Citizen Science Observations Reveal Long-Term Population Trends of Common and Pacific Loon in Urbanized Alaska

Common loons Gavia immer and Pacific loons Gavia pacifica are piscivorous waterbirds that migrate each spring from coastal and marine wintering areas to breed in northern freshwater habitats (Russell 2002; Evers et al. 2010). Approximately 2% of the world's common loon population (∼620,000 birds) and 50–70% of the world's Pacific loon population (∼125,000 birds) breed in Alaska (Groves et al. 1996; Russell 2002; Evers 2004). Although most inhabit remote and undeveloped areas, those within the southcentral region of Alaska (i.e., Municipality of Anchorage [Anchorage], the Matanuska-Susitna Valley [Mat-Su], and the Kenai Peninsula [Kenai]) inhabit an area with ∼55% of the state's human population (Alaska Department of Labor and Workforce Development 2017). Impacts to loons from human activities or development can include failed nesting attempts (Kelly 1992; McCarthy and Destefano 2011), death of a breeding adult (Sidor et al. 2003), and habitat or nest disturbance that reduces pair densities and lowers chick production (Badzinski and Timmermans 2006). Competition between co-occurring loon species for potentially limiting resources may also affect breeding opportunities (Haynes et al. 2014).

To maximize the spatial scope of loon monitoring in southcentral Alaska, the Alaska Department of Fish and Game (ADF&G) initiated the Alaska Loon Watch citizen science program in 1985 (ADF&G 2018). Citizen science engages nonprofessionals in scientific research and can be a valuable tool for monitoring many species, including loons (e.g., McNichol et al. 1995). The approach empowers citizen volunteers to collect ecological data on large spatial and temporal scales, with the added benefit of enhancing opportunities for science education (Dickinson et al. 2012). By engaging citizen scientists, biologists in southcentral Alaska were able to collect yearly loon presence information across an area nearly the size of Missouri. However, it was not until the program transferred to the U.S. Fish and Wildlife Service in the 1990s that it gained momentum as a mechanism for increasing awareness of loons nesting in the Anchorage community and other nearby southcentral subregions (Joint Base Elmendorf-Richardson [JBER], Kenai, Mat-Su, and areas that are undeveloped with a low human population density [Outlying]). The Alaska Department of Fish and Game once again took over citizen science coordination from 2005 to present. The 31-y program engaged 10 to 90 volunteers per year, with many volunteers participating for multiple and consecutive years.

Our objective was to analyze these 31 y of loon citizen science data to estimate long-term trends (percentage of change per year) in common and Pacific loon adult abundance and within-summer chick survival. In addition, we evaluated effects of lake size, human disturbance, and lakeshore development on loon abundance. Lastly, we examined the dataset for patterns consistent with interspecific competition, given the previously documented potential for competitive interactions to affect breeding of Pacific loons in northern Alaska (Haynes et al. 2014).

Southcentral Alaska comprises an expansive network of boreal freshwater lakes that attract both loon species. We examined loon trends in five subregions of southcentral Alaska with varying levels of human population density and activity: 1) Anchorage, 2) JBER, 3) Kenai, 4) Mat-Su, and 5) Outlying (Figure 1). Anchorage is the most populous city in Alaska, with nearly 300,000 residents and an extensive mosaic of urban and residential areas. The JBER is located just north of Anchorage and is an active U.S. Air Force and Army base. Kenai and Mat-Su comprise of residential and rural areas, whereas Outlying areas are undeveloped with a low human population density.

Common and Pacific loon breeding ranges overlap for much of Alaska, although their preferences for lake characteristics differ. Common loons breed on clear, oligotrophic freshwater lakes with floating bogs, small islands, and bays surrounded by forest and rocky shorelines (Barr 1996). Pacific loons typically nest on freshwater lakes in treeless or wooded habitats in arctic and subarctic tundra and taiga and generally prefer lakes with less emergent vegetation than other small-bodied loon species (Bergman and Derksen 1977).

Between 1985 and 2015, volunteer citizen scientists observed 346 lakes (selected based on citizen scientists' interest) and wetlands in our five subregions. Observed lakes ranged in size from less than 1 to more than 10,000 ha. We recruited volunteers via newsletters, outreach presentations, and word of mouth. Each year, we provided volunteers with a datasheet to record all sightings of adult, juvenile, and hatchling loons between May and September, along with information on nest locations and disturbances of adults due to human activities. Volunteers recorded observations throughout the season and returned datasheets, nest location maps, and photos by mail in the fall. Some volunteers participated for multiple consecutive years, whereas others participated occasionally or only once, making the data history at any given lake variable with respect to sample size and longevity of the observer. We transcribed all data received each year into an electronic database. We quality checked the transcription process before this analysis by comparing a random sample of one-third of all hardcopy records against the electronic database to look for data entry errors, which we deemed acceptable at 6% of all records examined.

The citizen science data we present are based on causal observations created by an unstructured survey protocol; such datasets are not easily analyzed using inferential statistics (Kamp et al. 2016). Therefore, we made several key assumptions about our loon data during analysis. First, we assumed perfect detection of all loons present on surveyed lakes; because loons are large, conspicuous, and vocal occupants over the entire summer, this is likely reasonable. Second, we assumed that surveyed lakes were representative of all lakes in the subregion (i.e., our dataset approximates a random sample); this is more likely to be true for subregions with large numbers of surveyed lakes (e.g., Mat-Su) than for subregions with less effort (e.g., Outlying). Lastly, we assumed that “missing” counts (years without observational records at a given lake) occurred at random throughout the dataset. In other words, the number of loons on a lake and trend in counts at that lake were unrelated to survey effort. Careful assessment of this final assumption was necessary, because greater survey effort (more participants and surveys) occurred during the first 15 y of the program. This pattern could lead to positively biased trend estimates; that is, if, over time, participants discovered and preferentially surveyed lakes with loons, but avoided surveying lakes with few or no loons (Kamp et al. 2016).

Because of wide variation in citizen scientist participation, the number of years of data available for each lake varied considerably. To include as many of the data as possible while reducing bias due to small sample sizes at individual lakes, we included all lakes in the analysis that met the following requirements: 1) the lake included at least one non-zero count for either species of loon over the 30-y interval, and 2) the lake was observed a minimum of three times over a span of at least 5 y. We also considered more conservative selection criteria that further reduced the number of lakes per subregion available for analysis. These alternative criteria produced similar results and are not presented here. In addition, we removed one large lake (Skilak Lake in the Kenai Peninsula subregion) from the analysis, despite meeting our inclusion criteria. Because of the lack of regular sampling effort, its extreme size (10,029 ha) compared to other lakes in our study, and high median count of loons (37 individuals), Skilak Lake had a disproportionate effect on trend estimates compared to all other lakes sampled.

We used generalized linear mixed models (Littell et al. 2006) assuming Poisson error distribution and log link function to estimate the trends by subregion in counts of adult common and Pacific loons from 1985 to 2015. We defined trend as the average rate of change in lake-specific counts (percentage of change per year) over the 30-y interval. In each analysis (loon species by subregion combinations), the maximum count of adult loons on a lake within a year was the response variable with year as the predictor variable. Individual lakes within subregions were included as random effects for both intercepts and slopes. Degrees of freedom for calculating confidence intervals on the estimated trends were based on the number of lakes in the region. In computing the regional trends, we weighted the effect of the counts for each lake by using the median abundance across years from that lake; this was based on the assumption that lakes with more loons made a greater contribution to the regional population count and thus had a greater effect on the true underlying trend. To illustrate among-lake variation in counts, we also fit a model to estimate trends separately for each lake.

As previously stated, our data were not collected with a predetermined, probability-based sampling plan, and the number of surveys conducted annually by volunteers decreased over time. Hence, if nonrandom sampling occurred over time (e.g., volunteers selectively dropped “less interesting” lakes where loons were less likely to occur), we could see positively biased estimated trends (Kamp et al. 2016). To assess this possibility, we used generalized linear mixed models (binomial error, logit link) to determine whether the proportion of lakes with zero counts changed over time. We fit models with no time pattern in zero counts, as well as models with a linear (on the logit scale) change. We then compared these models using the small sample-corrected version of Akaike's Information Criterion (AICc; Burnham and Anderson 2002).

We examined the effects of lake area, human disturbance, and lakeshore development on adult loon counts. We obtained lake area information from the U.S. Geological Survey (USGS) National Hydrography Dataset (USGS 2018). To quantify disturbance, we used citizen scientist reports on the frequency of seven types of potential disturbance (i.e., swimming, fishing, nonmotorized watercraft, motorized boats, jet skis, dogs, other) that were ranked as occurring daily (4), weekly (3), monthly (2), occasionally (1), or not occurring (0). For each lake–year combination, we used the maximum frequency across the seven disturbance types as an index of disturbance. We also tallied the number of disturbance types (out of a maximum of the seven listed above) as another measure of disturbance. We quantified development as the number of structures (e.g., cabins, homes, barns) within 150 m of the shoreline of each lake based on Google Earth^® aerial image. Aerial imagery was inconsistently available for lakes across all subregions between 1996 and 2015; therefore, changes in the number of structures over time is a component of these analyses, but we did not directly estimate the effect, if any, of increasing structure density on loon abundance.

To determine possible relationships between counts of adult loons and environmental factors, we used adult loon counts as the response variable and the natural log (ln) of lake area, maximum disturbance frequency, the number of disturbance types, and magnitude of shoreline development as predictor variables. Unfortunately, disturbance and development data were not available for all lakes, so we used separate subsets of the data within each subregion to investigate these effects. Consequently, models of disturbance and development cannot be compared directly (e.g., via AICc weights) with each other or with full-dataset models.

To investigate the potential for interspecific competition between loon species, we used two analyses. First, in models of Pacific loon abundance, we included common loon presence as a predictor variable. Second, we estimated the “persistence” probability of Pacific loons occurring at a lake, when common loons were either absent or present. We defined persistence as the probability of Pacific loons occurring on a lake if they had been observed on that lake in the preceding survey year. We estimated the effect of common loon presence on persistence probability of Pacific loons via logistic regression (Hosmer and Lemeshow 2000). The regression included an autoregressive error term for observations on the same lake to account for potentially autocorrelated residuals resulting from repeated observations on lakes where the same loons may breed for multiple years (Littell et al. 2006).

Citizen scientists observed loon chicks from hatch in June through fledge in September. In southcentral Alaska, both species typically hatch during the week of 25 June (n = 15 nests; ADF&G, unpublished data). Although fledging date is variable and depends on weather conditions, most common loon chicks fledge and are nutritionally independent by 11 wk (Barr 1996; Evers et al. 2010), whereas Pacific loons fledge after 9 weeks (Petersen 1989). We used this information to designate a “survival threshold” for broods of each species (16 September for common loons and 26 August for Pacific loons). If we observed chicks on or after the species-specific threshold date, we assumed they survived. Because we did not have a population of common and Pacific loons marked as chicks, our survival estimates only cover the period from hatch to fledge each year, and not annual survival. We treated each brood as a separate observation in generalized linear mixed models (binomial error structure, logit link), and used the ratio of fledged brood size to initial brood size as our estimate of survival and the response variable. For each species and subregion, we fit two models: the first estimated average chick survival probability across all years, and the second estimated a linear trend in chick survival through time.

We estimated that common loon counts increased 1–3% per year across four of five subregions of southcentral Alaska between 1985 and 2015 (Table 1; Figures 2A through 2E). With the exception of JBER (Figure 2B), there was substantial among-lake variation in trends. Lakes with few counts often had extreme trend estimates. In the Outlying subregion (Figure 2E), the confidence interval for the trend not only included zero but this sub-region also had the fewest surveys per year. For four of five subregions, analyses suggested a decline in the number of lakes surveyed that had no common loons (i.e., higher AICc weights for year-effect models and negative slopes; Table 1), which could indicate a potential positive bias in the estimated trend caused by preferentially including lakes with loons in later years. However, confidence intervals for three of four negative slopes included zero (Table 1). This indicates that any negative trend in the number of lakes surveyed without loons is relatively weak, leading us to conclude that any preferential surveys of lakes had little positive bias in adult trend estimates. In addition, common loons colonizing previously unoccupied lakes could also cause the pattern described above.

We estimated that Pacific loon counts declined by 3% per year in the Anchorage subregion (Figure 3A). Confidence intervals for trend estimates for this species in other subregions, however, included zero (i.e., less confidence that populations were changing) and were less precise than those for common loons (Table 1). This reflects the fact that Pacific loons were largely observed on fewer lakes than were common loons. Similar to results for common loon, we found no evidence suggesting that lakes without Pacific loons were preferentially dropped from the survey in later years (Table 1).

Lake characteristics (individual lake effect and natural log of lake area) were strong predictors of loon abundance. Lake size [as measured by ln(area)] was strongly related to loon counts in most areas for both species, but it did not explain loon count patterns as well as individual lake models (Table 2). Common loon abundance was positively related to ln(area), whereas Pacific loon abundance was negatively related to ln(area), except for the Mat-Su subregion, where the relationship between Pacific loons and ln(area) was positive (Table 3).

There was little evidence that adult loon counts were related to disturbance measures or the number of structures along a shoreline (Table 2), but our sample size was relatively small (Table S2, Supplemental Material). Across subregions, less than 50% of all citizen scientist observations recorded disturbance information, which weakens our ability to make strong conclusions. However, in the Kenai and Mat-Su subregions, common loon adult counts were positively related to human disturbance factors and the number of structures (Table S4, Supplemental Material). In Anchorage and the Kenai subregion, common loon counts were also positively related to the number of shoreline structures (Table 2). It is unlikely loons select lakes based on human occupancy, but rather loons and humans likely are attracted to similar lacustrine habitats (McCarthy and Destefano 2011).

Pacific loons also showed a mix of positive and negative patterns relative to disturbance (Table 3). In the JBER subregion, adult counts were negatively associated with the number of disturbance types and the number of structures, but in the Mat-Su subregion, Pacific loon counts were positively associated with disturbance frequency. Because of the small sample size of disturbance related information provided by volunteers, our findings should be interpreted with caution.

Models of adult Pacific loon abundance with common loon presence as a predictor were strongly supported (Table 3; Table S3, Supplemental Material). In all cases, Pacific loon abundance was negatively associated with the presence of common loons. Even after adjusting for lake size [e.g., common loon + ln(lake area); Table 3], where the two species have different preferences, common loon presence was a reliable negative predictor of Pacific loon abundance.

The persistence of Pacific loons was also negatively affected by the presence of common loons in each subregion (Table 4). For example, in all subregions combined, Pacific loon persistence probability was 0.88 when there were no common loons on the lake but only 0.50 when common loons were present (Table 4). Although this pattern supports the notion of competition, common loons did not always displace Pacific loons on all lakes (i.e., Pacific loon persistence probability ≠ 0 with common loons present). Some lakes, in fact, had both species present in some years. We did, however, note data patterns consistent with competitive displacement. For example, at Psalm Lake in the Anchorage subregion, Pacific loons were present every year from 1985 to 2007 (n = 23) when common loons were not present. However, after common loons were first recorded on Psalm Lake in 2009, Pacific loons were not recorded on the lake thereafter.

Of the 10,000 citizen science surveys completed between 1985 and 2015, less than 2% included brood observations past our predefined survival threshold of 16 September for common loons and 26 August for Pacific loons (Table S1, Supplemental Material). We assumed perfect detection of all chicks on observed lakes, because there were no instances of missed detections (e.g., the number of chicks observed per lake did not increase within a year). Chick survival estimates by subregion ranged from 0.64 to 0.87 for common loons and from 0.38 to 0.95 for Pacific loons, but precision was poor due to small sample sizes (Table 5). For a single subregion (Mat-Su), the model containing a year effect had a larger AICc weight than the null model, so we were able to estimate a trend for this subregion. Common loon chick survival in Mat-Su gradually declined over the 30-y study interval (Table 5; Figure 4).

Given our data and model assumptions, our findings indicate the common loon population has increased between 1985 and 2015 in many parts of southcentral Alaska, including Anchorage, JBER, Kenai, and Mat-Su. In addition, we found only weak evidence that observers dropped sites that lacked birds over time and preferentially focused on sites with birds (for origin of this concern, see Kamp et al. 2016). Consequently, we feel that common loon trend estimates are not biased, or only minimally biased, by selective observations. An increasing trend in common loon abundance is not exclusive to Alaska, because Maine, Massachusetts, Minnesota, Montana, and New York have all reported stable to increasing common loon populations since the early 2000s (Evers et al. 2013, 2015; Kovach et al. 2015; Schoch et al. 2015; Byrd et al. 2016). Byrd et al. (2016) suggested the positive trend in Maine indicated that common loons have acclimated to human disturbance and are remaining on lakes with heavy recreational use. The acclimation of loon adults to human disturbance and development may be a viable explanation for the population increase in Alaska's urban centers, as loon trends were positively correlated with disturbance (Kenai and Mat-Su) and shoreline structures (Kenai and Anchorage). However, we cannot rule out the artifact that volunteers preferentially surveyed larger lakes, which are positively associated with common loons, or tended to survey lakes with shoreline structures or recreational activity. Improved survey design and data collection are essential for making definitive conclusions related to the increasing trend, as well as the impact of environmental disturbance or shoreline development.

By comparison, our analyses suggest that the Pacific loon population in southcentral Alaska has decreased in Anchorage since 1985, while remaining stable on JBER and in Mat-Su. Unlike common loons, which require large lakes for running take-offs, the smaller bodied Pacific loons prefer small lakes and ponds (Solovyeva et al. 2017). The declining Pacific loon population in Anchorage may be attributed to the loss of wetland habitat due to human development and water diversions (U.S. Department of the Interior, Fish and Wildlife Service [USDOI] 1993), or possibly the increase of a superior competitor (see “Interspecific Competition” below). In Anchorage alone, freshwater vegetated wetlands have decreased by 59.5% between 1950 and 1990 (USDOI 1993). Similarly, over the past 50 y, wetlands within the Kenai lowlands have shrunk as the result of climate change, leading to wetland drying and successional changes (Klein et al. 2005). Although the effect of wetland drying on Pacific loons has not been studied directly in urbanized settings, it has been indicated as a potential threat to yellow-billed loons Gavia adamsii breeding in northwestern Alaska (USDOI 2014). In northern latitudes, climate change has the potential to alter loon habitat on breeding lakes through the reduction of water levels and diminished water quality, leading to reduced prey abundance (USDOI 2014).

Because of the small sample size of chick detections across subregions, we only had sufficient data to assess survival of common loon chicks in one of the five subregions. However, even with the constraint of sample size and our lack of a structured data collection method, common loon chick survival rates in southcentral Alaska appear analogous to annual chick survival rates in other parts of North America (Kenow et al. 2003). Until further information is obtained, a similar comparison of Pacific loon chick survival cannot be accomplished.

In Mat-Su, common loon chick survival during brood rearing appears to have gradually declined since 1985. The cause of this decline is not fully understood in southcentral Alaska, although New York state has experienced low chick survival with a stable adult trend and biologists attribute the pattern to either increased predation of eggs and chicks by bald eagles Haliaeetus leucocephalus or increased human pressures on breeding lakes (Schoch et al. 2015). Breeding Bird Survey data indicate that bald eagles in Alaska have increased annually (Matsuoka and Pardieck 2009), which may attribute to increased predation of eggs and chicks as well. In addition, it is understood that conventional observational studies, such as the Alaska Loon and Grebe Watch, are inadequate for identifying the specific cause(s) of loon chick mortality (Kenow et al. 2003). Thus, the collection of behavioral and biological responses to environmental factors would improve the strength of our results (Kenow et al. 2003; Merrill et al. 2005). For these analyses, we assumed perfect detection of all chicks on surveyed lakes. We had no instances where more chicks were observed later in a summer relative to lower counts earlier in the summer on the same lake (i.e., no record of chicks missed during a survey), which suggests that this assumption is reasonable.

Citizen scientist observations indicated that Pacific loon abundance is negatively associated with common loon presence. Given the notably smaller body size of Pacific loons relative to common loons, the results are consistent with the premise of interspecific competition. An analogous competitive situation between Pacific loons and larger yellow-billed loons, occurs in western Alaska (Haynes et al. 2014; Schmidt et al. 2014), where occupancy of Pacific loons decreased 10-fold in the presence of yellow-billed loons (Haynes et al. 2014). Although we adjusted for lake size, our abundance model cannot conclusively demonstrate that Pacific loons avoided or were competitively inferior to common loons, and an alternate explanation for the pattern could simply be differential habitat preferences (Schmidt et al. 2014).

However, our second analysis, which considered the persistence of Pacific loons on lakes observed over two consecutive years, provides stronger evidence that common loons have a competitive advantage over Pacific loons. All lakes included in the second analysis were initially occupied by Pacific loons. Only when common loons were present on a lake was there a marked reduction in probability that Pacific loons persisted on the lake in the second of two consecutive years. In a study of yellow-billed loons in northwestern Alaska, Earnst et al. (2006) suggests that inter- and intraspecific competition may be correlated to body size and dominant behavior, among other mechanisms. Common loons in southcentral Alaska may therefore be competitively superior and sometimes outcompete Pacific loons due to their larger body size and assertive behavior.

Citizen science programs are a cost-effective means of collecting data over large spatial and temporal extents and therefore may provide an early warning system for declining species (Kamp et al. 2016; Callaghan et al. 2017; Walker and Taylor 2017; Horns et al. 2018). Citizen science programs also encourage volunteers to participate in ecological studies that provide them a sense of ownership and a desire to preserve wildlife and habitat for future generations (Roelfsema et al. 2016). Unfortunately, without a probability-based sampling design and a structured data collection protocol, citizen science datasets can also have significant deficiencies. Drawbacks include a reliance on incomplete data, haphazard selection of sites, and no control over survey effort (e.g., Dickenson et al. 2010; Hochachka et al. 2012; Isaac and Pocock 2015; Kamp et al. 2016). Our unstructured loon dataset required us to attempt to create analyses that controlled for bias. Analyses required very strong assumptions and, when possible, follow-up analyses to assess the validity of both assumptions and interpretations (see “Methods”). Without such precautions, key relationships between variables, such as abundance estimates through time (trends), can be misleading and may even suggest patterns that are opposite to reality (e.g., Kamp et al. 2016). Our experience underscores the critical nature of understanding inherent flaws and limited interpretations that emerge from an analysis of unstructured data.

We strongly advocate researchers develop probability-based sampling designs and structured data collection protocols that work for a citizen science platform in advance of any data collection, to ensure higher information content and the potential for unbiased conclusions from citizen science data. This approach is critical for controlling the selection of study sites and maintaining equitable sampling effort within and between years. Sampling design and data protocols must be simple enough that volunteers can complete tasks asked of them but detailed enough that data can be clearly interpreted during the analysis process even decades into the future. In conclusion, unstructured citizen science programs present the allure of inexpensive, easily obtainable data; however, at the regional scale of our study, more information could have been gained by focusing our efforts on a structured approach, which avoids biased sampling and data deficiencies that result in weak inference.

Please note: The Journal of Fish and Wildlife Management is not responsible for the content or functionality of any supplemental material. Queries should be directed to the corresponding author for the article.

Table S1. Citizen scientists collected observational data of common loon Gavia immer chick survival across five subregions of southcentral Alaska: Anchorage, Joint Base Elmendorf-Richardson, Kenai Peninsula, Matanuska-Susitna Valley, and Outlying, 1985–2015. Data are summarized by subregion, location or town within the subregion, name of lake observed, year of observation, number of nesting pairs, number of chicks hatched per breeding pair, and number of chicks that survived past the defined survival threshold (16 September for common loon, 26 August for Pacific loon).

Found at DOI: https://doi.org/10.3996/082018-JFWM-078.S1 (46 KB XLSX).

Table S2. Citizen scientists collected data on anthropogenic and natural disturbances for lakes observed in the five subregions of southcentral Alaska: Anchorage, Joint Base Elmendorf-Richardson, Kenai Peninsula, Matanuska-Susitna Valley, and Outlying, 1985–2015. Anthropogenic disturbances include swimming, fishing, nonmotor craft, motor boats, jet skis, floatplanes, and dogs. Natural disturbances include eagles, mammals, and flooding. Disturbance is summarized by frequency of occurrence: daily, weekly, monthly, occasionally, or never occurred.

Found at DOI: https://doi.org/10.3996/082018-JFWM-078.S2 (386 KB XLSX).

Table S3. Presence or absence of common loon Gavia immer and Pacific loon Gavia pacifica on lake observed by citizen scientists in five subregions of southcentral Alaska; Anchorage, Joint Base Elmendorf-Richardson, Kenai Peninsula, Matanuska-Susitna Valley, and Outlying, 1985–2015. Adults and chicks are ranked as (1) if present or (0) if absent from an observed lake.

Found at DOI: https://doi.org/10.3996/082018-JFWM-078.S3 (253 KB XLSX).

Table S4. Number of structures located within a 150-m buffer of lakes observed by citizen scientists in five subregions of southcentral Alaska: Anchorage, Joint Base Elmendorf-Richardson, Kenai Peninsula, Matanuska-Susitna Valley, and Outlying, 1985–2015. Only lakes with a least three observations over a 5-y period and only years where Google Earth imagery was available were included in the analysis.

Found at DOI: https://doi.org/10.3996/082018-JFWM-078.S4 (55 KB XLSX).

Reference S1. Bergman RD, Derksen DV. 1977. Observations on arctic and red-throated loons at Storkersen Point, Alaska. Arctic 30:41–51.

Found at DOI: https://doi.org/10.3996/082018-JFWM-078.S5 (880 KB PDF); also available at http://pubs.aina.ucalgary.ca/arctic/Arctic30-1-41.pdf.

Reference S2. Byrd A, Evers DC, Taylor K, Gray C, Chickering M. 2016. Restore the call: Maine status report for the common loon. Biodiversity Research Institute, Portland, Maine. Science Communications Series BRI 2016-12.

Found at DOI: https://doi.org/10.3996/082018-JFWM-078.S6 (9.06 MB PDF); also available at http://www.briloon.org/uploads/BRI_Documents/Loon_Center/RCF/DRAFT%20Loon%20Status%20Report%20ME.pdf.

Reference S3. Evers DC. 2004. Status assessment and conservation plan for the common loon Gavia immer in North America. U.S. Department of Interior, Fish and Wildlife Service Biological Technical Publication, Washington, D.C.

Found at DOI: https://doi.org/10.3996/082018-JFWM-078.S7 (1 MB PDF); also available at https://wdfw.wa.gov/conservation/loons/common_loon_status_assessment.pdf.

Reference S4. Evers DC, Attix L, Osborne Gray C, Spagnuolo V. 2013. Restore the call: Massachusetts status report for the common loon. Biodiversity Research Institute, Gorham, Maine. Science Communications Series BRI 2013-22.

Found at DOI: https://doi.org/10.3996/082018-JFWM-078.S8 (4.69 MB PDF); also available at http://www.briloon.org/uploads/BRI_Documents/Loon_Center/RCF/6316%20MA%20Status%20Report.pdf.

Reference S5. Evers DC, Hammond C, Anderson C, Taylor K. 2015. Restore the call: Montana status report for the common loon. Biodiversity Research Institute, Gorham, Maine. Science Communications Series BRI 2015-15.

Found at DOI: https://doi.org/10.3996/082018-JFWM-078.S9 (5.03 MB PDF); also available at http://www.briloon.org/uploads/BRI_Documents/Loon_Center/RCF/FINAL%20MT%20Status%20Report%20050615.pdf.

Reference S6. Kelly ML. 1992. The effects of human disturbance on common loon productivity in northwestern Montana. Unpublished thesis. Bozeman: Montana State University.

Found at DOI: https://doi.org/10.3996/082018-JFWM-078.S10 (6.35 MB PDF); also available at https://scholarworks.montana.edu/xmlui/bitstream/handle/1/7174/31762101921714.pdf?sequence=1&isAllowed=y.

Reference S7. Kovach K, Paruk JD, Evers DC. 2015. Restore the call: Minnesota status report for the common loon. Biodiversity Research Institute, Portland, Maine. Science Communications Series BRI 2016-14.

Found at DOI: https://doi.org/10.3996/082018-JFWM-078.S11 (10.65 MB PDF); also available at http://www.briloon.org/uploads/BRI_Documents/Loon_Center/RCF/Minn%20Loon%20Report%2051316.pdf.

Reference S8. Matsuoka S, Paredieck K. 2009. North American Breeding Bird Survey, Alaska. U.S. Geological Survey Patuxent Wildlife Research Center, Laurel, Maryland.

Found at DOI: https://doi.org/10.3996/082018-JFWM-078.S12 (221 KB PDF); also available at https://prd-wret.s3uswest2.amazonaws.com/assets/palladium/production/s3fs-public/atoms/files/BPIF_annual_project_summaries_2009.pdf.

Reference S9. Schoch N, Evers DC, Sauer AK. 2015. Restore the call: New York status report for the common loon. Biodiversity Research Institute, Portland, Maine. Science Communications Series BRI 2016-13.

Found at DOI: https://doi.org/10.3996/082018-JFWM-078.S13 (8.59 MB PDF); also available at http://www.briloon.org/uploads/BRI_Documents/Loon_Center/RCF/51316%20NY%20loon%20report.pdf.

Reference S10. [USDOI] U.S. Department of the Interior, Fish and Wildlife Service. 2014. Species status assessment report yellow-billed loon. Listing Review Team, Fairbanks, Alaska.

Found at DOI: https://doi.org/10.3996/082018-JFWM-078.S14 (869 KB PDF); also available at https://www.fws.gov/alaska/fisheries/endangered/pdf/yellowbilled_loon_ssa.pdf.

We thank the hundreds of citizen scientist volunteers who submitted observational data over the duration of the Alaska Loon (and Grebe) Watch program. We also thank N. Tankersley, D. Tessler, T. Gotthardt, and T. Zeller who worked endlessly to promote the Loon Watch program and ensure its continued success. Sponsors of the Alaska Loon and Grebe Watch Program include ADF&G, U.S. Fish & Wildlife Service, the Alaska Zoo, U.S. Forest Service, the Alaska Natural Heritage Program, and the Department of Defense (JBER). Earlier versions of this manuscript benefitted from reviews by J. Fair, an anonymous reviewer, an anonymous Associate Editor and colleagues K. Christy and T. Gotthardt. Funding for this analysis was provided by a State Wildlife grant to ADF&G.

In memory of David Tessler March 1967–June 2017. Dave dedicated his life to promoting wildlife conservation, but always put his family first.