Portal de Periódicos da CAPES

Spatial and temporal associations between recovering populations of common raven Corvus corax and British upland wader populations

2010; Wiley; Volume: 47; Issue: 2 Linguagem: Inglês

10.1111/j.1365-2664.2010.01772.x

1365-2664

Arjun Amar, Stephen M. Redpath, Innes M.W. Sim, Graeme M. Buchanan,

Animal Ecology and Behavior Studies

Journal of Applied EcologyVolume 47, Issue 2 p. 253-262 Free Access Spatial and temporal associations between recovering populations of common raven Corvus corax and British upland wader populations Arjun Amar, Corresponding Author Arjun Amar Royal Society for the Protection of Birds – Scotland, Dunedin House, 25 Ravelston Terrace, Edinburgh, EH4 3TP, UK *Correspondence author. E-mail: arjun.amar@rspb.org.ukSearch for more papers by this authorSteve Redpath, Steve Redpath Aberdeen Centre for Environmental Sustainability, Aberdeen University & Macaulay Institute, Tillydrone Avenue, Aberdeen, AB24 2TZ, UKSearch for more papers by this authorInnes Sim, Innes Sim Royal Society for the Protection of Birds – Scotland, Dunedin House, 25 Ravelston Terrace, Edinburgh, EH4 3TP, UKSearch for more papers by this authorGraeme Buchanan, Graeme Buchanan Royal Society for the Protection of Birds – Scotland, Dunedin House, 25 Ravelston Terrace, Edinburgh, EH4 3TP, UKSearch for more papers by this author Arjun Amar, Corresponding Author Arjun Amar Royal Society for the Protection of Birds – Scotland, Dunedin House, 25 Ravelston Terrace, Edinburgh, EH4 3TP, UK *Correspondence author. E-mail: arjun.amar@rspb.org.ukSearch for more papers by this authorSteve Redpath, Steve Redpath Aberdeen Centre for Environmental Sustainability, Aberdeen University & Macaulay Institute, Tillydrone Avenue, Aberdeen, AB24 2TZ, UKSearch for more papers by this authorInnes Sim, Innes Sim Royal Society for the Protection of Birds – Scotland, Dunedin House, 25 Ravelston Terrace, Edinburgh, EH4 3TP, UKSearch for more papers by this authorGraeme Buchanan, Graeme Buchanan Royal Society for the Protection of Birds – Scotland, Dunedin House, 25 Ravelston Terrace, Edinburgh, EH4 3TP, UKSearch for more papers by this author First published: 08 March 2010 https://doi.org/10.1111/j.1365-2664.2010.01772.xCitations: 18AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinked InRedditWechat Summary 1. Recovering populations of predators and scavengers have often given rise to concerns about the impact they may have on prey species. Particularly, this is the case when the prey species are of economic or conservation importance. 2. Recovery of common raven Corvus corax populations in the UK and Europe has given rise to a conflict with some stakeholders over their concerns for both the protection of livestock and the possible detrimental impact on some upland bird species, particularly ground nesting waders. This has led to demands by some land managers for licences to lethally control ravens to protect upland breeding birds. 3. We used data from broad scale surveys of distribution and abundance of upland breeding birds in the UK carried out in 1980–1993 and 2000–2002 to test whether variation in raven abundance or change in raven abundance was negatively associated with changes in abundance of five species of waders. 4. We found no significant negative spatial or temporal relationships between ravens and any of the five species of waders. However, weak (0·05 < P < 0·1) negative relationships between raven abundance and trends of curlew Numenius arquata and lapwing Vanellus vanellus may warrant further investigation. 5. Synthesis and applications. Our study found no significant negative associations between raven abundance and population changes in upland waders, and so does not provide support to justify granting of licences for the lethal control of ravens in the interest of population-level conservation of these upland wader species. However, the near significant negative associations with lapwing and curlew merit further investigation. This study emphasizes the importance of making a thorough evaluation of the evidence base before making decisions regarding predator control. Introduction Across Europe, many predatory species are recovering from the adverse effects of persecution, over-hunting or toxic pesticides. Whilst many organizations view these changes as positive, increases in predator numbers do raise concerns amongst other groups, because of their perceived impact on prey species. This is especially the case for those predators which take prey of either economic or conservation importance (Redpath & Thirgood 1997; Landa et al. 1999; Stahl et al. 2002; Petty, Lurz & Rushton 2003; Amar et al. 2008). Such concerns can lead to conflicts between land users and conservation or statutory agencies, with land users often demanding the right to kill predators (Linnell et al. 2005; Treves & Naughton-Treves 2005). Therefore, licensing authorities must balance the need to safeguard populations of vulnerable, recovering predators, whilst at the same time acknowledging and addressing the concerns that farmers, hunters or conservationists may have over increasing predator populations. Such decisions need to be informed by evidence, and in particular the impact that predators are having on prey populations and livelihoods. In the UK, persecution of common raven Corvus corax (hereafter simply raven) by farmers and gamekeepers caused a rapid contraction in breeding range by the 20th century (Gibbons et al. 1994). More recently, however, both in the UK and across Europe, raven populations have increased rapidly (BirdLife International 2004). In the UK, the Breeding Bird Survey shows that from 1994 to 2007 the population increased by 134%, with increases of 267%, 155% and 34% in England, Scotland and Wales respectively (Risely, Noble & Baillie 2008). The species remains restricted largely to the uplands, and upland bird surveys carried between 1980 and 1993 and repeated in 2000 or 2002 found increases in ravens, although with considerable variation between survey areas (Sim et al. 2005). Ravens also occasionally kill lambs (Ratcliffe 1997) and in the UK, licences are granted to kill ravens to protect livestock. In Scotland, the number of licences granted increased by nearly 300% from 21 in 1998 to 61 in 2008 (Scottish Government, unpublished data). Concurrent with the increases in ravens reported by Sim et al. (2005), populations of many upland wader species have declined. Sim et al. (2005) reported widespread declines from 1980–1993 to 2000–2002 for three species of waders: lapwing Vanellus vanellus (L.), dunlin Calidris alpina (L.) and curlew Numenius arquata (L). For these species large per annum decreases, sufficient to cause at least a 50% population decline over 25 years, were found in several survey areas. For two other species of wader, golden plover Pluvialis apricaria (L.) and snipe Gallinago gallinago (L.), population changes varied between survey areas, with declines of over 50% in some areas, but increases of the same extent found elsewhere. Ravens are omnivorous scavengers and predators (Ratcliffe 1997). Their diet varies greatly between studies (reviewed by Ratcliffe 1997), and although full grown birds occasionally form part of their diet (Klicka & Winker 1991; Hendricks & Schlang 1998) these are thought to be more often scavenged than predated (Marquiss, Newton & Ratcliffe 1978). However, ravens prey regularly on the eggs and young of birds. Several studies in the UK have found the frequent presence of eggshells in raven castings, which included curlew and other wader species (Marquiss et al. 1978; Ewins, Dymond & Marquiss 1986; Marquiss & Booth 1986). Another study in southern Norway found that ravens were responsible for most of the June losses of golden plover clutches (Byrkjedal 1987). Thus, there is evidence to suggest that ravens could be an important predator of breeding wader, and some stakeholders have expressed concerns that ravens may have been responsible for upland waders declines. For example, Scottish Natural Heritage (SNH), the licensing authority in Scotland, received licence applications in both 2007 and 2008 for the control of ravens to protect upland breeding waders and wild gamebirds from 15 estates (SNH, unpublished data). No such licences have yet been issued, because SNH have not been satisfied that there is sufficient evidence to indicate that any wild bird populations are affected by increases in raven populations. These issues have been the source of much controversy, and were included within a petition submitted to the Scottish Parliament (http://www.scottish.parliament.uk/business/petitions/docs/PE449.htm), which led to a review on the impact of predatory birds on waders, songbirds, gamebirds and fisheries (Park et al. 2005). This review concluded that analyses to investigate changes in wader abundance in relation to predatory birds, including the raven, was a high research priority (Park et al. 2005). The greatest insights into the role of predators in limiting prey populations come from replicated predator removal experiments (Newton 1998). However, these types of experiments are logistically difficult to conduct at appropriately large spatial scales, particularly when the predators involved are legally protected. Where such studies are impractical, useful knowledge can be gained by examining correlations between changes in prey and predator abundance (Newton, Dale & Rothery 1997; Thomson et al. 1998; Amar et al. 2008; Chamberlain, Glue & Toms 2009). In this study, we take advantage of data provided by the Repeat Upland Bird Survey (RUBS) (Sim et al. 2005) to undertake such a study. The RUBS provides data from original and repeat surveys of upland breeding birds from a range of plots throughout Britain, providing data on both the abundance and population change of wader species and ravens from a number of discrete survey areas. In this study, we use the RUBS data to examine spatial and temporal associations between populations and trends in ravens and waders. Spatially, using data from the repeat surveys, we explore whether wader abundance correlates negatively with raven abundance. Temporally, we examine whether the changes in the abundance of any wader species are negatively associated with either raven abundance in the repeat survey or the change in raven abundance between the two survey periods on a plot. However, because any such relationship could in theory be as a result of difference in abundance or changes associated with environmental variables, we constructed full models incorporating broad habitat and topographical measures prior to examining the hypothesis that raven abundance or change was associated with wader abundance or changes in wader populations (Whittingham et al. 2006). Temporal associations, and in particular negative relationships between changes in wader and change in raven abundance, would provide the strongest correlative evidence that increases in raven populations might be responsible for declines in wader populations. This study presents the first attempt to test whether there are any negative associations between raven and upland wader populations with a view to helping inform policy on whether licenced control of raven populations is likely to benefit broader upland bird conservation objectives. Materials and methods The Repeat Upland Bird Survey Sim et al. (2005) conducted re-surveys of upland birds on plots distributed in a series of regionally discrete survey areas (Fig. 1). Survey areas were not necessarily representative of the UK uplands (Sim et al. 2005), because the rationale for survey area selection in the original surveys varied, and was associated with identifying areas with important upland breeding bird assemblages for legal protection under UK and EU conservation legislation. In many cases, these areas were simply known or suspected to hold relatively high densities of upland breeding birds, particularly waders. In others, they were areas directly threatened with large-scale land use change, such as afforestation. They do, however, represent plots from most of the main upland blocks of Britain. Thirteen survey areas covering a total of 1713 km2 were originally surveyed between 1980 and 1993, and were then resurveyed in 2000 or 2002. Thus, time between original and re-surveys varied from 9 years (e.g. Exmoor) to up to 21 years (Migneint). Any plots or areas of plots that were afforested between surveys were not resurveyed; for partially afforested plots, we recalculated original bird abundance within the plot removing those birds found in the areas that were subsequently afforested. Bird surveys (see Sim et al. 2005 for details) used either parallel transects (200/250 m apart) or were based on the methods of Brown & Shepherd (1993) which were broadly equivalent to 200 m transect surveys, as they involve surveying to within 100 m of every point in the plot. Surveys consisted of two visits, occurring between April and July. For analyses, we used the maximum count from either of the two visits (early and late) in both the original and re-survey. Original and repeat surveys were always conducted using the same methods. Figure 1Open in figure viewerPowerPoint Map of the UK showing and the location of the plots used in this study. Smaller images to the right and top, show the survey area locations in more detail with the key place names to provide greater clarity on the locations of plots. Please note that scales vary. Use of RUBS data Details of the number and average size of plots in the different survey areas are given in Table 1. Survey areas were the same as Sim et al. (2005) with a few exceptions. We split the North Wales area into two (Berwyn and Migneint), because the original surveys took place in different years. We excluded data from Staffordshire, South Scotland and South-west Scotland because they lacked survey data for ravens (Sim et al. 2005), and from East Flows because surveys used 500 m transects and were therefore not comparability with the methods used in the other survey areas. Where survey plots were contiguous or where they were separated by gaps of <200 m from a neighbouring plot, they were combined into a single plot for the purpose of this analysis. Altogether, we used data from surveys in 10 survey areas, from 118 plots covering 1285 km2. Table 1. Details of plots and covariates used in the analysis for each survey area Survey area Plots Plot size (km2) Total area surveyed (km2) Original raven abundance (per km2) Repeat raven abundance (per km2) Change (%) PCA1 scores PCA2 scores % <10° slope Altitude (m) Berwyn 5 14·4 (± 7·01) 71·9 0·18 (± 0·10) 1·33 (± 0·62) 656 −0·57 (± 0·06) 0·03 (± 0·02) 96 (± 18) 442 (± 11) Exmoor 9 16·1 (± 10·91) 144·6 0·00 (± 0·00) 0·00 (± 0·00) 0* −0·13 (± 0·06) 0·10 (± 0·02) 60 (± 6) 309 (± 21) Lake District 12 5·9 (± 0·75) 70·4 0·31 (± 0·09) 0·50 (± 0·11) 59 0·06 (± 0·05) 0·15 (± 0·03) 82 (± 4) 335 (± 34) Lewis & Harris 18 6·0 (± 0·40) 108·5 0·60 (±0·19) 0·58 (± 0·11) −3 −0·23 (± 0·09) −0·27 (± 0·08) 97 (± 1) 76 (± 6) Migneint 2 9·8 (± 1·35) 19·6 0·06 (±0·05) 0·60 (± 0·12) 907 0·03 (± 0·12) 0·26 (± 0·03) 82 (± 5) 425 (± 15) NE Scotland 12 18·2 (± 3·97) 218·9 0·06 (±0·02) 0·31 (± 0·11) 413 −0·38 (± 0·08) −0·00 (± 0·02) 61 (± 7) 509 (± 44) North Pennines 12 6·1 (± 0·04) 72·7 0·08 (± 0·03) 0·12 (± 0·04) 51 0·01 (± 0·06) −0·10 (± 0·10) 77 (± 6) 508 (± 34) North Yorkshire 10 6·3 (± 0·53) 62·8 0·04 (± 0·03) 0·07 (± 0·07) 58 0·00 (± 0·09) 0·03 (± 0·06) 81 (± 3) 494 (± 22) South Pennines 22 10·2 (± 3·57) 223·3 0·00 (± 0·00) 0·06 (± 0·03) NA −0·08 (± 0·05) −0·03 (± 0·04) 75 (± 3) 355 (± 13) West Flows 16 18·3 (± 4·01) 292·1 0·00 (± 0·00) 0·15(± 0·07) NA −0·43 (± 0·06) −0·14 (± 0·05) 89 (± 2) 205 (± 23) Data shown are mean (±1 SE). PCA, principal coordinates analysis; NA, not applicable, since ravens were not recorded in the first surveys. *Figures given in Sim et al. 2005 for raven change in Exmoor were incorrect. We had sufficient data to examine population changes for five wader species: curlew, golden plover, snipe, dunlin and lapwing. Counts of ravens (including flocks) were also made during the bird surveys, and we calculated raven abundance, for use in our analysis, using the maximum count between the two visits per total area (km2) of the plot. Environmental attributes of plots Environmental data extraction was undertaken using Map Info 6·0 (MapInfo Corporation 2000) or idrisi 32 (Clark Laboratories 2001). Plot boundaries were digitized and the area of land surveyed was calculated. Topographical information was extracted from a 50 m digital terrain model (Panorama, Ordnance Survey, Southampton, UK). The average altitude of each plot was calculated as the arithmetic mean of altitudes at all 50 m points across the plot. A slope model was produced in IDRISI, and the proportions of each plot with a slope of <10° was calculated. The proportions of different habitats in plots were extracted from the UK Land Cover Map 2000 (LCM2000; Fuller et al. 2005), a UK habitat map with 25-m resolution produced by classification of satellite images. We used data from five habitat types classified from LCM2000 subclass level 2: (i) dwarf shrub heath, (ii) open shrub heath, (iii) bog, (iv) rough grass and (v) acid grass. These five habitats types accounted, on average, for 82% of the area within each plot. We combined Dwarf Shrub Heath and Open Shrub Heath into its lower level subclass 1 – hereafter termed 'heather'. We then constructed a Principal Components Analysis (PCA) to describe the broad differences in habitat between plots. The first two axes of this PCA accounted for 76% of the variation in the data. The first axis, which explained 45% of the variance, described a gradient of plots from those dominated by heather to those dominated by acid grass, bog and rough grass (eigenvectors: Heather = −0·876, Bog = 0·233, Rough grass = 0·374, Acid grass = 0·196). The second axis, which explained 31% of the variance, distinguished plots which were dominated by bog habitat from those which were grass dominated (eigenvectors: Heather = −0·015, Bog = −0·880, Rough grass = 0·335, Acid Grass = 0·337). Statistical analysis Because data were collected from multiple plots clustered in survey areas, with differing numbers of plots in each area, it was necessary to incorporate this lack of independence in our analysis. Therefore, where possible we fitted a Generalized Linear Mixed Model (GLMM), with survey area fitted as a random term. Where models failed to converge using this approach, we used a Generalized Linear Model (GLM), with survey area fitted as a fixed effect. This latter approach is more conservative with respect to finding an association with the explanatory covariates, as more variation in the data is accounted for by survey area as a fixed effect. Models were fitted using either a Poisson error structure and a log link function, when examining wader abundance; or a binomial error structure and logit link function, when examining wader population change. All models were scaled to correct for over-dispersion. Denominator degrees of freedom in GLMM were estimated using Satterthwaite's formula (Littell et al. 1996). Additionally, as time span between original and repeat surveys varied (from 9 to 21 years) between survey areas and because there were small differences in the distances between transects in different survey areas, our 'survey area' term also controlled to some degree for the potential influence that these differences may have had on our measures of abundance or population change. All analyses were carried out in SAS version 9.1 (SAS Institute Inc 2004). In each analysis, we constructed full models, as recommended for hypothesis testing (Whittingham et al. 2006). Environmental variables, which included the first and second axis from the PCA describing habitat within each plot, and the average slope and altitude of each plot were included in the model together with our raven term (either abundance or change in abundance), the significance of which was then examined from the results of a Type III analyses. Thus, our models attempted to control for any additional influence of habitat or topography on wader abundance or change before testing the hypothesis that raven abundance was negatively associated with wader abundance or changes. Mean values are presented as mean ± 1 SE unless otherwise stated. Spatial association between wader and raven abundance For this analysis, we modelled wader density by using the counts of each wader species on each plot during the repeat surveys as the response variable and the log of the plot area as an offset. We weighted the analysis by the square root of plot size to account for our increased confidence in the abundance estimate on larger plots, as larger plots were more likely to be representative of the overall abundance and less susceptible to small-scale stochasticity. We carried out these analyses within a GLM, fitting survey area as a fixed effect, because of lack of model convergence when survey area was fitted as a random effect in a mixed model. Raven abundance was included in the full model, together with the habitat and typographical variables, and we examined the effect of raven abundance using a Type III analysis. Temporal association between wader change and raven abundance or change We used a binomial measure of wader population change on a plot as our response variable, by fitting the repeat survey count in each plot as the numerator, and the sum of the original and repeat counts in the plot as the denominator. This model therefore examines the degree of increase or decrease on a plot, and automatically weights appropriately for small or large counts. In this analysis, we excluded all plots with zero counts in the first episode because these 'colonization events' would have a disproportionately higher value in the response variable than plots showing large increases of pre-existing populations. However, numbers of plots excluded were relatively small, as follows (excluded plots/total plots for the species): Golden Plover n = 4/86; Lapwing n = 5/69; Dunlin n = 6/69; Curlew n = 7/95; Snipe n = 13/93). Plots that held no birds in both the original and repeat surveys were also excluded. Raven abundance in the repeat survey was included in the full model, together with the habitat and typographical variables, and we examined the effect of ravens abundance using a Type III analysis. We used this same model structure to examine whether changes in wader abundance were associated with changes in raven abundance, replacing raven abundance as the explanatory variable with changes in raven abundance, this measure was simply the abundance of ravens present in the re-survey minus the abundance of ravens present in the original-survey. Results Changes in raven and wader populations The full data on wader abundance and changes in abundance between the original and repeat surveys are presented elsewhere (Sim et al. 2005), with only the key data included here for completeness (Table 2). Raven populations showed increasing trends in all our survey areas, apart from on Lewis and Harris (Sim et al. 2005). On Exmoor, no ravens were recorded during either the original or repeat surveys. In the South Pennines and the West Flows, ravens were first recorded on the repeat visits, where none were counted originally. Elsewhere, there were increases of between 50% to 60% in the three northern England survey areas (Lakes, North Pennines and North Yorkshire), and increases of over 400% in northeast Scotland and in the two survey areas in Wales (Table 1). Table 2. Change (%) in the average number of each wader species counted in each survey area Survey area Golden plover Lapwing Dunlin Curlew Snipe Berwyn −89% (1·8, 0·2) −100% (2·8, 0·0) NA (0·0, 0·0) −78% (28·8, 6·2) −56% (1·8, 0·8) Exmoor NA (0·0, 0·0) −100% (0·1, 0·0) NA (0·0, 0·0) −63% (0·8, 0·3) +167% (0·3, 0·8) Lake District −95% (2·0, 0·1) −63% (6·5, 2·4) −100% (0·1, 0·0) −39% (15·3, 9·3) +67% (4·6, 7·7) Lewis & Harris +60% (22·7, 36·3) −19% (1·6, 1·3) +11% (30·3, 33·6) +133% (0·9, 2·1) +46% (3·5, 5·1) Migneint −60% (5·0, 2·0) −100% (2·5, 0·0) 0% (0·5, 0·5) −100% (15·5, 0·0) −50% (1·0, 0·5) Northeast Scotland −46% (18·3, 9·8) +17% (5·3, 6·2) −34% (3·2, 2·1) +8% (20·7, 22·3) +90% (2·0, 3·8) North Pennines +2% (28·5, 29·0) −26% (10·5, 7·8) −14% (4·3, 3·7) −29% (44·5, 31·5) 0% (4·3, 4·3) North Yorkshire −29% (32·3, 23·1) −45% (14·5, 8·0) −56% (3·6, 1·6) −41% (37·2, 22·1) −5% (4·3, 4·1) South Pennines +19% (12·6, 15·0) −9% (3·4, 3·1) −62% (2·6, 1·0) +116% (9·2, 19·9) −7% (1·5, 1·4) West Flows +38% (25·4, 35·0) −14% (0·7, 0·6) −5% (16·9, 16·0) −36% (3·3, 2·1) −64% (2·5, 4·1) Figures in parentheses show the average numbers counted in each survey area in the original and repeat surveys. No statistical tests of these changes were made. However, tests based on these data are given in Sim et al. (2005). Data in bold, are those survey areas where changes were significant according to Sim et al. (2005). The significance results from Sim et al.'s (2005) North Wales survey area are applied to both our Welsh survey areas. NA, not applicable, given that no birds were counted in either period. Spatial associations between wader and raven abundance Where survey areas had plots with none or very few individuals of a species recorded during both surveys, it was necessary to exclude these survey areas from the analyses to allow the models to converge (Table 3). After controlling for the effects of survey area, habitat and topographical variables, there were no significant negative relationships between raven abundance and any of the wader species (Table 3). However, there was a significant positive relationship between lapwing abundance and raven abundance (Table 3), indicating that lapwings were more abundant on plots with more ravens. Table 3. Outputs (parameter estimates, F values and significance) from full Generalized Linear Models, controlling for the influence of habitat [principal coordinates analysis (PCA)1 and PCA2] and topographical variables (altitude and slope), before testing for a spatial association between the abundance of the five wader species and the abundance of ravens during the resurvey Golden Plover Lapwing Dunlin Curlew Snipe Variable Estimate F value Estimate F value Estimate F value Estimate F value Estimate F value Intercept −6·621 *** −8·325 *** −8·439 *** −8·429 *** −6·304 *** PCA1 0·635 F 1,94 = 3·87* −0·427 F 1,91 = 0·58 0·466 F 1,80 = 2·14 −0·214 F 1,102 = 0·31 0·095 F 1,102 = 0·07 PCA2 0·245 F 1,94 = 0·41 −0·307 F 1,91 = 0·23 0·031 F 1,80 = 0·00 0·091 F 1,102 = 0·05 −0·149 F 1,102 = 0·10 Altitude 0·051 F 1,94 = 17·76*** −0·036 F 1,91 = 4·44* 0·051 F 1,80 = 6·83** −0·023 F 1,102 = 4·39* −0·019 F 1,102 = 2·31 Slope 1·391 F 1,94 = 3·08p=0·07 1·350 F 1,91 = 1·87 2·739 F 1,80 = 3·68* 2·306 F 1,102 = 12·28*** 0·636 F 1,102 = 0·65 Raven abundance 0·200 F 1,94 = 0·55 1·238 F 1,91 = 4·70* 0·064 F 1,80 = 0·07 0·147 F 1,102 = 0·12 0·154 F 1,102 = 0·23 Survey area F 8,94 = 16·59*** F 7,91 = 7·08*** F 6,80 = 9·29*** F 9,102 = 21·72*** F 9,102 = 11·82*** Survey area was fitted as a fixed effect. Results in bold were significant (*P < 0·05, **P < 0·01, ***P < 0·001). To enable models to converge we exclude data from study area lacking the species, or where counts were very low in both surveys, thus we excluded data from Berwyn plots for lapwing and dunlin, Exmoor plots for golden plover, lapwing and dunlin, Lakes plots for dunlin, and Migneint plots for curlew. Temporal associations between wader populations and raven abundance There were no significant relationships between abundance of ravens in 2000–2002 and the change in abundance of any of the wader species (Table 4). There was a marginal tendency for curlew to have declined more and for dunlin to have declined less on plots with more ravens in 2000–2002 (P = 0·09 in both cases). Thus, from these analyses there was no substantive evidence to suggest that any of the wader species showed a higher level of decline at plots with higher raven abundance in 2000–2002. Table 4. Outputs (parameter estimates, F values and significance) from full Generalized Linear Mixed Models, controlling for the influence of habitat [principal coordinates analysis (PCA)1 and PCA2] and topographical variables (Altitude and Slope), before testing for a temporal association between changes in wader abundance of the five wader species and the abundance of ravens during the resurvey Golden Plover Lapwing Dunlin Curlew Snipe Estimate F value Estimate F value Estimate F value Estimate F value Estimate F value Main effects Intercept 0·424 NS −0·857 * −0·397 NS −0·903 NS −0·571 NS PCA1 −0·383 F 1,51 = 2·51 −0·300 F 1,57 = 0·38 0·138 F 1,26 = 0·40 −0·592 F 1,72 = 3·660·05 −0·383 F 1,35 = 0·66 PCA2 0·738 F 1,63 = 5·08* 0·832 F 1,51 = 1·59 −0·104 F 1,42 = 0·07 0·331 F 1,71 = 0·81 0·206 F 1,53 = 0·13 Altitude −0·011 F 1,3 = 4