An experimental study of the effects of predation on the. breeding productivity of capercaillie and black grouse

Transcription

1 Ecology , An experimental study of the effects of predation on the Blackwell Publishing, Ltd. breeding productivity of capercaillie and black grouse R. W. SUMMERS*, R. E. GREEN, R. PROCTOR*, D. DUGAN*, D. LAMBIE*, R. MONCRIEFF*, R. MOSS and D. BAINES *Royal Society for the Protection of Birds, Etive House, Beechwood Park, Inverness IV2 3BW, UK; Royal Society for the Protection of Birds and Conservation Biology Group, Department of Zoology, University of Cambridge, Downing Street, Cambridge CB2 3EJ, UK; Centre for Ecology and Hydrology, Hill of Brathens, Banchory, Kincardineshire AB31 4BW, UK; and Game Conservancy Trust, The Gillett, Forest-in-Teesdale, Barnard Castle, Co. Durham DL12 0HA, UK Summary 1. The capercaillie Tetrao urogallus and black grouse Tetrao tetrix are declining in the UK, and low breeding success has been identified as the key factor in the decline of the former. To investigate possible causes, breeding productivity was studied in relation to predation, weather, vegetation changes and deer numbers over an 11-year period ( ) within native pinewood at Abernethy Forest, Scotland. The abundance of predators (crows Corvus corone and red foxes Vulpes vulpes) was experimentally manipulated in by culling. Productivity (chicks reared per female) was compared between forests with and without experimental predator management. 2. During predator control, the number of breeding crows was reduced from 10 pairs to one. The attempted reduction in red fox abundance was unsuccessful; only small numbers of adults were killed, and neither scat nor den counts declined significantly. 3. Predation on artificial nests containing six hen eggs and a hen egg filled with wax was measured as an index of predator activity from 1991 to Predation was lowest during the last three years of predator control, Predators could sometimes be distinguished by signs on depredated eggs. Predation on artificial nests by crows was highest during However, after predator removal stopped in 1997 few crows returned, and increased predation on artificial nests did not involve increased signs of crow predation. Pine marten Martes martes numbers increased during the study period and became significant predators of artificial nests. 4. The total number of capercaillie eggs and nests depredated by crows was estimated from the number of depredated capercaillie eggs found and the proportion of crowpredated hen eggs in artificial nests. The values ranged from 18 to 158 eggs over 3 years, equivalent to 3 23 capercaillie nests year Capercaillie productivity was low (< 1 chick per female) during and but higher during Compared with nine other forests in Scotland, changes in capercaillie productivity at Abernethy were different. Productivity at Abernethy was negatively related to June rainfall, and to the minimum daily predation rate on artificial nests by crows. There was also a significant interaction in that capercaillie were most productive when low rainfall coincided with low predation by crows on artificial nests. 6. The productivities of black grouse and capercaillie were positively correlated, but greater in the former. As in capercaillie, black grouse productivity was negatively related both to June rainfall and the minimum daily predation rate on artificial nests by crows, and there was an interaction. 7. Synthesis and applications. The long-term increase in crows and red foxes and the predicted increase in rainfall in Scotland may have negative effects on capercaillie and black grouse. In the short term, control of crows is likely to improve productivity. In the long term, increased woodland size and some reversal of fragmentation might decrease the access to woodland of predators associated with the interface between farmland and woodland. Ecological Society Key-words: artificial nests, crow, rainfall, red fox Ecology (2004) 41,

2 514 R.W. Summers et al. Introduction The capercaillie Tetrao urogallus L. and black grouse Tetrao tetrix L. are two of the fastest declining bird species in the UK (Gregory et al. 2002). For capercaillie, low breeding success associated with a changing pattern of spring temperatures and reduced chick survival in wet summers are possible causes, aggravated by fully grown birds colliding with deer fences (Moss 1986; Moss et al. 2000; Moss, Oswald & Baines 2001). For black grouse, agricultural changes on marginal land, expansion and growth of conifer plantations and increases in red deer Cervus elaphus L. and sheep Ovis aries L. numbers could all have affected the habitat (Baines 1996), while collisions with deer fences may have increased mortality (Baines & Summers 1997). Elsewhere, predation can reduce the productivity of capercaillie and black grouse (Marcström, Kenward & Engren 1988). Brood counts of capercaillie in the early years ( ) of the ownership of Abernethy Forest, Scotland, by the Royal Society for the Protection of Birds showed that productivity was too low to maintain adult numbers (Moss et al. 2000). Depredated capercaillie and black grouse eggs suggested that nest predation might be the cause. No predator control was initially carried out. This study aimed to assess which predators depredated capercaillie and black grouse eggs and if a reduction in predators would lead to improved productivity at Abernethy Forest. We compared years with and without predator culling. In addition, productivity was monitored in other forests with no change in predator control policy. In 1992, predator removal (crow Covus corone L. and red fox Vulpes vulpes (L.)) was initiated to try to reduce the levels of nest and chick predation. Crows and red foxes are both generalist predators whose numbers may be influenced by the availability of food outside the forest, with birds forming a minor part of the diet (Newton 1998). The breeding productivities of capercaillie and black grouse were related to the abundance of predators and other environmental variables. To allow for the possibility that annual changes in capercaillie productivity at Abernethy were due to wider regional effects, reference sites across the Scottish range of the capercaillie were surveyed to measure levels of productivity elsewhere. These spatial variations in productivity are examined elsewhere (Baines, Moss & Dugan 2004). Predator control was not the only change in management at Abernethy over the study period. Red deer numbers were reduced to encourage tree regeneration, resulting in shrubs being less frequently browsed. This might provide greater feeding opportunities for herbivorous insects (Baines, Sage & Baines 1994), which are eaten by capercaillie chicks (Kastdalen & Wegge 1985; Spidsø & Stuen 1988). In order to evaluate the effect of deer culls, the density of deer in the forest was estimated Correspondence: R. W. Summers, Royal Society for the Protection of Birds, Etive House Beechwood Park, Inverness IV2 3BW, UK ( ron.summers@rspb.org.uk). and shrub heights measured. Further, the amount of deer fences was reduced at Abernethy during the 1990s. Both chicks and adult capercaillie collide with fences, and such collisions are a major mortality factor in the latter (Moss et al. 2000). Therefore, changes in the lengths of fence line were also taken into account during the study. Methods STUDY SITES Abernethy Forest is the largest ancient native pinewood remaining in Scotland. The wooded area is about 33 km 2 and the trees are almost entirely Scots pines Pinus sylvestris L. The study was carried out in about 14 km 2 of the southeast part of the forest where the stands are primarily of native woodland with a shrub layer of heather Calluna vulgaris (L.) Hull, bilberry Vaccinium myrtillus L. and cowberry Vaccinium vitis-idaea L. (Summers et al. 1997). Capercaillie productivity was also estimated at nine reference sites (Fig. 1; Baines, Moss & Dugan 2004). While it was unlikely that habitat and predator control remained constant on these sites over the study period, we assumed that changes would not be contemporaneous with those at Abernethy while the number of sites was sufficiently large to provide otherwise comparable reference data. PRODUCTIVITY OF CAPERCAILLIE AND BLACK GROUSE In late July or early August, four to eight people walked line-abreast with three to seven trained dogs through different areas of Abernethy Forest covering km 2, to flush female capercaillie and black grouse and count associated chicks. The average number of chicks per female (including those with no chicks) gave a measure of productivity. Smaller teams of people and dogs were used in the reference sites in an attempt to locate at least 10 females at each forest for productivity estimates. At this time of year, the chicks are well grown and suffer little additional mortality before fledging, so the productivity estimates should be close to the number of independent young raised per female (Moss et al. 2000). To obtain an annual index of productivity from the reference sites, the following model was fitted using GLIM4 (Aitkin et al. 1990). The number of chicks per female at site i in year j was taken to be the product of a site factor g i and a year factor k j : chicks/female = g i k j, or log(chicks) log(females) = log(g i ) + log(k j ), and log(chicks) = log(females) + log(g i ) + log(k j ) The number of chicks was taken as the dependent variable with a Poisson error distribution, log(females) was an offset variable, and a log-link function was used

3 515 Breeding productivity in capercaillie and black grouse Fig. 1. The location of Abernethy Forest and other study sites in Scotland. 1, Abernethy Forest; 2, Nethybridge; 3, Carrbridge; 4, Cairngorm; 5, Ballater; 6, Aboyne; 7, Alford; 8, Aberfeldy; 9, Perth; 10, Auchterarder. (Aitkin et al. 1990). The year effects were obtained by back-transforming the year coefficients in the model to obtain the k j. The year effects were then scaled so that they had the same mean as the average breeding success (total chicks/total females) in all years at all sites. The productivities of capercaillie and black grouse were related to other variables using Poisson regression (Aitkin et al. 1990). The y variable was number of chicks, the natural logarithm of the number of females was an offset variable, and a scale parameter (residual deviance/ residual d.f.) accounted for overdispersion. Forward stepwise regression was used to examine combinations of x variables and two-way interactions, with each variable tested in turn against the response variable. The variable resulting in the most significant reduction in deviance was selected at each stage. ESTIMATES OF PREDATOR ABUNDANCE IN ABERNETHY FOREST Territorial crows were counted each spring by listening for calling birds during early morning surveys throughout the study area. Indices of abundance of red foxes were based on scat counts along 30 9 km of tracks. All counts and removal of scats were made by the same observer in April and July of each year. Only the numbers of scats accumulated between April and July were analysed because red foxes are most likely to affect breeding capercaillie and black grouse during this period. Changes in scat numbers were modelled using Poisson regression, accounting for track sections before testing for the effect of year. We searched for occupied red fox dens at 63 potential sites known since 1991 (single holes, groups of holes and piles of stones). Each was checked in early May and June. The number of incidental sightings of pine martens Martes martes (L.) per annum by eight observers during the course of management and research work gave an index of their abundance. All observers were present throughout the study period and had similar activities in the forest each year. PREDATOR CONTROL Crows were controlled during spring and summer from 1992 to 1996, largely by shooting and by using Larsen traps set in territories and areas visited by crows. Red foxes were controlled from 1992 to 1996 between February and July by shooting. A change was made to the usual deer culling practice at Abernethy of leaving the viscera where the animal was shot. From autumn 1992, the viscera were taken away, thus denying crows and red foxes a potential food supply. PREDATION RATES ON ARTIFICIAL NESTS An annual index of potential nest predation was obtained at Abernethy Forest from the rate of loss of artificial clutches of hen eggs. No attempt was made to mimic actual grouse nests. Forty-eight clutches comprising six fresh eggs and a wax-filled egg were placed on moss and/or bilberry on the ground in the same localities along two transects (5 25 and 4 35 km long) at the beginning of May and checked weekly for 4 weeks, the approximate incubation period of capercaillie (Cramp

4 516 R.W. Summers et al. & Simmons 1980). Each wax-filled egg had embedded string tied to a stake in the ground so that predators could not easily remove the egg. Signs left in the wax helped to identify the predator. The hen eggs were individually marked so that they could be attributed to a given nest if removed by a predator and the shell fragments found elsewhere. Daily predation rates by all predators, and minimum daily predation rates by crows and mammals, were calculated using the method of Mayfield (1975). A nest was judged to have been lost to predators when one or more of the hen eggs was damaged or removed. Standard errors were based on Johnson (1979). ESTIMATES OF NUMBERS OF CAPERCAILLIE EGGS DEPREDATED During and after the capercaillie nesting period (May June), the shells of depredated capercaillie and hen eggs were found incidentally during other work. Given that we knew how many hen eggs were taken from artificial nests, we could estimate the total number of capercaillie eggs depredated by crows from the ratio of depredated capercaillie and hen eggs found, using the Lincoln Index. We assumed that the proportion of wild capercaillie eggs taken by crows that we found subsequently was the same as the equivalent proportion for hen eggs taken by crows from artificial nests. The number of capercaillie eggs taken by all predators combined could not be estimated because we did not find depredated hen eggs away from artificial nests except for those depredated by crows. Some or all of the depredated hen eggs attributed to unknown predators may have been taken by crows, so we calculated minimum and maximum estimates assuming that either none or all of the hen eggs taken by unknown predators had been taken by crows. Dividing by the average capercaillie clutch size of seven (Proctor & Summers 2002) gave a minimum estimate for the number of capercaillie nests depredated. For this analysis, we excluded hen egg shells found at or near (< 20 m) artificial nest sites during nest checking. IDENTIFYING PREDATORS The results of egg predation were regularly found in the form of egg shell fragments. To identify the predator from the fragments, we compared eggs depredated by animals in captivity. Fresh hen eggs were put in the cages of captive red foxes (two cages, four animals), pine martens (four cages, four animals) and crows (two cages, four birds). The hen eggs were similar in size to capercaillie eggs (Cramp & Simmons 1980). Each day, shell fragments were collected and washed to remove dirt and any remaining egg contents, and dried. Shell remains comprised one or more broken fragments held together by the inner membrane. The fragments were separated and the mass of the largest fragment recorded to the nearest 0 1 g. In addition, shells of hen eggs filled with wax were given to captive red foxes, pine martens and crows to determine the marks they left. At many depredated artificial nests, all the hen eggs were taken and the wax egg was untouched, making it impossible to identify the predator. In 1999 and 2000, cameras with infrared trigger beams (Trailmaster system, Lanexa, Kansas, USA) were set at artificial nests in summer. This work was carried out after the assessment of the predation rate on artificial nests. DEER DENSITY Deer were counted in the south-east part of Abernethy Forest, an area of km 2. Counts were carried out in March and October each year from 1989 to 1999 by observers walking line-abreast through the forest counting flushed deer. Poisson regression was used to examine the effects of season and year, both being treated as factors. SHRUB HEIGHT Four widely spread sites within Abernethy Forest were selected for sampling potential effects that deer may have had on the shrubs (Ordnance Survey grid references NJ017157, NJ034154, NJ and NJ025135). At each site, heather and a mixture of bilberry and cowberry were sampled in August when bilberry was still in leaf. Heights of the shrubs were measured at the point where a polystyrene disk (24 cm in diameter and weighing 30 g) sliding on a pole rested on the vegetation. Twenty measurements were taken at 1-m intervals along fixed transects through each shrub type at each site. RAINFALL AND TEMPERATURE Data on rainfall in June were obtained from a rain gauge at Grianan, Tulloch (grid reference NH975165) on the western edge of Abernethy Forest. Moss (1986) found that capercaillie productivity was inversely related to the number of days of rain in early June. Therefore, the rainfall data were grouped into 10- and 20-day periods to examine potential effects of the time of rainfall through June. Annual June rainfall data were also obtained from Grantown-on-Spey, which has the longest running weather station close to Abernethy (12 km north), to examine long-term trends. To test for a relationship between productivity and spring temperatures (Moss, Oswald & Baines 2001), the average mid-april (11 20 April; Apr T 2 ) and late May (11 31 May) air temperatures were obtained for the years from Aviemore, 12 km south-west of Abernethy. Further, the mid-april temperatures relative to the early (1 10 April; Apr T 1 ) and late (21 30 April; Apr T 3 ) April temperatures were expressed in the equation: [(Apr T 2 Apr T 1 ) (Apr T 3 Apr T 2 )]/2 as an index of the timing of April warming.

5 517 Breeding productivity in capercaillie and black grouse Results CAPERCAILLIE AND BLACK GROUSE PRODUCTIVITY The annual productivity of capercaillie at Abernethy Forest ranged from 0 to 2 9 chicks female 1 (mean 0 71, n = 11; Fig. 2). Poisson regression showed that there was a significant interaction between site and year when Abernethy was compared with the reference sites (χ 2 = 166 8, d.f. = 10, P < 0 001; Fig. 3). During and , productivity at the reference sites was higher than at Abernethy Forest but during the opposite was true. This shows that the higher productivity at Abernethy Forest during was independent of the general pattern of productivity across Scotland. The productivity of black grouse at Abernethy Forest ranged from 0 4 to 4 7 chicks female 1 (mean 2 17, n = 11; Fig. 2), greater than that for capercaillie. Annual variations in black grouse productivity were significantly correlated with capercaillie productivity (r s = 0 84, P < 0 005, n = 11). PREDATOR ABUNDANCE AND CONTROL Crows The number of crows killed in the study area was similar between 1992 and 1994, and declined in 1995 and 1996 (Table 1). However, there was a large increase in the number of crows killed in the reserve as a whole in The number of breeding territories declined during , and there was little recovery of the breeding numbers after predator control stopped (Table 2). Red foxes The number of adult red foxes killed annually during the period of predator control varied between three and seven, whilst the number of cubs killed varied between one and 18 (Table 1). The index of abundance of red foxes based on numbers of scats in summer declined between 1992 and 1999 (r s = 0 79, P < 0 05, n = 8; Table 2). However, there was no significant change in abundance during Fig. 2. Annual breeding productivities (chicks per female) of capercaillie (circles) and black grouse (squares) at Abernethy Forest. Predator control took place between 1992 and Numbers of female black grouse (above) and capercaillie (below) are shown. Fig. 3. Breeding productivity (chicks per female) of capercaillie at Abernethy Forest (circles) and other forests (squares, with 95% confidence limits). Predator control at Abernethy Forest took place between 1992 and 1996.

6 518 R.W. Summers et al. Table 1. Numbers of crows and adult red foxes (cubs in parentheses) killed at Abernethy Forest during the period of control. Totals are given for the study area, and grand totals for the whole reserve Crow Red fox Crow Red fox Crow Red fox Crow Red fox Crow Red fox Total 36 3 (9) 42 3 (1) 37 7 (11) 16 3 (13) 11 6 (18) Grand total 74 3 (9) 69 3 (1) (11) 75 3 (13) 36 6 (18) Table 2. Numbers of crow territories, occupied red fox dens, fox indices (with 95% confidence limits), pine marten records, red deer density in March and October, lengths of deer fences, and deer fences reduced to stock fences at Abernethy Forest. Years with predator control are shown in bold;, no data Year Number of crow territories Number of red fox dens Red fox scat index Number of pine marten sightings Red deer density (no. km 2 ) March October Length of deer fences (km) Length of stock fences (km) ( ) ( ) ( ) ( ) ( ) ( ) ( ) the period of predator control, (r s = 0 6, NS, n = 5), when one might have expected a cumulative effect. The number of known occupied dens ranged from two to four, and showed no long-term change (Table 2). The removal of three to seven adult red foxes per annum (Table 1) had little effect on breeding numbers. Pine martens There were no sightings of pine martens between 1990 and One was seen in 1995 and sightings increased thereafter (Table 2). Thus, the pattern of sightings by the same number of observers from 1990 onwards showed a significant increase during the study period (r s = 0 93, P < 0 001, n = 10). EGG PREDATION BY CAPTIVE PREDATORS The fragments of shells from hen eggs eaten by captive red foxes, pine martens and crows showed that red foxes crunched the eggs so that no large fragments remained. The median size of the largest pieces was 0 44 g (interquartile range g). Even when a large part of the egg shell appeared intact it was found to be cracked and held together only by the membrane. Crows tended to make one hole through which the majority of the contents were removed. Some yolk remained in the shell. The distribution of the largest fragments was bimodal because crows sometimes broke the egg into two large pieces. The median size of all the largest pieces was 5 0 g (interquartile range g). Pine martens made larger holes in eggs than crows. However, they licked the egg clean so no pool of yolk remained. The median mass of the largest pieces was 1 2 g (interquartile range g). Crows made distinctive marks in the wax eggs, comprising paired dents separated by varying distances between the two dents of a pair. Sometimes the dents were together when the bill was closed, and up to 1 cm apart if the bill was gaping. The dents tended to be distributed around the equator of the egg since this part lies uppermost. The wax eggs given to a red fox were chewed and showed slicing marks in the wax. Pine martens left paired canine marks in the wax. Thus, it was possible to classify depredated eggs as crow-depredated or mammal-depredated from the size of the largest fragment, the presence of yolk (if found fresh) and from marks left in the wax egg. Only fresh eggs were used in these experiments. Clearly, a larger hole would be needed to extract the chick from a well-incubated egg. Therefore, we may have underestimated the number of wild capercaillie eggs taken by crows. PREDATION RATES ON ARTIFICIAL NESTS The predation rates on the artificial nests along the two transects were correlated (r s = 0 77, P < 0 05, n = 9). Overall, predation rates were high during ,

7 519 Breeding productivity in capercaillie and black grouse Table 3. Predation rates on artificial nests at Abernethy Forest. Years with predator control are shown in bold. Each transect had 24 nests Year Transect Daily predation rate on nests Standard error Percentage chance of predation in 28 days Number depredated Number taken by crows Number taken by mammals ANOVA, examining the effects of transect and year. MS F d.f. P Transect < Year < Interaction < low in 1994 and 1995, and high again during (Table 3). In many cases when an artificial nest was depredated, the clutch of hen eggs was removed and not found elsewhere and the wax egg had no marks on it, so the predator could not be identified. Cameras at artificial nests in 1999 and 2000 showed that on all 21 occasions when all the hen eggs were removed and the wax egg was untouched, the predator was a pine marten. Therefore, some of the unknown losses of artificial nests can probably be attributed to pine martens, at least after At other depredated artificial nests, the size distribution of the largest fragments of hen eggs indicated that crows were mainly responsible for predation where the predator was known ( Table 3). Some nests depredated by crows were identified from eggs found away (greater than 20 m) from the nest, where no signs had been left on wax-filled eggs. Therefore, some of the unknown category also referred to crows. Occasionally, the waxfilled egg was removed completely from the nest site, the attached string having been broken or pulled out of the wax. These were probably taken by mammals because considerable strength would have been required. Using the data for which artificial nest losses could be assigned to crows or mammals, minimum daily predation rates for each year showed that crow predation was greatest during (Fig. 4). The small peak in 1996 occurred when the number of breeding crows was low but non-breeding birds were present ( Table 2). Predation by mammals showed a contrasting pattern, and increased over the study period (Fig. 4). These indices were correlated with counts of predators, confirming their reliability as measures of predator activity. Thus, the daily predation rates by crows on artificial nests were correlated with the numbers of crow territories (r s = 0 88, P < 0 005, n = 9), and the daily predation rates by mammals on artificial nests were correlated with the number of pine marten sightings (r s = 0 79, P < 0 02, n = 9) but not with counts of fox scats (r s = 0 64, NS, n = 8). The correlation with pine martens indicated that they, rather than red foxes, were the main mammalian predators of artificial nests. Most depredated hen eggs were found where they were originally placed or within a few metres, but some were also found at shell dumps made by crows, up to 2 95 km away. Groups of depredated eggs sometimes included eggs from more than one artificial clutch. Depredated capercaillie eggs were also found casually and at shell dumps in crow territories. Numbers attributed to crows found each year varied from 0 to nine capercaillie eggs (mean 2 4 annum 1 ). Smaller numbers of crow-depredated black grouse eggs were found (mean 1 0 annum 1, range 0 4). Estimates of the number of capercaillie eggs that were depredated by crows were in 1991, in 1992 and in 1993 (Table 4). Thus, minimum estimates of capercaillie clutches lost to crows were three to five in 1991, in 1992 and eight to 17 in These estimates could only be made for

8 520 R.W. Summers et al. Fig. 4. Minimum daily predation rates by crows (circles) and mammals (squares) on artificial nests. The vertical lines show 95% confidence limits. Predator control took place between 1992 and Table 4. Numbers of artificial clutches and eggs lost to predators, numbers of depredated hen eggs found at greater than 20 m from their original site, numbers of depredated capercaillie eggs found, and estimates of total number of capercaillie eggs taken by crows. Maximum and minimum estimates of capercaillie eggs taken by crows (a and b) are given. The (a) estimate is based on the number of hen eggs from artificial nests known to have been taken by crows plus the number that potentially could have been taken by crows (i.e. those taken by unknown predators). Estimate (b) is based only on artificial clutches known to have been taken by crows., no count or estimate. Years with predator control are shown in bold Year Total number of artificial eggs (clutches) depredated Number of artificial eggs (clutches) lost to crows Number of hen eggs (dumps) found depredated by crows Number of capercaillie eggs (dumps) found depredated Number of capercaillie eggs (dumps) eaten by crows Estimate of the total number of capercaillie eggs ( 7 = nests) depredated by crows (a) (b) (5) 3 (3) (35) 107 (18) 12 (12) 4 (4) 2 (2) 35 (5) 18 (3) (36) 95 (16) 12 (6) 9 (4) 9 (4) 158 (23) 71 (10) (45) 123 (21) 9 (5) 5 (5) 4 (4) 118 (17) 55 (8) (20) 76 (13) 0 1 (1) 1 (1) (21) 18 (5) 0 2 (2) 2 (2) (28) 133 (23) 0 3 (3) 1 (1) (35) 24 (4) 0 1 (1) 1 (1) (37) 7 (2) (34) 29 (6) 0 1 (1) 1 (1) because no hen eggs from artificial nests were found away from nests in other years. DEER DENSITY A two-way (year season) Poisson ANOVA of red deer counts indicated significant variation among years (likelihood-ratio test statistic, χ 2 = 456, d.f. = 10, P < 0 001) and between the March and October counts (χ 2 = 241, d.f. = 1, P < 0 001). Counts in March and October were similar in 1989 and 1990, but the October counts declined by 1992 and remained low (Table 2). March densities also declined across years but not as steeply as the October counts. Combined October and March counts declined significantly from 1989 (r s = 0 53, P < 0 02, n = 22). The average March density was about five red deer km 2 by 1999, approximately half the 1989 density. FENCE REMOVAL AND MARKING In 1989, there were km of deer fences at Abernethy Forest. Totals of 23 2 km were removed and 4 9 km reduced to stock fence height (Table 2). The remaining deer fences were marked with chestnut paling, droppers and plastic netting in 1998 and 1999 to make them more visible. Thus, the potential for chick and adult mortality due to collisions with fences was substantially reduced over the course of the study. RAINFALL June rainfall at Abernethy Forest varied considerably between years (Fig. 5). By chance, the years with least rainfall ( ) coincided with the period of predator control. Average mid-april temperatures were around 5 C to 6 C during the first seven years of the study, and

9 521 Breeding productivity in capercaillie and black grouse Fig. 5. (a) June rainfall (mm) at Abernethy Forest and (b) average mid-april temperatures ( C) at Aviemore. Predator control took place between 1992 and Fig. 6. Average heights (cm) of heather and Vaccinium spp. at four localities (separate lines) in Abernethy Forest. There were significant site and year effects, and interactions, for both heather (F 3,532 = 64 1, P < for site; F 6,532 = 35 8, P < for year; F 18,532 = 3 0, P < for the interaction) and Vaccinium spp. (F 3,531 = 88 1, P < for site; F 6,531 = 8 6, P < for year; F 18,531 = 4 4, P < for the interaction). then peaked in This was followed by two cold mid-aprils in 1998 and 1999 (Fig. 5). SHRUB HEIGHTS The average height of heather at the four localities increased significantly from 41 cm in 1992 to 60 cm in 1999 (r s = 1 0, P < 0 001, n = 7). There was no corresponding increase in the height of the Vaccinium spp. (r s = 0 07, NS, n = 7; Fig. 6). RELATIONSHIPS OF CAPERCAILLIE AND BLACK GROUSE PRODUCTIVITY WITH ENVIRONMENTAL VARIABLES Univariate Poisson regression showed that capercaillie productivity at Abernethy Forest was negatively related to the predation rate of the artificial nests, and with the amount of rainfall in the first 20 days of June and in the whole of June (Table 5). Multiple Poisson regression showed that capercaillie productivity was significantly and negatively related to the overall rate of predation on artificial nests and also negatively related to June rainfall (Table 6a). The effect of the two-way interaction between these two variables was not significant ( χ 2 = 0 93, d.f. = 1, NS). When the variable selection procedure was repeated with the predation rates on artificial nests by crows and mammals as candidate variables instead of the overall rate, the selected model included a negative effect on capercaillie productivity of June rainfall; the negative effect of predation rate on artificial nests by crows was significant only when the two-way interaction between these variables was considered (Table 6b and Fig. 7). The interaction between predation rate by crows and June rainfall was positive, so that the magnitude of the negative effect of crow predation was reduced when June rainfall was high. Thus, capercaillie productivity was highest in dry Junes when crow predation on artificial nests was low. No relationship between capercaillie

10 522 R.W. Summers et al. Table 5. Univariate Poisson regression relationships between capercaillie and black grouse productivity, and independent variables Capercaillie Black grouse t Residual d.f. P t Residual d.f. P Artificial nest predation < NS Crow predation on artificial nests NS NS Mammal predation on artificial nests NS NS Number of breeding crows NS NS Red fox abundance (scats) NS NS June rainfall (first 20 days) < NS June rainfall (last 20 days) NS < 0 02 June rainfall (total) < < 0 02 Late May temperature NS NS Mid-April temperature NS NS Relative mid-april temperature NS NS Red deer density (March) NS NS Fence lengths NS NS Heather height NS NS Vaccinium height NS NS Table 6. (a) The relationship between capercaillie productivity at Abernethy Forest and June rainfall (mm) and the percentage of the artificial nests depredated in 28 days. Percentage of deviance accounted for = 90%. (b) The relationship between capercaillie productivity at Abernethy Forest and June rainfall (mm) plus the minimum daily predation rates on artificial nests by crows. Percentage of deviance accounted for = 92%. (c) The relationship between black grouse productivity at Abernethy Forest and total June rainfall (mm) plus the minimum daily predation rates on artificial nests by crows. Percentage of deviance accounted for = 84% Variable Estimate SE t P (a) Intercept June rain < 0 02 Predation on artificial nests < 0 01 (b) Intercept June rain < 0 02 Crow predation < 0 02 Interaction < 0 05 (c) Intercept June rain < 0 01 Crow predation < 0 05 Interaction < 0 05 rates on artificial nests by crows and mammals, no two combinations of variables were significant. However, there was a relationship between black grouse productivity and June rainfall, plus crow predation on artificial nests, plus their interaction (Table 6c and Fig. 7). Again, the negative effect of predation rate on artificial nests by crows was significant only when the two-way interaction between these variables was considered. Discussion THE EFFECT OF CULLING ON PREDATOR NUMBERS The cull of crows in the first two years of control slightly reduced the number of crows breeding in Abernethy Forest. It was only in 1994 that there was a noticeable and sustained decline. When crow culling stopped, breeding numbers recovered only moderately. The number of breeding crows in 1999 was less than half that prior to control. The slow return rate may reflect a long-term effect of the culling, the continuing policy to remove deer viscera, or a change in crow control on surrounding land. Red fox control did not reduce the number of occupied dens. The scat counts in summer showed some decline during the period of culling, but this continued when fox culling was discontinued. productivity and predation rate on artificial nests by mammals was detected, indicating that crow predation may have been more important than mammal predation. Poisson regressions were also carried out to examine patterns in black grouse productivity. Univariate analyses indicated significant negative relationships between black grouse productivity and rainfall in the last 20 days of June and in the whole of June (Table 5). In multiple regression, initially excluding predation THE EFFECT OF DEER ON THE VEGETATION We were concerned about potential confounding effects between the management of predator numbers and changes in vegetation associated with deer culling. Red deer density fell quickly from about 12 km 2 in 1990 (Table 2). During the period of reduced red deer numbers and presumed reduced browsing pressure, the height of heather increased from an average of 41 cm to 60 cm. The Vaccinium spp. showed no change in height,

11 523 Breeding productivity in capercaillie and black grouse Fig. 7. The relationship between capercaillie and black grouse productivity (chicks per female) and total June rainfall for different years (91 = 1991, etc.) at Abernethy Forest. Squares indicate years in which crows and red foxes were culled and circles show years with no predator control. The filled and open symbols indicate years when crow predation on artificial nests was above and below or equal to the median, respectively. The solid and dashed lines are the regression lines for the models (Table 6) when the average predation rate by crows was low (less than or equal to the median) and high (greater than the median), respectively. possibly because deer preferentially browsed them. Capercaillie chicks derive their invertebrate diet largely from foraging in bilberry (Kastdalen & Wegge 1985; Rolstad & Wegge 1989). Therefore, the reduction in the deer density may not have had a marked effect on food abundance for the capercaillie and black grouse chicks (Baines, Sage & Baines 1994). FACTORS AFFECTING CAPERCAILLIE AND BLACK GROUSE PRODUCTIVITY This study confirms Slagsvold & Grasaas (1979) and Moss (1986) in that June rainfall, particularly in the earlier part of the month, can adversely affect capercaillie productivity. High June rainfall also had a negative impact on the productivity of black grouse. Our data also support the importance of predation as a factor affecting capercaillie and black grouse productivity. In Sweden, Marcström, Kenward & Engren (1988) showed that the removal of mammalian predators (red foxes and pine martens) increased productivity of combined species of woodland grouse from 3 3 to 5 5 chicks female 1. However, there were few crows in their study area and their numbers were not controlled. Notably therefore, even when predators were not controlled, the Swedish productivity was still high relative to Scottish locations, and would have been adequate for maintaining numbers (1 1 chicks female 1 for Scottish capercaillie; Moss et al. 2000). In our study, estimates suggest that crows could have accounted for many capercaillie eggs in 2 years. Further, a Scotlandwide study showed that spatial variations in capercaillie productivity were related to a combination of crows and red foxes, but not pine martens (Baines, Moss & Dugan 2004). Thus, a difference between our results and Marcström, Kenward & Engren s (1988) appears to be in the different levels of crow predation. Crows are not generally part of the avifauna of boreal pine forests (Andrén 1992). Their occurrence in Scottish pinewoods is probably a consequence of forest fragmentation and favourable conditions on surrounding farmland that allow crows to feed both on farmland and in woodland, and thereby prey on woodland birds, as found in Sweden (Andrén & Angelstam 1988; Andrén 1992). HISTORIC AND FUTURE PERSPECTIVES Moss et al. (2000) concluded that poor breeding success rather than mortality of full-grown birds was responsible for the decline of capercaillie in Scotland. Moss, Oswald & Baines (2001) found a correlation between capercaillie productivity and relative temperature change in April. The present study found no significant relationship between capercaillie productivity and mid-april temperatures, or the relative mid-april temperature variable suggested by Moss, Oswald & Baines (2001). However, we found a relationship with June rainfall for both capercaillie and black grouse. This has implications for the future of these birds because Scotland s precipitation is expected, under future climatic conditions, to increase by 8% between 1990 and 2050 (Hill et al. 1999). Long-term weather data from Grantown-on-Spey, 12 km north of Abernethy Forest, suggests that wet Junes have become more frequent (Fig. 8). Other long-term changes in Scotland may also account for the decline in capercaillie breeding success. The numbers of red foxes and crows have increased against a background of reduced mortality because of declining numbers of gamekeepers (Hudson 1995), more plentiful food due a recovery of rabbits Oryctolagus cuniculus (L.) from myxomatosis (Hudson 1995), and carrion from increased sheep and red deer populations (Fuller & Gough 1999; Deer Commission for Scotland 2000). The reintroduction of capercaillie in Scotland in 1837 took place during a period of intense predator control so that, even though the forests were small and fragmented, capercaillie bred successfully and increased in numbers.

12 524 R.W. Summers et al. Fig. 8. Long-term changes in the total June rainfall (mm) at Grantown-on-Spey. r s = 0 289, 0 05 < P < 0 1, n = 37. MANAGEMENT IMPLICATIONS In reviews of the control of predators of birds, Côté & Sutherland (1997) and Newton (1998) found that predator removal generally had a positive effect on the hatching success and post-breeding numbers of the target bird. However, effects on the breeding numbers were less clear. Thus, predator removal can more easily fulfil the primary objective of game managers (larger shooting bags) but less often the objective of conservation managers (large or stable breeding numbers). However, given that the breeding success of the capercaillie and black grouse at Abernethy was related to an index of crow predation on artificial nests, and many capercaillie and black grouse eggs were eaten, the reinstatement of crow removal is a logical management prescription for increasing the productivity of capercaillie and black grouse at Abernethy Forest. Acknowledgements M. Canham, D. Clements, S. Franks, D. Gladwin and R. Parr helped with brood counts. G. Reid of Camperdown Wildlife Park and J. Usher-Smith of the Highland Wildlife Park kindly allowed eggs to be fed to captive animals in their parks. E. Kerr (Meteorological Office, Edinburgh) provided the weather data. A. Amphlett, I.P. Bainbridge, I.M. Cattadori, D. Gibbons, I. Newton, I. Storch, S. Taylor and M. Thompson commented on the drafts. We are very grateful to the Forestry Commission for financial support through a WIG II Special Management Grant We also thank landowners for granting permission to visit the reference forests. References Aitkin, M., Anderson, D., Francis, B. & Hinde, J. (1990) Statistical Modelling in GLIM. Clarendon Press, Oxford, UK. Andrén, H. (1992) Corvid density and nest predation in relation to forest fragmentation: a landscape perspective. Ecology, 73, Andrén, H. & Angelstam, P. (1988) Elevated predation rates as an edge effect in habitat islands: experimental evidence. Ecology, 69, Baines, D. (1996) The implications of grazing and predator management on the habitats and breeding success of black grouse Tetrao tetrix. Ecology, 33, Baines, D. & Summers, R.W. (1997) Assessment of bird collisions with deer fences in Scottish forests. Journal of Applied Ecology, 34, Baines, D., Moss, R. & Dugan, D. (2004) Capercaillie breeding success in relation to forest habitat and predator abundance Baines, D., Sage, R.B. & Baines, M.M. (1994) The implications of red deer grazing to ground vegetation and invertebrate communities of Scottish native pinewoods. Ecology, 31, Côté, I.M. & Sutherland, W.J. (1997) The effectiveness of removing predators to protect bird populations. Conservation Biology, 11, Cramp, S. & Simmons, K.E.L. (1980) The Birds of the Western Palearctic, Vol. 2. Oxford University Press, Oxford, UK. Deer Commission for Scotland (2000) Annual Report HMSO, Edinburgh, UK. Fuller, R.J. & Gough, S.J. (1999) Changes in sheep numbers in Britain: implications for bird populations. Biological Conservation, 91, Gregory, R.D., Wilkinson, N.I., Noble, D.G., Robinson, J.A., Brown, A.F., Hughes, J., Procter, D., Gibbons, D.W. & Galbraith, C.A. (2002) The population status of birds in the United Kingdom, Channel Islands and Isle of Man: an analysis of conservation concern British Birds, 95, Hill, M.O., Downing, T.E., Berry, P.M., Coppins, B.J., Hammond, P.S., Marquiss, M., Roy, D.B., Telfer, M.G. & Welch, D. (1999) Climate Changes and Scotland s Natural Heritage: An Environmental Audit. Report to Scottish Natural Heritage, Edinburgh. Hudson, P.J. (1995) Ecological trends and grouse management in upland Britain. Heaths and Moorlands: Cultural Landscapes (eds D.B.A. Thompson, A.J. Hester & M.B. Usher), pp HMSO, Edinburgh, UK. Johnson, D.H. (1979) Estimating nest success: the Mayfield method and an alternative. Auk, 96, Kastdalen, L. & Wegge, P. (1985) Animal food in capercaillie and black grouse chicks in south east Norway a preliminary report. Proceedings of the International Grouse Symposium, 3,

13 525 Breeding productivity in capercaillie and black grouse Marcström, V., Kenward, R.E. & Engren, E. (1988) The impact of predation on boreal tetraonids during vole cycles: an experimental study. Journal of Animal Ecology, 57, Mayfield, H.F. (1975) Suggestions for calculating nest success. Wilson Bulletin, 87, Moss, R. (1986) Rain, breeding success and distribution of capercaillie Tetrao urogallus and black grouse Tetrao tetrix in Scotland. Ibis, 128, Moss, R., Oswald, J. & Baines, D. (2001) Climate change and breeding success: decline of the capercaillie in Scotland. Journal of Animal Ecology, 70, Moss, R., Picozzi, N., Summers, R.W. & Baines, D. (2000) Capercaillie Tetrao urogallus in Scotland demography of a declining population. Ibis, 142, Newton, I. (1998) Population Limitation in Birds. Academic Press, London, UK. Proctor, R. & Summers, R.W. (2002) Nesting habitat, clutch size and nest failure of capercaillie Tetrao urogallus in Scotland. Bird Study, 49, Rolstad, J. & Wegge, P. (1989) Capercaillie habitat: a critical assessment of the role of old forest. Proceedings of the International Grouse Symposium, 4, Slagsvold, T. & Grasaas, T. (1979) Autumn population size of capercaillie Tetrao urogallus in relation to weather. Ornis Scandinavica, 10, Spidsø, T.K. & Stuen, O.H. (1988) Food selection by capercaillie chicks in southern Norway. Canadian Journal of Zoology, 66, Summers, R.W., Proctor, R., Raistrick, P. & Taylor, S. (1997) The structure of Abernethy Forest, Strathspey, Scotland. Botanical Journal of Scotland, 49, Received 16 June 2003; final copy received 4 December 2003