sign in sign up
<<
Home , Common buzzard, Common toad, Mistle thrush, Slow worm, Song thrush

Spatial Variation in the Breeding Success of the Common buteo in relation to Habitat and Diet

George Swan

September 2011

A Thesis submitted in partial fulfilment of the requirements for the degree of Master of Science and the Diploma of Imperial College

Contents List of Acronyms ...... 3 1. Introduction...... 1 1.1. and gamebirds: defining the conflict ...... 2

1.2. Aims and Objectives ...... 4

2. Background ...... 5 2.1. The Buteo buteo ...... 5

2.1.1. & morphology ...... 5 2.1.2. Distribution, population trends and current ...... 5 2.1.3. Reasons for reestablishment ...... 5 2.1.4. Reproduction and survival ...... 6 2.1.5. Habitat and diet ...... 6 2.2. Spatial variation in buzzard densities ...... 6

2.3. The role of in buzzards diet ...... 7

2.4. The role of habitat on buzzard densities and breeding success ...... 8

2.4.1. Forestry ...... 8 2.4.2. Farmland...... 8 2.5 Assessing raptor diet: pellet analysis, prey remains and direct observations ...... 9

3. Study site ...... 10 3.1. Forestry and Upland pasture ...... 10

3.2. Lowland farmland: Doune ...... 10

4. Methods ...... 12 4.1. Field data collection ...... 12

4.1.1. Habitat classification ...... 12 4.2.1. Prey Sampling ...... 12 4.2.2. Mapping buzzard ...... 14 4.1.3. Mapping core territories ...... 15 4.1.4. Assessing Buzzard diet ...... 16 4.2. Data Analysis ...... 18

4.2.1. Preparatory analysis ...... 18 4.2.2. Variable compilation ...... 19 4.2.3. Exploratory univariate analysis ...... 19 4.2.4. Investigating response type of buzzard to density ...... 20 1

4.2.5. Investigating the influence of prey size in breeding success ...... 20 4.2.6. Principal component analysis ...... 20 4.2.7. Modelling ...... 20 5. Results ...... 22 5.1. Exploratory analysis ...... 22

5.1.1. Variation in buzzard density on a spatial scale ...... 22 5.1.2. Variation in buzzard productivity on a spatial scale...... 22 5.1.3. Prey densities ...... 23 5.2. Results of buzzard dietary studies ...... 25

5.3. Investigating the type of functional response ...... 26

5.4. The influence of prey size in breeding success ...... 26

5.5. Principle component analysis ...... 27

5.5.1. PCA analysis: fledgling number ...... 29 5.5.2. PCA analysis: Nearest neighbour distance ...... 30 5.6. Direct observational data ...... 31

6. Discussion ...... 33 6.1. Overview ...... 33

6.2. The use of principle component analysis and multivarient modelling ...... 34

6.2.1. Breeding success ...... 35 6.2.2. Density ...... 35 6.3. Cameras on nests ...... 35

6.4 Study limitations and areas for future research ...... 37

6.4.1. Limitations of the methodology ...... 37 6.4.2. The effect of weather during the hatching period ...... 37 6.4.3. Limitations of the analysis ...... 38 6.5. Resolving the conflict ...... 39

7. References: ...... 42 8. Appendices ...... 55 Appendix 1.The regression formula for VSI to density conversion (from Lambin et al., 2000) ...... 55

Appendix 2. Full prey list from all dietry analysis methods ...... 56

Appendix 3:Examples of prey identification from remote cameras ...... 57

2

Appendix 4: All detenction functions for small from DISTANCE analysis ...... 58

List of Tables Table 1: A literature review of buzzard breeding success and nearest neighbour distance ...... 7 Table 2: Categories used to classify vegetation cover within the study areas...... 12 Table 3: The estimates of small abundance from DISTANCE analysis ...... Error! Bookmark not defined. Table 4: Linear regressions to calculate vole density as detailed in Lambin et al., 2000...... Error! Bookmark not defined. Table 5: List of variables with descriptions ...... 19 Table 6: Number of observations and percentage total diet of different prey items as estimated from prey remains collection and pellet analysis ...... 26 Table 7: The results of the Principle Componet Analysis showing the loadings of habitat and dietary variables ...... 28 Table 8: Competing generalisted liner models using the principle components to explain fledgling number and nearest neighbour distance ...... 29 List of Figures Figure 1: Map of study site showing all active nests...... 15 Figure 2: Example of buffer mapping from a forestry ...... 16 Figure 3: Photographs of prey remains and pellet analysis...... 17 Figure 4: A boxplot showing the nearest neighbour distance against the habitat categories...... 22 Figure 5: A boxplot showing the number of buzzard chicks against habitat type ...... 23 Figure 6: Scatter graphs showing per km² against; 1) the nearest neighbour distance (left) and; 2) the number of chicks raised to fledging (right)...... 25 Figure 7: Bargraphs showing the relationship between the dominant species found during the pellet analysis (A) and the prey analysis (B) against the mean number of fledglings ... 27 Figure 8: The relationship between PC1 ‘vole reliance’ and the number of chicks fledging...... 30 Figure 9: The relationship between ‘upland generalist’ and the nearest neighbour distance ...... 31 Figure 10: Bargraphs showing the proportions of mammals, and observed in all dietary methods in forestry (top left), upland pasture (top right), lowland pasture (bottom left) and all habitats (bottom right)...... 32 List of Acronyms VSI Vole Sign Index AIC Akaike Information Criterion NND Nearest Neighbor Distance GLM Generalized Linear Model PCA Principle Component Analysis SDRH Site Dependant Regulation Hypothesis PC Principle Component RSPB Royal Society for the Protection of Birds.

3

Abstract Widespread population increases of the Common Buzzard Buteo buteo has brought the species back into conflict with gamebird hunters. As the first step in addressing the conflict, a basic understanding of the factors that regulate buzzard breeding ecology is required. While variations in prey abundance have been associated with fluctuations in raptor breeding correlates, for generalist raptors like the buzzard the mechanisms of food limitation are poorly understood. To address this, the productivity and breeding density of 63 buzzard pairs was studied in relation to prey abundance, composition and diet across a range of different habitats in central . Concurrently, buzzard diet was determined using the two competing methods of prey remains and pellet analysis, while prey abundance was estimated from indices of vole sign and distance sampling for small passerines. Nest cameras allowed for a preliminary assessment of the prey and pellet dietary analysis methods with results suggesting that prey has been significantly underestimated in buzzard diet. Highly significant differences were observed in buzzard breeding success over spatial scales with low productivity being related to a reliance on vole prey. Highly significant differences were also observed in buzzard densities with the lowest densities being related to upland pasture and percentages of amphibians and birds. The results suggest that buzzards may be able to reach the highest densities and productivity in lowland pasture with access to prey. Future research is now needed to needed to assess how densities influence rates on gamebirds in both an upland and lowland context.

Acknowledgments

Firstly, a sincere thank you goes David Anderson and Katy Freeman from the Forestry Commission Scotland for putting up with me over the nesting season and for showing me that raptors are not just an interest but an obsession. My thanks are also extended to Keith Burgoyne and Mark Rafferety for the many trees that were fearlessly (Keith) and not so fearlessly (Mark) climbed and to Mike McDonnell and Duncan Orr-Ewing for the data and company that they provided. Collecting data from such a large number of buzzard nests is not easy and was only made possible by the brilliant mix of enthusiasm and dedication that you guys have toward these birds. I really enjoyed my time in Scotland and now consider myself a true ‘friend of the buzzard’.

In terms of analysis, it would almost be easier to write all the people who didn’t give me help at some point, but special mentions go to Rebecca Spriggs for her knowledge of R and saint like patience, Mike Hudson for his GIS genius and to Kyle, Cian, Jeremy and Anne for not just answering my incessant questions but actually caring about whether they gave the right answer.

I would also like to thank my two supervisors Steve Redpath and Nils Bunnefeld for the invaluable advice that they both provided through all aspects of this project.

Finally, I would like to thank my parents for supporting me, not just through this project but through the many unemployed years that lie ahead...

1

1. Introduction 1.1. Buzzards and gamebirds: defining the conflict

The relationship between predators and their prey has long been of interest to ecologists and wildlife managers. Through their very nature as hunters, predators produce some of the most controversial and polarised debates in wildlife management (reviewed in Woodroffe et al., 2005). While predatory species enjoy an elevated position in the public conciseness, they are also vilified by many as potential competitors for common, limited resources (Graham et al., 2004). Conflicts between predators and man can be observed the world over with examples such as the grey wolf Canis lupus depredation on in the northwestern United States (Bangs et al., 2005), brown bear Ursus arctos depredation on sheep in (Sagør et al., 1997), jaguar Panthera onca depredation on cattle in (Michalski et al., 2005) and Vulpes vulpes predation on gamebirds in Britain (Baker et al., 2006).

The attention of studies concerning raptors has often been parallel with issues related to species management, particularly the human-predator conflict that arises when a raptor is perceived as having the potential to reduce gamebird harvests (Thirgood et al., 2000b, Bro et al., 2006; Watson et al., 2007; et al., 2008). Conflicts of this nature are particularly acute during reestablishment or density increases (O’Toole et al., 2002; Henderson et al., 2004). Within the UK, concerns over the impact of raptors have often been raised by parties representing the interests of gamebird shooting (Arroyo & Viñuela, 2001; Arroyo, 2002).

Field sports play an important socioeconomic role within rural communities (Public & Corporate Economic Consultants 2006) as well benefiting biodiversity at both a landscape (Oldfield et al., 2003) and species level (Stoate & Szczur, 2001). However, persecution by gamekeepers (to enhance the densities of gamebirds) has severely reduced the abundance and distribution of many raptor species within the UK (Newton, 1979; Elliot & Avery, 1991; Brown & Stillman, 1998; Whitfield et al., 2003; Lovegrove, 2007), despite all raptor species being legally protected under the Wildlife and Countryside Act, 1981.

The debate between the protection of economic interests (in the form of gamebirds) and the protection of birds of prey has intensified over the last decade with repeated calls for landowners to be granted licences to control raptor species to protect gamebirds (Green, 2011). The issue has even attracted the attention of the in the form of the ‘European Concerted Action REGHAB’ (Reconciling Gamebird Hunting and Biodiversity) (Valkama et al., 2005).

Population increases coupled the opportunistic predatory nature of the Common Buzzard Buteo buteo (hereafter buzzard) has brought the species back into conflict with shooting stakeholders.

2

Indeed, in a recent members poll by the Scottish Gamekeepers Association 76% of respondents identified buzzards as having a negative effect on gamebirds (Green, 2011).This conflict arises because buzzards are seen as either having a direct impact on the numbers of birds available to shoot, or as causing in-direct mortality and financial loss due to disturbance (Harradine et al., 1997; UK Raptor Working Group, 2000; Kenward et al., 2001). As a result, although the reestablishment of the buzzard can be considered a conservation success story the species continues to bear the brunt of the relatively high level of illegal raptor persecution occurring within the UK (Mineau et al., 1999). In Scotland, the number of reported instances of buzzards being deliberately poisoned and shot has been increasing over the last 20 years, (Anon. 2009; Taylor et al., 2009).

As a generalist predator, buzzards are able to respond both functionally and numerically to changes in prey density. The generalist predator hypothesis (GPH) states that environments with a large selection of alternative prey species are able to support higher densities of generalist predators as they are able to alternate between prey species according to accessibility (Hanski et al., 1991; Bjǿrnstad et al., 1995). As gamebird species have both natural (Hudson, 1992) and artificial (Kenward et al., 2001) fluctuations in abundance, generalist predators have the potential of having a significant economic impact (Redpath & Thirgood, 1999; Thirgodd et al. 2000a, 2000b).

While previous studies have identified several gamebird species as being present in the diet of buzzards (Swann & Etheridge, 1995; Kenward et al., 2001), there is a need to consider how buzzard densities and diet are influenced by the wider landscape. Buzzards are generalist predators living in species rich communities and yet there may be a tendency, as with other predators, of describing their impact on gamebirds in terms of a simple single-predator single- prey interaction (Graham et al., 2004). This could lead to illegal persecution through the assumption that a reduction in predator numbers will lead to an increase in prey available to shoot (Bro et al., 2006). However, to answer the question whether high densities of buzzards impact significantly on the amount of gamebirds available to shoot, it is crucial to consider the ways in which buzzard predation rates (their functional response) and density (their numerical response) vary in relation to both gamebird density and the density of alternative prey species (Park et al., 2008). Buzzards are opportunistic predators, therefore it is expected that the proportions of prey species in their diet will vary with abundance (Tornberg & Reif, 2007). However, given that the artificial release of gamebirds begins around July (Kenward et al., 2001), it seems unlikely that the nesting densities of buzzards is a response to variations in gamebird abundance and more likely that it is as a result of alternative prey abundance (Graham et al., 1995; Swann & Etheridge, 1995). The influence of alternative (non gamebird) prey regulating 3 raptor breeding ecology on shooting estates has already been observed in relation to Circus cyaneus on managed for red Lapogus lapogus (Redpath & Thirgood 1999; Redpath et al., 2002).

1.2. Aims and Objectives

In order to address the impact of buzzards on gamebirds more information is needed as to the ecological details that govern buzzard densities and breeding success. To put the buzzard range expansion into context, the question of what constrains or indeed enhances buzzard densities and reproductive success will be addressed. The overall aim of this project is to analyse the spatial variation in breeding ecology in the context of: (1) habitat, (2) prey abundance and, (3) buzzard diet.

Objective 1: Identify how prey abundance regulates the breeding ecology of buzzards.

Hypothesis a: That there is a significant relationship between the breeding success of buzzards prey abundance estimates.

Hypothesis b: That there is a significant relationship between the density of breeding buzzards and prey abundance.

Hypothesis c: That buzzard will display a type II response to increases in abundance of common prey items.

Objective 2: Identify how dietary composition influences the breeding ecology of buzzards.

Hypothesis d: Nests with access to larger prey items will raise more chicks to fledging.

Objective 3: Undertake multivariate modelling using the results of a principle component analysis to explore how the variables habitat and diet interact to influence buzzard breeding ecology.

Objective 4: Investigate whether prey remains and pellet data is an accurate reflection of buzzard diet.

4

2. Background 2.1. The Common Buzzard Buteo buteo

2.1.1. Taxonomy & morphology The common buzzard Buteo buteo (Linnaeus, 1758) is one of 28 species of buzzard (also referred to as ‘hawks’) recognised worldwide (Ferguson-Lees & Christie, 2001). There are currently thought to be nine different of Buteo buteo divided into the western Buteo group and the eastern vulpinus group, however, while morphometric analysis allows for a discrimination between taxa, phylogenetic relationships are less clear (Kruckenhauser et al., 2004). Of the two species of the Buteo to be found in the UK (the common buzzard and the rough-legged buzzard buteo lagopus), it is only the common buzzard that is a permanent resident.

The common buzzard (hereafter buzzard) is a medium sized diurnal raptor with dark brown , a yellow cere and yellow feet. As adults they measure 51-57cm with a wingspan of 113- 128cm. Like many diurnal raptor species buzzards have reverse with the female 5-10% taller and 27-30% heavier than the male (Snow & Perrins, 1998).

2.1.2. Distribution, population trends and current conservation status The buzzard is the most widely distributed raptor throughout the Paleartic, one of the most abundant of all European raptor species (Bijlsma, 1997) and the UK’s most common of prey (Clements, 2002).

Until the beginning of the 19th century the buzzard was found throughout the UK but by the early stages of the 20th century it had become restricted to only western parts of Britain (Moore, 1957). However, since the 1950’s the population has been increasing steadily; from 8-10,000 in 1970 (Tubbs, 1974), to 15-16,000 pairs in 1983 (Taylor et al., 1988) to the most recent estimate of 44-61,000 in 2000 (Clements, 2002). At present, this number is likely to be substantially higher as the results of the British Bird Survey (Risely et. al., 2010) provide evidence of a significant and continuing increase (63% from 1995 to 2009) across the UK. As a consequence of its large range, increasing population trend and large population size the buzzard is evaluated as least concern under the IUCN red list criteria (IUCN, 2010).

2.1.3. Reasons for reestablishment The reestablishment of the buzzard across the UK has been accredited to two factors; firstly, an increase in productivity; attributed to both the recovery of rabbit populations after a crash in the 1950’s (Moore, 1957; Sim et al., 2000) and the banning of the organochlorine pesticide dichlorodiphenyltrichloroethane (DDT) which caused a thinning in the shells of raptor (Newton, 1979)and; secondly, higher survival rates due to a reduction in illegal persecution gamekeepers (Taylor et al., 1988; Sim et al., 2000).

5

The fact that every county within the UK currently has breeding buzzards (Clements 2002) can be viewed as an example of how, over the course of a relatively short time period a raptor species can re-establish itself over a large geographical area. Studies have gone as far as to use the success of the buzzard to provide an insight into the ability of similar re-bounding avian species (red kites and ravens) to recolonize lost territory (Smith, 2007).

2.1.4. Reproduction and survival Females lay a clutch of 2-4 eggs from early April through to early May over 6-9 days. After an incubation period of 36-38 days the eggs hatch asynchronously, at intervals of 48 hours and young spend 47-57 days in the nest before fledging (Tubbs, 1974). Immature birds will then disperse during their first autumn or the following spring (Walls and Kenward, 1995) and finally settle to breed in their third or fourth year. Unlike buzzards on the continent, most pairs in the UK remain in their home range throughout the year (Walls & Kenward, 1998), although continuous occupancy may be contingent on prey availability and mild winters (Halley, 1993).

2.1.5. Habitat and diet Buzzards are catholic in their choice of habitat. This can be attributed to the versatile range of hunting techniques that can be adapted to suit different habitats; (1) perching and scanning surrounding habitat then planning onto prey; (2) soaring over open terrain before dropping onto prey and (Hume, 2007); and (3) walking on the ground (Tubbs, 1974).As generalist predators buzzards consume a diverse range of prey items, principally mammals (mainly voles and ) but also birds, , amphibians, , and carrion (Dare, 1961; Newton et al., 1982; Graham et al., 1995; Swann & Etheridge, 1995; Sim, 2003; Selàs et al., 2007).

2.2. Spatial variation in buzzard densities

When attempting to understand spatial variation in the density of a species, it would seem appropriate to focus on the resources that would limit the carrying capacity of an area, for raptors, this is most often food resources (Newton, 1979). The spatial variation in food resources for birds of prey can be seen as a function of prey abundance and accessibility, with prey accessibility ultimately being contingent on first detection, then capture probabilities. This is particularly apparent during the breeding period as variations in prey may determine territory quality by influencing the ability of the females to produce eggs, the clutch size and the survival of the brood (Lack, 1954). In this way, natural variations in the abundance and accessibility of prey across habitat types can be expected to result in marked differences in raptor densities. Due to its extensive distribution (Clements, 2002), this is particularly true for the buzzard and significant differences in breeding density and fledgling success being observed across their range (Table 1).

6

Table 1: Buzzard breeding success and nearest neighbour distance from studies in Scotland, and

Author Fledgling NND Location Country success (km) Holling, 2003 2.4 - Lothian and Borders Scotland Jardine, 2003 1.8 - Colonsay Scotland Sim, 2003 1.7 0.8 Welsh Marches Wales Sim et al., 2001 1.2 - West Midlands England Austin & Houston, 1997 1.8 - Scotland Swann & Etheridge, 1995 2.5 1.1 Moray Scotland Swann & Etheridge, 1995 1.5 1.7 Glen Urquhart Scotland Graham et al., 1995 - 1.9 Langholm Scotland Dare, 1995 1.4 2.0 Wales Dare, 1995 1.4 1.5 Hiraethog Wales Newton et al., 1982 0.6 0.9 Cambrian Mountains Wales Picozzi & Weir, 1974 1.6 - Strathspey Scotland

High abundances of buzzards have been associated with the abundance of larger prey items, Newton (1979) for example an example of Skoma Island (Wales) where seven pairs nested within an area of 3.1km feeding both rabbits and seabirds. While buzzards are territorial, it would appear that, in the absence of human disturbance (Elliot & Avery, 1991), their density is principally limited by prey availability (Graham et al., 1995; Reif et al., 2004). So, while it would still be correct to describe buzzard as territorial, it is most likely that buzzards adjust their territorial behaviour to regulate their density to the resources available (Newton 1979).

2.3. The role of mammals in buzzards diet

Previous studies into the diet of buzzards in the have shown that mammals make up the majority of buzzards diet (Dare, 1961; Sim, 2003; Jardine, 2003). While most studies have indicated that rabbits are the most significant prey species (Kenward et al., 2001; Sim 2003; Jardine 2003), small mammals, specifically field voles, have also been shown to be a significant component (Dare, 1961; Graham et al., 1995). In their 1995 paper on breeding success and diet of buzzards in southern Scotland, Graham et al. described the overall importance of buzzard prey as being lagomorph > bird > vole. However, in his summary of European breeding studies Newton (1979) showed that breeding success of buzzards was correlated to vole abundance, particularly in poorer territories. This was confirmed more recently in southern Scotland by Swann & Etheridge (1995) who documented a highly significant correlation between the breeding success of tawny (vole specialists).

Field vole populations experience frequent population cycles in the northern parts of their range such as in Scotland and (Gibbs & Alibhai 1991; Lambin et al., 2000). While these 7 cyclical fluctuations have been documented to influence the reproductive success of both specialist and generalist raptor species (Korpimäki & Norrdahl, 1999; Salamolard et al., 2000; Redpath et. at., 2002; Petty & Thomas, 2003;), with generalist predators, there is the possibility of mitigating these effects through a functional response (predating other prey species at a higher intensity). Spidso & Selàs (1988) provided evidence that buzzard populations in Norway respond to crashes in vole populations by dramatically increasing the proportion of birds in their diet. This suggests that during vole population crashes, territories with access to alternative prey items will show less variation in breeding success. In their study of buzzard habitat selection in , Sergio et al. (2005) showed that at a landscape scale buzzards prefer high levels of habitat heterogeneity within their territories, a fact that was linked to a varied food supply. It is possible however, that declines in several of the common prey species of generalist raptors can occur in conjunction with one another causing even generalist predator populations to decline suddenly (Salafsky et al., 2007).

2.4. The role of habitat on buzzard densities and breeding success

2.4.1. Forestry Across the United Kingdom landscapes are utilised and managed for a variety of different reasons (for example economic gain and nature conservation) and, as a result are under a variety of pressures. Within central Scotland these pressures are most notably farming and commercial forestry. In his early study on the status of the buzzard within the , Moore (1957) observed that buzzards reach higher densities on farmland than within forested areas. Commercial forestry in Scotland may affect the abundance of forest dwelling buzzards through its effect on: (1), the number of prey; (2) the type of prey (Newton, 1979) and; (3) the availability of prey (Sonerud, 1997). Forestry practises such as clear cutting (the harvesting of a complete block of trees), have been recorded to cause decreases in generalist raptor species in both Fennoscandia (Widén, 1994) and (Penterianni & Faivre, 1997). However, within the UK it has been shown that clear cut and pre-thicket areas can also provide relatively high abundances of voles to avian predators (Petty, 1999).

2.4.2. Farmland Farming methods can also influence the density and breeding success of buzzards in many ways. Butet et al. (2010) showed that buzzards decrease significantly with the reduction of hedgerows, and areas. This can be attributed to a decrease of prey abundance at a landscape scale with an expansion of field margins and loss of habitat heterogeneity (Benton et al., 2003; Petrovan et al., 2011). Similarly, grazing intensity can have significant effects on the abundance of Cricetidae prey species; this has been demonstrated on both lowland pasture and also on upland habitats (Evans et. al., 2006).

8

2.5 Assessing raptor diet: pellet analysis, prey remains and direct observations

Knowledge of the diet of a raptor is essential to understanding its ecology. Dietary analysis can be achieved through several methods including: 1) monitoring individuals while hunting through radio tracking (Kenward, 1982) or direct observations; 2) recording the prey items brought to the nest during the breeding season (Lewis et al., 2004; Tornberg & Reif, 2007); 3) analysing raptor stomach contents (Pietersen & Symes, 2010) and; 4) identifying prey items from pellets or faecal samples (Tubbs & Tubbs, 1985; Korpimäki & Norrdahl, 1999; Redpath et al., 2002). The outputs from these studies can vary from simply confirming the presence of a prey species in a diet, to being able to quantify either the total number of individuals taken or the net energy gain afforded by predation on a specific prey species (Seaton et al., 2008). When dietary analysis is conducted in conjunction with demographic modelling of a prey population, more complex ecological processes can be studied such as the impact of predators in terms of prey species mortality (Thirgood et al., 2000a), or the functional responses of predators to fluctuations in prey abundance (Redpath & Thirgood, 1999). In this way dietary analysis of raptors can provide essential information to policy makers, conservationists and managers; indeed some conservation strategies now focus on prey as an essential component for maintaining populations of avian predators (Reynolds et al., 1992).

Assessments of buzzard diet have previously been attempted through recording either the prey remains present around the nest site (Jardine, 2003; Sim, 2003), or through recording both prey remains and prey items identified from pellet analysis (Graham et al., 1995; Swann & Etheridge, 1995). Often these methods are the only source of reliable data on diet as studying the prey selection of wide ranging raptors is extremely difficult. Although these two methods form the vast majority of all previous dietary analysis of buzzards, it is known from work on other raptor species that both methods (pellets and prey remains) have inherent biases (Simmons et al., 1991; Redpath et al., 2001; Tornberg & Reif, 2007). While it has been suggested that incorporating prey and pellet results can serve to rectify the biases of both methods (Simmons et al., 1991), studies that have addressed this since have reached different conclusions as to the accuracy of this approach (Redpath et al., 2001; Lewis et al., 2004; Sánchez et al., 2008).

It would seem appropriate therefore, to use direct observations to account for the biases of these two commonly used methods, something that few studies, with the exceptions of Selàs et al. (2007) and Tornberg & Reif (2007), have so far attempted with buzzards. Indeed, Selàs’s study in southern Norway showed that amphibians have been significantly underestimated in studies from both pellet and prey remains. To collect direct observational data, studies have used both remote cameras (Lewis et al., 2004) and observations from blinds (Redpath et al., 2001). However, as well as reducing data collection time, it has been shown that cameras allow for a 9 greater proportion of prey to be identified to genus and species levels when compared to observations from blinds (Rodgers et al., 2005). It is also the case that relatively recent advances in digital technology have created the capability to record nest activity over longer periods of time, allowing for cost effective observations of all prey items provisioned over a period of many days (Bolton et al., 2007).

3. Study site Fieldwork was conducted across 298.2km² of Stirling and the Trossachs, Central Scotland. The area contain three dominant habitat types, forestry, upland pasture and lowland pasture. The climate of the study area was cool and humid with a mean annual precipitation of 1344mm and an annual mean daily temperature of 4.9°c minimum and 11.9°c maximum [Ardtalnaig meteorological station annual means, 1971-2000, located approximately 30km north east of the centre of the study area; Meteorological Office (www.metoffice.gov.uk)]

3.1. Forestry and Upland pasture

The upland study area was located within or around the Queen Elizabeth Forest Park, part of Loch Lomond and The Trossachs National Park (56°10’N 4°23’W), a 167km² area consisting of three large planted forests owned and managed by the Forestry Commission Scotland. Although the area is predominantly planted with non-native coniferous forests, the park also contains other habitat types including native forest, rough grassland, sheep walk and heather moorland.

The native section of the forest was characterised by downy birch (Betula pubescens), silver birch (Betula pendula), scots (Pinus sylvestris) and sessile (Quercus petraea). The harvested section of the forest is characterised by coniferous trees primarily Sitka spruce (Picea sitchensis), Norway spruce (Picea abies), European larch (Larix decidua), larch (Larix x eurolepis), scots pine and lodgepole pine (Pinus contorta) managed on a 25-60 year rotation. The harvesting of timber creates clear areas of successional habitats: when trees are first removed a ‘clear cut’ area is formed, this will then be left for several years becoming grassland, after a period of time it is then planted to become ‘pre-thicket forest’, finally it reaches the point of becoming a thicket after 12-15 years. The Forestry Commission also creates ‘forest habitat networks’, areas of natural regeneration and grassland left for their biodiversity value.

3.2. Lowland farmland: Doune

The lowland section of the study area was centred on the town of Doune (56°11′24″N 4°03′11″W) within the Stirling district. Around the town the area was made up of small arable and pastoral farms, with sheep farming as the most common farming type. Most fields were bordered by hedgerows with many small patches of native and non-native forestry. In 10 the lowlands the native deciduous woodland was dominated by common oak (Quercus robur), beech () and sycamore (Acer psuedoplantus).

11

4. Methods 4.1. Field data collection

4.1.1. Habitat classification Within the study area, vegetation types were split into nine different categories following Austin et al. (1996) and after consultation with Katie Freeman (Habitats Manager for the Trossachs region of the Forestry Commission Scotland) (Table 2). Each habitat was then individually sampled for prey species abundance.

Table 2: Categories used to classify vegetation cover within the study areas. Adapted from the habitat classification in Austin et al. (1996), plant names follow Clapham et al., (1981).

Habitat type Description Habitat classification 1 Mature non- Forestry plantations after canopy closure and through to Forestry native mature tree stands. Characterized by spare or near absent woodland herb or shrub layer. The only open areas are rides between stands. 2 Mature native Corresponds either deciduous woodland or mixed Forestry woodland deciduous and as marked on 1:25 000 scale Ordinance Survey Landranger series maps. Within the study site oak and birch were the dominant species. 3 Clear-cut Areas where stands had been cleared within the last 3 Forestry areas years. Sites that had been replanted with saplings within the last year were also included in this category. 4 Pre-thicket Young forestry plantations 1-12 years old (before canopy Forestry forestry closure). Characterized by lush herb layer and shrub birch with much open space between lines of planting. 5 Rough Characterised by uncultivated or ungrazed areas with Upland grassland and native tree species before canopy closure. Grasses such pasture natural as Molinia and bracken Pteridium aquilinum dominated. regeneration These included the Forestry Habitat Areas created by the Forestry Commission Scotland. 6 Heather Characterised by cover species such as vulgaris, Upland moorland Vaccinium myrtillus and Nardus stricta. pasture 7 Upland Mountainous or hillside pasture on the boarders of the Upland farmland or National Park. Vegetation characterised by perianal pasture 'sheepwalk' grasses such as Festuca ovina and Agrostis tenuis. 8 Lowland Low lying intensively grazed or cultivated farmland. Lowland farmland Predominantly sheep farming with some silage fields. pasture 9 Other Rare vegetation types or land uses, for example quarries, Unspecified were not included.

4.2.1. Prey Sampling Scottish buzzards consume a wide variety of birds, mammals and reptiles including rabbits, voles, corvids, galliforms, small passerines and (Swann & Etheridge, 1995; Graham et al., 1995). This generalist diet necessitates a sampling method capable of estimating both bird and

12 diurnal mammal populations concurrently. For this reason both Distance sampling and vole sign index (VSI) sampling were chosen to be used in conjunction with one another.

4.2.1.1. Distance Sampling for prey species The abundance of avian and mammalian prey species was assessed through an adaptation of a transect sampling methodology recommended for distance analysis (Buckland et al., 2001) and commonly used in the Breeding Bird Survey (BBS). Distance analysis was judged as the correct method to obtain density estimates as it produces a detection function that compensates for differences in detection probability across species and habitats and, as a result, can produce density estimates representative of true population size (Buckland et al., 1993). Transects were selected in each of the eight habitat types by overlaying a 1km by 1km grid over mapped habitats and randomly selecting squares for sampling. Two transects were walked at each location running parallel at between 250-500m distance. In total eight 1km transects were walked through each of the defined habitat types between the hours of 06:00am and 10:00am. This time period was chosen for three reasons (1) it is recommended by the BBS methodology as being when birds are most active; (2) it has been used by previous studies for sampling lagomorph abundance (Trout et al., 2000) and; (3) it has been noted as the time of day when buzzards are most commonly observed, suggesting that is when they are most actively searching for prey (Vergara, 2010). Transects were walked slowly by a single observer and numbers of individuals of any avian or mammalian prey species, as listed by Graham (et al., 1995), Swann & Etheridge (1995), or Selàs et al. (2007), on both sides of the transect line were recorded. Avian prey was split into 7 different categories for ease of identification and recording: gamebirds, waders, columbids, small passerines, thrushes, corvids and other. Transects were not conducted on days with strong wind or heavy rain as this may have caused a reduction in activity.

4.2.1.2. VSI sampling As voles have been shown to be an important component in the diet of buzzards (Newton, 1979; Swann & Etheridge, 1995), the relative abundance of field and bank voles was assessed using a VSI sampling method adapted from extensive vole research conducted in the Kielder Forest, Northern England (Lambin et al., 2000; Petty et al., 2000). The same 1km transects as used for avian prey densities were walked with a quadrat (25m x 25m) thrown every 40m, creating a total of 25 different VSI recordings along the transect course (200 in total for each habitat type). Each quadrat was thoroughly examined for the presence of absence of vole sign, specifically fresh clippings. Quadrats were then scored 0, 1 or 2 depending on the sign of clippings present: old clippings = 1; fresh clippings = 2, making a maximum score of 50 for each transect line. Petty (1992; 1999) provided evidence that fresh clippings are an appropriate index indicator of vole abundance, as clippings do not remain fresh for longer than 1-2 weeks and so

13 accurately reflect recent vole activity. Although, this method does not allow a differentiation between species, previous studies indicate that the will be the most dominant species in open habitats and the the most common in the wooded areas (Hanski et al., 2001).

4.2.2. Mapping buzzard nests Previous nesting locations were taken from a long running program being undertaken by David Anderson (Forestry Commission Scotland) and Duncan Orr-Ewing (RSPB Scotland). Nests were located during the nesting period (20th April- 14th July 2011) by systematic foot searches of all know woodland containing nesting sites. This was necessary as buzzards will often use different nests within a single territory during different years (Tubbs, 1974). Additional nests were located from observations of territorial birds in previously unrecorded locations. To prevent misclassification of nesting attempts nests were only considered active if: 1) it was observed with both a substantial build up of fresh greenery around the nest cup and a territorial pair of birds; 2) the female was observed incubating eggs; 3) chicks were observed on the nest. The GPS co-ordinates of all active nests were recorded at nest site. When last year’s active nest was found to be inactive, a 500m radius around the nest was searched, if an active nest was not located after a thorough search (a minimum of 2 visits) the territory was considered to have been relinquished, or, if the buzzard pair was regularly seen together defending the area, it was counted as having already failed. This is because field work began in late April when pairs may already have abandoned nests either due to infertility or predation (Newton, 1979).

14

Figure 1: Map of study site showing all active nests. 4.1.3. Mapping core territories Once all active sites were located (Figure 1), the nearest neighbour distance (NND) for each nest was recorded using ArcGIS. Buzzard pairs defend an established territory while raising young (Hodder et al., 1998), for the purpose of this study they were assumed to conduct the majority of their hunting within the core territory surrounding the nest. To calculate core territories the NND/2 was averaged for each of the three nest categories providing three different buffer sizes. Once a buffer was established, the area (m²) of each of the eight habitat types within its radius was then mapped out using digitised ordinance survey maps (1:25,000), Forestry Commission Scotland GIS files and Google earth© images on ArcGIS (see Figure 2 for example). The areas of habitat within territories later allowed for prey abundance estimates to be calculated for each nest.

15

Figure 2: The habitat around 63 nests was mapped on ArcGIS and the clipped with the mean nearest neighbour distance (black circle). In this case the nest was classified as ‘Forestry dominated’ and a buffer with a radius of 550m (NND of habitat category/2) was used. 4.1.4. Assessing Buzzard diet

4.1.4.1. Pellet analysis and prey remains An assessment of buzzard diet was made from all active nests in the study (n=38). Pellets and prey remains were collected from within the nest cup and from plucking posts located within the vicinity of the nest (judged as a 50m radius). Nests that had recently failed (presence of fresh down or dead chicks) were included in this analysis but yielded few records. Prey remains were identified and recorded for each nest individually; the number of individuals of any prey species was recorded as the minimum number from the prey remains present, for instance, if the nest cup comprised entirely of vole fur, the minimum number of voles would have been estimated as being between 5 and 10. Prey remains identification and estimates of individual numbers were made with the assistance of David Anderson, an experienced raptor researcher.

16

Figure 3 Prey remains ( Talpa europaea and field vole agrestis) found within the nest (left), buzzard pellets before examination (right). Pellets were analysed in the field under a crude adaption of the method used by Clarke et al. (1997). It was assumed that each species recorded in a pellet was represented by no more than one individual. Bird identification from pellet analysis was confined to noting the presence or absence of any bird remains in pellets as feathers are difficult to identify to a species level (Graham et al., 1995). In the same way rabbits Oryctolagus cuniculus and brown Lepus europaeus were classed together as lagomorphs, because lagomorph hair can only be separated to species at the microstructural level (Wolfe & Long 1997). Pellets in general were less common than prey remains, as a consequence nests were often revisited several times to ensure that pellets were adequately represented. While the occurrence of prey species in pellets is unlikely to be comparative to the quantities of prey ingested, the results of prey and pellet analysis will be useful in comparing the dietary proportions of different territories.

4.1.4.2. Direct observations While pellet and prey remain data are easy to collect and record, several studies have shown that direct observational data is necessary to correct inherent biases and shortcomings in dietary analysis of raptors (Collopy, 1983; Redpath et al., 2001, Tornberg & Rief, 2007). To gather data on buzzard diet, six Memocam© nest cameras were installed on buzzard nests, two in each habitat category. Cameras were installed on nests chosen at random with all brood sizes encountered in the study being represented (1 brood of three, 1 brood of two and 4 broods of single chicks). Due to a technical malfunction, data from only one camera was gathered in the forest dominated territories. These cameras did not record constantly but had individually programmed motion detection triggers which were activated on the arrival of the parent bird on the nest.

17

To identify prey, a detachable memory card within the camera units was uploaded onto a laptop (15in screen) to view footage. Prey items were identified to the lowest taxonomic level possible, based on morphological characteristics such as feet, tails and body shape. On occasions that the prey item was only partly viewed or consumed quickly prey were categorised to a more general category of or genus. All unidentified prey items were small and consumed rapidly. For each nest, this data was collected for 100 ‘hunting hours’ over consecutive days usually between the hours of 06:00am and 22:30pm (when the parent bird left in the morning and returned to the nest at night). The target of 100 hours was chosen due of time constrains and was usually reached on the 9th day of recording. Recordings were made during June and July in the weeks preceding fledging. This period was chosen to maximise the number of prey being brought to the nest as both the male and female adult should be hunting constantly (Tubbs, 1974).

4.2. Data Analysis

The two measures of buzzard breeding ecology (density and breeding success) where analysed in three stages. Firstly, a large proportion of the analysis was directed at the preparation of data for use in models. Secondly, the data was explored with conventional univariate analysis methods. Thirdly, a multivariate analysis was conducted using extracted values from a principle component analysis to evaluate how the habitat and dietary variables interact to influence buzzard breeding ecology. All analyses were considered significant if P value is <0.05.

4.2.1. Preparatory analysis

4.2.1.1. Prey densities To produce density estimates (individuals/hectare) distance sampling data were analyzed using the program DISTANCE (Version 6.0, Thomas et al., 2009). All data from different habitats were tested for fit with the key functions of Uniform, Half-normal and Hazard-rate under conventional distance sampling (Buckland et al., 2001) (see Appendix Figures 2-9). To ensure reliable estimates of density, methods were designed to meet the three key assumptions of distance sampling: 1) all distances were measured accurately; 2) all individuals were detected at their initial location and; 3) all individuals on the transect line were detected. Density estimates were determined only for small passerines as they were the only group of prey with a sufficient sample size of >80 detections (Buckland et al., 2001). Vole abundances were determined from the VSI sampling through a liner regression detailed in Lambin et al. (2000) (See Appendix: Table 1). As the VSI sampling period took place over both the spring and summer months both were calculated and then averaged for each habitat type. The densities of both voles and small passerines within the core territory of each buzzard were then calculated from the area of each habitat type within each territory. To allow for comparable results between the different habitat

18 categories, the abundance of both voles and small passerines was calculated per km² of core territory.

4.2.2. Variable compilation Sixteen variables were used to describe nests comprising of five landscape variables, 5 dietary variables from the pellet analysis and 6 dietary variables from prey remains analysis (for full list and description see Table 5). The dietary variables had previously been reduced from twenty two by excluding variables with only one nest record.

Table 3: Variables used in the description of buzzard nests in relation to both landscape and dietary analysis.

Variable Description Response NSTSUC Number of chicks raised to fledging NND Distance of closest nesting attempt (used as a proxy for density) Explanatory Landscape variables %F Percentage of habitat made up of habitat types classed under 'Forestry' %UP Percentage of habitat made up of habitat types classed under 'Upland pasture' %LP Percentage of habitat made up of habitat types classed under 'Lowland pasture' VKM Voles density in an average km² of core territory SPKM Small passerines density in an average km² of core territory Pellet analysis variables PLL Importance of lagomorphs in diet as estimated from pellet analysis PLMO Importance of moles in diet as estimated from pellet analysis PLV Importance of voles in diet as estimated from pellet analysis PLB Importance of birds in diet as estimated from pellet analysis PLA Importance of amphibians in diet as estimated from pellet analysis Prey analysis variables PRL Importance of lagomorphs in diet as estimated from prey remain analysis PRM Importance of moles in diet as estimated from prey remain analysis PRV Importance of voles in diet as estimated from prey remain analysis PRB Importance of birds in diet as estimated from prey remain analysis PRA Importance of amphibians in diet as estimated from prey remain analysis PRC Importance of carrion in diet as estimated from prey remain analysis

4.2.3. Exploratory univariate analysis An analysis of variance (ANOVA) was fitted to analyse the influence of habitat type on nearest neighbour distance and laying date. Variation in productivity between habitat was explored using a Chi squared test on nesting success (measured as 0 or 1) and an ANOVA for number of fledglings. Further analysis on fledglings required Tukey’s Honest Significant Difference test (Tukeys HSD).

19

4.2.3.1. The influence of prey abundance on breeding ecology Analysis of the variation in prey abundance required an ANOVA to be fitted across the habitat categories and a Tukeys HSD to observe multiple comparisons. A Pearsons product movement correlation analysis (Pearsons) was then used to analyse the effect of prey abundance on productivity, laying date and density.

4.2.3.2. Exploratory dietary analysis Prey and pellet data was checked for normality before analysed using a Wilcoxon sign rank test (Wilcoxon) to address if techniques produce significantly different results.

4.2.4. Investigating response type of buzzard to vole density To investigate the type of response shown by buzzards to increases in vole density (Hypothesis D) the proportion of voles in diet (estimated from pellets) was tested against the vole abundance estimates for each territory. A generalised liner model (glm) with log link and quasipoisson errors was fitted with proportion of voles as a quadratic element to investigate how the proportion of voles in diet changed with vole abundance in territories.

4.2.5. Investigating the influence of prey size in breeding success In order to explore the role of prey size in breeding success a brief analysis was conducted concentrating on mammalian prey. Separate analyses were run for both prey remains and pellet analysis that ranked the dominant mammal prey species in the diet of each nest depending on size. All mammals <100g were ranked as 1 while all mammals .100g ranked as 2. An ANOVA was then fitted for both methods.

4.2.6. Principal component analysis A principal component analysis (PCA) was conducted on both the landscape and dietary variables to reduce the dimensionality of the dataset. Previous studies on raptor breeding ecology have shown that a PCA analysis can be appropriate to analyse the complex relationships within numerous variables (Kostrzewa, 1996; Krüger & Linderström, 2001; Baines et al., 2004; Lŏhmus, 2003; Rodriguez et al., 2010). Following a reduction of the variables, loadings were used to interpret the characteristics of the individual PCs. Only PCs with an eigenvalue >1 were be considered for further analysis (Crawley, 2007).

4.2.7. Modelling While the conventional univariate analysis allowed for an initial exploration of the data, model selection, a process in which several competing hypothesis are simultaneously confronted with data, is being increasingly promoted by environmental scientists (Burnham & Anderson, 2002; Johnson & Omaland, 2004). Model selection, based on likelihood theory, can be used to identify either the best single model or to make inferences based on weighted support from a set of competing models. This is appropriate for this study as the environmental variables that govern 20 buzzard densities and breeding success are likely monotonically related to many underlying factors as well as each other. Consequentially, it is likely to prove difficult to reject any of the hypotheses with a single univarite test. By using the approach of model selection, several models, each representing one hypothesis can be simultaneously evaluated in terms of support from the landscape and dietary variables.

Two generalised liner models (with poisson errors and log link) were built using the five principle components to explore the relationship between nesting density and breeding success with changes in the composition of habitat and diet. The R package ‘MuMIn’ was then used to run each model with all possible combinations of variables. From the results of this analysis the model with the lowest AIC was selected as the best model. AIC selects the best model taking into account the fit of the model and penalises for each extra parameter in the model (Sakamoto et al., 1986). The delta AIC (∆i) value provided for a measure of each model relative to the best model. A confidence set of models was obtained by extracting all model combinations with a ∆i

<2, as any models within 2∆i can be considered equally probable (Burnham & Anderson, 2002). The final group of models were those that supported the data strongest.

21

5. Results 5.1. Exploratory analysis

5.1.1. Variation in buzzard density on a spatial scale A total of 65 nesting attempts were recorded spanning three habitat categories: forest dominated territories (19), upland pasture dominated territories (25) and farmland dominated territories (22). Three nesting attempts (all upland pasture dominated) were excluded from further analysis after a review of the data showed them to be outside the main study area (>5km). The mean NND between all buzzard nests was found to be 0.96±0.50km (range 220-2499m, n=63) with the forestry, upland and lowland habitat categories producing NND means of 1.1±.043km, 0.99±0.51km and 0.79±0.52.4km respectively. The distances between nearest neighbours was tested between habitats with no significant differences observed (ANOVA,F2,60=2.08, df= 2, p=0.13) (See Figure 4).

Figure 4: A boxplot showing the nearest neighbour distance of buzzard nests against the habitat categories. Box represents interquartile range, bold lines represent the median, dotted line represent range, clear circles represent outliers. 5.1.2. Variation in buzzard productivity on a spatial scale From the 63 active nests monitored, this study noted a failure rate of 43% across all habitats, with forestry nests, upland pasture nest and lowland nests recording failure rates of 74%, 41% and 19% respectively. A Chi-squared showed that there was a statistically significant difference between the observed values and a uniform failure rate across the habitats (Chi-sq²= 12.88, df=2, P<0.001).

Further exploration of productivity revealed a mean fledgling number of 0.89±0.31 across all habitats with mean fledgling numbers of 0.26±0.45 in forestry, 0.73± 0.76 in upland and 1.41± 22

1.05 in lowland habitats. A highly significant difference was found between at least two of the habitat types (ANOVA: F2,60=2.08, df=2, p=<0.001). A Tukeys HSD was then used to show that there was a significant difference between lowland and forestry (diff=1.145, p=0.0001), upland and lowland (diff=-0.68, p=0.018) but no significance between upland and forestry (diff=0.46, p=0.168) (this is visualised in Figure 5).

Figure 5: A boxplot showing the number of buzzard chicks raised to fledging against habitat type. Box represents interquartile range, bold lines represent the median, dotted line represent range, clear circles represent outliers. 5.1.3. Prey densities The abundance of voles and small passerines was calculated for each of the eight habitat types (per km²) (Tables 4 & 5). An ANOVA showed that vole abundances per km² were significantly different across the habitat categories (F2,60=74.25, df=2, p<0.001). Significant differences were found between lowland and forestry territories (Tukey HSD: diff=-5.13, p<0.001), upland and lowland territories (diff=5.93, p<0.001) but no significant difference was found between upland and forestry territories (diff=0.80, p=0.31). Following a significant result from an ANOVA of small passerines per km² in the different habitat types (f=15.46, df=2, p<0.001), a Tukeys HSD revealed a highly significant difference between lowland and forestry territories (diff=-1.05, p<0.001), a significant difference between upland and lowland territories (diff=-0.65, p=0.003) but little significance between lowland and upland (diff=-0.40,p=0.08).

23

Table 4: The estimates of small passerine abundance from DISTANCE analysis. All models were run under a half-normal hermite polynomial distribution with observations binned and the largest 5% truncated.

Detection Encounter Dˉʰ Habitat type n probability rate estimate %CV df 95% CI Non-native mature forestry: 93 35 60.5 3.21 14.33 13.42 2.36 4.36 Native mature woodland: 279 68.5 30 7.21 11.59 68.61 5.72 9.07 Rough grassland: 134 20.9 76.7 5.38 15.82 8.48 3.75 7.7 Pre-thicket forestry: 224 23.1 76.1 3.88 11.52 12.04 3.02 4.98 Clear cuts: 67 19.3 80.1 0.64 23.04 16.98 0.4 1.04 Upland pasture: 114 80.6 11.8 1.65 8.41 128.17 1.4 1.95 Heather moorland: 121 28.6 69.8 2.5 13.31 14.16 1.88 3.31 Lowland pasture: 87 27.3 69.3 1.54 16.1 14.4 1.09 2.16

Table 5: The calculation of vole density from the regression detailed in Lambin et al., 2000 (See Appendix 1 for further details)

Habitat type VSI Spring Summer Average (Dˉʰ density density Estimate) Native mature woodland 2.63 44.35 51.64 47.99 Non-native mature 3.25 52.54 60.46 56.50 woodland Rough grassland 13.63 188.45 206.85 197.65 Pre-thicket forestry 13.88 191.72 210.38 201.05 Clear-cut areas 8.25 118.04 131.01 124.52 Heather moorland 7.63 109.85 122.19 116.02 Upland farmland 8.50 121.31 134.54 127.92 Lowland farmland 2.50 42.71 49.88 46.29

To begin to address hypothesis A, vole abundance was tested against buzzard productivity, revealing a highly significant negative relationship (Pearsons: df=61,t=-3.38,p<0.001) (Figure 6). This result allows the null hypothesis to be rejected, however the expected positive correlation was not found. To test hypothesis B, the influence of vole abundance was tested against a proxy for buzzard density; NND. This revealed a significant positive correlation (df=61,t=2.28,

24 p=0.025) (Figure 6). This again allows the null hypothesis to be rejected but is converse to the expected negative correlation.

Figure 6: Scatter graphs showing the vole abundance estimates in buzzard territories against the nearest neighbour distance (left) and the number of chicks raised to fledging (right). Regrassion lines were fitted and both the explanitory variables and the response variables were logged to normalise the data. The same analysis provided little evidence for an effect of abundance of small passerines in territories on NND (Pearsons: df=61,t=0.107,p=0.915) and productivity (df=61,t=- 1.57,p=0.121).

5.2. Results of buzzard dietary studies

A total of 170 fresh prey remains (average ±1sd=4.7±2.6) and 118 fresh pellets (2.2±1.4) were collected over the course of this study. Voles and birds formed 64.7% of buzzard diet in prey remains and 57.8% of diet in pellet analysis (Table 4). Lagomorphs, predominantly rabbits, (23.5% of prey, 22% of pellets) also formed a significant component of buzzard diet. The frequency of occurrence estimated by the two methods was investigated for disparity by plotting the medians to show that lagomorphs and birds occurred more frequently in prey remains and voles and mole occur more frequently in the pellets. This proved significant for voles (Wilcoxon: V=19,p<0.001), and birds (V=531, p<0.001) but not lagomorphs (V=60, p<0.68).

From the prey remains, voles where recorded the most often (29.4%) while lagomorphs made up the second highest group of observations (23.5%). In total, 21 different species of bird in buzzard diet including three gamebird species ( L. lagopus scoticus, 25

Lyrurus tetrix and Pheasant Phasianus colchicums). However of the bird species, it was corvids, constituting 13.6% of the prey remains found, that appeared to be predated at the highest level. From the pellets analysis, voles were the most common prey item (42.5%) followed by rabbits (22%), birds (15.3%) and moles (9.3%). Amphibian bones were also recorded (8.5%) but did not allow for a distinction between Common temporaria and Common bufo.

Table 6: Number of observations and percentage total diet of different prey items as estimated from prey remains collection and pellet analysis (for full list including direct observations see Appendix 2).

Prey remains Pellet analysis Total % Total % Amphibians 7 4.1 10 8.5 Reptiles 0 0 1 0.8 Birds 60 35.3 18 15.3 Red Grouse L. lagopus scoticus 1 0.6 - - Black grouse Lyrurus tetrix 1 0.6 - - Pheasant Phasianus colchicus 7 4.1 - - Woodpigeon Columba palumbus 4 2.4 - - Tawny Strix aluco 3 1.8 - - Small passerine species 16 9.5 - - Turdus species 4 2.4 - - Corvid species 23 13.6 - - Mammals 98 57.6 89 75.4 Erinaceau europaeus 1 0.6 1 0.8 Mustelid species 0 0 1 0.8 mole Talpa minor 5 2.9 11 9.3 Lagomorph species 40 23.5 26 22 Vole Cricetidae species 50 29.4 50 42.4 Mouse species 1 0.6 0 0 Soricidae species 1 0.6 0 0 Carrion 5 2.9 0 0 Total observations/items 170 118

5.3. Investigating the type of functional response

Although the results of a generalised liner model showed that vole number estimated from pellets increased with number of voles in the territory (t=5.007, p<0.001). When vole proportion was added as a quadratic element, the functional response type could not be confirmed (t=1.773, p=0.137).

5.4. The influence of prey size in breeding success

Hypothesis E was investigated by investigating the relationship between the size of the dominant mammalian prey against fledgling number. The results indicated that size of prey has a highly

26 significant influence on the number of fledglings from both pellet analysis data (ANOVA,

F1,31=21.42,df=31,p<0.001) and prey remains data (F1,31=13.04,df=29, p<0.001) (See Figure 4). Indeed, the mean number of fledglings raised predominantly on a diet of mammals <100g (voles) was estimated at 0.9 from assessments made by both methods (prey and pellets).This can be compared to the mean fledgling numbers from nests where the dominant mammal prey species was >100g (rabbits and moles) which was estimated at 1.73 from prey remains and 2.0 from pellet analysis. These results allow the null hypothesis to be rejected.

Figure 7: Bargraphs with 95% confidence intervals attached showing the relationship between the dominant mammal species found during the pellet analysis (A) and the prey analysis (B) against the mean number of chicks raised to fledging (mammals <100g=1, mammals >100g =2). 5.5. Principle component analysis

The PCA reduced the 16 variables to 5 factors (eigenvalues > 1) that retained 76.6% of the original variance (Table 8). The first principle component (PC1) was clearly interpreted as being ‘vole reliant’ nests. The analysis showed high positive loadings toward all vole influenced variables (voles abundance per km² VKM, proportion of vole in pellets PLV, and proportion of voles in prey PRV) as well as a negative loading towards lagomorphs in prey (PRL) and lagomorphs in pellets (PLL). PC2 was interpreted as rough upland habitat due to the high loading of percentage upland pasture (%UP), the low loading of percentage forest (%F), and the significant loading of bird in prey (PRB) and amphibians in pellets (PLA). PC3 was determined to be related to a generalist pasture as the proportions of bird remains in pellets (PLB) has a significant negative loading while proportion of mole prey (PRMO) and the proportion of carrion prey (PRC) were positive. PC4 was related to lowland territories with a diet of rabbits as there was a negative loading on both proportion of forest in territory (%F) and small passerines 27 per km² (SPKM) and a positive loading on the proportion of lagomorphs in prey (PRL). PC5 was taken to be reflecting dry pasture territories as there were significant loadings from both PRMO and proportion of mole in pellets (PLMO) and prey carrion (PRC) with negative loadings for proportion of amphibians in pellets (PLA), interestingly there was also high negative loadings for both measures of lagomorphs in diet.

Table 7: Importance of habitat and dietary variables of buzzard with respect to each varimax- rotated factor of the principal components analysis (PCA). Factor loadings > 0.3 are indicated in bold.

PC1 PC2 PC3 PC4 PC5 ‘Vole ‘Upland ‘Pasture ‘Lowland ‘Dry reliance’ generalist’ generalist’ rabbit’ pasture’ Habitat variables VKM 0.34 0.13 -0.01 0.02 -0.13 SPKM 0.28 -0.12 -0.16 -0.48 -0.13 %F 0.25 -0.26 -0.10 -0.45 0.15 %UP 0.30 0.31 -0.13 0.22 -0.16 %LP -0.37 -0.02 0.17 0.13 0.01 Dietary variables PLL -0.29 -0.16 -0.05 -0.20 -0.26 PLMO -0.07 0.39 -0.12 -0.33 0.36 PLV 0.36 -0.11 0.23 0.06 0.04 PLB 0.04 0.17 -0.49 0.41 -0.08 PLA 0.09 0.42 0.22 0.03 -0.34 PRL -0.31 -0.18 -0.06 0.36 -0.37 PRMO 0.06 0.36 0.40 -0.28 0.41 PRV 0.34 -0.16 0.05 0.14 -0.01 PRB -0.10 0.44 -0.18 -0.15 0.38 PRA 0.20 -0.15 -0.34 0.06 -0.17 PRC 0.11 -0.13 0.49 0.23 0.36 Importance of components: Standard deviation 2.465 1.484 1.323 1.092 1.016 Proportion of variance 0.380 0.138 0.109 0.074 0.064 Cumulative proportion 0.380 0.518 0.627 0.701 0.766

28

Table 8: Competing GLMs explaining fledgling number (with Poisson errors) and nearest neighbour distance (with quasi Poisson errors) of Buzzard in the study area. For each model, the AIC, the difference between the current model and the best model (Δ AIC), and the Akaike weights are given. Variables retained in all competing models are in bold.

Model/variables AIC Δ AIC weight Fledgling number: 'vole reliance' 76.58 0 0.22 (Null model) 78.4 1.51 0.1 'vole reliance' + Pasture generalist 77.62 1.54 0.1 Nearest neighbour distance: (QAICc) 'vole reliance’+'upland generalist'+'dry pasture' 38.77 0 0.15 'upland generalist'+'dry pasture' 39.15 0.38 0.12 'vole reliance'+'upland generalist' 40.05 1.27 0.08 'upland generalist' 40.11 1.34 0.08 'upland pasture' + 'pasture generalist' 40.72 1.95 0.06 (Null model) 44.96 6.19 0.01

5.5.1. PCA analysis: fledgling number When the PCA variables were tested against fledgling number the model that best explained the differences in breeding success and against landscape and dietary variables was the PC interpreted as ‘vole reliance’ (PC1) (Table 9). This model showed that PC1 has a significant negative effect on the number of chicks raised to fledging (0.348±0.035,df=27,t=-3.608,p<0.01) (Figure 8). However, the null model was within 2 Δ AIC (1.51 Δ AIC), indicating that the null hypothesis cannot be rejected (Burnham & Anderson, 2002).

29

Figure 8: The relationship between PC1 ‘vole reliance’ and the number of chicks raised to fledging. The dotted lines represent 95% confidence intervals. 5.5.2. PCA analysis: Nearest neighbour distance Finally, the relationship between NND and the PCA variables was explored through model selection, this identified five models within 2 Δ AICs, of which the PC related to ‘Upland generalist’ (PC2) featured in all. This principle component was statistically significant when modelled against NND (0.143±0.055,df=27,t=2.61,p<0.05). The model with the lowest AIC (see Table 9) was examined with ‘upland generalist’ (PC2) emerging as the only significant variable (O.119±0.08, t=2.212,p<0.05) (Figure 9).

30

Figure 9: The relationship between PC2 ‘upland generalist’ and the nearest neighbour distance, measured in meters. The dotted lines represent 95% confidence intervals. 5.6. Direct observational data

Due to a low sample size (n=5), the observational data collected from the nest cameras was not appropriate for complex quantitative analysis. However, the large number of prey records (n=226) from the 500 hours of footage did provide the opportunity for a rough comparison of the dietary techniques through a comparison of dietary proportions. This indicated that the proportions of diet estimated from prey and pellets in this study may have large biases ( Figure 10).

31

Figure 10: Bargraphs showing the proportions of mammals, birds and amphibians observed in all dietary methods in forestry (top left), upland pasture (top right), lowland pasture (bottom left) and all habitats (bottom right). In mammal species, voles only formed 20.5% of the diet, with the closest estimate coming from prey remains (29.4%) with pellet analysis providing an estimate over double that of observed (42.4%). Lagomorphs formed only 5.7% of overall diet, far lower than the estimates from the other methods (prey =23.5%, pellet=22%), despite being observed as prey items on 3 out of 5 nests. The source of the biases for proportions of mammal and bird prey may well have originated from a substantial underestimation of the perceived importance of amphibians in buzzard diet. From all five cameras amphibians were recorded as being a key component in buzzard diet (average 35.7%) comprising the highest proportion of prey taken in both ‘Upland Pasture’ nests (40% and 43%). Another interesting observation was the number of prey, adders (n=4) and slow (n=1) were observed being brought onto 3 separate nests yet only recorded once in both the prey and pellet analysis during the main dietary study.

32

6. Discussion

The overall aim of this project was to explore the factors that govern buzzard breeding success and density as a first step in understanding if buzzard can have a significant impact on gamebird harvests. The data confirms the importance of large prey items in breeding success and suggests that upland habitat may influence breeding densities.

6.1. Overview

Understanding the temporal and spatial variation in the reproduction of raptors requires large numbers of territories to produce reliable estimates (Newton, 1979). Reynolds et al. (2005) estimated the minimum threshold for accurate inferences of generalist raptors to be between 60- 100 nests. Although single year studies such as this one do not allow for a comparison of territory productivity on a temporal scale (Lŏhmus, 2003), the large sample size and the heterogeneity of the study site permit for an interesting snap shot of how habitat and prey availability constrain or promote buzzard productivity on a spatial scale.

The densities of buzzards observed in this study (NND mean= 0.96±0.50km) are comparable and often higher than those reported in previous studies (See Table 1). Although the exploratory analysis did not reveal a significant difference between the three habitat types, the results are similar to those of Newton et al. (1982) who recorded mean NNDs in Wales to be 1.13±0.04km in upland pasture (the observed was 0.99±0.51km) and 0.87±0.03 in lowland areas (the observed was 0.79±0.52). In this study NND was used as a proxy for density and probably gave an overestimate of abundance because (by definition) it did not account for the distances to other neighbours or gaps in the distribution of breeding pairs. However, it can be assumed that buzzard density within the study site was high as NND estimates are similar to those recorded by Sim (2003) in a population deemed to have one of the highest densities in (0.83±0.27). Indeed, the number of nesting attempts within the study site has remained fairly constant over the last 5 years (D. Orr-Ewing, RSPB, unpublished data) suggesting that the population is reaching an asymptote.

In terms of general productivity, although fluctuations are common over landscapes (Swann & Etheridge, 1995; Sim et al., 2001), this study found a surprisingly high degree of spatial heterogeneity. In the breeding attempts for the lowland pasture for example, only 19% of nests failed to rear young, this can be compared to the forestry nests from which a failure rate of 74% was recorded. Buzzard populations require a recruitment standard of 1.15 young per productive pair to ensure stability (Newton, 1979). Without taking emigration and immigration into account

33 it would appear that during this year at least, lowland pasture was the only habitat category to exceed this, making it a probable source population. Although similar failure rates have been reported, such low breeding success is rare in the literature (Table 1). At first glance, it would seem contradictory to the persistence of populations to attempt to breed in areas with such low success rates, however, a feasible explanation for this is the site dependant population regulation hypothesis (SDRH) (Rodenhouse et al., 1997). This predicts that the high densities of buzzards within the study area are causing individuals to inhabit increasingly poorer territories, decreasing the per captia population growth rate.

Although the SDRH explains why buzzards attempted to lay secondary habitat, it is likely that the effect of these areas was amplified by a poor vole year. In his early studies on avian reproductive ecology, David Lack hypothesised that birds are stimulated into breeding by the abundance of prey providing the energy requirements during the laying period (Lack, 1966). If this is assumed to be correct, there may have been a decrease in vole numbers between the beginning of the breeding season and first few weeks of hatching (the period in which most of the chick mortality was recorded). The decrease in vole abundance over this period would not necessarily have to have been substantial as in biomass, voles represent the lower end of prey importance (Graham et al., 1995). Indeed, voles may only prove an adequate primary food resource for buzzards at the higher end of their cyclical fluctuations (Lŏhmus, 2003). In an eleven year study on habitat and vole densities in relation to buzzard breeding success Lŏhmus (2003), observed that while buzzards in good vole habitat performed very badly in low vole years this was accounted for overall by a high breeding success in good vole years. The breeding success that recorded in low vole years was 0.76±0.54 SD, a result comparable to the 0.89±0.31 SD observed by this study.

6.2. Quantifying complex ecological relationships

Caution is often advised in attempting to draw inferences on the role of prey abundance in the breeding success of raptors. Observational studies that link prey abundance to raptor productivity often fail to take into account unforeseen factors such as prey accessibility (Preston, 1990; Widén 1994; Thirgood et al., 2002; Ontiverous et al., 2004; Salafsky, 2004; Whittingham & Evans, 2004) or alternative prey (Smout et al., 2010). However, by using a principle component analysis to reduce prey abundance variables, habitat variables and dietary variables over a large number of nests it is possible to build a clearer picture of the importance of individual species. In this study the explanatory analysis suggested three things: firstly, there was a positive correlation with vole abundance and their proportions in diet, secondly, that with higher abundances of voles the number of fledglings decreases, and thirdly, that a dominance of vole prey in diet may decrease the fledgling number. The PCA analysis allowed for this complex relationship to be 34 addressed within single analyses, clearly showing how variables combine to influence buzzard breeding ecology.

6.2.1. Correlates of breeding success One of the most noteworthy findings from the multivariate analysis was that vole reliant pairs of buzzards are less likely to raise multiple fledglings (Figure 8). Although this study recorded increases of vole abundance and utilisation over a spatial scale rather than a temporal one, it seems counterintuitive to observe a negative relationship between prey utilisation and predator productivity. This result suggests that during this year at least it was not the abundance but the type of prey that had the greatest influenced buzzard breeding success. These findings agree with those of Swann and Etheridge (1995) who observed that buzzards with access to persistent supplies of large prey items have consistently higher reproductive success than those whose diet is dominated by small . This is further confirmation of the key role that rabbits play in the breeding success of buzzards (Tubbs, 1974; Newton, 1979; Swann & Etheridge, 1995; Graham et al., 1995; Austin & Houston, 1997; Jardine, 2003; Sim, 2003).

6.2.2. Correlates of density While the exploratory analysis revealed no significant difference in the NND between habitat types, previous studies have associated buzzard density with both prey abundance (Newton, 1979) and habitat composition (Halley, 1993; Penteriani & Faivre, 1997). The modelled PC value suggested that decreases in buzzard density are associated with increases in the proportion of upland pasture within territories and the proportions of amphibians and birds in diet. These results are in keeping with previous studies that have recorded low densities in upland pasture (Newton et al., 1982; Dare & Barry, 1990) and with Spidso & Selàs (1988) who observed that birds are a secondary prey choice after mammals.

6.3. Cameras on nests:

Analysis of prey deliveries from the nest cameras provided a more extensive description of the diet of buzzards than the analysis of prey and pellet remains. As well as providing accurate data on both the range and proportions of the prey species in diet, it also allowed for addition information such as the provisioning rates, the age structure of prey and an estimation of prey biomass. Indeed, the observed provisioning rates for the nest that recorded the largest proportion of rabbits in prey were less than half of those recorded from the forestry nest (whose prey was mainly amphibians and voles) despite having two more chicks, clearly showing the advantages of prey with a larger biomass. Although provisional, the proportional data also suggests that a simple combination of prey and pellet methods would not generate the accuracy levels that Simmons et al., (1991) predicted. The use of nest cameras would be strongly recommended for future research of buzzard diet. 35

In comparison to the few previous studies of buzzard diet from direct observations the nest cameras proved highly successful at recording buzzard provisioning (Appendix 3). Tornberg and Reif (2007) for example, compiled 92 prey deliveries from 454 hours of footage with 25 % being unidentifiable, in contrast this study observed 263 prey deliveries in 500 hours with only 8.7% (n=23) being unidentifiable. With the exception of small prey items being possibly consumed instead of returned to the nest (Sonerud, 1992), the camera records created essentially true and unbiased data on the breeding diet of buzzards. Direct observations if collected from a larger sample over several nesting seasons would allow for the correction of false suppositions, for instance, Sim (2003) suggested that the majority of gamebirds observed in the diet of buzzards would have been scavenged, however, the five records of gamebirds (1 pheasant, 4 red grouse) from the cameras were all poults with no signs of rigor mortis.

As discussed previously, the high failure rate observed in this study suggests that it was a poor year for vole prey. In the studies of buzzard diet in northern Europe both Spidso & Selàs (1988) and Reif et al. (2001) recorded buzzard predation on birds to be significantly higher in poor vole years, however, the proportions of birds taken in this study are not higher than those observed by previous studies (Dare, 1961; Graham et al., 1995; Kenward et al., 2001; Sim, 2003; Jardine, 2003). Instead, a significant and largely overlooked source of prey identified from the nest cameras was and frogs (37.7% of all observations, n=96). Although Simmonds et al. (1991) stated that amphibians are overrepresented in the prey remains of generalist raptors, the results of this study agree with those of Selàs et al. (2007) who observed that amphibians were being underestimated in both pellets and prey remains. Indeed, studies of buzzard diet in upland areas rarely observe amphibians as significant (Graham et al., 1995; Swann & Etheridge, 1995; Sim et al., 2001), for example, from a two year study of 77 nests in Wales, Sim (2003) did not record one amphibian record from 253 prey remains.

The influence of small passerine density on buzzard breeding ecology was found to be negligible in this study. In retrospect, the probable reason for this is was the low importance of small passerines in spring and summer diet (3% direct observations, n=8). However, their contribution to buzzard survivorship cannot be ruled out as fledgling dependence on nesting territories can often extend over winter (Walls & Kenward, 1995) a period when voles are often inaccessible and amphibians are hibernating.

36

6.4 Study limitations and areas for future research

6.4.1. Limitations of the methodology

6.4.1.1 Sampling design The Vole Sign Index sampling and DISTANCE analysis provided coarse approximations of two prey groups within territories, although both were limited in their accuracy by the number of replicates. Indeed, over the 64km of transects surveyed over the study period there was not a single observation of a rabbit, highlighting the sampling limitations of randomised sampling on species than are not randomly dispersed by confined to a few high density sites (Thompson, 2004). If the methodology is to be repeated, a higher replicate size and the use of an automatic range finder have been shown to improve the accuracy of density estimates (Newson et al., 2008; Thomas et al., 2009; Ransom, 2011).

6.4.1.2. Accounting for winter weather In several studies of breeding raptors, winter weather has had a significant impact on breeding success (Steenhof et al., 1997; Gremel, 2001; Bloxton, 2002; Rodriguez & Bustamante, 2003). There is evidence that weather patterns not only affect the breeding success of raptors through direct effects such as hypothermia of chicks but also through causing declines in prey densities (Bloxton, 2002). While the winter of 2010/11 was recorded as the coldest in 31 years with several periods of prolonged snow cover (www.metoffice.co.uk), its effect on buzzards breeding ecology is difficult to surmise. Snow cover can restrict the accessibility of voles to buzzards during the winter while facilitating higher than average vole numbers during the buzzards breeding period (Selàs, 2001a). While temperature and snow cover in the winter have been shown to have no apparent effect on buzzard density, significant correlation has been observed between fledgling success and winter temperatures (Kostrzewa & Kostrzewa, 1991).

6.4.1.3. Accounting for stochastic events It is important to acknowledge the effect that weather over the hatching period has on raptor breeding success (Dawson & Bortolotti, 2000; Selàs, 2001b;). Most mortality in buzzard chicks occurs in the period directly after hatching (Sim, 2003) when the young birds are in down and most vulnerable to cold and wet weather (Tubbs, 1974). During May, the month when 90% of the chicks hatched, Glasgow (33km south) received 186mm of rain (almost double the monthly average) (www.metoffice.gov.uk). This period of bad weather cumulated with ‘Udo’ a severe storm on the 23rd of May that caused winds of up to 100mph across central Scotland and brought down whole blocks of conifer forest. Severe weather patterns such has this not only affect the breeding success of raptors through direct effects such as hypothermia of chicks but also through causing declines in prey densities (Bloxton, 2002). It has already been observed that precipitation during May and June may have a negative effect on buzzard broods (Selàs, 2001b)

37 and so it is fair to assume that the weather had a negative impact on the breeding success of the buzzards in this study. Further evidence for the impact of the storm on nesting raptors within the study area was observed within the resident population (monitored by RSPB) as 11 out of 22 nesting attempts failed to rear any young, resulting in 2011 having the lowest fledgling success since red kite recolonised the area in 1998. Of those 11 nests 7 are thought to have failed during the period of bad weather around the 23rd of May (D. Orr-Ewing, RSPB, unpublished data). The effect of climate on the breeding success of generalist raptors has been observed to be magnified in years when prey abundance is already low (Lŏhmus, 2003; Salafsky et al., 2005).

6.4.1.4. Accounting for intraguild exclusion and predation A possible constraint on buzzard density that was not accounted for within this study was that of intraguild predation (the killing of species that use similar resources) by other avian predators. It is possible that buzzards nest site selection was limited by the abundance of a potential predator, in this case golden , which are present in upland sections of the study area and have previously been observed to predate buzzard (Fielding et. al., 2003). The effect of intraguild predation by northern goshawks (Krüger, 2002) and owls (Sergio et al., 2005) has already been documented to reduce buzzard breeding densities.

6.4.2. Limitations of the analysis

6.4.2.1. The interpretation of PCA The use of principle component analysis in this study provided the opportunity to reduce the many habitat and dietary variables while retaining the majority of variance within them. One of its main advantages was that it allowed, through the creation of a composite variable, highly correlated data (such as the results of pellet and prey remains from the same prey group) to be reflected simultaneously. In this way the use of PCA provided a useful tool in exploring the many dimensions of the dataset. However, the interpretation of the principle components themselves (see Table 8) had the potential to be equivocal, and care has been stressed in their application (Crawley, 2007).

6.5. Areas for future research

6.5.1. Understanding the functional response type of buzzards to prey increases While this study failed to draw any conclusions on the type of response exhibited by buzzards to prey fluctuations, this is certainly an area for further study as understanding how generalists predators respond to fluctuations in prey density is crucial to understanding their impact (Holling, 1959). Although the functional response type has been documented for both peregrines (type II) and hen harries (type III) in relation gamebird densities (Redpath & Thirgood, 1999), recent literature on the subject has suggested that a ‘multispecies functional response’ approach is more appropriate for generalist predators (Smout et al., 2010). This is 38 necessary because generalist predators depend on the density of all of their major prey species and their response to a single species is presumably influenced by the abundance of alternative prey. The high diversity of prey species observed in this study suggests that in order to achieve an accurate understanding of the functional response of buzzard prey abundance several key species will have to be accounted for. Attempts at this are likely to be further confounded as the composition of these major prey species changes significantly over habitat types.

Another limitation of this study was its temporal scale. Longer term studies into the functional and numerical responses of raptors allow for clearer observations of trends and responses (Newton, 1979; Redpath & Thirgood, 1999). This appears to be particularly true for generalist predators; indeed, Lŏhmus (2003) showed that the relationship between buzzard productivity and landscape characteristics was distinct between years. Although limited to a single year snapshot, the methodology of this study was designed in such a way to provide a baseline for subsequent studies.

6.5. Educating the debate

The results of this study made two observations that are of importance to the debate concerning the impact of buzzards on gamebirds. Firstly, that buzzards have an increased breeding success in territories with access to larger prey. Secondly, that breeding density declines in relation to upland habitat and secondary prey items, in this case amphibians and birds. Although it has been suggested that buzzards in the northern sections of their range are vole specialists (Lŏhmus, 2003), the results of this study support the view that within the UK they are broad generalists, adapting their diet and density to prey availability (Tubbs, 1974). When these findings are applied to shooting estates which have; 1) voluntary habitat preservation and maintenance (Oldfield et al., 2003); 2) legal predator control providing high abundances of prey species (Tharme et al., 2001; Beja et al., 2009) and; 3) constant or annually high densities of gamebirds, it could potentially result in highly productive and relatively dense populations of buzzards. This is of concern for both conservationists and estate owners. For estate owners, the concern is an economic one as high densities of generalist avian predators have been observed to have an impact on the number of gamebirds available to harvest (Redpath & Thirgood, 1999; Thirgood et al., 2000a; 2000b). For conservationists the concern is one of illegal persecution in that, if this predator control were to include the illegal killing of raptors it could, in theory, create a localised sink in which high numbers of raptors are illegally killed. This has already been suggested to be restricting buzzard densities at local scales (Elliot & Avery, 1991; Swann & Etheridge, 1995; Gibbons et al., 1995; Holling, 2004).

39

It would appear that the next step in resolving the conflict would be address the issue directly by conducting research on buzzard diet within shooting estates. This would be necessary for both upland grouse moorland and lowland pheasant estates as the results of this study suggest buzzard densities, and therefore impact will be different for both. An accurate and unbiased assessment of diet will be essential for any such research and of the three possible methods, data collection through remote cameras is recommended.

40

41

7. References: Anon. (2009). Birdcrime: Offences Against Wild Bird Legislation 2009. RSPB, Sandy. www.rspb.org.uk/Images/birdcrime_tcm9-260567.pdf (Accessed on the 31st May 2011).

Arroyo, B. (2002). REGHAB Project (Reconciling Gamebird Hunting and Biodiversity): Conclusions from Workshop II. Centre for Ecology and Hydrology (CEH). http://hdl.handle.net/10261/8254 (Accessed on the 15st June 2011).

Arroyo, B., Viñuela, J. (2001). REGHAB Project (Reconciling Gamebird Hunting and Biodiversity): Conclusions from Workshop I. Centre for Ecology and Hydrology (CEH). http://hdl.handle.net/10261/8253 (Accessed on the 19th July 2011).

Austin, G. & Houston, D. (1997). The breeding performance of the Buzzard Buteo buteo in Argyll, Scotland and a comparison with other areas in Britain. Bird Study, 44, pp. 146-154.

Austin, G.E., Thomas, C.J., Houston D.C. & Thompson, D.B.A. (1996). Predicting the spatial distribution of buzzard Buteo buteo nesting areas using Geographical Information System and remote sensing. Journal of Applied Ecology, 33, pp. 1541-1550.

Bains, D., Moss, R. & Dugan, D. (2004). Capercaillie breeding success in relation to forest habitat and predator abundance. Journal of Applied Ecology 41, (1), pp.59-71.

Baker, P., Furlong, M., Southern, S. & Harris, S. (2006). The potential impact of red fox Vulpes vulpes predation on agricultural landscapes in lowland Britain. Wildlife Biology, 12, (1), pp.39-50.

Bangs, E.E., Fontaine, J.A., Jimenez, M.D., Meier, T.J., Bradley, E.H., Niemeyer, C.C., Smith, D.W., Mack, C.M., Asher, V. & Oakleaf. J.K. (2005). Managing wolf-human conflict in the north-western United States. Editors: Woodroffe, R., Thirgood S., Rabinowitz, E. People and Wildlife: Conflict or Coexistence?, Cambridge University Press, London, pp. 340–356.

Beier, P. & Drennan, J.E. (1997). Forest structure and prey abundance in foraging areas of Northern Goshawks. Ecological applications, 7, (2), pp. 564-571.

Benton, T., Vickery, J.A. & Wilson, J.D. (2003). Farmland biodiversity: is habitat heterogeneity the key? Trends in Ecology and Evolution, 18, (4), pp. 182-188.

Bijlsma,E. (1997). Buzzard. pp. 160-161 in (eds) Hagemeijer W.J.M., Blair M.J. The EBCC Atlas of European breeding birds, their distribution and their abundance. T. & A.D. Poyser, London.

42

Bjǿrnstad, O.N., Falck, W. & Stenseth, N.C. (1995). A geographic gradient in small mammal density fluctuations: a stastical modelling approach. Proceedings: Biological Sciences, 262, (1364), pp. 127-133.

Bloxton, T.D. (2002). Prey abundance, space use, demography, and foraging habitat of Northern Goshawks in Western Washington. Unpublished Masters Thesis, University of Washington. http://www.ruraltech.org/pubs/theses/bloxton/bloxton_ms_thesis.pdf (Accessed on the 21st August 2011).

Bolton, M., Butcher, N., Sharpe, F., Stevens, D. & Fisher, G. (2007). Remote monitoring of nests using digital camera technology. Journal of Field , 78, (2), pp. 213-220.

Bro, E., Arroyo, B. & Migot, P. (2006). Conflict between grey partridge Perdix perdix hunting and hen harrier Circus cyaneus protection in : a review. Wildlife Biology, 12, (3), pp.233-247

Brown, A.F. & Stillman, R.A. (1998). The return of the Merlin to the south Pennines. Bird Study, 45, pp.293-301.

Buckland, S .T., Anderson, D. R., Burnham. K. P., Laake, J. L., Borchers, D. L. & Thomas, L. (2001). Introduction to distance sampling: estimating abundance of biological populations. Oxford University Press, Oxford, UK

Buckland, S.T, Anderson,D.R., Burnham, K.P. & Laake, J.L. (1993). Distance sampling: estimating abundance of biological populations. Chapham & Hall, London, U.K.

Burnham, K. P. & Anderson, D. R. (2002). Model Selection and Multimodel Inference: a practical information-theoretic approach, 2nd edition. Springer-Verlag, New York

Butet, A., Michel, N., Rantier, Y., Comor, V., Hubert-Moy, L., Nabucet, J. & Delettre, Y. (2010) Responses of common buzzard (Buteo buteo) and Eurasian kestrel (Falco tinnunculus) to land use changes in agricultural landscapes of Western France. Agriculture, Ecosystems & Environment, 138, (3-4), pp. 152-159.

Clapham, A.R., Tutin, T.G. & Warburg, E.F. (1981). Excursion flora of the British Isles: 3rd Edition. Cambridge University Press, Cambridge

Clarke, R., Combridge, M. & Combridge, P. (1997). A comparison of the feeding ecology of wintering Hen Harriers Circus cyaneus centred on two heathland areas in England. Ibis, 139, pp. 4-18.

43

Clements, R. (2002). The common buzzard in Britain: a new population estimate. British Birds, 95, pp. 377-383.

Collopy, M. (1983). A comparison of direct observations and collections of prey remains in determining the diet of golden eagles. Journal of Wildlife Management, 47, pp. 360-368.

Crawley, M.J. (2007). The R Book. John Wiley & Sons Ltd, Chichester, England.

Dare, P.J. (1961). Ecological observations on a breeding population of common buzzard (Buteo buteo) with particular reference to diet and feeding habits, PhD thesis, unpublished, University of Exeter.

Dare, P.J. & Barry, J.T. (1990). Population size, density and regularity in nest spacing of buzzards Buteo buteo in two upland regions of North Wales. Bird Study, 37, (1), pp. 23-29.

Dare, P.J. (1995). Breeding success and territory features of Buzzards Buteo buteo in Snowdonia and adjacent uplands of north Wales. Welsh Birds, 1, pp. 69-78.

Dawson, R.D. & Bortolotti, G.R. (2000) Reproductive success of American Kestrels: The role of prey abundance and weather. The Condor, 102 (4) pp. 814-822.

Elliott, G.D. & Avery, M.I. (1991). A review of reports of Buzzard persecution 1975-1989. Bird Study, 38, pp. 52-56.

Evans, D.M., Redpath, S.M., Elston, D.A., Evans, A., Mitchell, R.J. & Dennis, P. (2006) To graze or not to graze? Sheep, voles, forestry and nature conservation in the British uplands. Journal of Applied Ecology, 43, (3) pp.499-505.

Ferguson-Lees, J. & Christie, D. A. (2001). Raptors of the world. Houghton Mifflin, Boston.

Fielding, A.H., Haworth, P.F., Morgan, D.H., Thompson, D.B.A. & Whitfield, D.P. (2003). The impact of Golden Eagles on a diverse assemblage. In Thompson, D.B.A., Redpath, S.M., Fielding, A.H., Marquiss, M. & Galbraith, C.A. (eds) Birds of Prey in a Changing Environment pp. 221–244. Edinburgh: The Stationery Office.

Gibbons, D. Gates, S., Green, E.R., Fuller, J.R. & Fuller, M.R. (1995). Buzzards Buteo buteo and Ravens Corvus corax in the uplands of Britain: limits to distribution and abundance. Ibis, 137, pp.75-84.

Gipps, J.H.W. & Alibhai, S.K. (1991). Field vole Microtus agrestis. Handbook of British Mammals, pp. 203–208. Blackwell Publishing, Oxford, UK.

44

Graham, I., Redpath, S. & Thirgood, S. (1995). The diet and breeding density of Common Buzzards Buteo buteo in relation to indices of prey abundance. Bird Study, 42, (2), pp. 165-173.

Graham, K., Beckerman, A.P. & Thirgood, S.(2004). Human-predator-prey conflicts: ecological correlates, prey losses and patterns of management. Biological Conservation, 122 ,(2), pp.159-171.

Green, S. (2011). Change the law on buzzards. Country Life (August Edition) p. 57.

Gremel, S. (2001). Spotted Owl monitoring in Olympic National Park: 2001 annual report. U.S. Dept. Int., National Park Service, Olympic National Park, Port Angeles, WA. http://www.reo.gov/monitoring/reports/nso/N110202_4_OLYNPS01.pdf (Accessed on the 22nd July 2011).

Halley, D.J. (1993). Population changes and territorial distribution of Common Buzzards Buteo buteo in the Central Highlands. Bird Study, 40, (1), pp. 24-30.

Hanski, I., Hansson, L. & Henttonen, H. (1991). Specialist predators, generalist predators, and the microtine cycle. Journal of Ecology, 60, pp. 353-367.

Hanski, I., Henttonen, H., Korpimäki, E., Oksanen, L. & Turchin, P. (2001). Small-rodent dynamics and predation. Ecology, 82, (6), pp. 1505-1520.

Harradine, J., Reynolds, N. & Laws, T. (1997). Raptors and Gamebirds: A Survey of Game Managers Affected by Raptors. British Association for Shooting and Conservation, Rossett, UK.

Henderson,I., Parrott, D. & Moore, N. (2004). Racing pigeons- impact of raptor predation. A report to Scottish Natural Heritage, Central Science Laboratory,

Hodder, K.H., Kenward, R.E., Walls, S.S. & Clarke, R.T. (1998). Estimating core ranges: a comparison of techniques using the common buzzard (Buteo buteo). Journal of Raptor Research, 32, (2), pp. 82-89.

Holling, C.S. (1959). Some characteristics of simple types of predation and . Canadian Entomologist, 91, pp. 385-398.

Holling, M. (2003). Buzzards (Buteo buteo) on the increase in Lothian and Borders, Scotland. In Thompson, D.B.A., Redpath, S.M., Fielding, A.H., Marquiss, M. & Galbraith, C.A. (eds) Birds of Prey in a Changing Environment pp. 173-178. Edinburgh: The Stationery Office.

45

Hudson, P.J. (1992). Grouse in Space and Time- the Population Biology of a Managed Gamebird. The Game Conservancy Trust, Fordingbrige, Hampshire

Hume, R. (2007). Complete birds of Britain and Europe (revised and updated). Dorling Kingersley Limited, London.

IUCN (2010). Buteo buteo. IUCN Red List of Threatened Species. Version 2010.4. . Downloaded on 05 June 2011.

Jardine, D.C. (2003). Buzzards (Buteo buteo) on Colonsay 1990-2000: numbers and breeding performance. pp.179-182. In Thompson, D.B.A., Redpath, S.M., Fielding, A.H., Marquiss, M. & Galbraith, C.A. (eds). Birds of Prey in a Changing Environment Edinburgh: The Stationery Office.

Johnson, J.B. & Omaland, K.S. (2004) Model selection in ecology and evolution. Trends in Ecology and Evolution, 19, (2), pp.101-108.

Kenward, R.E. (1982). Goshawk hunting behaviour, and range size as a function of food and habitat availability. Journal of Applied Ecology, 51, pp. 69-80.

Kenward, R., Hall, D., Walls, S. & Hodder, K. (2001). Factors affecting predation by buzzards Buteo buteo on released pheasants Phasianus colchicus. Journal of Applied Ecology, 38, (4), pp. 813-822.

Korpimäki E. & Norrdahl, K. (1991). Numerical and functional responses of kestrels, short- eared owls, and long-eared owls to vole densities, Ecology 72 (3) pp. 814-826.

Kostrzewa, R. & Kostrzewa, A. (1991).Winter weather, spring and summer density, and subsequent breeding success of Eurasian Kestrel, Common Buzzard and Northern Goshawks. The Auk, 108, pp. 342-347.

Kostrzewa, A. (1996). A comparative study of nest-site occupancy and breeding performance as indicators for nesting-habitat quality in three European raptor species. Ethology, Ecology & Evolution, 8, pp. 1-18.

Kruckenhauser, L., Haring, E., Pinsker, W., Riesing, M. J., Winkler, H., Wink, M. & Gamauf, A. (2004). Genetic vs. morphological differentiation of Old World buzzards (genus Buteo, ). Zoologica Scripta, 33, pp. 197–211.

Krüger, O. & Linderström, J. (2001). Habitat heterogeneity affects population growth rate in goshawk gentilis. Journal of Animal Ecology, 70, (2), pp.173-181.

46

Krüger, O. (2002). Interactions between common buzzard Buteo buteo and goshawk Accipiter gentilis: trade-offs revealed by a field experiment. Oikos, 96, pp. 441–452.

Lack, D. (1954). The natural regulation of animal numbers. Oxford University Press, Oxford.

Lack, D. (1966). Population studies of birds. Oxford University Press, Oxford.

Lambin, X., Petty, S. & Mackinnon, J. (2000). Cyclic dynamics in field vole populations and generalist predation. Journal of Animal Ecology, 69, pp. 106-118.

Lewis, S.B., Fuller, M.R. & Titus, K. (2004). A comparison of 3 methods for assessing raptor diet during the breeding season. Wildlife Society Bulletin, 32, pp.373-385.

Lŏhmus, A. (2003). Are certain habitats better every year? A review and a case study on birds of prey. Ecography, 26, (5), pp. 545-552.

Lovegrove, R. (2007). Silent fields: the long decline of a nation’s wildlife. Oxford University Press, Oxford.

Michalski, F., Boulhosa, L.P., Faria, A. & Peres, C.A. (2005). Human-wildlife conflicts in a fragmented Amazonian forest landscape: determinants of large felid depredation on livestock. Animal Conservation, 9, pp.179-188.

Mineau, P., Fletcher, M.R., Glaser, L.C., Thomas, N.J., Brassard, C., Wilson, L.K., Elliott, J.E., Lyon, L.A., Henny, C.J., Bollinger, T. & Porter, S.L. (1999). Poisoning of raptors with organophosphorus and carbamate pesticides with emphasis in Canada US and UK. Journal of Raptor Research, 33, pp. 1-37.

Moore, N.W. (1957). The past and present status of the buzzard in the British Isles. British Birds, 69, pp. 193-201.

Newson, E.N., Evans, K.L., Noble, D.G., Greenwood, J.J.D. & Gaston, K.J. (2008) Use of distance sampling to improve estimates of national population sizes for common and widespread breeding birds in the UK. Journal of Applied Ecology, 45, pp. 1330-1338.

Newton, I. (1979). The population ecology of raptors. T. & A.D. Poyser, Berkhamsted, UK.

Newton, I., Davis, P.E. & Davis, J.E. (1982). Ravens and buzzards in relation to sheep- farming and forestry in Wales. Journal of Applied Ecology, 19, pp.681-706.

O’Toole, L., Fielding, A.H. & Haworth, P.F. (2002). Re-introduction of the into the Republic of . Biological Conservation, 103, (3), pp. 303-312.

47

Oldfield, T.E.E., Smith, R.J., Harrop, S.R. & Leader-Williams, N. (2003) Field sports and conservation in the United Kingdom. Nature, 423, pp. 531-533.

Ontiveros, D., Pleguezuelos, J. & Caro, J. (2004). Prey density, prey detectability and food habits: the case of Bonelli’s eagle and the conservation measures. Biological Conservation, 123, (1), pp. 19-25.

Park, K.J., Graham, K.E., Calladine, J. & Wernham, C.W. (2008). Impacts of birds of prey on gamebirds within the UK: a review. Ibis, 150, (1), pp. 9-26.

Penteriani, V. & Faivre, B. (1997). Breeding density and landscape-level habitat selection of common buzzards (Buteo buteo) in a mountain area (Abruzzo Apennines, Italy). Journal of Raptor Research, 31, (3), pp. 208-212.

Petrovan, S.O., Barrio, I.C., Ward A.I. & Wheeler, P.M. (2011). Farming for pests? Local and landscape-scale effects of grassland management on rabbit densities. European Journal of Wildlife Research, 57, (1), pp. 27-34.

Petty, S. J. (1992). Ecology of the Strix aluco in the spruce forests of Northumberland and Argyll. PhD thesis, Open University.

Petty, S.J. (1999). Diet of tawny owls (Strix aluco) in relation to field vole (Microtus agrestis) abundance in a conifer forest in northern England. Journal of Zoology, 248, pp. 451–465.

Petty, S.J. &Thomas, C.J. (2003). Distribution, numbers and breeding performance of tawny owls (Strix aluco). in relation to clear-cutting in a conifer forest in northern England. Birds of Prey in a Changing Environment (eds) Thompson, D.B.A., Redpath S.M., Fielding A.H., Marquiss M., Galbraith, C.A.). pp. 111-129. The Stationary Office, Edinburgh, UK.

Picozzi, N. & Wier, D. (1974). Breeding biology of the Buzzard in Speyside. British Birds, 67, pp.199-210.

Pietersen, D.W. & Symes, C.T. (2010). Assessing the diet of the Amur Falco amurensis and Falco naumanni using stomach contents analysis. Ostrich, Journal of African Ornithology, 81, (1), pp.3 9-44.

Preston, C. (1990). Distribution of raptor foraging in relation to prey biomass and habitat structure. The Condor, 92, pp. 107-112.

48

Public & Corporate Economic Consultants. (2006). The Economic and Environmental Impact of Sporting Shooting. Commissioned report for British Association for Shooting & Conservation, the Countryside Alliance and the Country Land and Business Association, in association with Game Conservancy Trust. Cambridge, UK. http://www.shootingfacts.co.uk/ (Accessed on the 25th August 2011).

Ransom, J.I. (2011) Customizing a rangefinder for community-based wildlife conservation initiatives. Biodiversity Conservation, 20, pp.1603-1609

Redpath, S. & Thirgood, S. (1999). Numerical and functional responses in generalist predators: hen harriers and peregrines on Scottish grouse moors. Journal of Animal Ecology, 68, pp. 879-892.

Redpath, S., Clarke, R., Madders, M. & Thirgood, S. (2001). Assessing raptor diet: comparing pellets, prey remains, and observational data at hen harrier nests. The Condor, 103, pp. 184-188.

Redpath, S.M., Thirgood, S.J. & Clarke, R. (2002). Field vole Microtus agrestis abundance and hen harrier Circus cyaneus diet and breeding in Scotland. Ibis, 144, pp. E33–E38.

Reif, V., Tornberg, R., Jungell S. & Korpimaki, E. (2001). Diet variation of common buzzards in supports the alternative prey hypothesis. Ecography, 24, pp. 267-274.

Reif, V, Jungell, S., Korpimaki, E., Tornberg, R., Mykra, S. (2004) Numerical response of common buzzards and the predation rate of amin and alternative prey under fluctuating food conditions. Annales Zoologici Fennici, 41 pp.599-607.

Reynolds, R.T., Graham, R.T., Reiser, M.H., Bassett, P.L., Kennedy, D.A., Boyce,D.A., Goodwin, G., Smith, R. & Fisher, E.l. (1992). Management recommendations for the Northern Goshawks in the southwestern United States. United States Forest Service, General Technical Report, RM-217.

Reynolds, R.T., Wiens, D.J., Joy, S.M. & Salafsky, S.R. (2005). Sampling considerations for the demographic and habitat studies of Northern Goshawks. Journal of Raptor Research, 39, (3), pp.274-285.

Risely, K., Baillie, S.R., Eaton, M.A., Joys, A.C., Musgrove, A.J., Noble, D.G., Renwick, A.R. & Wright, L.J. (2010). The Breeding Birds Survey 2009. BTO Research Report 559. British Trust for Ornithology, Thetford.

49

Rodenhouse, N.L., Sherry, T.W. & Holmes, R.T. (1997) Site- dependant regulation of population size: a new synthesis. Ecology, 78, (7), pp. 2025-2042.

Roderiguez, C. & Bustamante, J. (2003). The effect of weather on Lesser Kestrel breeding success: can climate change explain population declines? Journal of Animal Ecology, 72, (5), pp.793-810.

Rodgers, A.S., DeStefano, S. & Ingraldi, M.F. (2005). Quantifying diets using remote cameras and observations for blinds. Journal of Raptor Research, 39, (3), pp. 303- 309.

Rodriguez, B., Siverio, F., Rodríguez, A., Siverio, M. Hernadez, J.J. & Figuerola, J. (2010). Density, habitat selection and breeding biology of Common Buzzards Buteo buteo in an insular environment. Bird Study, 57, pp. 75-83.

Sagør, J.T., Swenson, J.E. & Røskaft, E. (1997). Compatibility of brown bear Ursus arctos and free-ranging sheep in Norway. Biological Conservation, 81, (1), pp.91-95.

Sakamoto, Y., Ishiguro, M. & Kitagawa, G. (1986). Akaike information criterion statistics. KTK Scientific Publishers, Tokyo.

Salafsky, S.R. (2004). Covariation between prey abundance and Northern Goshawk fecundity on the Kaibab Plateau, Arizona. MSc Thesis, Colorado State University, Fort Collins, Colorado.

Salafsky, S.R., Reynolds, R.T. & Noon, B.R. (2005). Patterns of temporal variation in Goshawk reproduction and prey resources. Journal of Raptor Research, 39 ,(3), pp.237-246.

Salafsky, S.R., Reynolds, R.T., Noon, B.R. & Wiens, J.A. (2007) Reproductive responses of Northern Goshawks to variable prey populations. The Journal of Wildlife Management, 71, (&) pp.2274-2283.

Salamolard, M., Butet, A., Leroux, A. & Bretagnolle. V. (2000) Resonces of an avian predator to variation in prey density at a temperate latitude. Ecology, 81, pp. 2428-2441.

Sánchez, R., Margalida, A., González, M.L. & Oria, J. (2008). Biases in diet sampling methods in the Spanish Imperial Eagle Aquila adalberti. Ornis Fennica, 85, pp. 82-89.

Seaton, R., Hyde, N., Holland, J.D., Minot, E.O. & Springett, B.P. (2008). Breeding season diet and prey selection of the Falcon (Falco novaeseelandiae) in a plantation forest. Journal of Raptor Research, 42, (4), pp.256-264.

50

Selàs, V. (2001a). Predation on reptiles and birds by the common buzzard, Buteo buteo, in relation to changes in its main prey, voles. Canadian Journal of Zoology, 79, (11), pp.2086-2093.

Selàs, V. (2001b). Breeding density and brood size of the common buzzard Buteo buteo in relation to snow cover in spring. Ardea, 89, (3), pp.471-479.

Selàs, V., Tveiten, R. & Aanonsen, O. M. (2007). Diet of Common Buzzards (Buteo buteo). in southern Norway determined from prey remains and video recordings. Ornis Fennica, 84, pp. 97-104.

Sergio, F., Scandolara, C., Marchesi, L., Pedrini, P. & Penteriani, V. (2005). Effect of agro- forestry and landscape changes on common buzzards (Buteo buteo) in the : implications for conservation. Animal Conservation, 8, (1), pp. 17-25.

Sim, I.M.W., Campbell. L., Pain, D. & Wilson, J.D. (2000). Correlates of the population increase of Common Buzzards Buteo buteo in the West Midlands between 1983 and 1996. Bird Study, 47, pp. 154-164.

Sim, I.M.W., Cross, A.V., Lamacraft, D.L. & Pain, D.J. (2001) Correlates of Common Buzzard Buteo buteo density and breeding success in the West Midlands. Bird Study, 48, pp. 317-329.

Sim, I.M.W. (2003). Land use, common buzzards (buteo buteo) and rabbit (Oryctolagus cuniculus) in the Welsh Marches. pp. 351-371 In Thompson, D.B.A., Redpath, S.M., Fielding, A.H., Marquiss, M. & Galbraith, C.A. (eds). Birds of Prey in a Changing Environment. Edinburgh: The Stationery Office.

Simmons, R.E., Avery, D.M. & Avery, G. (1991). Biases in diets determined from pellets and remains: correction factors for a mammal and bird-eating raptor. Journal Raptor Research, 25, (3), pp, 63-67.

Smith, M., C. (2007). The return of the common buzzard to Warwickshire and it’s possible use as an indicator for the return of the and the red kite. Unpublished thesis, University of Manchester, Manchester.

Smout, S., Asseburg, C., Matthiopoulos, J., Fernández, C., Redpath, S., Thirgood, S. & Harwood, J. (2010). The Functional Response of a Generalist Predator. PLoS ONE 5, (5): e10761. doi:10.1371/journal.pone.00761

Snow, D.W. & Perrins C.M. (1998). The Birds of the Western Palearctic Concise Edition. Oxford University Press, Oxford, UK. 51

Sonerud, G.A. (1992). Functional responses of birds of prey: Biases due to the load-size effect in central place foragers. Oikos, 63, (2), pp. 223-232.

Sonerud, G.A. (1997). Hawk owls in Fennoscandia: population fluctuations, effects of modern forestry, and recommendations on improving foraging habitats. Journal of Raptor Research, 31, (2), pp. 167-174.

Spidso, T.K. & Selàs, V. (1988) Prey selection and breeding success in the common buzzard Buteo buteo in relation to small rodent cycles in southern Norway. Fauna Novegica, 11, (2), pp.61-66.

Steenhof, K., Kochert, M.N. & McDonald, T.L. (1997). Interactive effects of prey and weather on golden eagle reproduction. Journal of Animal Ecology, 66, pp.350-362.

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, (3), pp. 279-292.

Swann, R.L. & Etheridge, B. (1995). A comparison of breeding success and prey of the Common Buzzard in two areas of northern Scotland. Bird Study, 42, (2), pp. 37-43.

Taylor, K., Hudson R. & Horne G. (1988). Buzzard breeding distribution and abundance in Britain and Northern Ireland in 1983. Bird Study, 35, pp. 109-118.

Taylor, M.J., Sharp, E.A. & Giela, A. (2009). A Report of Investigations in Scotland: Pesticide Poisoning of in 2008. Science and Advice for Scottish Agriculture (SASA)., Chemistry Branch, Roddinglaw Road, Edinburgh, Scotland.

Thirgood, S. J., Redpath, S.M. & Graham, I.M. (2003). What determines the foraging distribution of raptors on heather moorland? Oikos, 100, pp.15-24.

Thirgood, S.J., Redpath, S.M., Campbell, S. & Smith, A. (2002). Do habitat characteristics influence predation on red grouse? Journal of Applied Ecology, 39, pp.217-225.

Thirgood, S.J., Redpath, S.M., Haydon, D.T., Rothery, P., Newton, I. & Hudson, P.J. (2000a) Habitat loss and raptor predation: disentangling long and short term causes of red grouse decline. Proceedings of the Royal Society London B, 267, pp.651-656

Thirgood, S.J., Redpath,S.M., Rothery, P. & Aebischer, N.J. (2000b). Raptor predation and population limitation in red grouse. Journal of Applied Ecology, 69, (3), pp.504-516.

Thomas, L., Laake, J.L., Rexstad, E., Strindberg, S., Marques, F.F.C., Buckland, S.T., Borchers, D.L., Anderson, D.R., Burnham, K.P., Burt, M.L., Hedley, S.L., Pollard, J.H.,

52

Bishop, J.R.B. & Marques, T.A. (2009). Distance 6.0. Research Unit for Wildlife Population Assessment, University of St. Andrews, UK.

Thompson, W.L. (2004). Sampling rare or elusive species; concepts, designs and techniques for estimating population parameters. (Ed) Island Press, London.

Tornberg, R. & Reif, V. (2007). Assessing the diet of birds of prey: a comparison of prey items found in nests and images. Ornis Fennica, 84, pp. 21-31.

Trout, R., Langton, S., Smith, G. & Haines-Young, R. (2000). Factors affecting the abundance of rabbits (Oryctolagus cuniculus) in England and Wales. Journal of the Zoological Society of London, 252, pp. 227-238.

Tubbs, C. (1972). Analysis of nest record cards for the Buzzard. Bird Study, 19, (2), pp. 97- 104.

Tubbs, C. (1974). The buzzard. David & Charles, Newton Abbot.

Tubbs, C.R., Tubbs, J.M. (1985). Buzzards and land use in the , Hampshire, England. Biological Conservation, 31, pp. 41-65.

UK Raptor Working Group, (2000). UK Raptor Working Group, Report of the UK Raptor Working Group. DETR/JNCC, Bristol, UK.

Valkama, J., Korpimaki, E., Arroyo, B., Beja, P., Bretagnolle, V., Bro, E., Kenward, R.E., Maǹosa, S., Redpath, S.M., Thirgood, S.J. & Viǹuela, J. (2005). Birds of prey as limiting factors of gamebird populations in Europe: a review. Biological Review, 80, pp. 171-203.

Vergara, P. (2010). Time-of-day bias in diurnal raptor abundance and richness estimated by road surveys. Revista Catalana d’Ornitologia. 26, pp. 22-30.

Walls, S.S. & Kenward, R.E. (1995). Movements of radio-tagged Common Buzzards Buteo buteo in their first year. Ibis, 137, pp.177-182

Walls, S.S. & Kenward, R.E. (1998). Movements of radio-tagged Buzzard Buteo buteo in early life. Ibis, 140, pp. 561-568.

Watson, M., Aebischer, N.J., Potts, G.R. & Ewald, J.A. (2007). The relative effects of raptor predation and shooting on overwinter mortality of grey partridges in the United Kingdom. Journal of Applied Ecology, 44, pp. 972-982

53

Whitfield, D.P., Fieldling, A.H., McLeod, D.R.A. & Haworth, P.F. (2003).The effects of persecution on age of breeding and territory occupation in golden eagles in Scotland. Biological Conservation 118, (2), pp.249-259

Whittingham, M.J. & Evans, K.L. (2004). The effects of habitat structure on predation risk of birds in agricultural landscapes. Ibis, 146, (2), pp.210-220

Widén, P. (1994). Habitat quality for raptors: a field experiment. Journal of Avian Biology, 25, pp. 219-223.

Woodroffe, R., Thirgood, S. & Rabinowitz, A. (2005). People and Wildlife: Conflict or Coexistence? Cambridge University Press.

Wolfe, A. & Long, A. (1997). Distinguishing between hair fibres of the rabbit and the mountain in scats of the Red Fox. Journal of Zoology (ZSL), 242, pp. 370-375.

54

8. Appendix Appendix 1.

Table 1: Liner regressions between the vole sign indices (VSI) and live trapping estimation of density. Values in brackets are standard errors of parameter estimates. Taken from Lambin et al., 2000.

Season Regression n R² F P Vole density Spring (March-May) Nha¯¹=13.10(1.76)*VSI+9.96(11.6) 36 0.62 55.12 <0.001 Summer (June to August) Nha¯¹=14.11(1.33)*VSI+14.60(9.34) 48 0.71 111.76 <0.001

55

Appendix 2. Full prey species list from direct observations, prey remains and pellet analysis methods

Direct Obs. Prey Pellet Total % Total % Total % Amphibians 94 35.7 7 4.1 10 8.5 Bufo bufo 57 21.7 6 3.5 - - Common Rana temporaria 37 14.1 1 0.6 - - Reptiles 5 1.9 0 0.0 1 0.8 Slow- fragilis 1 0.4 0 0 - - Adder berus 4 1.5 0 0 - - Birds 31 11.8 60 35.3 18 15.3 Teal Anas crecca 1 0.4 0 0 - - Red Grouse L. lagopus scoticus 4 1.5 1 0.6 - - Black grouse Lyrurus tetrix 0 0 1 0.6 - - Pheasant Phasianus colchicus 1 0.4 7 4.1 - - Unidentified wader 1 0.4 0 0 - - Woodpigeon Columba palumbus 0 0 4 2.4 - - Tawny owl Strix aluco 0 0 3 1.8 - - Great Spotted Dendrocopos major 0 0 1 0.6 - - Chaffinch Fringilla coelebs 0 0 5 2.9 - - Willow warbler Phylloscopus trochilus 0 0 1 0.6 - - Medow pipet Anthus pratensis 4 1.5 3 1.8 - - Prunella modularis 0 0 1 0.6 - - Great Tit Parus major 0 0 1 0.6 - - Coal Tit Parus ater 0 0 1 0.6 - - Robin Erithacus rubecula 0 0 1 0.6 - - Unidentified Small Passerine 8 3.0 3 1.8 - - Blackbird Turdus merula 3 1.1 1 0.6 - - Turdus viscivorus 4 1.5 2 1.2 - - Turdus philomelos 0 0 1 0.6 - - Jay Garrulus glandarius 0 0 1 0.6 - - Magpie Pica Pica 0 0 2 1.2 - - Jackdaw Corvus monedula 0 0 8 4.7 - - Crow Corvus corone 5 1.9 10 5.9 - - Rook Corvus frugilegus 0 0 2 1.2 - - Mammals 110 41.8 98 57.6 89 75.4 Hedgehog Erinaceau europaeus 0 0 1 0.6 1 0.8 Mustela nivalis 4 1.5 0 0 1 0.8 Stoat Mustela erminea 2 0.8 0 0 0 0 mole Talpa minor 18 6.8 5 2.9 11 9.3 Rabbit Oryctolagus cuniculus 15 5.7 39 22.9 - - Brown Hare Lepus europaeus 0 0 1 0.6 - - Unidentified lagomorph sp. 0 0 - - 26 22.0 Water Vole Arvicola terrestris 1 0.4 - - - - Unidentified Cricetidae species 54 20.5 50 29.4 50 42.4 Unidentified Apodemus species 8 3.0 1 0.6 0 0 Unidentified Soricidae species 8 3.0 1 0.6 0 0 Carrion 0 0 5 2.9 0 0 Sheep Ovis aries 0 0 2 1.2 0 0 Roe Capreolus capreolus 0 0 2 1.2 0 0 Red Fox Vulpes vulpes 0 0 1 0.6 0 0 Unidentified prey 23 8.7 0 0 0 0 56

Total observations/items 263 170 118

Appendix 3: Still images from one of the nest cameras (Upland pasture C2). Wood mouse Apodemus sylvaticus (top left), Unidentified duck sp. (top right), Mole Talpa minor (middle left), Weasel Mustela nivalis (middle right), Rana temporaria (bottom left), Common Toad Bufo bufo (bottom right). 57

Appendix 4:

Non-native mature forestry: Detection probability of small passerine birds as function of the distance to the observer (from line transect data in non-native forestry habitat, n=93). Columns - number of detections in different distance classes, Line - Fit detection function (Software Distance 6.0). (Chi²=2.98, df=2,P=0.23)

Rough grassland: Detection probability of small passerine birds as function of the distance to the observer (from line transect data in rough grassland habitat, n=134). Columns - number of detections in different distance classes, line - Fit detection function (Software Distance 6.0). (Chi²=0.067, df=2,P=0.97) 58

Upland pasture: Detection probability of small passerine birds as function of the distance to the observer (from line transect data in upland pasture habitat, n=114). Columns - number of detections in different distance classes, line - Fit detection function (Software Distance 6.0). (Chi²=0.72, df=4,P=0.95)

Pre-thicket forestry: Detection probability of small passerine birds as function of the distance to the observer (from line transect data in pre-thicket forestry habitat, n=224). Columns - number of detections in different distance classes, line - Fit detection function (Software Distance 6.0). (Chi²=8.47, df=2, P=0.014)

59

Heather moorland: Detection probability of small passerine birds as function of the distance to the observer (from line transect data in heather moorland habitat, n=121). Columns - number of detections in different distance classes, line - Fit detection function (Software Distance 6.0). (Chi²=3.21, df=3,P=0.359)

Clear cut: Detection probability of small passerine birds as function of the distance to the observer (from line transect data in clear cut habitat, n=67). Columns - number of detections in different distance classes, line - Fit detection function (Software Distance 6.0). (Chi²=1.39, df=3,P=0.706)

60

Lowland pasture: Detection probability of small passerine birds as function of the distance to the observer (from line transect data in lowland pasture habitat, n=87). Columns - number of detections in different distance classes, line - Fit detection function (Software Distance 6.0).Chi squared goodness of fit test: Chi²=2.44, df=3,P=0.48)

Native woodland: Detection probability of small passerine birds as function of the distance to the observer (from line transect data in native woodland habitat, n=279). Columns - number of detections in different distance classes, line - Fit detection function (Software Distance 6.0). (Chi²=3.95, df=2, P=0.14)

61