Meadow pipits, red grouse and the habitat characteristics of managed grouse moors

Authors


A. A. Smith (asmith@gct.org.uk).

Summary

  • 1The abundance of meadow pipits appears to be a good indicator of the breeding density of hen harriers on moorland managed for red grouse in Scotland. High densities of hen harriers can limit grouse populations at low density and reduce shooting bags, with repercussions for grouse moor management and conservation. We therefore examined the habitat characteristics of managed grouse moors, asking whether changes in vegetation could alter the ratio of pipits, and thus harriers, to grouse.
  • 2We examined grouse abundance and habitat on 69 sites of 1 km2 in upland Britain, of which pipit abundance was studied on 36. Similar data were collected on 73 sites of 25 ha within the Langholm estate in south-west Scotland, in order to make a within-estate and among-moor comparison.
  • 3The distribution of pipit abundance on the Langholm estate was similar to that seen in the extensive among-moor study. Pipit and grouse abundance were related to habitat in similar ways in the two data sets.
  • 4Pipits were the most frequent passerine, but their abundance was not related to grouse abundance. Pipit abundance declined with increasing muirburn and heather, but increased with grass cover. In a linear regression model Calluna, Sphagnum and muirburn cover accounted for 42% of the variation in pipit abundance across Britain.
  • 5Grouse abundance was influenced by the regional location of the grouse moor and to a lesser extent by its altitude. There were more grouse on English moors with less patchy habitats. Bird species declined with increasing Calluna and Sphagnum cover and habitat patchiness. Bird species diversity increased from west to east and on moors with more muirburn.
  • 6This study provides evidence that meadow pipits are influenced by habitat characteristics on moorland managed for grouse. This work also illustrates how habitat management may be able to mitigate human–wildlife conflicts: long-term increases in heather cover and management of this habitat by muirburn on grouse moors may reduce pipit numbers and thus reduce the ratio of hen harriers to grouse. However, the effects of such vegetation change on other components of the upland bird assemblage requires further investigation.

Introduction

Heather Calluna vulgaris (L.) Hull-dominated moorland is an internationally important habitat that supports an array of threatened flora and fauna (Anonymous 1995; Thompson et al. 1995; Barr 1997). Its area has been declining since the 1940s through afforestation and heavy grazing by sheep and deer (Huntings Surveys 1986; Sydes & Miller 1988; Scottish Natural Heritage 1994; Tudor & Mackay 1995). Available data suggest that the rate of heather loss has been lowest on moorland where red grouse Lagopus lagopus scoticus (Latham) are managed for shooting (Barton & Robertson 1997). This is not surprising as heather is the main food plant of red grouse (Jenkins, Watson & Miller 1963) and the habitat is managed to maintain red grouse numbers (Hudson 1992).

Although estimates of the area managed vary, grouse moor management is an important land use in upland Britain (Bunce & Barr 1988; Hudson 1992). Grouse management principally involves heather burning, and predator control and grouse parasite control (Hudson & Newborn 1995). Hudson (1995) and Thompson et al. (1995) have proposed government financial support for grouse moor management as one way of conserving large areas of heather moorland in private ownership. However, the illegal persecution of birds of prey on grouse moors presents an obstacle to such a solution (Thompson et al. 1995). As a result of the perceived threat to grouse shooting, hen harriers Circus cyaneus (L.) in particular have suffered intense persecution in the past, and continue to do so despite receiving full legal protection (Etheridge, Summers & Green 1997; Thirgood et al. 2000a).

Recent research demonstrates that the breeding density of hen harriers on moors managed for red grouse varies widely in the absence of persecution (Redpath & Thirgood 1997; Green & Etheridge 1999; Redpath & Thirgood 1999). Based on data from six Scottish moors where legal predator control was practised, hen harriers bred at high density on moors where meadow pipits Anthus pratensis (L.) and field voles Microtus agrestis (L.) were abundant. Meadow pipits are an important component of the diet of breeding hen harriers in Scotland (Watson 1977; Picozzi 1978; Redpath 1991; Redpath & Thirgood 1997). Thirgood et al. (2000b, 2000c) also demonstrated that predation by high densities of breeding hen harriers on Langholm moor in south-west Scotland was sufficient to limit grouse densities at a level at which driven shooting was not viable. This finding has important repercussions for the management of grouse moors and the conservation of heather moorland.

A better understanding of three issues raised by this recent research may offer ways of reducing this human–wildlife conflict.

  • 1What is the relationship between the habitat characteristics of managed grouse moors and the abundance of meadow pipits, red grouse and other upland birds?
  • 2Are meadow pipit and red grouse densities correlated? If they are not, it may be possible to manipulate upland habitats to increase or maintain red grouse numbers while decreasing meadow pipit and hence hen harrier numbers.
  • 3How representative of the British uplands is the intensively studied Langholm moor in terms of the abundance of meadow pipits and therefore potential hen harrier densities? This information will be useful in predicting where and at what density hen harriers might be expected to breed in the absence of persecution.

To address these issues we examined the relationships between habitat characteristics and the abundance of red grouse and meadow pipits intensively within Langholm moor in south-west Scotland and extensively among 69 different grouse moors throughout upland Britain.

Methods

This study drew on data gathered from previous intensive and extensive studies that used dedicated habitat surveys to provide a base number of sites from which to assess bird–habitat relationships (Hudson 1992; Redpath & Thirgood 1997). Subsets of these sites were then used in analyses depending on whether dedicated survey or long-term monitoring data were available for the bird species of interest.

Within-estate

This part of the study was conducted on the Langholm estate in south-west Scotland in May–July 1996. Langholm moor was managed for red grouse and sheep and consisted of a mosaic of Calluna-, grass- (mainly Molinia) and sedge- (mainly Eriophorum spp.) dominated patches (Redpath & Thirgood 1997). Within this moor, suitable habitat was delineated on a map and initially categorized as Calluna-dominant, grass-dominant or Calluna/grass-mix, based on 1988 aerial photographs. Within each of these categories, a roughly equal number of 25-ha sites was selected randomly to give a total of 73 sites. In each site, 16 quadrats of 4-m2 were laid out in a square grid pattern with each quadrat set 100 m apart. The presence (percentage ground cover) of Calluna (Calluna vulgaris), grass (graminoids, combining sedge, grass and rush species), muirburn (burnt Calluna stalks) and Sphagnum was recorded in each quadrat. Sphagnum presence was recorded as an index of site wetness. We also recorded whether Calluna or grass was dominant in each quadrat. An index of the mosaic pattern within sites was produced by counting the number of times the dominant habitat changed between two consecutive quadrats on the grid transects, giving values ranging from 0 (homogeneous) to 15 (heterogeneous).

An index of meadow pipit abundance was obtained by counting meadow pipits on two parallel 500-m transects traversing each of the 73 sites. Birds were counted in 100-m bands either side of the transect and the data presented as the number seen km−1. Transects were surveyed on foot once between May and early June, between 06:00 and 09:00 h in dry conditions with no or light wind (Thirgood, Leckie & Redpath 1995). Red grouse were censused on 22 of the 73 sites specifically chosen to represent a range of Calluna and grass cover. It was not possible to census more sites because of time constraints. Red grouse densities in July were estimated using pointing dogs and standard methods to obtain a total count of grouse (Jenkins, Watson & Miller 1963). Four parallel transects were walked across each site, with the dog quartering the ground in front of the observer and flushing all grouse encountered.

Among-moor

Data collection for this extensive study was in three parts: estimates of red grouse density collected between 1993 and 1997; a habitat survey in 1997; and estimates of bird abundance in 1997. Data on grouse and habitat characteristics were collected on 69 moorland sites (51 estates) where there was management for red grouse (predator control and heather burning). The distribution of these sites was determined by the availability of grouse abundance data, which had been estimated on these sites for 10–15 years in association with other studies (Hudson 1992). The estates on which the sites were located were originally chosen non-randomly on the basis that they were broadly distributed throughout the British uplands and the owners were willing to give access to their heather moorland for monitoring. Survey sites were chosen specifically to represent variations in altitude, grouse density and habitat typical of the individual moor, and of the moors on the estate.

In this study, the abundance of bird species other than grouse was estimated on 36 of these sites (29 estates; Fig. 1). These sites were chosen randomly within three regions (north of England, borders, highland Scotland) from the moors with grouse and habitat data. Most sites were 1 km2 but six of the grouse sites and three of the sites censused for other bird species abundance were 0·5 km2. Variation in census results because of edge effects from different sized areas were minimized because the habitat up to 0·5 km beyond the site perimeter was similar to that within the site (Gaston, Blackburn & Gregory 1999).

Figure 1.

The location of managed grouse moors surveyed in spring 1997. Sites are divided into those where habitat and grouse data were collected (open circles) and those where meadow pipit data were also collected (closed circles).

The habitat survey was designed to provide data directly comparable with those collected at Langholm and was conducted by three observers between late February and early April 1997. The survey progressed from south to north so that vegetation development was as similar as possible among regions. The presence of Calluna, grass (graminoids), muirburn (burnt Calluna stalks) and Sphagnum (spp.) was recorded using point quadrats (Bullock 1996). For each site of 1 km2, 100 point quadrats were evenly distributed along four parallel transects of 25 points each 250 m apart. Two parallel transects were covered on sites of 0·5 km2. The mosaic pattern within sites was examined by counting the number of times the dominant habitat changed between two consecutive point quadrats.

Meadow pipit abundance was estimated during late May to late June 1997. At each site, we recorded indices of meadow pipit abundance by walking on the same morning two parallel 1-km transects, 500 m apart, in conditions similar to above. Transects were walked in 20–30 min at constant speed and all birds within 250 m either side of the transect line were recorded (Bibby, Burgess & Hill 1992). We were able to count only two transects at each site because of time constraints. Latitude and time of year may have interacted to influence meadow pipit numbers (Thirgood, Leckie & Redpath 1995), therefore sites were surveyed in a random order. We minimized multiple observer effects by using only three observers and no one observer surveyed the majority of sites in a region. We also recorded the number and abundance of other bird species during these visits. We examined the influence of habitat variables on bird species number and species diversity as described by Simpson's index (Magurran 1988). Estimates of red grouse density were obtained using trained pointing dogs in April (as above). Data on red grouse abundance were available for these sites from 1993 to 1997 and the site mean of these counts was used in analyses to account for between-year variation in numbers.

The frequency of meadow pipits and red grouse observed per count was skewed and thus log-transformed [log10(x + 1)] before analysis. Meadow pipit abundance was described as numbers seen per km transect walked, and grouse abundance as numbers counted per km2. Examination of frequency plots and plots of residuals indicated grass, muirburn and Sphagnum were non-normally distributed. These proportions were normalized using arcsine (square-root) transformations for initial testing of relationships between bird abundance and habitat and other site variables using Student's t-tests and Pearson correlations. Linear regression models were then constructed with backward elimination of the untransformed variables when P > 0·05. The resulting minimal adequate models (MAM) contained only those variables that could explain significant amounts of the deviance in pipit and grouse abundance and observed bird species diversity. We explored quadratic relationships between meadow pipits, grouse and habitat variables by including the product terms of untransformed Calluna and grass variables in the stepwise regression models. Possible interactions between Calluna and grass variables were examined by including their product (Calluna × grass) in the regression models.

Among-moor sites were grouped by region representing either English (n = 29) or Scottish (n = 41) sites. Analysis of covariance (ancova) was used to examine the significance of latitude (northing) and longitude (easting) within models that controlled for the effects of region. ancova was also used to test whether there were significant differences in the relationships between the within-estate and among-moor bird abundance and habitat data.

Results

Within-estate

Meadow pipits were observed on all 73 sites and numbers ranged from 3 to 26 km−1 (median = 11·25, interquartile = 8·0–14·0). Date of survey had no effect on meadow pipit numbers (F1,71 = 0·91, P = 0·34). Adult red grouse density was estimated on 22 sites in July and grouse were found on 18 of these sites. We examined the correlations between meadow pipit and red grouse abundance and individual habitat variables (Table 1). Meadow pipits were significantly negatively correlated with muirburn. Sites with more muirburn were associated with more Calluna (r = 0·48, P < 0·001) and with less grass (r = −0·42, P < 0·001), and Calluna and grass were correlated (r = −0·71, P < 0·001). There was no correlation between meadow pipit and red grouse abundance (r = 0·10, P = 0·6). The MAM derived from the meadow pipit abundance data and habitat variable data contained muirburn and Calluna2 (MAM r2 = 0·16, F3,69 = 4·51, P = 0·006; Table 2). This indicated that there were fewer meadow pipits as heather cover increased and fewer pipits on areas with more muirburn. There were no linear or curvi-linear relationships apparent between grouse and Calluna or grass at Langholm. Only altitude was significant in the regression model containing grouse abundance and habitat data (MAM r2 = 0·23, F1,20 = 6·03, P = 0·023; Table 2), with more grouse found at higher altitudes.

Table 1.  Within-estate: Pearson correlation of meadow pipit and red grouse abundance with altitude, Calluna, grass (graminoids), muirburn (burnt heather stalks), Sphagnum occurrence and mosaic (the number of times the dominant habitat changed between two consecutive quadrats) on Langholm moor in 1996. Significant P given in bold
 Meadow pipit (73 sites)Red grouse (22 sites)
VariablerPrP
Altitude−0·050·66  0·550·008
Calluna−0·1950·098  0·220·32
Grass  0·0790·51  0·010·96
Sphagnum−0·120·31−0·060·79
Muirburn−0·330·004−0·130·56
Mosaic−0·070·57−0·260·25
Table 2.  Within-estate: linear regression analysis explaining variation in the abundance of meadow pipits (r2 = 0·16) and red grouse (r2 = 0·23) on Langholm moor in 1996. Slope is the regression coefficient for each variable. P-values for variables contributing to the minimal adequate models are shown in bold
 Meadow pipitRed grouse
Explanatory variableSlopetPSlopetP
Altitude−0·001−0·210·83  0·002  6·030·02
Calluna−0·341−0·340·73  0·114  0·220·83
Calluna2−0·688−1·950·05−2·539−1·140·27
Grass−0·038−0·460·65  0·294  0·490·63
Grass2−0·305−1·420·16−0·611−0·150·89
Calluna × grass  0·597  1·440·15  0·499  0·170·86
Sphagnum−0·248−0·640·53−0·125−0·060·96
Muirburn−3·858−2·980·004−3·569−0·600·56
Mosaic−0·007−0·860·39−0·022−0·560·59

Among-moor

Red grouse and meadow pipit count sites were located on managed grouse moors where Calluna presence varied between 46% and 94%. Meadow pipit abundance ranged from 1·5 to 30·5 km−1. There were no between-observer effects for the number of pipits recorded (F2,33 = 2·32, P = 0·11). Similarly, date of survey had no effect on meadow pipit numbers (F1,34 = 0·08, P = 0·78). Meadow pipit abundance at Langholm during this survey was 13·0 km−1, compared with a median of 11·3 km−1 (interquartiles 7·5–14·4) obtained during the within-estate study (Fig. 2).

Figure 2.

Frequency distribution of meadow pipit counts conducted on a range of grouse moors throughout upland Britain. The medians of these data (*) and of the data from Langholm estate during the within-estate (1) and between-estate (2) surveys are indicated.

We initially examined relationships between meadow pipits, red grouse and the habitat variables separately (Table 3). There was no relationship between red grouse and meadow pipit densities (r = 0·11, P = 0·53). Examination of plots of the original data and residuals suggested there were no non-linear relationships between meadow pipit or grouse numbers and the habitat variables. Meadow pipits were negatively correlated with muirburn (Fig. 3) and Calluna and positively with grass (Fig. 4), suggesting sites with more grass, less heather and less burning had more meadow pipits. Sites with more heather cover had less grass cover (r = −0·40, P = 0·001) but the amount of muirburn was not related to the presence of Calluna (r = −0·21, P = 0·09) or to the amount of grass (r = −0·01, P = 0·93). The geographical location of the sites was most closely related to grouse numbers, with moors in the south and east holding the most grouse (Fig. 5). These relationships between location and grouse densities occurred because the mean population density of grouse was almost 50% higher on grouse moors in England than those in Scotland. Once region was controlled for, neither easting (ancova, F1,66 = 0·03, P = 0·87) nor northing (ancova, F1,66 = 0·11, P = 0·74) had any effect on grouse densities. We therefore entered only region as a dummy variable into the regression analysis.

Table 3.  Among-moor: relationships between various geographical and vegetation factors and the numbers of meadow pipits and red grouse counted in 1997. Statistics given are Student's t and Pearson correlation coefficients, significant P given in bold
 Meadow pipit (36 sites)Red grouse (69 sites)
  • *

    Grouse densities are significantly higher in England than in Scotland.

VariableStatisticPStatisticP
Regiont = −0·61    0·55t = 0·77< 0·001*
Eastingr = −0·27    0·11r = 0·5< 0·001
Northingr = 0·03    0·86r = −0·55< 0·001
Altituder = −0·09    0·60r = 0·11     0·37
Callunar = −0·33    0·05r = −0·12     0·34
Grassr = 0·54< 0·001r = −0·12     0·16
Sphagnumr = −0·28    0·1r = −0·07     0·57
Muirburnr = −0·43    0·009r = 0·004     0·97
Mosaicr = 0·25    0·14r = −0·15     0·22
Figure 3.

The abundance of meadow pipits per km of transect in relation to the presence of muirburn (r2 = 0·19) in the among-moor study.

Figure 4.

Relationship between the presence of grass and the abundance of red grouse per km2 (r2 = 0·02) and meadow pipits per km of transect (r2 = 0·26) in the among-moor study.

Figure 5.

Relationship between the abundance of red grouse per km2 in the spring and latitude. Sites in England had higher grouse densities than sites in Scotland.

When the bird abundance data, habitat data and region variable were put into linear regression models the MAM explained 42% of the variation in pipit numbers (MAM F3,32 = 7·79, P < 0·001; Table 4). Muirburn explained 19% of this variation, Calluna a further 14% and Sphagnum an additional 9%, indicating there were more pipits on dry moors with low heather cover and little muirburn. The linear and quadratic terms describing grass and the interaction term of grass and Calluna variables were not significant in either the pipit or grouse models (Fig. 4 and Table 4). There were more red grouse on moors in England than in Scotland, region being the only significant variable explaining 34% of the variation in grouse density (MAM F1,67 = 34·39, P < 0·001; Table 4). A linear regression incorporating the same variables from just the English sites indicated that grouse abundance declined as habitat patchiness increased, mosaic explaining 16% of the variation in grouse abundance (MAM F1,28 = 4·87, P = 0·04). No explanatory variables entered the same model using just the Scottish sites.

Table 4.  Among-moor: linear regression analysis explaining variation in the abundance of meadow pipits (r2 = 0·42) and red grouse (r2 = 0·34) counted on moorland managed for red grouse across Britain in 1997. Region was entered as a categorical dummy variable with Scottish moors coded ‘0’ and English moors ‘1’. Slope is the regression coefficient for each variable. P-values for variables contributing to the minimal adequate models are shown in bold
 Meadow pipitRed grouse
Explanatory variableSlopetPSlopetP
Region−0·053−0·770·45−0·32834·39< 0·001
Altitude−0·001−0·630·54  0·001  1·362   0·18
Calluna−0·902−2·830·008−0·003−0·01   0·99
Calluna2  0·098  0·040·97  1·327  0·67   0·51
Grass  0·726  1·380·18  0·152  0·45   0·66
Grass2−5·680−1·140·26−0·436−0·14   0·89
Calluna × grass  5·432  1·110·27−2·976−0·79   0·43
Sphagnum−2·509−2·300·03−0·171−0·31   0·76
Muirburn−1·317−3·560·001−0·131−0·46   0·65
Mosaic−0·002−0·360·72−0·001−0·00   0·99

We recorded 36 bird species other than red grouse and meadow pipits on the among-moor surveys, although six of these species were each recorded on only one moor. We carried out separate correlations of bird species number and the diversity index with the main explanatory variables (Table 5). Species diversity increased from west to east and from north to south. Easting had a positive effect on diversity when region was controlled for (F1,33 = 4·70, P = 0·04), so region and easting were kept in the full model. Of the vegetation characteristics tested separately, Calluna and Sphagnum were negatively related with species number, while muirburn was positively correlated with diversityand grass cover negatively correlated with diversity. More bird species were present on drier moors with less heather and lower habitat patchiness, Sphagnum accounting for 15% of the variation, Calluna explaining 9% of the variation and mosaic explaining a further 11% of variation (MAM F3,32 = 5·80, P = 0·003; Table 6). Diversity of bird species increased from west to east and was greater on moors with more muirburn. Easting explained 22% of the variation in the diversity index, with muirburn explaining a further 9% of variation (MAM F2,33 = 7·43, P = 0·002; Table 6).

Table 5.  Relationship between various geographical and vegetation factors and the numbers of bird species and Simpson's diversity index across Britain in 1997. Statistics given are Student's t and Pearson correlation coefficients, significant P given in bold
 Species number (36 sites)Diversity (36 sites)
VariableStatisticPStatisticP
Regiont = 1·500·14 t = 2·120·04
Eastingr = 0·290·09 r = 0·470·004
Northingr = −0·310·07r = −0·350·04
Altituder = −0·280·09r = −0·160·35
Callunar = −0·460·004r = −0·280·11
Grassr = 0·090·62r = −0·330·05
Sphagnumr = −0·390·02r = −0·310·07
Muirburn r = 0·290·09 r = 0·440·007
Mosaicr = −0·030·86r = −0·190·27
Table 6.  Results from a linear regression analysis (stepwise entry, P to enter = 0·05) of factors explaining variation in bird species count (r2 = 0·35) and bird species diversity (r2 = 0·31) counted on moorland managed for red grouse across Britain in 1997. Region was entered as a categorical dummy variable with Scottish moors coded ‘0’ and English moors ‘1’. Slope is the regression coefficient for each variable. P-values for variables contributing to the minimal adequate models are shown in bold
 Bird speciesDiversity
Explanatory variableSlopetPSlopetP
Region     0·55  0·730·47   0·046   0·090·93
Altitude    −0·007−2·020·06−0·001−0·420·68
Easting    −0·001−1·780·08   0·001   2·400·02
Calluna −17·343−3·170·003−0·217−0·160·87
Calluna2−37·361−1·280·21−8·256−0·800·43
Grass      0·852  0·170·87   0·578   0·320·75
Grass2−83·464−1·450·16   2·175   0·180·86
Calluna × grass   73·035  1·410·1714·114   0·680·50
Sphagnum −24·925−2·080·05−5·812−1·230·23
Muirburn       6·779  1·680·10  3·268   2·040·05
Mosaic    −0·135−2·350·03−0·013−0·960·35

Comparison of within-estate and among-moor analyses

Meadow pipit abundance was explained by both Calluna and muirburn in the within-estate and among-moor analyses. Sphagnum appeared as a significant explanatory variable of pipit abundance only in the among-estate regression model (Table 4). Muirburn was significantly negatively correlated with pipit abundance in both data sets and the slopes of the linear regression lines were not significantly different (ancovaF1,105 = 0·02, P = 0·89). In both analyses meadow pipit abundance was negatively associated with Calluna but the association was curvi-linear within-estate and linear among-moor (Fig. 6). This may be because none of the among-moor sites had < 40% heather cover. Sites with between 0% and 40% heather cover in the within-estate data set were not associated with increasing or decreasing pipit abundance (r = 0·14, P = 0·34). On sites with > 40% heather cover a comparative analysis of both data sets showed similar negative relationships between pipits and Calluna within-estate and among-moor (ancovaF1,60 = 0·73, P = 0·73). The analyses of red grouse abundance within-estate and among-moor showed no similarities. The explanatory power of altitude in the within-estate model was replaced by region and, on English moors, habitat patchiness in the among-moor models.

Figure 6.

Relationship between meadow pipit abundance per km of transect and the presence of Calluna within-estate and among-moor.

Discussion

Bird–habitat relationships

Meadow pipit abundance varied between 1 km−1 and 30 km−1 of transect across a range of managed grouse moors in Britain. We found meadow pipit abundance on Langholm moor was not unusually high in this national context. Therefore, if the relationship between meadow pipit abundance and the breeding density of hen harriers shown by Redpath & Thirgood (1999) is widely applicable, our data suggest that the hen harrier densities seen at Langholm (up to 20 nesting females in 40 km2 of heather moorland) could occur on other moors in the absence of illegal persecution, provided suitable hen harrier nesting habitat was available (Redpath et al. 1998).

Meadow pipit abundance throughout the study was negatively associated with the amount of heather cover and the amount of muirburn. Increasing amounts of muirburn on Langholm were associated with more heather and less grass, while muirburn and heather cover were not correlated in the among-moor study. Thus there was some evidence that muirburn may have been influencing pipit abundance independent of this management techniques’ typical association with heather. Variation in meadow pipit abundance among moors was similar to pipit abundance on Langholm moor only in the range 40–100% heather cover. The results of both analyses showed that sites where Calluna was dominant and where most burning had taken place contained the fewest meadow pipits. Estimates of meadow pipit and red grouse numbers were not correlated either within-estate or among-moor and differences in vegetation could account for little of the variation in red grouse abundance. Red grouse were more abundant on English moors and there was evidence that they were also more abundant at higher altitudes.

As heather cover declined, grass cover increased both within-estate and among-moor. Meadow pipits utilize grassland for nesting cover and food and during the breeding season feed largely on Tipulids, Coleoptera and Diptera (Coulson & Whittaker 1978; Walton 1979; Allen 1995). Coulson & Whittaker (1978) found that Tipulids were abundant on blanket bog for short periods of time, whereas peaks in Tipulid abundance on grassland areas were less pronounced but lasted longer through the summer. Meadow pipits tended to nest on blanket bog next to grassland, apparently so that they could utilize the short peak in food abundance on the bog and then move to feeding in the grasslands (Coulson & Whittaker 1978). The longer periods of food availability on grasslands may account for the higher densities of meadow pipit observed in this habitat during the current study. Meadow pipits nest under cover at ground level and tussocks of rough grass may also be important in providing nest sites (Coulson & Whittaker 1978). Molinia, Eriophorum and Nardus were the most common graminoids recorded during the current study, and the first two often had rank growth forms, associated with light grazing. The impact of grass may extend into heather-dominated sward, which often has a grass component. Muirburn of these heather swards has been shown by some studies to reduce invertebrate diversity as well as removing habitat structure (Gimingham 1985). This may explain why muirburn was negatively associated with pipit abundance. Further research is needed on the feeding and nesting ecology of meadow pipits in the uplands and on the abundance of their invertebrate prey on moorland of varying habitat composition, so that the mechanisms driving the observed patterns in meadow pipit abundance can be understood.

A number of studies have examined associations between meadow pipit abundance and upland vegetation at regional scales. Meadow pipits were positively associated with heather moorland in the peatlands of Caithness and Sutherland (Stroud et al. 1987) and on the heather and grass grouse moors of the south Pennines (Yalden 1984; Haworth & Thompson 1990; Stillman & Brown 1994), but showed no significant habitat associations in the eastern Highlands (Brown & Stillman 1993). In reviewing these studies of the habitat associations of meadow pipits and other upland birds, Stillman (1995) suggests that because few species show strong positive associations with heather moorland other factors are more important in determining upland bird distribution and abundance. However, these regional surveys of upland breeding birds are not directly comparable with the current study as the vegetation surveys described broader habitat types than considered in the present study.

The absence of strong relationships between the abundance of red grouse and habitat characteristics either within-estate or among-moor may be because only a small proportion of available heather biomass is consumed by the birds (Jenkins, Watson & Miller 1963; Savory 1978). In addition, red grouse abundance is known to be affected by a variety of other factors, including heather nutrient status, intensity of parasite infection and abundance of predators, which were not taken into account in the current study (Moss 1969; Miller & Watson 1978; Hudson, Dobson & Newborn 1992; Thirgood et al. 2000b, 2000c). The patchiness of the moors as measured by the mosaic variable was only associated with the abundance of red grouse in England. This may have been because factors other than habitat and environment are more important in determining grouse abundance in Scotland (Hudson, Dobson & Newborn 1992). Differences in red grouse abundance between Scotland and England have been reported elsewhere, and are thought to be due to higher densities of gamekeepers in England that result in improved predator control and habitat management, although warmer June temperatures and increased productivity of heather may also be important factors (Hudson 1992; Smith, Redpath & Campbell 2000).

Three caveats should be placed on the findings of the current study. First, the data on meadow pipit abundance were collected in only one year. Meadow pipit numbers are known to vary between years and further research is needed to see how this temporal variation is related to vegetation. Secondly, the bird abundance data were collected from managed grouse moors where predators were rigorously controlled. Grouse densities would probably be lower in the absence of such control (Hudson 1992). Thirdly, as a consequence of working on managed grouse moors, we have little information on the densities of red grouse and meadow pipits on moors where grass is the dominant vegetation. However, we do have data collected from a range of sites within Langholm moor where habitats vary widely from grass to heather. These latter data indicate that there is a quadratic relationship between Calluna cover and pipit abundance, with the highest densities occurring on sites where heather and grass cover similar areas. If this pattern holds among-moor we might expect moors with an even mixture of Calluna and grass to have the highest densities of meadow pipits. The data on red grouse abundance within Langholm showed no significant relationships with any of the measured habitat variables. However, red grouse are reliant on heather for food, thus sites with no heather will have no grouse, as reflected in the regional-scale studies of upland bird habitat associations (Haworth & Thompson 1990; Brown & Stillman 1993; Stillman & Brown 1994).

Conservation implications

Our finding that meadow pipit abundance is correlated with habitat characteristics of moorland managed for red grouse has important implications for the management of these areas, as meadow pipit abundance appears to influence the settling densities of hen harriers (Redpath & Thirgood 1999). Modifying moorland vegetation by increasing heather cover and muirburn management may lead to fewer meadow pipits and therefore fewer hen harriers (Fig. 7). Such changes in vegetation may also have effects on the abundance of other species, and this study suggests that these vegetation changes may be deleterious to other components of the upland bird assemblage. Redpath & Thirgood (1999) found that much of the annual variation in breeding numbers of hen harrier within individual moors was explained by changes in the abundance of field voles. Redpath & Thirgood (1997) caught more field voles in grass-dominated areas than in either heather/grass-mixes or in heather-dominated areas. Field voles forage on fresh grass shoots, therefore decreases in vole abundance can be expected with reductions in grass on moorland sites (Hewson 1982). Reductions in field vole abundance may also affect a range of other vole predators, such as kestrel Falco tinnunculus (Village 1990), short-eared owl Asio flammeus (Lockie 1955; Village 1987), stoat Mustella ermina (McDonald & Harris 1997) and red fox Vulpes vulpes (Hewson 1984, 1986; Leckie et al. 1998).

Figure 7.

Summary of relationships between (a) meadow pipit abundance and heather cover and (b) meadow pipit abundance and male harrier breeding density. The lines for the relationship between pipits and heather cover are based on data in Fig. 4. The data in the pipit/harrier figure are taken from Redpath & Thirgood (1999) with a linear regression line (95% confidence interval) drawn between the points.

Heavy grazing by sheep and deer is associated with the loss of Calluna and a shift towards grasses such as Nardus and Molinia (Welch 1986; Hester et al. 1999). These typically long-term changes in habitat coupled with locally high raptor predation have been shown to explain the decline and restriction of a red grouse population (Thirgood et al. 2000c). Removal of sheep or reduction in grazing in upland areas can lead to increased vegetation height followed by increases in Calluna cover and decline in grasses (Ball 1974; Rawes 1981; Hill, Evans & Bell 1992; Hope et al. 1996). Reducing grazing pressure may therefore lead to fewer meadow pipits and hen harriers (Thirgood et al. 2000a). However, the effects of such management may only be witnessed over a number of decades. Data from the current study and from studies elsewhere suggest that reducing grazing to stimulate a change from grass- to heather-dominated moorland may initially increase meadow pipit and field vole abundance before these species decline as grass cover decreases (Hill, Evans & Bell 1992; Hope et al. 1996). Light grazing during the summer months and controlled heather burning may help speed up the process from grass- to heather-dominated moorland, although further experimental research is required to assess the effects on the associated fauna (Hudson & Newborn 1995).

The number of bird species seen on grouse moors was negatively related to the presence of Sphagnum spp., suggesting that increasing wetness is associated with fewer bird species. Fewer species were seen on moors with more heather cover and increasing patchiness. The relationship between the bird diversity index and easting and muirburn may also reflect increasing bird community richness with decreasing wetness. The literature on upland bird habitat associations suggests that the following species are associated with heather moorland: red grouse, black grouse Tetrao tetrix, merlin Falco columbarius, hen harrier, short-eared owl, whinchat Saxicola rubetra and stonechat Saxicola torquata, with some evidence of a positive association with heather for twite Acanthis flavirostris and ring ouzel Turdus torquata (Stroud et al. 1987; Haworth & Thompson 1990; Brown & Stillman 1993; Usher & Thompson 1993; Stillman & Brown 1994; Brown & Bainbridge 1995; Thompson et al. 1995). Other birds, including most waders, were more influenced by altitude, topography and surrounding land use than vegetation variables. It is not known currently how reducing the grass sward within heather-dominated moorland would influence these upland birds. Appropriate studies will need to take into account the effects of burning, grazing and predator control in addition to the effects of vegetation change.

In conclusion, the data collected in this study suggest that the habitat characteristics of managed grouse moors influence the abundance of meadow pipits and may therefore also influence the breeding densities of hen harriers. In contrast, habitat characteristics have little effect on the abundance of red grouse, at least within the range of habitats studied. These data suggest that habitat management to increase heather cover may in the long term help to reduce conflicts between hen harriers and red grouse. Further research is required to determine whether these relationships are applicable temporally as well as spatially and to examine in detail the mechanisms by which meadow pipits and red grouse are influenced by habitat. The impact of habitat change on other upland birds and on moorland biodiversity also needs to be assessed.

Acknowledgements

We thank the landowners and gamekeepers who gave permission to work on their moors and assisted with access. R. Foster, D. Howarth and D. Newborn collected much of the grouse census data. P. Robertson and P. Hudson helped initiate this project and, together with N. Aebischer, D. Baines, A. Brown, R. Green, S. Hartley, I. Newton, G.R. Potts and D. Welch, made helpful comments on previous drafts. This study was funded by the Department of the Environment, Transport and Regions (CRO196) and The Game Conservancy Scottish Research Trust.

Received 29 April 2000; revision received 8 September 2000

Ancillary