The data presented here show that Pacific dunlins wintering on the Fraser estuary in British Columbia have changed aspects of their winter ecology over the past few decades. In the late 1970s they showed classic temperate shorebird winter behavior, with a regular routine of feeding at low tide and ground-based roosting at high tide, and a mid-winter peak in body mass, attributable to fat storage. After the mid-1990s, the mid-winter peak in body mass disappeared, and dunlins began to spend several hours during the high tide period in over-ocean flocking. We interpret these changes as an adaptive response to greatly increased peregrine falcon presence during the winter since the 1970s.
Table summarizes the evidence that over-ocean flocking has become more frequent. For three reasons we feel that we can discount the possibility that over-ocean flocking occurred as frequently during the early (pre 1994) studies, but was missed because early observers were unaware of its occurrence, while later observers had been alerted to this phenomenon. (A) Save Zharikov, none of the later observers was aware of over-ocean flocking when they began their studies (pers. comm.). This is not surprising, as the descriptions of over-ocean flocking available at the time [15
] were not in mainstream publications, and none devotes more than a few lines to the description. (B) None of the 56 sets of systematic high tide surveys documented during the winters of 1979/80 (n = 27; [19
]) and 1989/90 and 1990/91 (n = 29; [26
]) records a low number of roosting dunlins, as would be expected if the birds were engaged in unseen over-ocean flocking. (C) Two observers with extensive experience predating 1994 (see Table ) both documented their first observations of over-ocean flocking in the mid-1990s. We thus feel confident in concluding that on the Fraser estuary this phenomenon has greatly increased in frequency, duration, or both, since the late 1970s.
The data presented here also confirm that in the late 1970s dunlins were heavier than in the late 1990s, the difference being greatest (~4 g) in November, and shrinking steadily over the course of the winter until the pre-spring migration period, when masses are the same. This convergence rules out the possibility that the body mass difference is attributable to some systematic bias (e.g. scale calibration). Our measurements of total body fat are consistent with all or most of the body mass difference between the decades being attributable to a decline in fat reserves.
The slight though statistically significant differences in mean culmen length and mean wing length between the decades are likely attributable to differences in details of the measurement (e.g. personnel, instruments, calibration) or sampling (e.g. habitat) procedures (cf. [28
]). In a sample of Pacific dunlins collected on the Fraser River estuary in 1992-1995 (i.e. between the samples reported here; see [31
]), means for both culmen (2.18%, relative to 1970s sample) and winglength (1.59%) are slightly larger than either sample reported on here. This indicates that the differences between samples do not represent an ongoing change in the size composition of the population, and supports the interpretation that the changes are attributable to minor procedural differences.
The seasonal pattern of mass decline between the decades shows a striking correlation with the seasonal pattern of peregrine occurrence on the Fraser estuary. The biggest mass drop coincides with the autumn peak in peregrine abundance, with the difference shrinking as the peregrine index declines during the winter. Mass in the decades is the same in March, when peregrines are at their annual (near zero) low. This correlation supports our hypothesis that Pacific dunlins have shifted emphasis from protection against starvation to protection against predation. Piersma et al. [9
] reached a similar conclusion in their analysis of changes in the mid-winter masses of golden plovers in The Netherlands. As the data are correlational, we must be cautious about inferring causation, or ruling out contributing roles for any of the many other factors that must have changed at this location between these decades.
Another hypothesis to explain the body mass reduction is that starvation risk has declined over recent decades as climate change reduced the severity of winter weather. To evaluate this we obtained weather records for the period 1970 - 2007 from the National Climate Data and Information Archive of Environment Canada. In Table we summarize the rate of change in average daily temperature, total precipitation, and maximum wind gust on the study area over the winter months. These data provide weak support at best for the idea that winter weather has become less severe. Though the change in mean daily temperature is slightly positive for each winter month over this period, ranging from 0.01 to 0.07°C y-1, neither the change in total monthly precipitation nor gustiness are consistent in direction across months, and the low r2 values in Table indicate that all three metrics are very noisy. In fact, the total net change over the 37 year record is much smaller than most of the year to year changes recorded.
Climate change during winter on the Fraser estuary
Neither does the seasonal pattern of change in body mass between the decades match very closely the pattern in those climate measures that do show evidence of change. The biggest seasonal mass change between the 1970s and 1990s occurs in the autumn, with the difference shrinking until March, when masses are the same. In contrast January shows the strongest rate of increase in temperature since 1970, followed by March. Changes in precipitation and wind match even less well. Overall, we feel that the reduction in winter body mass of Pacific dunlins is not well-explained by an hypothesis based on reduced winter severity. Neither is this hypothesis able to explain why over-ocean flocking has become commonplace.
We regard the changes in roosting behavior and body mass as adaptive adjustments to increased danger. Other interpretations are that the body mass decline is a non-strategic consequence of extra flight induced by harassment from numerous peregrines, or that it is a strategic reduction of body mass made to reduce flight costs because flight time has increased for a reason unrelated to predation danger. Differentiating between these competing hypotheses could be undertaken with an analysis of when strategic over-ocean flocking (or more generally, roosting behavior) ought to occur. The first study to make a strategic analysis of roost site choice is that by Rogers et al. [13
]. In their study area, distant roost sites required more travel than nearby roost sites, but birds suffered fewer disturbances there. They accounted for roost choice with a model that minimized total energy expenditure over the entire high water period, summing the energetic costs of both travel to the roost site and time spent in flight while there. We agree with this general approach, but feel that the choice is more appropriately analyzed in terms of maximizing survival than in minimizing energy expenditure. Over-ocean flocking becomes worthwhile when it reduces the probability of mortality [20
], taking account of the extra mortality that results from the extra foraging required.
Several factors may dispose dunlins on the Fraser estuary to more prolonged and frequent over-ocean flocking than at other sites. First, the winter population of peregrines is high relative to other locales, and the danger they pose to dunlins wintering there is particularly great, because kleptoparasitic competition from Bald Eagles (Haliaeetus leucocephalus
) forces peregrines to concentrate their foraging on dunlins instead of ducks, a favorite prey for peregrines elsewhere [32
]. Second, the nature of the Fraser River estuary is such that alternative coastal and inland roosting sites are also dangerous, due to the presence of peregrines, merlins (Falco columbarius
) and northern harriers (Circus cyaneus
), all of which prey on dunlins [18
]. Finally, the Fraser estuary is large and highly productive, and the winter in southwest British Columbia generally mild, which makes it possible to finance the extra energetic expenditure required. Dunlins cease over-ocean flocking during extended periods of freezing weather [18
], on days with heavy rain (see Fig. ) and on windless days, which suggests that increases in the energy requirement (freezing weather, rain) or the cost of the gliding and hovering mode of OOF flight (lack of wind) make it too expensive. Analyzing over-ocean flocking on the Fraser estuary and on other wintering areas in this way should lead to further understanding of when and why it is observed.