Guthrie S. Zimmerman

Harvest, Survival, and Abundance of Midcontinent Lesser Snow Geese Relative to Population Reduction Efforts

ABSTRACT We assessed the effectiveness of an extensive and unprecedented wildlife reduction effort directed at a wide-ranging migratory population of geese. Population reduction efforts that targeted several populations of light geese (greater snow geese [Chen caerulescens atlantica], lesser snow geese [C. c. caerulescens], and Rosss geese [C. rossii]) began in 1999 in central and eastern North America. Such efforts were motivated by a broad consensus that abundance of these geese was causing serious ecological damage to terrestrial and salt marsh ecosystems in central and eastern parts of the Canadian Arctic and subarctic regions along Hudson Bay. Starting in February 1999, special conservation measures (or, in the U.S., a conservation order) were added to the respective federal regulations that permitted hunters to take snow geese (in parts of Canada and the U.S.) and Rosss geese (in parts of the U.S.) during specified harvest periods outside of the hunting season. These measures were accompanied by increase or removal of daily kill and possession limits and by permissions to use previously prohibited equipment for hunting these species in certain regions of the continent. The intent was to reduce adult survival through increased hunting mortality, which was judged to be the most cost-effective approach to reversing population growth. Our principal goal was to assess the effectiveness of reduction efforts directed at the midcontinent population of lesser snow geese, which was thought to be the most serious threat to arctic and subarctic ecosystems of the 3 light goose populations. Our multiple objectives included the estimation and detection of change in the response measures of total annual harvest, harvest rate, survival rate, and abundance, using the 1998 hunting period (defined as 1 Aug 1998 to 31 Jul 1999) as a point of reference. We used information about hunter recoveries of leg-banded snow geese and estimates of regular-season harvest to estimate 1) conservation-order harvest and total annual harvest, 2) geographic and temporal distribution of recoveries by age class, 3) survival and recovery probability, and 4) abundance of snow geese each August using Lincolns (1930) method. We also modeled population growth to infer the form of population response to management efforts. Toward that end, we also proposed a method of estimating conservation-order harvest and tested for differences in band-reporting rate between Canada and the United States. Overall, the balance of evidence favored the conclusion that the midcontinent population has continued to grow during the conservation order, although perhaps at a reduced rate. We suggest that annual rate of population growth (), derived from estimates of annual population size in August, likely provides the most reliable inference about change in the midcontinent population. There was a decline in annual survival probability between these 2 periods from about 0.89 to about 0.83 among snow geese from the southern-nesting stratum (south of 60°N latitude), thought to compose about 10% of the midcontinent population. However, we detected no change in the much larger northern-nesting stratum (north of 60°N latitude), where annual survival remained at about 0.87 from 1989 to 2006. Thus, the conclusion that this population continued to increase during the conservation order was largely consistent with the finding that a weighted-survival probability for midcontinent snow geese essentially did not change between the period preceding (1989–1997) and during (1998–2006) the conservation order. Consistent with high survival rates were low harvest rates, which increased from 0.024 during 1989–1997 for northern geese to only 0.027 during 1998–2006 and from 0.031 to only 0.037 for southern geese. Despite the initial increase associated with the conservation order, harvest rates declined during the conservation order for geese from both strata. We suggest that the higher harvest rate evident for southern geese was related to their earlier fall migration and thus earlier exposure to harvest pressure. Migration by more abundant northern geese was later and resulted in a higher ratio of geese to hunters. Additionally, there was more harvest of southern geese in areas north of the Canadian prairies than there was of northern geese. Total annual harvest increased due to the conservation order but failed to exceed 0.75 million adults in any year during the assessment from 1989 to 2006. Harvest of both age classes exceeded 1 million in only 2 of 9 annual harvest periods since the conservation order started. These lower-than-expected harvests of adult snow geese combined with their low harvest rates of ≤0.048 during the conservation order suggested an August population size in excess of 15 million adult snow geese since 1998. We suggest that abundance of midcontinent snow geese was seriously underestimated in the past, and that this underestimate may have contributed to an overconfidence with which suggested harvest levels could achieve a goal of reduced survival and population reduction. Overall, all 3 populations of light geese now exceed numbers present when the conservation order was initiated. We are confident that the abundance and population growth rate of midcontinent snow geese (as well as by Rosss and greater snow geese) currently exceeds the ability of existing numbers of hunters to exert harvest pressure that is necessary to impose sufficient additive mortality and thus effectively influence population growth. It remains unknown how much more or how much longer such populations can increase towards carrying capacity, which we assume to be determined by the standing crop of arctic foods that they exploit, before density dependence can measurably slow the population growth rate. Estimation of carrying capacity in the large northern nesting stratum is among the key research needs that we propose. The situation that has emerged requires a review of perspectives about impacts of midcontinent lesser snow geese in the arctic, whether initial goals behind population management are still relevant, and whether alternative options from the initial array of management tools should be exercised.

Habitat use by bats in eastern Maine.

We described habitat use by bats in Acadia National Park and adjacent sites in eastern Maine. We analyzed frequency of bat passes with Anabat bat detectors to evaluate habitat use at microsite, single patch, and multipatch spatial scales, Most bat captures (91%) were of the genus Myotis, and most of those (74%) were males. Bat activity was concentrated over ponds and along gravel roads at the single-patch scale, and near lakes at the multipatch scale from late spring to late summer, During the early autumn, bat activity decreased overall and was distributed more evenly among patch types. Although our ability to predict areas of high bat activity declined during early autumn, our data indicated that bats appeared to avoid wetlands and sites with dense trees. Management of bat habitat should be considered at multiple spatial and temporal scales. We also recommend that resource managers consider bats when making management decisions concerning lakes, ponds, and gravel roads because bat activity was high at these sites throughout the late spring and summer.

TEMPORAL VARIATION IN THE VITAL RATES OF AN INSULAR POPULATION OF SPOTTED OWLS (STRIX OCCIDENTALIS OCCIDENTALIS): CONTRASTING EFFECTS OF WEATHER

Abstract We studied the demography of an insular California Spotted Owl (Strix occidentalis occidentalis) population in southern California for 12 years. We used model selection based on information theory to examine the relationship between weather and reproduction and survival. Mean annual fecundity was 0.139 (SE = 0.050) for subadult females and 0.345 (SE = 0.028) for adult females. Adult females had higher fecundity than subadult females during all years, and fecundity in both age classes was higher when a wet year preceded a dry spring (i.e. breeding season). A model incorporating these factors explained 100% of the estimated temporal process variation in fecundity. Mean apparent survival was 0.796 (SE = 0.012), 0.880 (SE = 0.041), 0.692 (SE = 0.062), and 0.368 (SE = 0.038) for adult, second-year subadult, first-year subadult, and juvenile (first-year) owls, respectively. We found no temporal process variation in survival. Using a Leslie projection matrix, we estimated the finite rate of population change to be 0.906 (SE = 0.018) over the entire period of study (1987–1998), which indicated that the population declined ≈9% per year during the study. That rate of decline was higher than a rate (λ1991–1998 = 0.921, SE = 0.020) we estimated for a shorter period (1991–1998) that matched the time interval used in a recent meta-analysis of Spotted Owl population dynamics. We believe that both the present estimates and those of the meta-analysis are valid, given their respective goals. The study population was characterized by relatively high, constant survival of territorial birds, low and variable annual reproduction, and relatively low juvenile survival. Because weather was strongly correlated with reproduction, fecundity rates for the species may decline during short-term droughts and when storms occur during the breeding season. Weather extremes may not, however, be sufficient to affect temporal variation in survival of Spotted Owls in this part of their range.

MULTISCALE HAbITAT SELECTION by RUffED GROUSE AT LOw POPULATION DENSITIES

Abstract. Theory suggests habitats should be chosen according to their relative evolutionary benefits and costs. It has been hypothesized that aspen (Populus spp.) forests provide optimal habitat for Ruffed Grouse (Bonasa umbellus). We used the low phase of a grouse populations cycle to assess the prediction that grouse should occupy aspen and avoid other forest types at low population density because of the presumptive fitness benefits of aspen. On the basis of our observations, we predict how the Ruffed Grouse population will increase in different forest types during the next cycle. In conifer (Pinus spp., Abies balsamea, Picea spp.)-dominated and mixed aspen—conifer landscapes, grouse densities were highest where forest types were evenly distributed. Within these landscapes, male Ruffed Grouse selected young aspen stands that were large and round or square. Although Ruffed Grouse selected young aspen stands strongly, contrary to prediction, they also used other forest types even when young aspen stands remained unoccupied. The relative densities of Ruffed Grouse in aspen and conifer forests indicated that the aspen forests carrying capacities for grouse was higher than the conifer forests at least during the low and declining phases of the grouses cycle. On the basis of our observations, we predict that Ruffed Grouse populations in aspen-dominated landscapes will have higher population densities and fluctuate more than will populations in conifer-dominated landscapes. We suggest that studies of avian habitat selection would benefit from knowledge about the relative densities among habitats at differing population sizes because this information could provide insight into the role of habitat in regulating populations and clarify inferences from studies about habitat quality for birds.

The influence of ecological factors on detecting drumming ruffed grouse

Abstract We surveyed drumming ruffed grouse (Bonasa umbellus) to estimate the probability of detecting an individual, and we used Bayesian model selection to assess the influence of factors that may affect detection probabilities of drumming grouse. We found the average probability of detecting a drumming ruffed grouse during a daily survey was 0.33. The probability of detecting a grouse was most strongly influenced by the temperature change during a survey (β̂temp change = 0.23, 95% probability interval [PI] = 0.13 ≤ β̂ ≤ 0.33) and its interaction with temperature at the start of the survey (β̂interaction = 0.01, 95% PI = 1.42 × 10−3 ≤ β̂ ≤ 0.03). Although the best model also included a main effect of temperature at the start of surveys, this variable did not strongly correlate with detection probabilities (β̂start temp = −0.03, 95% PI = −0.06 ≤ β̂ ≤ 9.80 × 10−5). Model assessment using data collected at other sites indicated that this best model performed adequately (i.e., positive correlation between observed and predicted values) but did not explain much of the variation in detection rates. Our results are useful for understanding the historical drumming index used to assess ruffed grouse populations and for designing auditory surveys for this important game bird.

Estimating migratory game-bird productivity by integrating age ratio and banding data

Context Reproduction is a critical component of fitness, and understanding factors that influence temporal and spatial dynamics in reproductive output is important for effective management and conservation. Although several indices of reproductive output for wide-ranging species, such as migratory birds, exist, there has been no theoretical justification for their estimators or associated measures of variance. Aims The aims of our research were to develop statistical justification for an estimator of reproduction and associated variances on the basis of an existing national wing-collection survey and banding data, and to demonstrate the applicability of this estimator to a migratory game bird. Methods We used a Bayesian hierarchical modelling approach to integrate wing-collection data, which provides information on population age ratios, and band-recovery data, which provides information on recovery probabilities of various age classes, for American woodcock (Scolopax minor) to estimate productivity and associated measures of variance. We present two models of relative vulnerability between age classes: one model assumed that adult recovery probabilities were higher, but that annual fluctuations were synchronous between the two age classes (i.e. an additive effect of age and year). The second model assumed that adults, on average, had higher recovery probabilities than did juveniles and that annual fluctuations were asynchronous through time (i.e. an interaction between age and year). Key results Fitting our models within a hierarchical Bayesian framework efficiently incorporates the two data types into a single estimator and derives appropriate variances for the productivity estimator. Further, use of Bayesian methods enabled us to derive credible intervals that avoid the reliance on asymptotic assumptions. When applied to American woodcock data, the additive model resulted in biologically realistic and more precise age-ratio estimates each year and is adequate when the relative vulnerability to sampling only slightly varies or does not vary among components of a population (e.g. age, sex class) among years. Therefore, we recommend using woodcock indices from our analysis based on this model. Conclusions We provide a flexible modelling framework for estimating productivity and associated variances that can incorporate ecological covariates to explore various factors that could drive annual dynamics in productivity. Applying our model to the American woodcock data indicated that assumptions about the variability in relative recovery probabilities could greatly influence the precision of our productivity estimator. Therefore, researchers should carefully consider the assumption of temporally variable relative recovery probabilities (i.e. ratio of juvenile to adults’ recovery probability) for different age classes when applying this estimator. Implications Several national and international management strategies for migratory game birds in North America rely on measures of productivity from harvest survey parts collections, without a justification of the estimator or providing estimates of precision. We derive an estimator of productivity with realistic measures of uncertainty that can be directly incorporated into management plans or ecological studies across large spatial scales.

New Insight To Old Hypotheses: Ruffed Grouse Population Cycles

Abstract We examined factors hypothesized to influence Ruffed Grouse (Bonasa umbellus) population cycles by evaluating 13 a priori models that represented correlations between spring counts of male Ruffed Grouse drumming displays and these factors. We used AICc to rank the relative ability of these models to fit the data and used variance components analysis to assess the amount of temporal process variation in Ruffed Grouse spring counts explained by the best model. A hypothesis representing an interaction between winter precipitation and winter temperature was the top-ranked model. This model indicated that increased precipitation during cold winters (soft snow cover for roosting) was correlated with higher grouse population indices, but that increased precipitation during warm winters (snow crust effect) was correlated with lower spring counts. The highest ranked model (AICc weight = 0.45), explained only 17% of the temporal process variation. The number of migrating Northern Goshawks (Accipiter gentilis), which has been correlated with grouse cycles in previous studies, does not adequately explain, by itself, the variation in annual population indices of Ruffed Grouse. Other factors not considered in our analysis, such as endogenous mechanisms, parasites, or interactions among factors may also be important, which suggest that mechanisms mediating the Ruffed Grouse cycle still require investigation.

State-Dependent Resource Harvesting with Lagged Information about System States

Markov decision processes (MDPs), which involve a temporal sequence of actions conditioned on the state of the managed system, are increasingly being applied in natural resource management. This study focuses on the modification of a traditional MDP to account for those cases in which an action must be chosen after a significant time lag in observing system state, but just prior to a new observation. In order to calculate an optimal decision policy under these conditions, possible actions must be conditioned on the previous observed system state and action taken. We show how to solve these problems when the state transition structure is known and when it is uncertain. Our focus is on the latter case, and we show how actions must be conditioned not only on the previous system state and action, but on the probabilities associated with alternative models of system dynamics. To demonstrate this framework, we calculated and simulated optimal, adaptive policies for MDPs with lagged states for the problem of deciding annual harvest regulations for mallards (Anas platyrhynchos) in the United States. In this particular example, changes in harvest policy induced by the use of lagged information about system state were sufficient to maintain expected management performance (e.g. population size, harvest) even in the face of an uncertain system state at the time of a decision.

Integrating grouse habitat and forestry: An example using the ruffed grouse Bonasa umbellus in Minnesota

Abstract We quantified forest stand attributes at ruffed grouse Bonasa umbellus drumming display sites to develop tree stocking guides as a tool for guiding ruffed grouse management. We estimated tree density and basal area surrounding grouse drumming sites and compared these with unused sites. We used model selection to assess predictions about whether tree density and basal area surrounding drumming sites varied by site classification (primary drumming site, alternate site, unused site) or forest type. We plotted the predicted values from the best model on tree stocking guides, which are tools commonly used by forest managers. Tree density and basal area varied by site classification and by forest type. Our results show that stem density was higher and basal area lower at both primary and alternate drumming sites compared to unused sites in all forest types. We also found that grouse sites in aspen stands had a greater stem density and lower basal area than grouse sites in pine and spruce/fir stands. Incorporating these results into a tree stocking guide suggested that management for grouse in aspen stands should attempt to maintain stands with average stem density and basal area for this species. In contrast, foresters who are managing for conifers and also wish to maintain some grouse habitat should favour wider spacing of trees in stands. Wider spacing will encourage the development of dense understory vegetation favoured by grouse as well as enhance the growth of quality saw-logs. Our study describes a method for incorporating habitat data on ruffed grouse and other wildlife into tree stocking charts, which are commonly used to facilitate management of forest stands.

Daily survival rates of ruffed grouse Bonasa umbellus in northern Minnesota

We radio-marked 56 ruffed grouse Bonasa umbellus in northern Minnesota, USA, during 1963–1965 and 2001–2002. Of these, we estimated the daily survival rate of 49 individuals (32 females and 17 males; of which 27 were grey phase and 22 were red phase) during the breeding and non-breeding seasons. We investigated whether daily survival varied by age, colour phase, gender, season and transmitter type. A model representing an interaction between colour phase and season fit the data best (AICc = 154.760) and was 77% more likely than any of our other a priori models. Daily survival rates were identical during the breeding season (0.998; SE = 0.002 for both colour phases), but higher for red-phased (0.994; SE = 0.003) than for grey-phased (0.980; SE = 0.007) birds during the non-breeding season. The daily estimate of grouse survival pooled across all individuals and seasons was 0.994 (SE = 0.002), which yielded an annual survival probability of 0.111 (SE = 0.082). The estimated annual survival rate was 0.010 (SE = 0.132) for grey-phased birds and 0.206 (SE = 0.146) for red-phased birds. There was no difference in survival rates between the two study periods. Our estimated annual survival rates were similar to other rates reported for ruffed grouse.