFE AND HUMAN-DIMENSIONS RESEARCH METHODS TO STUDY HUNTERS

Abstract Recreational hunting is the primary management tool used by natural resource agencies to control ungulate populations. Although free-ranging ungulates have been studied extensively in North America, relatively little is known about the field behavior of hunters or the factors that influence hunting behavior, except on small study areas where access is limited and controlled. We developed 3 integrated protocols to estimate hunter density, distribution, movements, habitat use, characteristics, and attitudes, which can be used on large areas with unrestricted access. We described how aerial surveys, in conjunction with distance sampling techniques and a Geographic Information System (GIS) database of landscape characteristics, provide estimates of hunter density and a map of hunter distribution and habitat use. We used Global Positioning System (GPS) units issued to hunters to systematically record hunter locations. Hunters also completed a simple questionnaire. We linked these data and used them to obtain detailed information on habitat use, movements, and activity patterns. Whereas aerial surveys are limited to discrete points in time and relate only to aggregations of hunters, data collected on hunters that carry GPS units can be used to study habitat use and distribution at different times of day for individual hunters. Finally, linked responses from a traditional mail or telephone survey to hunter location data collected via GPS units to assess how hunter characteristics (e.g., age, physical condition, attitudes) were related to field behavior. We applied these techniques during a white-tailed deer (Odocoileus virginianus) hunting season on a large tract (45,749 ha) of public land in Pennsylvania, USA, with unrestricted hunter access. We estimated density of 7 hunters/1,000 ha (95% CI: 4.2 to 10.3) in the morning and 6.3 hunters/1,000 ha (95% CI: 3.5 to 10.0) in the afternoon. We found that hunter density was negatively related to distance from roads and slope. Most hunters preferred stand hunting, especially in the early morning hours (0600–0800 hr; 72% stationary); more walked or stalked in the afternoon (1400–1600 hr; 58% stationary). The average maximum distance hunters reached from a road open to public vehicles was 0.84 km (SE = 0.03), and they walked an average of 5.48 km (SE = 0.193) during their daily hunting activities. We believe that the approaches we used for studying hunter behavior will be useful for understanding the connections between hunter attitudes and behavior and hence will allow managers to predict hunter response to changes in harvest regulations. Furthermore, our methods are more accurate than requesting hunters to self-report where they hunted. For example, we found that hunters reported that they walked >2.5 times farther from the nearest road (x̄ = 2.23 km, SE = 0.13) than actual distance recorded via GPS units (x̄ = 0.84 km, SE = 0.03). Our research provides wildlife managers with new knowledge on several levels. At the most basic level, we learned a great deal about what hunters actually do while in the field, rather than simply what they report. Second, linking field behavior with hunter characteristics will provide insights into the likely effects of changing hunter demographics. Finally, linking these data with traditional human-dimensions research topics, such as attitudes toward hunting regulations, may allow managers to better forecast the potential effects of regulation changes on hunter distribution and effort.

Modeling and Evaluation of Ear Tag Loss in Black Bears

Demographic models that use marked animals to estimate survival rates and population size assume no tag loss occurs, otherwise estimates are biased. Most studies of tag loss have assumed loss of 1 tag was independent of loss of the other, as did a prior study of ear tag loss in Pennsylvania black bears (Ursus americanus). We used permanently marked (tattooed) black bears to model ear tag loss rates so we could identify bears recovered missing both ear tags, and thus test the independence assumption. We found ear tag loss in male bears increased with time between tagging and recovery. Also, for males, the probability of losing a second ear tag was greater if it had already lost an ear tag. For a tagging-recovery interval of 0.5-<1 year, we estimated 3% of males lost both ear tags (95% CI = 2-4%); however, for an interval of 4.5-<5.5 years, we estimated 56% lost both ear tags (95% CI = 42-75%). We selected the same type of model for females, but ear tag loss rates were much lower. We estimated 2% of females lost both ear tags for tagging-recovery intervals of 0.5-<1 year (95% CI = 1-4%), and 5% of females lost both ear tags for intervals of 4-<5 years (95% CI = 1-18%). Comparison of survival estimates with and without a correction for ear tag loss suggests uncorrected annual survival estimates may be biased -6% for males and -1% for females. Black bears are a long-lived species with high loss rates of ear tags for males. Estimates of survival rates or population size that use mark-recapture type models should either incorporate ear tag loss in the model, especially for males, or use data from short time intervals (≤1 yr) to minimize bias from ear tag loss. In addition to ear tagging to identify individuals for mark-recapture studies, we recommend researchers tattoo bears on both inner sides of the upper lip.

Effect of undercounting and model selection on a sightability-adjustment estimator for elk

Aerial surveys of wildlife populations must correct for failure to observe all animals to obtain unbiased population estimates. We captured and radiocollared elk (Cervus elaphus) in northcentral Pennsylvania to develop a linear-logistic model of characteristics of elk groups associated with visibility bias during helicopter surveys. We used this sightability model in a Horvitz-Thompson sightability-adjustment population estimator. The number of elk in a group was positively correlated (P < 0.001), and the percent canopy cover was negatively correlated (P = 0.002) with the probability of observing a group of elk. Observations of elk groups of known size indicated that helicopter crews undercounted elk group sizes, and percentage of elk missed increased as percent canopy cover increased. Probably because of this undercounting, 3 of 6 population estimates from the sightability-adjustment estimator were less than the number of elk known to be alive on the study area. Simulations of elk surveys using our empirical data indicated that our population estimates may have been negatively biased by 20%, because of undercounting. The assumption of complete enumeration of sighted groups of animals should be verified when using this estimator. When all assumptions of the estimator were met during computer simulations, confidence intervals calculated under the assumption of asymptotic normality and log-normality did not provide nominal coverage. We found that a sightability model using only group size as an independent variable, even when the true sightability model also included percent canopy cover, provided population estimates with little negative bias (<1%), shorter confidence intervals, and the lowest mean square error (MSE). Further research is needed on selection of sightability models, and we recommend using a bootstrap technique to calculate confidence intervals. We believe the sightability-adjustment estimator is unsuitable for estimating population size for small populations (≤1,000 animals).

A test of the scent-station survey technique for bobcats

Scent-station surveys have been widely used to monitor bobcat (Felis rufus) populations, but relationships between bobcat abundance and the index derived from scent-station surveys have not been validated. In autumn 1988 and 1989 we reintroduced bobcats (n = 31) to Cumberland Island, Georgia. We conducted 15 scent-station surveys during September-February 1988, 1989, and 1990 to obtain scent-station indices (SSI) as we increased bobcat density. We found a positive relationship (r 2 = 0.45, P = 0.0066) between population size and SSI. However, because SSI variance also was correlated positively with SSI, we transformed data to meet the assumption of homoscedasticity for the regression model (r 2 = 0.73, P < 0.001)

Controlling for anthropogenically induced atmospheric variation in stable carbon isotope studies

Increased use of stable isotope analysis to examine food-web dynamics, migration, transfer of nutrients, and behavior will likely result in expansion of stable isotope studies investigating human-induced global changes. Recent elevation of atmospheric CO2 concentration, related primarily to fossil fuel combustion, has reduced atmospheric CO2 δ13C (13C/12C), and this change in isotopic baseline has, in turn, reduced plant and animal tissue δ13C of terrestrial and aquatic organisms. Such depletion in CO2 δ13C and its effects on tissue δ13C may introduce bias into δ13C investigations, and if this variation is not controlled, may confound interpretation of results obtained from tissue samples collected over a temporal span. To control for this source of variation, we used a high-precision record of atmospheric CO2 δ13C from ice cores and direct atmospheric measurements to model modern change in CO2 δ13C. From this model, we estimated a correction factor that controls for atmospheric change; this correction reduces bias associated with changes in atmospheric isotopic baseline and facilitates comparison of tissue δ13C collected over multiple years. To exemplify the importance of accounting for atmospheric CO2 δ13C depletion, we applied the correction to a dataset of collagen δ13C obtained from mountain lion (Puma concolor) bone samples collected in California between 1893 and 1995. Before correction, in three of four ecoregions collagen δ13C decreased significantly concurrent with depletion of atmospheric CO2 δ13C (n ≥ 32, P ≤ 0.01). Application of the correction to collagen δ13C data removed trends from regions demonstrating significant declines, and measurement error associated with the correction did not add substantial variation to adjusted estimates. Controlling for long-term atmospheric variation and correcting tissue samples for changes in isotopic baseline facilitate analysis of samples that span a large temporal range.

An Evaluation of Sex-Age-Kill (SAK) Model Performance

Abstract The sex-age-kill (SAK) model is widely used to estimate abundance of harvested large mammals, including white-tailed deer (Odocoileus virginianus). Despite a long history of use, few formal evaluations of SAK performance exist. We investigated how violations of the stable age distribution and stationary population assumption, changes to male or female harvest, stochastic effects (i.e., random fluctuations in recruitment and survival), and sampling efforts influenced SAK estimation. When the simulated population had a stable age distribution and λ > 1, the SAK model underestimated abundance. Conversely, when λ < 1, the SAK overestimated abundance. When changes to male harvest were introduced, SAK estimates were opposite the true population trend. In contrast, SAK estimates were robust to changes in female harvest rates. Stochastic effects caused SAK estimates to fluctuate about their equilibrium abundance, but the effect dampened as the size of the surveyed population increased. When we considered both stochastic effects and sampling error at a deer management unit scale the resultant abundance estimates were within ±121.9% of the true population level 95% of the time. These combined results demonstrate extreme sensitivity to model violations and scale of analysis. Without changes to model formulation, the SAK model will be biased when λ ≠ 1. Furthermore, any factor that alters the male harvest rate, such as changes to regulations or changes in hunter attitudes, will bias population estimates. Sex-age-kill estimates may be precise at large spatial scales, such as the state level, but less so at the individual management unit level. Alternative models, such as statistical age-at-harvest models, which require similar data types, might allow for more robust, broad-scale demographic assessments.