Several factors coinciding in time and location have contributed to the observed population growth rates of white geese.
Agricultural food resource subsidy in winter and migration
Snow goose population size was once thought to be limited by over-winter survival, due to the species’ apparent narrow definition of suitable winter habitat (salt marsh) and destructive foraging (grubbing) of vegetation resulting in depletion of food resources (Lynch 1975). Wintering habitats along the Gulf of Mexico (LSGO) and Atlantic coast (GSGO) were primarily restricted to coastal bulrush (Scirpus spp.) and cordgrass (Spartina spp.) salt marshes (McIlhenny 1932, Bellrose 1980, Anonymous 1981, Smith and Odum 1981). The area of habitat available for wintering LSGO on the coasts of Texas and Louisiana was somewhat more than 200,000 ha. After the 1940s, LSGO expanded their foraging range (Lynch 1975) by incorporating rice prairies immediately adjacent to coastal marshes. Overall, there was nearly 400,000 ha of land in rice production at that time. Although coastal marsh habitat loss or change has been implicated (e.g., oil and gas development, urban expansion, dredging and filling; Bent 1962, Robertson and Slack 1995). Lynch (1975) argued that these anthropogenic effects were a minor factor in the changing pattern of land use by the birds, particularly because refuges were established in coastal marshes to protect such habitats. Louisiana and Texas had 223,000 ha (550,000 acres) of protected marshes in the early 1970s (Lynch 1975). The reliance of LSGO on rice prairies has increased further since the mid 1960s (Bateman et al. 1988, Hobaugh et al. 1989, Widner and Yaich 1990). Agricultural and hunting management practices that provided large areas of shallow water suitable for night roosting played an important role in expanding this use to interior rice prairies. The increase in the area of rice cultivation and the use of nitrogenous fertilizers have been dramatic since that time, and LSGO and ROGO now forage in over 900,000 ha of rice fields in Texas, Louisiana, and Arkansas. Although the harvested rice area has declined in the last decade, the yield per hectare on the upper Texas coast has continued to increase due to second cropping and more efficient harvesting (Hobaugh, Stutzenbaker and Flickinger 1989, Robertson and Slack 1995).
Further north in the mid-continent region and in the mid-Atlantic coast, geese exploit other cereal grains (Alisauskas et al. 1988, Anonymous 1981, Hill 1992, Reed 1992). A major impact of this conversion of natural grassland and bottom land forest habitats to agricultural use is that the former depletion of winter food resources in salt marshes, which presumably resulting in mortality or sublethal effects on body condition, has been removed as a factor limiting survival and population growth.
Corn, wheat. barley, oats and rye cultivation in the mid-western and northern prairies provide additional nutrient and energy subsidies for LSGO. Snow goose preference for corn in Manitoba in spring minimizes the historical influence that drought in natural habitats may have had on condition of pre-breeding geese (Alisauskas and Ankney 1992, Davies and Cooke 1983). Due to the important role of stored reserves in determining breeding success (Ankney and MacInnes 1978), this more reliable food source would lead to higher average reproductive potential. In fall, availability of waste grain in harvested fields has delayed the southward fall movement of geese and blurred the definition of "winter" range (Alisauskas et al. 1988). This effect is additive to the effects of refuges (see below).
On the Atlantic coast, GSGO formerly wintered in a restricted area of coastal salt marsh. Compared with LSGO, their use of habitats other than coastal marshes is more recent and more limited (Anonymous 1981, Reed 1992), but GSGO have fed in winter in agricultural cropland since about 1970 (Hill 1992). Historically, spring staging was limited to the tidal brackish Scirpus spp. marshes of the St. Lawrence River. In the late 1960s GSGO began to forage in agricultural land in Québec during spring (Reed 1992) as a result of a growing population and the cumulative annual depletion of Scirpus spp. plants (Gauthier et al. 1988).
Refuges placed in close proximity to traditional feeding areas and the loss of salt marsh as a consequence of development (Hindman and Ferrigno 1990) apparently assisted this shift. Ironically, concentrations of GSGO feeding intensively on refuge impoundments may have pushed marsh plants beyond their threshold of regeneration. Geese may have been forced to seek off-refuge foods in agricultural lands. In addition, industrial and urban development limited available salt marsh food supplies, and where forage was available, hunting pressure was high. In effect, a positive feedback was established between broadened habitat use and the population increase of LSGO and GSGO on both wintering grounds and staging areas.
The effect of refugia on migration routes
National Wildlife Refuges, state refuges and other wildlife areas were established throughout the United States, especially from the mid 1930s to the 1970s to protect and restore wetland habitat for breeding and migrating waterfowl and other wetland-dependent birds (Bellrose 1980). The attractiveness of these protected environments to waterfowl was quickly evident. The subsequent interruption of migration of LSGO and northward shift in the fall was particularly rapid and dramatic in the central United States from North Dakota to Louisiana and Texas. This eventually led to a reduction in traditional long distance flights from northern staging areas to Texas (Johnsgard 1974), and from James Bay to Louisiana (Cooch 1955).
Hunters were also attracted to some refuges which provided hunting opportunities, and in these places mortality increased (Johnsgard 1974). High local harvest rates near some refuges, as a result of both anticipated and unforeseen events (e.g., firing lines at refuge boundaries) may have led to poorer survival of birds using refuges than those that fed and roosted elsewhere. However, during the 1970s, changes in hunting practices near refuges included a reduction of firing lines, creation of no-hunting zones, manipulation of croplands to provide food, and a restricted harvest of geese on refuges and off refuges. The management practice of half-day hunting (Schroeder 1963) was initiated to hold migrant geese longer to increase hunting opportunities and local harvest, but its success also appears to have influenced distribution. These factors led to such migration sites functioning as true refugia (Frederick and Klaas 1982). Long-term reduction in the hunter harvest is consistent with the hypothesis of disproportional growth of population units using refuges (Raveling 1978). These refuges may thus function as loci for population growth and exploitation of surrounding "new" agricultural foods. Note that these events took place in the mid 1970s, the same time as the other factors contributing to the overall population increase of white geese appear to have taken effect.
Lower harvest rates
Annual survival of LSGO adults from Hudson Bay increased from about 78% in 1970 to about 88% in 1987 (Francis et al. 1992). The principal cause of mortality of adult geese in recent decades is hunting (Owen 1980). The expansion of the winter range of mid-continent snow geese and the lengthening of fall migration stopovers at northern latitudes has had a significant effect on dispersion of geese, hunter access and exposure to hunting. Harvest in the central US has declined along with hunter numbers over the past 25 years (Fig. 2.9a). More importantly, harvest rates (measured as proportion of mid-winter indexed population) declined from near 40% to under 8% annually (Fig. 2.9b) (see Rockwell et al. 1997). Increases in harvest per hunter (Fig. 2.9c) have not compensated. Canadian harvests of lesser snow geese have also declined in recent decades (Fig. 2.9d). The increase in survival may have been due, therefore, to a reduced overall harvest rate. If so, the beginning of LSGO population “escape” from constraining effects of annual harvest occurred in the early 1970s.
GSGO harvest rates from the 1880s to the 1930s are not recorded but presumed to be high. Hunting of this subspecies was prohibited in the United States from 1931 to 1975 (Gauvin and Reed 1987) in order to increase survival and population growth. The population grew at a slow rate from 1910 to 1960. Significant growth did not occur until 1950 which suggested that hunting mortality alone was not responsible for low population growth. There is no evidence that habitat was in short supply (Reed, pers. comm.). Perhaps periodic weather related failures were in part, responsible for checking growth. Additionally, traditional use of coastal marshes (i.e., behaviour), lack of development pressure, and the low population numbers (i.e., little density dependent foraging competition) may have kept them from finding agricultural foods.
From 1967 to 1988, three distinct periods were recognized: the first (1967-1974) was characterized by legal seasons in Canada only, by variable but generally low harvest (mean 12,800 geese/yr) and a mean kill rate of about 10%; during the second (1975-1981) hunting occurred in both countries and mean harvest (51,600/yr) and mean kill rate (23.9%) increased; in the third (1982-1988) the mean harvest rose to 59,000 but the kill rate decreased to 18.5%. A more recent analysis covering 1989-1993 (Reed, pers. comm.) showed a continued increase in harvest (mean 74,500 geese/yr) and a further decrease in kill rate (mean 13.5%). The addition of a U.S. hunting season in 1975 resulted in larger harvests and, initially, in increased harvest rates but this was insufficient to stop steady population growth. Since the early 1980s, harvest rates have been decreasing, and the spring population has increased by an average of 10% annually (1980-1996) (Reed, pers. comm.). This inability of hunting to control numbers of GSGO occurred about the same time it ceased to control population size of mid- continent LSGO.
Climate amelioration in the Arctic
A general warming trend from 1961-1990 has occurred in the central and western Canadian Arctic regions (Cohen et al. 1994) which include major breeding areas of LSGO and ROGO. MacInnes et al. (1990) showed that LSGO nesting occurred progressively earlier in the Hudson Bay region from 1951 to 1986. They suggested it was, in part, due to climate amelioration. As reproductive success in Arctic geese is positively correlated with early spring melts (reviewed by Owen 1980), warming of nesting areas could have led to higher annual population growth rates of mid-continent LSGO. Long term trends in decadal climatic patterns have shown the decade from 1970 to 1980 to be an anomalous warm period in the southern Hudson Bay region (W. Skinner, unpublished data) during which time spring melt occurred almost 15 days earlier than in the previous and succeeding decades. Thus, rather than a steady climatic improvement, there may have been a brief window of more favourable conditions that reduced the frequency of reproductive failures (the "boom-bust" pattern) typical of Arctic goose productivity up to 1980 (Owen 1980). Boyd et al. (1982) were not able to relate the upward trend in numbers to weather patterns between 1964-79, but qualified this finding because of the measures of breeding success that they used. However, with the exception of poor or bust production in 1972 (LSGO, GSGO), 1974 and 1976 (GSGO) and 1978 (LSGO) they had periods of sustained good reproductive success, as measured by proportion of goslings to adults in fall flights and on the wintering grounds (Boyd et al. 1982, Gauvin and Reed 1987). Years of poor reproduction do not always coincide between low and high arctic areas. GSGO showed decreasing frequency of “bust” years (<10% juveniles in fall) up to 1988 (Reed 1990); three times in the decade 1956-65, twice in 1966-1975 and once in 1976-1988. But in the last eight years (1989-1995) there have been two bust years (Reed, pers. comm.). The striking feature of this long term data set is the eleven year stretch from 1975 through 1985 in which there were no bust years and in which juveniles in fall averaged 27%.
The southern shift of nesting range of LSGO
The distribution of breeding birds has changed dramatically since the late 1920s. As a result, the centre of LSGO breeding range in the eastern Canadian Arctic geese has moved south to areas with a less severe climate (i.e., rather than climate change in situ). Snow clearance, on average, is 2 weeks earlier at Cape Henrietta Maria than at Baffin Island (Kerbes 1975). Before 1940, all known nesting colonies of lesser snow geese in the eastern Arctic were north of 60·N (Cooch 1958, 1961), whereas by 1973, 40% of the entire population was nesting south of that latitude as far as 55·N (Kerbes 1975).
MacInnes et al. (1990) suggested that with an earlier average start to nesting and a longer growing season, higher average annual production would result in population growth of these southern colonies. Evidence from Cape Henrietta Maria (Hudson Bay Project, unpublished data) and La Pérouse Bay (Cooke et al. 1995) (Fig. 2.3) supports this assertion. However, the slow growth of each colony in the first 2 decades following establishment argues against this as the sole mechanism to account for growth. Northern colonies continue to experience occasional weather-related "busts" in production (e.g., 1972, 1978; Boyd et al. 1982, and 1983, 1992; Kerbes, pers. comm.). Some birds which would normally nest in northern colonies may be induced in some years to nest at southern latitudes by these late melt conditions (Hanson et al. 1972, Geramita and Cooke 1982, Hudson Bay Project, unpublished data).
LSGO populations have also expanded their breeding range in the central Arctic, both on islands and interior mainland (McCormick and Poston 1988, McCormick 1989, Alisauskas and Boyd 1994, Kerbes 1994, B. Bromley, pers. comm.) where the long-term climate trend has been a gradual warming (Cohen et al. 1994).