general enquiries on this form should be made...

General enquiries on this form should be made to:Defra, Procurements and Contracts Division (Science R&D Team)Telephone No. 0207 238 5734E-mail: [email protected]

SID 5 Research Project Final Report

SID 5 (Rev. 07/10) of 49

NoteIn line with the Freedom of Information Act 2000, Defra aims to place the results of its completed research projects in the public domain wherever possible. The SID 5 (Research Project Final Report) is designed to capture the information on the results and outputs of Defra-funded research in a format that is easily publishable through the Defra website. A SID 5 must be completed for all projects.

This form is in Word format and the boxes may be expanded or reduced, as appropriate.

ACCESS TO INFORMATIONThe information collected on this form will be stored electronically and may be sent to any part of Defra, or to individual researchers or organisations outside Defra for the purposes of reviewing the project. Defra may also disclose the information to any outside organisation acting as an agent authorised by Defra to process final research reports on its behalf. Defra intends to publish this form on its website, unless there are strong reasons not to, which fully comply with exemptions under the Environmental Information Regulations or the Freedom of Information Act 2000.Defra may be required to release information, including personal data and commercial information, on request under the Environmental Information Regulations or the Freedom of Information Act 2000. However, Defra will not permit any unwarranted breach of confidentiality or act in contravention of its obligations under the Data Protection Act 1998. Defra or its appointed agents may use the name, address or other details on your form to contact you in connection with occasional customer research aimed at improving the processes through which Defra works with its contractors.

Project identification

1. Defra Project code WM0318

2. Project title

Developing approaches to evaluate and mitigate the environmental impact of wild boar

3. Contractororganisation(s)

Food and Environmental Research AgencySand HuttonYO41 1LZYork

54. Total Defra project costs £ 280,202(agreed fixed price)

5. Project: start date................ 01/11/2008

end date................. 31/03/2011

SID 5 (Rev. 07/10) of 49

6. It is Defra’s intention to publish this form. Please confirm your agreement to do so............................................................................................YES x (a) When preparing SID 5s contractors should bear in mind that Defra intends that they be made public. They

should be written in a clear and concise manner and represent a full account of the research project which someone not closely associated with the project can follow.Defra recognises that in a small minority of cases there may be information, such as intellectual property or commercially confidential data, used in or generated by the research project, which should not be disclosed. In these cases, such information should be detailed in a separate annex (not to be published) so that the SID 5 can be placed in the public domain. Where it is impossible to complete the Final Report without including references to any sensitive or confidential data, the information should be included and section (b) completed. NB: only in exceptional circumstances will Defra expect contractors to give a "No" answer.In all cases, reasons for withholding information must be fully in line with exemptions under the Environmental Information Regulations or the Freedom of Information Act 2000.

(b) If you have answered NO, please explain why the Final report should not be released into public domain

Executive Summary7. The executive summary must not exceed 2 sides in total of A4 and should be understandable to the

intelligent non-scientist. It should cover the main objectives, methods and findings of the research, together with any other significant events and options for new work

1. Wild boar in their native range are associated with increased plant biodiversity, act as key prey species for large carnivores and as plant seed dispersers. However, overabundant populations of wild boar may have significant environmental and economic impact. Estimating local densities and assessing potential impact is thus crucial to managing wild boar populations.

2. Isolated wild boar populations occur in England as the result of farm escapes. Although the apparent lack of range expansion of these populations might be due to hunting pressure, wild boar might be in the initial phase of increase as observed elsewhere in Europe. In the latter case, Defra has the opportunity of adopting a proactive approach to limit the expansion and impact of wild boar populations whilst numbers are still relatively low.

3. This project aimed at developing surveillance methods to monitor density and abundance of wild boar and to detect range expansion, evaluating the feasibility and cost of electric fencing to prevent wild boar incursions, identifying methods to evaluate the impact of wild boar on the biodiversity of woodland habitats and applying these to limited woodland habitats.

4. The study was carried out in five sites, three in Gloucestershire and two in East Sussex. Three surveillance methods were developed to estimate relative local abundance: (i) activity signs, (ii) camera trap surveys and (iii) distance sampling using thermal imaging. Both distance sampling and activity signs showed that the number of wild boar observations and the relative activity signs such as trails and rooting were too few and variable to obtain precise local estimates. However, activity indexes obtained from camera grids indicated differences between sites. Density estimates using the thermal imaging were similar to those obtained from camera trap surveys, thus suggesting that the latter, less expensive method could be used to monitor population densities of wild boar in the UK.

5. Two methods were evaluated to monitor the range expansion of wild boar: (i) mapping sightings and road traffic accidents and (ii) confirmation of wild boar presence at bait points. The latter method was validated in 21 woodland sites around the Forest of Dean. The results indicated that wild boar expansion is still limited. Based on this method, a staged-approach

SID 5 (Rev. 07/10) of 49

decision tree was produced to monitor the potential expansion of wild boar.

6. The review of fencing to mitigate wild boar impacts suggested that although fencing and particularly electric fencing can successfully protect crops and localised areas from wild boar incursions, expensive maintenance is critical to its success.

7. Large-scale impact was assessed by quantifying rooting activity on permanent plots. The results indicated low frequency of rooting at all sites with a peak in spring and an estimated 0.7-4.4 % of woodland affected by rooting in any season. Rooting occurred relatively more than expected, based upon habitat availability, in broadleaf woodland than in conifer plantations.

8. Overall observations indicated that, at the landscape scale, wild boar rooting occurred more frequently in broadleaves woods, particularly oak and sweet chestnut of a mature structure and producing large fruits than in woodlands dominated by conifer plantations.

9. Rooting was associated with the presence of bluebells. However, additional, long term studies are required to establish whether repeated rooting affects the number and distribution of bluebells.

10. There was a positive association between total numbers of ground beetles and % rooting at the site level, probably reflecting a tendency for rooting to occur more frequently in woods where large numbers of ground beetles are already present. In contrast, the species richness and diversity of the ground beetle community showed no relationship with the amount of rooting. Other beetles and invertebrates generally showed no association with rooting; the only exception was woodlice, which tended to occur in greater numbers at sites where the amount of rooting was low.

11. The results on abundance estimates and impact suggest that local densities of wild boar are still relatively low compared to those in continental Europe. In four out of the five sites surveyed in the study the trend in relative densities of wild boar increased, while in one site the local density decreased. Anecdotal evidence indicates that locally wild boar are likely to be affected by hunting pressure.

12.The project realised the first tools to estimate local densities and impact of wild boar and provided a baseline for future studies on wild boar in the UK. Refining these tools to improve their accuracy will provide Defra with the means to facilitate the regional management of wild boar.

Project Report to Defra

Introduction and Policy Rationale

Defra has the responsibility to facilitate the regional management of wild boar (Sus scrofa) by providing local communities with advice and guidance on methods to alleviate human-wild boar conflicts (Defra 2008). In continental Europe overabundant populations of wild boar cause damage to crops and vehicle collisions, reduce abundance of plant and animal species and spread diseases. However, in their Eurasian native range, wild boar are also a key prey for large carnivores and play an important role as seed dispersers; at moderate densities, wild boar have been associated with increased plant biodiversity (e.g. Welander 2000, Massei and Genov 2004). As there are no large carnivores in UK, wild boar have no natural predators. Estimating local densities and assessing the potential for environmental and economic impact is thus crucial to manage wild boar populations.

The occurrence of wild boar in an area can be quantified in terms of presence/absence, relative density (derived by density indexes) or absolute density. Although several methods have been developed and applied to determine the absolute density of wild boar, the only way to validate the accuracy of these methods, for wild boar as well as for other species, is to compare the estimates obtained through different methods, with the actual number of animals in an area. The latter can only

SID 5 (Rev. 07/10) of 49

be obtained by removing all the animals from an area. Despite the success of many pig eradications from islands (e.g. Parkes et al. 2010) no study has compared the actual number of wild boar in an area with the number estimated through various methods. This is presumably due to the fact that total removal of a population requires a significant effort which very often cannot be justified unless the ultimate aim of the intervention is the eradication of the population. On the other hand, many theoretical studies have been carried out on methods such as distance sampling and capture-mark-recapture to estimate densities of wildlife populations. It is generally agreed that if different methods achieve similar conclusions when applied to the same population, the confidence in their accuracy increases (e.g. Engeman et al. 2001).

In the UK, wild boar were hunted to extinction by 1300 AD (Leaper et al. 1999) but in the ‘90s several populations became established in southern England as a result of escapes from farms (Wilson 2005, Goulding & Smith 2008). Modelling of trends in wild boar numbers and range expansion in the UK suggested that the species might continue to spread, albeit at slow rates, but also that well-established populations could become extinct due to hunting pressure (Goulding & Smith 1998, Moore & Wilson 2005, Holland et al. 2007). Although wild boar have the highest reproductive potential among ungulates (Massei et al. 1996, Bywater et al. 2010) population growth may be offset by hunting, particularly when small populations are found in relatively small, isolated woodlands.

The UK is thus in the unique position of being able to develop a pro-active approach to wild boar population management before potential conflicts escalate. The current project aimed at developing and testing methods to monitor growth, expansion and the relative environmental impact of wild boar in England. By addressing the risks associated with the presence of this species and by evaluating methods to mitigate specific impacts, this approach will also ensure that potential human-wild boar conflicts are prevented or contained before they become intractable. The results will provide Defra and the Forestry Commission with science-based evidence to inform management of wild boar in the UK and will also be used as baseline for future studies.

Objectives1. Develop surveillance methods to monitor density and abundance of wild boar and to detect

range expansion2. Evaluate feasibility and cost of electric fencing to prevent wild boar from damaging crops or

entering areas of high conservation interest3. Identify methods to evaluate the impact of wild boar on the biodiversity of woodland habitats

and apply these to limited woodland habitats within the scope of this study.

Objective 1. Develop surveillance methods to monitor density and abundance of wild boar and to detect range expansion

The study was carried out at five study sites where wild boar populations are well-established: Beckley/Bixley and Brede High Woods in East Sussex (the Sussex Weald), Penyard and Chase Woods (Ross-on-Wye), Ruardean (north Forest of Dean) and Oakenhill (south Forest of Dean). The following surveillance methods were developed and tested to monitor density and abundance of wild boar : (i) activity signs, (ii) camera grids and (iii) distance sampling using thermal imaging. The methods and results for each of these techniques and the conclusions for all three techniques are reported below.

1a. Activity signs The aim of this part of the study was to determine whether activity signs could be used to obtain

indexes of relative abundance of wild boar and to compare these indexes between seasons and between sites.

Methods Initial visits to the study sites in the Weald and in the Forest of Dean confirmed that wallows,

rubbed trees and pellets were too infrequent to be used to develop density estimators. Rooting may provide an indication of wild boar presence in an area but can vary significantly between seasons making it a poor predictor of wild boar numbers (Hone 1988). Thus, wild boar trails were used as the most reliable indicator of wild boar presence.

For each wood, forest rides and tracks (hereafter referred to as forest tracks) were mapped using Ordnance SurveyTM MastermapTM data series and ArcMap 9.3 GIS software (ESRI, California). In

SID 5 (Rev. 07/10) of 49

each wood, 50 x 1 m transects were located along forest tracks to obtain a density of one transect every 12-19 ha per wood, resulting in 27-66 transects per wood. The start point of each transect was randomly placed on forest tracks using Hawth’s Analysis Tools for ArcGIS. No independence was assumed between data from transects within and between tracks within a wood. A pilot survey was carried out in July 2009 and the number and location of wild boar trails that intercepted the transects were recorded. Due to the staff effort required for this survey, in winter 2010 the survey was modified by extending the length of 50% of transects to 200 m. Thus 200 x 1 m transects were used to record wild boar trails in early March (winter) and August (summer) 2010.

A residual maximum likelihood analysis (REML) was used to derive an activity index for each wood. A REML analysis, using season as a fixed effect, was also employed to compare the activity index between summer and winter.

The predicted number of boar trails xijk for season i and transect j was calculated as follows: xij = μ + Si + Tj + εij

where Si is a fixed effect for season, Tj is a random effect for transects and εij accounts for residual variability within season and transect.

Following Engeman (2005) and Engeman et al. (2002) a Passive Activity Index (PAI) for each season i was then derived as:

where ti is the number of transects within season i.

The proportion of 200 m transects with wild boar trails in 5 woods was compared between March and August 2010. As a preliminary analysis using a binomial test indicated that the proportion of 200 m transects with wild boar trails in summer was less than 9% in all 5 woods but varied between 9 and 59% in winter (see results), the comparison of PAIs between sites was restricted to PAIs calculated for winter only. A REML analysis was used to estimate PAIs and their relative variance after restricting the data to winter data and removing the fixed effect from the above model. Winter PAIs between woods were compared by Z-tests.

Results The proportion of transects with boar trails was consistently higher in winter than in summer: although the difference was significant for only 2 woods (P < 0.05, Beckley/Bixley and Ruardean), the same trend occurred for the remaining 3 woods (Figure 1).

0

10

20

30

40

50

60

70

Beckley/Bix. Brede Oakenhill Penyard/Ch. Ruardean

% o

f tra

nsec

ts w

ith w

ild b

oar

trai

ls

March 2010

August 2010

Figure 1. Proportion of 200 m transects with wild boar trails in five woods in England in March and August 2010.

SID 5 (Rev. 07/10) of 49

Thus the PAI indices for each wood were recalculated for trails for winter only (Table 1). There was no evidence of differences in winter PAIs between any of the woods (Z-test for all P>0.05)._______________________________________________Wood N. transects PAI SE_______________________________________________

Beckley/Bixley 32 1.398 0.347Brede 22 0.182 0.125Oakenhill 17 0.412 0.173Penyard 22 0.318 0.138Ruardeen 14 0.714 0.322_______________________________________________Table 1. Winter Passive Activity Index (PAI) and standard error calculated using the number of wild

boar trails on transects (200 x 1 m) in 5 woods.

1b. Camera trap surveys

The aim of this part of the study was to determine whether camera trap surveys could be used to obtain indexes of relative abundance as well as absolute densities of wild boar and to compare these indexes and densities between seasons and between sites.

MethodsIn each wood, motion-activated cameras (Penn’s Woods DS-04IR, DS06IR and Moultrie I60)

were placed on a grid pattern. Following Rowcliffe et al. (2008), cameras were placed at a density of one every 11-12 ha and were moved every ten days to cover each wood in 2-3 weeks and complete a wild boar survey (Figure 2).

Figure 2. Camera trap positions, at the Penyard-Chase study site, used to estimate wild boar abundance.

As fully randomized placement could result in cameras being positioned in areas of no visibility, cameras were positioned either along forest roads or in areas of relatively higher visibility within 50 m of the grid points.

Camera trap surveys were carried out in summer 2009, winter 2010 and summer 2010. A PAI on number of wild boar visits per 10 days was then calculated for each wood and compared between woods. One visit was defined as >1 photos of wild boar until there was a lapse of at least 10 minutes between consecutive photos: photos of wild boar taken > 10 minutes later were counted as a new visit.

SID 5 (Rev. 07/10) of 49

Preliminary observations with ear-tagged animals indicated that wild boar usually do not return to the same area within 10 minutes.

REML analyses were carried out on data obtained from camera trap surveys: in the first analysis, season (summer 2009, winter 2010 and summer 2010) was entered as a fixed effect, to investigate potential differences in PAIs between seasons, for each wood. A PAI was derived for each wood separately, as follows:

where ci is the number of cameras within season i and xij is the predicted number of pictures for camera traps j in season i and can be written as:xij = μ + Si + Cj + εij

where Si is a fixed effect for season and Cj is a random effect for cameras and εij accounts for residual variability within season and camera traps.To avoid carrying out too many comparisons, PAIs between woods were compared by Z-tests using only the most recent camera trap survey in summer 2010.A density estimator D was calculated (after Rowcliffe et al. 2008) for each wood and each survey separately, based on the number of wild boar visits per 10 days as follows:

where y/t = number of visits y per unit time t r and θ= radius and angle of the camera’s detection area v = speed of movements

D was then multiplied by group size to obtain the density (Rowcliffe et al. 2008). Independent

estimates of group size were obtained from another study, carried out in two of the sites used for this study, where wild boar were attracted to bait stations and remotely recorded by camera traps.Group size was calculated in January 2010 (n=24 observations) and in May 2010 (n=36 observations) by placing camera traps in Penyard and in Oakenhill. To minimize potential double counting, individual groups or animals were identified by a number of features including ear tags and GPS collars, ratio of females to piglets and physical traits such as body size and coat colour. The speed of movements, expressed as km/hr was obtained from wild boar (n=7) equipped with GPS collars which were programmed to record fixes and activity every 15 minutes in the Penyard-Chase site (Quy et al. in preparation).

Bootstrapping (Efron, 2000) was used to estimate the uncertainty associated with the density estimates by re-sampling 10,000 times the data from camera trap pictures at random and by estimating the corresponding density. Then, for each wood and each season, a mean density and a standard error were obtained from the bootstrapped data. A REML analysis was used to compare densities between seasons within woods. For the most recent survey, summer 2010, densities between woods were compared by Z-tests.

ResultsThe results obtained from the camera trap surveys suggested that in some woods PAIs differed

between seasons (Table 2). The PAI also indicated that between September 2009 and August 2010 wild boar numbers decreased significantly in Beckley/Bixley and increased in Penyard-Chase. In all other woods there was a trend towards increasing numbers of animals although the differences were not significant. In the most recent survey (August 2010) PAIs based on camera traps differed between Brede and Penyard (Z = 3.061, P = 0.002), Brede and Ruardean (Z = 1.929, p = 0.054), Oakenhill and. Penyard (Z= 3.260, p = 0.001) and Oakenhill and Ruardean (Z = 2.072, p = 0.038).

The following data were used to obtain density estimates following Rowcliffe et al. 2008): radius (r) of the cameras = 12 m, angle (θ) of the camera’s detection area =52 degrees (= 0.90757 radians), speed of movements (v) = 0.274 (+SD 0.052) km/hr. Group size in January was 2.83 (+SD 1.86) and in May was 2.11 (+SD 0.92) when counting only adults and 6.22 (+SD 5.28) when including juveniles.

_______________________________________________________________________

SID 5 (Rev. 07/10) of 49

Wood Month N. of cameras PAI SE P

________________________________________________________________________Beckley-Bixley Sep-09 45 1.053 0.261 0.029

Jan-10 1.541 0.264 Aug-10 0.528 0.261

Brede Sep-09 15 -- -- Jan-10 0.193 0.130 0.079 Aug-10 0.285 0.130

Oakenhill Sep-09 19 -- -- 0.62 Jan-10 0.119 0.139 Aug-10 0.222 0.135

Penyard Sep-09 34 0.248 0.244 0.032 Jan-10 0.482 0.248 Aug-10 1.121 0.240

Ruardean Sep-09 13 0.167 0.393 0.10 Jan-10 0.745 0.437 Aug-10 1.083 0.393

________________________________________________________________________Table 2. Passive Activity Index (PAI) and standard error calculated using number of wild boar visits per

camera trap in five woods in September 2009, January 2010 and August 2010. Only part of Brede and Oakenhill wood were surveyed in September 2009 and thus the PAIs were not calculated. P refers to between season comparisons.

The relative densities of adult wild boar in different sites varied between 1.8 and 7.6 animals per km2 in September 2009, between 0.8 and 11 animals per km2 in January 2010 and between 1.2 and 6.1 animals per km2 in August 2010 (Figure 3). When calculated per wood and season, wild boar densities between September 2009 and August 2010 showed a significant decrease in Beckley/Bixley (F 2, 82.6 = 3.69, P = 0.029,) a significant increase in Penyard-Chase (F 2, 62.9 = 3.64, P = 0.032) and remained stable between seasons in the other three sites, although the trends in all these sites appeared to increase with time.

In the most recent survey (August 2010) wild boar densities based on camera trap estimates differed between Brede and Penyard (Z= 2.285, P = 0.022) and between Oakenhill and Penyard (Z = 2.302, P = 0.021).

Figure 3. Estimated mean (+SE) numbers of adult wild boar per km2 in five English woodlands between September 2009 and August 2010. Numbers are derived from camera trap surveys and field estimates of mean group size.

1c. Thermal imaging

SID 5 (Rev. 07/10) of 49

The aim of this part of the study was to determine whether distance sampling based on thermal imaging could be used to obtain absolute densities of wild boar.

MethodsSince 1994 thermal imaging, in combination with distance sampling has been successfully used

to estimate deer densities, and in a few cases estimates of wild boar density were obtained from sites where both deer and boar coexisted. These include Chizé, a mixed oak-beech forest in central France, surveyed in 1999 to estimate the density of roe deer and the forest of Dean, which has been surveyed every 2-3 years since 1998 to estimate fallow deer numbers and where wild boar were first detected in 2005.

The present study anticipated that distance sampling based on thermal imaging would offer the best prospect to assess wild boar density. Estimates of wild boar density were obtained in Beckley and Bixley woods in winter 2009. However, numbers of boar were lower than expected, probably because of a recent increase in local culling, so another attempt to estimate numbers was made in 2010 in Penyard. Data from wild boar equipped with GPS collars, collected under Defra-funded project WM0408 were used to investigate diurnal behaviour patterns and evidence of avoidance of transects, a potential source of bias in distance sampling

The survey methods adopted were similar to those developed for deer (Gill et al. 1997; Mayle et al. 1999). The areas surveyed vary substantially in size and landscape composition (Table 3). The existing network of paths and tracks (referred to as ‘transects’) through each wood was used to make observations with the thermal imager. To ensure effective coverage, a ratio of 2.5 km of transects per km2 of woodland and 1.0 km per km2 of fields is normally recommended for deer surveys (Gill et al. 1997). Access permission could not be obtained for the entire contiguous block of woodland in Beckley; nonetheless, within the areas where access was possible, sampling intensity was higher than recommended. The ratio was exceeded in all sites except Chizé.

In two sites (Penyard and Beckley/Bixley) fields adjacent to the woods were surveyed at the same time as the woodlands. Boar activity was evident in some of the fields near the woodlands. As the GPS data (as well as signs and anecdotal reports) indicated that fields near the woodlands at Beckley/Bixley and Penyard were frequently used by wild boar, these fields were surveyed at the same time as the woodlands. Very little or none of the boundary adjoined fields at Chizé and the Dean.

During the survey, an attempt was made to identify all animals to species level. For the analysis, unidentified animals were assumed to be either boar or deer in the same ratio as those identified.

The surveys in Beckley and Penyard were carried out by Forest Research as part of this project. Surveys in the Dean and High Meadow were carried out by staff from the Dean Forest office. Forest Research collaborated with the Office National de la Chasse, France for the survey in Chizé.

Site Year Area (ha) Transect length(km) in survey area

Sampling intensity(km/km2)

Grid Ref Woods Fields Woods Fields Woods FieldsPenyard 362000 223000 2010 231 191 10.9 2.3 4.7 1.2Beckley/ Bixley 2010 231 191 2009 480 850 17.8 17.3 3.7 2.0Chizé 586000 122000 1999 2500 - 41.0 - 1.7 -High Meadow 356000 213000 2005 1200 - 26.0 - 2.2 -Dean 362000 212000 2008 5787 - 174.6 - 3.0 -

Table 3. Summary of survey locations and areas

Density estimates (D) from distance sampling were obtained as follows:

D = [E(S).n/L]/2ESW

where n/L = encounter rate, expressed as number of groups encountered per km transect E(S) = mean group size ESW = effective strip width.

The effective strip width is derived by fitting a function to the distribution of distances of each group from the transect. Thus animals detected at greater distances indicate a wider survey area and yield a larger ESW. To obtain density estimates using distance sampling, it is recommended that 50+ observations (of groups, not individuals) are used to fit a detection function.

SID 5 (Rev. 07/10) of 49

ResultsThe distributions of detection distances indicated a difference between Chizé and the UK

populations (Figures 4 and 5). Data from these sources was not combined, even though this meant using less than the recommended 50 observations for the detection function. The ESW derived from Chizé was smaller than that for the UK populations (Table 4). Since observations made at large distances can result in a poor fit of the detection function, it is normal practice to maximise fit by truncating the farthest observations. This resulted in the exclusion of three observations at Chizé and two (8 boar) from Penyard, in the latter case both groups were in fields beyond the woodland boundary. No wild boar were observed in the fields adjacent to Beckley/Bixley.

The estimated densities in the UK sites were all rather low, ranging from 0.4 to 5.4 animals per km2, with a relatively large coefficient of variation (Table 4). In some instances, the number of groups seen was higher than the number of animals estimated in the population because transects were sampled repeatedly and the same animals were encountered on several occasions. The densities in 2009-2010, when the current project was carried out, ranged between 2.2 and 5.4 wild boar per km2. The large coefficients of variation reflect the small numbers of observations from which the densities were derived. Since two observations were made in the fields adjacent to Penyard, an additional two animals could be added to the estimated population, on the basis that a mean number of 0.5 animals were observed each night and approximately 25% of the field area was visible. As noted below, the GPS data revealed that boar favour fields close to a woodland margin. However, because of the terrain, the edge of Penyard wood was often not clearly visible from the transects, suggesting that numbers in fields may have been under-estimated.

Figure 4. Proportion of wild boar groups detected at successive distances from the transect, for UK populations (combined) and Chizé.

Figure 5. Detection functions fitted to the distance data for UK populations and Chizé.

SID 5 (Rev. 07/10) of 49

No. ofgroups

No. nightsof survey

Transect lengthsampled

Encounter rateGroupsKm-1

Meangroup size

ESW (m)

Boar perKm2

Est. Pop.size, N

Coefficient of variation

Penyard 24 14.5 81.8 0.269 3.3

82.4

5.4 15 42.6

Beckley 6 7 30.8 0.195 1.8 2.2 10 60.1

High Meadow

1 4 48.0 0.021 24.0 3.0 36 101.0

Dean 8 19 318.0 0.025 2.9 0.4 25 52.3

Chizé 32 7 41.5 0.747 2.4 55.4 16.5 413 28.5

Table 4. Results of density estimates through thermal imaging and distance sampling. The Coefficient of Variation is expressed as the ratio between the standard deviation and the mean of the estimated density.

The survey in Chizé did not quite achieve the recommended sampling intensity (Van laere et al 2003).The GPS data from 11 radio-collared wild boar monitored by Fera were used to investigate diurnal activity patterns. The results revealed substantial differences during a 24hr period (F23,1497=16.79, p<0.001) and suggest wild boar were significantly more active at night There appears to be greater activity in winter than summer with a peak in activity before dawn in winter but after dawn in summer.

The GPS data was also used to investigate avoidance of transects and selection for concealment. Selection ratios, Si = Ln (Ui/Ai), were calculated where Ui = the proportion of all GPS fixes, and Ai = the proportion of study area, in habitat i. Habitats with relative use greater than availability yield values of Si > 0 and conversely Si < 0 for those with use less than availability. The results suggested that, at least in Penyard and Chase woods, animals have a tendency to avoid transects, up to a distance of about 20m, but this is more pronounced during the day, and almost negligible at night time.

1d Cost-effectiveness of different methodsBased on the above results, a preliminary assessment of the cost-effectiveness of the three

methods was carried out by evaluating the time spent by staff to complete field work associated with recording activity signs on transects, camera trap surveys and thermal imaging in 230-480 ha woods (or 400-1500 ha if the surrounding fields are included, as in the thermal imaging survey) (Figure 6). In addition, the different methods were evaluated in terms of advantages and disadvantages regarding feasibility, timing and constraints (Table 5).

0

10

20

30

40

50

60

Activity signs Camera grids Distance samplingActivity signs Camera traps Thermal imaging

Figure 6. Time (in hours) spent by staff to complete three different field surveys to assess wild boar population trends and densities. In this example, activity signs (trails) were detected on 200 x 1 m transects (n=33), camera trap surveys were carried out by positioning 22 camera traps; thermal imaging with distance sampling was carried out on 7 nights along 31 km of forest tracks.

SID 5 (Rev. 07/10) of 49

Hou

rs

All methods require a certain amount of training to carry out field work as well as additional training to input and analyse data. For instance, staff must be trained in recognising activity signs left by wild boar from those of other species, and in operating camera traps or the thermal imaging equipment. Training staff in the application of the different methods was estimated to have similar costs as training would be mainly focussed on using the equipment or on activity signs recognition. In terms of equipment, the method based on activity signs does not require any, whilst camera traps and thermal imaging must be purchased or hired. Current prices for a camera trap vary between £200 and £1,500 and the minimum price of a thermal imager is about £15,000. However, camera traps and thermal imagers also enable the collection of data on other species. In addition, local managers can share the cost of purchasing expensive equipment with other groups interested in managing wildlife populations, thus effectively reducing the expense of monitoring ungulate densities in an area.

Advantages DisadvantagesActivity signs(trails)

Used for relative index of abundance

Better results when used in winter than in summer

Better used on trails than on other activity signs

PAIs relatively easy to calculate

Not usable for absolute densities

At low densities, unable to detect differences between sites

Not reliable during the dry season

Difficult to tell old from recent trails

Camera traps

Usable all year

Used for relative index of abundance and to estimate absolute density

Can be used at low densities

Can be used to monitor other species

Provide immediate“snapshot” of animal presence

PAIs relatively easy to calculate

Require independent assessment of group size and speed to estimate absolute density

Initial costs of camera traps

Camera traps may be stolen

Theoretical framework in progress

Thermal imaging with distance sampling

Must be used in winter for best visibility

Can be used to monitor other species

Used to estimate absolute density

Best used for wild boar populations at medium-high density

Well-established theoretical framework

Not reliable in summer due to poor visibility

Expensive equipment

Low accuracy at low densities

Table 5. Advantages and disadvantages of the methods used to determine population trends or absolute density of wild boar in England.

DiscussionThe results of this study showed that the method based on camera trap surveys could be used

to detect differences in wild boar population abundance indexes between and within sites and to estimate densities of wild boar in English woodlands. As the number of wild boar groups observed and the number of activity signs such as trails were too few and variable, surveys based on thermal imaging and on activity signs proved to have a low precision in estimating abundance indexes or actual densities at least at the densities present in the study sites. In addition, both thermal imaging and activity sign surveys were restricted to winter due to better visibility (thermal imaging) or persistence of

SID 5 (Rev. 07/10) of 49

activity signs which were more detectable in winter than in summer. Conversely, camera trap surveys were less likely to be affected by season and could be used any time of the year.

Thermal imaging and line transects have been used to assess ungulates’ numbers in several studies (e.g. Focardi et al. 2002, Haroldson et al. 2003, Franzetti and Focardi 2006) generally in areas with relatively high densities of animals. For instance Focardi et al. (2002) used this method in a Mediterranean area and obtained densities of 8.5-9.9 deer per km2 and 10.6 wild boar per km2 with relative low coefficients of variation of 21% and 24% respectively and Hodnett (2005) advocated using thermal imaging in urban areas with overabundant populations of deer.

The accuracy of line transects and camera traps in estimating wild boar density could not be calibrated against distance sampling. However, density estimates obtained using the distance sampling were similar to those derived from camera trap surveys. In addition, in 2010 the Forestry Commission estimated a density of circa 3 wild boar per km2 in the Forest of Dean (Stannard pers. comm.) which falls between the estimates of 0.84 and 4.74 wild boar per km2 obtained from camera trap surveys in Oakenhill and in Ruardean woods, both part of the Forest of Dean. The consistent results in wild boar density obtained by using different methods was encouraging and suggested that the actual density of wild boar can be estimated. Future studies should focus on refining these methods to improve their accuracy and precision.Estimating the absolute density requires knowledge of the true number of animals in the study area, which could be possible only if all the above methods were applied and the animals were removed from that area (Hounsome et al. 2005). In the majority of cases this might not be possible or indeed desirable and it is likely to be an expensive exercise. However, if the eradication of a localized population of wild boar became necessary, for instance following a disease outbreak, all the methods used in this study might be applied prior to eradication.

The results of the current study suggested that, at current densities, the camera trap surveys had the best cost-benefit ratio compared to the other two types of surveys. Activity sign surveys were relatively inexpensive to conduct but relied on relatively few signs, on the subjective ability of observers to distinguish wild boar trails from those of deer and on the freshness of the trails, in turn affected by precipitations and thus easier to observe in winter than in summer. Similarly, the disappearance rate of other activity signs, such as the pellet groups of deer and wild boar, has been shown to vary dramatically between seasons and even habitat types (Massei et al. 1998). Therefore, at least at relatively low densities of wild boar, trails on transects should not be used as indexes of relative abundance in summer.

Thermal imaging and distance sampling resulted in large confidence intervals around the estimate of wild boar numbers and was also the most expensive method in terms of staff time. This method would be even more costly if the equipment had to be purchased. Although at current wild boar densities thermal imaging did not provide the best cost-benefit ratio, these conclusions might change if wild boar densities increase and the precision of the method can be improved.

The relatively small variation associated with the estimates derived from the camera trap surveys resulted in detectable differences, at least between some seasons or sites, in the relative indexes of abundance and in the estimated densities of wild boar. Due to the availability of newer, less expensive equipment suitable for field trials, animal population surveys based on camera trap surveys are increasingly employed in wildlife management (e.g. Rowcliffe et al. 2008, Rovero & Marshall 2009, Gerber et al. 2010). This has been complemented by a growing number of papers aimed at establishing a conceptual framework for the optimal use of camera trap surveys (e.g. Royle et al.2009, Gardner et al. 2010).

The densities of wild boar estimated in the current study are relatively low compared to those reported in continental Europe where they range from 2 – 7 animals/km² (Smiet et al. 1979; Ickes 2001; Pihal et al. 2010) to 10-11/km² (Focardi et al. 2002, Hebeisen et al. 2007) and up to 51 animals/km² (Franzetti et al. 2010). The densities recorded in the current study are also similar to those observed for populations subjected to hunting pressure in the Bialowieza Primeval Forest in Poland, where densities varied from 0.7 to 5.1 wild boar/ km² (Jedrzejewska et al.1994) and in coastal regions of California with densities of 0.7 to 3.8 wild pigs / km² (Sweitzer et al. 2000).

In summary, the results of this study suggested that local densities of wild boar in England are still relatively low compared to those in mainland Europe although in four out of the five sites surveyed in the study the trend in relative densities of wild boar increased.This study provided the first comparison of different methods to estimate wild boar densities and population trends for two consecutive years in five sites in England. The project also identified a method, based on camera trap survey, that could be used to obtain indexes of abundance and relative density estimates for wild boar populations in England. Future research should focus on refining this

SID 5 (Rev. 07/10) of 49

method to determine the optimum time of the year and the number and density of camera traps required to maximize the benefits of using camera trap surveys. Future studies should also establish the number of replicate surveys required to detect a set decline or increase in wild boar density.

1e Methods to detect range expansionWild boar populations may rapidly expand, as observed in Sweden (Thurfjell et al. 2009) and in

the US (e.g. Waithman et al. 1999). Therefore, the potential expansion of wild boar in England should be monitored to determine whether this species has colonised new areas. Detecting spread into new areas can be challenging, particularly in the short-term (2-3 years), since the edge of populations’ ranges are typically characterised by low densities. Large-scale range expansion has been assessed using presence records, often gathered by volunteers: this approach is particularly problematic because data validation is often impossible and absence of records does not necessarily mean absence of animals (MacKenzie et al. 2002, Ward 2005). However, recent advances in surveillance technology, such as camera traps and novel applications of analytical methods (e.g. likelihood-based models) have resulted in options now available to maximise detection probabilities in low-density areas. For instance, Mackenzie et al. (2002 and 2003) suggested that it is possible to estimate site occupancy, colonization and local extinction probabilities from remote surveillance data on animal presence, where the probability of detecting a target animal is less than one. However, the capabilities of these methods have yet to be tested for large mammals in England.

Surveillance is particularly important if a species such as the wild boar is to be managed to maintain populations within discrete geographical ranges or below critical density thresholds beyond and above which the species’ impacts may become unacceptable. This part of the project evaluated methods to monitor the potential expansion of wild boar in England.

MethodsA stepped approach for the quantification of wild boar range expansion was developed. This

process requires collating information on boar presence in new sites, confirmation of boar presence in these areas and quantification of the rate of expansion once boar presence has been confirmed.

Step1: Mapping sightings and road traffic accidentsA wide network of contacts with stakeholders was established to collate information on sightings

and road traffic accidents (RTA) involving wild boar. Data were provided by the Forestry Commission, Natural England (Mr Charlie Wilson) and by the RSPCA. Records were entered onto a spreadsheet, classified by location, date, type of record such as sighting, shot, or RTA. These records were reported on a GIS to provide a preliminary map of wild boar historical and present occurrence.

Step2: Confirmation of wild boar presence at bait pointsThis part of the study assumed that if an RTA is recorded or a wild boar is sighted for the first

time in an area where the species is not known to occur, camera traps associated with bait points could be used to confirm whether wild boar have established in an area. The method was first tested in woods with stable populations of wild boar and then used to determine the potential expansion of wild boar in woods surrounding the Forest of Dean (see Step3). In October 2009 and in March 2010 baited camera traps were deployed in woodlands where wild boar were known to be present to determine whether this technique could be used as a method of confirming sightings and reports. In September 2009 and February 2010 bait stations were established at the same sites used for the camera trap surveys. Each bait station was chosen at random from locations where wild boar had been photographed during the camera trap surveys. Seven sites (Beckley and Bixley, Penyard and Chase were split as they both comprise two adjacent woodlands) had a single bait station installed (Table 6). Two camera traps per site were placed at each bait station and left until wild boar had been detected or for up to 27 nights. In September 2009 boar had not been photographed in Brede or Oakenhill, so the October locations were chosen on the basis of recent rooting signs (<30 days). At each bait point, loose maize and maize placed in plastic pipes (these had been shown previously to attract wild boar but not other non-target species) was used as bait and replenished every 4 - 5 days. Data collected during the camera trap surveys carried out in August 2010 were also used by randomly choosing seven camera locations where pictures of wild boar had been taken (although baits was not present). The data from September 2009, February 2010 and August 2010 were used to simulate a sighting of wild boar in a new area and to calibrate the method developed to confirm their presence.

SID 5 (Rev. 07/10) of 49

Wood October 2009: number of nights cameras installed

March 2010: number of nights cameras installed

Beckley, Sussex 23 4Bixley, Sussex 23 4Brede, Sussex 18 4Penyard, Ross-on-Wye 27 9Chase, Ross-on-Wye 7 11Oakenhill, Forest of Dean 13 6Serridge, Forest of Dean 16 8

Table 6. Woods with bait station and number of nights camera traps were installed at each wood to confirm wild boar presence.

Wild boar occupancy was analysed by the program PRESENCE v.2 (Proteus Wildlife Research Consultants, New Zealand; http://www.proteus.co.nz) which estimates site occupancy rates when detection probabilities are less than one (MacKenzie et al. 2003). Following MacKenzie et al. (2002 & 2003) the variables used for the analysis were: the total number of sites surveyed, the number of distinct sampling occasions, the number of sites where the species was detected at time t, and the number of sites at which boar were detected at least once. The main output of this analysis was an estimate of the proportion of sites occupied by boar. Occupancy and detection probability were calculated also for September 2009, February 2010 and August 2010, were no food was provided.

Step3: Camera traps to determine range expansion In December 2010 baited camera traps were used to ascertain the presence of boar in 21

woods on the periphery of the Forest of Dean (FoD). According to MacKenzie and Royle (2005) the minimum number of surveys and sites required for precise estimates of occupancy (the proportion of sites occupied) varies according to occupancy levels and detectability of the target animal. Accordingly, the number of sites (21) and the minimum number of nights (14) that the cameras were installed were based on preliminary data on detection probability and site occupancy collected in the FoD and using the table provided by MacKenzie and Royle (2005). All woods selected were at least 3 km from the outer boundary of the FoD as the home range of wild boar radiotracked in Penyard and Chase woods in two-week periods (n = 6) did not extend further than 2.6 km (Lambert pers. comm.).

For each site, the following site-specific variables, which might have affected the presence of boar in a wood, were collected: 1. dominant (>50%) tree species: mixed broadleaf, conifers, oak, sweet chestnut, birch, beech; 2. structure of stand: standard, coppice, plantation, clearing; 3. presence of mast (chestnuts, acorns, etc) in the vicinity (<100m from bait station); 4. wood area (ha); 5. distance of the wood centre from the nearest FoD edge (km); 6. bait station distance from nearest habitation (km) and 7. bait station distance from nearest wood edge (interior point to nearest exterior edge).

The following variables, which might affect the probability of recording boar at the camera trap, were also collected: 1. boar signs within 100m (tracks, pellets, rooting, tusking, etc); 2. boar signs in wood; 3. bait station on or by animal trail; 4. bait station distance from road (km); 5. bait station distance from public footpath (km); 6. air temperature recorded at midnight (taken from data recorded on photos); 7.number of nights that bait remained at site; 8. any signs of shooting for wild boar in wood (high seats, cartridges, bait, etc). To obtain variables 1-3, one operator walked for 1-2 hours in each wood to record opportunistically activity signs by wild boar. Photos were checked for boar presence after the trial finished. The time period for analysis was the 24 hour period 1200h – 1200h of the following day. The program PRESENCE was used to analyse the results.

ResultsStep1: Mapping sightings and road traffic accidents.In total, 453 records were obtained; nearly half (222) of these records originated from the FoD,

where all shot animals are reported and entered on a database.

SID 5 (Rev. 07/10) of 49

http://www.proteus.co.nz/

Figure 7. Locations of boar records received between 1990 and 2011..

Records were also collected from 26 counties in England and Wales (Table 7, Figure 7). Besides the FoD, wild boar occurred in Devon, Dorset and East Sussex and in other southern counties (Table 7).

County Number of records Years recorded Number of sites where deemed resident

Bedfordshire 1 2001Cheshire 2 1990Devon 57 1990-2007 11Dorset 51 1990-2008 5Durham 1 1993East Sussex 21 1994-2010 11Essex 9 1990-1998Forest of Dean (Gloucs.)

222 1990-2011 1

Gloucestershire 2 2000-2006Hampshire 4 2006-2009 1Herefordshire 7 2000-2010 1Kent 16 1994-2006 3Leicestershire 1 1995Manchester 1? 1990Monmouthshire 2 2008-2009 1Norfolk 3 1992-2005North Yorkshire 13 1990-2005Northamptonshire 1 1990Nottinghamshire 1 1990Shropshire 1 1998Somerset 6 1990-2006 2Suffolk 11 1991-2009 1Tyne & Wear 5 2001Warwickshire 4 1990-2008 1West Sussex 4 2006-2008 3Wiltshire 3 2006-2007 1Worcestershire 1 2003

Table 7. Number of boar records, dates, and number of locations where wild boar were already regarded as resident between 1990 and 2011.

SID 5 (Rev. 07/10) of 49

Step2: Confirmation of wild boar presence at bait pointsIn October 2009 wild boar were not recorded at one site (Penyard).For all the other sites, it took

between 3 and 23 nights to obtain the first photo of a wild boar (Figure 8). The proportion of sites occupied and estimated by PRESENCE were very similar (0.7143 vs. 0.7324). The detection probability was 0.2654 (Table 8). In March 2010 wild boar were recorded at all sites, and first recording occurred between 1 and 9 nights (Table 8). The proportion of site occupancy observed and estimated by PRESENCE were both 1 (all sites), and the detection probability was higher than in October 2009. In July 2010 the detection probability, calculated on all camera traps recording wild boar, was 0.1868. Comparing data from September vs. October 2009 and February vs. March 2010 the detection probability was doubled when food was supplied.

Step3: Camera traps to determine range expansionIn December 2010 twenty-one woods (Table 9) had camera traps installed. At one site, camera

traps malfunctioned (Cuckoo Wood) and the site was discarded from the analyses. The woods were 3 - 15 km from the FoD (mean = 7.7 ± 0.8km). Four woods, Cobbler’s Plain, Highmeadow, Cadora Woods and Knockall’s Inclosure had signs of wild boar. Only one of these wood (Knockall’s Inclosure) recorded boar on camera, and then only once on the second night. In three other woods operators were unable to decide whether the activity signs observed could be attributed to wild boar. Penyard wood, used as control site, recorded boar on the 5th and 8th night.

The PRESENCE analysis indicated that as detection probability was so low (0.0036), no meaningful conclusions could be drawn about site occupancy.

Figure 8. The number of nights before wild boar were first recorded at bait stations during October 2009 and March 2010.

Camera trap month(* = bait stations)

First record range (mean nights)

Observed versus estimated site occupancy

Detection probability

Sept. 2009 4 - 13 (8) 0.7143 v 0.8749 0.1223Oct. 2009* 3 – 23 (9.5) 0.7143 v 0.7324 0.2654Feb. 2010 1 - 12 (5.3) 1 v 1 0.1744March 2010* 1 – 9 (4.1) 1 v 1 0.3750July 2010 1 – 6 (3.8) 1 v 1 0.1868

Table 8. Camera grid and bait station data: boar first recorded, site occupancy and detection probability.

SID 5 (Rev. 07/10) of 49

____________________________________________________________________

Wood name Grid Ref Stand type

Distance from FoD edge (km)

Boar sign in wood

Kings Wood SO 4712 Oak 15 No

FEDW Wood ST 5799Mixed broadleaf 10.7 No

Craig Llewellyn SO 4718 Sweet chestnut 13 NoCobbler’s Plain SO 4800 Sweet chestnut 12.5 Yes

Cuckoo Wood SO 5204Mixed broadleaf 7 No

Highmeadow SO 5314 Sweet chestnut 6 YesEast Wood ST 5699 Conifers 6.5 ?

Dancing Green SO 6320Mixed broadleaf 3 No

Hay Wood SO 7019Mixed broadleaf 7 ?

Queen’s Wood SO 6727 Oak 7.5 NoHuntley Hill SO 7019 Sweet chestnut 4.3 ?

Newent Woods SO 7021Mixed broadleaf 5 No

Collinpark Wood SO 7428Mixed broadleaf 9.6 No

Highnam Wood SO 7719Mixed broadleaf 8.8 No

St Pierre’s Great Wood ST 5092

Mixed broadleaf 8.9 No

Wentwood ST 4295 Conifers 12.9 No

Blaisdon Wood SO 6917Mixed broadleaf 4 No

Cadora Woods SO 5407Mixed broadleaf 3 Yes

Little Doward Wood SO 5316 Sweet chestnut 4 No

Knockall’s Inclosure SO 5411Mixed broadleaf 3 Yes

Haugh Wood SO 5936Mixed broadleaf 13.2 No

____________________________________________________________________ Table 9. Woods surveyed in December 2010 to determine potential range expansion of wild boar.

DiscussionThe results of mapping wild boar sightings and road traffic accidents indicated that wild boar

records are widely scattered across England and Wales, but mainly occurred south of a line between the Severn and the Wash. This confirmed previous data collected by Wilson (2005 and 2010) on wild boar status and dispersal in England. Boar have become established in some counties but are still very localised.

The British Wild Boar website (http://www.britishwildboar) lists sightings in other counties (e.g. Buckinghamshire, Cornwall, Hertfordshire) that have not been confirmed by the present study and that are provided by the general public. More advanced options, such as the US National Feral Swine Mapping System (NFSMS) (Corn et al. 2010) are available to monitor the distribution and expansion of wild boar in the UK. The NFSMS is a web-based interactive data collection and mapping system currently used to archive and display area data for the distribution of feral swine in the United States. Data on established populations of wild pigs are provided by agency personnel who work directly with the map on the NFSMS server using interactive mapping routines and Google Map protocols. With these tools, the managers are able to add new locations, revise existing ones or delete areas where

SID 5 (Rev. 07/10) of 49

http://www.britishwildboar/

feral swine are no longer present. Distribution data submitted by agency personnel are evaluated and filtered according to reliability criteria on a continual basis, and the digital geodatabase and distribution map are updated with verified additions on a monthly basis. This system could also be operated in England and used for instance to map pig farms and wild boar populations or to monitor distance of established wild boar population from disease outbreaks.

As wild boar might expand in England over the course of the next 20-30 years (Defra 2008) knowledge of their status will be beneficial to local communities and landowners responsible for their management. As Defra is to facilitate this through advice and guidance, maintenance of the database will assist in appropriate management.

The analyses indicated that even when wild boar populations are well established, and site occupancy may be 100%, detectability is comparatively low in most months (<0.3). This is probably the result of large home ranges and relatively low densities and the small likelihood of any boar or group of boar passing one camera trap in a wood during a survey period. When occupancy is high and detectability is low more surveys (nights of recording) should be done at each site (MacKenzie & Royle 2005). In the table provided by MacKenzie & Royle (2005), given a site occupancy of 0.7 and detection probability = 0.3 (from the October analysis), a minimum of seven sampling occasions (nights) would be required at each site.

The mean number of nights before first recording, which ranged between 3.4 and 9.5 nights, suggested that at any particular site boar may not be recorded for some time. However, when comparing data from September vs. October 2009 and February vs. March 2010 the detection probability was doubled when food was supplied. This indicated that, in case of an RTA or a sighting of a wild boar in a new area, the provision of bait would attract wild boar and enhance the probability of confirming the presence of the species on site.

The range expansion survey based on bait stations indicated very little movement away from the core area of the FoD as yet. Only one bait station recorded boar, and only on one occasion. However, preliminary surveys of the woods before bait station installation did reveal boar signs in at least four woods. Unless the boar sign is very fresh this may not indicate recent occupancy, and camera traps are needed to confirm that wild boar are still in the area.

Male boar often have larger home ranges than females (e.g. Baber & Coblentz 1986, Keuling et al. 2009). Normal dispersal distances are less than 20 km (Dardaillon & Beugnon 1987; Truvé & Lemel 2003), and in England the maximum distance moved by a radio-tracked animal has been a male which moved 20 km from the site where it was trapped (Moore 2004). Compared to other European ungulates wild boar disperse more frequently (Cargnelutti et al. 1992). Wild boar and feral pig annual home range varies widely throughout their geographical range – for males 2.8 – 25.7km2 and females 1.4 – 54.1/km2 (references in Moore 2004). Some individuals may have annual home ranges up to 154 km2, particularly in areas where hunting activity might force the animals to move away from the normal home range (Maillard & Fournier 1995, Keuling et al. 2008). With these large home ranges, some occurrences in sites close to established populations may only represent temporary movement within the home range rather than actual dispersal.

Based on these results, a decision tree is suggested as a tool that stakeholders could use to determine whether the presumed presence of wild boar in a new area requires confirmation and quantification (Figure 9).

The present study provides a useful baseline for future monitoring of wild boar expansion. The fact that wild boar detection at bait points was higher in spring than in autumn could be due to the availability of natural food that made bait points relatively unattractive in autumn. Future research should evaluate seasonal differences in wild boar detectability at bait points to select the optimum timing of the year to determine whether a new wild boar population has established. At higher levels of site occupancy, analysis of the site-specific variables and sampling-occasion variables could indicate which factors are important in (a) site occupancy and (b) detecting boar. Linkie et al (2007) have shown how this can be done in relation to sun bear Helarctos malayanus site occupancy, revealing useful information on habitat preferences.

Although the use of activity signs might be the less expensive option to confirm the presence of wild boar, the difficulty in recognising these signs for inexperienced staff and the fact that in some seasons activity signs are very scarce indicate that the baited camera trap approach is likely to be the best method to confirm the presence of this species. Baited camera traps therefore represent a useful, relatively inexpensive technique for investigating wild boar dispersal in England.

SID 5 (Rev. 07/10) of 49

Confirm wild boar presence?

YES NO

Activity signs present ?

YES NO

Presence confirmed?

YES NO

Density required?

YES NO

Initial information Presence confirm

ation D

ensity estimation

Map location of RTA or sighting

Stop

Stop

Baited cameras

Camera trap surveys

Stop

Stop

Figure 9. Decision tree of methods to monitor presence and local abundance of wild boar.

Objective 2. Evaluate methods to mitigate the impact of wild boar.

Current trends of human and wild boar population growth and landscape development indicate that conflicts between wild boar (or wild pigs) and human activities are likely to increase worldwide. These conflicts range from damage to crops and to the livestock industry to spread of diseases and vehicle collisions. Wild boar and pigs may also cause reduction in plant and animal abundance and richness, particularly where they occur as non-natives (Hone 2002, Engeman et al. 2003, Massei & Genov 2004). However, where native, wild boar are an important component of the animal community and in most of their range they are hunted as game. Therefore population control or eradication, often advocated by animal health authorities and farmers can be opposed by conservationists, hunters and by the tourist industry.

Non-lethal methods offer a more publicly acceptable solution to manaage human-wild boar conflicts. These methods include diversionary feeding, translocation, fertility control and fencing (Massei et al. 2011). Fencing has been widely employed to exclude wild pigs from crops, pastures and from sites of conservation interest although no review is available to summarise the findings of different studies.

The aim of this part of the project was to review information on the effectiveness of fencing, both electric and conventional, and highlight the most effective designs in terms of efficacy and costs.

MethodsInformation on fencing to exclude wild boar was obtained from published literature and reports,

and from guidelines and specifications produced by government bodies, research agencies and manufacturers. The review focussed on fencing to exclude wild boar from crops or from sites of

SID 5 (Rev. 07/10) of 49

conservation interest. In addition, researchers that investigated fencing to exclude wild boar or wild pigs were contacted by telephone or email and informally interviewed about their experiences. The results of the literature were summarised in a table which reported the aims of each study, specifications of the type of fencing used and the efficacy and the costs of fencing. Where possible, costs were split into materials, labour to set up the fence and maintenance. If more than one fence was studied in an experiment, the least cost-effective fences were excluded from the table. The views of experts and the specifications reported in guidelines or by manufacturers are also reported and discussed below.

ResultsThe review of the literature showed that a very large spectrum of fencing designs is available,

although few studies contain full details on the materials, costs and effectiveness. The most common types of fencing used to deter wild boar and pigs from entering an area are temporary electric, permanent electric and conventional (non-electrified), permanent fencing (Table 10). The differences in the design adopted depend on the size of the area to be protected, on the type of resource and on its temporal availability that determine whether exclusion should be permanent or only temporary. Designs also differ in number and type of strands tested and recommended and in the source of energy supply (batteries, solar-powered or mains).

Figure 10. Summary of results from publications on fencing to exclude wild boar and wild pigs from crops or areas of conservation interest. References as follows: 1= Vidrih & Trdan 2008 ; 2=Reidy et al. 2008; 3= Marsan, in Monaco 2010; 4=Santili & Mazzoni della Stella 2006; 5=Doupe et al. 2010; 6=Hone & Atkinson 1983, labour costs from Choquenot 1996 ; 7=Aviss & Roberts 1994.

The definition of effectiveness and cost varied significantly between studies. Effectiveness was in some instances poorly quantified (“very effective”) or referred to in terms of percentage damage reduction in the protected area. Other Authors pointed out that although the fencing was effective at reducing damage on a site, neighbouring unprotected areas might have suffered heavier losses. The cost of materials was generally reported although the costs of installation and maintenance were rarely quantified. In general, materials and installation of temporary electric fencing were found to be cheaper than those for fixed electric fencing and the cost of maintenance for all electric fencing was higher than that that for conventional, non electrified fencing.

Simple, non electrified fence employed in the exclusion of wild pigs often consist of a fence 110-120 cm high, made using a woven wire mesh 65-80cm high with strands of barbed wire along the top

SID 5 (Rev. 07/10) of 49

and the bottom, buried to a depth of 40-60cm (Tilley 1973, reviewed in Massei et al. 2011). Chain mesh can be used instead of lighter mesh if damage to the fence is frequent (Long & Robley 2004). These fences can also be ‘upgraded’ by electrifiying them. Outrigger wires can be added (Allen 1984) which can act as a deterrent, or the whole fence can be electrified (West et al. 2009, Khan 1990). Prefabricated or wire mesh fence coupled with electric wires have been found very effective against wild pigs (Hone & Atkinson 1983, Long & Robley 2004). For electric fencing, two to three live wires are commonly employed (Venamore and Hamilton 1978, Massei et al. 2011), although in moist conditions a single live wire may be sufficient (Wright 1972). Electric fencing may employ steel wire (Vassant 1994), polywire (Reidy et al. 2008) and a mixture of polytape and polywire (Vidrih & Trdan 2008).

A large number of guidelines on fencing to exclude wild boar and pigs are available, often produced by government agencies, farmer groups or conservation organisations; the most recent guidelines have been published by the Deer Initiative (2009). The most popular type of fencing recommended in guidelines and specifications appears to be conventional mesh fencing coupled with some electric fencing. In the UK, Natural England (2007) suggested that wire netting is unlikely to be effective on its own. Long & Robley (2004) recommended a prefabricated fence with small mesh sizes or wire netting to act as a strong barrier to prevent wild pigs charging through the fence, along with an electric wire at snout level to act as a tripwire for added protection. In Australia, ‘pig proof’ fence is made of netting (a rectangular wire mesh) and two outrigger wires that are pulsed and with earths attached every 500 metres (Caley 1999). In the UK the Forestry Commission recommended a mesh fence (maximum size 20x20 cm) buried 40-60cm deep, with a minimum height of 1.2m and 2 offset electric wires to deter wild boar (Trout & Pepper 2006). For containing wild boar, CALU,The Centre for Alternative Land Use, (2010) suggested a minimum fence height of 1.8m buried 30cm below the ground, with two strands of electric fencing to deter burrowing. However, Natural England (2007) indicated that these fences are not cost-effective and recommended that electric fencing should be preferred. For electric fencing, the number and type of strands recommended varies from one to two strands of smooth wire (McAtee 1939) to two to three wires (Service Public de Wallonie ,Belgium) and up to four to six alternating live and earthed wires (Forestry Commission, Trout and Pepper 2006). McKillop et al. 2003 also suggested alternating live and earth wires for best results.

Other considerations, such as the landscape and environment, must also be taken into account. In dry areas, wild pigs must be sufficiently earthed for an electric shock to be effective. Khan (1990) suggested digging a nearby channel with standing water in it, so the hooves would be fully conductive. In uneven terrain, the fence should follow the contours (Aviss & Roberts 1994), and when planning the position of a fence, factors such as the geography, nearby power lines, and seasonal and climatic influences must be considered (McKillop et al. 2003). The season is also important when protecting crops; for example, maize is highly attractive to wild pigs in the milk stage (Schley et al. 2008).

Many commercial businesses produced manuals and guidelines on electric fencing. The Gallagher website, for example, has a vast range of products and advice on electric fencing to suit different situations. The Gallagher PowerFence™ Manual contains suggestions and useful information on the types of electric fences available. For instance, for the exclusion of wildlife, live earth systems are the preferred design. In France, retailers recommend 3-4 wires at 20, 40 and 60cm from the ground to protect crops from wild boar along with an additional wire 40cm in front, 35cm high. The Patura electric fence catalogue suggests a temporary electric fence for the exclusion of wild pigs with two to three live wires at a height of 0.55m, and states that it is possible for one person to fence an area of one hectare in less than half an hour.

There were contrasting views among experts on the effectiveness of electric fencing to exclude wild pigs. S. Cahill (Spain, pers. comm.) reported that many people are not satisfied with electric fencing because of the high maintenance costs and the requirement for flat terrain, and therefore opted for fixed wire fencing instead. M. Sage (France, pers. comm.) actually observed wild boar crossing through electric fencing in single file: some animals would squeak before crossing the fence, thus indicating they were aware of the imminent shock, although this did not dissuade them from crossing. However, due to the improvement in technical quality and the decrease in the price of materials, other experts believe electric fencing can be effective provided it is well operated (Marsan, Italy pers. comm.; O.Keuling, Germany, pers. comm.). Experts’ recommendations were also given on the types of electric fencing. When protecting crops, it was advised to set up an electric fence around the whole plot, and before the crops were planted so the wild boar wouldn’t associate the area as a food place (D. Villanúa, Spain; pers. comm.; A. Marsan, Italy, pers. Comm.). M. Sage also suggested using white ribbon as the conducting wire so wild boar can see the fence, and therefore associate the electric shock with the fence.

SID 5 (Rev. 07/10) of 49

DiscussionElectric fencing has already been demonstrated as a potentially effective tool to protect crops

against wildlife such as badgers (Meles meles) (Poole et al. 2002) and white-tailed deer (Odocoileus virginianus) (Seamans & VerCauteren 2006). The main advantage of electric fencing over conventional fencing is that it is cheaper to erect and maintain (Campbell et al. 1990, McKillop et al. 2003). The conditioned avoidance response of electric fencing is not immediately lost if the power to the fence fails, and therefore daily checks (which leads to high maintenance costs) may only be essential during the initial learning period or when crops are their most vulnerable stage (McKillop et al. 2003, McKillop & Sibly 1988, Marsan in Monaco et al. 2010). As agricultural fields usually require regular access by large equipment and crops are seasonal resources, temporary fencing has the advantage over permanent fencing as it can be easily moved and reused (Reidy et al. 2008). However, maintenance is a major issue for the cost-effectiveness of electric fencing: regular checks are required to maintain the fence and reinforce the conditioned avoidance response and vegetation must be cleared to prevent the fence being shorted (e.g. McKillop & Sibly 1988, Campbell et al. 1990, Littauer 1993, Marsan in Monaco et al. 2010). Annual maintenance budgets are comprised almost entirely of labour costs, and electric fencing may become too expensive (Long & Robley 2004, Vidrih & Trdan 2008).

Overall, the scientific literature reports that fencing can successfully exclude wild boar and wild pigs although it can also shift the damage to to unprotected areas (Geisser & Reyer 2004). Comparisons between effectiveness and costs of different types of fencing are hampered by poor definitions or specifications. Effectiveness has been defined in terms of total exclusion of wild boar (Vidrih & Trdan 2008) or as the amount of damage reduction (Santili & Stella 2006). Costs are provided in different currencies and for different years and must be re-calculated referring to original exchange rates. Costs are also expressed in different ways : many studies only reported the cost of materials but did not mention installation and maintenance costs which may be a significant factor in determining whether electric fences are cost-effective. Other authors mentioned that maintenance costs can be absorbed by hunters and other volunteers (Santilli & Della Stella 2006, Marsan in Monaco et al. 2010) with the additional benefit that this would also reduce conflicts between farmers and hunters, as population control carried out by farmers is often opposed by recreational hunters.

The decision to fence, and what type of fencing to use, is context-specific and must be based on the size of the area to be protected, the impact of damage on the crop (which depends on the crop value) and the possibility of using alternative control methods. Several authors suggest combining fencing, electric or conventional, with a variety of techniques such as dissuasive feeding, to prevent damage and control wild pig populations (Calenge et al. 2004, Choquenot et al. 1996, West et al. 2010). Due to the high maintenance costs, fencing is not regarded as the best control technique for feral pigs except for enclosing relatively small and valuable areas, such as agricultural crops and stock or when the labour can be provided by volunteers (Long & Robley 2004, Natural England 2007).

In conclusion, many fencing types and designs are available, and it is unlikely that a single one will suit all situations. Further investigation should be carried out to quantify the cost-effectiveness of the different types of electric and conventional fencing. The actual costs should be split into materials, set up and maintenance and the latter should be expressed as man-hours as this indicator will not depend on the year the study was carried out or on currency. Future research should also focus on landscape evaluation to determine whether excluding wild boar from a few sites through fencing will increase the damage in unprotected areas.

Objective 3. Identify methods to evaluate the impact of wild boar on the biodiversity of woodland habitats, and apply these to limited woodland habitats within the scope of this study

Wild boar are omnivorous, but more than 90% of their food comprises plants (e.g. Genov 1981a; Massei et al. 1996; Schley & Roper 2003). In particular, wild boar obtain a large proportion of their diet from rooting for roots, bulbs, fern rhizomes, invertebrate larvae and earthworms (Kotanen 1995; Baubet et al. 2003; Schley & Roper 2003). Rooting creates a disturbed area of mounds and excavations of varying depths (Jones et al. 1994, Arrington et al. 1999). Conflicting evidence exists on the impact of wild boar on plant and animal species (Massei & Genov 2004). The scale, the intensity and the timing of rooting determines its impact on the environment (Kotanen 1995, Hone 2002). Rooting fluctuates in area, depth and frequency and is influenced by precipitation, availability of natural food resources and habitat type (Genov 1981, Hone 1988a, Welander 2000). Feeding opportunistically on crops and on other species of plants, vertebrates and invertebrates, wild boar have a direct impact on the flora and fauna of woodlands and on other habitats such as pastures (Howe et al. 1981; Singer et al. 1984; Kotanen 1995; Schley & Roper 2003). Furthermore, the rooting activity of wild boar may

SID 5 (Rev. 07/10) of 49

http://apps.isiknowledge.com.ezphost.dur.ac.uk/OneClickSearch.do?product=UA&search_mode=OneClickSearch&db_id=&SID=X1kPf5B5MmgbFgdabpN&field=AU&value=VerCauteren%20KC&ut=000237797300003&pos=2

have indirect effects on plants that are not eaten but that may die when uprooted (Massei & Genov 2004). As wild boar are a former native species to the UK, the impact on the biodiversity of woodland habitats should be more similar to that occurring in continental Europe where the species is native and widespread. Nevertheless, as wild boar have been absent for several centuries, they could have a negative impact on some ecosystems.

A three-year study on the ecological impact of wild boar in Sussex (Sims 2005) found that plant cover, diversity and species richness were significantly greater in rooted than in non-rooted areas. The number of individual plants in some species, such as bluebells Hyacinthoides non-scripta, was significantly reduced after rooting but was re-established rapidly in the following years following exclusion of wild boar. In parallel, the frequency of annual and perennial forbs was greater in rooted than in non-rooted plots and the cover of perennial graminoids was lower in rooted than in non-rooted plots. These results were similar to those found by Kotanen (1995) in California & Welander (2000) in Sweden. No other data are available for the UK, particularly on the effect of rooting over a number of consecutive years.

The aims of this part of the project were to develop methods to assess 1. the large-scale impact of wild boar on woodlands; 2. the impact of wild boar on plants and invertebrates and 3. the impact of wild boar on small mammals.

3a. Assessment of large-scale impactThe objectives of this element of the study were to (i) develop an easily-repeatable method to

assess the distribution, extent and timing of rooting in English woodlands; (ii) relate the spatial and temporal aspects of rooting to environmental variables and to wild boar densities, and (iii) compare results of this and other studies on the environmental impact of wild boar. Methods

The study was carried out in two wooded areas of England, where wild boar had been established since 1990 (Sussex Weald) and since 1999 (Forest of Dean). Two woodlands were surveyed in East Sussex: Beckley/Bixley Woods and Brede High Woods. Three woodlands were assessed in and around the Forest of Dean: Penyard/Chase, Serridge and Oakenhill (Table 11).

Wood Latitude & Longitude

Woodland area (ha) where plots established

Minimum and maximum altitude (m)

Number of plots established

Plot density

Beckley/Bixley 500 58/ N 00

38’E283 10-112 114 2.5ha

Brede 500 57/ N 00

33’ E257 4-85 90 2.9ha

Penyard/Chase 510 54/ N 20

34’W201 84-251 64 3.1ha

Serridge 510 49/ N 20

33’W373 37-222 106 3.5ha

Oakenhill 510 45/ N 20

32’W212 31-163 63 3.4ha

Table 11. Characteristics of woods and plots surveyed to assess the large-scale impact of wild boar.

The distance between each adjacent wood was measured by the distance of the closest point of each wood edge to the other: Penyard/Chase – Ruardean: 5.7 km; Ruardean – Oakenhill: 7.2km; Beckley/Bixley – Brede: 3.2 km. These woods are regarded as independent sites, as preliminary data on radio-tracking and on bait uptake showed that wild boar monthly movements in England were generally restricted to individual woods (Quy et al. in prep.).

Wild boar rooting activity was investigated by establishing permanently marked circular plots (10m radius; plot area = 314 m2) in the five woods and recording signs of rooting in subsequent surveys thereafter (Figure 10). Plots were initially uniformly distributed along a grid overlapped to the study area and placed at approximately one every 2.5-3.5 ha (Figure 10). Plots were established in May 2009, geo-referenced using a Garmin eTrex Venture HC (R) unit and the centre of the plot was marked with red tape to facilitate surveying.

For each plot the following data were recorded: slope, categorised as flat, slight (≤100), moderate (≥110≤690) or steep (≥700); elevation, determined from a GPS (Global Positioning System) fix at the centre of the plot; tree species in the immediate vicinity (50m radius) of the plot, categorised as one of six stand types on the basis of species dominance (≥50% of trees in vicinity). Stand types were defined as Oak Quercus spp., Beech, Birch Betula spp., Sweet Chestnut, mixed broadleaf (no

SID 5 (Rev. 07/10) of 49

dominant species) and conifers. The structure of the stand was categorised as standards, coppice, plantation or scrub. The field layer (ground cover) was visually examined and the dominant (≥50% of area) covering recorded. This layer was initially classified into 12 categories: bare, brash, bramble Rubus spp, bluebells Hyacinthoides non-scripta, bracken, fern, forbs, grass, ivy Hedera spp., leaf litter, moss, and pine needles.

Figure 10. Distribution of plots in Beckley Woods.

Plots were surveyed for rooting signs on four occasions, in July and December 2009, and March and July 2010 to quantify temporal and spatial patterns of rooting. On each visit, the field layer was recorded and rooting was assessed by noting the extent of the area (s) covered by rooting, the depth (each discrete area of rooting was measured at five random points to derive a mean depth) and the age (fresh, intermediate, old). Depth was also categorised as shallow (1-<4.9 cm), medium (5-15cm) or deep (>16cm). Age was determined by a modification of Killian’s (2009) categories (Table 12). A sketch map was drawn for each rooted plot, to enable subsequent rooting to be compared. Wild boar densities in each wood were obtained using camera trap surveys as described in Objective 1.

Category 1: fresh ≤ 7 days old

2: intermediate 8-30 days

3: old ≥31 days old

Soil Freshly overturned soil, still moist

Overturned soil shows some signs of weathering

Overturned soil is weathered and beginning to settle

Vegetation Overturned vegetation is still green in colour

Overturned vegetation is yellowing and/or wilted

Overturned vegetation is mostly dead

Roots Exposed roots are still intact and pliable; still alive

Exposed roots are intact but dry

Exposed roots are dry and brittle

Disturbed vegetation

Disturbed vegetation has not resprouted

Disturbed vegetation shows initial signs of regrowth (budding leaves and stems)

Disturbed vegetation shows signs of regrowth (extended shoots and stems)

New vegetation No new plants have sprouted up in areas of bare soil

Seeds may have just sprouted in areas of bare soil

Plants sprouted in areas of bare soil

Table 12. Classification of rooting according to age.

SID 5 (Rev. 07/10) of 49

Statistical tests were performed on SPSS 15.0 for Windows. Chi-square was used to analyse differences in the proportion of plots revisited, the proportion of plots rooted in each survey and the frequency of stand types and field layers utilised. A Generalized Linear Model (GLM) was used to compare the proportion of plots rooted in summer and winter.

The response (rooting / no rooting) at the plot level was binary, so the link function used in the GLM to account for this was a logit function. Due to the relative low number of plots rooted per sampling event and per site, the results were pooled across sites unless otherwise specified. All plots were treated as replicates, irrespective of whether they were from the same year or not, so that the estimates were overall estimates (over both years). A one-way ANOVA was used to compare the mean area of plots affected by rooting and the mean depth of rooting between surveys. A Spearman’s rank correlation was used to relate mean depth with mean area of rooted plots. The elevation of rooted and non-rooted plots was compared with a Mann-Whitney U-test. The proportion of plots rooted in each wood in each survey and the corresponding wild boar density were correlated using Spearman’s rank correlation.

ResultsA total of 437 plots was established in the five woods (range: 63 - 114 plots per wood). Out of

these, 89 (20%) plots were rooted at least once during the study period and 25 (28%) of the plots were rooted more than once during the study. In total, 64 plots were rooted once, 18 were rooted twice, 6 were rooted three times, and one was rooted on all four surveys (based on the age of fresh rooting) (Table 13).

Stand type Number and percentage of rooted plots revisited

Total number of plots rooted

Beech 0 1Birch 1 (11%) 9Conifers 3 (17%) 18Mixed broadleaf 3 (43%) 7Oak 9 (29%) 34Sweet chestnut 7 (40%) 20

Table 13. Number and percentage of plots revisited in each stand type

The percentage of plots revisited also differed between stand types, but this was not significant ((χ2 = 0.445, d.f. = 4, n.s.). There appeared to be a trend (29 - 43% of these plots) for broad-leafed stands (mixed, oak and sweet chestnut) to be revisited more often.

Survey period July 2009 December 2009 March 2010 July 2010

Number and percentage of all plots rooted (n=437) 37 (9%) 17 (4%) 50 (12%) 18 (4%)

Median age of rooting >31 days 8-30 days 8-30 days 8-30 daysMean area (m2) of rooting (± standard error) in rooted plot 64 (± 19) 102 (± 35) 111 (± 19) 82 (± 29)

Median slope of rooted plot Slight Slight Slight SlightMean depth (cm) of rooted plots (± standard error) 8.2 (± 0.8) 7.7 (± 1.3) 6.0 (± 0.8) 8.3 (± 2.9)Mean percentage of the area rooted in rooted plots 21 33 37 18 Table 14. Characteristics of plots rooted by wild boar across all woods between July 2009 and July

2010.

SID 5 (Rev. 07/10) of 49

The number of plots freshly rooted in each survey varied between 17 and 50 (4 - 12%) (Table 14). There were significant differences between the total number of plots rooted in each survey (χ2 = 26.823, d.f. = 3, P <0.01), with July 2009 and March 2010 numbers being comparatively high (Table 14). The number of plots freshly rooted in each wood on each survey varied between 0 - 9 plots for four of the woods, and 9 - 26 plots for Beckley/Bixley (Table 14).

The median age of rooting was more than a month in the first July survey, and less than a month in subsequent surveys (Table 14), indicating that most rooting activity had taken place earlier in the spring before the first survey. Signs of rooting were often indiscernible between one survey and the next, especially if a vegetative field layer had grown in the meantime. The mean area of rooting was between 64 - 111 m2 per rooted plot in the four surveys (Table 14). There were no significant differences in the mean plot area rooted between surveys (F3,118 = 1.007P> 0.05, one-way ANOVA).

Rooted areas varied in size, from 0.02 m2 to the maximum of 314 m2. The mean rooted area of 122 plots was 91.5 m2 ± 127 m2, in contrast to the median rooted area of 8.5 m2 ± 8.23 m2 (median absolute deviation). This indicated rooting was skewed to smaller rather than larger areas. However, 27 (22%) of the plots were in the highest size class (Figure 11). These were plots where extensive very shallow rooting had occurred, usually in oak or sweet chestnut stands. The mean depth of the rooted areas was 6.0 – 8.3 cm in each survey (F 3,118= 0.796, n.s., one-way ANOVA). There were 51 shallow-rooted plots, 61 medium-rooted plots and 10 deep-rooted plots.

Figure 11. Size distribution of rooted plots (m2).

Wood July 2009 December 2009 March 2010 July 2010Number of rooted plots; Total rooted area (m2); % woodland floor rooted

Number of rooted plots; Total rooted area (m2); % woodland floor rooted



Beckley/Bixley 10; 319; <1% 10; 1733; 4% 26; 4204; 12% 9; 419; <1%Brede 3; 323; 1% 0; 0; <1% 6; 1098; 3% 2; 316; <1%Penyard/Chase 9; 1001; 5% 1; 1; <1% 8; 315; 2% 4; 267; 1%Serridge 9; 591; 2% 6; 11; <1% 6; 132; <1% 2; 471; 2%Oakenhill 6; 227; 1% 0; 0; 0% 4; 76; <1% 1; 2; <1%

Estimated percentage of all woodland area rooted

1.89% 1.32% 4.44% 0.72%

Table 15. Characteristics of rooted plots in each wood during each survey and estimated percentage of woodland floor affected by rooting.

The frequency of stand types varied between woods. Conifer plots were less numerous in the Weald compared to the Forest of Dean, where they comprised >40% of the plots (Table 16). Birch plots were more common in Oakenhill, where heathland comprised a large proportion of the wood. Oak

SID 5 (Rev. 07/10) of 49

dominated stands were numerous in Beckley/Bixley and Serridge. As the number of rooted plots per stand type was too small to evaluate preferences within each wood, data from all rooted plots were pooled to enable use and availability of each stand to be compared across woods. Some stand types were more favoured for rooting than others (χ2 = 47.943, d.f. = 5, P <0.001). Compared to their availability, oak and sweet chestnut stands were preferred and conifers avoided (Figure 12).

The 12 field layers were amalgamated into 4 broad categories of interest, after initial analysis (Table 17). Bracken/brambles comprised >40% in 3 of the surveys, except for March 2010, after dieback had occurred. More rooting occurred in bluebell field layers than expected in March 2010 (χ2 = 26.512, d.f. = 3, P<0.01; Table 17; Figure 13). All other comparisons were not significant.

Wood Mixed Broadleaf

Oak Beech Sweet Chestnut

Birch Conifers

Beckley/Bixley 6 37 9 8 12 26Brede 21 18 10 12 12 27Serridge 0 40 10 4 4 42Oakenhill 2 21 5 0 24 48Penyard/Chase 14 11 6 13 2 54

Table 16. The percentage of plots within each stand type in each wood surveyed to determine large-scale impact of wild boar.

Figure 12. Percentage of plots in each stand type compared to percentage of plots rooted in each stand type.

July 2009 December 2009 March 2010 July 2010Field layer Field

layer %

Rooted Plot %

Field layer %

Rooted Plot %

Field layer %

Rooted Plot %

Field layer %

Rooted Plot %

Bracken/brambles 54 50 41 29 24 14 45 33Leaf litter 2 0 15 17 38 48 26 27Bluebells 0 0 0 0 7 22 <1 11Other 44 50 44 54 31 16 28 29

Table 17. The percentage of field layers in each survey’s plots and percentage of rooted plots in each field layer.

SID 5 (Rev. 07/10) of 49

Figure 13. Bluebell wood in the Sussex Weald, April 2009 (left) and rooting in bluebell field layer, June 2009 (right).

Density estimates were available for all wood and survey periods except for Brede and Oakenhill in September 2009. There was no correlation between density and the proportion of plots rooted (Spearman’s rho = -.157; n.s, n =13; Figure 14).

Figure 14. The proportion of plots rooted compared to estimated wild boar densities.

DiscussionThis study is the first to quantify spatial and temporal patterns of wild boar rooting in England.

Rooting was found in 20% of the plots with an average of 91m2 (29%) of the plot area rooted. By extrapolating this to each area covered by a woodland, then rooting by wild boar in English woodlands, at current local densities, affected between <1 – 2%, and exceptionally ~12% of a wood, depending on season. Boar preferred to root in specific stand (oak, sweet chestnut) and field layer (bluebells) types. There was also a trend for favoured sites to be revisited. The densities of wild boar found in UK woodlands are relatively low compared to those reported in continental Europe where they range from 2 – 7 animals/km² (Smiet et al. 1979; Ickes 2001; Pihal et al. 2010) to 51 animals/km² (Franzetti et al. 2010).

In Australia Hone (1988a) found ~2.7% of his study area rooted, in Poland ~4% of the study area was rooted (Jezierski & Myrcha 1975) in Sweden ~12% (Welander 2000) and in Hawaii ~10% (Ralph & Maxwell 1984) and 14 - 38% (Cooray & Mueller-Dumbois 1981). Bratton (in Howe et al. 1981) stated that as much as 80% of the surface area of mesic northern hardwoods (USA) was rooted annually. Sites were also rooted as many as 3 - 7 times during the growing season (Howe et al. 1981). Welander (2000) suggested that a six-fold increase in rooted area in one year in his study was dependent on food availability, particularly a high abundance of acorns and hazel nuts.

Although Belden & Pelton (1975) and Andersson & Stone (1993) found that the extent of rooting and population density are correlated, Hone (1988a, 1988b, 1995) showed that the relationship is not linear. There were large differences in rooting intensity between December 2009 and March 2010 in

SID 5 (Rev. 07/10) of 49

our study, when population size was likely to have remained the same as births are mainly occurring in late spring (Coats pers. comm.). An analysis of estimated density against the proportion of plots rooted at the same period also revealed no correlation. The study thus confirmed that rooting intensity is not correlated with population size, although the presence of rooting can be used to indicate the presence of wild boar in an area, especially in early spring.

Wild boar populations in the Weald and Forest of Dean were probably too low to directly affect large areas of woodland, with only an area of 1 - 2% being rooted in most seasons. Hone (1988a) had densities of <1 animal/km2 with the area rooted being ~2.7%. Therefore, even low densities of wild boar may impact on areas of woodland, and as they have preferences for certain areas (e.g. bluebells) they may have a disproportionate impact relative to their numbers.

Rooting may be divided into three types: superficial litter rooting, organic topsoil rooting and mineral soil rooting (Groot Bruinderink & Hazebroek 1996, Welander 2000). The depth classification showed that most rooting was of the first two types. Forty per cent of the rooting in this study was superficial, probably due to boar searching for mast.

In most woods there were no apparent seasonal (summer v winter) patterns, though this is probably a reflection of the small number of plots rooted in each wood on each survey. Beckley/Bixley did have a significant difference in rooting intensity between summer and winter, but December and March rooting patterns were clearly different within this winter period. As surveys were carried out at intervals of 2 - 4 months, precise temporal patterns were not evident. In July 2009 most rooting was older than a month, indicating that rooting had occurred during the late spring period, rather than concurrently in the summer. Welander (2000) carried out his surveys in April, June and August, and considered rooting to be most prevalent between mid-autumn to early spring. He considered rooting to be detectable for several years in forests, whereas in our study rooting signs soon became obscured. Rooting did not occur at a greater level in the autumn in this study, probably because mast was still available above-ground.

Temporal patterns of rooting have been found in most wild boar studies (Bratton 1975; Dardaillion 1987; Kotanen 1995; Focardi et al. 2000; Welander 2000; Baubet et al. 2003). Seasonal patterns of rooting vary according to a combination of food availability, habitat type and accessibility of below-ground resources. Schley & Roper (2003) reviewed wild boar diet studies and concluded that the availability of mast (acorns, beechnuts and sweet chestnuts) determines the consumption of other plant components. When energy-rich food such as mast becomes scarce, wild boar must rely on other, less preferred sources (Massei et al. 1996). In December mast is still readily available, but later in the year wild boar must root to find mast. In this respect, the preference of wild boar in this study to root in oak and sweet chestnut dominated stands is especially noteworthy. Conifers were largely avoided for rooting in this study. Preferred food may not have been available in more northerly deciduous forests at the end of the winter. In the current study, it was unclear why so little rooting took place in beech-dominated stands, but perhaps this was during a poor beech mast year (Groot Bruinderink & Hazebroek 1996). Keuling (2010) found that beech stands were avoided in winter, but preferred in autumn, possibly because of mast availability.

In spring bluebell sites were especially targeted by rooting as also reported by Goulding & Smith (1998), Goulding (2003) and Sims (2005). By summer, above-ground sources are available, including crops on adjacent farmland. Though agricultural land was not surveyed during this study, rooting was noticeable in spring on grassland adjacent to some woods and has been noted previously in England (Wilson 2004; Sims 2005).

Sims (2005) found summer was the period of lowest rooting intensity in the Weald, as the ground was hard and impenetrable. This was also shown by the current study, notwithstanding the initial July 2009 survey, when most rooting appeared to have been actually carried out earlier in the spring. An overall reduction in rooting from a peak in early spring through to early autumn has been generally noted (Wood & Roark 1980; Massei et al. 1996). Thurfjell et al. (2009) found that boar prefer to feed close to forest edges, and utilise ripe crops during the summer, preferring these to artificial feeding stations in forest. Wilson (2004) also found boar preferred fields adjacent to woodland, and that most damage occurred during January-March. In future surveys, placing sampling plots in adjacent farmland may clarify spatial and temporal patterns of agricultural landscape rooting.

Several studies reported that wild boar preferred rooting in some stand types (Wood & Roark 1980; Welander 2000, Keuling 2010). Bratton (1975) considered that there was no direct relationship between stand type and rooting, rather that the understory (field layer) is the main factor in rooting intensity. Mitchell et al. (2007) defined rooting areas in terms of microhabitats (swamp, track, ridge, forest, creek, road) and found swamps were the most adversely affected, with over 80% rooted by pigs in their 18-month study period. Wet areas have also been found to be preferred by boar in the USA

SID 5 (Rev. 07/10) of 49

(Bratton et al. 1982), France (Dardailion 1987) and Sweden (Welander 2000). This study found correlations between some stand types and rooting and between some field layer types and rooting frequency. Rooting was also noticeable along many forest rides (pers. obs.), but no plots happened to coincide with their occurrence.

A quarter of the plots were re-rooted during the study. This habit of revisiting favoured areas is well-known in wild boar and re-rooting may reduce plant cover if it re-occurs in the same area (Alexiou 1983; Groot Bruinderink & Hazebroek 1996; Welander 2000; Sims 2005). The majority of the plots were not re-visited, although those re-rooted occurred in more sensitive areas such as the bluebells. Disturbance of these areas resulted in some functional groups of plants, such as annual and perennial forbs being more favoured to re-establish compared to perennial graminoids (Kotanen 1995; Sims 2005). Bluebells are also affected by rooting, with abundance declining substantially in rooted areas, and prevention of re-rooting enabled bluebell populations to recover (Sims 2005). The effect of rooting on bluebells was investigated under session 3b below.

3b. Assessment of impacts on plant and invertebrate biodiversityThe main aim of this part of the study was to develop methods to assess the impact of wild boar

on plants, invertebrates and small mammals in English woodlands.

MethodsInvertebratesThe impact of wild boar on the floral and invertebrate diversity of woodlands was investigated

through vegetation surveys and sampling for invertebrates in woodland sites in East Sussex. Pilot studies were carried out in 2009 in 3 sites: one with no signs of boar activity (Rowland wood) and two with signs of wild boar rooting (Flatropers Wood (south) and Beckley Wood). At each site invertebrates were collected fortnightly from April to the end of July, in 6 flight interception traps and 15 pitfall traps. One flight interception trap was placed at either end of a 40m transect along which 5 pitfall traps were placed at 10m intervals. Samples were identified to order with the exception of bark beetles (Scolytidae), longhorn beetles (Cerambycidae), and ground beetles (Carabidae), which were identified to species.

Results from the interception traps showed that sample sizes were very large and variable with no evidence of differences between rooted and un-rooted areas within or between sites. Fewer Scolytids were found than expected, possibly due to the young age of trees at Beckley (e.g. 50yr oak) and although anticipated, there were no hoverflies (Syrphidae) amongst the Diptera. Pitfall samples were more consistent and indicated differences (although not significant) between rooted and non-rooted areas. Generally, pitfall traps were more cost effective than interception traps and allowed investigation at a range of levels: within and between habitats within a wood, and between woodlands.

The study was extended in March 2010 to include 12 sites in East Sussex (Figure 15), selected to include woodlands with different amounts of rooting activity both within and between sites.

Figure 1. Location of the study sites in East Sussex. Figure 15. Location of the study sites in East Sussex.

SID 5 (Rev. 07/10) of 49

The sites comprised stands of broadleaved trees the majority of which were generally unmanaged coppice growing on clay soils typical of the area. Invertebrates active on the ground surface and in the litter layer were sampled using 10 pitfall traps per site, arranged in two transects of 5 traps placed at 10 m intervals (Figure 16).

Where possible, within each site, transects were located in areas of differing rooting activity. Transects were established in early April 2010 with transect 1 laid out to sample across the cline between rooted and unrooted areas; transect 2 was laid out in a similar way in an area of different rooting activity (Figure 16). This “activity centred” arrangement maximised the chances of detecting an influence of boar rooting on invertebrates at both small (between traps) and medium scales (between transects) within woods.

Figure16. Schematic representation of transects with invertebrate traps and quadrats used to assess the impact of wild boar on plant and invertebrate species.

Figure 17. Armoured pitfall trap with lid in place to prevent wild boar disturbing the collection of invertebrates.

SID 5 (Rev. 07/10) of 49

In intensively rooted areas, specially designed, armoured pitfall traps were used to prevent disturbance by boar (Figure 17). The traps were filled with 50 ml of 50% ethanediol (antifreeze), which acts as a preservative and mild fixative, and they were emptied and refilled every two weeks between 13th April and 4th August. Invertebrates caught in the traps were stored in 70% industrial ethanol before sorting and counting. The number of individuals in each of the main invertebrate groups was counted and recorded, and ground beetles were identified to species. Ground beetles are a well-studied group which has been widely used as an indicator in studies of biodiversity; there are very good keys for British ground beetles and identification is relatively easy (Luff 2007).

Vegetation Two methods were used to assess boar impacts on woodland vegetation:1. vegetation around

pitfall traps and 2. stand characteristics and bluebells. For the “vegetation around pitfall traps” method, the abundance of vascular plant species was assessed in triplets of 2 x 2 m quadrats located at 10 m intervals along the transects used for pitfall traps (Figure 16). The pitfall traps were located in the central quadrat of each triplet. An additional triplet at the 50 m distance was used as a control to investigate whether the work associated with the pitfall trap had an effect on the vegetation assessments. Assessments were made in May and August to record the spring and summer floras. Cover of each plant species was scored as follows: 0, ≤ 3%, 4-10%, 11-25%, 26-50%, 51-75%, 76-100%. During August the presence of bluebells was recorded by observation of dead inflorescences and the amount of rooting, fresh disturbance, bare ground, vascular plants, bryophytes and over / understorey cover were recorded using the same 7-point scale.

The “stand characteristics and bluebells” method was used in May 2010, when bluebells were in full leaf, and their abundance was assessed by a broad survey of the woodland surrounding the pitfall traps. Thirty temporary 4 x 4 m quadrats placed at 30 m intervals along parallel transects 30 m apart were located by pacing along compass bearings. The length of the transects varied depending on the shape of the stand. All quadrats were 15 m or more from the woodland edge. If a quadrat fell on a footpath then it was moved further along the transect until it was at least 1 m away.

The percentage of each quadrat covered by bluebells was assessed using the following scale: 0, ≤ 3%, 4 ≤ 10%, 11 ≤ 30%, 31 ≤ 50%, 51 ≤ 100%. A simpler four point index was used to assess bramble, bracken, grasses and sedges, rushes, ericaceous species, shrubs (< 2 m tall) and all other herbaceous vascular plants (including ferns and bluebells).These were recorded as: none; rare = few isolated individuals; common = small amounts throughout; abundant = everywhere, obviously plentiful.The dominant species of tree or shrub in the overstorey and understorey was recorded within a distance of 10 m from the centre of each quadrat. Where several species were present and no individual species comprised > 30% of the cover the canopy was recorded as mixed. The current structure of the stand in the vicinity of the quadrat was recorded as high forest or some form of neglected coppice which were classified as: stored coppice (i.e. stools with large diameter stems, high forest in character, no standards); neglected coppice with standards; neglected coppice (i.e. small stems/stools, understorey only).

Rooting activityExtensive transect surveys beneath tree canopies in 2009 found that rooting activity was

patchily distributed, took place mostly in winter/spring and that little ground flora was present during summer. Rooting activity was assessed as follows in 2010:

(i) Rooting around pitfall traps. The rooting within the quadrats in which each pitfall trap was located was assessed on the 1st, 3rd, 7th and 8th collections. The presence of rooting was assessed using the categories: 0 = no sign of rooting or disturbance; 1 = possible old rooting, but uncertain; 2 = obvious existing rooting; 3 = apparently recent rooting.

(ii) Rooting in vegetation quadrats. During August the percentage area of old rooting, fresh rooting or disturbance was recorded for each quadrat in all triplets along the pitfall trap transects using the same class intervals as the vegetation around the traps.

(iii) Rooting associated with bluebells. Disturbance caused by rooting in the 4 x 4 m bluebell quadrats was assessed using the same scale as for bluebells. Rooting activity included that restricted to the litter layer as well as disturbance to the mineral soil. Only recent rooting activity that was obvious without extensive searching beneath the leafy cover of the ground flora was included. Recent activity was determined by its fresh unweathered appearance and the absence of established plant cover including bryophytes. The overall amount of disturbance on the plots was also recorded – including that due to rabbits, moles and unknown causes.

Results and Discussion

SID 5 (Rev. 07/10) of 49

InvertebratesLarge numbers of ground-dwelling invertebrates were caught of which 67,449 were identified to

species, family or order (Table 18). Overall, ants were the most common group (36% of individuals caught) and the majority of these were wood ants (Formica rufa), which occurred in very high numbers at one site.

Invertebrate group Order/family Count %Ants Formicidae 24,614 36.5Beetles: Coleoptera ground beetles Carabidae 12629 18.7 rove beetles Staphylinidae 4044 6.0 other adult beetles 3593 5.3 beetle larvae 2062 3.1Centipedes Chilopoda 1102 1.6Harvestmen Opilionida 5937 8.8Millipedes Diplopoda 2871 4.3Spiders Araneae 5630 8.3Woodlice Isopoda 4967 7.4Total 67,449 100

Table 18. Invertebrate groups caught by pitfall trapping at the woodland study sites between April and August 2010.

Beetles were the next most abundant group and more than half of these (12629) were ground beetles, representing 33 species. Between 10 and 20 ground beetle species were found at each site, but only the 4 most abundant species occurred at all sites: Abax parallelepipedus (67.3% of total individuals), Pterostichus madidus (20.4%), Carabus nemoralis (4.2%) and Nebria brevicollis (2.5%). Abax parallelepipedus is a large black beetle, 18-25 mm long, and is a typical generalist predator found in a range of habitats including woodlands, scrub and hedgerows.

The total numbers of ground beetles caught varied considerably between sites and at sites without wood ants there was a positive correlation between the total numbers captured and the percentage of the site that was rooted (Figure 18a; sites without wood ants: regression R2=0.87, P<0.001, N=10). Ground beetle numbers were particularly low at the two sites where wood ants occurred (21,770 wood ants were caught at Flatropers Wood (south) and 7 were caught at Flatropers Wood (north). However there were reasons to suspect there might have been higher numbers of wood ants at Flatropers Wood (north) in the recent past, but these had declined due to removal of an adjacent stand of conifers during the previous winter. At both of these sites ground beetle numbers were particularly low and it seems highly likely that this was due to the presence of wood ants.

Wood ants are well known to interfere with the foraging behaviour of ground beetles and other studies indicate that ground beetle populations are reduced where wood ants are abundant (Hawes et al. 2002; Dorosheva and Reznikova 2006). Ants were also caught at the other sites, but in all cases these belonged to two much smaller species, Stenamma westwoodi and Myrmica ruginodis (Bolton & Collingwood 1975) which are common in woodland but have no effect on ground beetle numbers.

The total number of ground beetle species (S) was not related to the amount of rooting or the presence of wood ants (Figure 18b).Ground beetle diversity measured by the Shannon-Weiner index (H’) or Simpson-Yule index (D) showed no relationship with either percentage rooting or the presence of wood ants.

The total number of ground beetles caught was also related to the percentage cover of bluebells (Regression R2=0.39; P=0.040, N=12), mainly due to higher numbers of Pterostichus madidus and P. niger in areas with greater bluebell cover. Denser vegetation, in this case bluebells, and a deeper litter layer provides a greater food supply for ground beetles and a refuge from predators (including other ground beetles), and reduces interference competition, all of which can result in higher population densities (Hawes et al., 2003; Brose, 2003; Phillips & Cobb, 2005).

SID 5 (Rev. 07/10) of 49

(a)

0

50

100

150

200

250

300

0 10 20 30 40 50

% rooting

Tota

l car

abid

s pe

r tra

p (9

5%C

L)

(b)

0

5

10

15

20

25

0 10 20 30 40 50

% rooting

Tota

l car

abid

spe

cies

(S)

Figure 18. The total number of ground beetles (carabids) (a) and the number of ground beetle species (b) sampled per trap at each of the 12 sites plotted against the percentage of the site rooted by boar. Open circle = Flatropers Wood (south) where there were large numbers of wood ants; grey circle = Flatropers Wood (north) where there were small numbers of wood ants; black circles = sites without wood ants.

Together the percent site rooted and the percent cover of bluebells explained 69% of the variation between sites in the total numbers of ground beetles caught (GLMM; P=0.001). This relationship was driven primarily by the numbers of Abax parallelepipedus, the most abundant species, but the total numbers of all other ground beetle species combined (i.e. without Abax) were also positively correlated with percentage rooting (Figure 19a) and cover of bluebells (regression R2=0.37; P=0.047) suggesting a more general relationship.

(a)

0

20

40

60

80

100

120

140

0 10 20 30 40 50

% rooting

Tota

l oth

er b

eetle

s pe

r tra

p (+

95%

CL)

(b)

0

20

40

60

80

100

120

140

0 10 20 30 40 50

% rooting

Tota

l woo

dlic

e pe

r tra

p(+

95%

CL)

Figure 19. Mean total number of other ground beetles (excluding Abax) (a) and woodlice (b) caught per trap at each site plotted against the percentage of the site rooted. Open & grey circles = sites with wood ants; black circles = sites without wood ants.

Amongst the other invertebrates groups only woodlice (Isopoda) showed a relationship with rooting with highest numbers captured at sites where there was only a small amount of rooting (Figure 19b). However, this relationship just failed to reach significance at the site (GLMM; P=0.070) and trap (GLMM; P=0.078) levels. There was no relationship between woodlice numbers and cover of bluebells or other vegetation.

Given the relatively low frequency of rooting at study sites it is unlikely that rooting increased ground beetle numbers directly. It is probably more correct to interpret the data as indicating that rooting was more frequent in woods where there were already higher numbers of ground beetles, for a variety of other reasons (Figure 20a).Thus large numbers of ground beetles (and perhaps wood ants) seem to indicate a richer habitat more likely to be foraged by boar. In contrast, a higher cover of bluebells, although good for certain ground beetles, was a relatively poor indicator of where rooting was likely to take place (Figure 20b).

SID 5 (Rev. 07/10) of 49

Figure 20. The percentage of each site rooted by boar plotted against the total numbers of ground beetles caught at the site (a) and the percentage cover of bluebells (b). Open & grey circles = sites with wood ants; black circles = sites without wood ants.

Despite establishing pitfall transects 1 and 2 in areas with high and low rooting within the same wood and positioning traps 1–5 on transect 1 along a gradient from rooted to non-rooted areas, there were no significant differences in total ground beetle numbers between transects 1 and 2 (ANOVA, P=0.34) or between traps 1–5 (ANOVA: P=0.49). Similarly there were no significant differences between transects 1 and 2 or traps 1–5 for individual ground beetle species, other types of beetles or other invertebrate groups. Consequently, this analysis did not detect any small or medium-scale associations between ground beetle numbers and rooting within sites and the same was true for other beetles and other invertebrate groups.

VegetationAll sites comprised stands of broadleaved trees most of which had been previously managed as

coppice or coppice with standards. They had not been recently cut and had a stored coppice or high forest structure (Table 19). A total of 14 tree species were recorded: oak and sweet chestnut was predominant in the overstorey, and hornbeam and sweet chestnut in the understorey.

Site Predominant Species Structure Rooting Bluebells

Overstorey Understorey Freq. %cover

Mill Wood - north OK / SWC HBM SC / nCS 3 0.83 45Mill Wood - south OK / SWC SWC SC / nCS 29 1.00 31Flatropers Wood - north

SWC SWC SC / HF 280.73 20

Flatropers Wood - south

OK HAZ HF 400.97 21

Beckley Wood OK HBM HF 1 0.77 9Rowland Wood OK HBM HF <1 0.90 2Burnthouse Wood OK HBM nCS 3 0.20 2Long Sowdens Wood BI HBM HF 3 0.67 15Twist Wood OK HBM / SWC nCS 6 0.93 35Maitland Plantation SWC SWC SC 11 1.00 38Coneyburrow Woods OK HBM nCS <1 0.57 17Rafters Wood OK HBM / SWC nCS / SC 5 1.00 30

Table 19. Vegetation and rooting characteristics of the study sites.Species: BI = birch; HBM = hornbeam; OK = oak; SWC = sweet chestnut; HAZ = hazel nut. Structure: HF = high forest; nCS = neglected coppice with standards; SC = stored coppice. Rooting = Total percentage of site rooted - estimated from survey. Freq. = Proportion of quadrats with bluebells.

SID 5 (Rev. 07/10) of 49

(a)

0

10

20

30

40

50

0 50 100 150 200 250 300

Total carabids per trap

% ro

otin

g(b)

0

10

20

30

40

50

0 10 20 30 40 50 60

% bluebells

% ro

otin

g

During May the ground flora of all woodlands was dominated by herbs with 78% of quadrats having abundant herb cover (mainly bluebell and anemone which senesce during June); only 3% had no herb cover. Ground flora during summer at all sites was sparse. Although other common woodland species such as bramble, bracken and shrubs were found at most sites, they were usually recorded on <50% of quadrats and were generally rare. Grasses and rushes were very uncommon being present on few quadrats at few sites, which probably reflects the deep shade cast by the over- and understorey canopy. Whilst ericaceous species were seen at some sites they were never recorded in the quadrats studied.

The detailed vegetation assessments made in May in the triplets along the pitfall transects reflected those of the broad survey: only 1% of quadrats had no vascular plant cover and over all sites mean cover was c. 50%. By August vascular plant cover had declined to <10% with about 90% of the forest floor being devoid of vegetation – including bryophytes.

A total of 50 plant species were recorded within the quadrats along the pitfall transects, the total number within woodlands varying from 12-27 (Table 20). There were generally fewer species in May than August. Most species were uncommon, occurring in <25% of the woodlands, at low frequencies with very low cover scores. Bluebell and anemone were the most common species having the greatest frequencies and mean cover scores (53.5% and 56.9% respectively). Other relatively common species included Dryopteris dilitata, Lonicera periclymenum, Pteridium aquilinum and Rubus fruticosus. Seedlings of Carpinus betulus, Castanea sativa and Fraxinus excelsior were also relatively common, but all were very small and contributed little to vegetation cover.

Site Number of speciesMay August Total

Mill Wood - north 9 15 16Mill Wood - south 8 11 12Flatropers Wood - north 9 14 15Flatropers Wood - south 11 24 27Beckley Wood 13 22 24Rowland Wood 18 17 21Burnthouse Wood 14 7 16Long Sowdens Wood 10 12 19Twist Wood 13 11 14Maitland Plantation 14 13 16Coneyburrow Woods 20 22 25Rafters Wood 10 12 15

Table 20. Number of plant species recorded in the quadrat triplets along the pitfall transects in spring and summer.

Rooting activityExtensive recent rooting activity was obvious during spring, hence differentiation between

recently rooted and other areas was easy. Distinguishing between recent and old rooting became progressively more difficult during summer. Although c. 30% of quadrats around the pitfall traps showed signs of fresh disturbance in August, boar were definitely the cause in only one quadrat and most disturbance in summer was caused by rabbits.

There was great variation between and within sites in the amount and distribution of rooting (Table 21). At some sites less than 20% of quadrats showed signs of recent rooting, whereas at Flatropers Wood (south) every quadrat had some recent rooting: over all sites about half of the quadrats were rooted. The area of each quadrat rooted also varied considerably, on some the area rooted was only 3% of the total (c. 0.7 x 0.7 m) but on others more than 30% of the quadrat was rooted (i.e. > 2 x 2 m). The estimated areas of each site rooted vary from <1% to 40%; for the majority of sites recent rooting had occurred over <10% of the area.

SID 5 (Rev. 07/10) of 49

Site Frequency % rootedArea (ha) %site

Mill Wood - north 14 47 6 2.73Mill Wood - south 26 87 33 28.92Flatropers Wood - north 25 83 33 27.78Flatropers Wood - south 30 100 39 39.47Beckley Wood 10 33 3 1.10Rowland Wood 4 13 3 0.40Burnthouse Wood 12 40 8 3.05Long Sowdens Wood 13 43 6 2.68Twist Wood 10 33 19 6.42Maitland Plantation 19 63 17 10.67Coneyburrow Woods 5 17 3 0.45Rafters Wood 18 60 8 4.58

Table 21. Recent rooting recorded during May 2010. Frequency = number of quadrats with recent rooting (max = 30). %rooted = percentage of quadrats with recent rooting. Area = for each quadrat with rooting damage the proportion of the quadrat affected (expressed as a percentage). %site = Overall percentage of the site rooted.

The relationship between stand structure and the amount of rooting recorded is shown in Table 22. There was an approximately even distribution of quadrats between high forest, neglected coppice with standards and stored coppice. Neglected coppice comprising trees of small stature with no overstorey component was a rare stand structure occurring on about 2% of quadrats in total.

Stand structure%rooting nCS SC HF nC Total0 16 13 19 <1 482 7 6 6 <1 198 4 3 2 <1 920 4 5 3 <1 1240 1 1 1 0 375 4 2 2 0 8Total 37 29 32 2 100

Table 22. Percentage of quadrats with each stand structure that had different amounts of rootingStructure: HF = high forest; nCS = neglected coppice with standards; SC = stored coppice; nC =

neglected coppice. %rooting = percentage of each quadrat rooted.

Over all stand structures 48% of quadrats had no rooting and most rooting was recorded below neglected coppice with standards. GLM analysis showed a marginally significant difference (p<0.05) in the probability of rooting in neglected coppice compared with high forest, and that rooting was not significantly related to understorey species.

Oak and sweet chestnut were the major overstorey species occurring on about half and a quarter of the quadrats respectively, and greater amounts of rooting were recorded in quadrats with these species (Table 23). The probability of rooting was significantly greater where oak and sweet chestnut were dominant overstorey species (p <0.05). Overall results suggest that rooting is related to stand structure and species present in the overstorey.

The frequency of rooting recorded around the pitfall traps was generally low with most quadrats being undisturbed by rooting on all 4 occasions (Table 24). The average frequency of rooting along the transects at each site was calculated in order to investigate whether the rooting activity immediately adjacent to the pitfall traps was consistent with that in the woodland overall. As most quadrats had no rooting and its presence on a large number was uncertain, the frequency of rooting was calculated by adding classes 2 and 3 (as defined on page 33 under “Rooting activity”) for each observation date and taking the average of these for each site over the 4 observations. There was a clear relationship between rooting along transects and rooting within the wood as a whole indicating that any effects of

SID 5 (Rev. 07/10) of 49

rooting found for vegetation and or invertebrates along the transects are likely to occur across the woodland in general.

Overstorey species%rooting BI OK SWC Other Total0 6 26 8 9 482 2 9 5 3 198 1 5 2 1 920 1 5 4 1 1240 0 2 1 0 375 1 4 2 1 8Total 11 51 22 16 100

Table 23. Percentage of quadrats with various dominant overstorey species that had different amounts of rooting Species: BI = birch; OK = oak; SWC = sweet chestnut; Other = other infrequent species.

Wild boar tend to root the same areas repeatedly (e.g. Alexiou 1983; Groot Bruinderink and Hazebroek 1996; Welander 2000; Sims 2005), but the frequency of rooting at any point can only be determined by repeated observations over several years. An 18 month pilot study conducted on 5 different rides in a Beckley Wood during the current project found that one ride was extensively rooted during spring of two consecutive years whereas the others suffered little damage. The areas of recent rooting at the sites studied were generally low: if this is representative of the rooting that takes place in any year and any repeated rooting is restricted to these areas, then the amount of disturbance caused by rooting could be low. However, more data should be collected to determine whether the observed patterns is typical for each wood.

Collection Rooting categoryNone Uncertain Old Recent

1st 48 13 31 82nd 34 22 40 47th 53 21 22 58th 42 14 32 13

Table 24. Amount of rooting recorded in each quadrat around pitfall traps on 4 occasions during summer 2010. Data are the percentage of quadrats in each category over all sites.

Bluebells

The study was carried out at the height of the flowering season well before the leaves had begun to senesce. Consequently the presence of bluebells is unlikely to have been missed and the cover recorded will probably be the maximum possible in the year. There were large differences within and between sites in the amount and distribution of bluebells, and the bluebell cover in each quadrat (Table 25). Although some quadrats had >50% cover, mean cover varied between 2 and 45%. There was a significant association between the presence of rooting and bluebells within a quadrat (Table 26), but this was only of marginal significance (p=0.044) and was of minor importance relative to interactions with site.

SID 5 (Rev. 07/10) of 49

Site Frequency %with bluebells %cover bluebellsMill Wood - north 25 83 45Mill Wood - south 30 100 31Flatropers Wood - north 22 73 20Flatropers Wood - south 29 97 21Beckley Wood 23 77 9Rowland Wood 27 90 2Burnthouse Wood 6 20 2Long Sowdens Wood 20 67 15Twist Wood 28 93 35Maitland Plantation 30 100 38Coneyburrow Woods 17 57 17Rafters Wood 30 100 30

_______________________________________________________________________

Table 25. Bluebell presence in woodlands. Frequency = number of quadrats with bluebells (max = 30).%with bluebells = percentage of quadrats with bluebells. %cover bluebells = mean percentage cover of bluebells on quadrats with bluebells present

%cover class of bluebells

%rooting 0 ≤3 4≤10 11≤30 31≤50 51≤100

0 67% 57% 53% 30% 31% 42%

1.5 14% 17% 15% 17% 24% 36%

7.5 10% 10% 15% 9% 7% 7%

20 8% 12% 3% 19% 16% 9%

40 0% 2% 12% 4% 2% 4%

75 1% 2% 3% 21% 20% 2%

Table 26. Percentage of quadrats with different amounts of bluebells that have different amounts of rooting. %rooting = percentage of each quadrat rooted.

Conclusions The study indicated that there is evidence for some significant associations between rooting

and woodland structure and species,and rooting and the presence of bluebells. Although these associations are weak, they are consistent with the expectation that rooting is likely to be greater in areas with overstorey trees such a oak and sweet chestnut which produce large fruits, and the concept that rooting will have an adverse effect on bluebells. The great variation between and within sites and the fact that the study was carried out over a short term prevented further conclusions to be drawn.

There was a positive association between total numbers of ground beetles and % rooting at the site level, probably reflecting a tendency for rooting to occur more frequently in woods where large numbers of ground beetles are already present. In contrast, the species richness and diversity of the ground beetle community showed no relationship with the amount of rooting. Other beetles and invertebrates generally showed no association with rooting: there was no evidence for a positive effect of rooting but equally no evidence for a negative effect. The only exception was woodlice, which tended to occur in greater numbers at sites where the frequency of rooting was low.

SID 5 (Rev. 07/10) of 49

3c. Impact on small mammals Rooting by wild boar could have a negative impact on ground-dwelling small mammals due to

direct predation, competition for seeds and invertebrates and disturbance of nests and tunnel systems built just below the soil surface (Singer et al. 1984, Focardi et al. 2000, Massei and Genov 2004). This study was carried out to develop methods to assess whether wild boar rooting affects small mammals’ activity.

Methods Three trials were conducted in November 2009, January 2010 and May 2010. For each trial, two independent trap sites were set up in Penyard woods : a control site, with no evidence of rooting and a treated site with signs of recent rooting. Sites were selected to be in deciduous woodland of similar tree species and similar coverage of ground vegetation but at least 150m apart to ensure independence. At each site, 50 Longworth traps (live traps) were set out in a 10 x 35m grid. Traps were placed in pairs, 5m apart from the next pair of traps. At each site trapping was carried out for 4 nights. Each trap was supplied with dry nesting material (hay) and food (wheat grain and mealworms or castors for shrews), replenished when necessary. The traps were checked twice daily, in the morning and in the afternoon. When an animal was trapped, the date, trap number and species were recorded. Animals were fur marked using fur clipping scissors to expose the dark under-fur and identify re-trapped individuals. Each animal was released at the point of capture and the trap re-set and re-stocked.

The impact of long term presence of wild boar on dormice (Muscardinus avellanarius) was investigated by placing, in early spring 2009, 50 dormice nest-boxes each in 12 woods known to have wild boar present and monitoring dormouse presence in June and October for 2 years. Results were compared with data from 12 similar woods (in terms of soil and woodland type) within the national dormouse monitoring scheme that had no wild boar. Data were recorded on number of sites in each group with positive evidence of dormice, number of nest boxes with evidence of dormice and number of dormice or empty nests found.

Results and discussionThe first trial took place between the 9th and the 13th of November. Only small numbers of small

mammals were caught in both sites: 6 wood mice (Apodemus sylvaticus) were caught in the non-rooted site and 4 mice (one yellow-necked mouse, Apodemus flavicollis and 3 wood mice) were caught in the rooted site.

To increase trapping success, the sampling method was modified by adding three ‘pre-bait’ night period before the trial, where the traps were fully stocked with food and bedding but the doors were locked so that the traps could not be triggered. This was intended to make animals more likely to come across the traps and to reduce neophobia. The subsequent trial, in January, was conducted as the November trial for 4 nights.

The second trial took place between the 18th and the 26th of January, a week after a heavy snow fall had thawed out. Trapped species included wood mice and yellow-necked mice as well as bank voles (Myodes glareolus) and shrew (Sorex spp.). The total numbers of individuals caught at each site were expressed as animals caught per trap per hundred trap nights (Figure 21).

Although the trapping success was higher than in November, the relative small numbers of mammals made it impossible to compare the differences statistically. However, the data indicated that there was no difference between number of individuals of Apodemus spp. (wood mice and yellow-necked mice) caught in the rooted and non-rooted sites but there were a higher number of shrews and bank voles caught in the non-rooted site than the rooted site. This suggests that boar rooting may negatively impact on the activity of shrews and bank voles but has little or no effect on Apodemus spp.

The third trial took place between the 30th of April and the 7th of May 2010 with the same method used in January 2010. The trapping was relatively unsuccessful due to severe interference by wild boar that were observed through motion-triggered cameras to move the traps. A total of 4 individuals (1 bank vole, 3 wood mice) were caught in the non-rooted site and 1 individual (wood mouse) was caught at the rooted site.

Singer et al. (1984) showed that two mammal species found predominantly among leaf litter, the red-backed voles (Clethrionomys gapperi) and the short-tailed shrew (Blarina brevicauda), were nearly eliminated from areas intensively rooted by wild pigs. Conversely, in the same area, the density

SID 5 (Rev. 07/10) of 49

of semi-arboreal small rodents (whose habitat was not affected by rooting) remained unchanged. These authors also suggested that repeated rooting decreases the food availability for small insectivores. Small mammals have been shown elsewhere to be affected by boar which compete for tree seeds and actively search and feed on rodents’ seed caches (Focardi et al. 2000).

Figure 21. Number of individual animals caught during the January 2010 study, expressed as catch per 100 trap nights, for Apodemus spp., bank voles and shrew spp.

The preliminary findings of the current study, in agreement with previous studies (Singer et al. 1984), suggest that rooting may affect shrews by exposing their invertebrate prey in the ground and making them available to birds, thus reducing prey availability for the insectivores. Rooting may also affect the bank voles as it destroys their tunnels through the surface layers. Conversely, more arboreal mice did not appear to be affected by rooting. A preliminary study carried out in the UK indicated that wild boar affected the numbers of bank voles and yellow-necked mice but not those of common shrews (Sorex araneus ) and wood mice (Jarvis pers. comm.).

The findings also indicated that, particularly in areas where wild boar densities might increase, more research should be carried out to expand on these preliminary results and to determine whether the activity, numbers and populations of small mammals could be affected by wild boar. Temporary protection, with boar-proof fencing of an area where recent rooting has occurred should overcome the problem of wild boar interference. In addition, food availability (seeds and invertebrates) could be quantified to assess the potential for indirect impact of wild boar on small mammal numbers.

For the dormice surveys, wild boar signs were very evident at the initial visits (late winter/early spring 2009/10) in sites where boar had been present for at least 10 years (based on anecdotal evidence). One 'boar absent' site reported boar presence during the study period (site closest to the boar area). Dormice were found only in non-boar sites in June 2009, but by October 2009 a similar proportion of boar and non-boar sites had evidence of dormice (Table 27).

Boar Present Boar AbsentNo of sites with dormice

N dormice or nests

N boxes with dormouse evidence

No of sites with dormice

N dormice or nests

N boxes with dormouse evidence

June 20090 0 0 6 15 13

October 20097 36 24 8 46 37

June 20107 14 10 8 24 18

October 20109 52 41 9 77 50

Table 27. Results from dormouse box inspections on sites with and without boar present. Twelve sites were monitored in each category with 50 nest boxes at each site.

SID 5 (Rev. 07/10) of 49

There was no evidence of dormice in 2-3 sites for both categories. Although there was a consistent pattern indicating that more dormice occurred in the sites that had no wild boar, the difference was only significant (p=0.048) for sites where dormice occurred. The highest number of dormice found on one site (16) was from a 'boar present' site where boar signs were evident all year. The difference between numbers of boxes with evidence of dormice was significant (P=0.018) when only sites with dormice were analysed.

These results suggest that there is no evidence to indicate that chronic presence of boar limits the presence of hazel dormice in the woodland type investigated (ancient semi-natural broadleaf woods). The study also indicated that this approach is effective at monitoring potential impact of boar on dormouse populations and that dormice use boxes within 6 months of them being placed in the field. Continued monitoring will enable confirmation of these conclusions.

Overall conclusionsThe results of Objective 3 on the large and small scale impact of wild boar indicated that, in

general, impacts at current population densities appear to be modest with possibly a few exceptions such as potential impact on bluebells. Bluebell sites were disproportionately rooted in early spring and rooting generally increased with increasing bluebell cover. Bluebells are a high profile species, being visually very obvious in spring. Thirty percent of the global population of bluebells occur in the UK, and the abundance of this species has declined markedly in southern England in recent years (HMSO 1994). There are already concerns about loss of this genetically specific species through hybridisation with the Spanish bluebell. Thus there would be further concerns if extensive wild boar rooting or re-rooting resulted in further declines of populations. Other plants, such as wood anemones Anemone nemorosa (Bialy 1996) and wild daffodil Narcissus pseudonarcissus (Gow 2002) may be also be affected by wild boar rooting and will be investigated in further studies.

The study also developed methods to quantify the environmental impact of wild boar at different spatial and temporal scales in woodlands. These methods represent simple and effective tools to establish patterns and severity of the environmental impact of introduced wild boar.

Future research should be undertaken to clarify the environmental impact wild boar in England on particular plant species or communities. Repeated surveys over a number of years and between seasons, in conjunction with monitoring wild boar density trends as well as diet, will clarify likely impacts and areas of possible concern. Future studies should include: (i) assessing the optimal number of permanent plots to monitor the frequency and timing of reoccurrence of rooting, (ii) determining the relationship between boar presence, rooting levels and ground flora and invertebrates, (iii) using the extensive rooting observed in spring to monitor effects on abundance and recovery of vernal species such as bluebell and anemone, (iv) evaluating the woodland edge/ride edge impact of rooting on plant species diversity, growth and flowering in relation to the development of habitat suitable for invertebrates, especially butterflies, (v) investigating the impact of rooting on deadwood invertebrates, (vi) investigating the impact of boar on ground nesting woodland birds,(vii) investigating the diet of wild boar (from samples of stomachs collected from culled animals) and compare availability and use of different plant and animal species and (viii) determining which factors affect the spatial and temporal intensity of rooting.

ReferencesAlexiou P.N. (1983) Effect of feral pigs (Sus scrofa) on subalpine vegetation at Smokers Gap ACT.

Proceedings of the Ecological Society Australia,12: 135-142.Allen L. (1984) Feral Pig Control Handbook. Queensland. Rural Lands Protection Board.Andersson S.J. & Stone C.P. (1993) Snaring to control feral pigs Sus scrofa in a remote Hawaiian rain

forest. Biological Conservation, 63: 195-201.Arrington D. A., Toth L.A. & Koebl J.W. (1999) Effects of rooting by feral hogs Sus scrofa L. on the

structure of a floodplain vegetation assemblage. Wetlands, 19: 535-544.Aviss M. & Roberts A. (1994) Pest Fences: Notes and Comments. Department of Conservation, New

Zealand, Threatened Species Occasional Publication No. 5.Baber D.W., & Coblentz B.E. (1986) Density, Home Range, Habitat Use, and Reproduction in Feral

Pigs on Santa Catalina Island. Journal of Mammalogy, 67:512-525. Baubet E., Ropert-Coudert Y. & Brandt S. (2003) Seasonal and annual variations in earthworm

consumption by wild boar (Sus scrofa scrofa L). Wildlife Research, 30: 179-186.Belden R.C. & Pelton M.R. (1975) European wild hog rooting in the mountains of east Tennessee. Proc

Twenty-ninth Annual Conf Southeast Assoc Game and Fish Commissioners: 665-671 Saint Louis MO: Columbia SC.

SID 5 (Rev. 07/10) of 49

Bolton, B. & Collingwoo, C.A. (1975) Hymenoptera: Formicidae. Handbooks for the Identification of British Insects, Vol. VI. Part 3(c). Royal Entomological Society, London, UK.

Bialy K. (1996) The effect of boar (Sus scrofa) rooting on the distribution of organic matter in soil profiles and the development of wood anemone (Anemone nemorosa L) in the oak-hornbeam stand (Tilio-carpinetum) in the Bialowieza primeval forest. Folia Forestalia Polonica Forestry, 38: 77-88.

Bratton S.P. (1975). The effect of wild boar on the ecology of the gray beech forest. Ecology, 56:1356-1366.

Bratton S.P., Harmon M.E. & White P.S. (1982) Patterns of wild boar rooting in the western Great Smoky Mountains. Castanea, 47: 230-242.

Bywater K. A., Apollonio M., Cappai N. & Stephens P. A. (2010) Litter size and latitude in a large mammal: the wild boar Sus scrofa. Mammal Review, 40: 212-220.

Calenge C., Maillard D., Fournier P. & Fouque C. (2004) Efficiency of spreading maize in the garrigues to reduce wild boar (Sus scrofa) damage to Mediterranean vineyards. European Journal of Wildlife Research, 50: 112–120.

Caley P. (1999) Feral pig biology and control in the Northern Territory. Department of Primary Industry Agnote J52. Northern Territories, Australia.

Campbell G., Mawson P. & Gray G. (1990) Electric fences for vertebrate pest management in Western Australia: a review. Agriculture Protection Board Technical Series No. 9.

Cargnelutti B., Spitz F. & Valet G. (1992) Analysis of dispersion of wild boar (Sus scrofa) in Southern France. In: “Ongulés/Ungulates 91”. Proceedings International Symposium Toulouse, France.

Choquenot D., McIlroy J. & Korn T. (1996) Managing Vertebrate Pests: Feral Pigs. Australian Government Publishing Service, Canberra.

Cooray R.G. & Mueller-Dumbois D. (1981) Feral pig activity In: Mueller-Dumbois D, Bridges KW, Carson HL (eds) Island Ecosystems: Biological Organization in Selected Hawaiian Communities. Hutchinson Ross, Stroudsburg, Pennsylvania, pp 309–317.

Corn J.L., Jordan T. R. & Madden M. (2010) Mapping the distribution of feral swine in the United States. 8th International Symposium on Wild boar and Other Suids, York UK1-4 September.

Dardaillon M. (1987) Seasonal feeding habits of the wild boar in a Mediterranean wetland the Camargue (Southern France). Acta Theriologica, 32: 389-401.

Dardaillon M. & Beugnon G. (1987). The influence of some environmental characteristics on the movements of wild boar. Biology of Behaviour, 12: 82-92.

Defra (2008) Feral wild boar in England: an action plan. Defra, 54 p.Dorosheva H. & Reznikova Z. (2006) An experimental study of behavioural mechanisms of

relationships between red wood ants and ground beetles. Entomological Review, 86: 139-148.Doupe R.G., Mitchell J., Knott M.J., Davis A.M. & Lymbery A.J. (2010) Efficacy of exclusion fencing to

protect ephemeral floodplain lagoon habitats from feral pigs (Sus scrofa). Wetlands Ecological Management, 18:69-78.

Efron, B. (2000) The bootstrap and modern statistics. Journal of the American Statistical Association, 95: 1293-1296.

Engeman, R.M., B. U. Constantin, M. Nelson, J. Woolard, & J. Bourassa (2001). Monitoring changes in feral swine population and spatial distribution of activity. Environmental Conservation, 28:235–240.

Engeman R., Pipas M., Gruver K., Bourassa J. & Allen L. (2002) Plot placement when using a passive tracking index to simultaneously monitor multiple species of animals. Wildlife Research, 29:85-90.

Engeman R.M,. Smith H.T., Schwiff S.A., Constantin B,. Woolard J., Nelson M. & Griffin D. (2003) Prevalence and economic value of feral swine damage to native habitat in three Florida state parks. Biological Conservation, 30: 319-324.

Engeman R. (2005) Indexing principles and a widely applicable paradigm for indexing animal populations. Wildlife Research, 32: 202–210.

Focardi S., Capizzi D. & Monetti D. (2000) Competition for acorns among wild boar (Sus scrofa) and small mammals in a Mediterranean woodland. Journal of Zoology, London, 250: 329-334.

Focardi S., Isotti R. & Tinelli A. (2002) Line transect estimates of ungulate populations in a Mediterranean forest. Journal of Wildlife Management, 66:48-58.

Franzetti B. & Focardi S. (2006) La stima di popolazione di ungulati mediante distance 367 sampling e termocamera a infrarossi. Ist. Naz. Fauna Selvatica, Documenti Tecnici, 26:1-88.

Gardner B., Reppucci J., Lucherini M. & Royle A.J. (2010) Spatially explicit inference for open populations: estimating demographic parameters from camera-trap studies. Ecology, 91: 3376–3383.

Geisser H. & Reyer H. (2004) Efficacy of Hunting, Feeding, and Fencing to Reduce Crop Damage by Wild Boars. Journal of Wildlife Management, 68:939-946.

SID 5 (Rev. 07/10) of 49

http://www.esajournals.org/doi/pdf/10.1890/09-0804.1

http://www.esajournals.org/doi/pdf/10.1890/09-0804.1

Genov P (1981) Food composition of wild boar in north-eastern and western Poland. Acta Theriologica, 26: 185-205.

Genov P (1981) Significance of natural biocenoses and agrocenoses as the source of food for wild boar (Sus scrofa L). Ekologia Polska, 29: 117-136.

Gerber B., Karpanty S.M., Crawford C., Kotschwar M. & Randrianantenaina J. (2010) An assessment of carnivore relative abundance and density in the eastern rainforests of Madagascar using remotely-triggered camera traps. Oryx, 44: 219-222.

Gill R.M.A., Thomas M.L. & Stocker D. (1997) The use of portable thermal imaging for estimating deer population density in forest habitats. Journal of Applied Ecology, 34:1273-1286.

Goulding M.J. & Smith G.C. (1998) Current Status and Potential Impact of Wild Boar (Sus scrofa) in the English Countryside: A Risk Assessment Report to MAFF March 1998.

Goulding M.J. (2003) An Investigation of Free-living Wild Boar (Sus scrofa L) in Southern England. PhD dissertation University of Sussex.

Goulding M., Kitchener A.C. & Yalden D.W. (2008) Wild Boar in (eds Harris S, Yalden D) The Mammals of the British Isles Handbook 4th edn. Mammal Society, Southampton pp 561-564.

Gow D. (2002) A wallowing good time – wild boar in the woods. Ecos, 23: 14-22Groot Bruinderink G.W.T.A. & Hazebroek E (1996) Wild boar (Sus scrofa L.) Rooting and forest

regeneration on podzolic soils in the Netherlands. Forest Ecology and Management, 88: 71-80. Hahn N. & Eisfeld D. (1998) Diet and habitat use of wild boar in SW Germany. Gibier Faune Sauvage,

15: 595-606.Haroldson B.S., Wiggers E.P., Beringer J., Hansen L.P. & McAninch J.B. (2003) Evaluation of

aerial thermal imaging for detecting white-tailed deer in a deciduous forest environment. Wildlife Society Bulletin, 31:1188-97.

Hawes C., Stewart A.J.A. & Evans H.F. (2002) The impact of wood ants (Formica rufa) on the distribution and abundance of ground beetles (Coleoptera: Carabidae) in a Scots pine plantation. Oecologia, 131: 612–619.

Hebeisen C., J. Fattebert E. Baubet & Fischer C. (2007) Estimating wild boar (Sus scrofa ) abundance and density using capture–resights in Canton of Geneva, Switzerland. European Journal of Wildlife Research, 54: 391-401.

HMSO (1994) Biodiversity The UK Action Plan. HMSO, London.Hodnett E. (2005) Thermal Imaging Applications In Urban Deer Control. Proceedings of the 11th

Wildlife Damage Management Conference. (D.L. Nolte, K.A.Fagerstone, Eds).Holland E.P., Aegerter J.N. & Smith G.C. (2007) Spatial sensitivity of a generic population model using

wild boar (Sus scrofa) as a test case. Ecological Modelling, 205: 146-158.Hone J. (1988a) Evaluation of methods for ground survey of feral pigs and their signs. Acta

Theriologica, 33: 451-465.Hone J. (1988b) Feral pig rooting in a mountain forest and woodland: Distribution abundance and

relationships with environmental variables. Australian Journal of Ecology, 13: 393-400.Hone J. (1995) Spatial and temporal aspects of vertebrate pest damage with emphasis on feral pigs.

Journal of Applied Ecology, 32: 311-319.Hone J. (2002) Feral pigs in Namadgi National Park Australia: dynamics impacts and management.

Biological Conservation, 105: 231-242.Hone J. & Atkinson B. (1983) Evaluation of fencing to control feral pig movement. Australian Wildlife

Research, 10:499–505.Hounsome T.D., Young R.P., Davison J., Yarnell R.W., Trewby .ID., Garnett B.T.,Delahay R.J. &

Wilson G.J. (2005). An evaluation of distance sampling to estimate badger (Meles meles) abundance. Journal of Zoology, 266: 81-87.

Howe T.F., Singer J. & Ackerman B.B. (1981) Forage relationships of European wild boar invading northern hardwood forest. Journal of Wildlife Management, 45: 748-754.

Ickes K (2001) Hyper-abundance of native wild pigs (Sus scrofa) in a lowland dipterocarp rain forest of peninsular Malaysia. Biotropica, 33: 682-690.

Jędrzejewska B., Okarma H., Jędrzejewski W. & Miłkowski L. (1994) Effect of exploitation and protection on forest structure, ungulate density and wolf predation in Białowieża Primeval Forest, Poland. Journal of Applied Ecology, 31: 664-676.

Jezierski W. & Myrcha A. (1975) Food requirements of a wild boar population. Polish Ecological Studies, 1: 61-83.

Jones C., Lawton J. & Shachak M. (1994) Organisms as ecosystem engineers. Oikos, 69: 373-386.Keuling O., Stier N. & Roth M. (2008) How does hunting influence activity and spatial usage in wild

boar Sus scrofa L.? European Journal of Wildlife Research, 54: 729-737.

SID 5 (Rev. 07/10) of 49

http://www.springerlink.com/content/1612-4642/

http://journals.cambridge.org/abstract_S0952836905006606

http://journals.cambridge.org/abstract_S0952836905006606

Keuling O., Lauterbach K., Stier N. & Roth M. (2009) Hunter feedback of individually marked wild boar Sus scrofa L.: dispersal and efficiency of hunting in northeastern Germany. European Journal of Wildlife Research, 56: 159-167.

Keuling O. (2010) Habitat use of wild boar in a forest agro-ecosystem in Northern Germany. 8th International Symposium on Wild boar and Other Suids, York, United Kingdom 1-4 September.

Khan M.H. (1990) Non-chemical methods of wild boar control. In: Wild Boar. Dylan, pp 157-160.Killian G. (2009) Development Implementation and Evaluation of Management Tools to Reduce Feral

Swine Damage at Avon Park. AFR Florida Annual Report Year 1 August 15.Kotanen P.M .(1995) Responses of vegetation to a changing regime of disturbance: effects of feral pigs

on a Californian Coastal Prairie. Ecography, 18: 190-199.Leaper R., Massei G., Gorman M. & Aspinal R. (1999) The feasibility of reintroducing Wild Boar (Sus

scrofa) to Scotland: Predictions from a Geographical Information System and a Population Viability Analysis. Mammal Review, 29: 239-259.

Linkie M., Dinata Y., Nugroho A. & Haidar I.A. (2007) Estimating occupancy of a data deficient mammalian species living in tropical forests: Sunbears in the Kerinci Sebilat region, Sumatra. Biological Conservation, 137: 20-27.

Littauer GA (1993) Control techniques for feral hogs. Pages 139–148. in Hanselka, C. W. and J. F. Cadenhead, editors. Proc. feral swine: a compendium for resource managers. Texas.

Long K. & Robley A. (2004) Cost-effective feral animal exclusion fencing for areas of high conservation value in Australia. A report for the Australian Government. Department of the Environment and Heritage , Canberra , Australia .

Luff M. J. (2007) The Carabidae (ground beetles) of Britain and Ireland. Handbooks for the Identification of British Insects, Vol. 4, Part 2 (2nd Ed.). Royal Entomological Society, London, UK.

Mackenzie D. I., Nichols J. D., Lachman G. B., Droege S., Royle J. A & Langtimm C. A. (2002) Estimating occupancy rates when detection probabilities are less than one. Ecology, 83: 2248-2255.

MacKenzie, D.I., Nichols J.D., Hines J.E., Knutson M.G. & Franklin A.D. (2003) Estimating site occupancy, colonization and local extinction when a species is detected imperfectly. Ecology, 84: 2200-2207.

MacKenzie D.I. & Royle J.A. (2005) Designing efficient occupancy studies: general advice and tips on allocation of survey effort. Journal of Applied Ecology, 42: 1105-1114.

Maillard D. & Fournier P. 1995. Effect of shooting with hounds on home range size of wild boar (Sus scrofa L.) groups in Mediterranean habitat. Ibex Journal of Mountain Ecology, 3:102–107.

Massei G., Genov P.V. & Staines B.W. (1996) Diet food availability and reproduction of wild boar in a Mediterranean coastal area. Acta Theriologica, 41: 307-320.

Massei G. & Genov P. (2004) The environmental impact of wild boar. Galemys, 16: 135-145.Massei G, Roy S. & Bunting R. (2011). Too many hogs? A review of methods to mitigate impact by wild

boar and feral pigs. Human-Wildlife Interactions, 5: 79-99.Massei G., Bacon P. & Genov P. (1998) Fallow deer and wild boar pellet group disappearance in a

Mediterranean area. Journal of Wildlife Management, 60:1086-1094.Mayle B., Peace A. & Gill, R. (1999) How Many Deer? A guide to estimating deer population size.

Forestry Commission field book 18. Forestry Commission, Edinburgh. 96pMcAtee WL (1939) The electric fence in wildlife management. J Wildlife Manage 3[1] 1-13.McKillop IG, Sibly RM (1988) Animal behaviour at electric fences and the implications for management.

Mammal Rev. 18: 91-103.McKillop I.G., Pepper H.W., Butt R. & Poole D.W. (2003) Electric Fence Reference Manual. Research

and Development Surveillance Report 607, Defra, London. Meterological Office (2011) Regional Climates. http://www.metoffice.gov.uk/climate/uk. Accessed 12

April 2011.Mitchell J., Dorney W., Mayer R. & McIIroy J. (2007) Spatial and temporal patterns of feral pig diggings

in rainforests of north Queensland. Wildlife Research, 34: 597-602.Monaco A., Carnevali L. & Toso S. (2010) Linee guida per la gestione del Cinghiale (Sus scrofa) nelle

aree protette. 2° Edizione. Quaderni di Conservazione della Natura, 34, Min. Ambiente – ISPRA.Moore N. (2004) The Ecology and Management of Wild Boar in Southern England. Defra Final Project

Report, VC0325. Moore N.P. & Wilson C.J. (2005) Feral wild boar in England: Implications of future management

options. A report on behalf of Defra European Wildlife Division.Parkes, J. P., Ramsey D. S., Macdonald N., Walker K., McKnight S., Cohen B. S., & Morrison S. A..

(2010). Rapid eradication of feral hogs (Sus scrofa) from Santa Cruz Island, California. Biological Conservation, 143:634-641.

SID 5 (Rev. 07/10) of 49

http://www.metoffice.gov.uk/climate/uk.%20Accessed%2012%20April%202011

http://www.metoffice.gov.uk/climate/uk.%20Accessed%2012%20April%202011

http://www.forestresearch.gov.uk/website/publications.nsf/WebpubsbyISBN/0855384050



Phillips I. & Cobb T. (2005) Effects of habitat structure and lid transparency on pitfall catches. Environmental Entomology, 34: 875-882.

Pihal R., Kamler J. & Homolka M. (2010) Comparison of four methods for estimation of wild boar population density in forest environment. 8th International Symposium on Wild boar and Other Suids York, UK1-4 September.

Poole D.W., McKillop I.G., Western G., Hancocks P.J. & Packer J.J. (2002) Effectiveness of an electric fence to reduce badger (Meles meles) damage to field crops. Crop protection, 21: 409-417.

Ralph C.J. & Maxwell B.D. (1984) Relative effects of human and feral hog disturbance on a wet forest in Hawaii. Biological Conservation, 30: 291-303.

Reidy M.M., Campbell T.A. & Hewitt D.G. (2008) Evaluation of electric fencing to inhibit feral pig movements. Journal of Wildlife Management, 72:1012–1018.

Rovero F. & Marshall A.R. (2009) Camera trapping photographic rate as an index of density in forest ungulates. Journal of Applied Ecology, 46: 1011-1017.

Rowcliffe J.M., Field J., Turvey S.T. & Carbone C. (2008) Estimating animal density using camera traps without the need for individual recognition. Journal of Applied Ecology, 45:1228-1236.

Royle J.A., Nichols J.D., Karanth K.U. & Gopalaswamy A.M. (2009) A hierarchical model for estimating density in camera-trap studies. Journal of Applied Ecology, 46:118-127.

Sáez-Royuela C. & Télleria J.L. (1986) The increased population of the wild boar (Sus scrofa L) in Europe. Mammal Review, 16: 97-101.

Santilli F. & Mazzoni della Stella R.M. (2006) Electrical fencing of large farmland areas to reduce crops damage by wild boars Sus scrofa. Agricoltura Mediterranea, 136: 7984.

Schley L., Dufrêne M., Krier A. & Frantz A.C. (2008) Patterns of crop damage by wild boar (Sus scrofa) in Luxembourg over a 10-year period. European Journal of Wildlife Research, 54:589–599.

Schley L. & Roper T.J. (2003) Diet of wild boar Sus scrofa in Western Europe with particular reference to consumption of agricultural crops. Mammal Review, 33: 43-56.

Seamans T.W. & VerCauteren K.C. (2006) Evaluation of ElectroBraid (TM) fencing as a white-tailed deer barrier. Wildlife Society Bulletin, 34:8-15.

Sims N. (2005) The ecological impacts of wild boar rooting in East Sussex. PhD University of Sussex.Singer F.J., Swank W.T. & Clebsh E.E.C. (1984) Effects of wild pig rooting in a deciduous forest.

Journal of Wildlife Management, 48: 464-473. Smiet A.C., Fulk G.W. & Lathiya S.B. (1979) Wild Boar Ecology in Thatta District: A Preliminary Study.

Pakistan Journal if Zoology,11 : 295-302.Sweitzer R.A., Van Vuren D., Gardner I.A., Boyce W.M & Waithman J.D. (2000) Estimating sizes of

wild pig populations in the North and Central Coast regions of California. Journal of Wildlife Management, 64:531 543.

Thurfjell H., Ball J. , Åhlén P., Dettki H. & Sjöberg K. (2009) Habitat use and spatial patterns of wild boar Sus scrofa (L.):agricultural fields and edges. European Journal of Wildlife Research, 55: 517–523.

Tilley L.G.W. (1973) Pig fencing in Mossman. Cane Growers Quarterly Bulletin, 36:132-133.Trout R.C. & Pepper H.W. (2006) Forest Fencing. Forestry Commission Technical Guide. Forestry

Commission, Edinburgh.Truvé J. & Lemel J. (2003) Timing and distance of natal dispersal for wild boar in Sweden. Wildlife

Biology, 9: 51-57.Van Laere G., Toïgo C., Gill R. & Gaillard J-M. (2003) Testing thermal imaging for the estimation of roe

deer density. 6th European Roe Deer Biology Conference, University of Minho, Portugal April 23-26.Vassant J. (1994) Les techniques de prevention des degats de sanglier. Office National de la Chasse,

Paris, France. Bulletin Mensuel, 191: 90-93.Venamore P.C. & Hamilton W.D. (1978) Don’t let feral pigs eat your profit. Queensland Agricultural

Journal, 104: 419-422.Vidrih M. & Trdan S. (2008) Evaluation of different designs of temporary electric fence systems for the

protection of maize against wild boar (Sus scrofa L., Mammalia, Suidae). Acta agriculturae Slovenica, 91: 343 – 349.

Waithman J. D., Sweitzer R. A., Van Vuren D., Drew J. D., Brinkhaus A. J., Gardner I.A. & Boyce W.M. (1999) Range expansion, population sizes, and management of wild pigs in California. Journal of Wildlife Management, 63: 298–308.

Ward A. I. (2005) Expanding ranges of wild and feral deer in Great Britain. Mammal Review, 35: 165-173.

Welander J. (1995) Are wild boar a future threat to the Swedish flora? Ibex Journal of Mountain Ecology, 3: 165-167.

SID 5 (Rev. 07/10) of 49

http://apps.isiknowledge.com.ezphost.dur.ac.uk/OneClickSearch.do?product=UA&search_mode=OneClickSearch&db_id=&SID=X1kPf5B5MmgbFgdabpN&field=AU&value=VerCauteren%20KC&ut=000237797300003&pos=2

http://apps.isiknowledge.com.ezphost.dur.ac.uk/OneClickSearch.do?product=UA&search_mode=OneClickSearch&db_id=&SID=X1kPf5B5MmgbFgdabpN&field=AU&value=Seamans%20TW&ut=000237797300003&pos=1


http://apps.isiknowledge.com.ezphost.dur.ac.uk/OneClickSearch.do?product=UA&search_mode=OneClickSearch&db_id=&SID=X1kPf5B5MmgbFgdabpN&field=AU&value=Packer%20JJ&ut=000176314300008&pos=5

http://apps.isiknowledge.com.ezphost.dur.ac.uk/OneClickSearch.do?product=UA&search_mode=OneClickSearch&db_id=&SID=X1kPf5B5MmgbFgdabpN&field=AU&value=Hancocks%20PJ&ut=000176314300008&pos=4

http://apps.isiknowledge.com.ezphost.dur.ac.uk/OneClickSearch.do?product=UA&search_mode=OneClickSearch&db_id=&SID=X1kPf5B5MmgbFgdabpN&field=AU&value=Western%20G&ut=000176314300008&pos=3

http://apps.isiknowledge.com.ezphost.dur.ac.uk/OneClickSearch.do?product=UA&search_mode=OneClickSearch&db_id=&SID=X1kPf5B5MmgbFgdabpN&field=AU&value=McKillop%20IG&ut=000176314300008&pos=2

http://apps.isiknowledge.com.ezphost.dur.ac.uk/OneClickSearch.do?product=UA&search_mode=OneClickSearch&db_id=&SID=X1kPf5B5MmgbFgdabpN&field=AU&value=Poole%20DW&ut=000176314300008&pos=1

Welander J. (2000) Spatial and temporal dynamics of wild boar (Sus scrofa) rooting in a mosaic landscape. Journal of Zoology London, 252: 263-271.

West B.C., Cooper A.L. & Armstrong J.B. (2009) Managing wild pigs: A technical guide. Human-Wildlife Interactions Monograph, 1:1-55.

Wilson C.J. (2004) Rooting damage to farmland in Dorset southern England caused by feral wild boar. Mammal Review, 34: 331-335.

Wilson C. J. (2005) Feral wild boar in England: status, impact and management. Defra European Wildlife Division Report.

Wilson C. J. (2010) Wild boar monitoring in England: monitoring novel species when the novelty has worn off. 8th International Symposium on Wild boar and Other Suids. York United Kingdom 1-4 September.

Wood G.W. & Roark D.N. (1980) Food habits of feral hogs in coastal South Carolina. Journal of Wildlife Management, 44: 506-511.

Wright J.W. (1972) Electric fence to keep out wild pigs. Queensland Agricultural Journal, 98: 371–372.

Websites CALU technical notes. Alternative Livestock. Wild boar production. June 2007.http://www.calu.bangor.ac.uk/Technical%20leaflets/040901%20wildboar.pdf. Accessed 2 October

2010.Deer Initiative 2009 Best Practice Guidance see http://www.wild-boar.org.uk/guide_list/ Gallagher PowerFence™ Manual (http://www.gallagher.com.au/electric-fence-manuals.aspx)Natural England Species Information Note SIN002 Feral wild boar. First Edition 18 June 2007.http://naturalengland.etraderstores.com/NaturalEnglandShop/SIN002. Accessed 2 October 2010.Patura electric fence cataloguehttp://www.allie.de/pdf/download/2010/GB/PatWZ10GB105_122.pdf

Prévention des dégâts de gibier. La cloture électrique. Service Public de Wallonie.http://environnement.wallonie.be/publi/dnf/cloture_electrique_sanglier.pdf. Accessed 2 October. 2010.Wild boar fencing – England and Wales best practice guides. The Deer Initiative. 27th July 2009.http://www.wild-boar.org.uk/pdf/WildBoar_fencing.pdf. Accessed 2 October 2010.http://www.clotures-electriques.com/plan_de_montage/cloture_sangliers.phphttp://www.sangliere.net/shop_int/sanglier.phphttp://www.britishwildboar.org.uk/Defra%20The%20Ecology%20and%20Management%20of%20Wild

%20boar%20in%20southern%20England.pdf

References to published material9. This section should be used to record links (hypertext links where possible) or references to other

published material generated by, or relating to this project.

Upsom M., Williams D., Straw N., Harmer R. & Kewitt A. 2010.Impacts of feral boar on woodland flora and invertebrates. 8Th International Symposium on Wild Boar & Other Suids,York.

Gill, R. and Brandt, G. Estimating density of British wild boar populations using thermal imaging. 2010. 8Th International Symposium on Wild Boar & Other Suids,York.

Watola G., S. Fox, M. Lambert, J. Coats, G. Massei & D. Cowan. 2010. The spatial & temporal scale of wild boar rooting in English woodlands. 8Th International Symposium on Wild Boar & Other Suids,York.

Presentations to FC Policy and Practice and Defra staff at Forest of Dean, Aug 2010.

SID 5 (Rev. 07/10) of 49

http://www.wild-boar.org.uk/guide_list/

general enquiries on this form should be made...

Documents