Habitat availability,hunting or poaching: what affects distribution and density of large mammals in western Tanzanian woodlands?

Game reserves (GR) in Tanzania have been found to support similar or lower densities of large mammals compared with National parks (NP). But as these areas usually differ considerably not only in regard to management but also to environmental factors, we assessed the relative importance of vegetation cover, species‐specific habitat preferences and legal (trophy hunting) and illegal off‐take for observed differences in species‐specific densities. In the Katavi ecosystem, open habitats were characteristic elements of Katavi NP, while Rukwa GR was dominated by miombo forest. In an inter‐specific comparison, density differences were moderately correlated with preferences for open habitats, and with estimates of combined legal and illegal off‐take but not with one of these separately. In a multiple linear regression, open habitat preference was found to explain 39.6% of the density differences between the two protected areas. This analysis suggests that the broad‐scale pattern of most species’ distributions is governed by differing vegetation cover but that several species are overexploited by illegal (elephant, giraffe, buffalo, bush pig, warthog) or combined off‐take (hippopotamus, eland, waterbuck), thus emphasizing the need for quota readjustments and a more efficient anti‐poaching control.     

Distribution of duck broods relative to habitat characteristics in the Prairie Pothole Region

Conservation programs for breeding ducks in the Prairie Pothole Region (PPR) of the United States and Canada require effective means of evaluating and characterizing breeding habitat across large landscapes. Extensive surveys of the distribution of duck broods in late-summer could help identify wetland basins with greater probabilities of occupancy. Broods are difficult to detect, however, rendering presence–absence data from single-visit surveys difficult to interpret, particularly when probability of detection is related to habitat features. Multiple-visit occupancy surveys offer a potential solution. From 20 July to 5 August 2007–2009, we conducted a 3-visit survey of wetland basins located on 167 10.4-km² study plots in the PPR. Our survey focused on broods of the 5 most common breeding duck species (Anas spp.). Our main objectives were to investigate ecological relationships between occupancy of wetland basins by broods and habitat characteristics and to examine if habitat-specific detection was of enough concern to warrant multi-survey approaches in the future. We surveyed 3,226 wetland basins during the study. Probability of occupancy of a wetland basin by a brood was positively related to the log of wet area for all 5 study species and was greater on wetlands located on plots with a greater proportion of herbaceous perennial cover for 4 of 5 species. For example, the median probability of occupancy for gadwall (Anas strepera) increased from 0.08 (90% Credible Interval [CrI]: 0.07, 0.10) to 0.28 (90% CrI: 0.23, 0.33) as wet area increased from 0.19 ha to 2.12 ha, and increased from 0.12 (90% CrI: 0.09, 0.16) to 0.20 (90% CrI: 0.16, 0.25) as proportion of perennial grass cover on the study plot increased from 0.03 to 0.99. Because occupancy and detection were both related to attributes of wetland basins, we concluded that the multiple-visit survey was a useful approach for identifying habitat relationships of duck broods. Our results indicated that most broods of the study species were found in 10.4-km² landscapes with greater densities of small- to mid-sized wetland basins and a greater proportion of herbaceous perennial vegetation. Our study provided new empirical support that could be used to help target conservation actions to the most productive landscapes for breeding ducks. © 2012 The Wildlife Society.     

Estimating the abundance of Long‐billed Curlews in Nebraska

ABSTRACT Long‐billed Curlews (Numenius americanus) are shorebirds of high conservation concern in North America. Populations have declined in the last 150 yr primarily due to habitat loss and conversion. We conducted a 2‐yr study to estimate the density and statewide abundance of breeding Long‐billed Curlews in Nebraska during 2008 and 2009. Surveys were conducted during the prenesting period in April when Long‐billed Curlews were likely to be detected. We used a simple random sample of roadside survey routes (N= 39), each consisting of 40 5‐min point‐counts at 800‐m intervals. We modeled detection probability and found that wind speed negatively affected detectability, but found no evidence of either year effects or observer bias. We estimated there were 0.0038 Long‐billed Curlews per hectare (0.38 Long‐billed Curlews per km²) along survey routes and, by extrapolation, estimated there were 23,909 (SE = 1685; 95% CI: 20,810—27,471) Long‐billed Curlews in Nebraska. Our population estimate suggests that ～15 to 22% of the United States population of Long‐billed Curlews is found in Nebraska. Curlews were not evenly distributed within our survey area, with the highest densities in the central Sandhills, an area dominated by grass‐stabilized sand dunes and shallow wetlands, and the lowest densities in predominantly agricultural areas. Because Long‐billed Curlews in Nebraska face many potential threats resulting from land‐use changes, monitoring of the breeding population should be continued. Our survey method was efficient and yielded interpretable results; other states or regions should consider adopting this approach for estimating curlew abundance.     

Encounter Rates From Road-Based Surveys of Rio Grande Wild Turkeys in Texas

ABSTRACT Traditional index-based techniques have indicated declines in Rio Grande wild turkey (Meleagris gallopavo intermedia; hereafter, wild turkey) populations across much of Texas, USA. However, population indices can be unreliable. Research has indicated that road-based surveys may be an efficacious technique for monitoring wild turkey populations on an ecoregion level. Therefore, our goal was to evaluate applicability of road-based distance sampling in the Cross Timbers, Edwards Plateau, Rolling Plains, and South Texas ecoregions of Texas. We conducted road-based surveys in each ecoregion during December 2007—March 2008 to estimate wild turkey flock encounter rates and to determine survey effort (i.e., km of roads) required to obtain adequate sample sizes for distance sampling in each ecoregion. With simulations using inflatable turkey decoys, we also evaluated effects of distance to a flock, flock size, and vegetative cover on turkey flock detectability. Encounter rates of wild turkey flocks from road-based surveys varied from 0.1 (95% CI = 0.0–0.6) to 2.2 (95% CI = 0.8–6.0) flocks/100 km surveyed. Encounter rates from surveys restricted to riparian communities (i.e., areas ≤1 km from a river or stream) varied from 0.2 (95% CI = 0.1–0.6) to 2.9 (95% CI = 1.5–6.7) flocks/100 km surveyed. Flock detection probabilities from field simulations ranged from 22.5% (95% CI = 16.3–29.8%) to 25.0% (95% CI = 13.6–39.6%). Flock detection probabilities were lower than expected in all 4 ecoregions, which resulted in low encounter rates. Estimated survey effort required to obtain adequate sample sizes for distance sampling ranged from 2,765 km (95% CI = 2,597–2,956 km) in the Edwards Plateau to 37,153 km (95% CI = 12,861–107,329 km) in South Texas. When we restricted road-based surveys to riparian communities, estimated survey effort ranged from 2,222 km (95% CI = 2,092–2,370 km) in the Edwards Plateau to 22,222 km (95% CI = 19,782–25,349 km) in South Texas.     

Detection Probability and Sources of Variation in White-Tailed Deer Spotlight Surveys

Abstract: As a first step in understanding structure and dynamics of white-tailed deer (Odocoileus virginianus) populations, managers require knowledge of population size. Spotlight counts are widely used to index deer abundance; however, detection probabilities using spotlights have not been formally estimated. Using a closed mark—recapture design, we explored the efficiency of spotlights for detecting deer by operating thermal imagers and spotlights simultaneously. Spotlights detected only 50.6% of the deer detected by thermal imagers. Relative to the thermal imager, spotlights failed to detect 44.2% of deer groups (≥1 deer). Detection probabilities for spotlight observers varied between and within observers, ranging from 0.30 (SE = 0.053) to 0.66 (SE = 0.058). Managers commonly assume that although road counts based on convenience sampling designs are imperfect, observers can gather population-trend information from repeated counts along the same survey route. Our results indicate detection rate varied between and within observers and surveyed transects. If detection probabilities are substantially affected by many variables, and if transect selection is not based on appropriate sampling designs, it may be impractical to correct road spotlight counts for detection probabilities to garner unbiased estimates of population size.     

Simulations inform design of regional occupancy‐based monitoring for a sparsely distributed,territorial species

Sparsely distributed species attract conservation concern, but insufficient information on population trends challenges conservation and funding prioritization. Occupancy‐based monitoring is attractive for these species, but appropriate sampling design and inference depend on particulars of the study system. We employed spatially explicit simulations to identify minimum levels of sampling effort for a regional occupancy monitoring study design, using white‐headed woodpeckers (Picoides albolvartus), a sparsely distributed, territorial species threatened by habitat decline and degradation, as a case study. We compared the original design with commonly proposed alternatives with varying targets of inference (i.e., species range, space use, or abundance) and spatial extent of sampling. Sampling effort needed to achieve adequate power to observe a long‐term population trend (≥80% chance to observe a 2% yearly decline over 20 years) with the previously used study design consisted of annually monitoring ≥120 transects using a single‐survey approach or ≥90 transects surveyed twice per year using a repeat‐survey approach. Designs that shifted inference toward finer‐resolution trends in abundance and extended the spatial extent of sampling by shortening transects, employing a single‐survey approach to monitoring, and incorporating a panel design (33% of units surveyed per year) improved power and reduced error in estimating abundance trends. In contrast, efforts to monitor coarse‐scale trends in species range or space use with repeat surveys provided extremely limited statistical power. Synthesis and applications. Sampling resolutions that approximate home range size, spatially extensive sampling, and designs that target inference of abundance trends rather than range dynamics are probably best suited and most feasible for broad‐scale occupancy‐based monitoring of sparsely distributed territorial animal species.