Recent changes in bird populations in British broadleaved woodland

Ibis (2007), 149 (Suppl. 2), 14–28

Blackwell Publishing Ltd

Recent changes in bird populations in British broadleaved woodland CHRIS M. HEWSON, 1* ARJUN AMAR, 2 JEREMY A. LINDSELL, 2 RICHARD M. THEWLIS, 1 SIMON BUTLER, 2 KEN SMITH 2 & ROBERT J. FULLER 1 1 British Trust for Ornithology, The Nunnery, Thetford, Norfolk IP24 2PU, UK 2 The Royal Society for the Protection of Birds, The Lodge, Sandy, Beds. SG19 2DL, UK

This paper presents estimated national and regional changes in numbers of breeding woodland birds in predominantly broadleaved woodland in Britain between the mid 1980s and 2003–04, derived from the Repeat Woodland Bird Survey (RWBS). A total of 405 sites across Britain for which historical data on bird populations were available were re-surveyed by BTO and RSPB fieldworkers during the RWBS. Changes were estimated separately for sites surveyed by the RSPB (n = 252) and BTO (n = 153) because of differences between the two groups in the field methods used in the original surveys. Changes between the 1960s and 1970s and 2003–04 were also estimated with the BTO data set only. The BTO and RSPB sites provide geographically complementary coverage of Britain and together provide much more widespread coverage than did the Common Bird Census (CBC), the former national monitoring scheme. Nationally, between the 1980s and 2003–04, six species declined significantly and three increased significantly based on the RSPB data, whereas eight declined and 13 increased according to the BTO data, with good correspondence between the changes recorded by the two data sets (only seven of 34 species showing trends differing in direction). Eight species showed large national declines (> 25%) according to both data sets: Garden Warbler Sylvia borin, Lesser Redpoll Carduelis cabaret, Lesser Spotted Woodpecker Dendrocopos minor, Spotted Flycatcher Muscicapa striata, Tree Pipit Anthus trivialis, Willow Tit Poecile montanus, Willow Warbler Phylloscopus trochilus and Wood Warbler P. sibilatrix. Eleven species showed large national increases (> 25%) in both data sets: Blackcap Sylvia atricapilla, Blue Tit Cyanistes caeruleus, Common Chiffchaff Phylloscopus collybita, Coal Tit Periparus ater, Great Spotted Woodpecker Dendrocopos major, Great Tit Parus major, Green Woodpecker Picus viridis, Goldcrest Regulus regulus, European Robin Erithacus rubecula, Eurasian Treecreeper Certhia familaris and Winter Wren Troglodytes troglodytes. Analysis of the longer-term changes using the BTO data set showed significant declines between the 1960s and 2003–04 in seven species and increases in 12 species. The large declines detected by the national monitoring schemes (CBC/BBS) for Lesser Spotted Woodpecker, Lesser Redpoll, Spotted Flycatcher, Tree Pipit, Willow Tit, Willow Warbler and Wood Warbler were all confirmed by the RWBS but the declines of the Common Bullfinch Pyrrhula pyrrhula, Marsh Tit Poecile palustris and Mistle Thrush Turdus viscivorus were not clearly supported. The declines detected by the RWBS for Garden Warbler, Pied Flycatcher Ficedula hypoleuca and Common Redstart Phoenicurus phoenicurus were not evident from the national monitoring schemes. Most of the increases detected by the monitoring schemes were confirmed by the RWBS. However, the large increases detected by the RWBS for Coal Tit, Goldcrest, Great Tit and Eurasian Treecreeper were unexpected. All long-distance migrants have declined whilst medium-distance migrants (Blackcap and Common Chiffchaff ) have increased. Also, rare and specialist woodland species have fared less well than common and generalist species.

*Corresponding author. Email: [email protected]

© 2007 The Authors Journal compilation © 2007 British Ornithologists’ Union

Recent changes in British woodland bird populations

In recent years, there has been growing concern about declines in woodland bird populations in the UK. The UK Government’s Wildbird Populations Framework Indicator for woodland birds fell by 20% between 1976 and 2001 (Defra 2005) and a number of woodland breeding species have been added to the red and amber lists of Birds of Conservation Concern (Gregory et al. 2002). These declines have been detected through the BTO Common Birds Census (CBC) (which gives information from 1966 to 2000) and the BTO/JNCC/RSPB Breeding Bird Survey (BBS) (which gives information from 1994 to the present). The data from these two monitoring schemes have been combined to give a continuous CBC/BBS trend (Freeman et al. 2007). The CBC was based on a non-randomly selected set of sites with a strong bias towards the south and east of Britain and there was considerable turnover of sites contributing (Marchant et al. 1990), so it is unclear how representative CBC population trends are of the whole country. These limitations of the CBC have largely been addressed in the BBS though this still suffers from small sample sizes of sites containing rare and localized species. The Repeat Woodland Bird Survey (RWBS) was designed to provide information on regional changes in woodland bird populations and assess the reliability of the combined CBC/BBS trends at the national scale. Surveys were conducted in broadleaved or mixed woodland sites (sites where at least 20% of trees were broadleaved species) that had been originally surveyed either as CBC plots or as part of an extensive survey of woodland bird populations undertaken by the RSPB in the mid-1980s. Identical survey methods to the original surveys were used within each site, thus allowing population changes to be assessed through direct comparison of counts made at the same sites in the different time periods. This paper reports population changes both nationally and regionally based on the 405 sites covered by the RWBS. Two sets of sites are used: (1) 253 sites surveyed by the RSPB in the 1980s and again in 2003–04 using point counts and (2) 152 sites surveyed using territory mapping as part of the BTO’s CBC scheme during at least one of three historical time periods (1960s, 1970s and 1980s) and again in 2003–04. The BTO data are thus not completely independent of the data used to calculate the initial CBC trends but by re-visiting the same sites for which historical information is available, and by stratifying site selection across regions, the potential problems of site turnover and regional bias in the original CBC

15

have been largely overcome. Although these BTO survey sites represent a non-random sample of potential woodland sites within Britain, the sites covered by the RSPB contrast in a number of characteristics and provide a complementary data set. In particular, the RSPB sites are generally within larger blocks of woodland situated in less human-dominated landscapes and provide better coverage of northern and western parts of Britain. The combined data set thus provides wider and more representative geographical coverage of Britain than did the CBC (Fig. 1). Population changes are estimated separately for RSPB and BTO data sets due to the different survey methods and response variables employed. However, comparisons are made of the results obtained from the two data sets in order to assess the robustness of conclusions reached. Thus, the different survey methods employed on the RSPB and BTO survey sites allow assessment of the generality of population changes independent of survey method and site type. Comparisons are also made between the findings of the present study and population changes determined by the CBC/BBS to highlight differences in the reported trends. METHODS Survey design Information on bird numbers was available for three historical periods, which are referred to as survey episodes. These were: 1965–72, 1973–80 and 1981– 88. Data for the first two episodes were available only for the BTO data set; data from the third episode were available for both the RSPB and the BTO data sets. All historical data from the RSPB point counts were collected from each site in a single year between 1982 and 1988. These same sites were again surveyed mostly in just a single year of the RWBS (2003–04), with around 20% of sites in each region (n = 56 in total) being surveyed in both years to provide information on interannual variation. Historical data were available for most CBC sites for multiple years, so two years were chosen at random for inclusion in the study when more than one was available from each of the survey episodes. Of the 153 CBC sites, 127 were re-surveyed in both 2003 and 2004; the remainder were surveyed in one year only. The re-survey consisted of just four visits for territory mapping (rather than the 8–10 visits used historically) and so the census maps for the selected historical years were re-interpreted, using four


16

C. M. Hewson et al.

Figure 1. (a) Location of the RSPB (open circles) and BTO (filled circles) RWBS study sites. Regional boundaries shown correspond to the regions outlined in Table 1. Details of the five areas indicated are shown in (b).

historical visits from the original 8–10 that most closely matched the periods of the resurvey. This made the numbers of ‘territories’ comparable with the results of the RWBS. To account for changes in phenology, in particular the fact that laying dates of birds in Britain advanced by an average of 8.8 days between 1971 and 1995 (Crick et al. 1997), visits that fell towards the early part of each period were selected whilst those that fell in the last 5 days were avoided. The four visits were then re-interpreted using the same criteria for assessing territory clusters as used to analyse the data collected during the RWBS (outlined below). Re-survey site selection More sites had been surveyed historically than could be covered by the re-survey. Selection of sites from those available was undertaken on the basis of


(1) geographical location, (2) the area of the woodland block within which each site was located and (3) in the case of the BTO survey sites, the historical episodes for which data were available. Selection was carried out to give the best possible geographical coverage and wood area distribution across the RSPB and BTO survey sites taken together and therefore the distribution of the two sets of sites was largely complementary. The sites originally surveyed by the RSPB were clustered into a number of broad localities (e.g. Forest of Dean, New Forest, Welsh Marches) visible as clusters of sites within each region in Figure 1(a,b). Re-survey sites were selected from each of these localities. The final selection of BTO survey sites was made from a pool of 249 sites given by Thewlis et al. (2003), which had been selected based on woodland type (pure coniferous woods were excluded), the availability of data from the historical episodes


Table 1. Regional distribution of sites included in the RWBS. The abbreviations shown are those used in this paper. The regions correspond to Government Office Regions (GORs) except that SE includes both the South East and London GORs and NE includes both North East and Yorkshire GORs. Region South East East of England East Midlands South West West Midlands Wales North West North East Scotland Total

Abbreviation

RSPB

BTO

All sites

SE EE EM SW WM Wa NW NE Sc

34 14 18 42 29 56 0 0 59 252

44 15 10 30 8 12 14 20 0 153

78 29 28 72 37 68 14 20 59 405

(maps for at least 1 and preferably 2 years in at least one of the historical survey episodes had to be available at BTO HQ) and species composition during historical surveys (plots had to contain one of the following species, to ensure adequate representation of species thought to have declined on the basis of CBC trends: Willow Tit Poecile montanus, Marsh Tit P. palustris, Tree Pipit Anthus trivialis, Spotted Flycatcher Muscicapa striata, Lesser Spotted Woodpecker Dendrocopos minor, Dunnock Prunella modularis, Willow Warbler Phylloscopus trochilus, Lesser Redpoll Carduelis cabaret, Hawfinch Coccothraustes coccothraustes and Common Bullfinch Pyrrhula pyrrhula). In practice, the species criterion was rarely decisive as virtually all plots had historical records of at least one of the key species. All potential re-survey sites were categorized by region (Table 1), and the area of the block of contiguous broadleaved woodland in which they were embedded was calculated using the digitized GIS layer of the National Inventory of Trees and Woodland (NIWT) (Forestry Commission 2003). All woodland types were considered except stands of pure conifer. Contiguous polygons of the same woodland type were considered as one unit for our purposes and polygons were considered contiguous if they were closer than 10 m to each other. Sites were then assigned to one of the following size classes of woodland blocks: < 20 ha, 20–50 ha, 50–100 ha, 100–500 ha and > 500 ha. In the case of the BTO sites, a further classification was used for woods of less than 10 ha that were embedded in a farmland matrix to allow us to include a sample of sites to determine whether population trends in small farm

17

woods were different from those in other woods. The results of these analyses are reported in Amar et al. (2006). Within each region and size class, sites were randomly selected from each RSPB locality and from the overall BTO pool. The BTO selection was then checked to ensure (1) that each historical survey period was covered by an adequate sample of sites and (2) that at least 20 small farm woods had been incorporated in the total sample. Substitutions were made with sites from the original pool until these criteria were satisfied. Following this initial selection, the combined RSPB and BTO selection was then overlaid onto a 100-km grid and substitutions of sites were made from the remaining pool, where possible, to ensure as even coverage as possible at this spatial scale in the overall site selection. The regional distribution of RWBS sites is shown in Table 1. In total, 253 woods were selected from the pool of RSPB survey sites and 153 from the BTO pool (Fig. 1). The BTO selection included 22 small farm woods. In total, 126 sites were chosen from the BTO selection to be surveyed in both years, with priority being given to those sites with the greatest number of years of historical data available. Characteristics of the re-survey sites A detailed examination of the characteristics and representativeness of the sites used in the RWBS is given in Amar et al. (2006). The percentage distribution of sites across regions was very similar to the distribution of broadleaved woodlands according to the National Inventory of Woodlands and Trees. Compared with the size distribution of ancient semi-natural woods in England and Wales given by Spencer and Kirby (1992), the RWBS sites oversampled woods of > 100 ha and under-sampled those < 20 ha. However, as these discrepancies are based on number of woods falling into each size class rather than the area of woodland contributed by each, it is likely that our sites are in fact a more representative sample of woodland by area than these apparent biases suggest. Secondly, as each of the sites included in the RWBS must have been at the very least 15 years old (due to the interval between the end of the historical survey period and the start of the RWBS) new woodlands would not be included and therefore are not represented in the RWBS sample. The size distributions of woodland blocks containing re-survey sites were similar for the BTO and the RSPB selections. However, a greater proportion


18

C. M. Hewson et al.

of BTO sites fell in woods of < 20 ha and a greater proportion of RSPB sites fell in woods of 100– 500 ha. Woodland blocks containing BTO sites also tended to be surrounded by more human-altered landscapes (e.g. arable/horticultural land, improved grassland, urban/suburban areas) than those containing RSPB sites. Differences in the geographical distributions of the two samples of sites (Table 1, Fig. 1) meant the RSPB survey sites provided better coverage than the BTO of the ranges of species associated with deciduous oak Quercus woodlands along the Atlantic seaboard of Britain (e.g. Pied Flycatcher Ficedula hypoleuca, Common Redstart Phoenicurus phoenicurus and Wood Warbler Phylloscopus sibilatrix). Therefore, the population changes for these species derived from the RSPB data set are likely to be more reliable. Bird survey methods To ensure data collected during the re-survey were directly comparable with the historical data available for each site, survey methods were made as similar as possible to those used in the original surveys. The two re-surveys therefore used two different census methods depending on the site – point counts in the case of the RSPB sites and four-visit mapping censuses in the case of the BTO surveys. For the RSPB surveys, a random number table was used to select points based on their grid reference within the area sampled by the original surveys, with no point being closer than 50 m to the edge of the wood and no two points being closer than 100 m of each other. Points were marked on a map, located in the field using a GPS and then marked with flagging tape during a reconnaissance visit, to allow easy relocation. In most cases, ten points were surveyed per site but there was some variation between sites and survey episodes. Two visits were made to each point, one in April or the first week of May and another in the last 3 weeks of May or first half of June. The timing of the two visits roughly corresponded with the historical survey but, where possible, was up to 1 week earlier to account for changes in phenology (Crick et al. 1997, Crick & Sparks 1999) since the 1980s. All bird counts were carried out in the morning, starting within 1 h of dawn and were completed by midday. Counts were not made in heavy rain or in strong or cold winds. The order in which the points were visited was reversed for the second visit. On arrival at the point, observers waited 5 min then started counting all individuals


seen or heard for 5 min. Birds were recorded in two distance categories although the distance bands are not used in the analyses presented here. Counts in visit 1 and visit 2 are treated separately in the analyses. For the BTO surveys, observers carried out a reconnaissance visit during which they familiarized themselves with the site using a map showing its boundaries on which they marked prominent features using a GPS. This facilitated orientation during the bird surveys. The same territory mapping methods were used as for the original CBC surveys (Marchant 1983) except that four visits were made to each site rather than 8–10. The visits were spread between late March/early April and early June; visit 1 was made by 16 April, visit 2 by 8 May, visit 3 by 29 May and visit 4 by 15 June in each year. Criteria modified from those given by Marchant (1983) were used for interpreting the census maps based on four rather than 8–10 visits. Full details of the criteria used can be found in Amar et al. (2006). Note that we use the word ‘territories’ in this report to signify the clusters of registrations identified during map interpretation. These clusters are interpretations influenced by the criteria applied and the application of these criteria by the map analyst, as well as by the degree of survey effort applied. They are therefore ‘pseudo-territories’ that give an index of abundance for each species at each site rather than an absolute value of the number of territories present. A sole analyst carried out virtually all of the map interpretation for the project (both historical and re-survey maps) and applied consistent criteria throughout, so our data are internally consistent, making comparisons within the data set robust. Statistical analysis All data were analysed using SAS version 8 (SAS Institute 1999). Generalized Linear Mixed Models (GLMMs) were used to estimate changes in the number of birds recorded per point (RSPB data) and ‘territories’ per ha (BTO data) between survey episodes (i.e. 1980s vs. 2003–04 for RSPB data and 1960s, 1970s and 1980s vs. 2003–04 for BTO data). The use of mixed models allowed us to control for pseudoreplication in the data and to control the use of data from different combinations of sites and years to calculate population changes between the survey episodes. This in turn allowed us to control for short-term fluctuations in populations when examining underlying population trends.


Analyses incorporated data from the historical survey episode in the 1980s and the contemporary survey (2003–04). These analyses were run separately for RSPB and BTO data, with different model structures (random terms) to account for the different survey structures (see later). Analyses incorporating data from all three historical survey episodes to assess changes between the 1960s, 1970s, 1980s and 2003–04 were also run for the BTO data (but results for change between the 1980s and 2003–04 derived from this model are not presented). For comparison, population changes were calculated for equivalent time intervals from the combined CBC/BBS index for all habitats by taking the mean of index values for the middle 2 years of each historical survey episode (1968/69, 1976/77 and 1984/85) and the mean of 2003/04. Model formulation The number of birds recorded at all points at a site on a given visit during the RSPB surveys, or number of ‘territories’ estimated from the BTO mapping census in a given year, were used as response variables. Only sites in which a species was recorded on at least one occasion in any episode were included in each analysis. In models of RSPB data, the log of the number of points surveyed was included as an offset to control for variation between sites and visits in the number of points at which counts were carried out. The log of plot area was included as an offset in models of BTO data to provide estimates of number of ‘territories’ per hectare from model outputs. Thus, model outputs gave parameter estimates in units of birds per point for RSPB data and ‘territories’ per hectare for BTO data. For all models using count data we assumed a Poisson error structure with a log link function. Denominator degrees of freedom were estimated using Satterthwaite’s formula (Littell et al. 1996). To examine changes in regional and national populations we specified the following variables as categorical fixed effects in the models: survey episode (1960s, 1970s, 1980s or 2003–04), region, and the interaction between survey episode and region (survey episode*region). Values for the numbers of birds per point or ‘territories’ per hectare counted were obtained through the back transformation of Least Square Means (LSMs) and were used to calculate proportional changes. The statistical significance of the differences in numbers between episodes or regions was examined through type III analyses

19

using the difference in Least Square Mean tests (DLSMs). Models of RSPB data also included visit (either early or late) as a permanent fixed effect. We included five random effects in the model: site, region, year, visit*year and site*survey episode. Site (a unique identifier for each wood surveyed) was specified in the model to allow comparison between episodes at the level of each specific wood to be incorporated in the analysis. Region was specified because of the clustering of many of the sites (Fig. 1). Year was specified to account for surveys being conducted in different years, which may have influenced the numbers counted. Visit*year was specified to account for the fact that visits within the same year were more closely related than visits between years. Lastly, site*survey episode was specified to account for the fact that some sites had surveys carried out in both 2003 and 2004. For three species in the RSPB data set (Hawfinch, Lesser Spotted Woodpecker and Willow Tit) with small sample sizes (few records at few sites) we modelled change in the probability of detection per point rather than in number of birds recorded per point. The response variable was number of points at which the species was detected divided by the number of points surveyed, and visit and episode were fitted as fixed terms using a generalized linear model assuming a binomial distribution with a logit link function. RSPB sites fell within seven regions: SE, SW, EE, EM, WM, WA and Sc (codes are listed in Table 1). For each species we examined the number of sites with records in each region. Where regions had few records for a species, there was often insufficient data to generate parameter estimates by region and models would not converge. Therefore, we excluded every region with fewer than ten visit records for a given species before running the models. We were therefore unable to generate change estimates for these regions. Additionally, models often did not converge where the data from any specific region in any episode were all zero, for example where there had been a complete disappearance of a species from a region. In these instances (for Lesser Redpoll and Spotted Flycatcher), we added a value of 0.01 to all of the counts prior to the analysis. As BTO data consisted of only a single count in each year surveyed, and did not have any strong clustering of sites, models were simpler than equivalent models for the RSPB data. Furthermore, because almost all sites had multiple survey years within each


20

C. M. Hewson et al.

survey episode there was information on interannual variation which could be incorporated in the analysis, resulting in more robust estimates of changes in populations between episodes (due to separation of the effects of site and year). Thus, only site and year were included as random effects, to account for nonindependence of visits to the same site (thus allowing direct comparison of data from a site in different survey episodes) and to account for differences between the years used to assess the population at each site during each survey episode. BTO sites fell within the following eight regions: SE, SW, EE, EM, WM, WA, NW and NE. Models were run to estimate population changes for each of these regions and to control for the distribution of sites between the regions when generating national population changes. Where models failed to converge with this full regional division, regions were combined. Only contiguous regions were combined and groupings were chosen to maximize the number of the original regions that were left ungrouped. In a few instances, secondary models were run in which the data from some regions were deleted rather than combined with others, in order to increase the number of the original regions that remained ungrouped and for which population changes could therefore be produced. However, all national population changes presented were produced from models including all data, with regional groupings determined as outlined above. RESULTS Changes between the 1980s and 2003–4 National percentage population changes between the 1980s and 2003–04 derived from BTO and RSPB data are shown in Table 2. For the RSPB data, there were six declines and three increases that were significant at P < 0.05 whereas for the BTO data, there were eight declines and 13 increases that were significant. The greater number of significant changes in the BTO results is probably attributable to the differences in the structure of the two data sets (see Methods). Figure 2 shows a comparison of the national population changes from the analyses of the two data sets. These show close agreement, with only seven of 34 species showing changes that differed in sign between the two data sets. Only one of these changes was significant at P < 0.05 (the increase of Chaffinch Fringilla coelebs according to the BTO data set;


Table 2. National percentage population changes between 1980s and 2003–04 derived from RSPB and BTO data. The combined CBC/BBS population trend for all habitats is shown for comparison. Species are shown in ascending order of the percentage change according to the RSPB data. Changes shown in bold type were significant at P < 0.05 and changes in italics were significant at P = 0.1. Species names in bold type indicate species included in the PSA Woodland Bird Indicator, with those designated as woodland specialists within the Indicator also in italics. Species

RSPB

BTO

Tree Pipit Hawfinch* Willow Tit Willow Warbler Lesser Spotted Woodpecker Lesser Redpoll Wood Warbler Common Redstart Garden Warbler Spotted Flycatcher Eurasian Jay Pied Flycatcher Eurasian Jackdaw Chaffinch Common Bullfinch Mistle Thrush Wood Nuthatch Long-tailed Tit Dunnock Marsh Tit Eurasian Siskin* Great Tit Blue Tit Coal Tit Song Thrush Common Blackbird European Robin Blackcap Goldcrest Woodpigeon Winter Wren Eurasian Treecreeper Great Spotted Woodpecker Common Chiffchaff Green Woodpecker

–85.4 –73.5 –72.1 –68.8 –58.9 –58.7 –55.0 –54.4 –39.4 –36.3 –26.8 –20.1 –19 –5.5 –2.0 8.8 9.8 12.7 13.0 26.5 28.4 31.8 32.5 48.7 52.2 64.3 71.3 79.8 87.5 88.5 91.0 95.1 123.1 190.7 269.3

–69.7 –17.4 –77.5 –74.2 –43.6 –88.9 –64 7.7 –25.6 –70.4 –19.9 –24.6 6.9 25.9 10.7 –12.5 78.7 130.8 –5.8 –27 107.6 51.2 30.8 74.0 –0.6 15.8 63.5 57.2 138.3 n/a 56.5 51.5 69.8 154.8 80.7

CBC/BBS –81.2 n/a –76.2 –63.0 n/a –94.4 n/a 12.5 –10.9 –75.8 –17.0 n/a 24.0 9.2 –20.3 –21.1 58.3 61.5 8.0 –22.8 n/a 21.2 13.0 11.7 –0.6 5.0 51.4 68.1 2.4 55.7 47.0 –15.3 79.4 198.1 125.4

*Hawfinch and Eurasian Siskin: BTO trend based on 22 and 36 ‘territories’, respectively.

additionally, the decline of Marsh Tit according to the same data set was significant at P < 0.1). Eight species showed large declines (> 25%) in both data sets, namely Garden Warbler Sylvia borin, Lesser Redpoll, Lesser Spotted Woodpecker, Spotted Flycatcher, Tree Pipit, Willow Tit, Willow Warbler and Wood Warbler. Additionally, Hawfinch showed a


Figure 2. Comparison of percentage changes in counts of each species at RSPB and BTO survey sites between the 1980s and 2003–04. [Species codes are as follows: B, Common Blackbird; BC, Blackcap; BT, Blue Tit; BF, Common Bullfinch; CC, Common Chiffchaff; CH, Chaffinch; CT, Coal Tit; D, Dunnock; G, Green Woodpecker; GC, Goldcrest; GS, Great Spotted Woodpecker; GT, Great Tit; GW, Garden Warbler; HF, Hawfinch; J, Eurasian Jay; JD, Eurasian Jackdaw; LS, Lesser Spotted Woodpecker; LR, Lesser Redpoll; LT, Long-tailed Tit; M, Mistle Thrush; MT, Marsh Tit; NH, Wood Nuthatch; PF, Pied Flycatcher; R, European Robin; RT, Common Redstart; SF, Spotted Flycatcher; ST, Song Thrush; TC, Treecreeper; TP, Tree Pipit; WO, Wood Warbler; WR, Winter Wren; WT, Willow Tit; WW, Willow Warbler.]

large decline according to the RSPB results and a moderate decline according to the BTO results, although the latter was based on only a very small sample of individuals and sites. Eleven species showed large increases (> 25%) in both data sets, namely Blackcap Sylvia atricapilla, Blue Tit Cyanistes caeruleus, Common Chiffchaff Phylloscopus collybita, Coal Tit Periparus ater, Great Spotted Woodpecker Dendrocopos major, Great Tit Parus major, Green Woodpecker Picus viridis, Goldcrest Regulus regulus, European Robin Erithacus rubecula, Eurasian Treecreeper Certhia familiaris and Winter Wren Troglodytes troglodytes. Additionally, Eurasian Siskin Carduelis spinus increased by more than 25% according to both data sets, but neither increase was significant at P = 0.05 and non-coniferous woods are marginal habitat for this species across most of its British range. The magnitude of these changes was in most cases similar in the two data sets, although Green Woodpecker increased more according to the RSPB results and Goldcrest increased more according to the BTO results. Additionally, Chaffinch, Wood Nuthatch Sitta europaea and Long-tailed Tit Aegithalos caudatus showed large increases in the BTO data but not in

21

the RSPB data, whilst Common Blackbird Turdus merula and Song Thrush T. philomelos showed large increases in the RSPB data but not in the BTO data. Comparison of the national change estimates between the 1980s and 2003–04 from the RWBS with equivalent estimates from the CBC/BBS for all habitats (Table 2) shows that the level of agreement between these two surveys varies between species. For instance, the large declines detected by the CBC/ BBS for Lesser Spotted Woodpecker, Lesser Redpoll, Spotted Flycatcher, Tree Pipit, Willow Tit, Willow Warbler and Wood Warbler were all confirmed by the RWBS results but the smaller decline of the Common Bullfinch was not and the declines of Marsh Tit and Mistle Thrush were supported by the BTO results but not by the RSPB results. Furthermore, the decline detected by the RWBS for Garden Warbler was greater than had been expected on the basis of the CBC/BBS trend, whilst the Common Redstart decline detected by RSPB results had not been expected based on the trends from these schemes. The large increases detected by the monitoring schemes for Blackcap and Common Chiffchaff were confirmed by the RWBS. The large increases detected by the RWBS for Coal Tit, Goldcrest and Eurasian Treecreeper were unexpected on the basis of the CBC/BBS trend whilst the increase for Great Tit was greater than expected. In the cases of all these increases, however, there is much closer agreement between the RWBS results and the most recent trends derived from the BBS alone. Tables 3 and 4 show the population changes between the 1980s and 2003–04 estimated by region. Although the number of sites contributing to the estimated change was small in some regions, there is broad agreement in the pattern of changes detected within each of the two data sets even at the regional level. For instance, the tables show that declines in species such as Lesser Redpoll, Tree Pipit, Willow Warbler and Wood Warbler occurred across all regions for which change estimates were available from both data sets. Also, the contrast between the change for Spotted Flycatcher in the SW compared with all other regions is supported by both BTO and RSPB results (see also Fig. 3). Garden Warbler decreased in most regions in both data sets, with the exception of an increase in Scotland evident in the RSPB data. Note that in addition to detecting an increase in Garden Warblers in Scotland, the RSPB results showed no decline in Willow Warbler there, in contrast to the results from both data sets for all other regions (see also Fig. 4).


22

C. M. Hewson et al.

Table 3. Percentage changes in number of birds recorded per point at RSPB survey sites in each region between the 1980s and 2003 –04. Figures in bold type represent changes significant at P < 0.05. Figures in italics represent changes significant at P < 0.1. ‘n/a’ represents regions for which the model could not generate individual estimates of change for a given species (any data from these regions were omitted from the models for the respective species). Species are listed in ascending order of the percentage national change according to this data set (see Table 2). Regions as in Table 1. Species Tree Pipit Willow Warbler Lesser Redpoll Wood Warbler Common Redstart Garden Warbler Spotted Flycatcher Eurasian Jay Pied Flycatcher Eurasian Jackdaw Chaffinch Common Bullfinch Mistle Thrush Wood Nuthatch Long-tailed Tit Dunnock MistleThrush Eurasian Siskin Great Tit Blue Tit Coal Tit Song Thrush Common Blackbird European Robin Blackcap Goldcrest Woodpigeon Winter Wren Eurasian Treecreeper Great Spotted Woodpecker Common Chiffchaff Green Woodpecker

SE

EE

EM

SW

WM

Wa

Sc

–96.1 –71.7 –10.9 –76.1 –45.3 –86.8 42.3 –44.6 n/a –28.8 –41.2 78.7 –38.8 –15.3 172.8 364.9 7.4 54.7 55.6 0.0 26.0 57.3 5.4 30.3 40.0 63.6 n/a 6.0 93.8 1.7 181.9 36.1

n/a –86.3 n/a n/a n/a –8.2 –99.2 192.6 n/a 322.0 18.4 –91.4 286.4 4.2 –29.9 –48.8 103.4 n/a 45.2 –23.5 540.2 10.0 –4.7 34.6 122.0 242.2 927.5 54.7 105.1 –12.2 147.6 2379.9

–93.4 –77.9 –98.1 n/a n/a –61.8 –15.3 –48.1 n/a 40.6 0.5 –70.5 61.6 –51.6 –53.9 –64.8 112.8 n/a 34.4 33.7 30.3 47.7 29.3 90.8 34.8 –20.8 n/a 56.1 100.1 n/a 225.2 n/a

–60.3 –71.3 –15.5 –67.0 –33.5 –60.2 370.9 –33.8 9.2 –89.8 –38.6 135.2 –23.7 –10.1 –4.0 –11.4 67.3 n/a –21.3 6.5 73.6 43.9 31.3 11.2 36.7 –11.4 –0.3 32.1 150.7 625.1 –7.1 0.0

–95.3 –79.4 –26.0 –40.2 –62.2 –22.1 7.0 –55.0 –44.1 n/a 49.5 n/a –16.1 110.0 72.8 220.1 –43.8 n/a 46.8 64.9 80.5 103.1 207.7 270.7 26.7 207.3 661.6 345.8 57.1 n/a 266.4 n/a

–71.0 –49.0 –33.3 –30.9 –51.2 –42.5 35.1 –50.9 –16.4 n/a 3.6 268.4 –2.3 118.0 119.9 –20.6 –6.1 n/a 110.1 176.0 –50.1 33.1 111.6 132.0 129.9 157.0 –24.0 245.3 228.6 282.3 431.6 49.2

–30.5 11.4 –28.9 –43.4 –70.6 264.4 –35.9 –8.7 n/a n/a 1.2 126.3 –24.5

There are some instances of contrasting findings between the two sets of results, particularly for Dunnock in the SE where the RSPB results showed a large significant increase and the BTO results showed a significant decrease. Other contrasting results were for Common Bullfinch in Wales, Dunnock in WM and Marsh Tit in EE and EM. Note that in all of these cases the RSPB results showed much more positive changes than the BTO results, but only the RSPB increase of Dunnocks in WM and the BTO decline of Marsh Tits in EE were significant at P < 0.05. Note also that the large increases in EE in the RSPB results for Eurasian Jackdaw, Eurasian Jay and Mistle Thrush (as well as Marsh Tit) contrast strongly with all other regional results from both data sets.


–28.1 24.2 n/a 6.6 –6.6 44.5 –2.4 90.4 190.9 35.4 264.0 162.3 –59.8 78.3 4.4 n/a 327.2 n/a

Changes between the 1960s/1970s and 2003–04 from BTO data Table 5 shows longer-term national population changes estimated from the BTO data set between the 1960s and 1970s and 2003–04 (together with equivalent changes derived from the CBC/BBS covering all habitats). Between the 1960s and 2003–04, seven species decreased significantly and 12 increased significantly at P < 0.05, and between the 1970s and 2003–04, there were five significant declines and 13 increases. There are many species for which the changes from the RWBS closely correspond to those from the CBC/BBS, such as Blackcap, Wood Nuthatch, Tree Pipit, Willow Tit and Willow Warbler.


23

Table 4. Percentage changes in the number of ‘territories’ per hectare recorded at BTO survey sites in each region between the 1980s and 2003–04. Figures in bold type represent changes significant at P < 0.05. Figures in italics represent changes significant at P < 0.1. ‘n/a’ represents regions for which the model could not generate individual estimates of change for a given species (but data from all regions were used to generate national indices – see Methods). Where a model converged based on fewer than three sites within a region or fewer than 40 ‘territories’ in total (all regions and both episodes combined), species are marked with * and details are given in the footnote. Species are listed in ascending order of percentage national population change according to this data set (see Table 2). Regions as in Table 1. Species Lesser Redpoll Willow Warbler Spotted Flycatcher Tree Pipit Wood Warbler Lesser Spotted Woodpecker Marsh Tit* Garden Warbler Eurasian Jay Hawfinch Mistle Thrush Dunnock Eurasian Jackdaw Common Redstart* Common Bullfinch Song Thrush Common Blackbird Chaffinch Blue Tit Great Tit Eurasian Treecreeper Winter Wren Blackcap European Robin Great Spotted Woodpecker Coal Tit Wood Nuthatch Green Woodpecker Long-tailed Tit Goldcrest Common Chiffchaff

SE

EE

EM

SW

WM

Wa

–91.4 –73.8 –87.2 –96.6 –85.7 –43.3 –17.1 –22.9 17.7 36.3 –40.9 –53.6 41.3 –15.0 16.6 23.4 16.1 –1.5 10.5 9.9 21.9 –1.1 56.6 34.7 26.4 –3.6 36.0 55.1 36.1 62.5 73.3

–90.8 –88.8 n/a n/a n/a –64.4 –65.2 –47.0 –30.6 n/a –0.3 –52.6 65.2 269.7 –33.9 –61.6 –27.4 19.6 –4.2 50.0 25.8 3.3 25.2 3.1 92.2 4.7 –66.5 252.0 120.8 504.5 178.8

n/a –70.1 –85.4 n/a n/a n/a 3.6 –11.7 –56.4 n/a 29.0 10.5 n/a n/a –17.8 –37.2 7.7 28.0 38.5 87.4 40.4 131.4 38.4 117.7 105.9 264.5 289.6 301.6 136.7 142.2 238.3

n/a –81.0 174.8 –65.9 n/a n/a –20.7 –49.2 –23.0 n/a –0.6 3.2 91.6 n/a 18.2 92.0 10.3 10.2 16.3 27.9 114.3 39.6 37.0 40.8 57.4 93.5 47.7 34.0 89.2 178.4 32.2

n/a –89.2 n/a n/a n/a n/a –80.0 23.1 –60.0 n/a –62.5 –12.2 n/a n/a 100.0 31.2 7.7 121.1 5.3 82.4 197.4 60.1 89.5 149.1 100.0 152.9 98.0 300.0 1500.0 50.8 105.3

n/a –50.8 –58.0 –37.0 –24.8 n/a –39.4 –41.8 17.9 n/a 1.5 69.8 –33.8 –14.1 –8.3 125.2 35.4 –14.3 24.6 14.7 –11.0 37.2 21.8 34.8 7.2 7.8 10.5 –14.5 –41.2 123.0 75.8

NW

NE

n/a –10.9 n/a n/a n/a n/a 40.0 –18.2 40.0 n/a 30.8 43.0 –74.0 94.4 14.3 46.7 66.7 101.3 148.7 138.7 83.1 109.4 82.5 109.7 124.9 151.8 450.7 n/a 161.5 291.2 175.8

n/a –74.9 n/a n/a n/a n/a 100.0 –12.2 –6.2 n/a –8.3 16.0 24.0 –72.3 43.7 29.4 32.5 –0.4 53.6 38.3 23.9 137.4 138.5 70.0 81.9 71.4 229.1 n/a 142.9 19.4 726.3

*Marsh Tit: only two sites in WM and one site in NE; Redstart: Only one site in EE and one in NE.

However, there are some where the relationships are less consistent. Most notably, the RWBS changes estimated for Common Redstart do not accord with those from monitoring schemes, but it should be noted that coverage of this species’ core range by the BTO RWBS sites was poor. Also, the increases shown for Treecreeper are at odds with the small declines shown by the CBC/BBS. DISCUSSION The RWBS confirmed the strong declines detected by the CBC/BBS in several woodland bird species. The UK Wildbird Populations Framework Indicator for

woodland birds divides the 33 species into woodland specialists and generalists. The trend line for the specialists shows a marked decline compared with that for the generalists (Defra 2005). Our data show a similar pattern, with, at the national scale, ten of the 19 woodland specialist species considered here declining as opposed to none of the 12 woodland generalists. One of the most striking features has been the decline of all seven long-distance migrants – Garden Warbler, Willow Warbler, Wood Warbler, Tree Pipit, Spotted Flycatcher, Pied Flycatcher and Common Redstart. By contrast, the two middledistance migrants (Blackcap and Common Chiffchaff) have increased strongly.


24

C. M. Hewson et al.

Figure 3. Regional variation in population changes of the Spotted Flycatcher between the 1980s and 2003–04. Open circles show the number of sites at which the species was stable or increased and filled circles show the number of sites at which the species declined in each region. Sites were assigned as either declining or stable/increasing based on changes in the maximum count (RSPB sites) or maximum number of ‘territories’ (BTO sites) at each site in each survey episode.

Although there was considerable variation in the population changes at the regional level, there were some clear patterns. For some species, population change was more positive in the north than in the south. These included Blackcap and Common Chiffchaff, both middle-distance migrants, and Tree Pipit and Willow Warbler, both long-distance migrants, as well as the largely resident Common Blackbird. Wood Nuthatch also increased more in


the north, related to expansion of its range in northern England/southern Scotland. However, Spotted Flycatcher did not perform as well in the north as some regions, notably southwest England, whilst the Green Woodpecker appeared to be doing better towards the south of its range. The results show mixed changes for the six species that are red-listed and the six that are amber-listed in Birds of Conservation Concern (BoCC) on the basis


25

Figure 4. Variation between regions in population changes of the Willow Warbler between the 1980s and 2003–04. Open circles show the number of sites at which the species was stable or increased and filled circles show the number of sites at which the species declined in each region. Sites were assigned as either declining or stable/increasing based on changes in the maximum count (RSPB sites) or maximum number of ‘territories’ (BTO sites) at each site in each survey episode.

of declines in their breeding populations between 1974 and 1999 (Gregory et al. 2002). Of these 12, no clear declines between the 1980s and 2003–04 were detected for Common Bullfinch, Dunnock, Mistle Thrush or Song Thrush according to the BTO or RSPB results, or for Marsh Tit according to the RSPB results. However, the regional results show declines for Common Bullfinch, Dunnock and Mistle

Thrush in at least parts of the south and east in both data sets and thus the national decline detected for these species by the national monitoring programmes may have been biased by the regional distribution of sites. The same is also apparent for Marsh Tit and Song Thrush according to the BTO results. However, as the roughly 20-year time interval considered here differs from the 25 years between 1974 and 1999 on


26

C. M. Hewson et al.

Table 5. National percentage population changes between 1960s and 1980s and 2003–04 derived from the BTO data set. Changes are derived from models incorporating BTO data from all survey episodes. The combined CBC/BBS trend for all habitats is shown for comparison. Changes shown in bold type were significant at P < 0.05 and changes in italics were significant at P < 0.1 (BTO RWBS results) or were very close to being significant at P < 0.05 (CBC/BBS). 1960s to 2003–04 Species Common Blackbird Blackcap Blue Tit Common Bullfinch Chaffinch Common Chiffchaff Coal Tit Dunnock Garden Warbler Goldcrest Great Tit Green Woodpecker Great Spotted Woodpecker Hawfinch Eurasian Jackdaw Eurasian Jay Lesser Spotted Woodpecker Long-tailed Tit Marsh Tit Mistle Thrush Wood Nuthatch Pied Flycatcher Lesser Redpoll Common Redstart European Robin Song Thrush Spotted Flycatcher Eurasian Treecreeper Tree Pipit Willow Tit Willow Warbler Wood Warbler Winter Wren

1970s to 2003–04

RWBS

CBC/BBS

RWBS

CBC/BBS

–1.4 124.7 70.4 –22.6 33.8 88.6 104.3 –39.4 –38.4 244.0 91.3 5.3 264.6 –60.0 614.0 17.7 12.8 195.8 –35.9 1.6 71.1 29.5 –90.3 –48.6 57.6 –17.1 –66.4 123.2 –79.6 –61.1 –81.4 –38.5 87.3

–18.2 128.2 34.8 –47.2 29.6 45.5 31.7 –34.5 –11.6 12.4 77.9 169.5 260.3 n/a 80.9 2.4 n/a 73.9 –60.4 –38.6 146.4 n/a –94.9 –6.0 40.9 –51.5 –84.8 –9.0 –81.0 –84.1 –62.4 n/a 70.3

–0.1 88.8 51.6 –27.4 44.2 154.7 84.4 –31.2 22.0 50.7 83.5 –5.5 122.3 –56.1 193.4 3.5 –36.5 127.9 –28.5 –23.5 124.8 90.3 –92.7 –9.4 49.3 –7.3 –53.9 96.4 –71.9 –77.9 –72.5 –31.7 53.1

–10.2 135.4 18.1 –49.7 23.0 178.9 –10.5 –30.9 35.8 –34.9 50.6 90.1 87.5 n/a 84.9 –1.5 n/a 34.1 –35.5 –39.0 116.2 n/a –97.3 111.3 35.6 –35.2 –81.6 –21.1 –80.9 –83.5 –56.9 n/a 16.9

which the BoCC has been based, the comparisons should be viewed with caution. Note also that the analyses for BoCC are based on trends across all habitats rather than solely woodland. This may be important in the case of species such as Common Bullfinch, Dunnock and Song Thrush for which a substantial proportion of the national populations are in farmland and other non-woodland habitats (Newson et al. 2005). However, overall there is considerable evidence that the national declines detected for these species may have been regional phenomena whose importance was amplified by biases in the distribution of CBC sites towards the south and east.


Increases detected for several species had not been anticipated on the basis of previously determined trends (e.g. for Eurasian Treecreeper in particular; also Common Blackbird, Blue Tit, Coal Tit and Goldcrest). With the exception of Goldcrest in both sets of results (and Eurasian Treecreeper and Coal Tit in the RSPB results), these increases were generally greatest away from the south and east, suggesting that the regional distribution of sites in the CBC may have prevented their detection. This is supported by the fact that the more recent BBS trends for these species do show substantial increases. There is generally a close agreement between the changes detected by the RSPB and BTO data sets


between the 1980s and 2003–04. Given the differences in the types and distributions of sites covered and the survey techniques employed to produce the two data sets, this strongly suggests that the findings are robust. The agreement is particularly good for the strongly increasing or decreasing species. In general the RSPB results have wider confidence intervals than those from the BTO survey. This is because between-year fluctuations in bird populations are better accounted for in the BTO models, as two randomly chosen years of data from each site were used to estimate populations in each historical time period and most sites were surveyed again in both 2003 and 2004. The effect of fluctuations between specific years cannot be accounted for so accurately in the RSPB models as each site was only visited in one year during the 1980s and (in most cases) one year in the RWBS (so the effect of site and year are confounded) and all sites within a region tended to be visited within a single year (so the effects of region and year are largely confounded). The possible causes of the population changes outlined here were examined using habitat data collected during the RWBS. The results of these analyses will be reported elsewhere (Amar et al. in prep.). However, Fuller et al. (2005) outlined the seven most likely potential causes of population declines, namely: (1) pressures on migrants during migration or in winter; (2) climate change on the breeding grounds; (3) reduction in invertebrates; (4) impacts of agriculture on woodland edges and on habitats outide woodland; (5) reduced management of woodland; (6) intensified grazing pressure from deer; and (7) changing predation pressure (due to increases in squirrels, corvids and Great Spotted Woodpecker). It is likely that these factors operate in different combinations on different species, although some, such as those relating to predation pressure or changes in woodland structure, could impact a variety of species. In conclusion, the results show that there have been more recent increases than declines in the abundance of woodland birds. However, the species that have showed contrasting trends are not random subsets of the species considered. In particular, longdistance migrants and woodland specialists have fared less well than residents, middle-distance migrants and woodland generalists. Also, uncommon species, such as Willow Tit and Lesser Spotted Woodpecker, have fared less well than common species, such as Blue Tits and Great Spotted Woodpecker. Although the ability to repeat historical fieldwork is a powerful tool in conservation biology, it relies on the availability

27

of historical data such as the two data sets upon which this paper is based. As neither of these data sets was collected at randomly selected sites some caution is needed in extrapolating our results, although the size and geographical scope of the data sets used in the RWBS are very large. The Repeat Woodland Bird Survey was funded by Defra, the Forestry Commission, English Nature, RSPB, BTO and the Woodland Trust. We also thank the following who contributed to the project: the members of the steering group (including Richard Brand-Hardy, Fred Currie, Phil Grice, Keith Kirby, Chris Quine and Richard Smithers) and David Noble and Richard Gregory for discussion at various stages; Chas Holt, Michael MacDonald and especially Greg Conway for help with data preparation and project planning; the BTO and RSPB fieldworkers (David Anning, Loyd Berry, Gareth Bonello, Steve Coney, Gary Cole, Paul Daw, Alasdair Dawes, John Day, John Dyda, Robin Eddleston, Allan Goddard, Matthew Harding, Richard Hibbert, Paul James, Paul Jacques, Graeme Lyons, Paul Medforth, David Morris, Daniel Piec, Mike Pilsworth, Alan Salter, Rosemary Setchfield, Hilary Swift, Chris Thaxter, Nick Unwin, Howard Vaughan, Graham Weedon, Robert Wheeldon) for collecting data during the RWBS; the BTO CBC volunteers; Professor David Elston and Dr Stijn Bierman from Biostatistics Scotland (BIOSS) for statistical advice; Colin Campbell from the RSPB Data Management Unit for producing the maps; and finally the land owners, managers and agents for permission to work on their sites and for their helpful cooperation which enabled the project to go ahead.

REFERENCES Amar, A., Hewson, C.M., Thewlis, R.M., Smith, K.W., Fuller, R.J., Lindsell, J.A., Conway, G., Butler, S. & MacDonald, M.A. 2006. What’s Happening to Our Woodland Birds? Long-Term Changes in the Populations of Woodland Birds. BTO Research Report 169 & RSPB Research Report 19. Thetford: BTO and Sandy: RSPB. Crick, H.Q.P., Dudley, C., Glue, D.E. & Thompson, D.L. 1997. UK birds are laying eggs earlier. Nature 388: 525 – 527. Crick, H.Q.P. & Sparks, T.H. 1999. Climate change related to egg-laying trends. Nature 399: 423 – 434. Defra. 2005. Sustainable Development – The Government’s Approach. http://www.sustainabledevelopment.gov.uk/ performance/progress/indicators/20.htm Forestry Commission. 2003. National Inventory of Woodland and Trees. Edinburgh: Forestry Commission. Freeman, S.N., Noble, D.G., Newson, S.E. & Baillie, S.R. 2007. Modelling population changes using data from different surveys: the Common Bird Census and Breeding Bird Survey. Bird Study 54: 61–72. Fuller, R.J., Noble, D.G., Smith, K.W. & Vanhinsbergh, D. 2005. Recent declines in populations of woodland birds in Britain: a review of possible causes. Br. Birds 98: 116 –143. Gregory, R.D., Wilkinson, N.I., Noble, D.G., Robinson, J.A., Brown, A.F., Hughes, J., Proctor, D., Gibbons, D.W. &


28

C. M. Hewson et al.

Galbraith, C.A. 2002. The population status of birds in the United Kingdom, Channel Islands and Isle of Man: an analysis of conservation concern. Br. Birds 95: 410 – 450. Littell, R.C., Milliken, G.A., Stroup, W.W. & Wolfinger, R.D. 1996. SAS System for Mixed Models. Cary, NC: SAS Institute Inc. Marchant, J.H. 1983. BTO Common Birds Census Instructions. Tring: British Trust for Ornithology. Marchant, J.H., Hudson, R., Carter, S.P. & Whittington, P. 1990. Population Trends in British Breeding Birds. Tring: British Trust for Ornithology. Newson, S.E., Woodburn, R.J.W., Noble, D.G., Baillie, S.R. & Gregory, R.D. 2005. Evaluating the Breeding Bird Survey for

producing national population size and density estimates. Bird Study 52: 42 – 54. SAS Institute Inc. 1999. SAS/STAT Users’ Guide, Version 8. Cary, NC: SAS Institute Inc. Spencer, J.W. & Kirby, K.J. 1992. An inventory of ancient woodland for England and Wales. Biol. Conserv. 62: 77–93. Thewlis, R.M., Noble, D.J. & Fuller, R.J. 2003. Scoping Study of Woodland Common Birds Census Plots to Assess Suitability for a Planned Resurvey of Woodland Plots. Thetford: British Trust for Ornithology.

Received 14 August 2006; revision accepted 7 June 2007.

Appendix. Number of sites that each species was recorded in during the 1980s and 2003–04 according to the RSPB and BTO data sets. Species are shown in ascending order of the percentage national change according to the RSPB data. For each data set, the number of sites at which each species was recorded in the 1980s is followed by the number of sites at which each was recorded during 2003–04. For the BTO data set, the second figure excludes sites for which there was no historical data from the 1980s (giving a total of 92 sites). Species Tree Pipit Hawfinch Willow Tit Willow Warbler Lesser Spotted Woodpecker Lesser Redpoll Wood Warbler Common Redstart Garden Warbler Spotted Flycatcher Eurasian Jay Pied Flycatcher Eurasian Jackdaw Chaffinch Common Bullfinch Mistle Thrush Wood Nuthatch Long-tailed Tit Dunnock Marsh Tit Eurasian Siskin Great Tit Blue Tit Coal Tit Song Thrush Common Blackbird European Robin Blackcap Goldcrest Woodpigeon Winter Wren Eurasian Treecreeper Great Spotted Woodpecker Common Chiffchaff Green Woodpecker

Anthus trivialis Coccothraustes coccothraustes Poecile montanus Phylloscopus trochilus Dendrocopos minor Carduelis cabaret Phylloscopus sibilatrix Phoenicurus phoenicurus Sylvia borin Muscicapa striata Garrulus glandarius Ficedula hypoleuca Corvus monedula Fringilla coelebs Pyrrhula pyrrhula Turdus viscivorus Sitta europaea Aegithalos caudatus Prunella modularis Poecile palustris Carduelis spinus Parus major Cyanistes caeruleus Periparus ater Turdus philomelos Turdus merula Erithacus rubecula Sylvia atricapilla Regulus regulus Columba palumbus Troglodytes troglodytes Certhia familiaris Dendrocopos major Phylloscopus collybita Picus viridis


RSPB

BTO

177, 107 30, 11 26, 7 248, 203 18, 10 97, 71 186, 147 177, 143 145, 128 145, 119 157, 154 98, 89 59, 51 252, 252 52, 60 200, 194 156, 172 170, 158 88, 87 135, 127 55, 53 250, 250 251, 251 232, 239 241, 245 221, 234 248, 252 175, 201 178, 220 123, 128 252, 252 210, 240 60, 90 136, 179 41, 46

27, 8 6, 5 23, 8 88, 54 19, 9 14, 4 22, 12 15, 10 65, 52 34, 21 71, 70 8, 7 35, 43 90, 92 91, 63 67, 66 58, 69 67, 80 79, 77 53, 46 4, 8 91, 91 91, 92 73, 83 85, 85 92, 92 92, 92 82, 90 63, 76 n/a 92, 92 68, 79 68, 86 73, 88 52, 66