Impacts of sudden winter habitat loss on the body condition and survival of redshank Tringa totanus

Authors


N. H. K. Burton, British Trust for Ornithology, The Nunnery, Thetford, Norfolk IP24 2 PU, UK (fax +44 1842 750030; e-mail niall.burton@bto.org).

Summary

  • 1Recent theoretical modelling has provided important insights into how habitat loss may affect local populations through impacts on individual fitness (survival, body condition, fecundity). Despite this, attempts to provide empirical evidence of such impacts on displaced individuals have been limited. Using a before-after-control-impact (BACI) approach, we report how a sudden loss of wintering habitat impacted on the body condition and survival of redshank Tringa totanus.
  • 2The intertidal mudflats of Cardiff Bay, UK, were inundated with freshwater in November 1999 following impoundment by a barrage, resulting in the displacement of c. 300 redshank to adjacent habitat on the Severn Estuary. Movements and the survival of these birds were monitored through observations of colour-marked individuals. Comparative survival rates were calculated for marked populations at the main recipient site, Rhymney, and a control site.
  • 3Displaced redshank had difficulty maintaining their mass in the first winter post-barrage closure: adults previously only recorded at Cardiff Bay were significantly lighter than those previously recorded at Rhymney.
  • 4Survival rates of displaced redshank also declined. The estimated annual survival of adult Cardiff Bay redshank fell from 0·846 in the 2 years pre-barrage closure to 0·778 in the 3 following years because of a significant decline in winter survival (P = 0·0006). In comparison, there was no significant change in the survival of adult Rhymney redshank, and adult survival at the control site was actually greater post-barrage closure than beforehand. The lack of decline in these rates and the similarity between those of Cardiff Bay adults pre-barrage closure and Rhymney adults indicate that the increase in winter mortality of Cardiff Bay birds resulted from their displacement.
  • 5Synthesis and applications. This study provides the first conclusive empirical evidence that habitat loss can impact individual fitness in a bird population. Adult redshank displaced from Cardiff Bay experienced poor body condition and a 44% increase in mortality rate. Without an increase in the recruitment of first-winter birds, such a change is likely to reduce substantially local population size. The results reported here should help to inform governments, planners and non-governmental organizations (NGOs) seeking to understand how developments might impact on animal populations.

Introduction

Habitat loss represents one of the major threats to conservation of biodiversity world-wide. On estuaries, large areas of intertidal mudflats and saltmarsh have been lost to land claim and the remaining habitat is now threatened by sea-level rise (Davidson et al. 1991; Austin & Rehfisch 2003). Given the importance of estuaries for both migrating and wintering waterbirds (Stroud et al. 2001), the problems presented by habitat loss have been the subject of considerable study. However, although theoretical modelling has emphasized the importance of studying individuals in evaluating impacts on local populations (Goss-Custard et al. 1994, 2002; Sutherland 1996; Goss-Custard 2003), there has been little empirical evidence of costs for individual fitness (i.e. survival, body condition, fecundity).

Most applied studies of the impacts of habitat loss on local populations of estuarine waterbirds have been based largely on count data (Evans 1978–79; Evans, Knights & Pienkowski 1979; Laursen, Gram & Alberto 1981; Lambeck, Sandee & de Wolf 1989; Meire 1991, 1996; McLusky, Bryant & Elliott 1992; Schekkerman, Meininger & Meire 1994; Hötker 1997). This provides information on how birds redistribute themselves or change their behaviour; Evans (1978–79), for example, reported that, following the loss of an area of intertidal mudflats, redshank Tringa totanus (Linnaeus) were forced to use supratidal feeding areas over high tide in order to meet their daily food requirements. However, such studies may be limited in establishing impacts on survival and thus understanding effects on populations.

To determine more clearly the fates of animals displaced by habitat loss, it is necessary to mark individuals. Two recent studies of estuarine habitat loss have used ringing to follow birds’ movements; in both cases, however, attempts to measure survival were methodologically limited. Lambeck (1991) used colour-ringing to determine where oystercatchers Haematopus ostralegus (Linnaeus) were displaced to following large-scale habitat loss in the Delta region of The Netherlands. Recoveries of dead birds suggested that displaced birds had experienced significantly higher mortality than those originally ringed elsewhere in the Delta, presumably as a result of the increased densities in recipient areas. Approximately half the oystercatchers from the former intertidal areas disappeared following the loss of habitat.

Colour-ringing was also used to determine the movements of dark-bellied brent geese Branta bernicla bernicla (Linnaeus) following the loss of saltmarsh at Rodenäs Vorland on the German Wadden Sea (Ganter & Ebbinge 1997; Ganter, Prokosch & Ebbinge 1997). This study revealed that, following the loss of habitat, long-distance movements were more frequent among displaced birds than control birds and that many displaced birds moved to less preferred sites that were apparently below their carrying capacity for the species and thus more able to support increased densities. Resighting rates suggested no difference in the survival rates of displaced and control birds, although sample sizes were small.

The present study reports the effects of a sudden loss of c. 200 ha of intertidal habitat at Cardiff Bay on the Welsh shore of the Severn Estuary, UK, following impoundment by a barrage and conversion to a freshwater lake in November 1999. Prior to its impoundment, the bay supported up to 300 wintering redshank (Burton, Rehfisch & Clark 2003), 49% of which had been individually colour-marked by October 1999. By observing these individuals, we were able to monitor this population through its displacement and, by using a before-after-control-impact (BACI) approach (Stewart-Oaten, Murdoch & Parker 1986), we investigated whether the loss of habitat impacted their survival. Additionally, we investigated whether the body condition of redshank displaced from the bay was adversely affected following barrage closure. To our knowledge, this is the first study to use standard mark–recapture analyses to investigate how habitat loss might impact survival in a bird population.

Methods

study species and site

Redshank winter in internationally important numbers on the Severn Estuary (Collier et al. 2005), favouring river mouths and other sites where there are freshwater inputs. Cardiff Bay (hereafter also the bay) (51°27′ N, 3°10′ W), which stands at the mouth of the Severn, was formed by the combined estuaries of the Rivers Taff and Ely and pre-barrage closure encompassed both intertidal mudflats and saltmarsh (Fig. 1).

Figure 1.

Main study area showing the intertidal mudflats of Cardiff Bay pre-barrage closure, the neighbouring Rhymney Estuary mudflats and their position on the Severn Estuary, UK. Asterisks indicate locations where colour-ringed redshank were recorded in winters post-barrage closure; the whole area shown was searched for colour-ringed individuals.

The mudflats of the Rhymney Estuary (c. 450 ha; hereafter Rhymney) (51°29′ N, 3°07′ W), 4 km to the north-east along the Severn (Fig. 1), supported, on average, c. 50% more redshank than Cardiff Bay in pre-barrage closure winters (Burton 2000a; Burton, Rehfisch & Clark 2003). On the Welsh shore of the Severn, further concentrations of redshank are found where the Rivers Usk, Ebbw and Wye flow into the estuary; on the estuary's English shore, large numbers are found at the mouths of the Rivers Parrett, Brue, Axe and Avon (Burton et al. 2003).

changes in densities

Numbers of redshank were monitored at Cardiff Bay and Rhymney from November 1989 and continued between autumn and spring each year until March 2003. Each site was divided into a number of count units and the birds on each unit were counted hourly (from 6 h before to 5 h after low tide) through two complete tidal cycles per month (exception for April, when only a single tidal cycle was covered).

Generalized linear models (GLMs) (proc genmod; SAS Institute Inc. 1999–2001) were used to examine how densities of feeding birds changed on each site over the 10 years before barrage closure and whether densities at Rhymney increased after the displacement of birds from the bay. Densities of feeding birds on each count (birds ha−1) were related to the year, month (August–April), tidal state (hour relative to low water at which the count was undertaken), mudflat count section and interaction between tidal state and mudflat. Models assumed a Poisson distribution for the number of birds, specified a log link function, treated all variables as categorical and used the natural logarithm of mudflat area (ha) as an offset. The problem of overdispersion caused by a combination of a large number of zero counts and several very high counts, typical of flocking species, was addressed by applying a scale factor estimated from the square root of the Pearson's chi-squared statistic divided by its degrees of freedom. Only those variables that were significant at the 5% level in explaining the variation in densities were retained in the final models.

marking and biometrics

Ringing activities began in January 1991 and continued throughout the study. Redshank were caught by cannon- or mist-netting at high tide roosts both within Cardiff Bay and at Rhymney. Each bird was aged, according to its plumage characteristics (Prater, Marchant & Vuorinen 1977), as either adult or first-year and fitted with a metal BTO (British Trust for Ornithology) ring. To help analyse variation in the birds’ body condition, the following measurements were taken: mass (g), wing length (maximum chord), foot length and bill length (to precisions of 1 mm, 0·5 mm and 0·1 mm, respectively) (Summers et al. 1988). The majority of measurements were taken by one observer.

Ringing data were also collated from five sites (within 7·3 km of each other; referred to collectively as the control site) in north Wales: Bangor Harbour, Llanfairfechan, Wig and two on the Ogwen Estuary (53°14′ N, 4°07′ W−53°15′ N, 4°00′ W, 205 km north-north-west of Cardiff). These provided control data for use in the survival analyses. As above, redshank were aged as either adult or first-year and ringed.

From November 1994 to October 1999, the majority of redshank caught at Cardiff Bay and some of those originally metal-ringed in the bay and then retrapped at Rhymney were given unique combinations of colour-rings so that they could be subsequently identified in the field. In total, 454 birds were individually colour-ringed, 396 in the bay (322 adults, 69 first-winter birds and five birds of indeterminate age) and 58 adults caught at Rhymney (39 of which had previously been caught in the bay).

analysis of redshank body condition

To determine whether displacement from Cardiff Bay adversely affected the body condition of the redshank that wintered there, individuals were first classified into three groups, those that had only ever been caught (or, having been colour-ringed, seen) at Cardiff Bay prior to barrage closure, those that had only ever been caught (or seen) at Rhymney and those that had been recorded at both sites. Comparison was then made of the masses of adults from each of these groups in winter 1999–2000, the first following barrage closure, using a GLM (proc glm; SAS Institute Inc. 1999–2001). The model controlled for month and measures of body size (wing length, foot length and bill length) and only used one observation for each bird. Previous work has shown that the redshank that formerly wintered in the bay were predominantly from the British breeding population, while the concentration found at Rhymney included a significantly larger proportion of birds from the Icelandic population (Burton et al. 2002). Thus Cardiff Bay redshank tended to have shorter wings and feet but longer bills (Summers et al. 1988). The model assumed a normal distribution and treated month and site group as class variables. Only those variables that were significant in explaining the variation in mass were retained in the final model. Likelihood ratio tests (LRTs) were used to investigate differences in the masses of redshank from each group.

survival analyses

Estimates of survival rates and recapture probabilities of adult redshank were calculated using mark–recapture methods. Survival rates were calculated separately for (i) birds originally caught and ringed at Cardiff Bay, using information from resightings of colour-ringed birds (pre-closure survival rates were also estimated independently using information from recaptures of metal-ringed birds); (ii) birds originally ringed at Rhymney, using information from recaptures of metal-ringed birds; and (iii) birds from the north Wales control site, also using information from recaptures of metal-ringed birds. Impacts were predicted to occur for adult redshank from Cardiff Bay, although potentially also for adults from Rhymney as it was expected that redshank densities might increase here after the displacement of birds from the bay. In each case, analyses aimed to determine whether survival rates differed in periods before and after barrage closure.

Survival rates of adult redshank from Cardiff Bay

To determine the survival rates of Cardiff Bay redshank, the bay and other areas of the wider Severn used by wintering redshank were searched extensively for colour-ringed birds (originally caught and ringed in the bay) twice a year, in February and October, from February 1996 to February 2003. The use of data from resightings of colour-ringed birds was preferred to using data from the recapture of ringed birds, as it provided higher resighting (i.e. recapture) rates (Bearhop, Ward & Evans 2003; Sandercock 2003; Robinson et al. 2005). Importantly, it also allowed the inclusion of data from sites away from the bay and Rhymney, and thus for the movement of birds displaced by barrage closure. This allowed for better estimation of true survival, rather than just the measure of survival and site fidelity to the Cardiff/Rhymney area (Sandercock 2003).

Data were only included for birds originally caught and ringed in Cardiff Bay and were restricted to those first caught, or subsequently seen, as adults in one of the February or October resighting periods. There were insufficient data to look at the survival rates of first-winter redshank, notably over the year following barrage closure, although data from birds originally ringed in the bay as first-winter birds but subsequently resighted (in October or February) as adults were included from when they were first seen as adults. As no new birds were captured in the bay following barrage closure (i.e. from 1999–2000 onwards) and because we wanted to measure the impact of the closure on birds that had formerly used the bay, the data set only contained resighting data for individuals first encountered as adults between February 1996 and October 1999.

Resighting data were analysed using Program mark (White & Burnham 1999). Cormack–Jolly-Seber (CJS) models (Seber 1982; Lebreton et al. 1992) were developed to estimate monthly survival rates (φm) (i.e. the proportion of birds surviving a month within a period) and resighting probabilities (p).

The validity of the CJS models depended on the assumption that all birds had an equal chance of being recaptured or resighted (Lebreton et al. 1992). To satisfy this assumption, models only considered individuals colour-ringed from October 1995, as earlier analysis revealed that birds marked using a previous colour-ringing scheme were identified less frequently because of the positions of rings (Burton 2000a, b). In spite of this, initial goodness-of-fit tests (of the model in which φm and p were assumed to vary with time) indicated that this assumption was still not valid (see Table 1 for results of all goodness-of-fit tests). This problem was not apparent in an earlier analysis of pre-barrage closure survival (Burton 2000a) but was thought to be because in the first year of data collection the River Ely was not surveyed for colour-ringed birds. Thus birds that were largely resident there at this time were subsequently seen to have had a lower chance of being resighted than those resident in the main part of the bay, and this would have biased survival estimates for that year. Further tests indicated that the exclusion of data from February and October 1996 was sufficient to overcome this violation (Table 1).

Table 1.  Results of goodness-of-fit tests carried out on adult redshank mark–recapture data. Tests from the program release (run through Program mark) check the validity of the CJS model. Test 3 checks whether previous capture history affects the future probability of survival or recapture; test 2 checks that survival rates and recapture probabilities are similar for different cohorts of birds (Burnham et al. 1987)
SiteDatesTest 2Test 3
Cardiff Bay (colour-ringed)February 1996 to February 2003inline image = 36·92, P = 0·0002inline image = 14·46, NS
Cardiff Bay (colour-ringed)February 1997 to February 2003inline image = 15·84, NSinline image = 12·69, NS
Cardiff Bay (metal-ringed)1990–91 to 1998–99inline image = 7·67, NSinline image = 9·34, NS
Rhymney (metal-ringed)1991–92 to 2002–03inline image = 8·12, NSinline image = 2·27, NS
North Wales (metal-ringed)1988–89 to 2002–03inline image = 26·88, NSinline image = 28·26, NS

In both this and subsequent investigations, a combination of LRTs and Akaike's information criterion (AIC), adjusted for overdispersion and sample size (QAICc; Burnham & Anderson 1998; White & Burnham 1999), was used to select the model that best described the data (typically that with the lowest QAICc value). Reduced models looked at whether p varied with time, either fully across every recapture period, between the October and February recapture periods or before and after barrage closure. Similarly, the models also investigated whether φm varied fully with time, between summer (February–October) and winter (October–February) or before and after barrage closure. The final chosen model was then used to calculate annual survival rates (φ) for comparison with those estimated for adults from Rhymney and north Wales.

To evaluate the accuracy of the pre-closure survival rates calculated from the resightings of colour-ringed birds, data from the recapture of metal-ringed birds were also analysed to provide independent estimates. Ring-recapture data (for redshank originally caught and ringed in the bay) were taken from catches undertaken at both Rhymney and Cardiff Bay between January 1991 and February 1999 within the defined winter period (October–February). Using Program mark, models were developed to describe annual survival rates and resighting probabilities and determine whether or not these parameters varied with time.

Survival rates of adult redshank from Rhymney

Survival rates were calculated for those adult redshank originally caught and ringed at Rhymney using ring-recapture data from catches undertaken at both Rhymney and Cardiff Bay between December 1991 and February 2003. Again, data were only included from catches undertaken between October and February each winter. In part this was to match the winter period defined for the survival analyses of Cardiff Bay birds, but also to exclude birds only caught on autumn and spring passage. Data were only included for individuals first encountered as adults between winter 1991–92 and October 1999. This was first in order to exclude data from redshank that had been displaced from the bay and, secondly, because we wanted to measure the impact of this possible influx of birds on those that had used Rhymney prior to the closure of the barrage. Models investigated whether φ and p varied with time and whether the φ of redshank from Rhymney differed before and after barrage closure.

Survival rates of adult redshank from north Wales

Survival rates were calculated for adult redshank in north Wales using ring-recapture data from catches undertaken between November 1988 and February 2003. As above, data were only included from catches undertaken between October and February each winter. Models investigated whether φ and p varied with time and also whether φ differed in 2000–02 to the 11 years before, to see whether any change in the survival of Cardiff Bay birds after barrage closure might have been simply a reflection of a regional trend, perhaps associated with a change in weather conditions.

As the majority of the redshank that wintered in Cardiff Bay were from the British race (Burton et al. 2002) it was also possible that any change in their survival post-barrage closure could have reflected a trend in this specific population rather than the birds’ displacement. To investigate whether this might have been the case, a further model tested whether the pre- and post-barrage survival of redshank in north Wales differed between British and Icelandic birds (this site held a more even mix of races). Birds were classified as having a greater than 50% probability of being either British or Icelandic, using the formulae described in Summers et al. (1988).

Results

the displacement of redshank from cardiff bay

Prior to 1999, individual redshank were highly faithful to Cardiff Bay, both within and between winters (Burton 2000a). Closure of the barrage displaced all but a few redshank and, thereafter, the bay was only used for roosting. The majority of colour-marked individuals moved to the Rhymney mudflats, just 4 km away. Despite extensive searches, none was recorded more than 19 km distant: 18 individuals were recorded on the River Usk and just one moved across the Severn to the River Axe (N. H. K. Burton & M. J. S. Armitage, unpublished data).

Table 2 summarizes the significance of variables in GLMs describing densities of feeding redshank at Cardiff Bay and Rhymney. Annual indices and mean winter (October–February) numbers derived from these models are shown in Fig. 2. Treating year as a continuous (rather than a class) variable in the models showed that there had been highly significant declines in the densities of feeding redshank at Cardiff Bay (F1,32085 = 58·65, P < 0·0001) and Rhymney (F1,22024 = 140·14, P < 0·0001) over the 10 years prior to barrage closure.

Table 2.  Likelihood ratio statistics and associated probabilities for variables in models describing densities of feeding redshank at Cardiff Bay (1989–90 to 1998–99) and Rhymney (1989–90 to 2002–03)
SiteMonthTidal stateMudflatTidal state × mudflatYear
Cardiff BayF8,33330 = 56·4F11,33330 = 2·8F18,33330 = 44·4F179,33330 = 4·6F10,33330 = 10·6
P < 0·0001P = 0·0013P < 0·0001P < 0·0001P < 0·0001
RhymneyF8,31960 = 123·1F11,31960 = 8·1F16,31960 = 186·3NSF14,31960 = 25·5
P < 0·0001P < 0·0001P < 0·0001 P < 0·0001
Figure 2.

Annual indices (± 1 SE) and mean winter numbers of birds derived from models relating densities of feeding redshank at (a) Cardiff Bay and (b) Rhymney to year, month, mudflat and tidal state. The dotted line indicates the date of barrage closure; the index value for 1999–2000 includes data from November 1999 to May 2000 only.

At Rhymney, densities increased following barrage closure, supporting the evidence from observations of colour-ringed birds that the majority of redshank from the bay was displaced to this site. Although densities were only slightly greater in 1999–2000, the first winter post-closure, than those in 1998–99 (F1,31960 = 3·02, P < 0·10), they rose significantly over each of the following 2 years. Observations of colour-ringed birds suggested that redshank from the bay settled at a number of sites east of Cardiff in 1999–2000 but subsequently became more concentrated at Rhymney (N. H. K. Burton & M. J. S. Armitage, unpublished data). Densities at Rhymney in 2002–03 were thus on average double those in 1998–99 (F1,31960 = 57·12, P < 0·0001), the observed increase of more than 1600 bird hours per tidal cycle more than accounting for the 900 bird hours recorded at Cardiff Bay in 1998–99, the difference being within the range of pre-barrage fluctuations at Rhymney. No overall trend was apparent for the Severn over this period (Maclean et al. 2005).

analysis of redshank body condition

The GLM used to analyse biometric data indicated that the mass of adult redshank in winter 1999–2000 (immediately post-closure) was significantly and positively related to wing length (F1,225 = 37·15, P < 0·0001) and foot length (F1,225 = 41·58, P < 0·0001), but not to bill length (F1,224 = 0·18, NS). Mass was also significantly related to month (F4,225 = 32·51, P < 0·0001) and, notably, the group to which an individual had been categorized (F2,225 = 3·45, P = 0·0336). ‘Cardiff Bay’ redshank were significantly lighter, controlling for body size and month, than those classified as ‘Rhymney’ birds (by a mean of 6·7 g, 4·3% lighter; F1,74 = 5·21, P = 0·0254) and those that had been recorded at both sites prior to barrage closure (by 7·9 g, 5·1% lighter; F1,161 = 6·82, P = 0·0099). Those from the latter two categories did not differ in mass (F1,209 = 0·04, NS).

survival analyses

Survival rates of adult redshank from Cardiff Bay, results from colour-ringed birds

Models describing monthly survival rates and resighting probabilities for adult colour-ringed redshank originally caught and ringed in Cardiff Bay are compared in Table 3. Resighting probabilities were high, equal to 0·862 (SE = 0·011) in the base model in which both φm and p were assumed to be constant over time.

Table 3.  Evaluation of mark–resighting models for adult colour-ringed redshank originally caught and ringed at Cardiff Bay, using data from February 1997 to February 2003. Models evaluated whether resighting rates p were constant (c), varied fully with time (t), between the October and February observation periods (O–F) or pre- and post-closure of the Cardiff Bay barrage (pre–post). Likewise, we also evaluated whether monthly survival rates φm were constant (c), varied fully with time (t), between the summer and winter periods (S–W) or pre- and post-closure (pre–post). Models are ranked according to QAICc value. In this and the following table, bold type indicates the best-fit model (i.e. that with the lowest QAICc value)
ModelQAICcParametersModel deviance
φt pc1730·0513492·35
φc pc1727·48 2512·06
φt pt1720·3324459·95
φc ppre–post1715·23 3497·80
φc pO–F1713·75 3496·32
φc pt1712·7213475·02
φc pO–Fpre–Fpost1708·20 4488·76
φc pOpre–Opost–F1704·73 4485·29
φSpre–Spost–W pOpre–Opost–Fpre–Fpost1701·39 7475·89
φS–W pOpre–Opost–Fpre–Fpost1700·07 6476·60
φc pOpre–Opost–Fpre–Fpost1698·81 5477·36
φpre–post pOpre–Opost–Fpre–Fpost1697·14 6473·66
φS–Wpre–Wpost pOpre–Opost–Fpre–Fpost1690·33 7464·83

Refinement of initial models indicated that, rather than vary fully with time, p varied between October and February and also before and after barrage closure. The most parsimonious model (that with the lowest QAICc value) needed only three survival parameters, indicating that φm was constant between summers (February–October) but differed between the 2 pre-closure and 4 post-closure winters. Monthly summer survival (φS) was estimated to be 0·980 (SE = 0·002). A LRT confirmed a significant drop in monthly winter survival (φW) from 0·998 (SE = 0·003) pre-closure to 0·978 (SE = 0·004) post-closure (inline image = 11·76, P= 0·0006).

Annual survival (φ), the product of whole winter survival (φW4) and over-summer return rates (φS8), was 0·846 (SE = 0·018) for the 2 pre-closure years but fell to 0·778 (SE = 0·019) in the 3 years thereafter; thus there was a 44% increase in long-term annual mortality.

Survival rates of adult redshank from Cardiff Bay, results from metal-ringed birds

Models describing pre-closure annual survival rates and recapture probabilities for adult metal-ringed redshank originally caught and ringed at Cardiff Bay are compared in Table 4a. Recapture probabilities were much lower than the equivalent resighting probabilities for colour-ringed Cardiff Bay redshank, equal to 0·085 (SE = 0·014) in the base model in which both φ and p were assumed to be constant.

Table 4.  Evaluation of mark–resighting models for metal-ringed redshank. Models evaluated whether resighting rates p were constant (c) or varied fully with time (t) and whether annual survival rates φ were constant (c), varied fully with time (t) or between the periods pre- and post-closure of the Cardiff Bay barrage (pre–post)
SiteDatesModelQAICcParametersModel deviance
Cardiff Bay1990–91 to 1998–99φc pc 759·17 2124·55
φt pc 759·06 9110·12
φt pt 726·3216 62·67
φc pt 712·99 9 64·05
Rhymney1991–92 to 2002–03φc pc1418·10 2263·86
φt pc1405·3012230·52
φt pt1347·6522151·59
φpre–post pt1332·9713156·10
φc pt1330·9212156·14
North Wales1988–89 to 2002–03φc pc2974·63 2538·90
φt pc2879·8915417·90
φc pt2826·9915365·01
φt pt2825·8728337·25
φpre–post pt2824·0916360·07

The most parsimonious model indicated that p varied fully with time but that φ was constant over the study period (i.e. 1991–98). Annual survival rate for this period was estimated to be 0·884 (SE = 0·043), a very similar rate to the pre-closure value calculated using data from colour-ringed birds.

Survival rates of adult redshank from Rhymney

Models describing annual survival rates and recapture probabilities for adult redshank originally caught and ringed at Rhymney are compared in Table 4b. The recapture probability for these birds equalled 0·086 (SE = 0·010) in the base model, in which both φ and p were assumed to be constant.

The best-fit model indicated that p varied fully with time but that φ was constant over the study period. No significant difference in survival was found between the 8 years before the closure of the Cardiff Bay barrage and the 3 years (2000–02) after (inline image = 0·03, NS). The annual survival rate for adult redshank originally caught and ringed at Rhymney was thus estimated to be 0·860 (SE = 0·021), a similar rate to that of Cardiff Bay adults pre-closure.

Survival rates of adult redshank from north Wales

Models describing the survival rates and recapture probabilities for adult redshank caught at the north Wales control site are compared in Table 4c. The recapture probability for these birds equalled 0·142 (SE = 0·010) in the base model in which both φ and p were assumed to be constant.

The final model indicated that p varied fully with time but that φ was actually greater in the years 2000–02 (i.e. the period following the closure of the Cardiff Bay barrage) than those before (inline image = 4·94, P = 0·0262). The annual survival rate in pre-closure years was estimated to equal 0·730 (SE = 0·017), compared with 0·932 (SE = 0·098) over the period 2000–02. These rates did not differ according to probable race (inline image = 1·71, NS).

Discussion

impacts of habitat loss on individual fitness

The probability that an individual is able to survive displacement by habitat loss will be affected by a number of factors. The availability and proximity of suitable habitat is a clear prerequisite, although the ability of individuals to relocate and use alternative sites may depend on the site-faithfulness exhibited by the species and whether displaced individuals have had prior experience of these sites. Should alternative sites be limited in quality or extent and already at or near capacity, increased densities may lead to intense competition for available resources (Goss-Custard 1985; Goss-Custard et al. 2002) and thus potentially increased mortality in the population.

Recent work has highlighted how changes in food supply (Camphuysen et al. 1996; Atkinson et al. 2003) and, indeed, densities (Durell et al. 2000, 2001) can impact the survival of waterbirds. However, despite significant modelling work (Durell et al. 2004, 2005; see also Stillman et al. 2000; Stillman 2003), attempts to provide empirical evidence that habitat loss actually does impact the fitness of birds displaced to remaining habitat have been limited (Lambeck 1991; Ganter & Ebbinge 1997; Ganter, Prokosch & Ebbinge 1997). Among other taxa, there is evidence that reductions in habitat quality associated with habitat loss can affect the survival of salmon Oncorhynchus spp. (Magnusson & Hilborn 2003). An experimental study on grey-tailed voles Microtus canicaudus (Miller), however, reported that a 70% reduction in habitat had no adverse effects on adult survival, reproductive rate, juvenile recruitment or population size (Wolff, Schauber & Edge 1997).

The present study found evidence that the loss of Cardiff Bay impacted on both the body condition and survival of the redshank that wintered there formerly. Analysis of biometric data showed that displaced redshank had difficulty maintaining their mass in the first winter post-closure: adults previously only recorded at Cardiff Bay were significantly lighter, controlling for body size and month, than those previously recorded at Rhymney. In The Netherlands, oystercatchers displaced by a loss of mudflats in the Delta region were similarly significantly lighter than those originally ringed at other neighbouring sites (Lambeck 1991). Redshank displaced from Cardiff Bay are likely to have faced intense competition for food, as a result of the increased densities at Rhymney (and other nearby feeding areas), as well as their limited experience of these sites: the redshank that wintered at Cardiff Bay were highly faithful to the site prior to its loss (Burton 2000a; Burton & Armitage 2005).

More significantly, the adult redshank displaced from Cardiff Bay also suffered increased mortality in the winters after barrage closure. The estimated annual survival rate of adult redshank originally caught and ringed at Cardiff Bay fell from 0·846 in the 2 years prior to barrage closure to 0·778 in the 3 following years because of a significant decline in winter survival rates. The local increase in the densities of redshank, in particular at Rhymney, may have impacted survival either directly, through starvation, or indirectly, by increasing the likelihood that birds may have been predated (Goss-Custard 2003), e.g. by sparrowhawks Accipiter nisus (Linnaeus) and peregrines Falco peregrinus (Tunstall). Although it was not possible to evaluate whether the increase in mortality among Cardiff Bay redshank was directly related to poor body condition (as survival rates were calculated from resighting data not recaptures), it does appear that it resulted from the birds’ displacement. First, there was no significant difference before and after barrage closure in the annual survival rate of adult redshank originally caught and ringed at Rhymney. Estimated annual survival for these birds was 0·860, similar to the value for Cardiff Bay adults prior to barrage closure. Secondly, estimated survival of adult redshank at the north Wales control site was actually higher in the post-barrage closure period than beforehand and this did not differ according to race, indicating that the drop in survival of Cardiff Bay birds was not the result of poor weather regionally or a decline in the survival of British redshank.

The survival rate of Cardiff Bay redshank over the winters subsequent to barrage closure may have been underestimated if birds had not been detected once displaced. However, the selected model accounted for a lower resighting rate following barrage closure and there is little evidence that birds moved to sites not surveyed. The majority of colour-marked individuals from the bay moved to Rhymney, just 4 km away; despite extensive searches of sites throughout the wider Severn, none was recorded more than 19 km distant (N. H. K. Burton & M. J. S. Armitage, unpublished data). If birds did move to sites further afield, not covered by surveys, it is perhaps more probable that they would have done this after the first (2000) breeding season following the barrage's closure. Over-summer return rates were found to be constant over the study period, however, and thus it seems that this did not happen. No redshank colour-ringed at Cardiff (in winter) was recorded away from the Severn in winter, despite there be a well-established network for reporting sightings, which by 2000 had resulted in 26 individuals being reported during the breeding season (Burton et al. 2002). There was therefore no evidence that survival was underestimated because of emigration beyond the area surveyed.

Although winter survival rates fell following barrage closure, it should be noted that the annual survival rate estimated for redshank displaced from Cardiff Bay is no lower than some other previous estimates for the species. Großkopf (1959, 1964) (see Boyd 1962) and Ottvall (2004), for example, reported adult survival rates of 71%, 75% and 80%, respectively, and Jackson (1988) rates of 77% and 75% for adult males and females, in studies based on colour-ring sightings in the breeding season. Thompson & Hale (1993), likewise, reported rates of 75% and 72% for males and females in a mark–recapture study of breeding birds, and Insley et al. (1997) a rate of 74% for birds at least three winters old.

It was not possible to distinguish changes in the survival of displaced redshank of the British and Icelandic races because of the small numbers of Icelandic birds that formerly used Cardiff Bay (Burton et al. 2002). However, as the population at Rhymney contained a higher proportion of Icelandic birds prior to barrage closure it might be supposed that Icelandic redshank from Cardiff Bay could have had a competitive advantage over British redshank following their displacement and thus experienced lower mortality. Impacts on the overall local numbers of the two races might have also differed as a result of differences in breeding productivity (Goss-Custard et al. 1995; Durell, Goss-Custard & Clarke 1997).

implications of reduced fitness for the local population

Numbers of redshank at Rhymney had fallen over the 7 years preceding barrage closure. As there had also been a regional decline in the species’ numbers over this period (Maclean et al. 2005), it seems likely that this was not solely the result of local factors and consequently the site may have been below its carrying capacity for the species at the time of the bay's loss. Certainly, the high survival rates of redshank at Rhymney (and at Cardiff Bay pre-closure) suggest that birds were not previously dying because of competition for food.

However, although there may have been some capacity for birds displaced from Cardiff Bay to settle at Rhymney, the poor body condition and reduced survival among these birds indicate that this site could not fully support them. By understanding how habitat loss may affect individual fitness, it is possible to predict impacts on local populations (Goss-Custard 2003). The reduction in annual survival from 0·846 to 0·778 reported here for Cardiff Bay redshank, for example, represents a 44% increase in mortality rate. Without an increase in the recruitment of first-winter birds, such a change is likely to reduce substantially the local population size.

Acknowledgements

Work investigating the impacts on waterbirds of the barraging of Cardiff Bay was in part funded by Cardiff Bay Development Corporation and latterly the Council of the City and County of Cardiff. Thanks are due to Graham Couchman, Peter Ferns and other members of ringing groups for their help in catching and marking redshank at Cardiff Bay and Rhymney. We also thank Dorian Moss and the SCAN Ringing Group for use of their data from north Wales. Mike Armitage helped to monitor the movements of marked birds and we thank all who provided sightings. Useful comments on the paper came from Phil Atkinson, Steve Freeman, Jeremy Greenwood, Rob Fuller, Jenny Gill and an anonymous referee.

Ancillary