One-third of North America’s bird species require immediate conservation action because of ongoing population declines and threats (North American Bird Conservation Initiative 2016), including climate change, pollutants, invasive species, and habitat loss and degradation (Robbins et al. 1989, Gurevitch and Padilla 2004, Both et al. 2006, Nocera et al. 2012, Hallmann et al. 2014, Stanton et al. 2018). Effective conservation planning and management actions for these species require accurate and cost-effective methods of monitoring species occurrence. However, for rare and even common species, nondetection error, when a species is falsely recorded as absent when it is present, can occur during occupancy surveys as a result of cryptic species behavior, low population density, or inadequate sampling effort due to limited time and resources (Gu and Swihart 2004). Nondetection error can result in inaccurate conclusions about species occupancy and mislead management actions (Kellner and Swihart 2014). In particular, reducing rates of nondetection error for at-risk species that typically occupy habitat types with higher risk of destruction should be a conservation focus to help species recovery.
The Chimney Swift (Chaetura pelagica) is one such species of bird that frequently occupies habitat at high risk of destruction or disturbance (COSEWIC 2007, Fitzgerald et al. 2014). Chimney Swifts are aerial insectivorous birds that nest and roost in human-made structures, including wells, silos, barns, and most commonly, old masonry chimneys (Steeves et al. 2014). This species is also experiencing widespread population declines in eastern North America (Cadman et al. 2007, Nebel et al. 2010, Smith et al. 2015, Michel et al. 2016). North American Breeding Bird Survey data indicate a population reduction of approximately 90% in Canada since 1970 (ECCC 2017), and data from the second Breeding Bird Atlas of Ontario (2001–2005) indicate substantial population declines and distributional losses compared to the first atlas (1981–1985; Cadman et al. 2007). Consequently, the Chimney Swift is listed as Threatened in both Canada and the Province of Ontario under the Species at Risk Act (2009) and Endangered Species Act (2009), respectively.
Though the cause(s) behind the population declines is not clear, reductions in the availability of aerial insects for food, as well as destruction of suitable nesting and roosting habitat, are frequently cited as the leading causes of Chimney Swift population declines (e.g., COSEWIC 2007, Nocera et al. 2012, Fitzgerald et al. 2014, Steeves et al. 2014). In particular, the availability of porous masonry chimneys that Chimney Swifts typically use for nesting and roosting is declining on account of chimney capping and screening to prevent animal entries, demolition of older buildings for redevelopment, and installation of metal liners and spark arresters for insurance purposes (Fitzgerald et al. 2014, Stewart et al. 2016, Wake 2016). For example, 47 out of 162 (29%) chimneys known to be used by nesting or roosting Chimney Swifts in London, Ontario between 2004 and 2013 were lost by 2015, mainly because of demolition or capping (Wake 2016). Although there is evidence suggesting that the availability of suitable chimney habitat is currently not limiting Chimney Swift populations in parts of their breeding range, the number of suitable sites is projected to continue to decline over the next several years as masonry chimneys are demolished or altered (Gauthier et al. 2007, Fitzgerald et al. 2014, Zanchetta et al. 2014). Thus, limiting the destruction of suitable nest and roost chimneys, as well as disturbance at these sites, is an important conservation objective for this species.
Reducing the incidence of habitat loss and disturbance requires a survey protocol that reliably determines if a chimney is used by roosting or nesting Chimney Swifts, and thus limits nondetection error, i.e., when a chimney occupied by Chimney Swifts is deemed unoccupied. Although examining the interior of a chimney or its cleanout for evidence of use, e.g., nests, eggshells, feathers, or guano, is the most reliable method of determining occupancy, access can be difficult. Therefore, occupancy is often determined by conducting surveys from the ground and observing if swifts enter or exit the chimney opening. Roost chimneys can typically be identified by conducting surveys around sunset during migration and the breeding season when groups of nonbreeding individuals conspicuously enter chimneys to roost communally (Stewart and Stewart 2010, Steeves et al. 2014; Bird Studies Canada, unpublished data). However, nest chimneys can be more difficult to confirm because of the cryptic and less predictable behavior of breeding Chimney Swifts. Only one breeding pair usually occupies a nest chimney, although one to two nonbreeding “helpers” have been documented at nest sites (Fischer 1958, Dexter 1969). Furthermore, a breeding pair makes inconspicuous trips in and out of the chimney throughout the day to build a nest, incubate eggs, or feed young (Kendeigh 1952, Fischer 1958, Stewart and Stewart 2010, Steeves et al. 2014).
We investigate factors influencing the probability of detecting Chimney Swifts at active nest chimneys in order to make recommendations on a survey protocol that reliably determines nesting occupancy of chimneys and thus reduces nondetection error. Previous studies have shown that the time between chimney entries and exits by nesting Chimney Swifts varies with weather, time of day, and nestling age (Kendeigh 1952, Fischer 1958, Zammuto et al. 1981, Stewart and Stewart 2010, 2013). In this study, we examine how temporal (time of day and season) and weather variables influence the probability of detecting Chimney Swifts at nest sites in southern Ontario, Canada. Our ultimate goal in doing so was to develop survey guidelines to reduce nondetection error during nest chimney surveys, which, in turn, will increase the number of active nest chimneys identified and subsequently conserved and protected as part of actions to help reverse or stabilize declining populations.
We recruited volunteers from Bird Studies Canada’s Citizen Science program, Ontario SwiftWatch, to monitor known nest sites of Chimney Swifts in seven urban centers throughout southern Ontario, including Barrie, Fort Erie, Kingsville, London, Stratford, Tillsonburg, and Toronto (Fig. 1).
Volunteers surveyed nest chimneys during the peak Chimney Swift nesting season (June–July) in 2016. All volunteers were experienced in identifying Chimney Swifts and with the monitoring protocols. Chimneys included in this study had been identified by volunteers as being occupied by nesting Chimney Swifts in a previous year, either through inspection of the chimney interior or ground observation of daytime activity, i.e., chimney entries and exits by Chimney Swifts. Observation of daytime activity at a chimney can be used to differentiate between roost and nest chimneys because Chimney Swifts typically return to roost chimneys only at sunset, whereas swifts visit nest chimneys throughout the day during the breeding season to nest build, incubate eggs, or feed young (Stewart and Stewart 2010).
At each nest chimney, volunteers conducted one to four standardized surveys between 1 June and 31 July, with up to one survey in each two-week period: early June (1 June–15 June), late June (16 June–30 June), early July (1 July–15 July), and late July (16 July–31 July). During each survey, one to two observers monitored the nest chimney for 60 min and recorded activity, i.e., Chimney Swift entered/exited chimney or no entry/exit, during each of the six 10-min intervals. Observers started surveys anytime between 0900 EST and one hour before sunset. If two or more surveys were conducted at a chimney, at least one survey was conducted during the day (0900 EST-1700 EST) and one in the evening (1700 EST-sunset). The mean number of days between consecutive surveys at sites was 14.6 days. Volunteers recorded air temperature (Celsius), wind speed (Beaufort scale), and proportion of cloud cover (to the nearest 10th) at the beginning of the survey. Surveys were not conducted if it was raining.
We used R version 3.3.1 (R Core Team 2016) with the unmarked package (Fiske and Chandler 2011) to conduct all analyses. To evaluate variables (see Table 1 for a list with associated mechanisms and predictions) affecting the detection probability of Chimney Swifts at nest sites, we fit dynamic occupancy models with removal design using the “colext” function in unmarked. Dynamic occupancy models are typically used in multiseason studies because of their hierarchical nature, i.e., repeated visits within repeated years. We opted to use dynamic occupancy models in our single-season study as we applied a similar hierarchical sampling protocol: we used repeated 10-min intervals during a 60-min survey, which was repeated at sites during different two-week periods, i.e., “seasons,” between 1 June and 31 July. We used a removal model because we were interested in determining the minimum length of time required to detect Chimney Swifts entering or exiting nest sites, rather than the frequency at which they are detected. In addition, we preferred a removal design because we suspected nonindependence between 10-min intervals within the 60-min survey on account of prolonged foraging and within-chimney bouts of nesting Chimney Swifts, especially earlier in the nesting season. In a removal occupancy model design, encounter events are only modeled until the first detection and sites only need to be observed until the species is first detected (MacKenzie and Royle 2005, Kéry and Royle 2016). For each 60-min survey in our study, we assigned each 10-min interval a binary value: Chimney Swift detected entering or exiting chimney = 1, Chimney Swift not detected entering or exiting chimney = 0. We then replaced observations during the 60-min survey with NAs if they occurred after the first detection. We scaled all continuous covariates by transforming data to a Z-distribution (mean = 0, standard deviation = 1). Independent variables were not strongly correlated (Spearman’s rank: -0.4 < r < 0.4).
We constructed 75 candidate models explaining variation in detection probability of Chimney Swifts entering or exiting nest chimneys. We listed all possible combinations of linear and quadratic date and time of day effects on detection probability, crossed with all combinations of weather variables, including air temperature, wind speed, and cloud cover. We did not include urban center as a variable in our models because we assumed the same patterns of detection across urban centers given their relatively close proximity within the Chimney Swift’s breeding range. Our candidate model set also included null and global (full) models. All multivariate models included only additive effects, i.e., no interactions.
Because we were interested in only detection probability (p), we did not include variables in the initial occupancy (Ψ), colonization (γ), or extinction (ε) components of the models, and thus, expected these parameters to be constant across the duration of the study. Initial occupancy in the model is the occupancy for the first “season” or two-week period in our study. We carried out model selection using Akaike’s Information Criterion (AIC). We considered the best model to be the model with ΔAIC equal to zero. We addressed model uncertainty using model averaging of the parameter estimates from the model confidence set (ΔAIC ≤ 6; Richards 2005) with the AICcmodavg package (Mazerolle 2017). We computed the model-averaged parameters using the natural average method (Burnham and Anderson 2002). We tested the fit of the most complex model using parametric bootstrapping and the chi-square goodness-of-fit test using the frequencies of detection histories; P values > 0.05 indicated adequate fit (Fiske and Chandler 2019).
We estimated the probability of detecting Chimney Swifts over a 60-min survey period as a function of covariates using predicted values derived from model-averaged estimates, while holding all other variables at their median values. We calculated detection probability across all 10-min intervals in a survey using ps = 1-(1-pi)6, where ps is the probability of observing at least one entry or exit of Chimney Swift during a 60-min survey, and pi is the probability of observing at least one entry or exit of a Chimney Swift during a 10-min survey interval. We also used the null model to calculate the average detection probability across a 60-min survey, including confidence intervals calculated using the delta method. We ranked the relative strength of the influence of all of the factors we considered through direct comparison of model-averaged coefficients, which was possible given that all covariates were standardised by z-transformation prior to modeling. We also estimated detection probability as a function of the length of survey using predicted values derived from model-averaged estimates.
Volunteers completed 136 surveys (63 daytime, 73 evening), with one to four surveys at 47 nest chimneys in seven urban areas in southern Ontario (Fig. 1). Chimney Swifts were detected during at least one survey at all but three chimneys (9 out of 136 surveys had zero detection during 60 min). Based on the null model, the probability of initial occupancy, colonization, and extinction were 0.96 (± 0.04 SE), 0.002 (± 0.052 SE), and 0.12 (± 0.05 SE), respectively. Average detection probability across a 60-min survey was 0.87 (95% CI: 0.80, 0.92). The most complex model adequately fit the observed data (χ² = 315, P = 0.99).
The best model (ΔAIC = 0) explaining variation in detection probability of Chimney Swifts included linear date and cloud cover (Table 2). However, the best model had an Akaike weight of 0.11, which is only 1.4 times higher than the second best model. Furthermore, the best model is closely followed by seven other models with a ΔAIC < 2, although most of these models included date and cloud cover effects.
Model-averaged coefficients for the model confidence set (ΔAIC ≤ 6) showed that cloud cover had the strongest influence, followed by linear date, air temperature, wind speed, and time of day (Table 3). Using model-averaged estimates, we found that cloud cover had a negative effect on detection probability (Fig. 2). Linear date had a positive effect on detection probability, where detection probability remained greater than 0.95 between 25 July and 1 August. However, the confidence limits for the model-averaged estimate of linear date slightly overlapped zero, and detection probability was 0.86 at the beginning of the nesting period and increased to 0.96 by the end of the nesting period (Fig. 2). We also made model predictions for the effect of length of survey on detection probability for early and late nesting periods under clear and overcast skies using model-averaged estimates of the candidate model set. Under these survey scenarios, only clear skies (0% cloud cover) during the late nesting period resulted in greater than 0.95 detection probability (Fig. 3).
Overall, we found that initial occupancy during the first “season” or two-week period was high (0.96), which was not surprising given that we only included chimneys known to be used for nesting in previous years. There were few changes in chimney use over the following two-week periods, including almost no colonization of sites (0.002) and some extinction (0.12), which might have been a result of nest failure. We found that detection probability of Chimney Swifts at nest chimneys was negatively influenced by increasing cloud cover. We also found weak evidence that detection probability was positively influenced by increasing date within the nesting period. Surveys conducted for 60 min under clear skies and during the late nesting period (late July) resulted in a mean detection probability of 0.95 or higher. All other scenarios resulted in lower detection probabilities (Figs. 2 and 3).
The results of our study suggest that the probability of detecting Chimney Swifts at nest sites is higher when cloud cover is low. Chimney Swifts are aerial insectivores that rely on the availability of flying insects for prey, including food to provision young, e.g., beetles, true bugs, caddisflies, stoneflies, ants, and bees (Fischer 1958, Nocera et al. 2012). Weather conditions that reduce insect abundance or availability will likely result in poorer foraging efficiency for swifts, and thus, fewer visits to nest chimneys. For example, studies of other aerial insectivores have shown negative effects of cloud cover on both aerial insect abundance and rates of food-provisioning to young (Lack and Lack 1951, Finlay 1976, Turner 1983). Cloudy conditions could lead to longer foraging bouts, possibly further away from nest sites, because of the challenges of finding or catching aerial insects, and ultimately reduce detection probability at nest sites. Chimney Swifts might also perceive darker skies as a cue for imminent inclement weather, resulting in extended incubation or brooding bouts inside the chimney (Steeves et al. 2014). We did not measure cloud type, e.g., stratus versus cumulus clouds, in our study, however, there could be an interaction of cloud type and cloud cover effects on detection probability. For example, Chimney Swifts might make fewer entries and exits at nest sites during periods of high cumulonimbus cloud cover than periods of high cirrus cloud cover, as the former typically precedes inclement weather. In addition, increasing cloud cover, particularly of darker, lower forming clouds, might negatively affect an observer’s ability to accurately detect Chimney Swifts entering or exiting a chimney because of reduced visibility.
Though linear date appeared in the top-ranked model (Table 2), support for this model was low (Akaike weight = 0.11) and several other models ranked similarly high (ΔAIC < 2). As well, the 95% confidence limits around the parameter estimate for linear date included zero (Table 3). However, linear date had a reasonably large effect size compared to the other less important covariates, and we know time of season is biologically important for detection probability of adult Chimney Swifts at nest sites based on previous studies (Zammuto et al. 1981, Stewart and Stewart 2010). In particular, we found that detection probability was highest later in the nesting season, i.e., late July. The month of July coincides with the nestling-rearing period of most Chimney Swifts in the northern portion of their breeding range, e.g., New York, Ontario, and Manitoba (Fischer 1958, Stewart and Stewart 2013, Steeves et al. 2014; Bird Studies Canada, unpublished data). Adults should be entering and exiting chimneys more frequently to feed growing young at this time compared to earlier in the season when breeding pairs are nest building or incubating eggs, i.e., early to mid-June. Consistent with our findings, previous work in Manitoba found that Chimney Swift adults visited nests more frequently in July during nestling rearing (6–8 visits per hour between mid-July and fledging) than in June during incubation (1–2 visits per hour; Stewart and Stewart 2010, 2013). We did not find evidence of time of day effects on the detection probability of Chimney Swifts entering or exiting nest chimneys, although previous work on Chimney Swifts and other insectivorous birds during the nestling-rearing period found that adults tended to visit nests more frequently in the early morning or early evening compared to mid-day (Pinkowski 1978, Walsh 1978, Zammuto et al. 1981).
Based on our findings, we recommend that observational surveys conducted from the ground that aim to identify active Chimney Swift nest sites in southern Ontario, Canada should be conducted for at least one hour during clear sky conditions, and as late in the nesting period as possible. One-hour surveys conducted under clear skies between 25 July and 1 August will likely have the highest detection probability of 95% or more. One-hour surveys conducted earlier in the nesting period will likely have lower probability of detection, but still 86% or higher. Surveys should be at least 60 min in duration because surveying for less time under most scenarios results in much lower probability of detection. If surveys occur earlier in the nesting period during June or in less optimal weather conditions, increasing the duration of the survey might result in higher detection probability. At seven chimneys included in our study, Chimney Swifts were detected during the first and last visits, but not during a middle visit, suggesting that a longer survey might have resulted in positive detection. Alternatively, three or more 20-min visits (or perhaps other arrangements totalling 60 min of observation) might also reduce the chances of nondetection error. During surveys, the chimney opening should be continuously observed for the entire survey period to reduce the chance of false negative results because Chimney Swifts entering or exiting a nest site can be easily missed (E. Purves, personal observation). Our findings suggest that time of day does not influence detection probability when surveys start between 0900 and one hour before sunset. Previous work has shown that frequency of nest visits by adult Chimney Swifts varies with time of day, but no statistical testing was conducted on these data (Zammuto et al. 1981). We also recommend avoiding surveying when cloud cover is greater than 10%. It is important to note that observational surveys conducted from the ground should only be considered the most reliable method of determining nest occupancy in cases where the chimney interior cannot be inspected for direct evidence of nesting.
Other factors not included in our models that could influence detection probability of Chimney Swifts at nest sites include predators in the vicinity of nests (Ghalambor and Martin 2000, Fontaine and Martin 2006), weather conditions preceding the day of surveying that affect nutritional demands of adults and young (Rose 2009), brood size, which positively affects the number of food-provisioning trips made by adult swifts (Lack and Lack 1951, Zammuto et al. 1981), additional “helper” adults attending young (Steeves et al. 2014), number of observers conducting the occupancy survey, and nest failure. In particular, rates of nest failure in Chimney Swifts are not well understood in Ontario, but in Manitoba, 19 of 30 monitored nesting attempts between 2007 and 2013 failed (63%; Stewart and Stewart 2013). Chimney Swifts at northern latitudes are typically single-brooded (Baicich and Harrison 1997, Stewart and Stewart 2010, 2013); however, renesting following early nest failure is possible (Dexter 1969). For example, in Kansas, up to 60% of breeding Chimney Swifts renest following a heavy rainfall event (Steeves et al. 2014). Any nests that fail because of extreme weather or other reasons would not likely be detected following the recommended protocol if nest failure occurred before late July and renesting was not attempted by the breeding pair. In future studies, the dynamic occupancy design used in our study could be used to estimate nest failure by modeling extinction probability. In addition, although adult Chimney Swifts have high nest site fidelity (Fischer 1958, Dexter 1992), traditionally occupied nest sites can sometimes go unoccupied during at least one year between occupied years (Bird Studies Canada, unpublished data). Thus, chimneys that are unoccupied during any given season might still provide available and suitable nesting habitat for Chimney Swifts in a future year.
In conclusion, we found evidence that cloud cover and, to some extent, date influence the probability of detecting Chimney Swifts at active nest chimneys, which has important implications for survey protocol design. The results of this study provide recommendations on a survey protocol that will increase the chances of accurately identifying nest chimneys, and thus, reduce the incidence of nondetection error during chimney surveys. Limiting nondetection error during chimney surveys is an important strategy for reducing Chimney Swift nesting disturbance and habitat loss, and might be important for reversing or stabilizing Chimney Swift population declines, particularly as the number of suitable nest sites continues to decrease over time.
We would like to first and foremost thank the dedicated Ontario SwiftWatch volunteers for their tireless efforts in monitoring and stewarding Chimney Swifts. Special thanks are owed to the Chimney Swift monitors in Barrie, Essex County, Fort Erie, London, Stratford, and Toronto, whose valuable time and effort made this project possible. We are grateful to I. Brinckmann, L. Goodridge, D. Riley, P. Tremblay, and W. Wake for helping to coordinate and recruit monitors in their communities for this project. Financial assistance for this project was provided by the Government of Ontario and the Government of Canada. We thank two anonymous reviewers for comments that improved the paper.
Baicich, P. J., and C. J. O. Harrison. 1997. A guide to the nests, eggs and nestlings of North American birds. Academic, San Diego, California, USA.
Both, C., S. Bouwhuis, C. M. Lessells, and M. E. Visser. 2006. Climate change and population declines in a long-distance migratory bird. Nature 441:81-83. http://dx.doi.org/10.1038/nature04539
Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach. Second edition. Springer, New York, New York, USA.
Cadman, M. D., D. A. Sutherland, G. G. Beck, D. Lepage, and A. R. Couturier. 2007. Second atlas of breeding birds of Ontario (2001-2005). Bird Studies Canada, Environment Canada, Ontario Field Ornithologists, Ontario Ministry of Natural Resources, and Ontario Nature, Toronto, Ontario, Canada.
Committee on the Status of Endangered Wildlife in Canada (COSEWIC). 2007. Chimney swift (Chaetura pelagica) COSEWIC assessment and status report. COSEWIC, Ottawa, Ontario, Canada. [online] URL: http://www.sararegistry.gc.ca/default.asp?lang=En&n=B2AFC099-1
Dexter, R. W. 1969. Banding and nesting studies of the Chimney Swift, 1944-1968. Ohio Journal of Science 69:193-213.
Dexter, R. W. 1992. Sociality of Chimney Swifts (Chaetura pelagica) nesting in a colony. North American Bird Bander 17:61-64.
Environment and Climate Change Canada (ECCC). 2017. Breeding Bird Survey results, Data-version 2015. Environment and Climate Change Canada, Gatineau, Québec, Canada. [online] URL: http://wildlife-species.canada.ca/breeding-bird-survey-results/
Finlay, J. C. 1976. Some effects of weather on Purple Martin activity. Auk 93:231-244.
Fischer, R. B. 1958. The breeding biology of the Chimney Swift Chaetura pelagica (Linnaeus). New York State Museum and Science Service Bulletin: 368.
Fiske, I., and R. Chandler. 2011. unmarked: An R package for fitting hierarchical models of wildlife occurrence and abundance. Journal of Statistical Software 43:1-23. http://dx.doi.org/10.18637/jss.v043.i10
Fiske, I., and R. Chandler. 2019. Overview of unmarked: an R package for the analysis of data from unmarked animals. R Project for Statistical Computing, Vienna, Austria. [online] URL: http://cran.r-project.org/web/packages/unmarked/vignettes/unmarked.pdf
Fitzgerald, T. M., E. van Stam, J. J. Nocera, and D. S. Badzinski. 2014. Loss of nesting sites is not a primary factor limiting northern Chimney Swift populations. Population Ecology 56:507-512. http://dx.doi.org/10.1007/s10144-014-0433-6
Fontaine, J. J., and T. E. Martin. 2006. Parent birds assess nest predation risk and adjust their reproductive strategies. Ecology Letters 9:428-434. http://dx.doi.org/10.1111/j.1461-0248.2006.00892.x
Gauthier, J., M. Dionne, C. Maurice, J. Potvin, M. D. Cadman, and D. Busby. 2007. Status of the Chimney Swift (Chaetura pelagica) in Canada. Technical Report Series No. 477, Canadian Wildlife Service, Environment Canada, Québec, Canada.
Ghalambor, C. K., and T. E. Martin. 2000. Parental investment strategies in two species of nuthatch vary with stage-specific predation risk and reproductive effort. Animal Behaviour 60:263-267. https://doi.org/10.1006/anbe.2000.1472
Gu, W., and R. K. Swihart. 2004. Absent or undetected? Effects of non-detection of species occurrence on wildlife-habitat models. Biological Conservation 116:195-203. http://dx.doi.org/10.1016/S0006-3207(03)00190-3
Gurevitch, J., and D. K. Padilla. 2004. Are invasive species a major cause of extinctions? Trends in Ecology and Evolution 19:470-474. https://doi.org/10.1016/j.tree.2004.07.005
Hallmann, C. A., R. P. B. Foppen, C. A. M. van Turnhout, H. de Kroon, and E. Jongejans. 2014. Declines in insectivorous birds are associated with high neonicotinoid concentrations. Nature 511:341-343. http://dx.doi.org/10.1038/nature13531
Kellner, K. F., and R. K. Swihart. 2014. Accounting for imperfect detection in ecology: a quantitative review. PLoS ONE 9(10):e111436. https://doi.org/10.1371/journal.pone.0111436
Kendeigh, S. C. 1952. Parental care and its evolution in birds. Illinois Biological Monographs 18:1-356. http://dx.doi.org/10.5962/bhl.title.50337
Kéry, M., and J. A. Royle. 2016. Applied hierarchical modelling in ecology: analysis of distribution, abundance and species richness in R and BUGS. Volume 1: prelude and static models. Academic Press, San Diego, California, USA.
Lack, D., and E. Lack. 1951. The breeding biology of the swift Apus apus. Ibis 93:501-546. http://dx.doi.org/10.1111/j.1474-919X.1951.tb05457.x
MacKenzie, D. I., and J. A. Royle. 2005. Designing occupancy studies: general advice and allocating survey effort. Journal of Applied Ecology 42:1105-1114. https://doi.org/10.1111/j.1365-2664.2005.01098.x
Mazerolle, M. J. 2017. AICcmodavg: model selection and multimodel inference based on (Q)AIC(c). R package version 2.1-1. [online] URL: https://cran.r-project.org/package=AICcmodavg
Michel, N. L., A. C. Smith, R. G. Clark, C. A. Morrissey, and K. A. Hobson. 2016. Differences in spatial synchrony and interspecific concordance inform guild-level population trends for aerial insectivorous birds. Ecography 39:774-786. https://doi.org/10.1111/ecog.01798
Nebel, S., A. Mills, J. D. McCracken, and P. D. Taylor. 2010. Declines of aerial insectivores in North America follow a geographic gradient. Avian Conservation and Ecology 5(2):1. http://dx.doi.org/10.5751/ACE-00391-050201
Nocera, J. J., J. M. Blais, D. V. Beresford, L. K. Finity, C. Grooms, L. E. Kimpe, K. Kyser, N. Michelutti, M. W. Reudink, and J. P. Smol. 2012. Historical pesticide applications coincided with an altered diet of aerially foraging insectivorous Chimney Swifts. Proceedings of the Royal Society B: Biological Sciences 279:3114-3120. https://doi.org/10.1098/rspb.2012.0445
North American Bird Conservation Initiative. 2016. The state of North America’s birds 2016. Environment and Climate Change Canada, Ottawa, Canada. [online] URL: http://www.stateofthebirds.org/2016/#_ga=2.15106750.946038313.1549553503-110771783.1549553503
Pinkowski, B. C. 1978. Feeding of nestling and fledgling Eastern Bluebirds. Wilson Bulletin 90:84-98.
R Core Team. 2016. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. [online] URL: https://www.R-project.org/
Richards, S. A. 2005. Testing ecological theory using the information theoretic approach: examples and cautionary results. Ecology 86:2805-2814. https://doi.org/10.1890/05-0074
Robbins, C. S., J. R. Sauer, R. S. Greenburg, and S. Droege. 1989. Population declines in North American birds that migrate to the neotropics. Proceedings of the National Academy of Sciences of the United States of America 86:7658-7662. http://dx.doi.org/10.1073/pnas.86.19.7658
Rose, A. P. 2009. Temporal and individual variation in offspring provisioning by Tree Swallows: a new method of automated nest attendance monitoring. PLoS ONE 4(1):e4111. https://doi.org/10.1371/journal.pone.0004111
Smith, A. C., M.-A. R. Hudson, C. M. Downes, and C. M. Francis. 2015. Change points in the population trends of aerial-insectivorous birds in North America: synchronized in time across species and regions. PLoS ONE 10(7): e0130768. https://doi.org/10.1371/journal.pone.0130768
Stanton, R. L., C. A. Morrissey, and R. G. Clark. 2018. Analysis of trends and agricultural drivers of farmland bird declines in North America: a review. Agriculture, Ecosystems and Environment 254:244-254. https://doi.org/10.1016/j.agee.2017.11.028
Steeves, T. K., S. B. Kearney-McGee, M. A. Rubega, C. L. Cink, and C. T. Collins. 2014. Chimney Swift (Chaetura pelagica), version 2.0. In A. F. Poole, editor. The birds of North America online. Cornell Lab of Ornithology, Ithaca, New York, USA. https://doi.org/10.2173/bna.646
Stewart, B. E., and R. E. A. Stewart. 2010. Nest site use and breeding success of Chimney Swifts in St. Adolphe, MB, 2007-2009. Blue Jay 68:124-132.
Stewart, B. E., and R. E. A. Stewart. 2013. Nesting site use, breeding success, and reproductive rates of Chimney Swifts in St. Adolphe, MB, 2010-2013. Blue Jay 71:166-182.
Stewart, R. E. A., T. F. Poole, C. Artuso, and B. E. Stewart. 2016. Loss and preservation of Chimney Swift habitat in Manitoba, 2007-2016. Blue Jay 75:11-15.
Turner, A. K. 1983. Food selection and the timing of breeding of the Blue-and-White Swallow Notiochelidon cyanoleuca in Venezuela. Ibis 125:450-462. http://dx.doi.org/10.1111/j.1474-919X.1983.tb03141.x
Wake, W. 2016. Loss of chimneys used by Chimney Swifts in London, Ontario, 2004-2013. Cardinal 243:33-38.
Walsh, H. 1978. Food of nestling Purple Martins. Wilson Bulletin 90:248-260.
Zammuto, R. M., E. C. Franks, and C. R. Preston. 1981. Factors associated with the interval between feeding visits in brood-rearing Chimney Swifts. Journal of Field Ornithology 52:134-139.
Zanchetta, C., D. C. Tozer, T. M. Fitzgerald, K. Richardson, and D. Badzinski. 2014. Tree cavity use by Chimney Swifts: implications for forestry and population recovery. Avian Conservation and Ecology 9(2):1. http://dx.doi.org/10.5751/ACE-00677-090201