The wetlands were mapped with aerial photography. The resulting 10-by-10-inch black-and-white contact prints were ground proofed to delineate basin perimeter, vegetation, and shoreline, and then classified with the system of Cowardin and others (1979). Figure 2 is an example of a field map generated using this method. It depicts Downing Bog, a wetland in the CHER site that was monitored during Study 2 and an additional 7 years after the study’s completion. The system and class of the wetlands were designated according to the dominant (30-percent areal coverage or greater) form of a vegetative life. The maps of all monitored wetlands were updated annually except for those in Aroostook County, where the wetlands monitored were mapped only in 1994. The wetlands were grouped into nine classes based on predominant vegetation type:

For surface water samples collected in July, specific conductance (mean μS/cm), an overall measure of nutrient content, and pH were reported for all wetlands except as noted in the table 2. These data were reported in 1979 for DIXM, 1982 for BEDD and CHER, 1984 for MOBU and MOEU, and 1994 for AAGR and AFOR. Samples from the midpoint of each wetland’s outlet were collected at 20–35 centimeters below the surface. In all sites except those in Aroostook County, specific conductance was measured with a Markson Science Inc. (Del Mar, CA), Model-10 meter; and pH, with a Fisher Scientific Company (Pittsburgh, PA), Model 640 portable, calibrated meter. In Aroostook County, samples were collected in containers rinsed in deionized water and transported on ice to the Environmental Chemistry Laboratory, Department of Geological Sciences, University of Maine–Orono for analysis. Detailed analytical procedures for water chemistry analyses for CHER and BEDD are in Longcore and others (2006), and those for AAGR and AFOR are in Longcore and others (1998).

Each brood-rearing wetland was visited at least eight times per 7–10-day period to ensure black duck broods could be followed from hatching (Class Ia) to fledging (Class IIc-III; 34–44 days; Gollop and Marshall, 1954). Brood use, including all Anatidae species, was determined by observing wetlands for 2 hours each visit from May to early September. Observations began 0.5 hour before sunrise or ended 0.5 hour after sunset from elevated platforms (Longcore and Ringelman, 1980). The species of a brood was identified by observing the accompanying female along with the physical characteristics and plumage of ducklings; color images of downy young in Kortright (1942) were useful to confirm species. The number and age class of ducklings and their species was used to identify individual broods. Many wetlands were visited a total of 10–11 times to document the fledging date of a brood (table 3). For most wetlands, a single observer was adequate for a count; however, as many as 6 or 7 observers were needed to cover large wetlands such as the highly productive Downing Bog (fig. 2).

The onset and length of breeding periods can vary among species and years depending on environmental variables, especially weather that affects breeding ground habitats such as ice-out. Breeding pairs of black ducks usually arrive on Maine wetlands in early April and commence egg laying by mid-April; nests started later than May 10 are considered renests (Coulter and Miller, 1968). The timing of the initial visit to each wetland differed among study sites and years but were reasonably consistent overall (table 4). Visits starting in late May at the DIXM site reflected the team’s unfamiliarity with the habitat, especially in the first year, 1977. More time was spent looking at wetlands without broods because not all the wetlands were evaluated for their suitability for broods. Also, fewer than 10 elevated observation platforms were erected in 1977 (Longcore and Ringelman, 1980). The initial date of wetland visits in 1993 was identical for AAGR and AFOR sites but was about one week later in 1994 for the AFOR site related to logistics of locating suitable, scattered wetlands in the more remote, forested habitats.

The ages of ducklings were estimated by evaluating their shape and size, pattern of down replacement, and degree of feather emergence (Kortright, 1942; Gollop and Marshall, 1954), characteristics that Evrard (1996) evaluated and found reliable. The generalized duckling plumage classes and sub-classes of Gollop and Marshall (1954) are as follows: Ia, a bright ball of fluff; Ib, a fading ball of fluff; lc, gawky-downy; IIa, first feathers; llb, mostly feathered; llc, last down; and III, feathered-flightless. Refer to Gollop and Marshall (1954) for additional descriptors. Images of developing Branta canadensis (Linnaeus, 1758) (Canada goose) goslings by Yocom and Harris (1965) depicted plumage classes, following the same age classes designated by Gollop and Marshall (1954), which were used to determine relative age. Grice and Rogers (1965) in Massachusetts published descriptions of wood duck duckling plumages in different age classes that were used to confirm age classes of ducklings we observed.

Flow charts were constructed for each brood on each wetland for each count to track the development of individual broods. After counts, the species, age, and number of ducklings in each brood were compared, noting the time of observation and the direction of brood movements to avoid double counting broods. The plumage development of a previously seen brood was then projected forward to the next observation period to seek a match for age and number of young. This was done with a two-part “slide rule.” The first part used a linear calendar that covered the period from the date the earliest brood hatched through the date the latest brood attained flight (Gollop and Marshall, 1954). The second part was a series of sliding strips, one for each species, that depicted the width (in days) of the seven age classes (Gollop and Marshall, 1954, app. 1). If no obvious match was found, a brood of the expected age, but with fewer ducklings, was sought. Broods were given a ±4-day leeway to limit the possibility of misclassifying a brood as “not seen previously” when it had been seen before. If initial aging of a brood was uncertain, an additional leeway of ±1 age class was allowed. This methodology may slightly underestimate the total number of broods, especially on wetlands with large numbers of broods of the same species, but ensures reliable, conservative estimates of brood size.

The size of a wetland, its amount and distribution of vegetative classes, and its water chemistry, especially acidity and specific conductance, which reflects nutrient content, affect which wetlands are used by brood-rearing Anatidae species.

The mean number of visits to wetlands was variable among studies, partly because numbers of wetlands were different among sites and years (table 2). The lowest mean number of visits was during the initial field work (1977, DIXM) when field crews were unfamiliar with the wetland habitats. The number of wetlands increased during the 4-year study at the DIXM site because of beaver activity (Ringelman, 1980).

Eleven duck species and one goose species were observed in the study sites (table 5). Black duck, ring-necked duck, and Lophodytes cucullatus (Linnaeus, 1758) (hooded merganser) raised broods on wetlands monitored in all four studies. Aix sponsa (Linnaeus, 1758) (wood duck) used all study sites but MOEU, and Anas crecca (Linnaeus, 1758) (green-winged teal) used all but BEDD. Mareca americana (J.F. Gmelin, 1789) (American wigeon) and Spatula clypeata (Linnaeus, 1758) (northern shoveler) used only AAGR. Only one or two mallard broods used the DIXM, BEDD, and CHER sites, whereas many were concentrated on the AAGR site. A few Spatula discors (Linnaeus, 1766) (blue-winged teal) broods used the CHER and MOBU sites, but most were counted in Study 4. A single Mergus merganser (Linnaeus, 1758) (common merganser) brood was observed in each of the DIXM, MOBU, and AFOR sites; four were counted on the CHER site. Bucephala clangula (Linnaeus, 1758) (common goldeneye) was observed only at BEDD and in Study 4. Canada geese raised broods only at the MOBU site and at both sites in Study 4 (AAGR and AFOR). Of the 1,907 broods observed on all sites, the black duck made up 35.4 percent, whereas wood duck made up 13.9 percent; ring-necked duck, 12.9 percent; hooded merganser, 8.5 percent; mallard, 7.3 percent; green-winged teal, 6.1 percent; and blue-winged teal, 3.7 percent (table 5). Each of the other four duck species made up less than 1.0 percent, or fewer than 20 broods. The 159 Canada goose broods made up 8.3 percent of the total broods.

Of the black duck and mallard broods monitored during the four studies, 69.2 percent and 66.9 percent, respectively, were followed to fledging. These are the highest percentages among Anatidae species monitored because black duck and mallard were the focus of these studies (table 6). The highly visible Canada goose, with 56.6 percent of broods followed to fledging, was next highest. Final counts were more difficult to obtain for ducklings of other species, which tend to scatter (wood duck; McGilvrey, 1969) or dive (hooded merganser, common goldeneye). For these three species, the percent of broods with final counts ranged between 47.8 percent (hooded merganser) and 35.7 percent (common goldeneye). Because the ring-necked duck initiates nesting later, only 21.5 percent of its broods reached Class IIc-III before each field season ended (table 6).

Detecting all broods of Anatidae species in boreal forest habitats of the Northeast with aerial or ground surveys is challenging. Some species are more readily detected (mallard, black duck, green-winged teal, ring-necked duck, and common goldeneye) than others. The wood duck and hooded merganser are poorly detected from fixed-wing aircraft. The three merganser species (hooded, common, and Mergus serrator [Linnaeus, 1758] [red-breasted]) are also not readily identified and are therefore lumped as “mergansers” during aerial surveys (Mark Koneff, Chief of the Branch of Migratory Bird Surveys, US Fish and Wildlife Service, oral commun., February 4, 2021).

Technicians participating in the ground surveys at the DIXM site in 1977 and the single-visit counts at Downing Bog in 1985–91 gained more understanding of the study area the longer the study area was observed. Pagano and Arnold (2009) found that novice observers on the ground during brood studies in prairie pothole habitats of North Dakota averaged 0.11 lower detection probabilities than more experienced observers.

Pagano and Arnold (2009) also reported that detection probabilities varied among duck species, time of observations (morning versus evening), wetland characteristics, and most critical, the duration of the survey (1 minute versus 5 minutes). Sauer (2002) cautions that survey methods must address two fundamental issues—an “accommodation of detectability” and identification of “sampling frames”—to ensure that all parts of the target population have a chance to be sampled. The 2-hour quiet observation survey either started 0.5 hour before sunrise or ended 0.5 hour after sunset, which ensured early and late-feeding broods could be detected. Whenever possible, intruding on a wetland was avoided. Multiple observers were deployed for counts to ensure wetland coverage. To visit wetlands during the entire brood period (early May to late September), 8 visits were made to each wetland every 7–10 days per study site (table 3). Counts were purposefully initiated in early May to observe Class I-aged broods; thus, age and size of broods did not affect counts as noted by others (Ringelman and Flake, 1980; Giudice, 2001).

Disturbance on or near a wetland causes females with ducklings to become secretive and sneak into the vegetation or even leave the wetland. Gabor and others (2000) found that sounds of an approaching helicopter will cause female dabbling ducks with broods to become motionless, leave a wetland, or swim under cover; hooded mergansers move to open water and dive. They recommended that quiet observation from elevated platforms be used to acquire reliable dabbling duck brood densities, duckling age, and number of ducklings. In 1995, some of the field crew in Nova Scotia verified their observational abilities by comparing counts made by a helicopter crew to that of a crew on elevated platforms (table 7). The helicopter crew missed 40–69 percent of the broods that were recorded by the crew on elevated platforms on five wetlands.

Waterfowl brood data from these four studies are reliable within a study site but attempting to determine a long-term trend for the 18-year period of Anatidae brood counts in Maine should be done with care. The occurrence and distribution of many wetlands are related to the activities of American beaver, so brood data are confounded over years, locations, and the changing conditions within study site wetlands. It is germane, however, to revisit any one of these study sites to evaluate changes over time. For example, the DIXM site would be ideal for determining the effect of land use, especially wetland dynamics, because of two previous studies by Spencer (1963) and Ringelman (1980) that published historical information for broods and wetlands for 1958–60 and 1977–80, respectively. For Study 2, the CHER site brood production was extraordinary (annually up to 1 brood per 2 ha of habitat) on Downing Bog. After the initial 3 years of the study, a single visit to count broods was made for an additional 7 years. Brood numbers declined during these 7 years; a future study to understand why may inform monitoring frequency, wetland changes, and the effect of a Haliaeetus leucocephalus (Linnaeus, 1766) (bald eagle) taking up residency. Data from Study 3 (MOBU and MOEU) is a robust baseline for determining the effects of changing impoundment management objectives, which could cause population changes in species (waterfowl, fish, amphibians). As for Study 4 (AAGR), environmental regulations applied to modify eutrophication in wastewater settling ponds (Christina Reservoir, Josephine Lake, nearby lagoons) may have already diminished waterfowl brood use since the study ended in 1994.

Each of the four studies contained 1–2 wetlands that accounted for a large proportion of total females with broods counted (table 8). For DIXM, Chase Bog (82 ha, class PFW) and Gilmore Meadows (5.7 ha, class PEW) accounted for 53.5 percent of the 185 broods counted during Study 1. For the CHER and BEDD sites, 56.0 percent of the 241 broods counted between 1982 and 1984 were on the Downing Bog (117 ha, class PSS). For the MOBU and MOEU sites, the Magurrewock complex (144 ha, class PEW) and Barn Meadow (64 ha, Class PEW) accounted for 37.2 percent of the 411 broods counted during Study 3. For the AAGR and AFOR sites, the Christina Reservoir (113 ha, class PUB) and Josephine Lake (45 ha, class PUB) wetlands accounted for 36.2 percent of the 849 broods counted during Study 4. These two wetlands were exceptionally rich in nutrients from the potato processing plant that maintained these settling basins (Longcore and others, 1998).

For some individual species, however, broods were not concentrated only on one wetland. For example, although 69.7 percent of black duck broods in Study 2 were on Downing Bog (CHER site), an additional 11.8 percent (9 broods) were on Snake Flowage (4.9 ha, class PFW). Longcore and McAuley (2004) counted four usual-sized broods (8–10 ducklings) and two large broods (one of 18–20 Class Ia-b ducklings and one of 20 Class Ib ducklings) in 1982. The size of the large broods likely resulted from a post-hatch amalgamation for two black duck broods. Thus, six black duck broods used Snake Flowage in 1982 along with three hooded merganser, two wood duck, and one green-winged teal brood. This wetland was a beaver flowage that had been dewatered several years before Study 2 began and then reflooded by beaver to create an excellent PFW habitat.

Introductions of the Canada goose established breeding populations in Maine. Palmer (1949) remarks that the Canada goose nested in Maine during the colonial period and cites several references from the 1800s into the mid-1900s. Although the Canada goose was recorded in Maine as early as 1605 (Rosier, 1887), the first authentic records of breeding in the state (nest of five eggs) was in 1939 in Penobscot County (Mendall, 1945). Then, in 1944, two broods were seen with adults in northern Piscataquis County near the Maine-Quebec border. These breeders were thought to be wild-strain birds, but the source could have been cripples from hunting, or escaped or feral birds.

The breeding geese at MOBU probably began from pinioned (18) and wing-clipped (16) birds released at Magurrewock marsh in 1952 (Moosehorn National Wildlife Refuge, 1952). In 1959, two Canada goose broods were documented at the refuge (Moosehorn National Wildlife Refuge, 1959), which constituted the first known goose nesting in eastern Maine since the 1940s. By 1965, the breeding population was well established; there were as many as 15 broods, based on young produced, and the pairs dispersed 20 miles from the refuge (Moosehorn National Wildlife Refuge, 1966). By 1990, the breeding population was stable with a minimum of 33 broods accounting for over 100 young annually (Moosehorn National Wildlife Refuge, 1991).

The Maine Department of Inland Fisheries and Wildlife initiated a Canada goose transplant project in 1965 by releasing 56–78 geese at each of three Wildlife Management Areas and 14 geese at MOEU (Spencer and Corr, 1976). Goose releases continued annually for 10 more years with a total of 2,341 geese from New York, New Jersey, and Connecticut being released mostly throughout southern Maine (Weik, 2005). During 1981–85, 1,723 more geese were moved from Connecticut into mostly northern Maine (Weik, 2005); some were released into Aroostook County townships where goose broods were counted in 1993–94. Canada geese are considered to be distributed statewide and broods are now associated with almost all release sites or adjoining areas (Vickery and others, 2020).

During Study 3, 97 Canada goose broods were recorded at MOBU, and during Study 4, 62 at AAGR and AFOR. The mean (± standard error [SE]) size of broods that reached Class IIc-III was 3.15±0.21 in MOBU (62 broods) and 3.79±0.30 in AAGR (28 broods), which may reflect the higher conductance of wetlands in the AAGR site that affected gosling survival. Mean (± SE) brood size with sites combined was 3.34±0.17 (90 broods). This value is lower than what Schummer and others (2011) reported (3.89±0.36; 28 broods) for the same years (1977–94; table 9). These data provide a reference point for comparison to future data that may describe distributional changes of Canada geese throughout Maine.

Nearly 69 percent of black duck broods and 66 percent of mallard broods reached fledging compared to 52 percent of blue-winged teal, 50 percent of green-winged teal, 47 percent of hooded mergansers, and 36 percent of wood ducks (table 6). Because black ducks and mallards were the major focus of the four studies discussed in this report, more effort was expended to obtain a final count of older broods, including returning for extra visits to wetlands with these species and timing the consecutive daily observations to coincide with the predicted date the brood reached Class IIc. When the observation period ended, some of the ring-necked duck broods had not reached Class IIc or III because the accompanying female had initiated nesting later, thus only 21 percent of the species’ broods were monitored to fledging. Difficulties with re-sighting broods on forested wetlands because tree canopies impair visibility, and the behavior of wood duck ducklings to scatter (McGilvrey, 1969) and hooded merganser and common goldeneye to dive (Bellrose, 1976) may have contributed to the lower percentages observed for these species.

During 1977–94 mean (±SE) brood sizes for black duck monitored to age Class IIc-III was 4.36±0.11 (468 broods); that of mallard, 4.61±0.30 (93 broods); that of wood duck, 3.67±0.18 (97 broods); that of hooded merganser 3.10±0.20 (78 broods); that of ring-necked duck, 4.34±0.29 (53 broods); that of green-winged teal, 4.78±0.28 (59 broods); and that of blue-winged teal, 4.81±0.33 (37 broods; table 9). In previous years and locations, average brood sizes were larger. Mendall (1958) reported a mean of 5.2 Class III ducklings per brood for 141 ring-necked duck broods during 1939–54 in Maine. Survival of 135 web-tagged wood duck broods in Massachusetts during 1952–53 averaged 5.8 ducklings per brood (Grice and Rogers, 1965). In Maryland during 1964–67, Class IIb aged wood duck broods averaged 5.2 for 223 broods (McGilvrey, 1969).

Schummer and others (2011) calculated the mean (±SE) Class III brood size for black duck (4.55±0.10), mallard (3.96±0.30), hooded merganser (4.10±0.23), common goldeneye (4.14±0.19), wood duck (3.63±0.15) and ring-necked duck (4.46±0.31) based on 51 years of counts in Maine during 1955–2007. During this period, 6–57 broods per year (23 on average) were counted, totaling 1,169 broods by 2007. No density values were presented for locally delineated wetland types (McCall, 1972) and no explanatory variable, including wetland type, received much support in their models. Survey year explained the most variability; however, only two counts were done per year per wetland, either by canoe or stationary ground blinds, and numbers of wetlands surveyed ranged from 20 to 63 per year. The actual effort Schummer and others (2011) expended per wetland per year, because of their acknowledged “cursory nature of our brood surveys,” likely affected results, but to what extent is unknown.

When the 1977–94 brood sizes reported by Schummer and others (2011) and Longcore and Bunck (2024) are compared (table 9), the mean (±SE) brood sizes are identical for black duck, are nearly identical for wood duck and ring-necked duck, but are 0.6 ducklings higher for mallard than for the Schummer and others (2011) data. The common goldeneye brood size (3.4±1.12) based on 5 broods averaged substantially lower than that of Schummer and others (2011; 4.36±0.28; 95 broods detected on 17 wetlands). The hooded merganser brood size from the Longcore and Bunck (2024) data was 0.9 ducklings fewer, but number of single-duckling Class III broods was about the same, 15 versus 13. The two sources of data were from different wetlands except for Arnold Brook Lake in Aroostook County, which Schummer and others (2011) monitored sporadically from 1985 to 1989.

The Canada goose mean (±SE) brood size with sites combined was 3.34±0.17 (90 broods). This value is lower than what Schummer and others (2011) reported (3.89±0.36; 28 broods) for the same years (1977–94; table 9).

It is prudent to remember that the average size of broods at fledging is subject to certain biases (Cowardin and Blohm, 1992); total brood loss and larger, early-hatched broods from larger clutches may affect ultimate size. The surveys of this report extended from May 1 to September 2 through the breeding season, which probably mitigated some effect except for loss of a total brood.

After the initial 3 years of brood evaluations, Downing Bog was monitored for broods (table 10) with one evening visit annually for an additional 7 years as part of Study 2, which evaluated effects of acid rain on conductance and pH on invertebrate diversity and numbers (Longcore and others, 2006). Downing Bog was of particular interest because of its low conductance (18 μS/cm) and pH (5.1). The wetland supported substantial numbers of black duck broods, counts of which provided an additional sample of final brood count (76) that was used to calculate mean size of Class IIc-III broods.

This single-visit effort revealed the benefit of multi-visit counts. The total number of Class IIc-III broods observed during the multi-visit counts in 1982–84 (90 of 135 total observed broods; 66.6 percent) was 22.7 greater than the total observed during single-visit counts in 1985–91 (97 of 221; 43.9 percent). Furthermore, a single visit may reflect unique annual habitat conditions from beaver activity or sparse rainfall and an aerial predator (like bald eagle) taking up residence near the wetland.