Observer aging and long‐term avian survey data quality

Long‐term wildlife monitoring involves collecting time series data, often using the same observers over multiple years. Aging‐related changes to these observers may be an important, under‐recognized source of error that can bias management decisions. In this study, we used data from two large, independent bird surveys, the Atlas of the Breeding Birds of Ontario (“OBBA”) and the North American Breeding Bird Survey (“BBS”), to test for age‐related observer effects in long‐term time series of avian presence and abundance. We then considered the effect of such aging phenomena on current population trend estimates. We found significantly fewer detections among older versus younger observers for 13 of 43 OBBA species, and declines in detection as an observer ages for 4 of 6 vocalization groups comprising 59 of 64 BBS species. Consistent with hearing loss influencing this pattern, we also found evidence for increasingly severe detection declines with increasing call frequency among nine high‐pitched bird species (OBBA); however, there were also detection declines at other frequencies, suggesting important additional effects of aging, independent of hearing loss. We lastly found subtle, significant relationships between some species' published population trend estimates and (1) their corresponding vocalization frequency (n ≥ 22 species) and (2) their estimated declines in detectability among older observers (n = 9 high‐frequency, monotone species), suggesting that observer aging can negatively bias long‐term monitoring data for some species in part through hearing loss effects. We recommend that survey designers and modelers account for observer age where possible.     

Humans versus autonomous recording units: a comparison of point‐count results

ABSTRACT There is growing interest in the potential use of autonomous recording units (ARUs) to obtain point‐count bird survey data in the absence of human observers. To determine possible reasons for differences in the point‐count performances of humans and ARUs and to better understand the possible limitations of ARUs, we compared the results of point‐counts conducted by human observers with those conducted using ARUs. Human observers and ARUs recorded birds at 56 point‐count stations in three different habitats (green mixed‐conifer forest, burned mixed‐conifer forest, and mixed riparian cottonwood bottomland). Combined, human observers and ARUs generated a total of 858 detections of 86 different species. We found that 9.7% of detections were recorded by ARUs only, 40.9% by human observers only, and 49.4% by both ARUs and humans. The mean number of species detections per point was significantly greater for human observers (13.8) than for ARUs (9.1), and did not differ significantly among habitat types. Birds not detected using ARU recordings included those with songs or calls too distant to be recorded (52.7%), those detected by human observers using only visual cues (14.8%), and those too difficult to identify from recordings (10.3%) or simply overlooked (8.8%). About two‐thirds (68.7%) of birds detected using ARUs, but missed by human observers in the field, were simply overlooked in the field; most of the rest were the result of misidentification in the lab. The failure of ARUs to record a large proportion of the detections recorded by human observers, combined with problems with detecting or identifying sounds in the lab and the extra time and cost associated with use of ARUs, suggest that they would not provide a cost‐effective means of gathering data for traditional point‐count surveys.     

Estimates of regional population densities of badger Meles meles, fox Vulpes vulpes and hare Lepus europaeus using walked distance sampling

Walked spotlight transect surveys with distance sampling were used to estimate regional population densities of badger (Meles meles), fox (Vulpes vulpes) and brown hare (Lepus europaeus) in south-west England (Cornwall, Devon, Gloucestershire, Herefordshire) and Wales (Pembrokeshire, Borders, North Wales). All regions were surveyed during spring 2006 with English regions re-surveyed in autumn 2006. In each region, surveys were conducted in a random sample of 19.6 km² areas (mean areas per region: spring = 19, autumn = 25). Within each survey area, a semi-random transect was established in each of a random sample of fields (open habitat almost exclusively pasture). Transects were subsequently walked at night with spotlights (mean transects per survey area: spring = 21, autumn = 21). Each area was surveyed twice during a season. Total transect length per region ranged from 137 to 193 km in spring and 230 to 250 km in autumn. The mean density of species per region was: badger 1.5–4.8 km⁻², fox 1.0–4.0 km⁻², hare 0.4–4.6 km⁻². The study has provided baseline estimates of regional densities against which any future equivalent surveys can be compared. It has also illustrated the practical application of large-scale walked distance sampling to surveys of British mammals.     

Probability of detection of Flammulated Owls using nocturnal broadcast surveys

ABSTRACT Broadcasts of conspecific vocalizations are used to increase detection rates on surveys of secretive bird species, yet the assessment of detectability necessary to fully interpret such survey data is frequently lacking. We used radiotelemetry to evaluate the probability of detection of 17 radio‐tagged male Flammulated Owls (Otus flammeolus) using conspecific broadcast surveys in Idaho during 2005 and 2006. Probability of detection among the 11 paired, five unpaired, and one unknown pairing status owls was 100% during the pair‐bonding and incubation periods of the breeding season, after which it declined to less than 15% during the postfledging period. Paired males showed a different pattern than unpaired males. Following hatching of eggs, detectability of paired males declined gradually over a 6‐week period, whereas detectability of unpaired males dropped sharply for a 4‐week period before increasing during the postfledging period. We suggest that surveys for Flammulated Owls be conducted during the pair‐bonding and incubation periods when high detectability permits stronger inference concerning the presence or absence of owls based on broadcast survey detections.     

Efficacy of eBird data as an aid in conservation planning and monitoring

One of the world's largest citizen science projects is eBird, a database that has been used primarily to address questions of bird distributions and abundance over large spatial scales. However, addressing finer‐scale questions is also possible, depending on survey coverage and whether assumptions and limitations are matched to the scale of inferences. Our objective was to determine if the eBird database could be used to develop estimates of bird abundance and diversity comparable to those from standardized shorebird surveys. We compared a year of standardized shorebird surveys by trained observers at Snook Islands Natural Area located in Palm Beach County, Florida, to a year of eBird observations from the same site. Total species richness derived from eBird (25 species) was higher than that from standardized surveys (20 species). Similarly, we found the Shannon diversity index calculated from eBird was higher (2.81) than the same index calculated from standardized surveys (2.21; P < 0.001). The higher diversity and species richness may reflect the greater effort of eBird participants (35,289 person‐hours) compared to our standard surveys (2126 person‐hours). We found only a slight difference in parameter estimates between data obtained from eBird and from standardized surveys. Potential use and value of eBird as a tool for land managers and conservationists may be greater than currently realized, but studies conducted in a wider range of ecosystems and locations are needed to develop generalizations.     

Experimental Investigation of Observation Error in Anuran Call Surveys

ABSTRACT Occupancy models that account for imperfect detection are often used to monitor anuran and songbird species occurrence. However, presence—absence data arising from auditory detections may be more prone to observation error (e.g., false-positive detections) than are sampling approaches utilizing physical captures or sightings of individuals. We conducted realistic, replicated field experiments using a remote broadcasting system to simulate simple anuran call surveys and to investigate potential factors affecting observation error in these studies. Distance, time, ambient noise, and observer abilities were the most important factors explaining false-negative detections. Distance and observer ability were the best overall predictors of false-positive errors, but ambient noise and competing species also affected error rates for some species. False-positive errors made up 5% of all positive detections, with individual observers exhibiting false-positive rates between 0.5% and 14%. Previous research suggests false-positive errors of these magnitudes would induce substantial positive biases in standard estimators of species occurrence, and we recommend practices to mitigate for false positives when developing occupancy monitoring protocols that rely on auditory detections. These recommendations include additional observer training, limiting the number of target species, and establishing distance and ambient noise thresholds during surveys.     

Associations between distance to forest and spatial and temporal variation in abundance of key peatland breeding bird species

Capsule Negative forest edge effects were detected for Willow Ptarmigan (Red Grouse) Lagopus lagopus and Dunlin Calidris alpina.

Aims To investigate the effects of distance to forest on the abundance and changes in abundance of four key peatland breeding bird species, and to measure changes in predatory bird numbers, in the peatlands of northern Scotland.

Methods Bird surveys were carried out in 2000 at 34 plots, covering 197 km² of peatland, and 80 forestry point‐count sites, first surveyed in 1988. Habitat data were also collected in 2000. We used multi‐model inference to investigate the associations between forest distance and other habitat variables, and the abundance, and changes in abundance, of four bird species of economic or conservation importance: Red Grouse, European Golden Plover Pluvialis apricaria, Dunlin and Common Greenshank Tringa nebularia.

Results There was strong evidence that distance to forest was negatively associated with Dunlin abundance and changes in Red Grouse abundance, but only weak evidence for negative associations with Golden Plover abundance and changes in Dunlin abundance. There was no evidence of a forest distance effect on Greenshank. Among predatory birds, there were no significant increases either on peatland plots or in new forestry plantations.

Conclusions This study provides evidence that, for a given habitat quality, Dunlin densities are lower, and Red Grouse declines more likely, near to forest edges, but weak evidence only that Dunlin declines are more likely, and Golden Plover abundance lower, near to forests. These results suggest that for at least two key peatland breeding birds, forest removal is likely to benefit birds breeding on adjacent unafforested peatland.     

Body condition, territory ownership and age-related reproductive performance in Spruce Grouse Dendragapus canadensis hens

Adult Spruce Grouse Dendragapus canadensis hens show better reproductive performance (i.e. larger clutch size, earlier clutch initiation, higher frequency of renesting) than year-lings. Body and breast muscle weights, condition indices and relative body composition were compared between adult and yearling birds from northeastern Ontario, Canada, to determine whether differences in endogenous reserves mobilized for egg production and (or) territory ownership could explain these differences in reproductive performance. Adults and yearlings showed similar declines in body-weight, breast muscle weight and condition indices from the pre-laying to the post-laying period, but no significant interage differences emerged at any stage of the breeding season. Changes in body composition, especially gut structure, were similar in all birds; this result is consistent with a dietary shift in both age classes from a fibrous winter diet to more succulent foods in spring. Non-territory-holding (replacement) yearlings also had similar body-weights and breast muscle weights and were in similar condition to territory holders (residents) in the pre-laying period. My findings do not support the hypothesis that differences in body condition and territory occupancy can account for differences in age-related reproductive performance in Spruce Grouse.     

Differences in bird communities in postfire silvicultural practices stands within pine forest of South Korea

We examined differences in bird communities in relation to characteristics of habitat structure in a pine forest, Samcheok, South Korea. An unburned stand, a stand burned 7 years earlier and then naturally restored, and a stand where Japanese red pine Pinus densiflora seedlings were planted after the fire were used for the survey. Habitat structure was dramatically changed by postfire silvicultural practices. Number of stand trees, shrubs, seedlings, snags, and vegetation coverage were significantly different among study stands. We made 1,421 detections of 46 bird species during 23 separate line transect surveys per stand between February 2007 and December 2008. The mean number of observed bird species and individuals, bird species diversity index (H′), and Simpson’s diversity index (D _s) were highest in the unburned stand and lowest in the pine seedling stand. There were more species and individuals of forest-dwelling birds in the unburned stand than both burned stands. Canopy and cavity nesters, foliage searchers, bark gleaners, and timber drillers were significantly higher in the unburned stand. In the pine seedling stand, densities of birds that prefer open field and shrub cover were higher. Stand structure was simplified in the pine seedling stand by postfire practices. Because of differences in habitat structure and bird communities, postfire practices in the burned stand should be re-evaluated. Also, management strategies for pine forest after forest fires are needed based on results of long-term experiments.     

Testing the importance of auditory detections in avian point counts

ABSTRACT.   Recent advances in the methods used to estimate detection probability during point counts suggest that the detection process is shaped by the types of cues available to observers. For example, models of the detection process based on distance-sampling or time-of-detection methods may yield different results for auditory versus visual cues because of differences in the factors that affect the transmission of these cues from a bird to an observer or differences in an observer's ability to localize cues. Previous studies suggest that auditory detections predominate in forested habitats, but it is not clear how often observers hear birds prior to detecting them visually. We hypothesized that auditory cues might be even more important than previously reported, so we conducted an experiment in a forested habitat in North Carolina that allowed us to better separate auditory and visual detections. Three teams of three observers each performed simultaneous 3-min unlimited-radius point counts at 30 points in a mixed-hardwood forest. One team member could see, but not hear birds, one could hear, but not see, and the third was nonhandicapped. Of the total number of birds detected, 2.9% were detected by deafened observers, 75.1% by blinded observers, and 78.2% by nonhandicapped observers. Detections by blinded and nonhandicapped observers were the same only 54% of the time. Our results suggest that the detection of birds in forest habitats is almost entirely by auditory cues. Because many factors affect the probability that observers will detect auditory cues, the accuracy and precision of avian point count estimates are likely lower than assumed by most field ornithologists.     

Detection and vocalisations of three owl species (Strigiformes) in temperate rainforests of southern Chile

Conspecific broadcasts are effective to increase detection of owls. To determine the most appropriate time of the year to survey owls, we played conspecific owl vocalisations monthly in a temperate rainforest of southern Chile. From 12 broadcast points surveyed we recorded detections of Glaucidium nana, Strix rufipes and Tyto alba. Glaucidium nana presented a bimodal detection curve throughout the year and we recorded two regular vocalisations in response to broadcasting: contact pair call and territorial call. Strix rufipes and T. alba both showed a peak of detection between February and May. Strix rufipes presented three vocalisations: territorial call, contact pair call and female contact pair call while T. alba uttered two vocalisations: territorial call and twittering call. We recommend surveys during the end of the breeding season (austral summer–autumn) when detection is higher in most owls. Surveys should also take into consideration the variability of the vocalisations and include covariates in monitoring to evaluate occupancy/detection models.     

Lethal and sub‐lethal impacts of respiratory cryptosporidiosis on Red Grouse,a wild gamebird of economic importance

Respiratory cryptosporidiosis was first diagnosed in Red Grouse Lagopus lagopus scotica, the UK subspecies of Willow Ptarmigan, in 2010. In the next 3 years, respiratory infection by Cryptosporidium baileyi had manifested itself in Grouse on half the moors in northern England and 80% of moors in the North Pennine Hills. In this first account of the impact of respiratory cryptosporidiosis on the population dynamics of a wild bird we fitted 111 diseased and 67 healthy Grouse with radio‐transmitters at two North Pennine moors where disease prevalence averaged 8.1% and monitored their survival and fecundity between autumn 2013 and autumn 2015. Six‐month natural survival rates (excluding shooting) were 0.70 in healthy Grouse, but only 0.44 in diseased females, and 0.22 in diseased males. Some 39% of diseased birds died from their infection, whereas 28% of healthy birds were shot. A similar proportion of each group were killed by predators, either by Stoat Mustela erminea or raptors. Diseased females bred 8 days later than healthy females, but clutch size, egg volume and nesting success did not differ in relation to disease status. Productivity was 43% lower among pairs with a diseased member than in healthy pairs, but appeared impaired only if the female was diseased, not the male. Differences in productivity were related to chick survival rather than the proportion of pairs that reared broods, with chick survival being lower in the 10 days after hatching and again when chicks were 20–50 days old. This latter period was when respiratory infection among chicks was first noticed and the onset of infection may have been a contributing factor to higher mortality during this period. Described levels of respiratory infection reduced the number of birds available to shoot in August by 6%, which represented a mean annual loss of £0.9 million in revenue across managed grouse moors. Likely reductions in shoot economics could escalate should prevalence increase. This disease is a welfare concern and potentially a conservation concern, too, should infection cross to other bird species occupying the same moors.     

Factors affecting detectability of river otters during sign surveys

Sign surveys are commonly used to study and monitor wildlife species but may be flawed when surveys are conducted only once and cover short distances, which can lead to a lack of accountability for false absences. Multiple observers surveyed for river otter (Lontra canadensis) scat and tracks along stream and reservoir shorelines at 110 randomly selected sites in eastern Kansas from January to April 2008 and 2009 to determine if detection probability differed among substrates, sign types, observers, survey lengths, and near access points. We estimated detection probabilities (p) of river otters using occupancy models in Program PRESENCE. Mean detection probability for a 400-m survey was highest in mud substrates (p = 0.60) and lowest in snow (p = 0.18) and leaf litter substrates (p = 0.27). Scat had a higher detection probability (p = 0.53) than tracks (p = 0.18), and experienced observers had higher detection probabilities (p > 0.71) than novice observers (p < 0.55). Detection probabilities increased almost 3-fold as survey length increased from 200 m to 1,000 m, and otter sign was not concentrated near access points. After accounting for imperfect detection, our estimates of otter site occupancy based on a 400-m survey increased >3-fold, providing further evidence of the potential negative bias that can occur in estimates from sign surveys when imperfect detection is not addressed. Our study identifies areas for improvement in sign survey methodologies and results are applicable for sign surveys commonly used for many species across a range of habitats. © 2010 The Wildlife Society     

Using bioacoustics to enhance the efficiency of spotted owl surveys and facilitate forest restoration

The California spotted owl (Strix occidentalis occidentalis) is an older-forest associated species that resides at the center of forest management planning in the Sierra Nevada and Southern California, USA, which are experiencing increasingly large and severe wildfires and drought-related tree mortality. We leveraged advances in passive acoustic survey technologies to develop an acoustically assisted survey design that could increase the efficiency and effectiveness of project-level surveys for spotted owls, allowing surveys to be completed in a single year instead of in multiple years. We deployed an array of autonomous recording units (ARUs) across a landscape and identified spotted owl vocalizations in the resulting audio using BirdNET. We then evaluated spatio-temporal patterns in spotted owl vocalizations near occupied territories and the ability of a crew naïve to the location of occupied territories to locate spotted owls based on patterns of acoustic detections. After only 3 weeks of acoustic surveys, ≥1 ARU within 750 m of all 17 occupied territories obtained spotted owl detections across ≥2 nights. When active surveys using broadcast calling were conducted near ARUs with spotted owl detections by surveyors naïve to territory occupancy status and locations, surveyors located owls in 93% to 100% of occupied territories with ≤3 surveys. To further improve the efficiency of spotted owl surveys, we developed a statistical model to identify and prioritize areas across the Sierra Nevada for different survey methods (active only, acoustically assisted, no surveys) based on the expected probability of occupancy predicted from remotely sensed measurements of tree height and historical occupancy. Depending on managers' tolerance for false negatives, this model could help identify large areas that might not benefit from surveys based on low expected occupancy probabilities and areas where acoustically assisted surveys might enhance survey effectiveness and efficiency. Collectively, these findings can help managers streamline the survey process and thus increase the pace of forest restoration while minimizing potential near-term adverse effects on California spotted owls.     

Occupancy modeling of ruffed grouse in the Black Hills National Forest

Ruffed grouse (Bonasa umbellus) are a popular game bird and the management indicator species for quaking aspen (Populus tremuloides) in the Black Hills National Forest (BHNF), which requires development of a robust monitoring protocol to evaluate trends in ruffed grouse populations. We used roadside drumming surveys in spring 2007 and 2008 to estimate ruffed grouse occupancy and detection probabilities in the BHNF while simultaneously assessing the influence of sampling and site covariates on these processes. Ruffed grouse occupancy estimates were constant between spring 2007 and 2008 (Ψ = 0.12, SE = 0.03) and were positively influenced by the amount of aspen surrounding the site. Detection probability estimates were constant between spring 2007 and 2008 (p = 0.27, SE = 0.06) and were influenced by survey date in a quadratic form and negatively influenced by wind speed and time of the survey. Collectively, our results demonstrated that ruffed grouse occupancy and detection probabilities in the BHNF were low. Occupancy could be increased by increasing the extent of aspen. To improve monitoring efficiency and maximize probability of detecting ruffed grouse, ruffed grouse monitoring should be conducted during the peak of drumming (mid-May), during favorable weather conditions such as low wind speeds and little precipitation, and during early morning, near sunrise. © 2010 The Wildlife Society.     

ESTIMATES OF SPERM WHALE ABUNDANCE IN THE NORTHEASTERN TEMPERATE PACIFIC FROM A COMBINED ACOUSTIC AND VISUAL SURVEY

We estimate the abundance of sperm whales in a 7.8 million km² study area in the eastern temperate North Pacific using data from a ship-based acoustic and visual line-transect survey in spring 1997. Sperm whales were detected acoustically using a hydrophone array towed at 15 km/h and 100 m depth. The hydrophone array was towed for 14,500 km, and locations were estimated acoustically for 45 distinct sperm whale groups. Whales producing slow clicks (>2-s period) were detected at greater distance (up to 37 km), and the estimation of effective strip widths was stratified based on initial click period. Visual survey effort (using 25° binoculars and naked eyes) covered 8,100 km in Beaufort sea states 0–5 and resulted in only eight sightings. The effective strip width for visual detections was estimated from previous surveys conducted using the same methods and similar vessels in the eastern Pacific. Estimated sperm whale abundance in the study area was not significantly different between acoustic (32,100, CV = 0.36) and visual (26,300, CV = 0.81) detection methods. Acoustic techniques substantially increased the number of sperm whales detected on this line-transect survey by increasing the range of detection and allowing nighttime surveys; however, visual observations were necessary for estimating group size.     

Sampling Plant Diversity and Rarity at Landscape Scales: Importance of Sampling Time in Species Detectability

Documenting and estimating species richness at regional or landscape scales has been a major emphasis for conservation efforts, as well as for the development and testing of evolutionary and ecological theory. Rarely, however, are sampling efforts assessed on how they affect detection and estimates of species richness and rarity. In this study, vascular plant richness was sampled in 356 quarter hectare time-unlimited survey plots in the boreal region of northeast Alberta. These surveys consisted of 15,856 observations of 499 vascular plant species (97 considered to be regionally rare) collected by 12 observers over a 2 year period. Average survey time for each quarter-hectare plot was 82 minutes, ranging from 20 to 194 minutes, with a positive relationship between total survey time and total plant richness. When survey time was limited to a 20-minute search, as in other Alberta biodiversity methods, 61 species were missed. Extending the survey time to 60 minutes, reduced the number of missed species to 20, while a 90-minute cut-off time resulted in the loss of 8 species. When surveys were separated by habitat type, 60 minutes of search effort sampled nearly 90% of total observed richness for all habitats. Relative to rare species, time-unlimited surveys had ∼65% higher rare plant detections post-20 minutes than during the first 20 minutes of the survey. Although exhaustive sampling was attempted, observer bias was noted among observers when a subsample of plots was re-surveyed by different observers. Our findings suggest that sampling time, combined with sample size and observer effects, should be considered in landscape-scale plant biodiversity surveys.     

Designing occupancy surveys and interpreting non‐detection when observations are imperfect

Aim Conservation practitioners use biological surveys to ascertain whether or not a site is occupied by a particular species. Widely used statistical methods estimate the probability that a species will be detected in a survey of an occupied site. However, these estimates of detection probability are alone not sufficient to calculate the probability that a species is present given that it was not detected. The aim of this paper is to demonstrate methods for correctly calculating (1) the probability a species occupies a site given one or more non‐detections, and (2) the number of sequential non‐detections necessary to assert, with a pre‐specified confidence, that a species is absent from a site. Location Occupancy data for a tree frog in eastern Australia serve to illustrate methods that may be applied anywhere species’ occupancy data are used and detection probabilities are < 1. Methods Building on Bayesian expressions for the probability that a site is occupied by a species when it is not detected, and the number of non‐detections necessary to assert absence with a pre‐specified confidence, we estimate occupancy probabilities across tree frog survey locations, drawing on information about where and when the species was detected during surveys. Results We show that the number of sequential non‐detections necessary to assert that a species is absent increases nonlinearly with the prior probability of occupancy, the probability of detection if present, and the desired level of confidence about absence. Main conclusions If used more widely, the Bayesian analytical approaches illustrated here would improve collection and interpretation of biological survey data, providing a coherent way to incorporate detection probability estimates in the design of minimum survey requirements for monitoring, impact assessment and distribution modelling.     

甘肃莲花山斑尾榛鸡的春季领域行为

斑尾榛鸡(Tetrastes sewerzowi)为我国特有松鸡科鸟类,仅分布于我国中西部地区的高山针叶林中,为国家Ⅰ级重点保护野生动物。斑尾榛鸡雄鸟在春季繁殖季节占据并保卫领域。2012年4~5月,在甘肃省莲花山自然保护区,利用录音回放和放置标本模型模仿入侵者的方法,研究斑尾榛鸡的领域行为,重点关注斑尾榛鸡雄鸟领域行为的个体差异,以及与配对时间和领域质量的关系。研究发现,斑尾榛鸡的领域行为依据对入侵者驱赶程度不同分为观察确认、警告、追逐、对峙、跳起打斗5个阶段。斑尾榛鸡领域行为中会发出警告鸣叫,分析表明可分为4种类型。通过对6只无线电遥测斑尾榛鸡个体的实验研究,发现其个体间领域行为的强弱存在显著差异;配对前后领域行为的强弱也存在显著差异;而这种竞争力与雄鸟领域内柳树盖度显著相关。     

Modelling habitat selection of the cryptic Hazel Grouse Bonasa bonasia in a montane forest

The Hazel Grouse Bonasa bonasia is strongly affected by forest dynamics, and populations in many areas within Europe are declining. As a result of the ‘wilding’ concept implemented in the National Park Bavarian Forest, this area is one of the refuges for the species in Germany. Even though the effects of prevailing processes make the situation there particularly interesting, no recent investigation about habitat selection in the rapidly changing environment of the national park has been undertaken. We modelled the species–habitat relationship to derive the important habitat features in the national park as well as factors and critical threshold for monitoring, and to evaluate the predictive power of models based on field surveys compared to an analysis of infrared aerial photographs. We conducted our surveys on 49 plots of 25 ha each where Hazel Grouse was recorded and on an equally sized set of plots with no grouse occurrence, and used this dataset to build a predictive habitat-suitability model using logistic regression with backward stepwise variable selection. Habitat heterogeneity, stand structure, presence of mountain ash and willow, root plates, forest aisles, and young broadleaf stands proved to be predictive habitat variables. After internal validation via bootstrapping, our model shows an AUC value of 0.91 and a correct classification rate of 87%. Considering the methodological difficulties attached to backward selection, we applied Bayesian model averaging as an alternative. This multi-model approach also yielded similar results. To derive simple thresholds for important predictors as a basis for management decisions, we alternatively ran tree-based modelling, which also leads to a very similar selection of predictors. Performance of our different survey approaches was assessed by comparing two independent models with a model including both data resources: one constructed only from field survey data, the other based on data derived from aerial photographs. Models based on field data seem to perform slightly better than those based on aerial photography, but models using both predictor datasets provided the highest predictive accuracy.