M. WATSON*, N. J. AEBISCHER, G. R. POTTS and J. A. EWALD. The Game Conservancy Trust, Fordingbridge Hampshire, SP6 1EF, UK

Transcription

1 Ecology , The relative effects of raptor predation and shooting on Blackwell Publishing Ltd overwinter mortality of grey partridges in the United Kingdom M. WATSON*, N. J. AEBISCHER, G. R. POTTS and J. A. EWALD The Game Conservancy Trust, Fordingbridge Hampshire, SP6 1EF, UK Summary 1. Factual information is key to resolving conflicts between raptor conservation and gamebird management, especially when the conservation status of one of the species involved is threatened. The grey partridge Perdix perdix is a UK Biodiversity Action Plan species because of a marked decline in abundance caused by agricultural intensification. Recently, the number of raptors present on farmland and the commercial shooting of red-legged partridges Alectoris rufa have both increased. To inform conservation action, the relative impacts of these two factors on grey partridge populations urgently require quantification. 2. On our study site areas of low density of grey partridges coincided with areas of high raptor density. However, these areas were managed intensively for shooting and for two areas that suffered local partridge extinction, the 3-year average percentage of partridges shot exceeded 50%. 3. Grey partridge mortality to raptors between autumn and spring lay between 9 5% of autumn density (assumes losses to raptors occurred before shooting) and 15% of post-shooting density (if all losses to raptors were post-shooting). A deterministic model suggested that this rate of loss would reduce the equilibrium density of spring pairs by 11 26% relative to a situation without raptors. In comparison, shooting losses across the study area amounted to 35 39% of autumn density, more than double the losses to raptor predation, with a predicted reduction of 68 85% in equilibrium density of spring pairs. 4. Synthesis and applications. Shooting based on large-scale releases of red-legged partridges acts in a density-independent manner and can lead to local grey partridge extinction. Removing the grey partridge from the UK quarry list would be counterproductive, as most action to boost wild grey partridge densities is carried out by enthusiasts with shooting as the incentive. However, it is imperative that managers of intensive shoots based on gamebird release adopt measures to reduce shooting pressure on wild grey partridges at low density. These include training shooters to distinguish between grey and red-legged partridges and implementing a warning system (whistle) to alert the gun line when birds of the non-target species are approaching. Such voluntary measures are effective in addressing overshooting, which has greater implications for grey partridge conservation than raptor predation. Key-words: gamebird shooting, grey partridge, human-wildlife conflict, raptor predation, sustainable harvest Ecology (2007) 44, doi: /j x Ecological Society Present address and correspondence: M. Watson, c/o The Estate Office, Holkham, Wells-Next-The-Sea, Norfolk NR23 1AB, UK. waranimi@hotmail.com

2 973 Grey partridge mortality Introduction Agricultural intensification has been a major factor in the decline of farmland birds throughout Europe (O Connor & Shrubb 1986; Marchant et al. 1990; Tucker & Heath 1994; Newton 2004) and the decline of the grey partridge Perdix perdix L. in the United Kingdom was one of the first to implicate it (Potts 1980). However, grey partridge is also affected by nest predation and shooting, which have been quantified, and by raptor predation, which has not (Potts 1980, 1986). Since Potts diagnosis of the causes of partridge decline, the abundance of sparrowhawks Accipiter nisus L. and common buzzards Buteo buteo L. has increased in the United Kingdom, while the manner in which shooting is conducted has also changed. The grey partridge decline has continued to a point where traditional shooting is not worthwhile (Aebischer 1997), and commercial lowland shooting has turned towards the release of large numbers of pheasants Phasianus colchicus L. and red-legged partridges Alectoris rufa L. Small numbers of grey partridges may also be released but red-legged partridges are much preferred for economic reasons (Tapper 1992). Against falling incomes from farming and forestry, commercial shooting has become an increasingly important land-use in the United Kingdom and management for game, if carried out sensitively, can benefit biodiversity (e.g. Aebischer 1997; Stoate & Szczur 2001). Commercial shooting often results in unintentional density-independent mortality of wild grey partridges because the number of shoot days depends on the number of gamebirds released, irrespective of wild grey partridge density. Considerable effort has been made to impose voluntary restrictions on shooting grey partridges at low densities (Tapper 2001). The debate about the balance between game management and raptor conservation has intensified in recent years (Thirgood et al. 2000a; Allen & Feare 2003). This conflict was considered so acute as to warrant European Concerted Action within the 5th Framework Program of the European Union, resulting in workshops and reviews by scientists and wildlife managers under the title Reconciling Gamebird Hunting and Biodiversity (Valkama et al. 2005). Where shoot management includes the illegal killing of raptors this can threaten the population status of rare species, for example the hen harrier Circus cyaneus L. in the United Kingdom and France (Etheridge, Summers & Green 1997; Bro, Arroyo & Migot 2006). Conversely, raptors kill gamebirds, and when predators are numerous relative to prey, may threaten the population status of the prey species (Thirgood & Redpath 1997; Thirgood et al. 2000b, 2000c; Aebischer, Ewald & Potts 2002). A long-term study of the grey partridge in a 62-km 2 area of Sussex showed that the observed decline in abundance was caused by reduced fecundity due to the effects of herbicides on chick insect food, compounded by a loss of hedgerow nesting habitat (Potts 1980; Rands 1985). In turn, there was an increase in incubating female and clutch losses through predation, resulting from reduced control of nest predators by gamekeepers as they switched to providing shooting from released birds instead of wild game (Potts 1980; Tapper, Potts & Brockless 1996). Recent analysis of Sussex monitoring data found that overwinter losses had increased with the estimated abundance of predatory mammals and raptors, and the population decline has continued (Fig. 1, Aebischer, Ewald & Potts 2002). This was the first time that raptors had been implicated in the partridge decline in the United Kingdom, although a similar process has been suggested in France (Bro et al. 2001; Bro, Arroyo & Migot 2006). Aebischer, Ewald & Potts (2002) emphasized the correlative nature of their analysis, noting that potential raptor predation was confounded with other factors, not least release of gamebirds for shooting. The present study investigated the relationship between partridge losses over winter, raptor dispersion and shooting pressure, and sought to quantify the relative magnitude of partridge losses to raptors, shooting and other factors. Methods study area The study area comprised a range of chalk hills ( m a.s.l.) bounded by the rivers Arun and Adur on the South Downs in West Sussex, southern England ( N, W). It falls naturally into east and west halves, separated by the Findon Gap, and totalling 32 km 2. This is the core area within the 62 km 2 used for the Game Conservancy Trust s long-term studies of the grey partridge (see Potts 1986). Most of the land is intensive arable, predominantly wheat and oilseed rape with some ley grassland, permanent pasture and small areas of woodland. Gamekeepers are employed on around two-thirds of the study area to manage the shoots based on released pheasants and red-legged partridges. Predator control is targeted at reducing losses of released gamebirds rather than maximizing the reproductive potential of wild game. There was no illegal control of raptors during the study. partridge abundance and dispersion Grey partridge abundance was measured by intensive counts following Potts (1980, 1986). Spring grey partridge counts were conducted in March 2000 and 2001 once pairs had formed and before the crops were too high to observe them. An autumn count of coveys (flocks) was conducted in September 2000 after harvest. Counts were conducted by driving each field systematically in a four-wheel-drive vehicle and observing with binoculars during a period of 2 h at dawn and dusk. The location of each pair (spring) or covey (autumn) was plotted on a 1 : map. Because of movement restrictions imposed to limit the spread of foot and

3 974 M. Watson et al. Fig. 1. Decline in grey partridge breeding density (line, spring pairs per km 2, right axis) and increase in autumn raptor sighting rate (bars, sightings per km 2, left axis) in the Sussex study area, mouth disease, the 2001 spring pair count was completed only for the west side of the study area. raptor dispersion Raptor presence was measured as number seen per unit time. Raptors were observed systematically from a grid of 40 points approximately 1 km apart that covered all 32 km 2 of the study area. At each point, four watches each of 1 h duration were conducted in the course of a season, one in each quarter of the daylight hours (early/ late a.m., early/late p.m.). The sequence of watches was determined at random, avoiding consecutive watches at the same or adjacent points. Three rounds of observations were made, in spring 2000 (May June), autumn 2000 (August September) and winter 2001 (January), with two to six observations per day. Preliminary analysis of the spring and autumn counts found no difference in sighting rate according to time of day or location [analysis of variance (anova), joint main effects: F 42,117 = 1 18, P = 0 24], so the number of watches was reduced to one at each point for the winter round, to save time and increase data-gathering efficiency. The observation periods were chosen to represent raptor occurrence during the nesting, post-breeding and winter periods. The species recorded were sparrowhawk, goshawk Accipiter gentilis L., buzzard, kestrel Falco tinnunculus L., peregrine Falco peregrinus T., merlin Falco columbarius L., hobby Falco subbuteo L., marsh harrier Circus aeruginosus L. and hen harrier. To avoid overemphasizing conspicuous species such as buzzard at the expense of cryptic ones such as sparrowhawk, a single sighting was taken as one observation regardless of how long the individual remained in view. Sightings were allocated to the observation point, not the location where the raptor was flying, because of the difficulty in estimating distant locations precisely and consistently. causes of partridge mortality To quantify partridge losses to raptors and other factors, carcass searches were conducted on foot across the study area at a rate of 5 km 2 per day. This involved a tour of each field margin, checking hedgerow bottoms, and a complete search of small woods by walking on parallel lines, the distance apart being set by the limit of visibility determined by the ground vegetation. Each 5-km 2 patch was covered once per month from January 2000 to March Due to foot and mouth restrictions, searches in 2001 were limited to the west side of the area. Carcasses that had saliva, teeth-marks and bitten feathers were assigned to fox Vulpes vulpes L. and carcasses with plucked feathers and faecal splash were assigned to raptors. Other potential causes of loss were collisions with fences and cars, disease and shooting. Losses to shooting were obtained from shooting bag records for the site. Foxes are known to scavenge carcasses of dead birds, often picking them up and carrying them away completely (Corbet & Southern 1997). The scavenging rate was determined in seven trials, whereby 30 red-legged partridge carcasses per trial were placed in situations that simulated raptor kills as realistically as possible, and examined for sign at 24 h and 72 h. Each carcass was placed on the ground, the breast opened and several breast feathers scattered. The carcasses were stored frozen beforehand and defrosted before placement. To minimize the effect of the trial on fox behaviour, all carcasses remaining after 72 h were removed. gamekeeper carcass survey A UK network of gamekeepers submitted fresh raptor kills of grey and red-legged partridges for post-mortem examination by a gamebird veterinarian. The gamekeepers were given a protocol for carcass collection, handling and documentation. It was emphasized that they needed to search for carcasses in all months of the year so that the seasonal pattern of losses to raptors could be established. This was important in subsequent interpretation, when addressing the issue of compensation between mortality to raptors and shooting. statistical analysis Spatial patterns in partridge density and raptor sighting rate were examined using the Geographical Information System (GIS) Mapinfo Professional 7 5 with Vertical Mapper version 3 1 (MapInfo Corporation Inc., Troy, NY, USA). Because of the sparseness of the partridge and raptor data, we pooled information over a full year from spring 2000 to spring Density contours were constructed for partridge counts pooled over spring 2000 (actual counts of spring pairs) and autumn 2000 (number of adult males, equivalent to spring pairs; Potts 1986), according to the following procedure: at each point of a 250-m grid across the study area, the density of grey partridge pairs was calculated as (number of pairs within 1 km of the grid point)/(area surveyed within 1 km of the grid point), then an inverse distance-weighting smoothing algorithm was applied to produce a grid of smoothed values. This

4 975 Grey partridge mortality procedure yielded a spatial moving average such that observations nearer the grid point had greater weight than more distant ones. Contour levels for partridge density were chosen to produce five contour regions that sought to balance clarity with subdivision into five zones of approximately equal size. The actual density of pairs within each contour region was calculated as (number observed within the contour region)/(area of the contour region)/2. Raptor sightings included kestrel (52% of observations), buzzard (26%), sparrowhawk (14%), hobby (2%), merlin (2%), peregrine (2%), marsh harrier (1%), hen harrier (< 1%) and goshawk (< 1%). We excluded kestrels, hobbies and merlins from the analysis as they are not considered predators of adult grey partridges (Cramp & Simmons 1979; Village 1990). A contour map of the raptor sighting density was constructed for comparison with partridge density, using the same smoothing procedure applied to pooled sightings from the spring, autumn and winter raptor counts. To correlate partridge and raptor presence, we calculated the density of partridge pairs and the density of raptor sightings within each of the five partridge contour regions on the west side of the study area and the five on the east side, then calculated Pearson s correlation coefficient. Grey partridge overwinter losses were obtained by subtracting the number counted in spring from the number counted in autumn. We assumed that emigration and immigration were negligible, because the study area was effectively an island surrounded by unsuitable habitat for partridges. After deducting the number shot plus 10% of this number to allow for wounding losses (Birkan 1977; Potts 1980), the remainder was split pro rata to the proportion of partridges killed by raptors and foxes corrected for scavenging rate. The estimated number of raptor kills was divided by autumn stock size to give a first measure of the proportion of grey partridges lost to raptors during winter. It was a minimum measure because it assumed that raptor losses occurred before shooting. A second, maximum, measure was obtained by dividing the estimated number of partridges lost to raptors by post-shooting stock size (this assumes that losses occurred after shooting). In reality, a small but unknown fraction of losses would have been due to causes other than predation, so these measures of losses to raptors would be slightly high. We examined the spatial and temporal pattern of shooting and its effect on partridge overwinter losses across the whole study area by comparing the proportion of grey partridges shot with the numbers available in the autumn. Shoot data were obtained from shoot managers and comprised totals of each species shot per day and numbers of gamebirds released by species. Two estates that had commercial levels of shooting, known here as farm 1 and farm 2, were examined in detail. The shooting bag data (records of the number of birds shot) for grey partridges at farm 1 were totals per day, so we divided them equally among the game drives used on that day to provide a measure of the shoot pressure by area. Farm 2 shooting bag data required no manipulation because shooting pressure was even over the estate. The results were interpreted in terms of harvest sustainability by reference to an existing gamebird harvest model (Potts 1980, 1986). A proportional estimate of overwinter loss to raptors was incorporated into the existing model (Potts 1980, 1986) to examine the relative effect of raptor predation on the equilibrium density of spring partridge pairs under different management and shooting pressure scenarios. The existing model included overwinter raptor losses but did not separate them from other losses, so our approach was to calculate losses found in this study and add them back into the model to estimate the situation as it would be in the absence of raptors. The model was based on females, which suffer less raptor predation than males (Brüll 1964), so the estimates of losses to raptors were corrected using the sex ratio from the gamekeeper carcass survey. The equilibrium pair density was estimated by averaging spring pair densities for the last 5 years of the 28-year model run. To estimate what would happen in the absence of raptors, the model was run with the minimum estimate of raptor losses (all pre-shooting), and with the maximum one (all postshooting). Within the original model, overwinter losses were modelled as a single density-dependent relationship covering all forms of loss. To partition out the effects of raptor predation from this general density-dependent relationship would be extremely complex, and subject to many assumptions especially about the sequence and timing of events. Instead, when the basis for model adjustment in the absence of raptors was the assumption of all losses occurring after shooting (maximum estimate), we calculated the equilibrium density of spring partridge pairs twice, first applying the correction for raptor predation before the density-dependent overwinter losses, then applying it after them. These two equilibrium levels were averaged as an approximation to the situation in which raptor predation and other density-dependent losses occurred together, as happens in reality. We assumed that raptor predation pressure remained constant across the range of management scenarios considered. Results spatial patterns in raptor sighting rate and grey partridge density In 360 h of observation, the sighting rate of sparrowhawks was 0 11 h 1, buzzards 0 22 h 1 and other raptors (excluding kestrel, merlin and hobby) 0 03 h 1. Contour maps of counts aggregated over (Fig. 2) showed that grey partridge density and raptor sighting rate were significantly negatively correlated (r 8 = 0 88, P < 0 001, Fig. 3). The pattern was similar to that identified from raptor sightings accumulated over 30 years of annual four-day counts during monitoring of partridges in September (Aebischer, Ewald & Potts 2002). The areas of high raptor sighting rate and low partridge

5 976 M. Watson et al. Fig. 2. Contour maps showing the spatial pattern in distribution of raptor sightings (from standardized observations at fixed points) and of grey partridge pairs (from counts) on the Sussex study area mortality to raptors Fig. 3. Raptor sighting rate in relation to grey partridge density in each of the 10 grey partridge contour regions defined in Fig. 2, Sussex (r 8 = 0 88, P < 0 001). The values from the west side of the study area are filled circles, the values from the east are open squares. Correlation is not evidence of causality. Using the proportions from the sample of grey partridges found dead on the study area (40% raptor, 60% fox), an estimated 16 of the 41 non-shooting losses would have been lost to raptors and 25 to foxes. However, the fact that foxes scavenge some raptor kills, but not vice versa, creates a bias in assigning causes of mortality. Tests of secondary scavenging showed that on average 31% (n = 7 trials of 30 carcasses, SE = 5 0) of simulated raptor kills had fox sign left on them or were removed completely after 72 h. This means that the number of kills attributed to raptors from sign alone was only 69% of the true total. Adjusting the proportions killed with this correction gave 16/0 69 = 23 (56%) killed by raptors and 18 (44%) killed by foxes. The 23 raptor kills represented 9 5% of the 243 grey partridges counted in autumn. The resulting short-term (withinyear) reduction in spring pair density due to raptors was 17% [23/( )]. This is a maximum because it assumes that all partridges killed by raptors would otherwise have survived. density also coincided with areas of intensive shoot management. partridge mortality overwinter A total of 243 grey partridges were counted in September 1999 in the core study area of 32 km 2. A total of 116 were counted in the following March. Assuming no net emigration or immigration, these autumn and spring counts showed that 127 (52%) grey partridges died over the 1999/2000 winter. Shooting bag data indicated that 78 (61%) of these dead partridges were shot. Allowing for an extra eight wounding losses, the remaining 41 (32%) must have died from some agent other than shooting. No disease was found or suspected, and there were no known collisions with the few overhead wires or with fences, leaving predation by raptors and foxes as the most probable cause of death. Of the 49 dead partridges unaccounted for, kill searching during the winter period between the counts produced remains of only five carcasses (two killed by raptors and three killed by foxes), giving a search efficiency of 10%. timing of raptor predation and sex bias of prey Gamekeepers from 19 estates across the United Kingdom sent 45 grey partridge carcasses that had been killed by raptors for post-mortem examination. Of these, 39 carcasses were recovered outside the breeding season (September March), of which 31 (84%) were from February and March (Fig. 4). This suggested that more than four-fifths of overwinter mortality to raptors came after the shooting season had ended on 1 February. Of 42 carcasses where sex could be determined, 16 were females (38%). mortality to shooting Across the study area, 78 of 243 grey partridges present in the autumn were shot, giving an average shooting rate of 32%. A 10% adjustment for wounding gave estimated overall shooting losses of 86 birds (35%). More detailed information follows for two farms on which the shooting intensity was much higher.

6 977 Grey partridge mortality Fig. 4. Collection date of 45 grey and 22 red-legged partridges killed by raptors and collected by gamekeepers for post-mortem examination from 19 farms throughout the United Kingdom. Farm 1 This was a large commercial shoot based on the release of tens of thousands of red-legged partridges. Bag data showed a contrast in shooting pressure between two areas. On one half (area A) the average percentage of wild grey partridges shot in 1997/1998, 1998/1999 and 1999/2000 was 67%; the average percentage shot on the second half (area B) was 37% over the same period (Table 1). The difference in shooting pressure arose mainly because the number of drives on area B averaged 23 (41%) fewer than area A per year. The average percentage shot fell to 16% for the 3 years after 1999/ 2000 because the shoot manager adopted a policy of avoiding shooting grey partridges, such that the bag contained only those shot by mistake. Farm 2 The time-series of bag data reflects two periods of shooting with associated release of grey and red-legged partridges, and (Table 2). During 1992, 1993 and 1994, when only red-legged partridges were released and the shooting rate of wild grey partridges Fig. 5. Model of the effects of harvesting (shooting) on wild grey partridges at low density from Potts (1986). The spring density (dotted line) declines as harvest (solid line) increases, where harvest rate is the percentage of autumn stock shot. Three-year averages ± SE are shown for bags (squares) and spring stock (circles) relative to harvest rate for farm 1 areas A and B before shooting restrictions were imposed ( : unsustainable), and for area B after they were imposed ( : sustainable). could be evaluated, the average annual percentage shot was 70%. After 1995 grey partridge releases resumed, and it was not possible to distinguish between wild and reared partridges in the bag. Evidence that wild birds continued to be shot is derived from the bag for 2000/2001, when all released birds were wing-tagged: two of eight (25%) wild birds were shot compared to 73 of 355 released (21%). There was no attempt to reduce shooting pressure. sustainability of shooting Sustainable harvest models suggest that at low density on modern farmland the optimum sustainable yield was 20% of autumn density, and that extinction occurred beyond 50% harvest (Fig. 5). The 3-year average bags on farm 1 for areas A and B in were outside the equilibrium yield curve and hence unsustainable relative to average densities Table 1. Count and shoot data from farm 1 (areas A and B) from 1997/1998 to 2002/2003. After season 1999/2000 restrictions were imposed on shooting grey partridges Year Area code 1997/ / / / / /03 A B A B A B A B A B A B Area (km 2 ) Grey partridge spring pairs Spring pair density per km Wild grey partridges in autumn Wild 2007 grey The partridges Authors. shot Shot Journal (harvest) compilation density per km Percentage shot Grey partridges shot per year Grey partridges released 300 None None None None None Drives per year No of shoot days

7 978 M. Watson et al. Table 2. Count and shoot data for farm 2 from 1979/1980 to 2000/2001, showing numbers of wild and released grey partridges (gp) in autumn, number and percentage of grey partridges shot, number of released red-legged partridges (rl), number and percentage of red-legged partridges shot Year gp wild gp rel wild + rel gp shot % wild + rel shot rl rel rl shot % rl shot 1979/ / / / / / / / / / / / / / / / i95/ / / / / / (Fig. 5). On area A, where the average percentage shot exceeded 50%, grey partridges went extinct (Table 1), as expected from the model. However, the implementation of a policy of avoiding shooting grey partridges after 1999/2000 reduced the average percentage shot to below 20%. In contrast to area A, the average spring density of 2 3 pairs per km 2 after 1999/2000 on area B was slightly higher than the average spring density of 2 0 pairs per km 2 over the 3 previous years. Relative to the sustainable harvest model, this meant that the equilibrium yield curve was within the range of variation of the average proportion shot on area B in , and the same was true when comparing expected equilibrium pair density with the average observed pair density in (Fig. 5). The implication is that the new shooting regime was sustainable and that shooting rates and partridge densities were now in balance. At farm 2, both periods of partridge releasing coincided with declines in the wild grey partridge stock (Table 2). Conversely, in the period when shooting was suspended completely, grey partridges numbers ceased declining. In subsequent years, shooting pressure rose again in response to releasing, and was known to exceed the 50% threshold in With no attempt made to reduce shooting, wild grey partridges were extinct by the spring pair count of 2001, again in line with model expectations. evaluating raptor predation and shooting pressure Data were available on raptor predation rates as well as shooting bags for the study area as a whole during As seen above, kills by raptors accounted for 9 5% (23 of 243) of the autumn stock, corresponding to the minimum estimate that assumed pre-shooting raptor losses. In this situation, overall shooting losses (bagged + wounded) were 39% [86/(243 23) = 0 391]. The maximum estimate of losses to raptors (all losses post-shooting) was 15% [23/(243 86) = 0 146]; in this case overall shooting losses (bagged + wounded) was 35% (86/243). modelling the long-term impact of raptor predation and shooting With a 1 : 1 sex ratio among juveniles (Potts 1986), 115 of 243 birds in the autumn were females (47%). The ratio of the percentage of females in autumn relative to that in the gamekeeper carcass survey gave a correction factor that adjusted the losses to raptors to reflect female losses for use in the Potts (1986) partridge model. The estimated values for female losses to raptors were (minimum) 9 5 (38/47) = 8% and (maximum) 15 (38/47) = 12%. The model was first run with a starting density of 10 pairs in spring, modern farming with herbicide use, no predation control and 5 5 km per km 2 of nesting cover (scenarios 1 9 in Table 3). These options best represented the Sussex situation under modern agriculture. With no shooting and no raptor predation these conditions resulted in a predicted equilibrium spring density of pairs per km 2 (scenarios 2 3). Including raptor predation, the resulting spring density was 6 5 pairs per km 2, a reduction of 11 26% (scenarios 1 vs. 2 and 3). Further model runs were made to examine other scenarios. With traditional density-dependent driven

8 979 Grey partridge mortality Table 3. Grey partridge population equilibrium levels from the simulation model under 10 different scenarios, comprising combinations of management, shooting and raptors. The model was run assuming first that raptor predation occurred before the density-dependent overwinter losses, and secondly (where relevant) that it occurred after them; such losses include raptor predation, so these assumptions come into play when correcting for the absence of raptors Equilibrium levels (pairs/km 2 ) Management Shooting Raptors Run 1 Run 2 Average 1. Modern farming 1 None Present Modern farming 1 None Absent (preshoot) NA Modern farming 1 None Absent (post-shoot) Modern farming 1 Dens-dep 3 Present Modern farming 1 Dens-dep 3 Absent (preshoot) NA Modern farming 1 Dens-dep 3 Absent (post-shoot) a. Modern farming 1 Fixed 39% 3 Present 0 7 NA 0 7 7b. Modern farming 1 Fixed 35% 4 Present Modern farming 1 Fixed 39% 4 Absent (preshoot) NA Modern farming 1 Fixed 35% 4 Absent (post-shoot) Traditional partridge 2 None Present Traditional partridge 2 None Absent (preshoot) NA Traditional partridge 2 None Absent (post-shoot) Traditional partridge 2 Traditional 3 Present Traditional partridge 2 Traditional 3 Absent (preshoot) NA Traditional partridge 2 Traditional 3 Absent (post-shoot) Herbicide use, no predation control, 5 5 km/km 2 of nesting cover. 2 No herbicide use, predation control, 8 0 km/km 2 of nesting cover. 3 Traditional density-dependent shooting of wild partridges only. 4 Set at the 1999/2000 estimates of 39% (when losses to raptors were assumed to occur before shooting) or 35% (when they were assumed to occur after). 5 In brackets we indicate whether the correction for the absence of raptors assumed that losses to raptors occurred before shooting (9 5%, minimum estimate) or after shooting (15%, maximum estimate). Shooting always preceded the density-dependent overwinter losses. shooting of wild partridges only and no raptor predation, the predicted equilibrium spring density was pairs per km 2, which declined to 6 4 pairs per km 2 with raptors, a 11 26% reduction (scenarios 4 vs. 5 and 6). In the absence of raptors, traditional shooting reduced the spring density by 1% (scenarios 5 vs. 2, 6 vs. 3), reflecting minimal shooting effort at low densities. In the presence of raptors, traditional shooting reduced the spring density by 2% (scenarios 4 vs. 1). To simulate intensive release-based shooting, we set shooting losses at the average observed in 1999/2000 (35 or 39%, dependent on when raptor losses were assumed to have occurred see above). This left spring pairs per km 2 with raptor predation and spring pairs per km 2 with no raptor predation (scenarios 7 9, respectively). In the absence of raptors, intensive shooting reduced spring equilibrium density by 68 85% (scenarios 8 vs. 2, 9 vs. 3). With raptors present, the reduction was 80 89% (scenarios 7a and 7b vs. 1). The combined effect of intensive shooting and raptors was a reduction of 85 90% (scenarios 7a vs. 2, 7b vs. 3). Considering an ideal situation with maximum habitat enhancement (8 km per km 2 of nesting cover, no herbicide use), intensive legal predator control and traditional shooting (Table 3, scenarios 10 15), the predicted equilibrium spring density of grey partridges in the absence of raptor predation was pairs per km 2 (scenarios 14 15). Adding raptor predation gave a spring density of 26 2 pairs per km 2, a 1 13% reduction (scenarios 13 vs. 14 and 15). Without shooting, the spring density in this situation would have been pairs per km 2, so shooting reduced spring density by 49 54% (Table 3, scenarios 13 vs. 10, 14 vs. 11, 15 vs. 12). Discussion The data on the impact of overshooting on grey partridges were unequivocal. Our stock-yield model predicted extinction when the proportions of wild grey partridges shot exceeded 50%. Two sources of shooting bag data showed that when average annual shooting mortality exceeded this level (farm 1, area A and farm 2), extinction was the observed outcome. In neither case was the overshooting deliberate, but resulted from intensive commercial shooting regimes based on largescale gamebird releasing. How typical of UK lowland gamebird shoots is this pattern of unwittingly excessive wild grey partridge harvest? While the level of grey partridge releasing has declined nationally in the United Kingdom, that of red-legged partridges has doubled since 1990 (Aebischer 2002). Thus, it is likely that the situation in Sussex was repeated elsewhere in the United Kingdom. Aebischer & Ewald (2004) estimated that grey partridges are shot on a quarter of farms where they are at low densities, and that the average shooting mortality is 12%, suggesting that some managers moderated or abandoned shooting at low densities. In France, wild grey partridge shooting has been limited or stopped altogether in some areas where their density has fallen to low levels (Reitz 1996; Bro et al. 2001). However, 2 million grey partridges were released for shooting in France in 1995, and releasing is practised extensively in regions of

9 980 M. Watson et al. greatest population decline (Reitz 2003; Bro, Arroyo & Migot 2006). Clearly, in France as well as the United Kingdom, the risk of overshooting exists as a factor in partridge population decline, and is complicated by landowners management responses to low stock levels. Most authors agree that releasing has negative consequences for biodiversity owing to increased risk of disease transmission from reared to wild stock, the risk of genetic dilution and the reduced need for habitat stewardship (Tapper 2001; Viñuela & Arroyo 2002; Bro, Arroyo & Migot 2006). Traditional shooting of wild grey partridges in the United Kingdom was density-dependent, in that the proportion shot was calculated on the basis of annual counts (Potts 1980). With large gamebird releases, the number of shoot days is set from the number of birds released regardless of the wild stock. Thus the shooting pressure can be far higher than a small wild stock can withstand. Nevertheless, as the results from farm 1 demonstrate, even on commercial shoots it is possible to take precautionary measures to avoid shooting wild grey partridges. Adopting such measures reduced shooting pressure to below 20%, i.e. below the optimal sustainable yield for that stock density. It was followed by stabilization of numbers, indicating that commercial shooting can coexist with grey partridge conservation in the United Kingdom. Note that avoiding overshooting a wild stock of grey partridges became possible only once the red-legged partridge was adopted as the release species, as the two could then be differentiated by the hunters. Obviously, this management option is not available in situations where the vulnerable native species is the same as that released, for example, for red-legged partridges shot in large numbers in Spain and Portugal. Our data on the effects of raptor predation were less robust. Carcass searching allowed us to identify the cause of death of only 10% of grey partridges that died during the winter. However, a separate multi-site study of mortality causes of 150 radio-tagged grey partridges found proportional allocations of predation losses (unadjusted for fox scavenging) of 41% to raptors and 59% to foxes (Watson 2004), remarkably similar to our estimates from carcass searches in Sussex (40% and 60%, respectively). This gives us confidence that, despite the low number of carcasses found in Sussex, the estimate of losses to raptors used in our model was realistic. Among radio-tagged partridges there was nil mortality to disease or starvation and a single case of mortality to fence collision, which again supported our assumption that all non-shooting mortality was due to predation. An independent measure of maximum partridge losses to raptors was available from an empirical relationship in Watson s (2004) study: ln(raptor kills) = 0 48 ln(autumn density after deducting birds shot) Substituting the Sussex post-shooting partridge density in this equation estimated post-shooting raptor losses at 0 154, remarkably similar to our own estimate of 15% in the Sussex study. This is further evidence that the parameterization of the model was realistic. The gamekeeper carcass survey indicated that twothirds of all raptor kills, equal to 84% of kills outside the breeding season, were found in February March. This strongly supports the hypothesis that losses to raptors occur primarily after the shooting season, implying that the true impact of raptors and of shooting is closer to the evaluations obtained on the basis of the 15% maximum estimate of partridge losses than to those derived using the minimum estimate of 9 5%. The accuracy of the gamekeeper survey is corroborated by data from Watson s (2004) multi-site study, whereby 72% (n = 29) of grey partridges found killed by raptors outside the breeding season died between end January and end March. The model results confirmed that low-density grey partridge populations are very sensitive to overwinter loss, especially raptor loss that occurs in late winter. A reduction of up to 26% in spring equilibrium density caused by raptor predation is likely to be sufficient to at least hamper recovery of low-density populations. Traditional density-dependent driven wild partridge shooting carried out at low partridge densities had little impact on subsequent spring pair densities, but intensive shooting based on released gamebirds had a serious impact of at least 68% to 85% reduction. In combination the effect of raptors and intensive shooting was extremely severe, exceeding 85%. In this study, we concentrated on modelling partridge losses to raptors outside the breeding season. In France, Bro et al. (2001) found that losses to raptors during the breeding season could be severe, but that the mortality was probably caused by harriers Circus spp. Harriers are present on the Sussex study area only in winter and even then at very low densities. The gamekeeper carcass survey (Fig. 4) indicated that only 18% of UK raptor kills were during the breeding season, when foxes and mustelids are the main predators of adult partridges (Potts 1980, 1986). In terms of model outcomes, this means that the impact of raptors is slightly underestimated, again suggesting that evaluations based on the maximum estimate of partridge losses to raptors are likely to be closest to the true picture. A feature of the grey partridge is its high potential fecundity (Potts 1986). The causes of the declines that reduced partridge numbers to the levels examined in these model scenarios were reductions in brood and chick survival due to habitat degradation exacerbated by nest predation (Potts 1980, 1986). In large-scale, replicated and controlled experiments, nest predation reduced equilibrium spring pair densities by 62% (Tapper, Potts & Brockless 1996). In other words, nest predation was 3 6 times as important as raptor predation during the non-breeding season in determining breeding densities. The last ideal model runs suggest that habitat enhancement and legal predator control can have dramatic positive effects on partridge stocks. This has been demonstrated in practice in a partridge recovery experiment, where grey partridge spring density increased by an annual average of 66% over the 2 years following the

10 981 Grey partridge mortality start of management (Aebischer & Ewald 2004). In the model, with 1 5 times more nesting cover and low nest losses, raptor predation reduced the equilibrium spring pair density by 13%, at most. Although the pattern of density-dependent predation at high partridge densities is not clear, this result is suggestive of a reduced impact of raptor predation at high densities, due presumably to the lack of a numerical response. Few studies have demonstrated such a response to gamebird prey and these have been exclusively on grouse in northern latitudes killed by specialist predators when alternative prey availability was low (Valkama et al. 2005). This study has looked at raptor predation using data from grey partridges as prey. A complementary approach would be to examine the functional response of the raptors (cf. Redpath & Thirgood 1999 and Thirgood et al. 2000b for hen harriers and red grouse). Although we were unable to distinguish reliably between the kills of different raptor species, during the 2 years of the study we saw sparrowhawks attacking or feeding on partridges 25 times and a buzzard once only. Previous studies of raptor predation on partridges have been conducted at different seasons, using different methods and at different densities of raptors, partridges and alternative prey, so may not be comparable. For example in northern France, predation caused 73% of mortality to female partridges during the breeding season (Bro et al. 2001). Of this, 29% was due to raptors and 64% to mammals. Most raptor predation was attributed to hen harriers and marsh harriers, as predation rates were correlated positively with harrier abundance. In earlier studies, partridge kills comprised 0 25 ± 0 03% of the sparrowhawk diet in Schleswig-Holstein (Uttendörfer 1939). Of 210 birds brought to buzzard nests in Moray, four were partridges, i.e. 2% of the birds killed (Swann & Etheridge 1995). In a recent study of prey brought to nests in southern England, Smart (2002) estimated that buzzards killed 2 3% of grey partridges available. In this study, farms 1 and 2 held the highest density of raptors on the study area. In this particular case, the areas favoured by raptors coincided with areas of highest shooting pressure. This may have been because these areas had woodland that provided the only suitable places for raptors to nest, as well as cover crops and grain feeders that attracted concentrations of passerines that were the main prey of sparrowhawks. It is also possible that casualties among the released gamebirds provided a source of carrion that increased food availability for buzzards during the winter. This coincidence is possible support for Newton s (1986) suggestion that raptors benefit from landscapes managed for game. Certainly, raptors would have been absent if illegal raptor control was a corollary of intensive management for game on this study area. While our data provide evidence of a potential conflict of interests between grey partridge conservation and commercial shoot management, they imply that there is little conflict between raptor conservation and this type of shooting, as raptors were tolerated. The rapid recolonization by buzzards west to east in lowland England supports this view (Clements 2000). Our results are of wide relevance to the real or perceived conflict between gamebird management and raptor conservation (Kenward 1999; Thirgood et al. 2000a; Valkama et al. 2005). They suggest that on commercial shoots based on released gamebirds and in the absence of specialist gamebird predators, overshooting is a much greater risk to wild partridge stocks than raptor predation. Thus in addition to Valkama et al. s (2005) recommendation for more studies that measure raptor predation rates and any compensation between mortality factors, there is a need to take into account the variety of gamebird management practices. Under releasing regimes where there are no wild gamebirds, raptor predation losses may usually be offset by releasing more birds (Kenward et al. 2001). Ultimately, shooting of released non-native species such as pheasants or red-legged partridges (in the United Kingdom) should not threaten the conservation status of the native grey partridge. While some might suggest from these findings that partridges of both species should be removed from the quarry list, we argue that this would be counterproductive in the United Kingdom as grey partridge density is highest on estates managed specifically as wild partridge shoots: shooting provides the incentive (and private funding) for the management package that allows grey partridges to thrive in these places (Aebischer 1997). Thus, a third of the estates participating in the Game Conservancy Trust s Partridge Count Scheme have autumn densities that exceed 20 wild grey partridges per km 2, greater than that required to achieve the UK Species Action Plan target (Aebischer & Ewald 2004). However, it is imperative to reduce the accidental kill rate of grey partridges at low densities on intensive commercial shoots based on released gamebirds. Training the shooters to identify grey partridges and have a warning system (whistle) to alert them when birds of the non-target species are approaching the gun line are all that is required. We believe that such voluntary precautionary measures can be effective when implemented, and they need to be implemented nationally as a matter of routine. The threshold below which precautionary measures should be adopted is 20 birds per km 2 in autumn and this has been included as part of recommendations for the Game Conservancy Trust s Partridge Recovery Programme (Tapper 2001). Together with the agricultural reforms launched in the United Kingdom in March 2005, which effectively reduce the cost of habitat management, there is now the opportunity to reverse the fortunes of this charismatic farmland species. Acknowledgements We thank the farmers, landowners and gamekeepers on the Sussex Study area and elsewhere who allowed access to their land and supplied bag records. We are grateful to four anonymous reviewers for help in improving the manuscript.

11 982 M. Watson et al. References Aebischer, N.J. (1997) Gamebirds: management of the grey partridge in Britain. Conservation and the Use of Wildlife Resources (ed. M. Bolton), pp Chapman & Hall, London. Aebischer, N.J. (2002) The National Gamebag Census: recent trends and the effect of foot-and-mouth disease. Game Conservancy Annual Review, 34, Aebischer, N.J. & Ewald, J.A. (2004) Managing the UK grey partridge Perdix perdix recovery: population change, reproduction, habitat and shooting. Ibis, 148 (Suppl. 2), Aebischer, N.J., Ewald, J.A. & Potts, G.R. (2002) Preliminary results from using GIS to examine spatial variation in grey partridge demography over 30 years in Sussex, UK. 24th International Congress of the International Union of Game Biologists (eds C. Thomaides & N. Kypridemos), pp Hellenic Hunting Federation of Macedonia-Thrace, Thessaloniki. Allen, D.S. & Feare, C.J. (2003) Birds of prey and lowland game management. Birds of Prey in a Changing Environment (eds D.B.A. Thompson, S.M. Redpath, A.H. Fielding, M. Marquiss & C.A. Galbraith), pp The Stationery Office, Edinburgh. Birkan, M. (1977) Analyse des tableaux de chasse de perdrix (Perdix perdix L. et Alectoris rufa L.). Courbes d éclosion, structure et dynamique des populations, plan de chasse. Ecologie Du Petit Gibier Et Aménagement Des Chasses (Analysis of partridge (Perdix perdix L. and Alectoris rufa L.) bags. Hatching curves, population structure and dynamics, shoot management. (Small Game Ecology and Shoot Management) (eds P. Pesson & M. Birkan), pp Gauthier-Villars, Paris. Bro, E., Arroyo, B. & Migot, P. (2006) Conflict between grey partridge hunting and hen harrier protection in France: a review. Wildlife Biology, 12, Bro, E., Reitz, F., Clobert, J., Migot, P. & Massot, R. (2001) Diagnosing the environmental causes of the decline in grey partridge Perdix perdix survival in France. Ibis, 143, Brüll, H. (1964) A study of the goshawk Accipiter gentilis and sparrowhawk Accipiter nisus in their ecosystem. Report on the Working Conference on Birds of Prey and Owls, pp International Council for Bird Preservation, London. Clements, R. (2000) Range expansion of the common buzzard in Britain. British Birds, 93, Corbet, G.B. & Southern, H.N. (1997) The Handbook of British Mammals, 2nd edn. Blackwell, Oxford. Cramp, S. & Simmons, K.E.L. (eds) (1979) The Birds of the Western Palearctic, vol. 2. Oxford University Press, Oxford. Etheridge, B., Summers, R.W. & Green, R.E. (1997) The effects of illegal killing and destruction of nests by humans on the population dynamics of the hen harrier Circus cyaneus in Scotland. Ecology, 34, Kenward, R.E. (1999) Raptor predation problems and solutions. Journal of Raptor Research, 33, Kenward, R.E., Hall, D.G., Walls, S.S. & Hodder, K.H. (2001) Factors affecting predation by buzzards. Buteo buteo on released pheasants Phasianus colchicus. Ecology, 38, Marchant, J.H., Hudson, R., Carter, S.P. & Whittington, P. (1990) Population Trends in British Breeding Birds. British Trust for Ornithology, Tring. Newton, I. (1986) The Sparrowhawk. Poyser, London. Newton, I. (2004) The recent declines of farmland bird populations in Britain: an appraisal of causal factors and conservation actions. Ibis, 146, O Connor, R.J. & Shrubb, M. (1986) Farming and Birds. Cambridge University Press, Cambridge. Potts, G.R. (1980) The effects of modern agriculture, nest predation and game management on the population ecology of partridges (Perdix perdix and Alectoris rufa). Advances in Ecological Research, 11, Potts, G.R. (1986) The Grey Partridge: Pesticides, Predation and Conservation. Collins, London. Rands, M.R.W. (1985) Pesticide use on cereals and the survival of grey partridge: a field experiment. Ecology, 22, Redpath, S.M. & Thirgood, S.J. (1999) Numerical and functional responses in generalist predators: harriers and peregrines on Scottish grouse moors. Journal of Animal Ecology, 68, Reitz, F. (1996) Les perdrix en 1995 dans le Nord, le Bassin Parisien et le Centre: à suivre! (Patridges in 1995 in the North, the Parisian Basin and the Centre: watch this space) Bulletin Mensuel de l ONC, 208, Reitz, F. (2003) Le statut communal de la perdrix grise et de la perdrix rouge en France: resultats d une enquête. (The status by parish of grey and red-legged partridges in France: survey results.) Faune Sauvage, 258, Smart, S. (2002) The diet of breeding sparrowhawks Accipiter nisus and common buzzards Buteo buteo. Unpublished MSc Thesis, University of Reading, UK. Stoate, C. & Szczur, J. (2001) Could game management have a role in the conservation of farmland passerines? A case study from a Leicestershire farm. Bird Study, 48, Swann, R.L. & Etheridge, B. (1995) A comparison of breeding success and prey of the common buzzard Buteo buteo in two areas of northern Scotland. Bird Study, 42, Tapper, S.C. (1992) Game Heritage: an Ecological Review from Shooting and Gamekeeping Records. Game Conservancy Ltd, Fordingbridge. Tapper, S.C. (2001) Conserving the Grey Partridge. Game Conservancy Trust, Fordingbridge. Tapper, S.C., Potts, G.R. & Brockless, M.H. (1996) The effects of an experimental reduction in predation pressure on the breeding success and population density of grey partridges Perdix perdix. Ecology, 33, Thirgood, S.J., Redpath, S.M., Haydon, D.T., Rothery, P., Newton, I. & Hudson, P.J. (2000c) Habitat loss and raptor predation: disentangling long-and-short term causes of red grouse declines. Proceedings of the Royal Society London, Series B, 267, Thirgood, S. & Redpath, S.M. (1997) Red grouse and their predators. Nature, 390, 547. Thirgood, S.J., Redpath, S.M., Newton, I. & Hudson, P. (2000a) Raptors and grouse: conservation conflicts and management solutions. Conservation Biology, 14, Thirgood, S.J., Redpath, S.M., Rothery, P. & Aebischer, N.J. (2000b) Raptor predation and population limitation in red grouse. Journal of Animal Ecology, 69, Tucker, G.M. & Heath, M.F. (1994) Birds in Europe: Their Conservation Status. BirdLife International, Cambridge. Uttendörfer, O. (1939) Die Ernährung der Deutschen Raubvögel und Eulen und Ihre Bedeutung in der Heimischen Natur (The diet of German raptors and owls and their importance in the natural environment). Neumann, Neudamm. Valkama, J., Korpimaki, E., Arroyo, B., Beja, P., Bretagnolle, V., Bro, E., Kenward, R.E., Mañosa, S., Redpath, S.M., Thirgood, S.J. & Viñuela, J. (2005) Birds of prey as limiting factors of gamebird populations in Europe: a review. Biological Reviews, 80, Village, A. (1990) The Kestrel. Poyser, Calton. Viñuela, J. & Arroyo, B.E. (2002) Gamebird Hunting and Biodiversity Conservation: Synthesis, Recommendations and Future Research Priorities. Report to the European Community, Brussels. Watson, M. (2004) The effects of raptor predation on grey partridges Perdix perdix. Unpublished DPhil thesis, University of Oxford, Oxford. Received 3 May 2006; final copy received 16 April 2007 Handling Editor: Simon Thirgood