Brenda A. Mayle, Andrew J. Peace and Robin M. A. Gill

Forestry Commission Field Book 18 ARCHIVE

Field Book 18

How Many Deer?

A Field Guide to Estimating Deer Population Size

Brenda A. Mayle, Andrew J. Peace and Robin M. A. Gill Forest Research

Edinburgh: Forestry Commission © Crown copyright 1999 Applications for reproduction should be made in writing to The Copyright Unit, Her Majesty's Stationery Office, St Clements, 2-16 Colegate, Norwich NR3 1BQ

ISBN 0 85538 405 0 FDC 153:149.6 (410)

KEYWORDS: Census, Fallow deer, Forestry, Muntjac deer, Population estimation, Populations, Red deer, Roe deer, Sika deer, Wildlife management

Acknowledgements The authors would like to thank the many people involved at various stages during the planning and production of this book: Stephen Buckland and Simon Hodge for comments on earlier drafts of the manuscript; James Cordery and Colin Lavin for providing examples for a number of methods in addition to text comments; the Deer Commission for Scotland for providing information and the area map for open hill counts, and results from distance pellet group sampling; the Ministry of Agriculture, Fisheries and Food for allowing reference to work funded by them; George Gate and John Williams for photographic material and artwork; June Bell and Mavis Peacock for typing drafts of the text. Particular thanks go to the Forest Enterprise ranger staff who have supported the field work for this publication, and especially those who have monitored faecal pellet decay in the sites given in Appendix 7.

Enquiries relating to this publication should be addressed to:

The Research Communications Officer Forest Research Alice Holt Lodge Wrecclesham, Farnham Surrey GU10 4LH

Front cover: top left Group of fallow does in birch woodland [F o r e s t L if e P ic t u r e L ib r a r y : 1005495020]. top right Roe doe in a woodland ride [F o r e s t L if e P ic t u r e L ib r a r y 1012420020], bottom left Sika stag in a spruce plantation [N o r m a n H e a l y ], bottom right Red deer on the open hill [A l a s t a ir B a x t e r ], centre Muntjac doe in a grassy ride [Ia n W y l l ie ].

Back cover: Searching for faecal pellet groups in an open grassy habitat within Kielder Forest [T e c h n ic a l D e v e l o p m e n t B r a n c h , Fo r e s t Re s e a r c h ], Contents

Page Acknowledgements ii Preface v List of plates vi List of figures vii List of tables viii

1. THE NEED TO MANAGE DEER 1 Why count deer? 2

2. CHOOSING A SUITABLE CENSUS AND SAMPLING METHOD 4 Objectives 4 Sampling 5 Relationship between accuracy precision and bias 5 Sample size 7 Probability sampling 7 Line transects 13

3. POPULATION ESTIMATION METHODS 15 Direct counts 15 Daylight counts 18 1. Open hill counts 18 2. Drive counts 21 3. Static census 23 4. Vantage point counts 24 5. Aerial counts 27

Night counts 29 6. Spotlight counts 30 7. Thermal imaging: direct counts 31 8. Thermal imaging: distance sampling 32

Other direct methods 38 9. Individual recognition (mark-resighting) 38 10. Change-in-ratio counts 40

Indirect methods of measuring deer populations 43 11. Impact levels 43 12. Track /slot counts 45

iii Faecal pellet counts 46 13. Index of deer presence 49 14. Clearance plot counts (FAR) 50 15. Standing crop plot counts 52 16. Standing crop strip transect counts 55 17. Standing crop line transect counts 58

Using cull information 60 Population reconstruction from mortality data 60 18. Balance sheet 60 19. Life tables 62 20. Cohort analysis 63 21. Population modelling 65

4. USING CENSUS DATA 66

5. SUPPORTING INFORMATION 68 Appendix 1 Random number table 68 Appendix 2 Example habitat stratification 69 Appendix 3 Vantage point count form 71 Appendix 4 Thermal imaging record form 72 Appendix 5 List of equipment suppliers 74 Appendix 6 Faecal pellet group identification 75 Appendix 7 Faecal pellet group decay assessment 78 Appendix 8 Faecal pellet group defecation rate assessment 88 Appendix 9 Plot faecal pellet group count form 89 Appendix 10a Transect faecal pellet group count form 90 Appendix 10b Deer density assessed from transect faecal pellet group count 91 Appendix 11 Cohort analysis form 92

References 93

iv Preface

Within Great Britain and across The book is divided into four key much of the Northern hemisphere sections and a fifth section deer populations have been comprising supporting material. expanding in size and distribution. Section 1 considers the need to In Britain this has been enhanced by manage deer and the importance of an increase in the proportion of estimating the size of deer woodland cover, and hence suitable populations. In Section 2 guidance habitat for deer. There has also been is given on choosing a suitable insufficient recognition by land census and sampling method, with owners and managers of the need to emphasis on the importance of limit deer numbers. The management establishing initially the underlying of deer populations is a complicated objectives of the required survey(s). subject with no two sites being exactly the same. To be most Section 3 describes the population effective, detailed knowledge of the estimation methods, numbered 1 to deer population range, size, rates of 21, and includes guidance on change and habitat use are required selection of the most appropriate before decisions about management method. Each follows the format: objectives and methods are made. Method, Data recorded, Equipment required and Health and safety (H & S) This Field Book aims to provide considerations, Advantages and information on the various methods Disadvantages. Guidance on the ease available for determining deer of use, performance and levels of population size, and the suitability costs associated with each method is of each method for a given situation. given in Table 3.1 and within each Although aimed primarily at description. A worked example is woodland deer managers, methods provided in most cases. The examples suitable for a range of situations given can be calculated using a are described. Results from research simple pocket calculator. Results are by the authors and their coll­ presented as deer densities per 100 eagues, information in the avail­ hectares (or per square kilometre, able literature, and the practical km'2). Software is available from the experiences of friends and coll­ authors at Forest Research, Alice eagues in the field are all drawn Holt Lodge, Wrecclesham, Farnham, together and used as examples to Surrey GU10 4LH, to help speed up illustrate methods. Methods of some of the calculations. greatest use to practising deer managers (in terms of time and cost Section 4 reinforces the advantages effectiveness) are described in more of using census data in the detail. The authors hope that the development of effective deer combination of experience, research management strategies. The sup­ and observations that make up this porting information in Section 5 Field Book will help to promote includes guidance on identification, sound deer management and assessment and recording; example encourage those new to the subject record forms and addresses of to approach it in an informed and equipment suppliers are also objective manner. provided.

v List of plates

Plates 1-15 in colour section between pages 48 and 49.

Plate 1 Example of an area suitable for vantage point count, looking across a valley. [B. M a yle]

Plate 2 Handheld thermal imager. [R. G il l ]

Plate 3 Thermal image of roe deer. [P il k in g t o n O p t r o n ic s Lt d ]

Plate 4 Close up of deer slots in soil. [41876]

Plate 5 Deer path through bottom of hedge. [41850]

Plate 6 Fused pellet group. [41732]

Plate 7 Pellet groups from rabbit, sheep, squirrels and three deer species. [41839]

Plate 8 Red deer pellet group. [A . C h a d w ic k ]

Plate 9 Muntjac pellet group. [41554]

Plate 10 Roe deer pellet group. [41735]

Plate 11 Fallow deer pellet group. [41481]

Plate 12 Sika deer pellet group. [51971]

Plate 13 Sheep pellet group. [41843]

Plate 14 Muntjac latrine. [41551]

Plate 15 Roe deer faecal pellet group, almost decayed. [42071]

vi List of figures

Figure 1.1 Deer presence in the UK

Figure 2.1 Relationship between accuracy, precision and bias

Figure 3.1 Flowchart for selecting appropriate census method

Figure 3.2 Deer Commission for Scotland: red deer count areas

Figure 3.3 Map of drive count area showing positions of observers and direction of beat

Figure 3.4 Drive count record form

Figure 3.5 Vantage point count map for Abhain Bheag, Carradale

Figure 3.6 Vantage point count record form for density estimation of woodland deer in hilly terrain

Figure 3.7 The distribution of transect routes to sample the Mundford block

Figure 3.8 Distance estimate through rangefinder

Figure 3.9 Calculation of the perpendicular distance from a deer to the transect line

Figure 3.10 Measurement of the perpendicular distance of observed deer from a transect route when plotting deer on a map: three typical scenarios

Figure 3.11 Location of transect lines within a block for strip and transect sampling

Figure 3.12 Cohort analysis for roe does, Alice Holt Forest

Figure A6.1 Typical deer pellet shapes

Figure A7.1 Three moisture zones in Great Britain, based on ecological site classification List of tables

Table 2.1 Habitat structure classes based on differences in shelter and vegetation cover

Table 2.2 Habitat stratification based on vegetation structure classes

Table 3.1 Suitability of census methods to estimate deer population size

Table 3.2 Results from open hill counts (small block)

Table 3.3a Average body length of each species/sex used to estimate distance (cm)

Table 3.3b Width across ears as a proportion of nose-tail length (%)

Table 3.4 Index of deer presence from browsing and grazing impacts

Table 3.5 Number of plots required to achieve ±20% accuracy for standing crop counts

Table 3.6 Sample plot allocation (proportional sampling)

Table 3.7 Estimation of roe deer population size at Glen Righ from faecal pellet standing crop counts

Table 3.8 Estimation of deer population size at Lochaber from strip transect faecal pellet counts (800 m)

Table 3.9 Predicted population changes using a Leslie matrix model

Table A7.1 Decay lengths for red deer

Table A7.2 Decay lengths for roe deer

Table A7.3 Decay lengths for sika deer

Table A7.4 Decay lengths for fallow deer

Table A7.5 Decay lengths for muntjac deer

Table A7.6 Faecal pellet group decay assessment form

Table A7.7 Pellet decay: example of 12 months' records

Table A8.1 Defecation rates for British deer species

Table A10.1 Roe deer density determined from transect faecal pellet group count

Table A10.2 Red and sika deer density determined from transect faecal pellet group count

viii The Need to Manage Deer

Within Great Britain and across habitat for deer. There has also much of the Northern hemisphere been insufficient recognition by deer populations have been landowners and managers of the expanding in size and distribution need to actively manage deer. Hence (Gill, 1990; Harris et al., 1995). This populations have increased in both has been enhanced by changes in numbers and distribution, such that land-use practices which have led to deer of one species or another occur an increase in the proportion of in almost every 10 km square of woodland cover, and hence suitable Great Britain (Figure 1.1). Despite

Figure 1.1 Deer presence in the UK. (Prepared by H. Arnold, J. Gaunt and A. Morton, ITE, Monks Wood) 1 being our largest land mammal their problems in maintaining particular secretive nature and tendency to habitats to owners and managers avoid humans makes them very with sporting interests. The 'Deer difficult to observe, even at high Initiative', a partnership of densities. There is generally, organisations in England and Wales therefore, a lack of awareness of the led by the Forestry Commission, size or density of deer populations aims to encourage responsible and in many areas. humane management of deer (Forestry Commission, 1995), and As woodland habitats are the formation of co-operative increasingly planted close to urban management groups where areas, the potential for contact with necessary. and conflict between human and deer populations increases, Why count deer? emphasising the need for proactive management. The main reasons for estimating the size of a deer population are: The management of deer populations is a complex subject • To determine whether the with no two sites being exactly the population is increasing, stable or same. To be most effective it requires in decline. detailed knowledge of the deer • To identify the level of culling the population range, size, rates of population can sustain. change and habitat use, before decisions are made about Anyone wishing to establish new management objectives and how to woodlands will need to know achieve them. For populations which species are present and at which reside exclusively in one least an index (i.e. an indication) of particular area or habitat type it will numbers of deer when deciding be easier to define the range, for whether protection against deer example where the population is damage is necessary. For example, held in a deer park, within an area individual tree protection (e.g. tree bounded by an intact deer fence or guards) against fallow, red and within the boundaries of a large sika deer needs to be 1.8 m tall, woodland. Seasonal movements of while to protect against roe and the deer also need to be taken into muntjac browsing, 1.2 m is account when defining the adequate. Using the incorrect height boundaries of a 'deer range'. of protection will lead to greater costs if lower protection would have For populations ranging across large been adequate, or loss of leading areas of land, decisions on shoots, deformed trees and even management objectives for the replanting if taller protection should whole population may need to be have been used. made within a consortium of landowners and managers across At very low deer densities (<5 km 2) whose land the deer roam. This will the risks of damage will be low invariably include people with (especially for large areas of diverse interests and objectives, planting). Damage risk generally from farmers with crop damage, increases with increasing deer conservation site managers with presence, but is influenced by

2 species-specific behaviour and Accurate methods of estimating the habitat. For example, fallow deer are population size will also be required a herding species and move around by those wishing to reduce a their range in large groups. They population. This may be necessary may cause high impact within a for a number of reasons: very short period of time by feeding on newly established or coppiced • To reduce crop or woodland trees or by trampling and lying on damage. agricultural crops. • To decrease levels of road traffic accidents occurring in the area. For managers who are considering • To improve the quality of the cropping a deer population to herd. produce a regular income it will be important to know the sex-class and Without an accurate estimate of age-class ratios and rate of increase population size and its rate of of the population as well as the increase, it will not be possible to population size. Census sampling determine how many deer need to methods that provide this be removed to either hold the information and an indication of the population stable or reduce it in accuracy of the population estimate size. to be calculated will be required.

3 / Choosing a Suitable Census and i s s s f Sampling Method

Objectives will also be influenced by the objectives. If these include an The success of any survey to estimate assessment of age-class or sex-class deer population size should be ratios then direct methods will have judged against the underlying to be used. In a forest habitat where objectives. Furthermore, clear direct observation is severely objective setting is essential for the restricted then indirect methods will selection of the most appropriate provide a greater opportunity to census method and choice of estimate deer populations or at least sampling design. The questions to provide an index of their numbers. be answered must be clear at the outset. For example: Often a quick answer is required but quick methods of estimating deer • How many deer are there in a populations usually involve a number specific woodland block? of assumptions which may not be • How accurate will this answer be valid for the particular population in for a fixed cost? question. In general, more accurate • Are estimates required for population estimates require more different habitat types? time to collect, but greater confidence • Is the deer population increasing can be placed in them. or decreasing? • Is an index of deer numbers, If deer numbers are to be monitored rather than an actual count, over a number of years it is sensible acceptable in deciding this? to keep the census and any sampling methods as consistent as possible, An index can be thought of as a using the same census technique measurement related to the actual and making assessments when number of deer present, such as the conditions are similar, e.g. at the number of deer tracks in snow, and same time of the year, under similar can be extremely useful for broad weather conditions, using the same decision making. Where there is little observers. All this will help to concern about current population reduce any bias, although significant size, or impact levels, or the changes in the habitat or likelihood of expansion, methods disturbance levels over time may providing only a minimum estimate affect deer behaviour and presence may be acceptable. and hence the suitability of any particular method. Results should Once the objectives of the study be stored so that they are readily have been decided it should be retrieved, accessible and clearly possible to match these to one of the understood by others in the future. several population estimation methods described in Section 3, and Key points quantify the amount and type of • Make a list of objectives and data that will need to be collected. questions to be answered. • Select a suitable census method The choice between direct and indirect and sampling design to produce methods of assessing deer numbers the answers required.

4 • Use of inappropriate methods is Relationship between accuracy inefficient, may be costly, and will precision and bias produce results that are either too detailed, unreliable or not Once the sampling method and pertinent to the main objectives/ design have been chosen, care should questions. be taken to ensure that all observers follow the instructions given. The Sampling aim is to obtain a result that is precise, accurate and unbiased. Several census methods require data Deviation from the chosen method to be collected by sampling. On a and sampling design may cause the small scale it may be possible to final result to become imprecise, count all the deer inhabiting an inaccurate and biased, resulting in a isolated woodland, and consequently waste of time and money. have no need to sample. However a complete census is generally not Precision, accuracy and bias can be feasible, especially in woodlands, defined as follows: and sampling methods will have to be used. Factors which need to be • Precision is the closeness of taken into account in choosing a repeated estimates to each other. sampling design include the • Accuracy refers to the closeness resources available, the accuracy of the estimated mean to the true level required and the likelihood of value. bias. • Bias is a systematic component of error, such as observer bias. The sampling methods discussed below apply equally to estimates of Figure 2.1 shows four possible deer, their faecal pellets and their scenarios that can occur when tracks. sampling.

(c) Precise Accurate Unbiased

m= true mean e= estimated mean x= samples

Figure 2.1 Relationship between accuracy, precision and bias

5 Figure 2.1a shows a successful method used on a population of sampling design where the average known size. However some simple of the samples (e) gives a good precautions can be applied during estimate of the true mean (m). the sampling period. For example, if Figure 2.1c is the most desirable it is obvious that deer on the outer result (precise and accurate) as a edges of a strip transect are going good estimate of the true mean can undetected then the width of the be obtained from fewer samples strip should be reduced. Observer with a resultant saving in sampling bias can be detected by checking the costs. Figures 2.1b and 2.Id show the results from different observers effect of bias: the estimated mean for using the same method on the same the samples is not close to the actual site. Consistent differences between mean and in these cases the true observers will suggest that bias is number of deer are consistently occurring but not who is biased. All over- or underestimated. observers should therefore be retrained. More often than not, the Bias presence or absence of bias will only be confirmed by using a second If bias exists anywhere in the sampling technique with a further estimation procedure then the increase in cost. Many of the census estimated deer population size may techniques listed in this Field Book not be very close to the true size. appear in the scientific literature Increasing the number of sampling where the effects of bias have been units will not remove any bias that is measured on known deer populations. present although it may produce a Many papers suggest methods that more precise estimate (Figure 2.1d). can measure the extent of any bias Bias can arise from many sources and some produce models that during the process of estimating convert biased results to unbiased deer numbers. During sampling it estimates of the population size. could be introduced by lack of randomisation of the sample plots Ultimately, if bias persists in the or, in a line transect, by deviation deer population estimate then there from the fixed route to sample a are several options. In the worst more attractive or accessible habitat case the chosen sampling method type. Mark-resighting methods will must be replaced with a more suitable be biased if poor capture and method to control the particular marking techniques lead to an circumstances causing the bias, and increased mortality rate of marked the results already obtained must be individuals compared to unmarked rejected. Alternatively it may be ones or if the methods assume a acceptable to use the biased estimate 'closed' deer population during the as an index of the true population period between marking and number. This decision will depend resighting, when it is in fact 'open' to on the amount of observed bias, the migration, immigration, mortality value of having at least some idea of and recruitment. the population size, and the consequence of any wrong decisions Detection and correction for bias can made as a result of using a false be extremely difficult. Generally estimate of population size. For the bias in the final population estimate methods described we highlight can only be detected by testing the factors which may lead to bias.

6 The precision of unbiased estimates Most projects are undertaken with a of a deer population can be fixed budget and this may limit the improved by increasing sampling size of the sample that can be taken. intensity. However a gain in Studies will consist of fixed costs, precision will nearly always come at such as administration and a cost. By comparing relative planning, and a cost for each precision and cost the surveyor can sampling unit. The maximum make important decisions on sample size one can afford to take is whether the planned level of calculated as follows: resources are sufficient to provide a precise enough estimate and hence Maximum Total budget - Fixed costs reduce the likelihood of wasting sample size Cost of each sampling unit resources on a study that either fails to deliver sufficiently precise results If this maximum sample size is or produces overly precise results. bigger than the number that is Even decisions on whether or not to needed to achieve the required carry out a survey at all can be made precision then there is no problem. based on the expected precision for There is no point in taking more a given cost. observations than necessary to achieve the desired precision. If the Key points maximum sample size is smaller • Avoid introducing bias during than the number needed for the the sampling procedure. specified precision, then there are • Increasing the sample size will three possible choices: increase precision but may not • Take as many observations as can deal with bias or improve be afforded, knowing the results accuracy. will not be as precise as hoped • Most surveys will involve a for. compromise between cost, • Use a different, cheaper sampling accuracy and precision. design which is more likely to give the required precision. Sample size • Do not carry out the survey.

At an early stage of the design a Probability sampling decision must be made about 'How large a sample is required to The sampling methods listed below achieve a given precision?'. Too use the concept of probability small a sample will produce an sampling. In order to apply this estimate that is too imprecise to be concept the study area for which the useful, while a sample that is deer population is to be estimated too large is both inefficient and has to be defined and subdivided costly. Recommended sample sizes into equal sized units called are given for each census sampling units. These sampling units method that requires a sampling (which, for example, could be design. For the reader who is 7 m x 7 m quadrats, 2 m x 100 m familiar with statistical techniques, strips or 1 hectare plots), should be books on sampling theory, such as distinct, non-overlapping and Cochran (1977), give formulae to together constitute the whole study estimate sample sizes for a given area. They then form the basis for precision. the selection of the sample.

7 If faced with a choice between appearing in any position. different sampling units then a guiding rule is to try to select one Example that returns the greatest precision Suppose a forest block is divided for the available resources. Generally, into 500 equally sized plots the precision of the overall estimate of (sampling units) and each plot is the deer population is related to the uniquely identified by a number square root of the number of sampling from 1 to 500. To take a random units selected for the sample. To double sample of 10 plots first choose a the precision of the estimate the number random starting position in the of units selected must be quadrupled. random number table and combine For a fixed sample size, say 5% of the digits in a long string, say the total study area, fewer larger 173649374655238475601374110.... sampling units will normally give less accurate results than more As the largest plot number is 500, smaller sampling units. However, it which has 3 digits, group the is generally cheaper to sample a random numbers into groups of larger sampling unit than a smaller three: 173 649 374 655 238 475 601 374 one. If sampling units are too small, 110.... numbers may be biased by edge effects (observers may tend to Plot 173 is the first plot in the sample. include all individuals that straddle Plot 649 is ignored as there is no plot the sample boundary). One solution with this number. is to choose square sampling units Plot 374 is the second plot in the which have a smaller edge to area sample. ratio than rectangular plots. Plot 655 is ignored. Plots 238 and 475 are the third and Simple random sampling fourth plots in the sample. Simple random sampling is a Plot 601 is ignored. method in which the sampling units Plot 374 is a repeat and is ignored. are chosen independently and with Plot 110 is the fifth plot in the equal probability. If there is prior sample. knowledge of the variability Continue until 10 plots have been between sampling units, such as selected. certain habitat types being more likely to give higher responses than Results from the sample are used to others, then simple random calculate four important statistics: sampling will not give the best results. One of the methods • The sample mean described later may be more precise. - average count per plot Choosing a simple random sample is • The sample variance usually done using random numbers - a measure of variability which are obtained from a computer between plot counts or random number table (Appendix 1). • The standard error of the mean A random number table is just a list - a measure of precision with of digits 0,1,2...9 where the number which the population mean is at any point in the table has been estimated chosen at random so that each digit • Approximate 95% confidence has exactly the same chance of interval of the mean

8 - a measure of precision with Approximate 95% confidence which the population mean is interval of the mean (95% Cl) estimated. = sample mean ± 2 x s.e

= 5 ±1.16 Now if 5, 4, 2, 6, 8, 3, 4, 6, 5, and 7 deer are observed in the 10 plots the = 3.84 - 6.16 deer following information is normally calculated from the sample data. The 95% confidence interval can be interpreted as follows: we are The sample mean 95% confident that the true The mean number of population mean of the 500 deer per sampling unit sampling units lies between 3.84 and 6.16 deer per sampling unit. Sum of all sample deer Sometimes 80% confidence intervals Number of sample plots are used and are approximated as follows: 5 + 4 + 2 + ...+ 7 80% Cl - sample mean ± 1.3 x s.e. 10 Within simple random sampling the selection of the sample is left to = 5 deer chance and any prior knowledge of differences in deer numbers within the study area is ignored. However, The sample variance this knowledge can be used to The variance of deer per sampling unit increase the precision obtained for a fixed number of samples by using a stratified sampling design. Sum (samples - sample mean)2 Number of sample plots -1 Stratified sampling Where there is likely to be variation (5-5)2 +(4-5)2 +(2-5)2 +...+(7-5)2 across the study area in the factors 10-1 being measured (e.g. different decay rates of pellet groups in different 30 habitats or variation in visibility of 9 deer depending on tree species and age-class) it is important to stratify. = 3.33 deer To do this, the study area, consisting of varied sampling units is divided The standard error of the mean (s.e.) into strata, each of which contains a The standard error of the group of similar sampling units. A mean number of deer per simple random sample is taken sampling unit independently from each stratum (as above) and the total population Variance of deer per sampling plot estimate calculated by combining the Number of sample plots estimate of the stratum means. As sample size can vary between strata this provides scope for allocating resources between strata and in a number of cases leads to a gain in = 0.58 deer precision. Table 2.1 Habitat structure classes based on differences in vegetation and cover

Habitat class Resources available to deer Establishment/ 1 m Increasing food availability. restock Little cover/shelter. Pre-thicket 1-3 m High food quality and abundance. Increasing cover. Thicket 3-10 m Canopy closure and reducing food availability except where gaps in trees occur. Excellent cover. Pole-stage 10+ m Poor food availability except where gaps occur. Cover good depending on thinning level. Pre-felling 10+ m In open crops food good as ground cover increases. Cover from disturbance good, but from weather may be poor. Open ground No crop Food variable depending on availability of herb/ shrubby species. Shelter poor but topography may be sufficient for smaller deer species.

Stratification is normally based on (after Ratcliffe and Mayle, 1992) habitat or ecological features with gives an example of site which deer density is likely to be stratification for Alice Holt Forest correlated. Decisions can be made in North Hampshire/Surrey. from large scale vegetation or Stratification was based on crop habitat maps of the study area. structure classes and types in Tables 2.1 and 2.2 show different preparation for faecal pellet counts woodland habitat strata determined and undertaken by someone with from the availability of food and no previous knowledge of the site. shelter for the deer. Table 2.2 shows the potential range of open and In stratified sampling, sampling woodland habitat strata. Within units should be allocated to strata each of the structure classes it may based on the variability (variance) be necessary to subdivide the strata that is expected within each stratum. using specific knowledge of the site Strata with greater than average or deer population. Some areas may variation should be allocated have high visitor access or other proportionally more sampling units. disturbances such as game shooting Similarly, if sampling costs vary while other areas may include from stratum to stratum, then the rutting stands at specific times of the more expensive it is to sample a year. Forestry operations other than stratum, the fewer the number of clearfelling, however, do not often sampling units should be allocated cause high levels of disturbance to to it, down to a minimum sample the deer or lead to unusual size. This concept can be solved behavioural patterns. Appendix 2 mathematically and is known as

10 Table 2.2 Habitat stratification based on vegetation structure classes

Open habitat Woodland habitat Bare ground Establishment / Restock Grass ride/glade Conifer Broadleaved Riparian Pre-thicket Pre-thicket Moor/heath Thicket Thicket Arable crops Pole-stage Pole-stage Grass pasture Pre-fell Pre-fell Mature / retention Mature / retention Short rotation coppice Shrub and ground vegetation below conifer and broadleaved tree crops on the same site are likely to differ due to the differential effect of shading and so stratification needs to consider crop species. optimum allocation. We will introduce we will allocate 2/5 of the samples and recommend the use of to habitat A, 2/5 of the samples to proportional allocation which, on most habitat B and 1 / 5 of the samples to occasions, is nearly as efficient as habitat C. optimum allocation. In fact the two methods become equivalent when How many sampling units should costs and variability are similar be sampled? Table 3.5 (page 47) between strata. Proportional suggests a sample size of 35 if we allocation can be defined as follows: have a high density roe deer population and we require the A method of selecting samples estimate to be within 20% of the from different strata such that the true total with 95% confidence. (An numbers chosen from the strata are initial site visit to determine an proportional to the population index of population size may be sampling units in those strata. necessary before decisions on required sample sizes can be Example made.)

We wish to estimate the total Therefore 14 samples are taken from number of roe deer faecal pellet habitat A, 14 from habitat B and 7 groups in a study area of 5 ha from habitat C. (50 000 m2) to within 20% of the true total with 95% confidence. If Counts of deer pellets per plot are: we choose a sampling unit of 50 m2 then there are 1000 possible habitat A 3,1,5,4,2,0,3,4,4,5,2,0,5,4 sampling units in total. In this study area there are three distinct habitat B 3,5,4,6,5,2,3,6,4,0,3,6,7,2 habitat types which are known to influence deer use: habitat A covers habitat C 0,2,1,4,2,4,1 2 ha, habitat B 2 ha and habitat C 1 ha. With expected differences in We then calculate the mean and habitat use, stratified sampling variance for each stratum using the would appear to be the most equations described previously. appropriate sampling method. Applying proportional allocation

11 Number of Number of Mean number Variance sampling sample of pellet units plots groups

Habitat A 400 14 3.0 3.08 Habitat B 400 14 3.0 3.85 Habitat C 200 7 2.0 2.33 Total 1000 35

These results are then combined to give an overall population estimate as follows.

The mean number of = Sum (stratum mean x number of stratum sampling units) faecal groups per plot Total number of sampling units

(3.0 x 400) + (4.0 x 400) + (2.0 x 200) 1000

= 3.2 pellet groups

Standard error of the sample mean

Sum (stratum sampling unit proportion2 x stratum variance/number of stratum sample plots)

I (400/1000)2 x 3.08/14 +(400/1000)2 x 3.85/14 + (200/1000)2 x 2.33/7

= 0.30 pellet groups

So the estimated total number of roe deer faecal pellet groups for the 5 ha

= Number of sampling units in the 5 ha x mean number of groups per sampling unit

= 1000 x 3.2

= 3200

With estimated standard error

= Number of sampling units in the 5 ha x standard error of the sample mean

= 1000 x 0.30

= 300 and approximate 95% confidence interval

= Population total ± 2 x population standard error

= 3200 ± 600

= 2600 - 3800

This study has achieved its objective, estimated total of 3200 with 95% The true number of faecal pellet confidence, groups lies within 20% of our

12 Other sampling methods sampling theory described above as In most cases the appropriate choice the size of the sampling area does of simple random sampling or not need to be known, and during stratified sampling will lead to an the survey not all deer need to be efficient, if not optimal, estimation detected. As the perpendicular of the deer population. However distance from an observed animal to there are a number of other the transect line is recorded this sampling methods such as cluster method is more commonly known sampling and two-stage sampling as distance sampling. that could be used in their place. For a fixed cost, the precision obtained Line transects should be placed at by these methods depends, as in the random within the study area such above sampling methods, on both that there is no relationship between the variability between sampling the transect line and the deer, i.e. units and sub-units and the costs of deer are not avoiding or associating sampling them. Details of these with the line. However, when using calculations are not given in this this sampling method for direct Field Book but can be obtained from count census methods, it is usually any sampling textbook; see for necessary to use the existing road example: Cochran, 1977; Levy and network, for practical reasons. This Lemeshow, 1991; Thompson, 1992. provides acceptable results if roads are well distributed across the Key points survey area, and deer are not • Divide the study area into avoiding or attracted to them. These sampling units and sample using line transects can be thought of as the concept of probability sampling. long narrow quadrats where every • The size of sampling unit should deer on or near the centre line of the be large enough to record transect is detected but where deer sufficient individuals but small away from this line will only be enough to increase precision for a detected with a probability that is a fixed cost. function of their perpendicular • In simple random sampling, the distance away from it. The sample size needed to give probability of observing an results to a given precision is individual deer is represented by the dependent on the variability detection function which is between sampling units. estimated from the distances. This • Stratified sampling is preferred to function is used to calculate deer simple random sampling if the density using the total length of the population of sampling units is transect lines, and the number of heterogeneous but can be deer observed within a set distance, subdivided into parts (strata) of the transect line. Further details which are fairly homogeneous. are given in Buckland et al. (1993). • If sampling costs and within strata variability are similar If deer occur in groups then the across strata then sample using distance to the centre of the group proportional allocation. should be recorded together with Line transects the number of individuals in the group. Average group size in the Sampling using line transects population can then be estimated is unlike the finite population and the density of individuals 13 computed by multiplying the require the use of specially density of groups and the average developed computer programs. group size. When using this The program DISTANCE equation it is necessary to check and (available on the Worldwide Web take into account, any influence of http:// www.mwpa.st-and.ac.uk/distance/) group size on the probability of is recommended for these detection of the group. Larger calculations and further details on groups are more likely to be the theory, sample sizes required detected than smaller ones. (This and analysis of distance sampling can be done using the specialist are available in Buckland et al. software mentioned below.) (1993).

If detection rates are likely to Key points vary between different parts of • Transect lines should be the study area then it will be independent with respect to necessary to stratify the area as the distribution of the deer discussed previously: see Stratified within the sampling area. sampling. Densities can be • Perpendicular distances must estimated for individual strata be measured accurately. and combined to give a total • Adequate analysis cannot be estimate of deer. The computation carried out without specialist of the detection function and software. estimation of population density

14 .7

[■i Population Estimation Methods

Population estimation methods, also costs of each method. This is also called census methods, fall into three provided in a small summary box main categories: direct methods, within each numbered method. indirect methods, and the use of cull Practical limitations of application data. Greater confidence is often and potential biases which may placed in methods involving actual influence accuracy are highlighted observations of deer. However, in each method description under studies of marked populations and Advantages/Disadvantages, to help comparisons of different methods the reader choose the most have shown that some of the most appropriate method. frequently used direct observation methods are the least accurate in Direct counts many situations, with only 10-33% of the true population being In some habitats it is possible to see recorded (Anderson, 1953; Langbein, and directly count deer. However, 1996; Ratcliffe, 1987). For all even in open moorland, animals methods the best, i.e. most accurate, resting in hollows may be easily population estimate will be obtained missed, and deer may take flight by sampling across the whole range before accurate assessments of (winter and summer) of the numbers, age-class and sex-class are population being considered. made. Counting may be easier when animals are concentrated together in The choice of method to be used will groups, but this may also make depend upon the following factors: accurate assessment of numbers more difficult when group sizes • Whether an index or an accurate are very large. Dividing the deer population assessment is required. range into sections or a grid, and • Habitat type: in open habitats avoiding counting when weather deer are generally seen more and visibility are poor usually easily than in wooded habitats. increases precision. • Seasonal and daily behaviour patterns which may influence In smaller areas and where it is which sex- and age- classes of possible to recognise individual deer, deer will be found in the area. the total number can be estimated • Terrain which may be more suitable over time by keeping a record of the for some methods than others. different individuals seen on each visit. • The resources available in terms Eventually no new animals will be seen of cash, labour and expertise. (unless new animals are constantly • The timescale in which the entering the population through births information is required. and immigration) and the list of • The limitations of weather. individuals will equal the population size. However, for most deer species The flowchart in Figure 3.1 identifies it is difficult to confidently tell which method(s) are likely to be individuals apart without additional most practicable in any given marking unless very many hours are situation, while Table 3.1 provides a spent carefully observing and quick guide to the performance and recording individual features.

15 In wooded areas it is particularly observations need to be continued difficult to observe and identify until there are 50 separate individuals and there will be observations. animals which are rarely (if ever) seen. To be confident about Methods which allow the population estimates based on unobserved part of the population observations, a large number of to be estimated are more efficient in individual observations are terms of time. Mark-resighting required. Approximately 5 separate techniques are a way of determining sightings of each individual deer are this, but catching and marking deer required before the observer can be is an extremely expensive and time- 90% confident about the total consuming operation, and hence population estimate. For example if these techniques are generally only 10 individual deer have been seen, used for research.

Figure 3.1 Flowchart for selecting appropriate census method 16 Table 3.1 Suitability of census methods to estimate deer population size

1. Open hill ***** ***** *** * ***** 1-7 days 2. Drive counts *** *** *** * **** 1-7 days 3. Static census *** **** **** * **** 1-3 days 4. Vantage point *** **** **** ***** **** 1-7 days 5. Aerial counts **** ***** * **** *** 1-2 days 6. Spotlight ** *** *** **** *** 1-7 days counts 7. Thermal imaging: ***** ***** ** **** ***** 1-3 days direct counts 8. Thermal imaging: ***** ***** ** **** * 3-5 days distance sampling

9. Mark-resighting **** **** * * ** 3-24 months 10. Change-in- *** **** *** *** *** 6-9months ratio counts

11. Impact levels * *** *** *** *** l-5days(aop) 6-12 months (habitat) .

12. Track/slot * *** ***** ***** **** 1-4 days counts 13. Faecal pellet * **** ***** ***** **** 1-4 days index 14. Faecal pellet ***** ***** ***** ** **** 2-3 months clearance 15. Faecal pellet **** ***** ***** *** *** 4r-12 months standing crop 16. Faecal pellet *** **** ***** *** **** 4-12 months strip transects 17. Faecal pellet **** ***** ***** *** * 4-12months line transects 18. Balance sheet ** *** *** ** *** 6-9 months 19. Life tables *** **** **** ** ** 5 + years 20. Cohort analysis **** **** ***** ** ** 5 + years

21. Population ***** ***** *** ** ** 1 + years modelling

Note Methods 10-21 require information from previous culls. Key ***** excellent **** good *** fair ** poor * very poor

Direct counts may be carried out by and may involve few or many day, or by night with the aid of people. The section that follows spotlights or night vision equipment, describes the most useful methods.

17 Daylight counts Each block is subdivided into sections (bounded by prominent physical 1. Open hill counts features such as rivers and high ridges), and each can easily be The Deer Commission for Scotland covered by a census team in a day. (DCS) uses this technique to count Block size depends upon the number red deer and, where they occur on of members in the counting team open hills, sika deer populations (usually 8-10). Experienced counters (Stewart, 1976). can cover 1200-2000 hectares in a day. Method Counts are carried out in late winter The whole of the red deer range in and early spring (January-April) Scotland is divided into 49 discrete when the deer have been driven blocks (Figure 3.2), each encompassing onto low ground by snow cover. the winter and summer ranges of They are usually in their poorest both stags and hinds. The perimeter physical condition and so do not of each block should have natural or move far during the day. Classif­ man-made barriers sufficiently ication is also easiest at this time. substantial to ensure there is relatively little interchange of animals Team members should have binoculars between neighbouring blocks (e.g. (minimum of 7 x magnification and deer fences, lochs, roads and railways). 30 mm objective), a telescope

Figure 3.2 Deer Commission for Scotland: red deer count areas. Blocks 31-36 and 49 are unsuitable for open hill counts

18 (x 20-25), a pocket radio, a notebook should be checked and collated at and a 1 inch (1 : 50 000) Ordnance the end of each day, and any Survey map with their own and possible double counts eliminated. adjacent areas clearly indicated. As Counting is continued until the the count team move through the whole block is completed, and the sections they observe and record the results from each day are collated to number of stags, hinds and calves in provide an overall estimate of each group, the time of observation, population size, age-class and sex- the direction of movement and any class ratios for the block. particularly recognisable animals (outstandingly good or bad heads, Data recorded Numbers of hummels or injured animals). animals, by sex- and age-class, in Movements of animals into each group, time observed and neighbouring sections should be direction of movement. reported by radio to neighbouring counters to reduce the likelihood of Equipment required Binoculars, double counting. telescope, pocket radio, notebook, compass, 1 :5 0 000 Ordnance Survey On open ground where deer can be maps of area. observed easily without disturbance and without causing them to group H & S considerations Ensure team up with other animals, counters can members suitably dressed for work individually. However in prevailing weather conditions (risk broken ground or scrub woodland of exposure in open habitats in late they should work in pairs: one winter). observing and recording from a suitable vantage point, the other moving the deer. When disturbed, Advantages individual deer tend to herd and this • Suitable for very large areas. facilitates counting. During periods of • Probably one of the most reliable severe weather when large methods. numbers of deer (up to 1000) tend to • Species composition, sex- and herd together on lower ground, age-class ratios can be determined. counters should work in pairs, preferably splitting the deer into Disadvantages groups of approximately 200 to count, • Only applicable to open or and cross-check each other's results. mainly open ground. • Requires good visibility. Counters should work down or • Only a minimum population size across wind so that disturbed can be estimated. animals tend to move onto areas • Precision cannot be determined already counted. The finishing line and influenced by observer bias for the day's count should be kept as in sex- and age-class classification short as possible to reduce the (-10% red deer stags and up to likelihood of deer already accounted 30% calves misclassified as hinds; for moving onto the adjacent area to Lowe, 1969). be counted the next day or vice • Results specific to count days, versa. The following day's count is and liable to influences of started at the previous day's seasonal changes in climate and finishing point. Counters' records deer behaviour.

19 • Large numbers of people Table 3.2 Results from open hill required for a number of counts consecutive days. Subarea Stags Hinds Calves • Requires careful, detailed organisation. a 338 353 156 • Blocks are counted only every b 2 4 1 5-10 years. c 28 41 16 d 96 178 67 Performance as an estimate ***** e 8 11 5 f 2 16 9 Performance as an index ***** g 50 199 82 Inexpensive equipment costs *** h 6 0 0 Inexpensive labour costs * Total 530 . 802 336 Simple data analysis ***** Example 2 Data collection period 1-7 days Larger areas may require sub­ dividing into 25 or more subareas. Example 1 Even with large numbers of DCS An area of open hill approximately and local stalkers it is often not 10 x 30 km was counted over two possible to count such blocks within consecutive days during January by a few days and inclement or a 10-man team of DCS staff and local unsuitable weather will often delay stalkers. It was divided into eight or interrupt counts such that a 30 x subareas (a-h). The results are 50 km block may be counted over shown in Table 3.2. There was wide 8-10 days within a 30-day period. variation in the number of animals Care must be taken to ensure that counted in each block with almost animals have not moved from 50% of .the whole population being previously counted to uncounted counted in block a. The total areas. One such area was counted population estimate for the whole during March-April 1994 and totals area was 530 stags, 802 hinds and were 3825 stags, 6925 hinds and 336 red deer calves. 1325 red deer calves.

20 2. Drive counts Individually recognisable animals are recorded, as a means of identifying Method specific individuals or groups and The area to be counted should be reducing double counting. clearly identified on 1:10 000 map. Large numbers of observers are The operation should be co-ordinated positioned around and within the from a high spot providing as wide area to be counted, such that each a view over the area as possible to person only views in one direction enable the beat line to be directed and is able to see to the next via pocket radios, particularly observer (Figure 3.3). Each observer through woodland habitats. At the has a map specifying the area they end of the operation all records are to observe and the position of are collected and the minimum neighbouring observers. A beat population size calculated, using team then moves gently through the details of individual animals, time of area to move animals out of cover to sighting and direction of movement enable them to be counted. Beaters to ensure that double counting is should be spaced about 10 m apart minimised. and be able to see the person either side of them. Observers record This method is more successful for details of animals moving out of small woodland blocks and for cover as specified, including time larger species such as red deer than and direction of movement. for the smaller species (roe and Members of the beat team also muntjac) which are more difficult to record animals seen moving back flush from cover or classify when past them, to one side only. moving at speed.

■ ■ ■ ■ *------Direction of beat Open moor , Observer position

Figure 3.3 Map of drive count area showing positions of observers and direction of beat: 14 refers to observer number 14 (see Figure 3.4)

21 Data recorded For each observer and especially when no antler growth beater, numbers of animals seen, by present; Winder, unpublished). species, sex- and age-class, time of • Results specific to count day, liable observation and direction of travel. to influences of seasonal behaviour, weather and disturbance prior to Equipment required Maps of the area, and on the day. indicating observer positions and • Large numbers of people (60-120) viewing area (Figure 3.3), and record required as observers and beaters. forms (Figure 3.4); radios and • Requires careful, detailed whistles for beaters. organisation. • Large numbers of beaters are H & S considerations Potential for difficult to co-ordinate in thick beaters to become lost in thick cover, cover and undulating terrain. particularly in large areas. Bright Performance as an estimate *** clothing and roll-calls reduce risk. Performance as an index *** Inexpensive equipment costs *** Advantages Inexpensive labour costs * • Suitable for small or large areas. Simple data analysis ***» • Suitable for open ground and woodlands. Data collection period 1-7 days • Species composition, sex- and age-class ratios can be estimated. • Completed in 1 day. Example This method has been used to Disadvantages estimate the size of a population • Requires good visibility on the day. of red deer in thicket conifer • Deer are very difficult to flush woodland in south-west England from cover (even with dogs) (Winder and Chanin, unpublished). particularly adults; and juveniles A minimum of 56 observers plus remain in cover for the first few 40 beaters were used for an area of weeks of life. 3 km2 (Figure 3.3). Thirty-nine • Only minimum population size individual hinds and 13 stags (plus estimated. 1 roe deer) were seen, giving a • Precision cannot be determined minimum population estimate of and accuracy is influenced by 52 deer in 3 km2 or 17.3 deer km'2. observer bias in age- and sex- (Results from faecal pellet counts class classification (e.g. young in the same area suggested a males recorded as females population density of 13 km 2.)

Figure 3.4 Drive (J. Holmes. View Left ei rigdt along track. count record form. RED Roe Obs point: observation Obd CP4 Deer point time ttoch M F calf M 1 F 1 Kid 1 (?) point; cps : cows, ? (?) 13*50 6 1 2 Seen in wood ponies, sheep; deer? : 1 4 ft r- 14-00 3 '2 f\ Seen leavfno uoocf. species unknown; (T): 14-05 I v 1 ...... sex- or age-class ; unknown crossing stream. All same animals lc fr I unknown extra

22 3. Static census Advantages • Can be used in all habitat types. Method • Species composition, sex- and This method is similar to drive counts age-class ratios can be estimated. in that observers are placed at • Completed in 1 day. locations around and within the area (woodland block) to be counted prior Disadvantages to the periods of greatest deer activity • Requires good visibility. (dusk and dawn). They observe and • Only minimum population size record deer as they move out of cover can be estimated. to feed. Observation periods should • Accuracy influenced by observer be a minimum of 2'/2-3 hours to bias in age and sex classification. ensure that animals which may have • Results specific to 1 day and liable settled down to ruminate just before to influences of seasonal behaviour, the observers arrived will have weather and disturbance levels started feeding again before the prior to and on the day. observers depart. Observers record • Many observers (30-40) required. sex- and age-class details for all deer • Requires detailed organisation. seen, their position, time of observation and direction of Performance as an estimate *** movement, and any unusual features Performance as an index **** (to enable double counting to be Inexpensive equipment costs **** eliminated). The total number of Inexpensive labour costs * animals seen (minus any records Simple data analysis **** which are considered to be double counts) indicates the minimum population size. Particularly recognisable animals should be Example recorded as a means of reducing double counts. The method is most A static census in a large forest in suited to early spring (March/April) North Wales involved 34 observers. when vegetation and cover are They were positioned before dawn limited allowing easier observation of and collected 2'A-3 hours later. The the deer, but it is generally not time, position and direction of recommenced as a method of movement was recorded for all deer assessing overall population size. seen (does 1 year+, bucks, fawns and unidentified). Conditions were ideal Data recorded Species, sex-class, with calm, dry, cool weather. A age-class, time of observation and minimum of 131 deer were seen (95 direction of travel for all animals seen. does, 14 fawns, 8 bucks and 14 unclassified), but as only 25% of the Equipment required Maps of the forest was visible to the observers this area, indicating observer positions was estimated to represent at least and viewing area, record forms and 524 deer (19 km'2). Faecal pellet group binoculars. counts carried out on the same day by the same people, in all habitat types H & S considerations Ensure obs­ except thicket conifer, indicated ervers are suitably dressed for 629-895 deer (22-25 km'2) had been prevailing weather (risk of exposure using the area over the previous in wintry conditions). 5 month period (decay length).

23 4. Vantage point counts contiguous blocks (40-100 ha) which can be viewed from’ a suitable This method has been used vantage point, generally from across successfully to determine densities of or above a valley (Plate 1). At the red, sika and roe deer in coniferous vantage point the observer first forests in Scotland (Ratcliffe, 1987; clearly marks on the map the area Ratcliffe and Mayle, unpublished). which can be readily seen and As it involves viewing down into an blocks out areas which cannot be area from a high or 'vantage' point it viewed into, such as small blocks of can only be used in hilly areas. different age structure and dead ground in the lee of ridges Method (Figure 3.5). The area is first identified and stratified (see Sampling in Section 2) The delineated area is methodically into habitats. Randomly selected scanned using binoculars (7-10 x 50 representative areas of each habitat, mm) for any deer movement. A suitable for the technique, are telescope (15-60 x 60 mm) is used to identified. These should be large classify any deer located into age-

Figure 3.5 Vantage point count map for Abhain Bheag, Carradale. Numbers 1-10 refer to observed deer (see Figure 3.6). Arrows indicate direction of movement

24 and sex-classes. This is possible up area. Population size can then be to distances of 1 km for red or sika estimated from habitat specific deer assuming visibility is good. Roe densities and knowledge of the area deer can only be confidently of each habitat in the forest. classified up to 0.5 km away. All deer observed and their direction of Vantage point counts should be movement are recorded on the map carried out when there is limited and the record form (Appendix 3). ground vegetation (March to May). From a good vantage point deer can Where there is very little be seen moving through even disturbance to the deer during the extremely dense habitats, although day, particularly in large woodland this may be for only short periods of blocks, it may be possible to observe time as they cross open spaces (such deer moving at most times of the as areas of check, rides, drainsides) day. However, peak deer activity within the area. Deer moving into the periods tend to be around dusk and area from blocks which had initially dawn, particularly where human been identified as not capable of disturbance is high and so the most being viewed into should be noted suitable periods for observing deer (to prevent double counting) but not are 06.00 - 10.00 and 16.00 - 21.00 included in the final estimate. hours.

Watches should last at least 2'h hours If the vantage point is on a ride or to ensure that animals which had track, viewing may be possible from settled to ruminate just before the a vehicle. (Care must be taken to observer arrived are rising to feed ensure that the view is not restricted again before the watch is completed. by the vehicle's internal structure, At the end of the count period the such as window frames.) minimum number of deer seen (taking account of the possibility of Data recorded Species, sex- and observing the same animals more age-class, and direction of move­ than once as they move in and out of ment of all animals. cover) is related to the area being observed and expressed as a density Equipment required Binoculars, per square kilometre. Three or four telescope, map of area (1:10 000). counts should be carried out for the same area over consecutive mornings H & S considerations Ensure and evenings. The maximum observer suitably dressed for number of deer recorded in the prevailing weather conditions (risk habitat for any single observation of exposure if viewing from outside period should be used as the habitat a vehicle in late winter). Preferably specific density estimate, particularly two observers present. for sedentary species such as roe deer. Ideally other areas Advantages representative of the same habitat • Species composition, sex- and structure should also be sampled and age-class ratios can be estimated. the habitat specific density calculated • Seasonal deer use may be as the mean of the densities indicated by carrying out counts determined for each representative. in the same area during different This should be repeated for each of seasons. the different habitat types in the total • Only 1-2 observers required.

25 Disadvantages Example • Only suitable in hilly terrain. Vantage point counts to estimate • Requires good visibility. sika deer densities and population • Generally only a minimum size were carried out in an area of population density estimated. east Scotland, during 1988. Figure • Precision cannot be determined 3.5 shows an area of thicket unless a number of representatives observed during one count period; of each habitat are sampled. Figure 3.6 is the completed vantage • Bias may be introduced if the area point count record form. During the observed is not typical of the full 3 hour observation period 14 habitat type being sampled. individual sika deer were seen: • Results are specific to count days 5 adult males, 8 adult females and and liable to influences of 1 unclassified. The area observed seasonal changes in climate and was 47 ha, so that the density deer behaviour. recorded was 29.8 km'2. Three areas *** Performance as an estimate of thicket were observed during the Performance as an index **** morning and evenings over a 2 day Inexpensive equipment costs **** period. Estimated densities varied Inexpensive labour costs ***** from 12.5 to 29.8 deer km'2. Simple data analysis ****

Data collection period 1-3 days

Forest Csjrad'aJc' Conservancy ...... Observers) .. AHO...... Location Abhain.Bhea^ Area ..4.7 hu Structural Type .Thicket. Yr “I'."'. Date 1^88 TimconQZ:4.Q Timeoff J.0.-.40.. Weather Conditions and Visibility Poor, at start..

DEERSEEN Adult Adult Yearlings Yearlings Call/ Time Location Uodaulfied Remarks Spp Male Female Male Female Kid

7-4R 1 SIKA 1 8-10 2 ■ 1 8-34 3 ■■ 1 Visibility imprevi/iy 3 10 4 " 1 1 Stag laging down 3 1 5 5 ■■ 2 q -2o 6 4 Groups 5qb camt together 3 |a! Up burnsfde 3-36 male. 3 seen again fVedint) 354 7 “ 1 10-03Group4 nowon feel feeding (005 0 1 very pregnant 10-27 q 1 Dear h/o.2 also up 5 10-35 10 feeding

5 8 1 = 14 /4-7ha /

= 2 4 8 /too ha

Figure 3.6 Vantage point count record form for density estimation of woodland deer in hilly terrain

26 5. Aerial counts Advantages • Large areas of deer range can be Method covered in a relatively short Where deer are resident on open period of time. ground, or mainly open habitat, it • It is possible to view down into may be possible to observe them open woodland habitats, readily from the air using either a particularly when using a light aircraft, microlight or helicopter. microlight, without disturbing The observer(s) counts animals seen the deer. as the aircraft flies predetermined • Estimate of minimum population transects over the area. The width of size is likely to be better than for the piece of ground viewed depends visual methods from the ground. upon the height at which the aircraft • Species composition, sex- and age- is travelling and how far from the class ratios can be determined if aircraft the observer(s) can see deer can be seen sufficiently (delimited by fitting streamers to the clearly. wing struts). Therefore any change in • Only 2-4 people required. flight height needs to be corrected for in the calculations. The use of thermal Disadvantages imaging equipment and videos or • Most suited to flat or rolling photography to record images terrain (see above). detected (for later checking) is • Requires good visibility and advised as this enables double flying conditions. counting to be eliminated. The • Accuracy influenced by height method is described in detail by of aircraft above the ground, Sutherland (1996) and Norton- and tendency of deer to bunch Griffiths (1978). Detection of animals and run off wildly, making from the air is better following snow assessment of numbers, age- and cover as they show up more clearly sex-classes difficult. (Microlights both thermally and visually. The cause less disturbance.) method is most suitable for large deer • Accuracy may be influenced by populations, and is not recommended the presence of other animals, e.g. where the area and population are sheep (see Example below). small. • Results specific to day of count and liable to influences of Data recorded Species, sex- and seasonal changes in climate and age-class, and direction of travel of deer behaviour. all animals seen. • Hire of aircraft may be expensive and operation requires much Equipment required Aircraft and planning. pilot, map of area with transects clearly marked, 1-2 observers, **** photographic /video equipment or Performance as an estimate thermal imaging equipment. Performance as an index ***** Inexpensive equipment costs * H & S considerations Flying may Inexpensive labour costs **** be dangerous in some terrains Simple data analysis *** particularly where weather is unpredictable. Data collection period 1-2 days

27 Example open hill counts (see 1) in the study areas. At the first site, 137 deer were The Deer Commission for Scotland, seen from the air while 136 were in collaboration with Scottish Natural Heritage, investigated the counted on the ground. At the use of helicopter mounted thermal second site, 2 observers counted 428 imaging equipment for aerial and 452 deer from the air while the surveys at two sites during 1993 ground count was 404. The (Reynolds et a l, 1994) and found discrepancy was believed to be due close agreement between total count to sheep being misclassified as deer numbers for this and traditional in the thermal imaging results.

78 Night counts To date, imagers originally designed for weapons targeting applications Deer tend to move out onto open have been found to be best suited areas to feed at night and are for deer density estimation (see often more easily approached, Appendix 5 for suppliers). These particularly when the observer is in are portable and provide good a vehicle. Night vision equipment image resolution and magnification. can be used to observe and count Less expensive night vision deer in areas where daytime equipment, including low-cost observation is not possible. Low thermal imagers, spotlights or image cost night vision equipment, such as intensifiers, are less likely to reveal spotlights and image intensifiers, partly concealed deer, reducing the have a limited application as their number of detections and increasing shorter detection distances will tend the risk of disturbing animals prior to lead to more disturbance to the to detection. Such equipment is not deer, which will influence the recommended. population estimate. In view of the fact that thermal Thermal imaging energy will not penetrate foliage, an Thermal imaging equipment (Plate estimating technique such as 2) is sensitive to the long-wave distance sampling (8) is the most radiant energy emitted by warm- appropriate method to use in bodied animals and their woodland. However direct counts surrounding habitat, revealing can be used in the following surface temperature differences with situations: varying brightness on a viewfinder or video screen (Plate 3). Modern • When surveying open areas thermal imagers are sensitive to (fields, parks or moorland). very small temperature differences • When a large proportion of the (<0.1°C), making deer particularly population emerges from conspicuous even if partly woodland to use nearby open concealed. In the open, deer may be ground at night. visible at distances of up to 2 km, with species and sex differences This arises in some red, fallow and becoming increasingly discernible at occasionally roe deer populations. closer range. A decision to base an estimate on counts in this instance should only Thermal imaging equipment can be be made if the woodland has been used hand held from the ground or a extensively surveyed and a large vehicle, or may be mounted on an proportion of the deer (>75%) was aircraft. Since foliage will conceal observed outside the wood. many deer, aircraft mounted systems are more suited to open In practice, most areas to be areas or woodland with a sparse surveyed include a proportion of canopy. An imager used from the woodland and open ground and ground provides a lateral view population size is best estimated giving relatively better penetration using a combination of both distance into most stand types. sampling and direct count methods.

29 6. Spotlight counts Advantages • Relatively quick method. Method • Low labour requirement (2—4 Deer moving out onto open areas at people). night can be counted, with care, • Low cost equipment. with the aid of spotlights, which pick out the eyes of any animals (up Disadvantages to 300 m away) facing the observer. • Only suitable in areas with a good Deer passing close to the observer system of rides, roads or tracks. (up to 50 m) may be seen sufficiently • Requires good visibility. clearly to sex or age them. Usually • Only minimum population size the area to be counted is identified can be estimated and accuracy and visited at night with one person probably poor. driving the vehicle (on sidelights to • Difficult to determine age- and reduce disturbance), another operating sex-class. the spotlight (either roof mounted or • In areas where night shooting handheld) and a third recording all occurs deer may respond by fleeing observations. If a sampling method rapidly at the sound of a vehicle. (such as distance sampling) is used • Results specific to day of count then population density and and liable to influences of confidence limits may be seasonal changes in climate and determined. However, the method is deer behaviour. only suitable in areas where deer are undisturbed at night, as they will Performance as an estimate ** flee rapidly at the sound of a vehicle Performance as an index *■** or glimpse of a light in areas where Inexpensive equipment costs *** night shooting (legal or otherwise) Inexpensive labour costs **** occurs. This method is limited in Simple data analysis *** use and therefore not recommended. Data collection period 1-7 days Data recorded Where possible, species, sex- and age-class of all Example animals seen. Perpendicular distance to observed deer (if Spotlighting was used in a lowland distance sampling). broadleaved woodland as a means of estimating kid:doe ratios Equipment required Spotlight, during September-October 1979 vehicle with roof hatch or open and January-March 1980 (Rowe, back, 1-3 observers / assistants, map unpublished). During both periods of the area and record forms. too few individuals were seen to provide good estimates of kid:doe H & S considerations Risks asso­ ratios, in part due to the fact that ciated with driving at night with 30% of all deer seen could not be sidelights only (to reduce classified as male, female or kid, and disturbance). Risks can be reduced it was not possible to distinguish by reconnoitre of the route during between adult and yearling females. daylight to identify hazards such as No attempt was made to determine low branches and potholes. population size based on these observations.

30 7. Thermal imaging: direct Advantages cou n ts • Accuracy high. • Species composition can be Counts may be made from a number estimated. of locations through or around the • Sex- and age-classes can be survey areas. Sites should be chosen estimated if males in velvet. that offer the most extensive view, • Group sizes can be estimated. bearing in mind the range of the • Nocturnal behaviour can be imager and any hedges, hollows or observed. convex slopes that may conceal deer. • Camouflaged animals easily At each location, it is important detected. to mark on a 1 : 10 000 map the • Open areas can often be viewed area visible. from public highways.

If the total area to be surveyed is Disadvantages relatively small and all locations can • Not suitable within woodland. be assessed with little risk of deer • The availability of thermal moving between them, then the imaging equipment is currently estimated population size is simply limited and costs are very high: the sum of all deer counted. If up to £40K (see Appendix 5 for however there is the possibility of suppliers). movement between areas, then • Usually limited to small woods or population size should be estimated areas with little woodland cover. from the mean density of deer seen • Results specific to count days and on each area multiplied by the total liable to influences of seasonal area. It is advisable to repeat the changes in climate and deer exercise on subsequent nights, and behaviour, unless total deer home take the mean of each night's range is sampled. estimate. It is important to try to • Distinguishing between similar eliminate observer disturbance species and shapes (sheep and when moving between areas. deer) can be difficult.

Data recorded Number, species, Performance as an estimate ***** sex- and age-class of all animals seen. Performance as an index ***** Inexpensive equipment costs ** Equipment required A high re­ Inexpensive labour costs **** solution thermal imager. For vehicle censuses a 4-wheel drive vehicle Simple data analysis ***** with a roof hatch is recommended. A compass with a luminous dial is needed to take bearings.

H&S considerations Particular care should be taken when walking through woodlands at night while viewing through the thermal imager. Routes on steep or unstable ground should be avoided by vehicles. All routes should be checked for hazards in daylight immediately prior to sampling.

31 8. Thermal imaging: distance contiguous areas that differ in sam p lin g vegetation structure as separate strata (see Stratified sampling in Method Section 2), and calculate a detection The estimation of population function for each. However, it is density using distance sampling usually only necessary to separate relies on the calculation of a detection forest areas from any open ground function from the distances immediately adjacent to the measured between groups of deer forest. and the transect line. The detection function defines the probability of It is important to ensure that the detecting deer at a given distance extent of possible diurnal and from the transect, and makes it seasonal movements are considered possible to estimate density from when planning a survey. Transect the number of deer encountered routes surveyed on any one night along the route sampled. In view should encompass areas known or of the fact that visibility depends likely to be used both day and night on vegetation density, it may be by the deer, to avoid the risk of over- necessary to treat any large or under-detection.

Figure 3.7 The distribution of transect routes to sample the Mundford block. -- Forest rides/roads used as transect routes; these should be well distributed across the survey area

32 It will be acceptable to survey only improved by subdividing the area summer or winter ground if it is safe into strata and sampling each area to assume there are no deer remaining separately. Transect routes may be on the unsurveyed habitats at that any length (300-2000 m is ideal) and particular time of year. If not, it will should start and end at the forest be necessary to survey both summer edge or at road/ride intersections. and winter habitats. The use of two or more routes in close proximity should be avoided The deer population range and area on the same night, to avoid the risk to be surveyed should be clearly of disturbing deer from a previous identified on a map on which the route. There should be 10-25 transect transect lines should be marked routes within each woodland (Figure 3.7). For practical reasons (precision will tend to increase with transects should follow existing the number of transects sampled). forest roads and paths. It is All routes should be marked and important that the routes surveyed clearly identified on a 1 :10 000 map are representative of the forest as a of the area. whole, i.e. that they pass through the range of stand types that occur in Transects should be followed either the survey area. on foot (walking into wind and without using a torch) or using a Sampling intensity needs to be high vehicle (using only sidelights and at enough to ensure that areas of high 5-15 kph), taking care to minimise density are not over- or under­ disturbance. Vegetation either side represented in the sample. To of the transect should be scanned for achieve this the minimum ratio of the presence of deer. When passing ride/road length sampled (in km) to dense vegetation vehicle speed forest area (in km2) should be at should be kept to 5 kph and scans least 2.5 although more is prefer­ repeated approximately every 5 able. Roads /rides should penetrate seconds. Where the route passes all areas of the forest, with no large patches of dense vegetation it area >~300 m wide or >~10 ha is usually easier to focus on one side unpenetrated. This effectively of the vehicle, and return to the ensures that about 30% or more of other side of this part of the route on the woodland area is sampled. The another night. In this situation the density of roads/rides sampled can part of the route sampled in such a be reduced in larger forest blocks way is considered as half the length. (-2000 ha) provided routes are well Scanning from back-front gives distributed throughout the area. In more uniform coverage than from this event, precision is likely to be front-back.

Table 3.3a Average body length of each species!sex used to estimate distance (cm) — Females Unsexed Fallow 135 125 130 105 Muntjac 80 80 80 65 Roe deer 100 100 100 80 Red deer 170 150 160 120 Sika deer 140 125 132 105 *

33 For each group of deer observed, the away in reaction to the observer and perpendicular distance from the preferably on an animal at the centre transect line should be estimated of the group. If a side-on view and the number of deer counted and proves impossible then use the map classified by species, sex and age, method described below or measure where possible. Distances should be the width across ear tips and assume estimated before the deer move away, that this is a proportion of body or if they have already moved, the length (see Table 3.3b). distance to their position before they took flight should be estimated. If Table 3.3b Width across ears as a pro­ the group is scattered, the distance portion of nose-tail length (%). to an animal close to the group Ear width measured when facing either centre should be estimated. directly towards or away from observer Distances should be measured by one of the following two methods. Fallow 30 Rangefinder Muntjac 29 Measure length of body, nose-rump, Roe deer 25 standing side on, against Red deer 28 rangefinding marks (Figure 3.8 and Sika deer 27 Table 3.3a). This should be measured before the group moves

Distances can be estimated if thermal imager is equipped with an internal rangefinder: Distance (m) = k x Object length (m) Length in viewfinder The value of k depends on the type of imager and lens being used and is calculated from the angle between rangefinding marks (0): *=[2tanf I''

In this example, using a Pilkington life imager with a x9 lens (0 = 0.57°) and assuming body length is 1.00 m for an adult roe deer, the distance is:

100.5 x = 48 m 2.1

Figure 3.8 Distance estimate through rangefinder

34 Detection distance (yj) is calculated Map from average body length (Table Mark the position of the group 3.3a), body length as measured and the observer on a 1:10 000 through the rangefinder, and the map as accurately as possible. angle through the rangefinding Perpendicular distance can be marks. measured from the map afterwards. Where the road bends, the shortest The compass bearing0 of the group perpendicular distance should be from the observer (<|)i) and the taken (see Figure 3.10). This method bearing0 of the direction of the route is preferable if the group is distant (200 m), the route bends or body made must also be recorded. The length cannot be measured. perpendicular distance is then equal to sin((f>i-cj)2 )yui (Figure 3.9). A Species may be identified from body computer program for converting proportions (leg:body length), alarm body lengths and angles into call, group size, antler shape and perpendicular distances is available gait, particularly when taking flight. from the authors.

%

Figure 3.9 Calculation of the perpendicular distance from a deer to the transect line

Perpendicular distance estimated Observation line

Figure 3.10 Measurement of the perpendicular distance of observed deer from a transect route when plotting deer on a map: three typical scenarios

35 It is more difficult to identify species as observation distance increases, Data recorded Number, species, particularly >500 m. Sex can only sex- and age-class of all animals in a usually be distinguished reliably group, and distance of the group when males are in velvet. Closer from the transect line. approach usually aids identification, Equipment required For wood­ but should only be attempted after land censuses, a high resolution the distance from the group has thermal imager with built in been estimated. rangefinder (for distance sampling). (In daylight an optical or laser Each road/ride should be surveyed rangefinder may be used.) For 1-3 times, on different nights, vehicle censuses a 4 wheel drive preferably covering the same length vehicle with a roof hatch is each time. The length surveyed recommended. A compass with a should be measured from the map luminous dial is needed to take (to the nearest 100 m) and recorded bearings, and two-way radios for on the census form (see Appendix driver-observer communications 4). Transects should not be sampled are also recommended. in bad weather (storms, high winds, heavy rain or falling snow) when H & S considerations Particular deer will make more use of cover care should be taken when walking and be less visible. through woodlands at night while viewing through the thermal For an area of -1000 ha it should be imager. Routes on steep or unstable possible to carry out a census over ground should be avoided by 5-8 nights by foot or 2-3 nights by vehicles. All routes should be vehicle. checked for hazards in daylight immediately prior to sampling. To estimate a 'detection function' with sufficient accuracy, a sample of Advantages approximately 50 groups is required. • Accuracy and reliability good if However 70 is recommended. assumptions valid. Where insufficient observations are • Species composition can be obtained (in small woodlands or estimated. where deer density is low) data may • Sex- and age-class ratios can be be pooled from other areas to derive estimated if males in velvet. a common detection function. • Low labour requirement (1-2 people). Deer population density is • Areas of high deer use can be estimated using the program identified. DISTANCE (available on the • Nocturnal patterns of behaviour Worldwide Web http:/www.ruwpa. and habitat use can be observed. st-and.ac.uk/distance/). The data analysis and calculation of Disadvantages population densities is explained in • Woodland must have an Gill el al. (1997). extensive ride/road network >2.5 km km'2, which is not associated with deer habitat use. • Requires the use of specialist computer software to analyse 36 data and calculate population Thetford Forest during April/May density. 1996. Transects within the woodland • Results specific to count days and block followed forest roads, tracks liable to influences of seasonal or firebreaks (Figure 3.7). Twenty- changes in climate and deer nine transects totalling 61.6 km were behaviour. sampled. The use of transects in • Training in the use of equipment, close proximity (<500 m) on the sampling method and analysis same night was minimised to reduce are recommended. the risk of detecting deer disturbed • The availability of thermal from a previous route. Routes were imaging equipment is currently travelled two or three times if limited and performance necessary, until sufficient data had and prices (£7000-£40 000) of been obtained (>50 observed equipment are variable (see groups) to calculate a population Appendix 5). density. As deer also used fields • Bias may occur where deer adjacent to the woodland block are avoiding or attracted to these areas were surveyed and the roads /rides. This is detectable in results added to the estimate the data analysis. obtained from distance sampling.

Performance as an estimate ***** All observed groups of deer were Performance as an index »**»» classified by species, sex- and age- Inexpensive equipment costs ** class; group size and distance from the observer were estimated. The Inexpensive labour costs **** data were analysed using the Simple data analysis * procedures outlined by Buckland et Data collection period 3-5 days al. (1993), using their software 'DISTANCE'.

Example The overall population estimate for the Mundford block was 23.2 knr2 Thermal imaging was used to (95% confidence interval 17.9 - estimate the size of the deer (red, roe 30.0 knv2). and muntjac) population in the Mundford block (-1200 ha),

37 Other direct methods Various models to analyse mark-recapture data and estimate The following methods can be used population size are available to estimate population densities, or (Greenwood, 1996). Although to provide an index of population applicable to both open and closed size. However, they are of limited populations open population use to practical deer managers due models produce less precise to the resources required to achieve estimates of population size as they sufficiently large sample sizes or the make fewer assumptions. However, low accuracy levels achieved. although more precise, if closed population models are applied to 9. Individual recognition open populations they will provide (mark-resighting) biased estimates (generally higher). However if the period between Method capture and resighting is sufficiently In populations where it is possible to short and at a time when it is individually recognise a sufficiently reasonable to assume that there has large number of the population been no loss or gain of animals to (from either natural markings or the population from emigration, capture-marking programmes) the immigration, births or deaths then population size can be estimated closed population models can be from the proportion of recognisable used. The use of radio-telemetry will individuals in the observed sample. also assist in determining the In most deer species it is difficult to number of marked animals in an observe and recognise individuals open population. unless many hours are spent observing the population Where one cannot be sure about (particularly in woodlands), and so whether the population is open or animals require capture and closed it is better to use an open marking for this method. Care must model as an imprecise but unbiased be taken that marking effort is estimate is generally preferred to distributed equally across the whole a precise but biased estimate. The survey area, and that it does not Jolly-Seber model is most suitable introduce bias, by altering for analysing open deer behaviour patterns or mortality populations, but tends to produce levels. If the population is totally imprecise density estimates unless a enclosed, or it can be assumed that high proportion of the population there are no gains (births or is marked (Strandgaard, 1972). immigration) or losses (deaths or Capture-resighting techniques are emigration) during the period therefore most suitable for research between marking and resighting projects, due to the effort and costs then the population is assumed to involved with observing and/ be 'closed'. For most wild deer or marking individual deer. populations these assumptions are Additionally a licence to capture not valid and so the population is and mark deer is required from the considered to be 'open'. statutory conservation bodies.

38 Data recorded Individual deer used the joint hypergeometric identity (number/mark), dates seen, maximum likelihood estimator total number of deer seen on each (White and Garrott, 1990) which is occasion. suited to observation data rather than recapture data. Deer were caught Equipment required If marking and released onto the study area each deer, clearly coloured collars and ear March. Repeat censuses shortly after tags; equipment and licence to catch release were then used (up to 36 per and handle the animals. year) to estimate the ratio of marked : unmarked deer on each day. H & S considerations Risks of in­ Counts from each census day were jury to staff and deer during pooled and population size estim­ catching operations. Additional ated as follows: risks associated with chemical immobilisation. Estimated population size =

Advantages Number of Sum of total • Precision can be estimated. marked deer x deer seen in +1 • Sex- and age-class ratios can be alive all censuses estimated. r Sum of marked deer -i L seen in all censuses +1 J

Disadvantages For example: • Precision dependent upon Total number of deer marked = 80 validity of assumptions, and Censuses liable to bias if assumptions not April May June Sum valid for the model used. Total deer seen 100 79 120 299 • High labour (observation time Marked and deer capture) and catching deer seen 30 40 29 99 and marking equipment costs. • Licence required to catch and Estimated 8Qx3QQ mark adult deer. population size — = 240 deer • Careful, detailed organisation required and staff skilled in The method assumes that the catching and handling deer. population is 'closed', that loss of • Complex software required for marks is accounted for, all animals data analysis. have the same probability of being caught and resighted, and that each Performance as an estimate ***» census is independent. Performance as an index **** Inexpensive equipment costs * At Chedington Wood the population Inexpensive labour costs * size varied from 46 km'2 in spring 1967 to 76 km'2 in 1975, falling again to 34 Simple data analysis ** km2 in 1979/80; 95% confidence Data collection period 3-24 months intervals varied from ± 0 to 12. In most years the percentage marked (of the minimum number alive) in each Example spring was >78%. The lowest Gill et al. (1996) estimated population proportion of marked animals in any size for a marked roe deer population year was estimated to be 64% in 1966 at Chedington Wood, Dorset. They while the maximum was 89% in 1976.

39 10. Change-in-ratio counts 'hinds' in all cases) and data gathered before and after the hind By observing and recording sex, age- cull. class or marked-unmarked ratios, just prior to and following a known The method assumes males and cull, it is possible to calculate a females are seen with equal population estimate before the cull probability, which may not be valid based on the change in ratios and where there are behavioural and known mortality. range use differences between sexes. Sample sizes need to be large to be The method assumes the population confident of the accuracy of the is 'closed' (i.e. that there are no ratios calculated, particularly where gains/losses unaccounted for the change in ratio before and after during the period between surveys). the cull is small. Where the This is valid if the range of a population is not 'closed' (i.e. there population is well defined and the are losses from poaching and cull period between surveys is short. natural mortality), the assumptions are not valid and the population size Method will be underestimated. Conner et al. The sex ratio of the population (1986) discuss sample sizes and is estimated from observations calculation of variance estimates to directly before and after the male determine accuracy. cull and before the female cull (e.g. at the beginning of July and If there are problems in accurately early October for red deer stags). classifying individuals to sex then The population size before the another feature such as cull can be estimated from the with/without antlers can be used. observed sex ratios and known cull Buckland (1992) proposed this between the survey dates using this method as a means of checking bias equation: in open hill red deer counts (1) and suggested that it may be particularly useful for checking on estimates of open hill red deer populations where which have access to forestry blocks and therefore an unknown A N j = Estimated population size proportion of the population may be before the cull. hidden at any given time. Rs = Number of stags removed between surveys 1 and 2. Data recorded Number of animals R = Total number of stags and by species, sex- and age-class. hinds removed between survey 1 and 2. Equipment required Binoculars, Pi = Proportion of stags before telescopes. the cull. P 2 = Proportion of stags after the H & S considerations Risks asso­ cull. ciated with lone working: minimise risks by set daily reporting Population size before the hind cull procedures and provision of mobile can be estimated using the same phones. equation (replacing 'stags' with

40 Advantages of adult males to adult females was • Suitable for large areas. 1:2 (0.33:0.67) and the overall ratio • Sex- and age-class ratios can be (including kids) was 1 male : 1.8 estimated. females (0.36:0.64). • Provides a check on open hill count bias. The population (N) at the time of the • Low equipment costs. first observation was estimated from

Rd-Rx Pi Disadvantages N = • Large sample of observations P.-P2 required (most suited to open habitat). where • Good visibility required. • Precision influenced by observer Rd Number of does culled. bias in sex and age classification. • Precision influenced by seasonal Based on the overall sex ratio counts changes in deer behaviour. (adults + kids), • Two separate surveys required. 74 - (74 x 0.64) *** N = Performance as an estimate 0.67- 0.64 Performance as an index **** Inexpensive equipment costs Based on adult only sex ratios, Inexpensive labour costs *** Simple data analysis *** N _ 53-(53x0.67)_ 21p 0.75 - 0.67 Data collection period 6-9 months The widely differing results indicate Example the problems associated with this method due to seasonal Changes in sex ratio pre- and post- deer behaviour and possible culling were used to estimate the misclassification. During October, it size of a woodland roe deer is possible some animals were living population before the doe cull. On a out on the fields and had not moved day during early October, 20 back to the woodland. The observers were positioned for 3 proportion of bucks seen during the hours around the outside of the first observation may have been woodland block to observe deer reduced due to seasonal behaviour, coming out onto fields to feed. and there may have been bias in age Another two observers were inside estimation during the second the wood. A total of 187 individual observation, with well-grown kids deer were seen in ratios of 1 adult counted as adults. Care was taken male : 3 adult females (0.25:0.75). to ensure no double counting The overall ratio (including kids) occurred; however, it is possible was 1 male : 2 females (0.33:0.67). that this may have led to an underestimate in the number of deer During the doe cull, 53 adults and actually observed, as animals 21 kids were removed. On the assumed to have been seen twice on second observation at the end of the same day may actually have March, 288 deer were seen; the ratio been separate individuals.

41 Table 3.4 Index of deer presence from browsing and grazing impacts (after Mitchell and Kirby, 1990) a J !f ill!! fii! li I I & & 11 1 1 £ 1 1 | "- 3 -13 « d, E i o m C l Hi ■H > i -d tio o> 42 l m J u JS O •a 1 = O K P ■= I S i H I jr I £ 1 eu I

Optimum grazing intensity Indirect methods of 11. Impact levels measuring deer populations Browsing and grazing impact levels may be used to provide an index of Where deer use woodland or other deer presence: high, medium, low concealing habitats for a part or (Table 3.4, opposite). Impact levels most of the time, direct methods of on a site will be influenced by the population estimation are likely to number of deer present and their be inaccurate. Ratcliffe (1987) and feeding behaviour, habitat type, Langbein (1996) suggest population and the availability (presence and underestimates of a factor of 4. whether protected or not) of Indirect methods of determining preferred and vulnerable plant population size and density should species. For these reasons no simple be considered. Signs such as relationship has been found tracks/slots (Plate 4), fraying and between deer numbers and impact browsing, scrapes and faecal pellet levels. There does, however, appear groups have been investigated to be a threshold density for any (Dzieciolowski, 1976), with varying habitat, below which very little levels of success. Usually the area is damage seems to occur, and above sampled to determine a mean level which damage/impact will be of sign presence and this is then noticeable (Gill, 1992). This related to previous information on threshold density will also depend deer density, presence and decay of upon land management objectives. the sign in other known In many situations unacceptable populations. If the presence of the impact levels (to the crop /habitat) sign is not expected to be similar will be the reason for deciding to across the whole site, stratification estimate deer population size. (see Stratified sampling) will be required before sampling begins. Method

Scrapes and fraying stocks. Scrapes Impact levels can be measured by and fraying stocks tend to be determining browsing / grazing associated with territorial marking levels in sample plots taken across and may be more frequent where the whole deer range or within competition for access to females is various habitat strata, depending greatest (e.g. where there is a high upon the detail required and male : female ratio) or in open areas. resources available. Tree crop impact Therefore they will not be levels (% trees damaged) can be dispersed evenly throughout an determined using the 'nearest area. Additionally, there are no neighbour' technique (Pepper, background data indicating the 1998). Other vegetation may be number of scrapes or fraying stocks assessed for the proportion associated with an individual grazed/browsed, or % cover (e.g. by territory. They are therefore not a species, groups), and a score practical method of assessing deer allocated to each factor which can densities and population size, but then be totalled to provide an index are useful in indicating where of overall deer impact. Ferris-Kaan territorial males may be resident. and Patterson (1992) describe methods for monitoring vegetation, and sampling design and layout

43 are discussed in Sampling in Section Performance as an estimate * 2. The number and size of samples Performance as an index *** taken will depend upon the Inexpensive equipment costs *** methods to be used and resources Inexpensive labour costs *** available. Generally, samples should Simple data analysis *** be allocated at random within each habitat stratum. Data collection period 1-5 days crop 6-12 months Data recorded Deer species pre­ habitat sent, type of tree protection (if any), proportion of damaged trees, impact Example scores for other vegetation. In a study of farm woodland Equipment required Record forms, damage, the proportion of trees quadrats for vegetation assessments. damaged by deer or rabbit was recorded separately for each H & S considerations Risks asso­ woodland visited (Mayle, un­ ciated with lone working: minimise published). Protection used on the risks by set daily reporting trees varied between woodlands procedures and provision of mobile and within woodlands with some phones. woods having individual tree protection of different heights and Advantages design for the various tree /shrub • Applicable in all habitats. species planted. Deer browsing • Low labour requirement levels varied from 0% in woodlands (1-2 people). where treeshelters of adequate • Low equipment costs. height (1.8 m for fallow deer) were used to 84% where trees had not Disadvantages been protected against deer • Imprecise. browsing. • Only an 'index' of deer presence can be established. • Seasonal deer behaviour and impacts will cause bias and influence choice of sampling date. • Impacts from sheep, rabbits, hares, etc. may be difficult to separate from deer impacts.

44 12. Track/slot counts Advantages

In areas where there is reliable snow • Suitable in most habitats. cover, deer tracks or 'slots' have been • Relatively quick. used as a means of estimating relative • Low labour requirement. deer densities by comparing the number of tracks entering with the • Low equipment costs. number leaving an area on the next Disadvantages day (Dzieciolowski, 1976). This • Dense, grassy vegetation may assumes that an animal passes through obscure pathways. the same place on consecutive days. • Only of use to provide an index However, as this assumption is not of presence or activity. necessarily true, the method is best used to provide an index of deer • No sex or age classification activity rather than numbers (Pucek et possible. al., 1975). The method has also been • Unreliable as an estimate as main tried in areas where soil type assumption often not valid. (sandy/loam) enables clear imprints or slots (Plate 4) to be formed, by either Performance as an estimate * counting the number of regularly used Performance as an index *** pathways crossing the woodland Inexpensive equipment costs ***** boundary (Plate 5) or the number of Inexpensive labour costs ***** slots formed in prepared ground over Simple data analysis **** a specified period of time. Data collection penod 1-4 days Method

Deer pathways crossing woodland Example boundaries are best counted during Mayle et al. (in prep.) found spring /early summer when ground correlations between faecal pellet vegetation is not too dense and group counts and trackway counts pathways can be more easily seen. The per 100 m in sites with mainly fallow observer walks around the outside of the and roe deer present. Faecal pellet woodland block and records the number decay was assumed to be 6 months of deer pathways crossing the boundary across all sites and for both species, for each 100m (paces). Fencing around providing density indices of 1 to 10 the woodland boundary may restrict km2 for roe and 1 to 22 km'2 for deer access and hence the number of fallow deer. The number of pathways. The average number of trackways per 100 m tended to be 4 pathways per 100 m is then used as an times greater for roe than for fallow index of deer presence. at the same deer density index; i.e. at 10 deer km'2 there were Data recorded Number of path­ approximately 8 pathways for roe ways crossing the woodlands but only 2 for fallow deer sites. boundary per 100 m.

H & S considerations Risks asso­ ciated with lone working: minimise risks by set daily reporting procedures and provision of mobile phones.

45 Faecal pellet counts Estimate of population density and size Faecal pellet groups are one of the most obvious signs that deer are Faecal pellet group counts to present in an area. They are found in estimate deer population density fall all habitat types and can be used into two main categories: faecal either to provide an index of deer accumulation rates (FAR) or clearance presence or an estimate of population counts and faecal standing crop counts density in different habitats from (FSC). The choice of method will which total population size across depend on the level of accuracy the deer range can be calculated. required, the time available, how Faecal pellet group counts are best soon the result is required, and the carried out in spring and autumn. A density of deer in the area. faecal pellet group is defined as a Clearance counts (involving two cluster of 6 or more pellets produced site visits) tend to be more accurate at the same defecation. as variation due to pellet group decay will be either reduced to a Two groups lying close together or minimum or eliminated. The latter one on top of the other can usually will occur where plots are revisited be distinguished by differences in before fresh pellet groups marked colour, size and texture. Groups during the first visit have found on the edge of a transect or disappeared. Clearance counts are plot should be counted or rejected most practical at high deer densities based on the number of pellets in the (above 30 km'2). group lying inside or outside the search area. When groups lie exactly At lower deer densities, standing on the edge they should be crop counts are most practical in alternately included and rejected. terms of effort and accuracy. Three Occasionally individual pellets are sampling regimes can be used: plots, seen but these are often part of a as for clearance counts; transects; group which has been defecated and line transects. Before standing while the animal has been moving, crop counts are carried out decay causing them to be scattered in a line should be monitored in each of the or string ('stringers'). Such habitats to be sampled, and for each individual pellets should be noted of the deer species being considered and care taken when searching to (see Appendix 7). Where a site is follow the line of these pellets across visited regularly (monthly) decay the boundaries of the transect or monitoring can be included during section of the plot being searched. the visit, otherwise separate visits may be required. Prior to fieldwork it is advisable to check that staff involved use the Both methods use information on same search technique, and are faecal pellet group decay (Appendix equally competent at identifying 7) and defecation rate (Appendix 8) faecal pellet groups to species where along with the faecal pellet group more than one species of ungulate is count results to estimate population present (Appendix 6). It will density, and so the overall accuracy probably not be possible to identify of the population estimate will be all groups to species and so 'possible influenced by the accuracy roe/muntjac' etc. categories will associated with the estimate of also need to be recorded. defecation and decay rates, as well as by the sampling regime. Although number of pellet groups per plot; they may be carried out at any time this is influenced by deer density, of the year, faecal pellet counts are daily defecation rate and decay time best done during late winter /early for the pellet groups. A pilot study spring, after snow has melted and or previous information may be before vegetation growth starts. used to indicate expected mean number of pellet groups per plot. Sampling regimes will vary Table 3.5 provides examples of the depending upon the approach to be minimum number of plots required used and the resources available. For to achieve ±20% precision for all methods it is recommended that all varying deer densities and decay habitats (woodland and open) used lengths based on results of previous by the deer population are sampled faecal pellet group count studies. and that the total deer range is first stratified (see Stratified sampling in Where the area to be counted has Section 2) to increase the precision of been stratified, plots should be the final population estimate. As allocated proportionally (page 11) to faecal pellet group decay will also each habitat with a minimum vary between habitats, stratification number of 6 plots in each habitat should also consider variation in type. Plots should ideally be pellet group decay. allocated to representatives of each stratum at random as in the The number of sample plots following example. required depends upon the precision required and resources Example of plot allocation available. When using faecal pellet group counts, deer population size Habitat 1 (total area: 550 ha) is is usually estimated to ±20%. To represented by 15 blocks (a-o) with achieve this, at least 100 faecal pellet individual areas as listed in groups need to be counted in total. Table 3.6. The running total for the If the precision required is ±10%, habitat area is listed beside the plots then the sample size should be and used with the random number increased so that at least 400 pellet table (Appendix 1) to determine groups are counted in total. The within which block the samples number of plots required to ensure should be taken. As the total area is 100 pellet groups are counted 550 ha, we can use the 3 digit groups depends upon the expected mean within the table. Suppose the 13th

Table 3.5 Number of 7 m x 7 m plots required to achieve ±20% accuracy for standing crop counts

Decay length Mean pellet Minimum (months) groups per plot number of plots to sample3

<10 9 1.8 55 10-20 4-6 1.3 77 40-50 3 6.9 15 20-27 3-4 3 35

' Minimum of 6 plots per habitat stratum.

47 row in the table is randomly blocks of each habitat type and selected. The numbers are 797, 844, where deer habitat use is likely to 747, 041, 433, 658, 062, 922, 410, 107, vary little between representative 154, 252, 679, 874, 785, 468, 084, 348. blocks of any given habitat type, As the total area is only 550 ha, the stratified systematic sampling may first three numbers are ignored; the be used, e.g. for roe in first rota­ fourth number, 041, identifies the tion upland forests. A large block in which the first plot is placed representative block of each habitat - block C (running total 45); 433 should be selected and sample plots identifies the block in which the located along a transect line passing second plot is placed - block N through the block, starting at a (running total 460), and so on giving readily located point at the edge of the sample plot allocation shown in the block. The plots should be Table 3.6. Within each block, plots spaced equally along the transect a should be randomly sited using the minimum of 50 m apart. Transects random number table to decide should start at the edge of a block or position within the block (e.g. 79 m compartment to ensure that the north and 551 m east from the south­ habitat edge is included within the west comer). sampling regime. As 'edges' are often heavily used by deer, they In areas where the deer range is should be sampled in proportion to composed of large homogeneous their presence in the area.

Table 3.6 Sample plot allocation (proportional sampling) for habitat I

a 15 15 b 5 20 c 25 45 1 d 20 65 2 e 55 120 1 f 10 130 g 110 240 1 h 25 265 1 i 25 290 j 60 350 1 k 6 356 1 4 360 m 35 395 n 65 460 2 0 90 550 1

48 Plate 1 Example of an area suitable for vantage point count, looking across a valley.[B. M ayle]

Plate 4 Close up o f deer slots in soil. Plate 5 Deer path through bottom 141876] of hedge. [41850] Plate 7 Pellet groups from rabbit, sheep, squirrels and three deer species. [41839] Plate 8 Red deer pellet group. Plate 9 Muntjac pellet group. [A. C hadw ick] [41554]

Plate 10 Roe deer pellet group. [41735]

Plate 11 Fallow deer pellet group. [41481] Plate 12 Sika deer pellet group. [51971]

Sheep

Plate 13 Sheep pellet group. Plate 14 Muntjac latrine. [41551] [41843]

Plate 15 Roe deer faecal pellet group, almost decayed. [42071] 13. Index of deer presence Advantages

Method • Applicable in any habitat type. • Not restricted by weather (other The deer range can be sampled than snowfall). using either a specified number of • Quick method (especially if plots or a specified length of transect transects used). within representatives of each • Low labour requirement (1 person). habitat. Plots are generally allocated • Low equipment costs. at random within the habitat, while the start point and directions of Disadvantages transects may be allocated at • Only an index can be calculated. random or systematically, e.g. • No age or sex classification diagonally SW to NE across a wood possible. (see Sampling in Section 2). Each plot or transect is systematically Performance as an estimate * searched and the number of pellet Performance as an index **** groups present is recorded by Inexpensive equipment costs ***** species, where possible. Where more Inexpensive labour costs ***** than one species of deer or sheep or Simple data analysis **** goats are present it will probably not be possible to identify all pellet Data collection period 1-4 days groups to species (Appendix 6). The results (mean or total number of Example pellet groups per species per transect or plot) can be compared A transect method was used to between habitats within and obtain an index of deer presence in between sites to provide an index of woodlands in East Anglia during deer presence (high, medium, low). 1997. For each woodland visited a transect (minimum length 1 km) was Data recorded Number of faecal walked through the middle pellet groups per plot or transect following the compass bearing which length by species. bisected the wood through its longest axis. All pellet groups found within Equipment required Compass, 0.5 m either side of the line being measuring tape and pegs (to set up walked were identified to species, plots). where possible, and recorded. If the transect was less than 1 km then a H & S considerations Strong dis­ second transect at right angles to the posable gloves should be worn first was completed. The mean when searching the undergrowth for number of pellet groups per 100 m faecal pellet groups. was calculated for each deer species.

49 14. Clearance counts (FAR) marked to enable any decay between visits to be estimated. Ideally six Clearance counts are most suitable fresh groups per habitat type should in sites of high deer density be marked. The procedure should be (>30 km 2) where faecal pellet groups continued until all sample plots have quickly accumulate on the ground. been checked and cleared.

Method All plots should be visited a second The deer range should be clearly time and the number of pellet identified on a map (1:10 000) and groups accumulated since the first stratified into the constituent habitat visit counted. The timing of the types. Sample plots should be second visit should be such that the allocated to each habitat as previously number of pellet groups counted described (page 11). Plots up to 7 m will be maximised and accuracy x 7 m in size can be effectively and efficiency increased. Where searched by one person and 10 m x 10 previous decay data are available m plots are suitable where two people the second visit should be just before are involved. Larger plots are more the expected decay time (but before likely to contain pellet groups on the there is a need to make large second visit. adjustments to the results due to numbers of deer dying naturally Plots should be located from the or culled between visits). By map using a compass bearing from a monitoring the presence of the readily identified point and marked pellet groups for each following this into the area for the habitat it will be possible to decide specified distance. The distance into when the most suitable time for the the area can be paced while the plot second visit will be. Tables should be marked accurately with a A7.1-A7.7 in Appendix 7 indicate stake or peg placed at each corner to decay lengths measured for faecal ensure that it will be easily located pellet groups since 1995 in a range of on the second visit. habitat types. Decay length will tend to be shorter in open habitats The plot should be carefully and and particularly if there has been systematically searched for faecal heavy rainfall, or generally warm, pellet groups, which should be moist conditions. If no decay removed. Dividing the plot into 1 m information is available a period of strips which are systematically 2-3 months is usually acceptable. searched is the most practicable searching technique. If two people On the second visit the marked work together they should start at groups should be examined to diagonally opposite corners and determine whether there has been both search the whole plot. The any loss of pellet groups due to number of groups located during decay between the two visits. If so, this initial search may also be then the figure for accumulated recorded, to allow a population faecal pellet groups should be estimate based on standing crop adjusted to take account of the data (see 15 and 16). decay. Deer density for each habitat is calculated from the mean number Fresh pellet groups found within or of pellet groups per hectare, species- close to the plot should be clearly specific defecation rate (Appendix 8)

50 and the number of days between the No indication of sex- or age-class clearance and second count, using ratios. the following equation: Habitat specific decay should be monitored. Number of deer per ha = Delay in obtaining estimate (2-3 months). Number of pellet groups per ha Two site visits required. Time between visits x Defecation rate (days) (pellet groups per day) Performance as an estimate ***** Performance as an index ***** The use of a computer program Inexpensive equipment costs ***** speeds up the calculation of Inexpensive labour costs population size and confidence ** **** intervals from clearance count data. Simple data analysis Suitable programmes are available Data collection period 2-3 months from the authors.

Example Data recorded Number of pellet groups, by species, accumulated for Clearance counts were carried out in each plot on the second visit and any a deciduous (oak and hazel) decay or loss of marked groups. woodland with a sparse understorey to estimate the resident roe deer Equipment required Orienteering population size. Previous studies compass, tape measure, tent pegs or at the site indicated a decay period pins, posts, gloves. of 1-2 months. Plots totalling 0.37 ha in area were cleared during March H & S considerations Strong dis­ and left for 10 days. On the second posable gloves should be worn visit, a total of 40 pellet groups were when searching vegetation for or counted in the 0.37 ha. No decay handling faecal pellet groups. occurred during the 10 day period between clearance and checking. Advantages Assuming 20 pellet groups per day • Applicable in all habitats and defecation rate for roe deer and weathers (except snowfall). 10 days between visits, the • Accuracy and reliability of population size was estimated as estimate is higher than for faecal follows: pellet standing crop counts. • Population estimate for a specific Number of deer per ha = period of time equal to the period between visits. Number of pellet groups per ha • Low labour requirement (1-2 Time between visits x Defecation rate people). (days) (pellet groups per day) • Low equipment costs. • Easily repeated. 108.108 = 0.54 per ha 10x20 Disadvantages • Species identification may The population density of roe deer be difficult where two or in the area was therefore estimated more ungulates are present as 54 km'2. (Appendix 6).

51 The period between visits in this At the site, plots should be located example (10 days) is much shorter from the map by following a than generally recommended, due compass bearing, from a readily to prior knowledge of decay on the identified point, into the area for the site and high deer densities. A required distance. The plot should period of 3 months or more be marked out using a tape measure (depending on habitat specific decay and pegs. It is not necessary to length) is recommended. permanently mark plots unless you wish to return to the same plots in 15. Standing crop plot counts following years.

This method is best suited to areas of Each plot should be carefully and medium deer density (10-30 km ). systematically searched for faecal pellet groups, as previously Method described (page 50), ensuring that As with 14 (Clearance plots), once the method of searching is the same the deer range has been stratified, as that used to monitor faecal pellet plots should be allocated decay. Dividing the plot into 1 metre proportionally to the various strips for searching is the most habitats, and allocated randomly to practical searching technique. All representatives of the strata and faecal pellet groups found should within each representative. be identified to species (where possible) and recorded (Appendix Plot size and number will depend 9). If two people work together they upon the level of accuracy should start working diagonally required and resources available. opposite each other, with both Plots measuring 7 m x 7 m are searching the whole plot. Results recommended as these can be should be compared at the end of effectively searched by one person the plot to ensure that searching without the task becoming too standards are the same. tedious. Larger plots can be searched effectively by two people. If The procedure should be continued more, smaller plots are used, until all sample plots have been proportionally more time will be checked. The mean number of pellet spent finding plot locations. Search groups per plot is then calculated for time per plot will depend on pellet each habitat and deer density group density, plot size and how calculated from the mean number of much ground vegetation is present. pellet groups per hectare, habitat- For example, the search time for specific decay rates (Appendix 7) one person searching 7 m x 7 m plots and species-specific defecation rates in an upland conifer forest was (Appendix 8) using the following up to 95 minutes per plot with a equation: maximum of 35 pellet groups found in one plot (Smith and Mayle, 1994). Number of deer per ha = However, where ground vegetation is reduced two or three 8 plot Number of pellet groups per ha Average decay Defecation rate transects can be searched by two time (days) for x (pellet groups people in a day (Lavin, personal a pellet group per day) communication).

52 From a knowledge of the area of • Only one site visit required to each habitat type within the deer count pellet groups. range and the habitat specific • In upland habitats it is possible densities it is then possible to for 2 people to visit and check calculate an overall population size two or three 8 plot transects in a and density within stated day (assuming 7 m x 7 m plots). confidence limits (usually calculated • Low labour requirement (1-2 to 80% confidence intervals) (see people required). Stratified sampling example). The • Low equipment costs. program 'Pellet 2' available from the • Habitat use can be investigated. authors is recommended for these calculations although these can be Disadvantages done by hand. • Habitat and species specific decay rates need to be monitored Data recorded For each plot, prior to counts to enable deer number of faecal pellet groups per density to be calculated. deer species or unknown category. • Accuracy influenced by accuracy Habitat and species specific faecal of defecation and decay pellet decay length. estimates. • No sex- or age-class information. Equipment required Orienteering • Species identification may be compass, measuring tape and tent difficult where two or more pegs/pins (to mark out plot). ungulates are present (Appendix 6). H & S considerations Disposable gloves should be worn when Performance as an estimate **** searching vegetation for and Performance as an index ***** handling faecal pellet groups. Inexpensive equipment costs ***** Inexpensive labour costs *** Advantages Simple data analysis *** • Suitable for large areas. • Suitable for most habitat types. Data collection period 2 monthsa • Not restricted by weather (other a Previously recorded decay ratescan than snowfall). be used to reduce the collection period • Confidence intervals for the to 4r-7 days. population estimate can be calculated. • More precise than clearance Example counts. Standing crop counts were made for • Population densities and size a roe deer population in a North estimated for the number of Scotland upland forest. Results are animals continuously using the shown in Table 3.7. area over a period of time equal to decay length (usually at least 3-6 months).

53 Table 3.7 Estimation of roe deer population size at Glen Righ from faecal pellet standing crop counts

Habitat strata Mean n SD SE Decay Deer Area Total SE and count number of days (n ha'l) (ha) deer pellets

Pre-thicket 3,7,16/13,1,10,10,12 9.0 8 5.07 1.79 183.0 0.50 100.0 50.2 10.00 Restock 8,3,2,3,0,6,5,1 3.5 8 2.67 0.94 91.5 0.39 90.0 35.1 9.48 Thicket 10,5,7,7,9,8,9,5 7.5 8 1.85 0.65 183.0 0.42 220.0 92.0 8.03 Pre-fell 2,3,1,0,1,2,3,2 1.8 8 1.04 0.37 183.0 0.10 340.0 33.2 6.94 Open 1,1,2,0,2,2,1,1 1.2 8 0.71 0.25 91.5 0.14 200.0 27.0 5.58 Total 184 pellet groups 950.0 238.4 18.26

Population density (deer km 2; mean ± SE) = 25.1 ± 1.9. 80% Cl = 238 ± t35 x 18.26 = 238 ± 24 (215-262 deer).

54 16. Standing crop strip transect compass. Walk 10 m along the counts bearing and record the number of pellet groups, by species, in the first This sampling method is best section (1 m wide x 10 m long) on the suited to areas of low deer density survey form (Appendix 10a). A i m (1-10 km'2). It is a long thin plot long cane or rule should be used to 500-2000 m x 1 m in size. As the measure the transect width, and the procedure involves recording pellet compass bearing followed rigidly to groups found within each 10m ensure minimum bias or selectivity is length of lm wide transect it is introduced. Laying a string trail effectively a plot sampling method (using a Walktac: see Appendix 5) or with plots located next to each other. pulling a length of wire behind as the compass bearing is followed enables Method the midpoint of the transect to be The deer range should be identified determined. Pellet groups found on and, if considered necessary, the edge of a transect or plot should stratified. The length of each transect be counted or rejected, based on the will depend upon the level of number of pellets in the group lying precision required. For a more inside or outside the search area. precise population estimate a longer When groups lie exactly on the edge, transect is required, i.e. 2000 m they should be alternately included rather than 500 m. For each habitat, a and rejected. Individual pellets may representative compartment should belong to 'stringers' which pass be selected. Transect lines should be across the strip. Care should be marked on the map so that they pass taken when searching to ensure these through representative areas of the pellets are not missed. habitat but do not run parallel to streams, rides, topographic or other Walk the next 10 m section and features which may influence deer record the second section. Repeat habitat use. until 500 m or 2000 m have been covered, depending on the accuracy Where the compartment or block is required. Add up the total number on a valley side and deer may use of pellet groups and read the deer the lower areas more than the upper density from Appendix 10b, Table areas, transects should run A10.1 or A10.2, depending on the downhill. Transects are best located deer species and defecation rate (see at random using random number Appendix 8 and Table A8.1). The tables (Appendix 1) to determine the density is more accurately start point from a specified (e.g. calculated using this equation: north-west) corner of the block. The number of transect lines will depend Number of animals per ha = upon their individual lengths and Number of pellet groups per ha the total transect length to be Defecation rate Average decay sampled. In the example shown in (pellet groups x time (days) Figure 3.11, three lines are required per day) for a pellet group to achieve 2000 m sample length. This can then be used with At the site the start point should be knowledge of the area of each located and the map bearing habitat to estimate overall deer transferred to an orienteering population size and density.

55 Although potentially a quicker Advantages method than standing crop plot • Suitable for large areas. counts, in practice where there is ground vegetation, searching times • Suitable for most habitats. are similar to plot counts. Where • Not restricted by weather (other there is no ground vegetation (e.g. than snowfall). under thicket and prefell Sitka spruce), this method is quicker than • Population estimate is for a plot counts. Unless a number of period (3-6 months at least). transects are sampled in each • Large areas sampled relatively habitat, confidence intervals cannot quickly, particularly where be estimated. ground vegetation is sparse.

Data recorded Number of pellet • Habitat use may be indicated. groups, by species, in each 10 m • Low labour requirement length. Habitat and species-specific (1-2 people). faecal pellet decay lengths. • Only one site visit to count pellet Equipment required Orienteering groups. compass, Walktac (see Appendix 5, list of equipment suppliers) or wire • Low equipment costs. and a measuring tape, 1 m cane. • Most useful in sites of low deer density. H & S considerations Disposable gloves should be worn when searching for and handling faecal pellet groups.

Figure 3.11 Location of transect lines within a block for strip and transect sampling

56 Disadvantages Performance as an estimate *** • Confidence intervals cannot be Performance as an index ** ** estimated unless a number of Inexpensive equipment costs ***** transects are sampled in each Inexpensive labour costs *** habitat. Simple data analysis **** • Habitat and species-specific decay needs to be monitored Data collection period 4-12 monthsa before counts and finding a Previously recorded decay rates can be sufficient fresh pellet groups may used to reduce the collection period to be a problem. 4-7 days. • Precision influenced by accuracy of defecation and decay Example estimates. Strip transects were used to estimate • Unless sampling protocol is population density and size for red followed strictly, there may be a and roe. deer in a 907 ha area in tendency to veer off the transect north-west Scotland during line towards pellet groups (this March/April 1997. Two men will cause bias, and possibly completed seven 800 m x 1 m over-estimation of population transects over a 2'/2 day period size). (3 transects per man per day). The • No sex- or age-class information. results are given in Table 3.8.

Table 3.8 Estimation of deer population size at Lochaber from strip transect faecal pellet counts (800 m)

Number of Deer density Area Total deer pellet groups (km~2) (ha) ■ Roedeer R ed d eer Roedeer R ed d eer Roedeer R ed d e er Native 4 12 2.4 5.7 211 5.0 12.0 pinewood Establishment 48 6 28.6 1 2.811 1 >26.8 I?■ 5.0 107 28.7 5.4 Establishment 42 15 25.0 J 7.1 J 1

Pre-thicket 19 6 11.3 2.9 121 13.7 3.5

Thicket 25 12 14.9 "j 5.7] \| >14.9 1f 4.8 200 29.8 9.6 Thicket 25 8 14.9 J 3.8 J1

Pre-fell 30 4 17.9 1.9 268 48.0 5.1 Total 907 125 35

Population density of roe deer 13.78 km '2. Population density of red deer 3.86 km '2.

57 17. Standing crop line transect Pellet group density = counts Total number of pellet groups detected 2 x Effective Total length of Faecal pellet groups may also be 'half-width' x transect line counted using the distance sampling of the strip walked techniques described in Section 2. Transect length will depend upon Deer density is then estimated based expected pellet group density and on knowledge of defecation and required accuracy. decay rates, as for plot counts (14).

Method Data recorded Number of pellet groups and distance of each group The area to be sampled should be from the line for each deer species. stratified as previously discussed Habitat and species-specific pellet (see Stratified sampling) and a group decay lengths. random system of transects superimposed on each area to be Equipment required Orienteering sampled (see Figure 3.11). compass, Walktac or wire and measuring tape. At the site the start point should be located and the map bearing of H & S considerations Disposable the transect transferred to an gloves should be worn when orienteering compass. The compass searching vegetation for and bearing should be followed for a handling faecal pellet groups. predetermined distance, while pellet Advantages group presence, for each deer • Suitable for large areas. species, and distance from the line is • Suitable for all habitats. recorded. As with strip transects it • Not limited by weather (except will be necessary to locate the line of snow cover). the compass bearing using a Walktac, • Precision can be determined. string, wire or by having two • Sample sizes are greater than for observers, one a few metres behind strip counts. the other, so that the line is accurately • Relatively quick compared to plot located before the distance to the counts especially at low densities. observed pellet group is measured. • Potentially more accurate than The distance recorded should be the plots in habitats where dense perpendicular distance from the ground vegetation occurs or there 'centre' of the faecal pellet group (the is variation in ground vegetation. mid point between the two most • Low labour requirement (2 people). widely spaced pellets in the group) to • Low equipment costs. the transect line (Figure 3.11). Disadvantages Computer models are then used to • Accuracy depends upon accurate estimate the probability that a pellet measurement of perpendicular group located within the area will be distance from pellet group to the detected and hence estimate pellet transect line. group density in the area (Buckland, • No sex-/age-class information. 1992; Buckland et al., 1993). • Slower than strip counts. • Requires use of specialist Pellet group density is estimated as computer software to analyse data follows: and calculate population density. 58 Performance as an estimate **** m apart. Greater survey effort was Performance as an index ***** allocated to blocks where sika deer density was considered to be Inexpensive equipment costs ***** highest. Faecal pellet groups were Inexpensive labour costs *** counted and pellet group density for * Simple data analysis each block estimated. Density was Data collection period 4-12 monthsa calculated based on an assumed defecation rate of 25 pellet groups a Previously recorded decay rates can be per day and decay times which had used to reduce the collection period to been monitored for 16 months prior 4-7 days. to the faecal pellet count. Densities ranged from 1.38 (95% Cl, 0.86-2.2) Example deer km-2 to 20.94 (95% Cl, The method was used by the Deer 17.21-25.49) deer km 2. Commission for Scotland to estimate sika deer population sizes in a part To achieve this level of accuracy of south Scotland. The area was (95% confidence intervals), 26 man- divided into eight similarly sized days of effort were required for 10.35 blocks, within which transect lines km of transects. were placed in a zigzag fashion, 200

59 Using cull information 1992), it is possible to reconstruct the population size in a number of ways. Clear and concise mortality and fertility records for a deer population 18. Balance sheet are invaluable as a means of retrospectively calculating population This method relies on having size. They can also be used as a basis previously determined the number of for predicting likely future changes individuals in the population (by sex- in population size. The methods and age-class) using a visual method generally rely on knowing the sex (open hill counts), together with and age of all dead animals, and so knowledge of sex- and age-class accuracy will be influenced by the specific deaths. Although the number of animals which die and are technique provides a quick estimate not found, any emigration and the to predict population size, the estimate accuracy of age determination. will be influenced by the accuracy of the initial population estimate, any bias in The number of animals culled in any sex-class or age-class classification year can only provide an indication (see 1. Open hill counts), and the of the minimum population size proportion of deaths which are not before the cull, e.g. if the cull was recorded. The method also assumes 120 animals in total then the there is no immigration/emigration, minimum population must have i.e. a 'closed' population. been 120 animals before the cull. For deer populations using woodland Method habitats a cull of 10-25% is usually The deer population range (for both required to keep the population males and females) is first identified stable, depending upon natural and marked on a map. During the mortality and fertility levels. spring and autumn of year 1 visual counts to estimate population size A regular cull (mortality) level, with and the number of individuals in constant sex- and age-class ratios, each sex-/age-class are carried out. achieved over a number of consecutive All deaths (natural and culled) are years may indicate that the population recorded by sex- and age-class, and is stable or growing, particularly if the adult population in year 2 is there have been no changes in culling predicted by subtracting known effort or habitat quality. However, deaths in any age-/sex-class from more detailed information about both the observed population in year 1. the animals culled or dying and the remaining population is required Data recorded Age- and sex-class before making any attempt to estimate of all deer seen during the census, true population size. found dead or culled.

Population reconstruction Equipment required Binoculars, from mortality data telescope, map of area.

If all animals dying (culled or H & S considerations Disposable natural mortality) are found and gloves should be worn when handling aged (for ageing methods, see and ageing lower jaws of dead animals. Chapman and Chapman, 1997; Ratcliffe, 1987; Ratcliffe and Mayle,

60 Advantages does (kids and adults) were culled • Suitable for all habitats. over winter and the population • Not limited by weather. censused visually, using the same • Sex- and age-class ratios estimated. observers and technique in the • Provides comparison data for following March. Two hundred and other census methods. eighty-eight deer were observed at • Encourages detailed record this time (104 males, 184 females). collection. Given the initial observations and the known cull of 74 females, the Disadvantages expected number of deer to be seen • Requires visual census and cull in the following spring would be 113 data. (62 males and 51 females). • Accuracy influenced by accuracy However, the observed animals of original census and mortality were 104 males and 184 females, data. suggesting that either the initial • Assumption of 'closed' pop­ survey was inaccurate and a large ulation probably invalid. number had been missed, or that 175 new animals had moved into the Performance as an estimate ** area during the winter. Performance as an index *** Inexpensive equipment costs *** Assuming no immigration over Inexpensive labour costs ** winter, from the spring observation *** Simple data analysis data (104 males and 184 females) Data collection period 6-9 months and the known cull of 74 females, the minimum precull population must have been 362 or almost twice Example the number observed in October. A woodland population of roe deer This emphasises the problems was observed on one day during associated with single sample visual October prior to the doe cull. A total censuses, and the main reason why of 187 individual deer were seen (62 they are not recommended. males, 125 females). Seventy-four

61 19. Life tables Data recorded Sex- and age-class If the age and sex of all deer dying of all deer seen during the census, (natural or culled) over a number of found dead or culled. years is recorded it is then possible Equipment required For roe and to reconstruct the population for any fallow, age determination (see year in the past. The exact number of Chapman and Chapman, 1997; deer in any age class in any given Ratcliffe, 1987; Ratcliffe and Mayle, year can then be calculated. The 1992). probability of an animal surviving from birth to any given age can then H & S considerations Disposable be estimated and used to calculate gloves should be worn when handling other parameters. If it can be and ageing lower jaws of dead animals. assumed that these age-specific survival rates are still valid then Advantages they can be used to estimate current • Suitable for all habitats. population size (Lowe, 1969; • Sex- and age-class ratios can be Caughley, 1977; Boyce, 1995). In estimated. reality it is rarely possible to find all • Low labour requirement. dead animals, but it can be assumed • Low equipment costs. that most are found (e.g. in parks) • Encourages detailed record keeping. then accuracy will be high. Disadvantages Life tables are termed 'time-specific' • Accuracy influenced by bias and or 'static' when calculated from data accuracy in age determination on age composition of a population methods and the proportion of at a specific point in time animals dying and not recorded. representing all age-groups in the • Assumption of 'closed' population. If calculated by population (no emigration or following an age-group or 'cohort' immigration) may not be valid. through time, they are known as • Retrospective rather than current 'cohort' or 'dynamic' life tables (see population estimated. 20). Static life tables are only valid • Requires data collection over a where the age-class distribution of period of years. the population remains stable over • Requires experience in fitting the period equal to the oldest cohort mathematical models. in the population. Where juvenile survival rates fluctuate from year to Performance as an estimate *** year (e.g. roe deer: Ratcliffe and Performance as an index **** Mayle, 1992) age-class distribution Inexpensive equipment costs **** will vary between years and static Inexpensive labour costs ** life tables are not valid. Simple data analysis **

Data collection period 5+ years

62 20. Cohort analysis • Low labour requirement. • Low equipment costs. For a given deer population, from • Encourages detailed record keep­ knowledge of age at death it is ing. possible to determine year of birth • Useful for retrospectively check­ and then accumulate over a period of ing the accuracy of other years the total number of individuals methods. born in any specified (cohort) year (Ratcliffe, 1987; Ratcliffe and Mayle, Disadvantages 1992). Appendix 11 (record form) • Accuracy influenced by biases in shows the format for data recording. age determination and the The number of animals dying in each proportion of animals dying and age class are entered in the vertical not recorded. column for each year. Lines are • At least 5 years before a population totalled horizontally to give the estimate can be calculated. number of animals dying from any • Retrospective (but can be used particular cohort. The total animals predictively with 18, 19 and 21). recovered from a given cohort year can then be used, with fertility and Performance as an estimate **** mortality data, to calculate the Performance as an index **** number of young which must have Inexpensive equipment costs ***** been born in that year, and hence the Inexpensive labour costs ** number of reproducing females in ** the population. Information on adult Simple data analysis sex ratios allows a minimum Data collection period 5+ years population size to be calculated for the cohort year. At least 5 consecutive Example years' worth of mortality data should Figure 3.12 shows the cohort be collected before population analysis for roe does from Alice Holt reconstruction is attempted as only Forest, south-east England (Ratcliffe by then can it be assumed that a and Mayle, 1992). By 1988, 26 reasonably large proportion of female deer had been recovered animals born in year 1 will have been from the 1978 cohort. As a 1:1 sex recovered. The method can be ratio is common at birth, 26 male applied wherever there is good kids would probably also have been fertility, mortality and sex ratio data born in 1978. Survival data suggest for the population, and is 0.75 survival to the winter, therefore particularly useful where most 52 -5- 0.75 = the total number of kids deaths are from culling. born in 1978, i.e. 69. The number of reproducing females required to Data recorded Sex and age of all produce these kids depends on deer found dead or culled. fertility levels. If this is 1.5 kids per H & S considerations Disposable doe, then 69 -h 1.5 = 46 breeding age gloves should be worn when handling females must have been present. and ageing lower jaws of dead animals. Roe populations generally have a 1:1 sex ratio, therefore there would Advantages probably have been 46 males of the • Suitable for all habitats. same age. • Sex- and age-class ratios can be estimated.

63 Figure 3.12 Cohort analysis for roe does, Alice Holt Forest

The number of non-breeding population just after the birth pulse (yearling) does present = the in 1978 can be estimated as 46 adult number of animals recovered from males, 46 adult females, 28 yearlings the 1977 cohort minus the number (males + females) and 69 kids. shot as kids (i.e. 18 - 4 = 14). If we assume there were the same number The total minimum population of yearling bucks present, the total estimate for 1978 is therefore 189.

64 21. Population modelling Disadvantages • Accuracy influenced by accuracy Computer models (Leslie, 1945 and of initial population estimate, 1948) based on estimates of current age-class ratios and age-specific population size, fertility and age- fertility and mortality data specific mortality can be used to (which may vary annually). predict post-breeding population • Requires use of computer and size in future years. The accuracy of suitable software. predictions relies on accurate estimates of initial population size, sex and age- Performance as an estimate ***** class ratios, age-class specific fertility Performance as an index ***** and mortality levels. One such model Inexpensive equipment costs *** is available from the authors. Inexpensive labour costs ** Models which include habitat quality influences on fertility and Simple data analysis ** mortality levels are now becoming Data collection period 1+ years available (Buckland et al., 1998). Population modelling is used as a routine and integral part of deer Example population management by Forest A roe deer population of 238 animals Enterprise in North Scotland. was modelled based on the following assumptions: Advantages • Suitable for all habitats and 1:1 sex ratio. Fertility: yearlings 0.9 species. kids per doe, adult (2-6 years) 1.14 • Sex- and age-class ratios can be kids per doe, adult (7-9 years) 0.4 estimated. kids per doe. Survivorship: kids • Encourages detailed record­ 0.88, yearlings 0.9, adults (2-6 years) keeping. 0.95, adults (7 and 8 years) 0.3. • Different culling scenarios can be quickly evaluated and tested Table 3.9 shows the predicted against true population population change for females only information. for 5 years based on this information. • Low labour requirement. The population increases at • Predictive approximately 25% per year. • Implications of changes in mortality, fertility and age structure are easily shown.

Table 3.9 Predicted population changes (females only) using a Leslie matrix model Year Total Age-class population K 1 2 3 4 5 6 7 8 9 0 119 34 24 18 14 9 7 6 5 1 1 1 149 43 30 22 17 13 9 7 6 2 0 2 187 54 38 27 21 16 12 9 7 2 1 3 234 68 48 34 26 20 15 11 9 2 1 4 291 84 60 43 32 25 19 14 10 3 1 5 364 106 74 54 41 30 24 18 13 3 1

65 £Ju | Using Census Data

Accurate census information enables interval. Setting the cull too high objective decisions to be made about will often become evident as it deer population management. These becomes increasingly difficult to decisions will be based on sound achieve. However, if too low a cull is information about which species are set then the population and its present and their estimated densities impacts will remain unacceptable or and numbers. increase, and a larger cull will be required in the next year to reduce Where an index of deer numbers the population to the target level. If indicates that the population is at a confidence intervals are large then a medium to high level then it would second census method on the same be advisable to protect newly site may help to increase the planted or coppiced areas against precision of the estimate. Once deer browsing. As discussed in target culls have been set decisions Section 1, it will be essential to know can then be made about how these which deer species are present to are to be achieved, and by whom, ensure effective protection is used. particularly where the population is Advice on the height of individual being managed by a Deer tree protection or fence Management Group. specifications to protect against specific deer species is available in Knowledge of population size also Hodge and Pepper (1998), Pepperet enables priorities to be set where a al. (1985) and Pepper (1992). manager is dealing with a number of deer populations, particularly if Accurate population estimates resources are limited. It may be within clearly stated confidence possible to 'let' stalking for less intervals are required where vulnerable areas while concentrating management objectives are to the limited resources on priority areas. reduce numbers (in order to decrease impact levels and road Deer numbers should be assessed traffic accidents or improve herd regularly (every 1-2 years ) prefer­ quality) or to crop deer populations. ably by the same people using the Target culls can then be set based on same methods. Monitoring in this the overall population size and way allows the effectiveness of a predicted recruitment rate, given culling strategy to be measured estimated from counts of adult against management objectives. females and young (e.g. 1) or from Regular population estimates should cull data (see 20 and Ratcliffe and be an integral part of any long-term Mayle, 1992). Cull setting should strategy to manage a deer population also take into consideration the and their habitat. Where only an upper and lower limits of the 95% index of numbers is determined then confidence interval, particularly any population trends will become where the confidence limits are apparent after a minimum of 8 years. wide. A cull based on the overall This may help clarify management estimate will be too low if the objectives for the site and deer population is, in fact, at the upper population. However, knowing just limit, and too high if the population the trend is not sufficient information is at the lower limit of the confidence on which to base a cull. 66 Collection of census data on deer without changes in habitat structure numbers and sex ratios together to allow forward planning of with data on, or estimations of, resource requirement and allocation. recruitment rate enables a whole series of management scenarios to As was stated at the beginning of be simulated by computer ( see 21). this book, deer populations have Different culling regimes to produce been and are likely to continue to the same result can be considered. increase in size and distribution in The effect of the stalker falling ill the foreseeable future. Managers and only achieving half of his target need to know which species are cull, or retiring and not being present and how many there are. replaced, can be investigated. It is Only then can they effectively possible to model the outcome of not manage both the deer and their achieving the target cull, and impacts. The authors hope that this specifically not achieving the target book provides the initial step number of males or females. towards sound and effective long­ Additionally population changes term deer management strategies over time can be predicted with or for the future.

67 P Supporting Information

Appendix 1. Random number table

079 551 301 310 461 159 818 853 608 672 203 590 754 571 895 971 265 887 487 075 504 870 468 884 660 070 970 729 505 616 943 880 980 449 970 409 490 778 260 570 039 047 972 907 421 029 676 568 387 019 426 817 271 470 315 968 428 832 661 594 416 539 672 408 300 742 695 026 349 690 930 535 254 686 475 647 147 101 322 432 425 735 897 061 661 978 018 646 115 106

674 156 836 810 092 594 189 346 340 857 794 334 717 195 470 261 229 514 819 969877 363 635 512 765 016 377 008 940 535 023 904 040 318 094 488 803 813 142 229 099 369 477 025 346 217 930 416 903 937 570 930 893 054 942 388 420 880 037 040 065 883 681 130 316 224 853 197 886 907 031 387 184 016 138 497 263 958 010 994 205 843 968 938 684 754 341 563 200 631

922 488 098 387 143 398 273 748 002 499 491 213 351 281 067 612 330 630 994 864 659 974 305 796 692 855 174 998 282 548 794 344 764 264 210 215 797 844 747 041 433 658 062 922 410 107 154 252 679 874 785 468 084 348 991 861 683 506 573 712 508 300 662 367 341 618 781 025 301 828 014 333 956 895 763 081 950 731 882 981 665 096 260 745 347 212 418 152 877 611

687 338 000 717 523 884 423 725 569 815 977 095 229 727 889 052 264 915 340 101 648 387 543 674 248 467 202 589 050 491 731 523 800 020 846 249 562 962 795 832 028 606 047 005 468 092 583 507 102 249 063 636 305 076 792 015 035 793 243 078 629 521 525 764 691 353 807 819 731 771 749 295 707 849 431 375 911 060 853 385 982 331 979 776 432 333 267 921 693 066

637 131 431 839 538 010 585 553 989 508 228 001 802 899 481 359 400 094 707 823 501 481 137 874 879 124 132 433 318 401 809 846 076 139 274 890 991 060 513 016 462 822 340 366 637 799 777 069 970 624 932 612 439 803 015 368 866 592 747 885 752 528 679 683 603 719 024 258 810 900 359 765 675 769 369 939 184 869 707 327 412 174 932 215 172 678 448 854 145 141

014 268 317 352 641 718 804 564 970 709 771 127 515 609 327 728 508 422 958 358 005 900675 184 274 833 465 888 643 220 400 249 033 834 257 014 777 600 444 723 017 754 082 491 001 836 058 004 845 742 991 154 035 672 854 144 044 321 442 431 556 823 945 255 051 330 357 243 471 389 002 682 338 861 601 345 147 322 815 243 694 671 345 883 437 938 928 581 100 228

690 811 307 910 255 184 979 534 919 601 985 059 861 922 836 326 065 142 627 550 060 058 401 663 424 122 491 117 986 347 420 066 566 946 788 731 551 477 437 142 134 126 574 169 685 565 451 448 084 837 845 669 254 615 133 515 561 071 023 496 045 432 844 397 006 348 908 382 643 069 711 307 538 271 986 475 739 518 005 216 100 798 522 746 920 635 404 574 184 690

892 938 887 902 151 345 599 033 991 722 213 949 631 060 620 351 058 068 920 510 765 708 105 258 789 960 910 589 210 166 976 259 243 898 525 866 516 098 213 244 481 718 673 361 353 847 867 480 690 938 699 333714 587 640 324 072 607 732 490 162 193 809 888 980 459 710 321 946 073 553 508 291 089 858 333 561 201 978 748 813 487 071 395 034 656 658 630 590694

68 Appendix 2. Example habitat stratification

In Alice Holt Forest, Hampshire/ Broadleaved establishment Surrey, a very diverse forest, 12 Broadleaved trees aged 0-3 years. habitat types were recognised on the Mainly oak about 1 m high. Ground basis of age-class of trees and flora of bramble, hazel, bracken, vegetation types, which would holly (Ilex aquifolium) and Yorkshire probably be differentially used by fog (Holcus lanatus). the deer. These were as follows: Broadleaved pre-thicket Conifer establishment Broadleaved trees aged 4-15 years. Conifers 0-5 years old. Mainly Mainly oak about 2.5 m high, Norway spruce (Picea abies) and interspersed with regenerating Corsican pine (Pinus nigra) less than hazel. Ground flora of Yorkshire 1 m high. Ground flora of hazel fog, bramble, bilberry (Vaccinium {Corylus avellana), bracken (Pteridium myrtillus), willowherb (Chamaenerion aquilinum), bramble (Rubus angustifolium), dog's mercury fruticosus) and wavy hair grass (Mercurialis perennis) and bracken. (Deschampsia flexuosa). Broadleaved thicket Conifer pre-thicket Broadleaved trees aged 15-25 years. Conifers aged 10-20 years old. A mixture of oak (Quercus spp.) with Mainly Corsican pine, Norway some Norway spruce and hazel. spruce, hybrid larch (Larix eurolepis) Profuse shrub vegetation with do^'s and western hemlock (Tsuga mercury, bramble, wild strawberry heterophylla). Ground flora of (Fragaria vesca), ground ivy bracken, wavy hair grass and (Glechoma hederacea) and willowherb. raspberry (Rubus idaeus). Broadleaved pole-stage Conifer pole-stage Broadleaved trees aged 25M0 years. Conifers aged 20-30 years old. A mixture of oak, alder (Alnus Mainly Norway spruce, western glutinosa) and beech (Fagus hemlock, Corsican pine, hybrid sylvatica). Canopy and understorey larch and Douglas fir (Pseudotsuga species of holly, hawthorn (Crataegus menziesii). Ground flora of bracken, spp.), yew (Taxus baccata) and hazel. bramble, wavy hair grass and dead Ground flora of bramble, dog's needles. mercury and creeping soft grass (Holcus mollis). Conifer pre-felling Conifers more than 30 years old. Broadleaved pre-felling Mainly Corsican pine, western Broadleaved trees older than 40 hemlock, Norway spruce, Japanese years. A mixture of oak, ash larch (Larix kaempferi). Canopy (Fraxinus excelsior), sweet chestnut closed in most places with a reduced (Castanea sativa) and western red ground flora; scattered patches of cedar (Thuja plicata). Canopy and short bracken and wavy hair grass. understorey species of hawthorn, yew and hazel. Ground flora of bramble, honeysuckle (Lonicera periclymenum), stinging nettle (Urtica dioica), wavy hair grass and bracken.

69 Recently felled broadleaved Pasture woodland A semi-permanent grassland used Originally oak woodland with for hay and grazing. Ground hawthorn, hazel and field maple vegetation of dock (Rumex spp.), (Acer campestre). Newly growing creeping thistle (Cirsium arvense), coppiced stumps and ground flora meadow buttercup (Ranunculus including bramble, raspberry and acris), yarrow (Achillea millefolium) bracken. and grasses (Gramineae).

70 Appendix 3. Vantage point count form Density estimation for woodland deer occupying hilly terrain

Forest ______Observer(s) __

Location______Area_____ ha Structural type Planting year.

Date______Time on ______Time off

Weather conditions and visibility______

DEER SEEN Time Location Species Adult Adult Yearlings Yearlings Calf/ Unclassified Remarks male female male female kid

71 Appendix 4. Thermal imaging record form 5 Q a to 4| J "1 PH

RouteBlock SpeciesRoute Numbers Body Bearing Direction CommentsPerp. length in group (deg°) (deg°) dist. (m) (m) (>r C *3 w - T h - 72

See page 73 for explanatory notes. Thermal imaging record form: explanatory notes

Forest: Name of forest or woodland. Weather: Mention weather, especially if it is stormy, rainy or windy, or changes during the course of a night. Block: Letter (e.g. A or B) referring to a part of the forest named above (only needed if estimate required for part of the forest). Route: Route label, e.g. 1, 2, etc. Length: Length of route or line in metres (measured from 1:10 000 map, e.g. 1500). If observations were made from only one side of the vehicle, then this distance should be divided by 2. Routes from which no deer were seen should also be included. Species: For example, R=red deer, S=sika deer. Number: Number of deer in group followed by numbers of sex-/age-classes, if identifiable. X=total, ; =males, _ =females, J=juveniles, r+J=groups of does/hinds plus calves. If the figure put under I > + +J (or t-( +])) then the remainder is assumed unclassifiable. Body: Length of body, nose-rump, standing side on, measured in graticules, e.g. 1.5. Assumed to be measured on animal at centre of the group. If the group was of mixed sex or included adults+juveniles, enter the sex/age measured under 'comments'. If only the head is visible measure ear-ear width or tine-tine width and mention this in 'comments'. This measurement should be taken from where the group was first sighted. If the animals were approached (e.g. to help identify them) before the body length was measured then the distance moved should be estimated and noted in 'comments'. Bearing: Bearing0 of animal from observer. Direction: Bearing0 of the direction of the route from where the observation was made. Perp. dist. Distance at right angles (perpendicular) to route/line. Comments: Any other information. For example: behaviour, or compartment label; enter 'Ride' if on ride or forest road or 'Field' if in neighbour's field; enter direction (N, S, E, W) group took flight, if this may take them near another route/line.

NB • For all groups, enter 'Body' and 'Bearing' and mark the position of the group on 1:10 000 map as accurately as possible, labelling group with the serial number. • Enter 'Direction' if the route ahead is straight. • If body length < 0.5 graticules or the route bends and the bearing of direction of travel would give a misleading estimate of the perpendicular distance, then mark observer position on the map as well, and measure/confirm perpendicular distance from the map afterwards.

73 Appendix 5. List of equipment suppliers

Thermal imaging equipment Transect equipment

Pilkington Optronics Ltd Walktac 1 Linthouse Road, Glasgow Stanton Hope C51 4BZ 11 Seax Court, Southfields Laindon, Basildon, Essex Agema Infrared Systems Ltd SS15 6LY Arden House, West Street Leighton Buzzard LU7 7DD

GEC Marconi Christopher Martin Road Basildon, Essex SS14 3EL

Insight Vision Systems Merebrook Hanley Road,Malvern Worcestershire WR13 6NP

Suitable equipment is also available for hire from the authors.

74 Appendix 6. Faecal pellet group identification

Deer faecal pellets are short, There is also minimum confusion cylindrical or almost spherical, and when trying to distinguish between often have a small point at one end faecal pellet groups of the largest (Figure A6.1). They have a smooth (red; Plate 8) and smallest (muntjac; surface compared to faeces of Plate 9) deer species. Greatest carnivores or rabbits and hares and confusion occurs between muntjac when fresh are usually black and and roe (Plate 10), roe and fallow covered in a thin, shiny layer of (Plate 11), fallow and sika (Plate 12), mucus which quickly dries. In and sika and red deer especially if spring/summer when lush new red/sika hybrids are present. vegetation is available they may be Confusion, particularly with red sticky and softer and individual and roe deer pellets, can also occur pellets will often become fused if goats or sheep (Plate 13) are also together (Plate 6). The number of resident in the area. pellets defecated in a group at one time will vary due to sex, age-class 'Typical' pellet groups for all species and diet of the individual animal are shown in Plates 6-13 and and may be as many as 150 for described in the identification fallow deer (Smith and Mayle, section below. However individual 1994). However, the average pellet and pellet group form will number of pellets in a fresh group is vary due to diet (site and seasonally 40-60. influenced), age- and sex-class of the animal. For example, pellets produced by mature red deer stags tend to be larger than those produced by hinds and individual pellet shape may also vary. Within any species pellets produced by juveniles will be smaller than those produced by adults.

The most effective means of becoming confident at the identification of pellets is to collect fresh pellets from known species. This can be from deer farms, areas where only one species is known to Figure A6.1 Typical deer pellet shapes occur, after watching an animal defecate, or from the rectum of shot Where only one species of deer or deer. Then study their form and ungulate is present faecal pellet ideally get a colleague to place these identification is not a problem. Deer out in the field to enable a 'blind' and rabbit or hare faecal pellets are test for identification. sufficiently different for there to be no confusion about identification between these species groups (Plate 7).

75 In any event, unless there is only Muntjac deer one species of ungulate present it is Pellets are black, shiny and striated, not usually possible to allocate spherical or cylindrical in shape and 100% of all pellet groups found with from 6-13 mm long x 5-11 mm confidence. Most, usually 80-90% broad. Pellets may be pointed at one of pellet groups, will be allocated to or both ends, rounded concave or a specific species and the remainder flat at the other end or rounded at will be allocated to an unknown or both ends. Fresh groups contain possible category, e.g. red/sika, 20-120 pellets which may adhere to which can then be dealt with each other. In some areas the same separately or assigned to the spot may be returned to for specific species in the same defecation, forming a latrine site proportions as the identified faecal (Plate 14) (Chapman, 1991). pellet groups (Buckland, 1992). Chinese water deer Deer faecal pellet identification Pellets similar to those of muntjac, but usually longer - 10-15 mm x Red deer 5-10 mm. Black or dark brown and Pellets are up to 30 mm long and cylindrical, with a point at one end 13-18 mm across (Staines, 1991; (Farrell and Cooke, 1991). Bang and Dahlstrom, 1974), cylindrical in shape, often with a Sheep point at one end and rounded or Similar to, but less regular than slightly concave at the other, being goat, pellets measure 12-17 mm x described as 'acorn-shaped' by approx. 10 mm. Although initially Staines (1991). Usually black and cylindrical, compression during shiny when fresh, they become deposition may cause them to duller and dark brown as they age. become more angled and sometimes pyramidal in shape Sika deer (Bang and Dahlstrom, 1974). Large Pellets are similar to those of red deposits may build up in some sites and roe (Ratcliffe, 1991) and fallow (Jewell and Bullock, 1991). (Lowe, 1977). Goats Fallow deer Pellets are similar in shape and size Pellets are cylindrical, usually to those of sheep, red and roe deer, pointed at one end and concave at 10-20 mm long, cylindrical with the other: 16 x 11 mm in males and pointed or concave ends. Tend to be 15x8 mm in females (Chapman and more symmetrical than deer or sheep Putman, 1991). pellets and large accumulations build up in traditional shelter spots Roe deer (Bullock, 1991). Pellets are elongated and cylindrical in shape, approx. 14 x 6 mm, up to Rabbits and hares 18 x 9 mm and often stick together Rabbit and hare pellets are very when deer have been feeding on similar, slightly flattened, spherical readily digested food during the and firm, being larger in hares summer (Staines and Ratcliffe, (15-20 mm diameter) than in rabbits 1991). (-10 mm diameter). No obvious

76 mucus covering, so not adherent and becomes paler as exposed to the and coarse fragments of plants can sim. Rabbits use faecal pellets as be clearly seen on the surface. territory markers hence large Colour varies according to diet, but numbers may be found in one spot tends to be paler than deer pellets, (Bang and Dahlstrom, 1974).

77 Appendix 7. Faecal pellet group decay assessment

The presence of faecal pellet groups moist, dry, see Figure A7.1) for the in the habitat will be influenced by five major deer species in Great both the number of deer, their Britain. defecation rate and the length of time that pellets remain on the These figures are presented as a ground (decay time). Pellet groups guide to indicate the variability in are subject to a number of influences decay lengths, and to provide the and may disappear due to microbial reader with an 'estimate' of decay or invertebrate attack, the which can be used in conjunction mechanical forces of rain, wind or with their own faecal pellet group physical disturbance (by passing count data, where they do not yet animals) or by being covered by have their own site specific decay vegetation or falling leaf/needle data. The figures can be adjusted litter. Deer diet and hence pellet accordingly, based on 'local' and form and shelter from climatic regional information on climatic influences provided by vegetation variation between years. For on the site, together with soil example, if summer 1997 was wetter moisture levels, will also influence than 1995 then length of time to decay. Pellet group decay will decay could be expected to be less therefore be species, site and habitat for an open habitat. specific and should necessarily be measured for each habitat in sites Monitoring faecal pellet where faecal pellet groups are to be group decay counted to provide an estimate of deer population size. Where decay lengths of <6 months are expected faecal pellet group Mitchell and McCowan (1984) decay need only be monitored for showed that red deer pellet group the 6 months prior to standing crop decay could vary from 95 to 450 counts (i.e. from September/October days between sites and seasons, and onwards if standing crop counts are Smith and Mayle (1994) recorded to be carried out in March/April). decays as short as 12 days for fallow Where decay lengths are generally deer faecal pellet groups in a site in >6 months (see Tables A7.1-A7.5) north-west Wales. we advise that decay length is monitored for 12 months or more. Studies to investigate and model the influence of habitat, season and Faecal pellet groups to monitor decay weather on faecal pellet group decay should be collected and placed out are under way in Forest Enterprise on a day when 'typical' weather sites. Six fresh pellet groups have conditions for that month prevail. been monitored in each habitat for Extreme weather conditions should each month, as described below. be avoided. They should be placed in Initial results are given in Tables a position where 'typical conditions' A7.1-A7.5 and are for decay length within the habitat prevail (e.g. if 75% estimated from faecal pellet groups of a pre-thicket area is closed canopy, monitored between March 1995 and 75% of the groups should be placed August 1997 in a range of habitat under closed canopy). types in three climatic zones (wet,

78 Fresh faecal pellet groups should be The presence of each pellet group collected from the habitat in should be ascertained and recorded question or neighbouring habitats. A on a decay assessment form (Table minimum of 4-6 groups, each A7.6). A group is still present if six or containing at least 40 pellets, should more pellets remain (Plate 15). The be placed and marked in the habitat, searching technique used should be away from the edge but in a site consistent across habitat types and the which can be readily monitored. The same as that used for the faecal pellet date of placement should be groups standing crop or clearance crop recorded and the pellet groups counts (i.e. if vegetation is parted or visited regularly (weekly during the last leaf fall moved to count the periods of high rainfall, monthly number of pellets in a group when during 'normal' weather). Fresh monitoring decay, the same search pellet groups should be placed out method should be used for the each calendar month. clearance or standing crop counts).

Figure A7.1 Three moisture zones in Great Britain, based on ecological site classification

79 Any tendency to search more that only pellet groups from 4 diligently for pellet groups which months prior to a March standing are being monitored for decay rate crop count would be expected to be (as the observer will remember present, then only the previous 4 where these were initially placed) months data should be used to should be avoided. calculate decay length.

For each pellet group, the number of The following indicates the days to decay should be recorded importance of accurate decay and the average number of days to information: decay calculated for the six groups in each habitat for each month. The Assuming the same number of overall average number of days to faecal pellet groups per hectare had decay is calculated by totalling the been found in three habitats (A, B 12 monthly averages and dividing and C) with decays of 12, 95 and 450 by 12. Table A7.7 gives an example days, then the deer densities of B of 12 months records for one habitat and C would be 1/8 and 1/37.5 type. Where decay length is so short that of A.

80 Table A7.1 Decay lengths for red deer

Climatic zones (see Figure A7.1) Habitat Dry Moist Wet | Location Days to Location Days to Location Days to decay decay decay O pen Bare ground Grass ride /glade Sunart 81 Moor/heath Lochaber 444 Sunart 146 Grass/pasture W oodland Establishment/restock Sunart 149 (up to 1 m top height) C on ifer Pre-thicket Thetford 96 (up to 3 m top height) Thicket Thetford 106 (up to 10 m top height) Pole-stage Thetford 103 Pre-fell Thetford 99 Lochaber 177 Doilet a 510 B road leaf Pre-thicket (up to 3 m top height) Thicket (up to 10 m top height) Pre-fell Mature Thetford 77

a Pellet groups still present after 1 year.

81 Table A7.2 Decay lengths for roe deer

Climatic zones (see Figure A7.1) H abitat D ry M o ist W et Location D ays to Location D ays to Location decay decay BIB O pen Bare ground Grass ride/glade Ayrshirea 638 Riparian Moor/heath Grass/pasture Hampshire 64 Hamstedey 98 W oodland Establishment/restock Hampshire 138 Ayrshire 189 Hamsteriey3 352 (up to 1 m top height) C on ifer Pre-thicket Thetford 78 Ayrshire 202 Hamsteriey 371 (up to 3 m top height) Thicket Thetford 91 (up to 10 m top height) Pole-stage Thetford 77 Hamsteriey 370 Pre-fell Thetford 92 Ayrshire a 416 Hamsteriey 537

B road leaf Pre-thicket (up to 3 m top height) Thicket Hampshire 165 (up to 10 m top height) Pre-fell Hamsteriey 196 Mature Thetford 75 Hampshire l b 111 Hampshire 2C 129

‘ Pellet groups still present after 1 year. b Ground vegetation grassy. c Ground vegetation leaf litter.

82 Table A7.3 Decay lengths for sika deer

Climatic zones (see Figure A7.1) Habitat Dry Moist Wet Location Days to Location Days to Location Days to decay decay decay O pen Bare ground Peebleshire 135 Grass ride/glade Riparian Moor/heath Grass / pasture W oodland Establishment/ restock (up to 1 m top height) C on ifer Pre-thicket Peebleshire0 134 (up to 3 m top height) Thicket (up to 10 m top height) Pole-stage Peebleshire 219 Pre-fell B road leaf Pre-thicket (up to 3 m top height) Thicket (up to 10 m top height) Pre-fell Mature

‘ Pellet groups still present after 1 year.

83 Table A7.4 Decay lengths for fallow deer

Climatic zones (see Figure A7.1) j Habitat Dry Moist Wet ! Location Days to Location Days to Location Days to decay decay decay O pen Bare ground Ae 240 Grass ride/glade New Forest 99 West Glamorgan 57 Riparian West Glamorgan 89 Moor /heath Dolgellau340 Arable crops Grass/pasture Ae 42 W oodland Establishment/restock Dolgellau 124 (up to 1 m top height) C on ifer Pre-thicket West 114 Dolgellau 198 (up to 3 m top height) Glamorgan Thicket Dolgellau 270 (up to 10 m top height) Pole-stage New Forest265 Dolgellau3434 Pre-fell Dolgellau 171 Broad leaf Pre-thicket (up to 3 m top height) Thicket (up to 10 m top height) Pre-fell Mature New Forest 219

a Pellet groups still present after 1 year.

84 Table A7.5 Decay lengths for muntjac deer

Climatic zones (see Figure A7.1) Habitat Dry Moist Wet Location Days to Location Days to Location Days to decay decay decay O pen Bare ground Grass ride/glade Riparian Moor/heath Grass/pasture W oodland Establishment/restock (up to 1 m top height) C on ifer Pre-thicket Thetford 89 (up to 3 m top height) Thicket Thetford 99 (up to 10 m top height) Pole-stage Thetford 102 Pre-fell Thetford 128 B roadleaf Pre-thicket (up to 3 m top height) Thicket (up to 10 m top height) Pre-fell Mature Thetford 100

85 Table A7.6 Faecal pellet group decay assessment form 86 Table A7.7 Pellet decay : example of 12 months' records NX X rH CN CO LO VO (S02Q^£S

rH CN CO LO VO X 'j s u o u n s C'gjs X X X X rH CN CO LO VO X rH CN CO VO m X X X X X X x CN CO (N CO in VO 87 CN CO LO vO CN CO LO VO

x: Pellet group no longer present. Appendix 8. Faecal pellet group defecation rate assessment For all sampling methods it will be within a given period of time. Mean necessary to have an estimate of daily defecation rate can then be defecation rate for the species being calculated. This is rarely practicable considered to enable deer density to and so defecation rates determined be calculated, based on faecal pellet from captive animals in similar group counts. habitats, or from the literature, are most usually used. For a captive Defecation rate will vary under the population of known size it is influences of sex- and age-class, possible to determine defecation habitat type and diet (influenced by rate from clearance plots (14), a season). Ideally mean defecation knowledge of the number of animals rate should be determined for the present and the area available to population under consideration, by them. This method has been used to following individual animals of determine some of the rates given in different age- and sex-classes and Table A8.1. counting the number of defecations

Table A8.1 Defecation rates for British deer

Species Rate per day Method of Reference determination Red 25 (19-29) clearance plots Mitchell and McCowan, 1984 (24-33) observations Mitchell et al., 1983 Sika 25 (26.3) clearance plots Burkett, unpublished, 1995 (24) observations Benson, unpublished, 1995 Fallow 21.4 clearance plots Mayle et al., 1996 Roe 20 (17-23) clearance plots Mitchell et al., 1985 Muntjac 7.5 observations Chapman (personal communication) Chinese water deer unknown Appendix 9. Plot faecal pellet group count form

Observer's name Forest Location Date Starting point Bearing Structure type Plot area

Plot number 1 2 3 4 5 6 7 8 9 10 11 12 Total groups

Dung groups:roe Dung groups:red Dung groups:fallow Dung groups:sika Dung groups:muntjac Dung groups:roe? muntjac Dung groupsiroe? fallow Dung groups:red? sika Dung groups:roe? sheep

jr each Total------.-----. Number----i------1 i------1 Average number pecies (row) groups I___ I ‘ of Dlots I____ I - I____ I of erouDs/Dlot

89 Appendix 10a. Transect faecal pellet group count form

Record number of pellet groups in each 10 m section of the 1 m wide transect in a separate box. Pellet groups should be identified to individual deer species.

Transect (m) 10 20 30 40 50 60 70 80 90 100 Total 0-100 100-200 200-300 300-400 400-500 Total= 500-600 600-700 700-800 800-900 900-1000 Total= 1000-1100 1100-1200 1200-1300 1300-1400 1400-1500 Total= 1500-1600 1600-1700 1700-1800 1800-1900 1900-2000 Total=

90 Appendix 10b. Deer density assessed from transect faecal pellet group count

Table A10.1 Roe deer density (figures rounded) determined from transect faecal pellet group count

Dung groups Deer density per 500 m transect per 2000 m transect per 100 ha

5 20 1.4 10 40 2.7 15 60 4.1 20 80 5.5 25 100 6.8 30 120 8.2 35 140 9.6 40 160 11.0

Density based on a defecation rate of20 groups per day and a decay rate of 12 months (365 days):

1 Deer = 7300 pellet groups (20 x 365 days)

The relationship for alternative decay rates can be simply calculated by multiplying (x) the deer density by 365 and then dividing (-r) by the new decay rate (days).

Table A10.2 Red and sika deer densities (figures rounded) determined from transect faecal pellet group count

Dung groups Deer density per 500 m transect per 2000 m transect per 100 ha

5 20 1.1 10 40 2.2 15 60 3.3 20 80 4.4 25 100 5.5 30 120 6.6 35 140 7.7 40 160 8.8 45 180 9.9 50 200 11.0

Density based on a defecation rate of25 groups per day and a d ecay rate of 12 months (365 days):

1 Deer = 9125 pellet groups (25 x 365 days)

The relationship for alternative decay rates can be simply calculated by multiplying (x) the deer density by 365 and then dividing (-r) by the new decay rate (days). 91 Appendix 11. Cohort analysis form 92 REFERENCES

Andersen, J. (1953). Analysis of a Caughley, G. (1977). Analysis of Danish roe-deer population. vertebrate populations. Wiley, Danish Review of Game Biology 2: Chichester. 127-155. Chapman N.G. (1991). Muntjac deer. Bang, P. and Dahlstrom, P. (1974). In: Corbet, G.B. and Harris, S., Collins guide to animal tracks and eds. The handbook of British signs. Collins, London. mammals. Blackwell Scientific Publications, Oxford, 525-532. Boyce, M.S. (1995). Populations and management. In: S.H. Berwick Chapman, D. and Chapman, N.G. and V.B. Saharia, eds. The (1997). Fallow deer. Coch-y- development of international Bonddu Books, Machynlleth, principles and practices of wildlife Powys. research and management. Asian and American approaches. Oxford Chapman N.G. and Putman, R.J. University press, New Delhi, (1991). Fallow deer. In: Corbet, 263-292. G.B. and Harris, S., eds. The handbook of British mammals. Buckland, S.T. (1992). Review of Blackwell Scientific Publications, deer count methodology. Oxford, 508-518. Unpublished report to the Scottish Office. Cochran, W.G. (1977). Sampling techniques, 3rd edn. Wiley, New Buckland, S.T., Anderson, D.R., York. Burnham, K.P. and Laake, J.L. (1993). Distance sampling. Conner, M.C., Lancia, R.A. and Estimating abundance of biological Pollock, K.H. (1986). Precision of populations. Chapman and Hall, the change in ratio technique for London. deer population management. Journal o f Wildlife Management 50: Buckland, S.T., Trenkel, V.M., Elston, 125-129. D.A., Partridge, L.W. and Gordon, I.J. (1998). A decision support Dzieciolowski, R. (1976). Estimating system for red deer managers in ungulate numbers in a forest by Scotland. In: Goldspink, C.R., track counts. Acta Theriologica 21: King, S. and Putman, R.J., eds. 217-222. Population ecology, management and welfare of deer. Manchester Farrell, L. and Cooke, A. (1991). Metropolitan University and Chinese water deer. In: Corbet, Universities Federation for G.B. and Harris, S., eds. The Animal Welfare, 82-87. handbook of British mammals. Blackwell Scientific Publications, Bullock, D. J. (1991) . Feral goat. In: Oxford, 532-537. Corbet, G.B. and Harris, S., eds. The handbook of British mammals. Blackwell Scientific Publications, Oxford, 541-547.

93 Ferris-Kaan, R. and Patterson, G.S. Harris, S., Morris, P., Wray, S. and (1992). Monitoring vegetation Yalden, D. (1995). A review of changes in conservation British mammals: population management of forests. Forestry estimates and conservation status of Commission Bulletin 108. British mammals other than HMSO, London. cetacean. JNCC, Peterborough.

Forestry Authority (1995). Managing Hodge, S. and Pepper, H. (1998). deer in the countryside. Forestry The prevention of mammal damage Practice Advice Note 2. Forestry to trees in woodland. Forestry Commission, Edinburgh. Commission Practice Note 3. Forestry Authority, Edinburgh. Gill, R.M.A. (1990). Monitoring the status of European and North Jewell, P.A. and Bullock, D.J. (1991). American cervids. GEMS Infor­ In: Corbet, G.B. and Harris, S., mation Series No. 8. Global eds. The handbook of British Environment Monitoring System, mammals. Blackwell Scientific United Nations Environment Publications, Oxford, 547-552. Programme, Nairobi, Kenya. Langbein, J. (1996). The red deer of Gill, R.M.A. (1992). A review of Exmoor and the Quantocks. Deer damage by mammals in north 9: 492-498. temperate forests: 1. Deer. Forestry 65(2): 145-169. Leslie, P.H. (1945). On the use of matrices in certain population Gill, R.M.A., Johnson, A.L., Francis, mathematics. Biometrika 33(3): A., Hiscocks, K. and Peace, A.J. 183-212. (1996). Changes in roe deer (Capreolus capreolus L.) Leslie, P.H. (1948). Some further population density in response notes on the use of matrices to forest habitat succession. in population mathematics. Forest Ecology and Management Biometrika 35: 213-245. 88: 31-41. Levy, PS. and Lemeshow, S. (1991). Gill, R.M.A., Thomas, M.L. and Sampling of populations: Stocker, D. (1997). The use of methods and applications. portable thermal imaging for Applied probability and statistics. estimating population density in Wiley Series in Probability and forest habitats. Journal of Applied Mathematical Statistics. Wiley, Ecology 34:1273-1286. Chichester.

Greenwood, J.J.D.L. (1996). Basic Lowe, V.P.W. (1969). Population techniques. In: Sutherland, W.J., dynamics of the red deer (Cervus ed. Ecological census techniques. elaphus) on Rhum. Journal o f Cambridge University Press, Animal Ecology 38: 425-457. Cambridge, 11-109.

94 Lowe, V.P.W. (1977). Sika deer. In: Pepper, H.W. (1992). Forest fencing. Corbet, G.B. and Southern, H.N., Forestry Commission Bulletin eds. The handbook of British 102. HMSO, London. mammals. Blackwell Scientific Publications, Oxford, 423-428. Pepper, H.W., Rowe, J.J. and Tee, L.A. (1985). Individual tree Mayle, B.A., Doney, J., Lazarus, G., protection. Arboricultural Leaflet Peace, A.J. and Smith, D.E. 10. HMSO, London. (1996). Fallow deer (Dama dama L.) defecation rate and its use in Pepper, H.W. (1998). Nearest determining population size. neighbour method of quantifying Supplemento alle Ricerche di wildlife damage to trees in Biologia della Selvaggina XXV, woodland. Forestry Commission 63-78. Practice Note 1. Forestry Authority, Edinburgh. Mitchell, B., McCowan, D. and Campbell, D. (1983). Faecal Pucek, Z., Bobek, B., Labudski, L., deposition as indicators of site Milkowski, L., Morow, K. and use by red deer. Institute of Tomek, A. (1975). Estimates of Terrestrial Ecology Annual Report density and numbers of 1982. ITE, Cambridge, 85-87. ungulates. Polish Ecological Studies 1 : 121- 135. Mitchell, B. and McCowan, D. (1984). The defecation Pyatt, D.G. and Suarez, J.C. (1997). frequencies of red deer in An ecological site classification for different habitats. Institute of forestry in Great Britain. Forestry Terrestrial Ecology Annual Report Commission Technical Paper 20. 1983. ITE, Cambridge, 15-17. Forestry Commission, Edinburgh.

Mitchell, B. and Kirby, K. (1990). Ratcliffe, P.R. (1987). The Forestry 63(4): 333-354. management of red deer in the commercial forests of Scotland Mitchell, B., Rowe, J.J., Ratcliffe, related to population dynamics P.R.R. and Hinge, M. (1985). and habitat changes. PhD Thesis, Defaecation frequency in roe London. deer (Capreolus capreolus) in relation to the accumulation Ratcliffe, P.R. (1991). Sika deer. In: rates of faecal deposits. Journal o f Corbet, G.B. and Harris, S., eds. Zoology, London (A) 207:1-7. The handbook of British mammals. Blackwell Scientific Publications, Norton-Griffiths, M. (1978). Oxford, 504-508. Counting animals, 2nd edn. African Wildlife Leadership Ratcliffe, P.R. and Mayle, B.A. Foundation, Nairobi. Available (1992). Roe deer biology and from the African Wildlife management. Forestry Com­ Foundation, PO Box 48177, mission Bulletin 105. HMSO, Nairobi, Kenya. London.

95 Reynolds, P., Duck, C., Youngson, D. Staines, B.W. and Ratcliffe, P.R. and Clem, D. (1994). An (1991). Roe deer. In: Corbet, G.B. evaluation of airborne thermal and Harris, S., eds. The handbook imaging for the census of red of British mammals. Blackwell deer (Cervus elephus) populations Scientific Publications, Oxford, in extensive open habitats in 518-525. Scotland. In: Forests and wildlife - towards the 21st century. Stewart, L.K. (1976). The Scottish red Proceedings of 21st International deer census. Deer 3(10): 529-533. Union of Game Biologists Congress, Halifax, Nova Scotia, Strandgaard, H. (1972). The roe deer Canada 1993. Canadian Forest ('Capreolus capreolus L.) Service, Petawawa National population at Kalo and the Forestry Institute, Chalk River, factors regulating its size. Danish Ontario, vol. 2, 162-168. Review of Game Biology 7:1-205.

Smith, D.F.E. and Mayle, B.A. (1994). Sutherland, W.J. (1996). Ecological Fallow deer density and census techniques. Cambridge habitat use at Coed-Y-Brenin. University Press, Cambridge, Unpublished Forestry Commission 336. Research Division closure report. Thompson, S.K. (1992). Sampling. Staines, B.W. (1991). Red deer. In: Applied probability and statistics. Corbet, G.B. and Harris, S., eds. Wiley Series in Probability and The handbook of British mammals. Mathematical Statistics. Wiley, Blackwell Scientific Publications, Chichester. Oxford, 492-504. White, G.C. and Garrott, R.A. (1990). Analysis of wildlife radio-tracking data. Academic Press, London.

96

Searching for faecal pellet groups in an open grassy habitat

This Field Book provides information on the various methods available for determining deer population size. It draws together research, practical experience and observations with the aim of offering guidance for a wide range of woodland situations.

How Many Deer? is recommended to all those interested in promoting sound deer management, from landowners and deer managers to students and those involved in research.

Forestry Commission objectives are to:

• p ro te ct Britain's forests and woodlands

• exp an d Britain's forest area

• en h an ce the economic value of our forest resources

• conserve and improve the biodiversity,

lan d scap e and cultural heritage of our forests Published by an d w oodlands The Forestry Commission 231 Corstorphine Road • d evelo p opportunities for woodland EDINBURGH recreation EH 12 7AT

in crease public understanding and IS B N 0 85538 405 0 community participation in forestry £14.00