Environmental effects on body size of Canada Geese

Auk, The, Jan 1998 by Leafloor, James O, Ankney, C Davison, Rusch, Donald H

JAMES O. LEAFLOOR,(1,4) C. DAVISON ANKNEY,(2) AND DONALD H. RUSCH(3)

ABSTRACT.-Canada Geese (Branta canadensis interior) breeding on Akimiski Island, Northwest Territories, and those breeding on the Ontario mainland southwest of James Bay, share a common wintering range and are considered to be part of the same population, but differ significantly in size (e.g. skull length of mainland birds averages ca. 8% greater than that of Akimiski birds). We collected eggs in each area and raised the goslings in a common environment to determine environmental effects on differences in body-size traits observed in wild birds. We found no differences in asymptotic size or growth periods for skull, culmen, and tarsus length among birds from different origins (P > 0.05) but significant differences between sexes (P

VARIATION IN BODY SIZE of birds can be influenced by genetics (Boag and van Noordwijk 1987) and by environmental factors (James 1983; Richner 1989; Cooch et al. 1991a, b; Larsson and Forslund 1991; Sedinger and Flint 1991; Rhymer 1992; Lindholm et al. 1994). Thus, differences in body size between populations of the same species do not necessarily indicate genetic differences (Boag and van Noordwijk 1987, Larsson and Forslund 1991). For example, body size of Canada Geese (Branta canadensis) generally decreases with increasing latitude (Aldrich 1946), presumably in response to shorter growing seasons in the north (Dunn and MacInnes 1987).

Skull lengths of Canada Geese from Akimiski Island are significantly smaller than those of conspecifics nesting on the mainland less than 200 km away, and they are smaller than expected based on the latitude at which the geese nest (Leafloor and Rusch 1997). Canada Geese that nest in these two areas share a common wintering range and are considered to be part of the Southern James Bay population (Leafloor and Rusch 1997, Trost et al.1997). We equate differences in skull length with differences in overall body size because these variables are significantly correlated in B. c. interior (Moser and Rusch 1988), but we recognize that univariate measures usually are not the best indicators of body size (e.g. Willig et al. 1986, Alisauskas and Ankney 1987, Rising and Somers 1989, Freeman and Jackson 1990). Such differences in body size conceivably could result from local genetic differentiation in the absence of gene flow between the island and mainland populations (e.g. Mayr 1963, Shields and Wilson 1987, Van Wagner and Baker 1990), and/or from environmental factors differentially affecting gosling growth and subsequent adult size in each area (e.g. Ehrlich and Raven 1969; Cooch et al. 1991a, b; Larsson and Forslund 1991).

We measured growth of Canada Goose goslings hatched from eggs collected from Akimiski Island, Northwest Territories and near Kinoje Lake, Ontario (Fig. 1) and raised in a common environment to evaluate the effect of environment on differences in body size. We also compared measurements from a sample of known-aged goslings raised naturally on Akimiski Island with those that we raised in captivity. Hereafter, we use "island" to refer to Akimiski Island and "mainland" to refer to the area south of Attawapiskat, Ontario shown in Figure 1.

METHODS

We randomly collected one Canada Goose egg from 40 different nests on Akimiski Island (53 deg 12'N, 81 deg 30'W) and 40 eggs from 26 different nests (maximum two eggs per nest) near Kinoje Lake (51 deg 30'N, 81 deg 45'W) on 27 and 28 May 1993. We measured maximum length and width of all eggs with digital calipers ( /- 0.1 mm). Eggs were placed in portable incubators and flown to Lake St. Clair National Wildlife Area, Ontario (42 deg 30'N, 82 deg 30'W) on 28 May, where they were kept in a larger incubator until goslings hatched between 2 and 14 June. We marked all goslings at hatching with individually numbered size 1 Monel web tags. When goslings were dry, we weighed them with a Pesola spring scale ( /- 1 g) and measured head length, culmen length, and tarsus length with digital calipers ( /-0.1 mm; Dzubin and Cooch 1992). Thereafter, goslings were weighed and measured every three days until age 21, then every four days until 8 September, and finally every seven days until 13 October. We stopped measuring tarsus length when three consecutive measurements indicated no further growth. Geese were killed to provide tissue for genetic analyses, and sex of each gosling was confirmed postmortem.

We kept goslings indoors under heat lamps for the first few days (until they learned to feed on commercial chow by themselves) and then moved them outdoors to portable wire cages (2 x 3 x 1 m). Goslings had unlimited access to commercial chow (Purina Duck Grower, 18% protein) and were allowed out of their cages to feed on a mowed lawn ad libitum during the day. Goslings also fed on grass that grew beneath their wire cages, and we moved the cages every few days to ensure a constant supply of growing grasses. As goslings approached fledging in midAugust, we transferred them to larger pens (ca. 7 x 3 x 2 m) that provided access to water from a creek but no access to vegetation. Thereafter, goslings were fed only Purina Duck Grower ad libitum until 13 October.

Visual inspection of individual growth curves suggested that an asymptotic curve would accurately represent growth trajectories of goslings. We used the nonlinear regression program in SYSTAT (Wilkinson 1989) to fit growth data to the following asymptotic regression equation:

where X is the structural measure of interest (e.g. skull, culmen, tarsus), age is the age of the gosling in days, and a, b, and c are parameters estimated by the model for asymptote, shape, and the x-intercept, respectively. We used starting values of a = 120, 50, and 90; b = -0.03, -0.03, and -0.05; and c = -14, -15, and -5 for skull, culmen, and tarsus curves, respectively. For each structure, we used estimates of b and c to calculate the time required to grow to 50% and 90% of asymptotic size for each gosling using the following general equation: