The Effects of Supplemental Feeding on Wintering Black-capped Chickadees (poecile Atricapilla) in Central Maine: Population and Individual Responses

—In a remote area of central Maine, I established bird feeders stocked with black oil sunflower seeds to supplement the food of wintering Black-capped Chickadees (Poecile atricapilla). The chickadees discovered experimental feeders established in late October within two weeks; feeders established in mid-January were discovered more slowly. Weekly censuses showed that chickadee abundance was significantly higher in the presence of the feeders. Mark-recapture analysis revealed that as many as 110 and 70 chickadees were using the feeders at the two most intensively studied sites over 2-day periods. Ambient temperature had no influence on the rate at which banded chickadees visited the feeders. The visitation rate of banded chickadees was higher during the first third of the winter; perhaps competition for feeder access increased as increasing numbers of unbanded chickadees used the feeders as the season progressed. The frequency of feeder use varied markedly among the chickadees at each feeding site; this variation could not be explained by age of the birds (first-winter versus adult birds). Variability in feeder use was also apparent for individual birds over time. Received 24 July 2000, accepted 2 February 2001. Acquiring food is a particularly acute challenge that wintering passerines must meet in cold environments. Winter temperatures may dip beneath the lower critical temperature (Withers 1992) for many passerines, necessitating an increase in the metabolic rate at a time when daylength is short and food abundance is low. It is not surprising that the experimental supplementation of food results in enhanced winter survivorship of Blackcapped Chickadees (Poecile atricapilla), demonstrating that food can be a limiting resource (Brittingham and Temple 1988, 1992a, b; Desrochers et al. 1988; Egan and Brittingham 1994). The U.S. Fish and Wildlife Service has estimated that 63 million United States citizens feed birds (Caudill and Laughland 1998). For much of the heavily human-populated eastern seaboard, bird feeders are likely to be encountered by passerines at some time during the year; feeders are now a part of the landscape for birds. The objective of this study was to examine the effects of supplemental feeding on Black-capped Chickadees in a remote region of central Maine with a sparse human population, where no other sources of supplemental food were available for wintering chickadees. I examined both population and individual responses to supplemental food. 1 Dept. of Biology, Colby College, Waterville, ME 04901; E-mail: [email protected] MATERIALS AND METHODS This study was conducted along Long Falls Dam Road on the eastern shore of Flagstaff Lake, Maine (458 109 N, 708 019 W) at altitudes ranging from 350– 440 m. Most of the study area was Maine Reserved Land, protected second growth forest. There were no human dwellings along this portion of the road and the closest bird feeders were 15 km distant. The forest was dominated by conifers: red spruce (Picea rubens), balsam fir (Abies balsamea), eastern white cedar (Thuja occidentalis), eastern hemlock (Tsuga canadensis) and eastern white pine (Pinus strobus). Common deciduous trees were paper birch (Betula papyrifera), American beech (Fagus grandifolia) and quaking aspen (Populus tremuloides). A National Weather Service station was located at Long Falls Dam, adjacent to Long Falls Dam Road. Mean high and low monthly temperatures for the study period were 24.9 and 213.78 C for December, 25.4 and 217.48 C for January, and 24.2 and 215.78 C for February. In October 1995, a total of 16 sites was established along 19 km of Long Falls Dam Road. Consecutive sites were at least 1.0 km apart. Each site had similar topography and forest composition (primarily red spruce and balsam fir). The 16 sites were grouped into four blocks of four consecutive sites, with each of four treatments randomly assigned once in each block. These treatments were (1) Continuous, in which sunflower seed was provided ad libitum from 25 October 1995 until 12 March 1996; (2) Early, in which sunflower seed was provided ad libitum until 11 January 1996 when the feeders were taken down; (3) Late, in which food was provided only between 11 January and 12 March 1996; and (4) Control, where food was never provided. Only three replicates of the Early treatment were conducted because of theft of the feeders at one site. 66 THE WILSON BULLETIN • Vol. 113, No. 1, March 2001 Supplemental food was provided using Magnumt sunflower seed feeders, two feeders per site except as noted, according to treatment (Continuous, Early, Late). The feeders were made of 6-mm hardware cloth, measured 18 cm in diameter and 30 cm tall, and permitted several birds to feed at once. Each feeder was suspended from a wire cable strung between two trees at a height of approximately 2 m. The two feeders were placed within 5 m of each other to facilitate simultaneous observation. Each site was visited weekly (except for two weeks in late December). The feeders were filled each week with black oil sunflower seeds. During my 2-wk absence I placed a third feeder at the two sites that received the heaviest use to ensure that the food would not be depleted. At the end of my 2wk absence, every feeder had some seed remaining. Each week I conducted a morning survey along Long Falls Dam road. At each of the 16 sites I stopped my car and spent three min censusing chickadees. At many of the sites, no birds had been banded so no attempt was made to identify banded birds for these counts. For sites with feeders where bird abundance was high, I recorded the maximum number of chickadees that I could see or hear at one time within the 3-min observation period. As block effects were not evident (one-way ANOVA, F3,281 5 1.112, P 5 0.344), the four blocks were pooled to increase the degrees of freedom for the error term. A repeated-measures two-way ANOVA was used to test for differences in numbers of birds in each of the four treatment types, for differences among dates, and for the expected interaction between date and feeder treatment. Scheffé post hoc contrasts were used to identify where significant differences lay. On 9, 11, 18 and 29 November 1995 I mist netted birds feeding at the four Continuous sites. Each Blackcapped Chickadee was banded with a unique combination of an aluminum U.S. Fish and Wildlife Service band and two color bands. Chickadees were aged as either HY (first winter) or AHY (adult) according to the amount of wear on the rectrices (Pyle 1997). I visited the four Continuous sites repeatedly for 2d periods every week of the study period except for the 2-wk hiatus in December. I chose two Continuous sites (Sites 1 and 12) for my most intensive observations. During each week’s observations I made at least four 30-min observations at each site. I visited once in the early morning (07:00–09:30), once in the late morning (09:30–12:00), once in the early afternoon (12:00–14:00) and once in the late afternoon (14:00– 16:30) over the 2-d period. During each 30-min observation I stood at a point equidistant from the two feeders. The observation distance varied from 5–8 m, determined by optimal viewing location within the vegetation. During those 30 min I recorded the visit of every Black-capped Chickadee that came to the feeder and took a seed. Birds that came to the feeder and were displaced or frightened away before feeding were not counted. I used 7 3 42 binoculars and a hand-held audio tape recorder to observe feeders and record data. I measured the temperature at the beginning and end of each 30-min period, and used the mean of those two measures in subsequent analyses. Individual Black-capped Chickadees were identified with various degrees of precision. Some birds that came to the feeders did not provide me a sufficient view to determine if bands were present; these chickadees were recorded as C (chickadee). I recorded unbanded birds that visited a feeder as uC (unbanded chickadee). Other birds were obviously banded but viewing angles prevented me from seeing all three bands; these birds were recorded as bC. Finally, some birds that came to the feeder could be identified by their unique combination of color bands. For each 30-min period I calculated the total number of successful feeder visits made by the chickadees. However, two confounding effects, variable chickadee numbers and variable feeding rates, make such data difficult to interpret. To separate these two effects, I used the mark-recapture analysis program NOREMARK (White 1996) to estimate the total number of chickadees using the feeders each week. This software explicitly allows for resighting of marked but imprecisely identified individuals (bCs in the present analysis). The software also does not assume that the probability of resighting for each marked individual is the same. NOREMARK does incorporate the assumption of a closed geographic population. Before generating population estimates, I had to classify the C chickadees, i.e., those that I could not see well enough to know if they had been banded or not. I classified C chickadees for each 30-min observation period by totaling the number of uC bird visits and the number of banded bird visits (bC birds plus all visits by uniquely identified birds). The proportions of unbanded and banded birds were used to apportion the unclassified chickadees to either the banded or unbanded class. Because the site fidelity of banded chickadees was very high, I was able to document patterns of feeder use by individuals in the population of banded birds. Using estimates of total feeder visits by all banded birds at each site, I tested for differences in feeder use as a function of time of day using the early and late morning and early and late afternoon criteria defined above. I also tested for seasonal differences, dividing the study period into early (9 December–11 January), middle (11 January–8 February) and late (13 February–12 March) winter periods. In each case, one-way ANOVA was used to test for significant differences overall and pair-wise Scheffé contrasts were used to identify means that were significantly different. Unpaired t-tests were used to test for differences in mean number of feeder visits of adult versus first-winter chickadees.