We examined 70 years (1931–2000) of black brant Branta bernicla nigricans abundance on Humboldt Bay, California. We used linear regression to convert count data to a standard variable (use-days) for evaluating hypotheses that explain temporal trends in brant use. Winter and spring brant-days on Humboldt Bay declined sharply in the mid-1950s and continued to decline through the mid-1980s, but have since increased. Evidence suggests that this trend may have been driven largely by changes in temporal patterns of hunting pressure on Humboldt Bay. We found little convincing support for alternative hypotheses such as changes in eelgrass Zostera marina condition over time, effects of non-hunting disturbance, and correlation with trends in abundance at the flyway level. Our study affirms the appropriateness of current hunting regulations for brant in California, but poses a challenge to wildlife managers who wish to provide hunting opportunities without displacing brant from important staging and wintering areas.
During spring migration, black brant Branta bernicla nigricans (in the following referred to as ‘brant’) feed on eelgrass Zostera marina in shallow bays and estuaries along west coast North America (Reed et al. 1998b, Moore et al. 2004), where they accumulate nutrient reserves critical for successful reproduction (Ankney 1984, Ebbinge & Spaans 1995). Humboldt Bay is the most important spring staging area in California based on peak use data, and the fourth most important site in the Pacific flyway (Moore et al. 2004). Lee (2001) estimated that approximately 60% of the flyway population used the bay in spring 2001, probably because of its high eelgrass abundance and relative isolation from other staging areas (Moore et al. 2004). Our goals in this paper are to: 1) use historic counts to describe trends in brant use of Humboldt Bay, 2) evaluate use-patterns in relation to variation in hunting disturbance or Pacific flyway population trends, and 3) recommend an annual protocol for monitoring brant use.
Single winter censuses for Humboldt Bay have been conducted since 1931, with more frequent fall-, winter- and spring-counts since the 1970s. Data indicate that winter and spring brant use declined suddenly in the mid-1950s, continued to decrease through the mid-1980s, and then increased through 2000. Midwinter survey data reveal similar declines at winter and spring sites in Oregon and Washington (Drut & Trost 2000). Proposed hypotheses to explain changes in brant use primarily include disturbance due to hunting and other human factors and eelgrass habitat degradation (Einarsen 1965, Smith & Jensen 1970, Henry 1980, Subcommittee on Pacific Brant 1992). Investigation of these hypotheses has been hampered, however, by the sporadic nature and inconsistent timing of surveys, and lack of historical data for eelgrass condition or non-hunting human disturbance. We found predictive relationships between seasonal use-day estimates and count data, thus enabling us to recreate a historical record of brant use-days on Humboldt Bay that could be used to empirically test two possible hypotheses concerning use at a key staging area.
Material and methods
Humboldt Bay, located on the northern California coast, is the second largest estuary in the state, with a water surface area of 62.4 km2 at mean high water (MHW; Proctor et al. 1980). The bay has three main sections, two of which receive 99% of the bay's brant use, i.e. Arcata Bay and South Bay. These areas consist of extensive tidal flats, accounting for 65–70% of the total MHW area of the bay (Barnhart et al. 1992). Eelgrass, which varies in extent annually, occurs below about +0.3 to +0.4 m (relative to mean lower low water; MLLW). In 1997, eelgrass covered approximately 1,044 ha of which 309 ha were in Arcata Bay and 720 ha were in South Bay (Terra-Mar 1997). A long narrow channel (Arcata Channel) connecting two sections of the bay contains about 15 ha of eelgrass, and supports about 1% of annual brant use (Humboldt Bay National Wildlife Refuge (HBNWR), unpubl. data). Shoot and biomass densities are also consistently higher on South Bay than on Arcata Bay, where an estimated 78–90% of the total biomass occurs (Keller 1963, Waddell 1964, Harding & Butler 1979, Bixler 1982).
Sources of brant abundance data
Moffitt (1931–1941, 1943) conducted annual February brant surveys along the entire coast of California during 1931–1942. We used U.S. Fish and Wildlife Service (USFWS) mid-winter survey data for 1940–1974. Midwinter counts were usually conducted in January; however some were as late as February or March (Henry 1980; Marty Drut, USFWS, pers. comm.). We obtained additional count data from Bentley & Christianson (1957, cited in Henry 1980), Denson (1961), Denson & Murrell (1962), and Monroe (1973, cited in Henry 1980). During 1975–1978, Henry (1980) conducted weekly counts of brant on Humboldt Bay during fall, winter and spring, and estimated weekly, monthly and seasonal brant use (i.e. number of brant use-days). Since 1978, USFWS personnel at HBNWR used this protocol to continue the surveys. Except for USFWS mid-winter aerial surveys, counts were ground based.
Analysis of winter and spring brant use
Two factors made it impossible to directly compare all brant data. First, while data from 1975–2000 provided direct estimates of seasonal brant use-days, surveys in many of those years occurred on South Bay only (Table 1). We estimated total Humboldt Bay use-days in these years by dividing use on South Bay by 0.83, as this was the average proportion of total use that occurred on South Bay in 11 years between 1975 and 2000 (range: 0.78–0.94; Moore et al. 2004).
Individual counts of black brant during winter and spring on Humboldt Bay (Arcata and South bays). In months when multiple counts were conducted, the peak count is reported.
Secondly, data prior to 1975 typically consisted of 1–4 counts during winter or spring, but they have not been used to estimate seasonal use-days. We used linear regression on South Bay data during 1975–2000 (N = 24 years) to identify predictive relationships between individual counts conducted at different times during winter and spring, with seasonal and total use-days in the same year. For example, we examined whether a single count conducted in the second week of February (as per Moffitt's protocol through the 1930s) could predict the estimate of use-days occurring in the same winter or spring. If a strong relationship existed, we extrapolated pre-1975 winter or spring brant use-days by entering available count values into a regression equation with slope coefficients that were estimated using the post-1975 data. We defined a strong relationship to be the regressions in which > 50% of the variation in use-days was explained by count data (i.e. adjusted R2 > 0.50). Presumably, we could have used a slightly different cutoff (e.g. R2 > 0.40 or 0.60) with similar results. We defined winter use as that occurring during January and February, and spring use as that occurring from March through May. We excluded December in winter use-day estimates because: 1) several of the years used to identify regression relationships (1975–2000) contained no December data, 2) pre-1975 December counts provided little additional predictive power to estimate winter use-days because they were conducted in years when both January and/or February were also counted, 3) it is not clear from historical counts whether brant in December were late-fall migrants, winter residents or early-spring migrants, and 4) relative to brant numbers in January and February, numbers in December have been low since the mid-1950s (see Table 1), and were typically low before 1975 also (Moffitt 1932, 1935, 1936); therefore, December use-days probably contributed relatively little to the number of use-days each year.
Different regression models were used to estimate brant use-days in different years, with independent variables defined by the dates for which pre-1975 count data were available (Appendix I). To estimate use-days from Moffitt's data in the 1930s and 1940s, our independent variable was the number of brant recorded during the second week of February. If only monthly peak numbers were reported in a particular pre-1975 year, we described the relationship between monthly peak counts and brant use-days in post-1975 years. In some instances, we did not know the exact date on which a pre-1975 count was conducted, nor whether it was a peak estimate. So, we assumed that such a count was a peak estimate if it exceeded peak values in post-1975 years. When the number of brant on a pre-1975 survey equaled zero, and we only knew which month the survey was conducted, we assumed this count was from the first week of the month, when numbers would be at their lowest (zero-counts were always from January or February). When multiple counts were available to predict use-days in a particular pre-1975 year, we used stepwise regression to select the combinations of counts that produced the best predictive model (i.e. that which maximized the adjusted R2).
Once we had estimated historical brant use-days for each year, we used linear regression to determine whether annual variation on Humboldt Bay corresponded with the historical timing of hunting seasons or annual harvest estimates based on post hunting-season interviews with hunters (M. Drut, pers. comm., California Department of Fish and Game, unpubl. data). We also used regression to investigate whether annual trends in brant use-days on Humboldt Bay correlated with annual Pacific flyway population estimates (Drut & Trost 2000).
Analysis of fall brant use
Single counts were conducted on Humboldt Bay during fall migration (October–November) in seven years during 1956–1975. Weekly counts were used to estimate brant use-days on South Bay in 17 years during 1976–2000. We estimated bay-wide use (Arcata Bay and South Bay combined) in years following 1975 by dividing use-days on South Bay by 0.96, since the average percentage-use for that area ranged within 94–97% in fall (HBNWR, unpubl. data). Using the same methods as for winter and spring data, we identified linear relationships between fall use-days and peak counts from October and November in post-1975 data, from which we estimated fall use-days for the seven years prior to 1975 (Adjusted R2 = 0.62, F2,13 = 13.17, P < 0.001 ). We used a Mann-Whitney test to examine whether fall use-days were lower in years when hunting seasons took place in fall.
Winter and spring brant use
Strong relationships existed between estimates of winter brant use-days and individual brant counts conducted in January or February (Adjusted R2 = 0.74–0.99, F1–2, 18–22 > 54.0, P < 0.001; see Appendix I), enabling extrapolation of winter brant use in all pre-1975 years with count data (Fig. 1). Winter brant use-days were much greater prior to the 1950s than in the decades following, with estimated use-days before 1954 commonly ranging within 1–2 million. By contrast, winter use-days since then have frequently measured below 100,000 use-days and reached a winter low of about 2,500 use-days in 1979.
January and February count data did not correlate well with spring use-days estimates, so we could only estimate spring use-days in 13 years between 1931 and 1974 when March or April data were available. Like winter use-day estimates, confidence intervals were relatively narrow for these 13 spring use-day estimates because regression relationships between March–April counts and spring use-days were strong (Adjusted R2 = 0.72–0.95, F1–4, 18–22 > 29.9, P < 0.001; see Fig. 1). Unfortunately, only four of the 13 years pre-date 1954, which makes long-term evaluation of spring use-days in relation to hunting-season timing difficult. However, inspection of available spring counts (see Table 1) and use-day estimates (see Fig. 1) suggests that spring use-days on Humboldt Bay may have commonly exceeded 1 million use-days before 1954 (see Fig. 1), with estimates as high as 4–5 million use-days during this time.
In spite of the relative lack of March and April count data, we estimated the number of annual brant use-days for the entire goose season (January–May) from 1931 (see Fig. 1). However, the precision and apparent accuracy of use-day estimates depended on the number of counts available in a given year. In years when both winter and spring counts were available (see Table 1), estimates were generated from regression models with high adjusted R2 values (see Appendix I), and thus had relatively narrow confidence intervals (see Fig. 1). By contrast, when only winter counts (January or February) were available, total use-day estimates were generated from regression equations with lower adjusted R2 (see Appendix I), and had wider confidence intervals (see Fig. 1). Further, when only one or two counts were available, estimation accuracy may be sensitive to annual variation in the timing of migration. Annual estimates of total use-days (see Fig. 1) ranged within 1–6 million use-days before 1954. By contrast, the total number of use-days following 1953 have usually been < 1 million, and reached a low of about 285,000 total use-days in 1985.
The timing of a major decline in brant use coincides with the initiation of winter hunting seasons on brant in California that ran from November or December through the end of January or mid-February (Subcommittee on Pacific Brant 1992). Prior to 1953, hunting seasons usually ran from October or November through mid to late December, although they did occasionally run through the first week of January, and until 20 January in 1946 (Fig. 2). From 1954 to 1983, winter and spring brant use on Humboldt Bay steadily declined (Fig. 3). Then, in the winter-spring of 1983/84, the hunting season reverted to the fall (see Fig. 2). Winter and spring use has steadily increased since then (see Fig. 3), peaking in 1998 at about 461,000 and 1,207,000 use-days, respectively. However, these estimates are still short of estimated pre-1950 levels. Harvest estimates were not available prior to 1958, so we could not directly relate harvest levels to the sudden decline in brant use in the mid-1950s. However, for the period 1958–1983, we found no relationship between winter harvest and winter use-day estimates in the same year (R2 = 0.08, F1,14 = 1.25, P = 0.28; Fig. 4), nor between harvest in one year and change in winter use-days in the following year (R2 = 0.03, F1,14 = 0.46, P = 0.51).
We investigated an alternative explanation for the downward trend in brant use of Humboldt Bay, which is that the flyway population simply decreased during this time. During the period of brant decline on Humboldt Bay (1960–1983), we found no significant relationship between the total number of brant use-days on the bay and the Pacific flyway brant population index (log transformed with one outlier removed; R2 = 0.002, F1,22 = 0.04, P = 0.85; Fig. 5). Similarly, no relationship existed between population size and number of brant use-days during the period of increasing brant use on Humboldt Bay, which occurred from 1984 to 2000 (R2 = 0.08, F1,14 = 1.29, P = 0.27). Unfortunately, flyway data are not complete prior to 1960, so we could not test whether the dramatic decrease in use of Humboldt Bay in the early 1950s was associated with a flyway level decline.
Fall brant use
We compared data collected during 1956–1982 when hunting took place in January and February but not in October or November, with data collected during 1983–1999 when fall-hunting took place (see Fig. 2). In years when fall-hunting was absent, brant use was variable, but higher on average than during years when brant were hunted (Mann-Whitney test: x̄nohunt = 5,507, x̄ hunt = 1,671, Z = −2.42, P = 0.016, N = 24; Fig. 6). Fall harvest was positively correlated with fall use-day estimates, indicating that hunters were more successful in years with more birds (Fig. 7). More interestingly, fall harvest was negatively related to fall use-days in the subsequent year with marginal statistical significance (see Fig. 7), suggesting that harvest intensity (either by number reduction or by related disturbance) may have a direct effect on fall brant use.
During 1975–1999 (the years when observed use-day estimates were available), the majority of the fall use occurred in November (mean = 91%, range: 76–100%, N = 16). This information is consistent with reported departure dates from Izembek Lagoon (Dau 1992), and early reports suggesting that brant rarely arrived on Humboldt Bay before November (Moffitt 1932, 1935, 1938, 1940, 1941, 1943).
Historical brant use of Humboldt Bay
Historically, Humboldt Bay has been the most important area in California for wintering and migrating brant. During 1931–1942, Humboldt Bay accounted for 36–84% (x̄ = 68%) of the brant observed in California during February (Moffitt 1943). Based on USFWS midwinter inventories (Drut & Trost 2000), this value ranged within 12–100% during January 1941–1952 (x̄ = 60%), and according to incomplete winter estimates of the entire flyway in 1951 and 1952 (Drut & Trost 2000, Leopold & Smith 1953), Humboldt Bay may have harboured as much as 15–20% of the Pacific flyway brant population in January. From the mid-1950s through the mid-1980s, however, winter and spring brant use declined precipitously on Humboldt Bay, then increased from the mid-1980s through 2000, but still remains below pre-1950s levels. Similar declines occurred at other winter- and spring-staging sites in Oregon and Washington (Drut & Trost 2000), and coincided with an increase in winter use at new sites along the western coast of mainland Mexico in the early 1960s (Smith & Jensen 1970, Drut & Trost 2000), and with initiation of winter hunting seasons that extended through mid-February rather than the end of December (Subcommittee on Pacific Brant 1992).
Pacific flyway mid-winter population size did not explain the historical variability in brant use of Humboldt Bay. No correlation was found between January–May brant use on Humboldt Bay and the flyway population size during any period in 1960–2000. Therefore, the decline in use on Humboldt Bay during 1954–1983, and the subsequent increase during 1984–2000, was not simply tracking any such trend in the population as a whole.
Eelgrass condition has been suggested as a factor affecting brant use of Humboldt Bay based on few empirical data and anecdotal data (Einarsen 1965, Henry 1980). In the winters/springs of 1937/38, 1940/41, 1951/52, 1952/53, 1957/58 and 1997/98, substantial numbers of brant fed in salt marshes and pastures surrounding Humboldt Bay (Moffitt 1938, 1941, Leopold & Smith 1953, Murrell 1962; HBNWR unpubl. report 1998). In all cases, this behaviour was attributed to poor feeding conditions on the bay, where eelgrass was observed covered with slimy or silty deposits, and/or infected with Labyrinthula (implicated in the wasting disease of eelgrass on the Atlantic coast in the 1930s; Moffitt & Cottam 1941). Eelgrass in these years was greatly reduced in its extent and severely depleted by brant. However, in incidents prior to 1954, estimated brant use on Humboldt Bay remained high. Similarly, in the winter/spring of 1997/98, the greatest number of brant use-days in the last 25 years was recorded. Reduced food abundance on Humboldt Bay may thus affect brant feeding behaviour and habitat use, without reducing use of the overall area. The steady decline in brant use from the mid-1950s through the mid-1980s, if induced by poor eelgrass condition, would presumably have resulted from a long-term change in eelgrass habitat for which no evidence exists. In fact, several studies suggest that eelgrass was healthy and abundant from 1959 through 1962 (Murrell 1962, Keller 1963), in 1972 (Harding 1973), in 1975–1977 (Henry 1980), and in 1980–1981 (Bixler 1982). We therefore consider it unlikely that historical trends in brant use of Humboldt Bay could be explained by changes in quality or size of eelgrass habitat.
The most widely held view is that winter hunting disturbance from the 1950s to the 1980s reduced winter and spring brant use of Humboldt Bay and other sites in California, Oregon and Washington, and may have driven brant to new wintering sites in Mexico (Denson 1964, Smith & Jensen 1970, Henry 1980, Subcommittee on Pacific Brant 1992). During 1931–1953, when winter and spring brant use of Humboldt Bay was consistently high, the hunting season began in October or November, and typically ended by the end of December (see Fig. 2). This could have affected fall migrants, for which there are no data, but not spring migrants. During 1954–1983, the hunting season consistently ran through late January or mid-February (Subcommittee on Pacific Brant 1992), and declining winter and spring use of Humboldt Bay occurred during this time. Since 1983, hunting seasons in California have been limited to fall months, during which time winter/spring use-days have increased and fall use-days have decreased on Humboldt Bay. In Washington and Oregon also, hunting seasons have been more restrictive since 1983; during 1984–1986 there were no open seasons in these states, and since then seasons have been shorter and have ended earlier than before 1984 (Subcommittee on Pacific Brant 1992). Midwinter survey data (Drut & Trost 2000) suggest that winter brant numbers in Washington have increased since the mid-1980s.
We cannot definitively conclude whether decreased brant use of U.S. sites was a cause or consequence of increased use in Mexico. However, it is not surprising that hunting might have had such an impact on brant use on Humboldt Bay and other U.S sites in general. Shifts in local and flyway-wide distributions due to hunting pressure have been documented in several other waterfowl populations (references in Fox & Madsen 1997, Bechet et al. 2003). Hunting activities apparently interrupt foraging time, increase energetic costs due to extra flying time, and displace birds to less profitable feeding areas (Fox & Madsen 1997, Madsen 1998) such that birds are unable to build adequate fat and nutrient reserves (Madsen 1995, Madsen & Fox 1995, Feret et al. 2003), a prerequisite to successful breeding in brant (Prop & Deerenberg 1991, Ebbinge & Spaans 1995) and other northern geese (Madsen 1995Madsen 2001, Black et al. 1991, Prop & Black 1998). In addition, disturbances near traditional grit sites and roosts may limit the birds' ability to replenish gizzard grit thus reducing digestive efficiency. Wild geese are thought to adjust their foraging routines on temporal and spatial scales to best achieve requisite fat and nutrient reserves (Owen & Black 1990, Prop et al. 2003). Geese that are unable to meet their daily energetic needs are more likely to initiate movements to new areas (e.g. younger birds; Black 1998, Black et al. 1991), thus establishing new migratory traditions (Black et al. in press).
Detailed accounts of hunter disturbance (Murrell 1962, Henry 1980) describe south Humboldt Bay as being essentially unavailable to brant during a large portion of the hunting season. A narrow sand spit that separates South Bay from the ocean provides important roost and grit habitat to brant (Lee 2001), and is crossed by brant to enter the bay. Heavy hunting pressure along this spit prevents access to the bay and sand habitat, while intensive hunting from scull boats and offshore blinds precludes use of sanctuary areas within eelgrass habitat. Such disturbance effects in January, and especially in February, could have consequences for brant use in March and April as well, if winter residents make up a significant fraction of the spring population. Lee (2001) found that spring-staging birds arriving to Humboldt Bay in late January through February stayed approximately 30–50 days on average. Alternatively, social facilitation may act such that brant on the bay in February attract later arrivals. This also seems plausible, given that later spring migrants are comprised of a greater fraction of juveniles (Henry 1980, Reed et al. 1998b, Lee 2001) who may need to learn the location of important staging areas from earlier arriving adults. Though unstudied in brant, related learning processes have been suggested for juvenile Canada geese Branta canadensis (Williams & Kalmbach 1943, Surrendi 1970, Raveling 1976).
Hunting, in addition to its disturbance effects, might also reduce the number of use-days in an area simply by reducing local population numbers through hunting mortality. Adult brant show high fidelity to winter and spring-staging sites (Reed et al. 1998a), so high hunting mortality as seen in some years during 1958–1983 (see Fig. 3) could have long-term negative impacts for the number of use-days on Humboldt Bay. Harvest data do not exist for Humboldt Bay prior to 1958, so we could not rigorously evaluate whether extreme harvest mortality caused the sudden drop in brant use-days in the early 1950s. However, for years when harvest data were available, we found no relationship between winter harvest intensity and winter use-day estimates, suggesting that overharvest may not have been the primary driver of winter or spring population trends on Humboldt Bay, at least since 1958. Furthermore, an ‘overharvest hypothesis’ does not predict increasing use of other areas in the flyway (i.e. Mexico). The coincident rise in use of Mexican sites in the 1960s therefore suggests that northbound migrants were not reduced by harvest on Humboldt Bay, but may have been displaced, so that for the past several decades brant have spent most of January and February in Mexico rather than in the U.S. For southbound migrants, data do suggest that fall harvest intensity may have affected fall use-days in the subsequent year. Why might there be a relationship between harvest intensity and brant use-days in fall, but not winter and spring? One possibility is that while brant use-days are much higher in winter than in fall, harvest intensity has been similar in these two seasons. Fall-harvest numbers during 1983–2000 (Median = 304; 1st-3rd quartiles: 290–440) were of comparable magnitude to winter harvest estimates between 1958 and 1983 (Median = 1,490; 1st-3rd quartiles: 684–2,235), whereas fall brant use during 1984–2000 (Median = 1,638, 1st-3rd quartiles: 819–2,086) was roughly 20 times lower than estimated winter use during 1958–1983 (Median = 30,028, 1st-3rd quartiles: 21,064–38,722). Thus, we might expect a given level of harvest to have a greater measurable effect on fall brant use-days.
Human population and industrialization surrounding Humboldt Bay has increased, as has related sources of non-hunting human disturbance to brant. These changes have been quantified by Henry (1980) and Schmidt (1999), who identified the activities of clam fishermen, recreational and commercial boaters, oyster culture, low-flying aircraft, and vehicle traffic (especially off-road vehicles) on the east shore of South Bay (South Spit) as the most significant contributors. In addition, a temporary camping settlement on South Spit was described as a major deterrent to brant use at roost and grit sites during the first half of the 1990s (HBNWR, unpubl. reports 1991–1995). We agree with previous authors that these factors are likely to have played a role in reducing winter and spring brant use-days following the early 1950s, and suggest that persistent human disturbance could prevent brant use-days from returning to historical levels. However, these non-hunting disturbances are probably not solely responsible for the overall trends in brant use-days observed over the past 70 years. Based on our historical estimates, Humboldt Bay suffered a steep drop in brant use in 1953 or 1954, which persisted and declined further through the mid-1980s. This decline definitely coincided with a change in timing of winter hunting seasons, whereas there is no indication that other sources of disturbance also occurred in a punctuated and then persistent manner. Furthermore, the number of winter and spring brant use-days have increased since the cessation of winter hunting in the mid-1980s, despite probable increases in non-hunting human disturbance.
While January and February brant use has increased in response to fall hunting seasons, December use (while hunting is occurring) has not similarly responded, having remained consistently low on Humboldt Bay since at least the 1960s. Occasional high counts in December during the 1930s and 1950s, along with large January counts during the 1940s, suggest that larger numbers of brant arrived earlier, at least in some years, than they do today. Increased winter use along the mainland coast of Mexico has presumably resulted in fewer brant moving up into California before January, so while winter use on Humboldt Bay has increased in the absence of hunting disturbance, the arrival of brant is still somewhat delayed each year.
Since 1956, fall use has been low and irregular on Humboldt Bay, although higher on average in years without fall hunting. During fall migration, most brant fly non-stop from Alaska to Mexico (Dau 1992), and early reports suggest that relatively few brant stopped at Humboldt Bay in fall (Moffitt 1932). The variability of fall use in non-hunting years can probably be explained by the fact that Humboldt Bay is not a major fall-staging ground, so numbers are more likely to depend on annual variation in weather patterns and the number of birds unable to make the otherwise non-stop migration (Dau 1992).
Collection of long-term data on single species is valuable for management, but may be costly and difficult to maintain. Conversely, studies of shorter duration often preclude analysis of historical trends in population size, distribution and habitat condition and use. We suggest that accurate predictions of brant use-days (Ra2 > 0.90) at Humboldt Bay can be derived from single mid-monthly counts from January through April, thus lowering the costs and facilitating the establishment of a long-term monitoring program, though such a protocol should be revalidated and calibrated periodically via thorough estimation of seasonal brant use-days.
Using this method, we extrapolated historical brant use of Humboldt Bay, which enabled us to identify long-term trends and investigate factors thought to affect them. We suggest that 30 years of hunting disturbance during early spring migration resulted directly in declining brant use during 1953–1983. Other forms of disturbance and habitat conditions probably contributed relatively little to this decline. This conclusion has been assumed by wildlife managers, who have responded by maintaining a fall hunt in California since 1984. The results of our study empirically affirm the appropriateness of the current policy; relieving hunting pressure has increased brant use of Humboldt Bay. However, this poses a challenge to managers of how to provide hunting opportunities without driving brant away from important migration areas.
We thank K. Forrest, R.J. Guadagno and M. Drut for providing us with, and interpreting, historical brant survey reports. We are grateful to D. Yparraguirre who helped us see the many complexities involved in waterfowl management. We also thank Dan, along with T. Fox, D. Ward, S.E. Sheaffer, and two anonymous reviewers for their helpful comments on earlier versions of this manuscript.
Regression equations used to estimate winter, spring or total brant use-days during 1931–1974 (all F1–5, 18–23 > 27, P < 0.001). Equations were derived from 1975–2000 data, but were based on count dates from pre-1975 years. For example, in 1949 one count was conducted during the second week of January. Therefore 1975–2000 data were used to determine the relationship between winter use-days and a count conducted in this week. The regression relationship was then used to predict winter use-days in 1949 from that year's count.