Interpreting the Songbirds of Great Duck Island, Maine

Songbirds use islands for breeding and migrating. Great Duck Island is located 10 miles from Mount Desert Island, Maine; it is about 200 acres large and consists of a variety of habitat types. A point count survey was conducted in 2017 to better understand the island’s breeding songbirds. During the summer of 2022 I repeated this study. Monitoring the island can reveal how songbirds might change their use of the island and what this might indicate about population trends or the suitability of the island and similar islands for breeding, especially in the face of climate-derived changes to island landscapes.

Every morning, from June 9 to August 21, I conducted a point count survey. I walked a roughly 1.5 mile transect that contained 12 different points. Points were selected based on congruence with the 2017 study and expected productivity. At each of these points I would stop for three minutes and record every songbird (order Passeriformes) species I saw or heard. Every bird was also attributed to a quadrant based on cardinal direction (Figure 1.). The survey began within a 30 minute window before sunrise and took an average of 1.5 hours to complete. At the start of the survey, environmental conditions were recorded, including temperature, cloud cover, wind speed, and wind direction. Points were not surveyed under low-visibility conditions, e.g. fog and rain. 

A songbird was deemed a  “probable breeder” if it was recorded within the same quadrant seven or more days apart. Additionally, if definitive evidence of breeding was observed (i.e. fledgling birds, nest sites, adults feeding young, etc.), it would also be considered a “probable breeder” (Schuver 2017).

No effort was made to estimate the distance of detected birds from the survey point. To preserve independence, care was taken to not double count birds. This was not difficult to decipher save for a few points that were close in proximity. See discussion for more details.

Using a trimble, I mapped the most prominent distinctions between habitat types within a 50 meter radius of each survey point. I then assigned each quadrant one or two habitat types based on the vegetation composition I mapped. Habitat types include forest, meadow, blowdown, shrub, marsh, and berm (Schuver 2017).

The survey resulted in 2798 observations and 33 songbird species. The most common species were song sparrow (745 observations), savannah sparrow (450 observations), and common yellowthroat (433 observations). A total of 17 species were seen more than 7 days apart, satisfying the requirement for “probable breeder.” An additional three species were also deemed probable breeders based on qualitative evidence of breeding (see discussion). The breeding list includes one species of special concern in the state of Maine: nelson’s sparrow. 

The quadrants with the most observations were 10D (173 - Marsh), 12B (126 - Meadow), and 11C (101 - Shrub). The quadrants with the most amount of species observed were 10D (17 - Marsh), 11A (16 - Forest), and 6D (14 - Forest).

Use the slide tool to interact with the map to see how different points had different numbers of observations and species of songbirds

More species and more probable breeders were identified in 2022 than in 2017, though this is likely a product of sampling effort. The three most abundant species were the same in 2017 and 2022, and both studies identified nelson’s sparrow as a probable breeder. There were no species on the 2017 breeding list that were excluded in the 2022 list; there was the addition of gray catbird, northern cardinal, and american redstart in 2022.

The survey did not identify Northern Cardinal, Gray Catbird, or Red Crossbill as probable breeders. However, I have considered them probable breeders due to qualitative observations. Fledglings of both northern cardinal and gray catbird were observed, and red crossbills were observed nearly every day of the season; the only thing preventing them from meeting the requirements of the survey was the fact that the overwhelming majority of them were flyovers, in which case I did not attribute them to a specific quadrant.

Based on the frequency to which species were observed in different habitats, it appears that blowdowns and meadow support the greatest quantity of birds while the forest supports the greatest diversity. A chi-squared revealed that a songbird's use of forest vs non-forest differs significantly across families. That means that habitat plays a decisive role in what species are present or absent. Coastal islands in the Gulf of Maine are experiencing drastic changes to their vegetative composition, red and white spruce facing astonishing decline. The spruce forest on GDI is becoming thinner every year, a fact only exacerbated by the large population of snowshoe hare that was introduced to the island and is now preventing seedling spruce trees from growing. Blowdowns and meadow are becoming increasingly standardized on the island. This transition away from forest may benefit some species, like song sparrow and common yellowthroat, but will eradicate the breeding presence of a great deal of species. This may not appear to be anything of note in the context of a small island like GDI, but when you consider the thousands of other islands along the coast of Maine that support the same types of changing landscapes, it quickly becomes an alarming issue.

Song sparrow was the most ubiquitous both in what habitat it was found in and the total number of observations. Savannah sparrows were a lot more habitat dependent, being found most richly in the lush meadow habitats of points 1 and 12. At the end of July I started noticing impressively large groups of fledgling birds, some consisting of upwards of 20 birds. Common yellowthroats were also able to thrive in the blowdowns where savannah sparrows could not.

I extended the survey into August with the hopes of documenting fall migrants. This was only partly accomplished. I overestimated the number of migrant species that would be easily detectable. Migrant species were most often observed in the woods on the northern part of the island, which were a mix of softwoods and hardwoods. None of my survey points were located in this part of the island, so I undoubtedly missed a lot of migrants. On some days I had time to walk through these woods after I had completed the survey, but was not nearly as thorough as I would have liked to be. Though sometimes difficult to track, there was discernable migrant activity that would be worth monitoring in the future.

The most rare species identified during the survey was a red-headed woodpecker, but is excluded from the study because it belongs to the order Picidae. Despite the abundance of blowdowns and seemingly excellent habitat for them, woodpeckers have remained strangely absent from the island. Red-headed woodpeckers are regionally uncommon, making the appearance of one on GDI all the more unusual. The bird observed was an adult with a bright iridescent plumage including an entirely red head and blueish-black wings that had a clean vertical white patch on the edge. The apparent absence of woodpeckers may after all be in part a matter of little effort to locate them; by the end of the season I had seen several downy woodpeckers and a northern flicker. Despite this, it seems unusual that they would not be more conspicuous given the amount of suitable habitat that appears to be available. 

There were also several rare passerines identified over the course of the season that were not associated with the survey. These include prairie warbler, lark sparrow and mourning warbler.

Nelson's sparrow is a species of special concern in the state of Maine. Their breeding habitat is restricted to coastal salt marshes. GDI has a limited amount of this habitat to offer but it is enough to attract several birds every year. The species was detected infrequently during the survey, mostly due to the fact that there was only one point that aptly surveyed the marsh habitat. Incidentally I heard as many as two or three birds singing at once.

While the receding spruce forest may not directly affect nelson's sparrows, rising sea levels might. The island may very well be split in half by the ocean right where the marsh currently is, negating it of any use for breeding nelson's sparrow.

While this study was intended to replicate the one done in 2017, there are several essential differences that must be taken into account when comparing the two sets of data. The first is the discrepancy between survey points; at the time of deciding what points I would choose to survey, I was not aware that the geographic locations of the 2017 survey points were available. Future studies would benefit from replicating survey points, or using an entirely different survey method. The frequency of surveying was also markedly different; Schuver sampled each point 8 times, ending in mid July, while I sampled each point 60 times and ended in mid August. The relationship between species richness and total individuals is non-linear, meaning the difference in results could simply be explained by the increase in sampling effort ( Gotelli, J. 2017).

One of the most critical things to understand is that the study is designed to determine if birds are persisting at a particular location, using longevity as a metric for breeding. This study is not designed to address habitat preferences; the same points were sampled every day, resulting in pseudoreplication: individual birds inevitably represent multiple observations, and in fact the study is designed for this purpose. In order to make statistically sound investigations into species’ preferences in habitat type, I would encourage future researchers to employ random stratified sampling. Categorizing habitat should occur before points are established, not after, and each habitat type should be surveyed the same number of times. Choosing one random spot from each habitat type to survey every day would be a sufficient model; not only would it ensure the data was random, but it would also resemble independence; given the small size of the island it would be impossible to expect no individual bird would be observed twice, but it would greatly increase the total number of unique birds observed. Determining if a bird was a probable breeder would become not a matter of repeated observation at a specific point, but within a specific habitat.

Great Duck Island houses a diverse set of habitat types that are undergoing rapid changes. Some birds, like song sparrows and common yellowthroats, seem well adapted to the shift of more blowdowns and meadows, but more often songbirds rely on the forests that are now disappearing. It is important to look at species composition across other Maine islands and examine how vegetation is changing to understand what that means for songbirds. 

I would like to thank John Anderson for advising my time on Great Duck Island. I also owe a great deal of thanks to the College of the Atlantic W.H. Drury Jr. Research Fund and the Nature Conservancy for allowing me to spend my summer on the island and study the land​. 

The foundation of the maps presented in this story map are created  by Eleanor Gnam; she took the data I collected with a trimble and curated the base map I would modify for nearly all of the maps found here. Gordon Longsworth was essential in teaching me how to use GIS and advising the map making process.

Lastly I would like to thank the rest of the 2022 Great Duck crew for their support and good spirit. 

Gotelli, J. and Aaron M. Ellison. (2017). A Primer of Ecological Statistics, 2nd ed., Sinauer Associates, Inc., Publishers, Sunderland, Massachusetts U.S.A. Ch. 13, pp. 449–453.

Schuver, A. (2017). A Survey of Passerines of Great Duck Island, Maine. (unpublished)

Special Thanks to ESRI -  https://www.esri.com/en-us/home 

Contact Levi Sheridan - lsheridan24@coa.edu