Factors influencing Willow Tit Poecile montanus site occupancy: a comparison of abandoned and occupied woods

Authors


*Corresponding author. Email: arjun.amar@rspb.org.uk

Abstract

The British Willow Tit Poecile montanus kleinschmidti underwent a decline of 85% between 1970 and 2003. The cause of this decline is unknown. However, several hypotheses have been put forward to account for it: competition from other tit species, predation by Great Spotted Woodpeckers Dendrocopos major and habitat change. In order to test these, woods that are currently occupied by Willow Tits were paired with woods (within 50 km) that had been abandoned by Willow Tits five or more years previously. Point counts for other tit species (potential competitors) and woodpecker species (potential predators) were carried out at ten evenly spaced points throughout each wood. Habitat variables were collected within a 50-m radius of where a Willow Tit was located (in the occupied woods) or where maps showed a Willow Tit had been located (for abandoned woods). No evidence was found for differences in numbers of potential competitor or potential predator species in abandoned and occupied woods. Soil water content was found to be higher at occupied sites. No other habitat features differed between the two categories of site. The drying up of British woods could therefore be implicated in the Willow Tit decline and this warrants further investigation.

The British Willow Tit Poecile montanus kleinschmidti declined by 85% between 1970 and 2003 (Eaton et al. 2005). Since an initial dramatic reduction in numbers in the 1970s, the decline has been less severe but has continued to be steady and constant. Analysis of the British Trust for Ornithology Common Bird Census (CBC) data has shown that it is in woodland and farmland that the decline has occurred (Siriwardena 2004). In wet habitats, Willow Tit populations have remained stable. As well as an overall reduction in numbers, many sites across Britain have lost their Willow Tits altogether and this species has been lost completely from several counties (Gibbons et al. 1993, Smith et al. 1993). It is thus vital that the cause of the decline be elucidated and, ultimately, acted upon.

Many woodland bird species have suffered dramatic declines in recent years (Fuller et al. 2005, Hewson et al. in press). However, no single hypothesis has been able to explain these declines and it is likely that different factors are acting on different species (Amar et al. 2006). For the decline of the Willow Tit in British woodland, three main hypotheses have been put forward (Maxwell 2002, Siriwardena 2004): (1) competition with other species for nest-sites, (2) predation by the Great Spotted Woodpecker Dendrocopos major and (3) changes in woodland habitat.

Competition for holes excavated by Willow Tits has been observed (Maxwell 2002, A. Lewis pers. obs.). Willow Tits nest in holes that they excavate in standing dead wood. These holes can be taken over by Blue Cyanistes caeruleus and Great Tits Parus major late in the building process meaning that the Willow Tits must find another suitable stump and excavate another hole (Maxwell 2002). Thus, recent increases in these behaviourally dominant species (1970–2003: Blue Tits +28% and Great Tits +63%) (Eaton et al. 2005) may have caused an increase in nest-hole oustings experienced by Willow Tits. This, in turn, may have reduced breeding success. Maxwell (2002) monitored Willow Tits for 5 years in Lanarkshire, Scotland, using both natural nests and specially constructed, wood shaving-filled nestboxes. Of the 30 pairs monitored, 20 were ousted from their initial nests, 18 by Blue Tits and two by Great Tits.

Predation by Great Spotted Woodpeckers on Willow Tit nests usually occurs at the chick stage (our pers. obs.). Willow Tits are a single-brooded species and such a predation event can therefore result in total breeding failure for that year. Great Spotted Woodpecker numbers have increased dramatically (1970–2003: +231%, Eaton et al. 2005) and this may have caused increased predation pressure on Willow Tit nests (Fuller et al. 2005). This, in turn, may have reduced overall breeding success and driven the population decline.

On mainland Europe, the Willow Tit is primarily associated with coniferous and upland forest, whereas in Britain it largely inhabits broadleaved woodland (Snow & Perrins 1998). Many changes in British woodland have occurred in recent years and, inevitably, these will have impacted woodland birds (Fuller et al. 2005). However, it is difficult to determine which species have been affected and by which changes. Some of these changes may be responsible for the Willow Tit decline. A reduction in coppicing and maturation of woodland, for example, has resulted in canopy closure, which, in combination with increased deer browsing, has resulted in a decreased shrub layer, the key habitat where Willow Tits forage and nest (Perrins 1979). In addition, the proportion of young stands and the amount of birch in British woodland have decreased since the 1980s (Fuller et al. 2005). Willow Tits are often associated with young stands of trees and birch is favoured for nesting. Any decrease in the amount of standing dead wood suitable for nesting would also be detrimental to Willow Tits.

The competition and predation hypotheses have been tested using CBC data. Siriwardena (2004) analysed, with respect to habitat type, the relationship between the local abundance of Willow Tits and that of key potential nest competitor and predator species. No negative correlations were found between Willow Tit abundance and that of Blue and Great Tits. A negative correlation was found between Willow Tit abundance and that of the Great Spotted Woodpecker but only in farmland. There was therefore little evidence that predation by the Great Spotted Woodpecker was likely to have contributed directly to the national population decline in woodland. The hypothesis relating to habitat change has not yet been tested.

This study aimed to test all three hypotheses and, in particular, to provide the first empirical test of whether habitat change within woodlands could have driven the decline of the Willow Tit. In the south and east of England, a survey was carried out at woodland sites which had been occupied by breeding Willow Tits during the last 2 years or had been recently abandoned by them. If competition or predation was responsible for sites being abandoned, it is predicted that there will be higher numbers of competitors (Blue and Great Tits) or predators (Great Spotted Woodpeckers) at the abandoned sites relative to the occupied sites. However, if habitat changes are responsible for the sites being abandoned, it is predicted that there will be habitat differences between the abandoned and occupied sites.

METHODS

Study sites

Nine pairs of sites were identified through a variety of sources such as recent surveys and county records. Each pair consisted of a site recently abandoned by Willow Tits and a site that had been occupied within the last 2 years. Sites in each pair were located within 50 km of each other and were in south and east England in the following counties; Norfolk, Northamptonshire, Buckinghamshire (n = 2), Oxfordshire, Kent (n = 2), East Sussex and West Sussex (Fig. 1, Table 1).

Figure 1.

Map showing the south of England, with major towns and the location of the abandoned (open circles) and occupied (closed circles) woodland sites used in this study.

Table 1.  Pairs of woods abandoned and currently occupied by Willow Tits (WT).
Wood nameCountyWT statusLast WTYears occupiedNo. of years surveyedCBC survey periodArea (ha)
  1. ‘Last WT’ refers to the year when Willow Tits were last recorded at the site exhibiting territorial behaviour during the breeding season. Occupied sites had records of breeding Willow Tits within the previous 2 years (2003–04). ‘Area’ refers to the area at each site surveyed for bird species (i.e. Willow Tits, potential predator species and potential competitor species). Counties fell into three geographical clusters referred to as ‘County Groups’ 1, 2 and 3. Additional information from RSPB wardens at Blean Woods and Brakeybank Woods means that ‘years occupied’ is longer than the CBC Survey Period.

BoylandNorfolk101996 3 91994 –2002 12
BrandonNorfolk112005    19
ShortNorthants.201994 2171977 –2000 35
LaundimerNorthants.212005   100
Coleshill CommonBucks.20199114251976 –2000 39
Little LindfordBucks.212005    50
Beacon HillBucks.201996 3271975 –2001 44
QueensBucks.212003    70
Sydlings CopseOxfordshire201990 7181980 –1999 25
Wychwood ForestOxfordshire212003   112
BrakeybankKent30199014231983 –2006 51
Marshley HarbourKent312005    84
BleanKent30198810241982 –2006131
ChurchlandKent312005    33
Brightling ParkE. Sussex301994 2121991 –2002 63
BuchersE. Sussex312005    47
Barn's CopseW. Sussex301991 1121989 –2000 28
Warnham NRW. Sussex312005    47

The area surveyed at each site ranged from 12 to 131 ha and there was no difference in size between occupied and abandoned sites (paired t-test, P > 0.05). The study took place over a wide geographical area and this paired design aimed to reduce any differences between the categories of site which could arise due to regional variation. Occupied sites were identified from county bird reports, reserves records and CBC plots that were known still to hold territorial Willow Tits or had done so between 2003 and 2005. These were paired with abandoned sites which were defined as such if they had been abandoned by Willow Tits for over 6 years. Seven of the nine abandoned sites were originally surveyed as part of the CBC (or were partly made up of a CBC site) and two were RSPB reserves. The CBC (1962–2000) was a mapping census which involved volunteers choosing a particular plot of land, visiting it 8–10 times during the breeding season and recording the locations of all species of birds seen or heard (Marchant et al. 1990). Where possible, the sites were visited every year. However, this depended on the availability of the volunteer surveyor. RSPB reserves have detailed maps showing the locations of breeding Willow Tits for every year. Detailed maps of past Willow Tit registrations were therefore available for all abandoned sites. The most recent registrations were used for this study.

Bird surveys

Point counts were carried out between 2 and 30 April 2005 in order to compare the number of potential competitor and predator species at occupied and abandoned sites. Points were located by placing a 100 × 100-m grid over a 1 : 12 500 map of the site. As many points as possible, with a maximum of ten, were located by eye at the intersections of the grid in a way that optimized the spread of points over the site. All counts were started within 1 h of dawn and were completed by 11:00 h. They were not made in heavy rain or strong winds. All ten points were completed on a single morning. On arrival at the point, 5 min was allowed for the birds to settle prior to beginning the count. Each count lasted for 5 min and numbers of Blue Tits, Great Tits and Great Spotted Woodpeckers heard or seen were recorded with no limit on the distance from the point. Birds flying over the area were discounted.

At occupied sites, all the locations where Willow Tits were recorded during the point count survey were noted. However, if no registrations were obtained in this way, a survey of the whole wood was carried out using a tape lure of Willow Tit song and call. This was played for 2-min periods at approximately 200-m intervals throughout the wood. Once a Willow Tit had been located, tape-luring was stopped. This procedure was also adopted at abandoned sites to confirm that Willow Tits were no longer present.

Habitat surveys

To ensure that data were collected at a relevant spatial scale, habitat variables were measured around past or present Willow Tit locations within a wood rather than throughout the wood itself. In order to select sampling points at abandoned sites, the original census maps were examined and a single point selected from each cluster of registrations. Registrations further than 200 m from each other were treated as separate clusters. All abandoned sites had multiple registrations with the exception of Brightling Park which had one. Due partly to the nature of the Willow Tit surveying at occupied sites, no points fell within 200 m of each other and thus all positive locations were sampled. All occupied sites had only one positive registration with the exception of Wychwood Forest which had three.

In June–August 2005, habitat was sampled within a 50-m radius circle centred on the past and present Willow Tit locations. Within this circle, habitat variables were measured within 37 5-m-radius circles located at every alternate intersection of a 10 × 10-m grid. This was with the exception of ‘basal area’ which was measured from the centre point of the 50-m-radius circle using a relascope (Hamilton 1975). Habitat variables were only collected at intersections with ‘suitable’ habitat. ‘Suitable’ habitat was characterized by the presence of woody stems growing either within the circle or present within the vertical column of the circle. This strategy thus avoided measuring variables in habitat with no woody vegetation such as open glades or large bodies of water. Any point falling onto ‘unsuitable’ habitat was randomly re-allocated onto vacant intersections within the 50-m circle. This resulted in a total of 37 points always being surveyed.

In every wood, 29 variables were measured within each 5-m circle. All trees (> 5 cm diameter at breast height (dbh)) were counted, identified to species and assigned to a size category according to their dbh (Size 1: 5–10 cm, Size 2: 11–20 cm, Size 3: 21–30 cm and Size 4: > 30 cm). A study in 2004–05 in the English Midlands of 33 nests (A. Lewis unpubl. data) found that Willow Tits nest in willow (Salix spp.) (n = 11), elder (Sambucus nigra) (n = 9), birch (Betula spp.) (n = 7), alder (Alnus spp.) (n = 4), rowan (Sorbus spp.) (n = 1) and hawthorn (Crataegus spp.) (n = 1). The cumulative number of these tree species within each 5-m circle was calculated to give the proportion of ‘nest tree species’. Circles where no trees were present were excluded from this analysis when calculating a mean for the 50-m-diameter circle. The Midland study found nests in dead wood of between 6.5 and 16 cm (diameter at nest-hole) that was either attached or suspended above ground, and that was at an angle of 45° or less to the vertical. The number of pieces of standing dead wood of this nature that were > 5 cm in diameter was therefore recorded. In addition, the number of dead wood pieces belonging to the ‘nest tree species’ category was recorded and the proportion of dead wood pieces belonging to this calculated. The number of active coppice stools was also recorded. Soil water content (volumetric percentage) in the upper 5 cm of the soil was measured using a theta probe and soil moisture meter (Delta-T Devices). Three measurements were taken at alternate intersections within 1 m of the centre of the 5-m circle. The median values from each point were used to calculate a mean for each 50-m circle. Data from one occupied site are missing due to a logistical problem that was encountered in the field.

Canopy was defined as vegetation at a height of > 10 m. Canopy height was recorded using a Laser Rangefinder (Bushnell Yardage Pro Sport) and adding the observer's own height. Canopy cover was measured using a canopy frame. This is a metal grid of 10 × 10 cm, divided into 16 squares of 2.5 × 2.5 cm. It is held 60 cm above the observer's head and the number of squares more than half covered by canopy is then recorded and divided by 16.

The percentage cover of vegetation in three height bands (0.5–2, 2–4 and 4–10 m) and ground cover (bare ground, bracken, bramble, grass, forbs, moss, leaf litter and brash) was estimated by eye by imagining that the 5-m subplot was viewed from above. The basal area of woody stems was recorded from the centre of the 50-m-diameter circle as a count of the number of stems of each tree species exceeding a threshold subtended angle to the observer using a simple viewing stick and cross piece (Hamilton 1975).

The landscape composition within 1-, 3- and 10-km-radius circles around each site (with the site as the centre-point) was obtained from the Land Cover Map 2000 (using subclasses Level 2) (Barr et al. 1993). Three standard categories were used: ‘coniferous woodland’, ‘deciduous woodland’ and ‘suburban/rural development’. In addition, the categories ‘total woodland’ (combining ‘deciduous’ and ‘coniferous’ woodland categories) and ‘total farmland’ (combining ‘arable cereals’, ‘arable horticulture’, ‘improved grassland’, ‘non-rotational horticulture’ and ‘set-aside grass’) were created. The distance from the centre of the site to the nearest water body, defined as a reservoir, pond, stream or spring, was measured from the Ordnance Survey maps.

Data analysis

Statistical analyses were performed using SAS Version 9.1 (SAS Institute 2003) and generalized linear mixed models were implemented using the GLIMMIX macro (Littell et al. 1996).

The nested structure of the field study was represented in the model by using ‘circle’, ‘site’, ‘county’ and ‘county group’ as random terms. Habitat sampling points were located within a 50-m-diameter circle of the past/present Willow Tit location. These circles were located within woodland sites. These woodland sites were paired within a county and these counties were located within one of three county groups (Table 1). Thus, ‘circle’ was nested within ‘site’, ‘site’ was nested within ‘county’ and ‘group’ and, finally, ‘county’ was nested within ‘county group’. The only exception to this was basal area where ‘circle’ was omitted as a random term as there was only one measure per 50-m circle. Habitat data were analysed using a different error structure according to the variable being tested. For percentage values, an arcsine square root transformation was carried out. The data were then analysed using a normal error structure with an identity link function. For count data, a Poisson error structure with a log link function was used. In order to make maximum use of the data recorded from the 37 habitat points, each data point was used as the response variable and the Willow Tit status of the site (occupied or unoccupied) as the explanatory fixed effect. Using the random terms accounted for the lack of independence in these data. The analyses were therefore not compromised by issues of pseudoreplication.

For those variables with a single measure per site (distance to water and surrounding landscape features), a simplified version of the model was used. This had only ‘county’ as a random effect. For soil water content, all values were used in the model, despite the fact that data from one occupied site were missing. The effect of predator and competitor number on a site's Willow Tit status was tested by using the total number of birds counted at a site with the log number of points surveyed as the offset.

In addition, a simple paired t-test (using the site pairs) was carried out using the mean value from each site to confirm the findings from the mixed models. For soil water content, the site pair with the missing values was omitted.

RESULTS

Numbers of Blue and Great Tits (both separately and combined) were not significantly different at abandoned and occupied sites. There was also no significant difference in Great Spotted Woodpecker numbers at abandoned and occupied sites (Fig. 2).

Figure 2.

The mean number of individuals of each potential competitor (tit) species and Great Spotted Woodpeckers (GSW) per point at woodland sites abandoned by Willow Tits (0, open circles) and occupied by Willow Tits (1, closed circles). B+G = Blue Tit and Great Tit numbers combined. The error bars represent 95% confidence intervals.

Soil water content was significantly higher at occupied sites (GLIMMIX: F1,9 = 10.69, P = 0.009, paired t-test: t = 2.96, P = 0.02) (Fig. 3). No other habitat variables differed significantly between occupied and abandoned sites (Table 2). In particular, there was no difference in the amount of low understorey cover or in the amount of standing dead wood (including that of the appropriate species and size for nesting Willow Tits). There was also no difference in the proportion of habitat in the surrounding landscape at either 1, 3 or 10 km from the site centre (Table 3).

Figure 3.

Mean soil water content (volumetric percentage) for each site abandoned by Willow Tits (0, open circles) and occupied by them (1, closed circles). The data are arranged in site pairs and in county groups (group 1, n = 1; group 2, n = 4; group 3, n = 4). The error bars represent 95% confidence intervals. Sites occupied by Willow Tits had a significantly higher soil water content (F1,0 = 10.69, P = 0.009). Note, measurements were not collected at one occupied site due to a logistical problem.

Table 2.  Variables measured at woodland sites occupied by Willow Tits and woodland sites abandoned by them. The mean value per 5-m circle and the standard error are shown for each with the exception of basal area (which is the mean value and standard error per 50 m circle). Only soil water content was found to differ significantly between the two categories of site.
 AbandonedOccupied
Trees
  No. trees 5–10 cm dbh3.8 ± 0.63.9 ± 0.7
  No. trees 11–20 cm dbh2.4 ± 0.61.6 ± 0.1
  No. trees 21–30 cm dbh0.5 ± 0.10.9 ± 0.3
  No. trees > 30 cm dbh0.5 ± 0.10.4 ± 0.1
  Tree diameter16.4 ± 0.717.2 ± 1.9
  Total no. trees6.4 ± 0.77.7 ± 1.0
  No. nest tree species2.1 ± 0.63.9 ± 1.1
  Proportion trees 5–10 cm dbh51.6 ± 4.146.3 ± 7.1
  Proportion trees 11–20 cm dbh23.6 ± 2.324.7 ± 4.8
  Proportion trees 21–30 cm dbh10.6 ± 1.616.2 ± 5.4
  Proportion trees > 30 cm dbh14.2 ± 2.212.8 ± 3.5
  Proportion nest tree species33.9 ± 5.843.7 ± 10.6
  Number of coppice stools1.9 ± 0.71.9 ± 0.5
  Basal area8.2 ± 1.95.3 ± 2.2
Dead wood
  No. pieces dead wood0.7 ± 0.10.7 ± 0.2
  No. nest species dead wood0.2 ± 0.10.3 ± 0.2
Canopy
  Canopy height (m)10.0 ± 1.111.3 ± 1.4
  Canopy cover0.5 ± 0.10.6 ± 0.1
% Cover
  Vegetation at 0.5–2 m57.7 ± 4.866.6 ± 4.9
  Vegetation at 2–4 m52.9 ± 2.948.5 ± 6.3
  Vegetation at 4–10 m64.2 ± 5.458.5 ± 5.2
  Bare ground6.5 ± 0.84.2 ± 1.6
  Bracken11.0 ± 4.712.6 ± 6.6
  Bramble20.0 ± 2.217.1 ± 6.0
  Grass23.9 ± 6.134.6 ± 8.0
  Forb32.3 ± 6.548.2 ± 8.8
  Moss15.2 ± 3.015.9 ± 5.5
  Leaf litter42.6 ± 8.038.4 ± 8.1
  Brash7.1 ± 1.212.1 ± 2.9
  Soil water content21.1 ± 1.930.0 ± 3.1
Table 3.  The proportion of four habitat types measured within 1-, 3- and 10-km-diameter circles around woodland sites occupied by Willow Tits and sites abandoned by them. The mean values and the standard error are shown for each. No variables were found to be significantly different between abandoned and occupied sites.
 AbandonedOccupied
1 km
  Broadleaved woodland26.0 ± 9.424.6 ± 7.6
  Coniferous woodland5.0 ± 5.07.4 ± 5.1
  Total woodland31.0 ± 9.032.0 ± 8.4
  Suburban/rural development0.0 ± 0.06.6 ± 3.5
  Total farmland54.0 ± 10.047.6 ± 6.9
3 km
  Broadleaved woodland19.4 ± 4.214.4 ± 2.0
  Coniferous woodland4.0 ± 2.05.8 ± 4.5
  Total woodland23.4 ± 5.420.3 ± 4.0
  Suburban/rural development6.2 ± 2.211.0 ± 4.2
  Total farmland58.07 ± 6.6655.26 ± 6.5
10 km
  Broadleaved woodland14.9 ± 2.116.1 ± 1.8
  Coniferous woodland2.0 ± 0.55.4 ± 3.7
  Total woodland16.9 ± 2.321.5 ± 2.9
  Suburban/rural development8.0 ± 1.88.7 ± 1.9
  Total farmland58 ± 3.955.8 ± 4.3

DISCUSSION

In the absence of any significant relationship between Willow Tits and their potential competitor and predator species from the CBC analysis, Siriwardena (2004) suggested that habitat change was the most likely cause of the national decline. Our study set out to test this hypothesis by comparing sites that were occupied by Willow Tits with sites that had been recently abandoned by them. Although the number of sites in our sample is small, the results show that occupied sites had significantly higher soil water content than abandoned sites. This supports previous observations that Willow Tits are associated with damp and wet habitats (Perrins 1979, Snow & Perrins 1998). These findings are also consistent with those of Siriwardena (2004), who showed that Willow Tits have declined dramatically in woodland and farmland but not in damp habitats (where they are also more abundant). A study in the English Midlands (A.J.G. Lewis et al. unpubl. data), an area where Willow Tits still occur in good numbers, also showed that in more mature woodland, the probability of site occupancy was positively related to soil water content. Alder, Willow and Birch, which Willow Tits favour for nesting (Perrins 1979), are more likely to occur under damper conditions and the higher humidity found in such areas may enhance the decay process of dead wood. Such conditions may also present favourable feeding opportunities for Willow Tits. To date, no quantitative studies investigating the mechanisms underpinning this association have been undertaken. No other habitat features varied between occupied and abandoned sites, despite some having direct relevance to Willow Tit ecology such as the abundance of dead wood. However, it should be recognized that because our study was limited to only nine pairs of sites, there is the potential for type II errors. Thus, any lack of differences found here does not necessarily mean there are no differences, simply that we may not have had sufficient power to detect them.

Willow Tits are dependent on standing dead wood for excavating nest-holes and thus any decrease in this substrate could have caused a national population decline. If this were the case, abandoned sites would be expected to have lower amounts of standing dead wood than occupied sites. The abundance of dead wood in British woodlands is thought to have increased in recent decades (Kirby & Buckley 1994, Fuller et al. 2005, Kirby et al. 2005, Amar et al. 2006, Smith 2007), possibly through events such as the storms of 1987 and 1990 (Kirby et al. 1998), Dutch Elm Disease in the 1970s (Osborne 1983) and the drought of 1976 (Peterken 1996). However, this trend does not necessarily represent an increase in standing dead wood that is suitable for excavation by Willow Tits. Our study looked specifically at differences in the amount of standing dead wood of the appropriate tree species, size and angle for nest-hole excavation. No difference was found between abandoned and occupied woods in this respect and therefore the prediction that abandoned sites have lower amounts is not supported. This study therefore found no evidence to support the hypothesis that a reduction in the quantity of standing dead wood has caused the national Willow Tit decline. No other woodland habitat variables differed between abandoned and occupied sites, including the amount of vegetation cover in the 0.5–2-m or 2–4-m height bands, the area of the wood important for foraging Willow Tits (Perrins 1979, our pers. obs.) and surrounding landscape.

This study looked for differences in potential competitor abundance between sites that are currently occupied by Willow Tits and sites that had been abandoned by Willow Tits. If competition has caused the national Willow Tit decline, numbers of competitor species are expected to be higher at abandoned sites than they are at occupied sites. However, no significant differences were found between the two categories of woods in this respect. Numbers of competitor species were actually higher at occupied sites, although this difference was not significant. Siriwardena (2004) also found a positive relationship between potential competitor species and Willow Tit numbers although these were also not significant. Both these studies therefore strongly suggest that competition with other tit species has not been a major factor in the national Willow Tit decline.

Although competition with other tit species may not be responsible for the national Willow Tit decline, Maxwell (2002) showed clearly that competition can be important on a local scale. In his study area, two-thirds of the Willow Tit population lost their nest-holes to Blue and Great Tits over a period of 5 years, although it is not known whether all of these oustings resulted in complete breeding failure. It is possible that in some suburban areas, numbers of tits have increased dramatically due to a proliferation of bird feeders and nestboxes. This, in combination with the fact that natural nest-sites can be in short supply (because, for example, older trees are removed from parks and gardens), may mean that competition for nest-holes can result in local Willow Tit declines.

The results of this study and those of Siriwardena (2004) show that it is unlikely that competition for nest-sites has caused the national Willow Tit decline. However, they do not rule it out completely. Competition could still be implicated if Blue and Great Tit behaviour varies according to the abundance of nest-holes in a wood. If nest-sites were limiting in the abandoned woods, competition could still be higher in these woods despite there being an equal number of other tits in the occupied woods. Thus, in theory, the Willow Tit decline may have been due to the interaction between densities of other tit species and the abundance of suitable dead wood, rather than to either factor in isolation. In order to test this it would be necessary to quantify, from a large number of sites, Blue and Great Tit densities, the abundance of natural nest-holes and the dead wood suitable for nesting Willow Tits. However, in the absence of such data, although both the national data (Siriwardena 2004) and the data presented here suggest that competition is not a factor in the national Willow Tit decline, neither dataset is able to rule it out definitively.

If increased chick predation by Great Spotted Woodpeckers has caused the national Willow Tit decline, numbers of this species are expected to be greater in woods that have been abandoned by Willow Tits. In addition, a relationship is expected between increased Great Spotted Woodpecker numbers and decreased Willow Tit numbers nationally. A slightly higher number of Great Spotted Woodpeckers was recorded in abandoned woods. However, this was far from significant. Using CBC data, Siriwardena (2004) showed that the only potentially important negative correlation between Willow Tit and Great Spotted Woodpecker abundance was in farmland. It is therefore unlikely that Great Spotted Woodpecker predation has caused the decline of the Willow Tit in woodland directly although it is possible that increasing numbers of the Woodpeckers in farmland has been a factor.

In this study, the only difference found between abandoned and occupied sites was a lower soil water content at abandoned sites. If a reduction in woodland soil water content has caused the national decline, it would be expected that woods have become drier over recent decades. A study by Amar et al. (2006) showed that, overall, there was no change in numbers of water features in woodlands between the 1980s and early 2000 although in five localities there had actually been a significant increase. Kirby et al. (2005) recorded a loss of some wet habitats within woodland but no overall change in mean Ellenberg wetness scores for woodland ground flora. There is therefore currently no strong evidence that woods have become drier although there has been no large-scale study to test this using the method employed herein. Peach et al. (2004) showed that in intensive agricultural landscapes, one factor contributing to declines in Song Thrush Turdus philomelos populations was drying of surface soils which limited the birds’ access to key summer invertebrate prey. However, more studies over a wider geographical scale are required to explore the relationship between Willow Tit site occupancy and soil water content.

Acknowledgments

This study was funded by the Royal Society for the Protection of Birds, English Nature and Forestry Commission England. We thank the volunteers of the British Trust for Ornithology and wardens of the RSPB for carrying out the original mapping censuses which have been so useful in this study. We are also grateful to the landowners of the sites for giving us access permission and to the many local birdwatchers and county recorders who provided information on the past and present location of Willow Tits. We thank Ken Smith, Jeremy Lindsell, Fred Currie and Phil Grice who provided help with this project. We also thank Laura Daniells and David Wood who helped to gather the field data and the RSPB Conservation Data Management Unit for providing the Landcover data.

Ancillary