The relationship between weather and wader distributions on non-estuarine coasts is described and used to predict how future wader distributions may respond to climate change. The distributions of eight out of nine species of wader commonly wintering on the non-estuarine coasts of Britain altered between two similar surveys, in 1984/85 and 1997/98, that covered 78% and 38% of Britain's 12 594 km of non-estuarine coastline, respectively. These eight species moved at least in part either eastwards along the winter isotherms or northwards. These changes in distribution broadly coincide with a distributional shift towards the species’ respective breeding grounds and are correlated with the local winter weather over the period: increasingly mild extreme temperatures and changes in mean rainfall, mean wind speed and wind-chill. Based on the scenarios for Britain's climate in 2020 and 2080, it is predicted that the distributions of the waders will move away from the west. The non-estuarine coasts of Britain hold particularly high proportions of the international flyway populations of Ringed Plover Charadrius hiaticula, Sanderling Calidris alba, Purple Sandpiper Calidris maritima and Ruddy Turnstone Arenaria interpres that are all expected to show continuing decline to 2080. Overwintering waders appear to be good indicators of the effects of climate change.
There is little doubt that change in weather over decades has affected a wide range of biota (Parmesan & Yohe 2003, Root et al. 2003). Crick (2004) and Rehfisch and Crick (2003) review the phenological, demographic and range changes that have been recorded in birds, with recent climatic change. There is evidence of differential population change in the internationally important numbers of waders wintering in Britain's estuaries according to region between the 1970s and 1990s (Austin et al. 2000). Wader distributions mostly moved eastwards as winters became milder between the mid 1980s and late 1990s (Austin & Rehfisch in press), possibly as a consequence of the effects of weather on the waders. Rain, low temperatures and high winds lead to increased metabolic rates and heat loss in waders (Piersma 1994, Wiersma & Piersma 1994) and to their invertebrate prey becoming less available due to burying themselves more deeply in the sediments and being less active (Evans 1979, Pienkowski 1983, Zwarts & Wanink 1993).
Britain's 12 594 km of non-estuarine coastline holds internationally important numbers of 11 species of overwintering wader (Rehfisch et al. 2003b). (Holding 1% of a particular flyway population of waterbird qualifies a geographical area as being internationally important for the species. In this instance, the East Atlantic Flyway covers the Atlantic seaboard of Europe and Africa.) The only two nationwide surveys of the waders that winter along this coastline, the 1984/85 Winter Shorebird Count (WSC) (Moser & Summers 1987) and the 1997/98 Non-estuarine Coastal Waterfowl Survey (NEWS) (Rehfisch et al. 2003b), show that the distribution of waders has changed.
Using WSC and NEWS data the relationship between weather and non-estuarine wader distributions, defined here as site-specific wader numbers, is quantified in order to describe a relationship that is of interest in its own right and to predict how changes in weather under the UKCIP98 (UK Climate Impacts Programme) scenarios (Hulme & Jenkins 1998) could affect wader distributions by 2020 and 2080 on Britain's non-estuarine coasts. Although such predictions must be treated with caution (see Discussion), they can be useful indicators of the need for conservation action and the kinds of action that may be appropriate. The analyses are limited to seven species listed in Table 1 with 7–41% of their flyway population present on Britain's non-estuarine coasts (Rehfisch et al. 2003a, 2003b), together with Northern Lapwing Vanellus vanellus and Dunlin Calidris alpina that were encountered on over 10% of the count stretches during each survey.
Table 1. Models relating changes in wader numbers between 1984/85 and 1997/98 to geographical position.
Survey × Longitude Eastings 100 km
Survey × Latitude Northings 100 km
Survey × Longitude × Latitude (100 km)2
These models assume Poisson error terms and have used P-scaling to deal with the over-dispersion of the data. In each data cell the parameter value is followed by its standard error, χ2 and significance level. (Longitude and latitude are based on the British National Grid in metres, where 0 is the bottom corner of 100-km square SV southwest of The Scillies. The range of eastings (longitude) is 69–656 km and northings (latitude) 7–1217 km).
NEWS count methodology is detailed in Rehfisch et al. (2003b). It broadly followed WSC methodology (Moser & Summers 1987). Waders on each count section were counted once between 1 December and 31 January. The 1984/85 WSC and 1997/98 NEWS covered, respectively, 78% and 38% of Britain's non-estuarine coastline.
Daily data for the weather variables known to affect the metabolic rates of waterbirds and the availability of their invertebrate prey were obtained from the British Atmospheric Data Centre (http://www.badc.rl.ac.uk) for meteorological recording stations within 5 km of Britain's coast in 34 regions, loosely based on coastal counties. Average values, November to February inclusive, for each region were calculated for minimum daily temperature (°C), rainfall (mm/day), wind speed 10 m above ground level (knots) and wind-chill [wind speed/(minimum daily temperature + 10 °C)]. Wind-chill is an indication of the risk of exposure faced by the waders (Wiersma & Piersma 1994). To ensure that all wind-chill values remained positive within the range of temperatures considered (−4.3 to 9.7 °C), 10 °C was added to temperature throughout. Temperature, rainfall and wind speed helped to explain the change in estuarine wader distributions (Rehfisch & Crick 2003) and are included in the 2020 and 2080 UKCIP98 predictions (Hulme & Jenkins 1998). Weather data for a region were used only if available for at least three out of five winters preceding the counts. The data for February 1985 and February 1997 were excluded as the two wader surveys took place in January 1985 and January 1997. The weather data were averaged for each of the two time periods, 1980/81–1984/85 and 1993/94–1997/98, for incorporation into the models.
Wader counts in each survey were assumed to follow a Poisson distribution, and related to covariates via a generalized linear model (McCullagh & Nelder 1989) and a logarithmic link function.
Two sets of covariates were separately built into the model, to test for (a) geographical variation in the proportional change between surveys, and (b) any relationship between wader numbers and local conditions. Thus, if cij is the count of a species at site i in survey j, we have the following models:
where si is a site-specific effect allowing for differences in the quality of individual sections due to size, habitat, disturbance, etc.; α is the survey effect representing proportional change in the number of birds recorded between the surveys; xi is a vector of three geographical variables (longitude, latitude and longitude × latitude) determined by the location of site i; and yij is a vector of five weather measurements (the four given above and the squared values of wind-chill) relevant to site i in survey j.
In both (a) and (b), β is a vector of estimable parameters and µ a model intercept. Variables in x and y were successively removed, in a stepwise fashion beginning with that making the least significant contribution to the model, until all remaining variables were significant at the 5% level. Significant components of x, interactions between location and survey, demonstrate shifts along latitudinal or longitudinal axes in the relative numbers of birds. Significant components of y indicate weather variables correlated with bird numbers.
The problem of over-dispersion caused by a combination of a large number of zero and small counts, and several very high counts, typical of flocking species, was addressed in two stages. First, sections that did not hold a species during both counts were excluded from the analyses, as their habitat may have been unsuitable for the species. Secondly, a scale factor estimated from the square root of Pearson's χ2 statistic divided by its degrees of freedom was employed to adjust standard errors for over-dispersion. Fitted values were calculated to help interpret the effect of the model interaction terms.
Predicting the future distributions of non-estuarine waders
The models developed from the weather variables were used to predict the future wader distributions for the UKCIP98 medium–low (ML) scenarios (ensemble mean of scenarios) for 2020 and 2080 and high (H) scenario (scaled general circulation model) for 2080. The ratios between the numbers of waders predicted for each of these scenarios to the numbers predicted from the 1961–90 baseline data give an indication of the proportional changes in wader numbers that should occur in Britain between the two respective periods if the predicted climate changes are accurate, and if the waders continue to follow their present association with weather.
Identifying changes in wader distribution
There was a significant difference between the numbers of waders recorded by WSC and NEWS (Table 1), although the effect was weak for Sanderling Calidris alba ( = 3.59, P = 0.06), the species that showed no evidence of geographical variation between surveys. Between the two surveys, the proportions of the total Purple Sandpiper Calidris maritima and Ruddy Turnstone Arenaria interpres numbers in Britain increased in the northwest only, that of Ringed Plover Charadrius hiaticula in the east and northeast, those of Lapwing and Common Redshank Tringa totanus in the east and that of Eurasian Curlew Numenius arquata in the east and southeast. The distribution of Eurasian Oystercatcher Haematopus ostralegus changed in a complex manner both northwestwards and southeastwards, whereas that of Dunlin moved away from the southwest.
Describing wader distributions according to weather
Ringed Plover, Lapwing, Curlew, Redshank and Turnstone numbers on particular stretches of the coast during the two surveys decreased with increasing mean rainfall (Table 2). Lapwing and Dunlin numbers increased, whereas Sanderling and Purple Sandpiper numbers decreased with increasing mean minimum temperature. Oystercatcher, Ringed Plover, Purple Sandpiper, Curlew, Redshank and Turnstone numbers were affected by wind-chill. Ringed Plover numbers were positively correlated with wind-chill; the effect of the interactions between temperature and wind speed in the models for the remaining five species can be seen in Figure 1.
Table 2. Models relating wader numbers during 1984/85 and 1997/98 to weather variables.
These models assume Poisson error terms and have used P-scaling to deal with the over-dispersion of the data. In each data cell, the parameter value is followed by its standard error, χ2 and significance level.
Wind-chill = mean wind speed/(mean minimum temperature + 10 °C).
If the UKCIP98 weather scenarios are correct and the waders continue their present associations with weather, between 1961–90 and 2020 ML or 2080 ML and H, Lapwing (increases by a factor of 1.3–1.7, 2.2–3.5 and 2.9–8.9, respectively) and especially Dunlin (Fig. 2) are expected to benefit from the predicted rise in mean minimum temperature, increasing everywhere but especially in southern Britain. By contrast, by 2080 ML and H, Oystercatcher (0.8–1.0 and 0.7–1.0, respectively), Ringed Plover (0.6–0.9 and 0.3–0.7, respectively) and Redshank (0.6–0.9 and 0.4–0.8, respectively) numbers are expected to decline slightly, and those of Sanderling (0.5–0.6 and 0.2–0.4, respectively), Purple Sandpiper (Fig. 3), Curlew (0.3–0.7 and 0.2–0.5, respectively) and Turnstone (0.2–0.7 and 0.1–0.4, respectively) are expected to decrease by over 30%. The largest declines are predicted in the west of Britain.
These analyses would not have been valid had unusually high or low wader numbers been recorded in the UK during the WSC or NEWS winters, possibly as a result of unusual weather. Based on comparisons made between the Wetland Bird Survey (WeBS) indices for 12 species of wader for each survey winter and the two immediately preceding and following, there was no evidence of any such differences (Rehfisch et al. 2003b).
Between 1984/85 and 1997/98, with the exception of Sanderling, all species of wader for which data were analysed changed significantly in number between the surveys and their distributions moved at least in part either eastwards or northwards (Table 1). The estuarine component of the overwintering populations of seven out of nine species studied, including Oystercatcher, Ringed Plover and Dunlin, also showed evidence of similar eastwards shifts along the winter isotherms using a similar analytical approach, but one that only tested for changes in distribution along a longitudinal axis (Austin & Rehfisch in press).
Weather helped to explain the wader distributions in 1984/85 and 1997/98 (Table 2). Rainfall decreases prey availability to waders (Pienkowski 1981, 1983, Selman & Goss-Custard 1988) and thus the decrease in Ringed Plover, Lapwing, Curlew, Redshank and Turnstone numbers with increasing mean rainfall follows expectation (Table 2). Wader numbers would have been expected to decrease with increasing mean wind speed as wader survival may decrease as average winter wind speed increases (Insley et al. 1997), probably as a result of exposure to wind increasing the energy requirements of waders (Wiersma & Piersma 1994) that may be close to their metabolic limits in northern latitudes (Piersma 1994). The relationship between wader numbers on particular stretches of coast and wind speed was complex. Above freezing point the numbers of Oystercatcher and Purple Sandpiper tended to be higher at low wind speeds (Fig. 1). The opposite tended to be true for Ringed Plover, Curlew and Turnstone, whereas Redshank numbers were affected by a complex interaction between wind speed and temperature.
Lapwing and Dunlin are at the northwesterly extreme of their wintering distribution in Britain and Ireland. Lapwing, being particularly intolerant of cold conditions (Kirby & Lack 1993), respond to periods of cold weather by making large-scale westerly movements from the Continent to Britain and from Britain to Ireland. Furthermore, long periods of cold weather can lead to major mortality incidents in waders and decrease their year-to-year survival rates (Clark 1982, Insley et al. 1997), probably as a consequence of increased heat loss and energy requirements at a time when invertebrate prey may be less available (Evans 1979, Pienkowski 1983, Zwarts & Wanink 1993). Thus, the increase in Lapwing and Dunlin numbers with increasing mean minimum temperature follows expectation (Table 2). The increase in Oystercatcher, Ringed Plover, Sanderling and Purple Sandpiper numbers with decreasing mean minimum temperature is unexpected. This may be because their rocky shore invertebrate prey is not usually made less accessible by cold weather, unlike invertebrates living in soft sediments. Turnstone numbers, a wader that finds its prey by probing under rocks and algae, were hardly affected by mean minimum temperature. Again, this species’ specialized approach to feeding is unlikely to be affected by the cold.
These results provide broad support for the idea that waders tend to winter closer to their breeding grounds as winters have become milder, in this instance by about 1.5 °C, between 1984/85 and 1997/98. The distributional shifts in the British wintering grounds of Ringed Plovers (towards Scandinavian/Baltic breeding grounds), Lapwings (towards breeding grounds in Scandinavia), Purple Sandpipers (towards breeding grounds in Canada), Dunlins (towards breeding grounds in northern Fenno-Scandia/northern Russia) and Turnstones (towards breeding grounds in northeast Canada/Greenland) fit the hypothesis (Table 3). The observed shifts for Redshanks and Curlews are towards some of their breeding grounds but are less clear cut. The exact boundaries on the British wintering grounds of the three separate breeding populations of Oystercatcher are unclear (Cramp & Simmons 1982). Although lack of detailed information clearly limits interpretation of these results, the complex change in distribution of Oystercatchers may reflect a tendency for the geographically segregated subpopulation wintering in north-western Britain to be moving northwest towards breeding grounds in Iceland, the Faeroes and northern Britain, and for that population wintering in southeastern Britain to be moving southeast towards breeding grounds either in southern Britain or in The Netherlands.
Table 3. The main breeding and wintering areas of the populations of the studied waders (Cramp & Simmons 1982). Breeding grounds in bold type fit the breeding ground distributional shift hypothesis.
Breeding grounds relative to wintering grounds
Direction of distributional shift of wintering non-estuarine coast populations*
Change in distribution between WSC and NEWS: these analyses.
complex: NW and SE
W Britain/English Channel
Siberia (few Greenland)
W Britain (86% Scotland's birds)
E Britain (14% Scotland's birds)
Britain (60% of Iceland's birds)
How future climate may affect wader distributions
The UKCIP98 climate change scenarios provide predictions for future winter precipitation, diurnal temperature range and mean wind speed in each 10-km square in Britain in 2020, 2050 and 2080 (Hulme & Jenkins 1998). For example, under the 2080 Medium–High scenario, it is predicted that compared with the 1961–90 baseline, mean winter temperature will increase by on average 2.5–3 °C, precipitation will increase by 18–22% and wind speed hardly change (−1 to 1%).
If the waders maintain their present relationship between weather and distribution, the numbers of Sanderling, Purple Sandpiper (Fig. 3) and Turnstone overwintering in Britain should decrease everywhere between 1961–90 and 2080. Based on the conservative 2080 ML scenario, even in their strongholds of the Western Isles, Orkney and Shetland, where they are expected to decline on average less than elsewhere, their numbers may decrease by 35–60%. This would follow the 16–21% decline in the numbers of these species on British non-estuarine coasts recorded between 1984/85 and 1997/98 (Rehfisch et al. 2003b); if correct, and if populations are maintained elsewhere, this prediction implies that Britain should hold less than the 9–53% of their flyway populations recorded in 1997/98 (Rehfisch et al. 2003a). Curlew numbers are expected to decline by over 40% in its strongholds of Shetland and Orkney, and Ringed Plover numbers may decline by up to 36% in the Western Isles, which held 41% of its British population in 1997/98 (Rehfisch et al. 2003b). The predicted declines in Oystercatcher and Redshank numbers by 2080 are relatively minor. As a result of the predicted increase in winter temperature by 2080, numbers of Lapwing and Dunlin (Fig. 2) should increase everywhere but especially in southeast England. However, the predicted reduced intertidal area in southeast England resulting from increased relative sea-level in conjunction with coastal squeeze should make conditions less suitable for these birds (Austin & Rehfisch 2003).
CONCLUSIONS AND CONSERVATION IMPLICATIONS
Changes in population size could be giving the impression of a shift in distribution as the species’ ranges expand or contract (Thomas & Lennon 1999, Tryjanowski & Sparks 2001). It is possible, for example, that decreasing populations of Ringed Plover, Sanderling, Purple Sandpiper and Turnstone could be giving the appearance of northwards movement simply as a result of birds retreating into their core wintering range. The following examples show that this is unlikely. Ringed Plovers do not traditionally winter to the northeast of Britain, the direction of its predominant change in distribution. Dunlin and Redshank numbers on non-estuarine coasts have increased and yet their distributions are also moving to the northeast and east, respectively. The shifts of six of the eight species that have changed distribution include an easterly component in the direction of the winter isotherms that run along the longitudinal axis in northwest Europe. This would be as expected if the winter distributions of the waders were being affected by weather. Analyses in 2004/05 of wader count data from northwest Europe should help confirm that the birds are moving outside of Britain.
This is the first paper to demonstrate that the large-scale change in the winter distributions of several species of internationally important waders on non-estuarine coasts can be related to changes in climate. The evidence from the waders on the non-estuarine coast fits the hypothesis that the waders are wintering closer to their breeding grounds, which benefits them by allowing earlier return to breeding grounds, thereby facilitating the attainment of better territories and higher reproductive success (Kokko 1999).
These analyses do not make it possible to determine whether the predicted changes in temperature resulting from climate change will lead to changes in total flyway population size, but they demonstrate that it is likely that there will be changes in wader distribution. The tentative predictions made in this paper of the future distributions and numbers of waders on Britain's non-estuarine coast show that Britain is likely to hold many fewer Ringed Plovers, Sanderlings, Purple Sandpipers and Turnstones, the species that are found predominantly on this habitat, although some other species may become more common if their flyway populations are large enough. This change is expected to occur as a greater proportion of the East Atlantic Flyway populations of these, and perhaps other, waders start to overwinter in more northerly and easterly areas of continental Europe as they become suitable wintering grounds with increasing winter temperatures. Areas including Iceland, Scandinavia, the Waddensee and even the Baltic Sea, which are closer to the wader breeding grounds, should become increasingly attractive as the risk of cold weather mortality decreases with increasing minimum winter temperature. However, there are caveats. As wader food intake rates tend to be lower at night (Sitters 2000) and as northern areas of the Northern Hemisphere have fewer hours of daylight during the winter, this may be a disincentive to waders wintering there. The predictions may underestimate the size of the overwintering wader decline in Britain if their total populations decline. If the increase in precipitation projected for Britain (Hulme & Jenkins 1998) is widespread within Europe it could lead to smaller future flyway populations of Ringed Plover, Lapwing, Curlew, Redshank and Turnstone (Table 2), but perhaps the most likely causes of wader population decline are changes to wader habitat brought about by human activities, including landclaim and shellfisheries (Piersma & Lindström 2004), and the effects of climate change on wader breeding grounds (Rehfisch & Crick 2003). As Knogge et al. (2004) state, fauna tend not to rank very highly against our economic interests.
The results of the analyses reported in this paper have implications for the conservation of waders in Britain and probably beyond. First, as Britain holds 9–53% of the flyway populations of Ringed Plover, Sanderling, Purple Sandpiper and Turnstone, it is urgent to confirm that their apparent decline is due to redistribution into other wintering grounds, rather than a flyway population decrease. If the 15–21% decrease in their British population size between 1984/85 and 1997/98 (Rehfisch et al. 2003b) is due to the latter then its cause needs to be elucidated rapidly to allow conservation action to be implemented. Any population decline could be due to the changes in winter weather observed over the last two decades or declining organic inputs into coastal waters (Burton et al. 2002) that may be making Britain and elsewhere less suitable for the relevant species.
Secondly, wader conservation policy has often been based on protecting a set of sites that have been designated for holding nationally or internationally important numbers of overwintering waders or other waterbirds. If the waders’ distributions continue to change rapidly, the rationale for fixed site conservation needs to be developed further (Rehfisch 2000, Boere & Taylor 2004).
Thirdly, climate change will not only affect waders on their wintering grounds, but it will also affect them on their northerly breeding grounds and on staging posts used during migration (Rehfisch & Crick 2003). A flyway approach towards determining where the major climate change-induced pressures will act is recommended. This would help to target research and conservation work towards the components of the waders’ life cycle that are likely to be most affected by climate change, and if necessary to start the process of collecting the information required to parameterize models that would make it possible to estimate any likely conservation implications (but see Piersma & Lindström 2004).
It is probably wise to end on a pessimistic note. Although, like other fauna, waders have historically had to adapt to changes in weather, they may not be behaviourally plastic enough to be able to adapt to a rate of change in weather over the next decades that is expected to be far greater than anything previously encountered according to the geological record. Even if these birds had an unlimited capacity for changing their wintering distributions, they might be limited by the capacity of their invertebrate prey and preferred intertidal habitat to keep up with the rapidly changing weather and rising sea-levels associated with climate change. The adaptability of the waders will be further tested by changes to their breeding grounds as the arctic thaws and the tundra starts to be colonized by scrub, invaded by new species of fauna that may compete with them for breeding space and present a greater variety of predators, as well as change their highly abundant but relatively species-poor prey community (Rehfisch & Crick 2003). There is no evidence of such plasticity in the much-studied migratory behaviour of Red Knot Calidris canutus (Piersma & Lindström 2004).
The study would not have been possible without the combined efforts of the counters who supply the data to WeBS. The British Atmospheric Data Centre allowed access to their weather data and the UKCIP allowed their baseline data and predictions to be used for research purposes. Thanks are due to Chris Feare, Heidi Mellan and Ron Summers for their comments and help during the preparation of this manuscript. This work was funded by WeBS, a partnership of the BTO, WWT, RSPB and JNCC (on behalf of EN, SNH, CCW & EHS(NI)).