African Forest Biodiversity: a Field Survey Manual for Vertebrates

A Field Survey Manual for Vertebrates

Edited by Glyn Davies Earthwatch supports field research projects in over 40 countries around the world, in a wide range of disciplines and habitats. The organisation has supported research in African tropical forests for over 20 years, and accepts applications for funding from researchers across the continent.

Earthwatch Europe runs a professional development programme designed to build the capacity of African institutions working to conserve biodiversity. Our African Fellowship Programme places African conservationists, scientists and NGO workers on field research projects relevant to their professional lives. The programme provides training in an African context; through the unique experience of practical participation on a project in another African country, and working in an international team with other African conservation professionals from across the continent.

Publication of this manual has been made possible through a generous donation from Rio Tinto plc. African Forest Biodiversity: a field survey manual for vertebrates

Editor Glyn Davies Assistant Editor Michael Hoffmann

Authors Leon Bennun* Department of Ornithology National Museums of Kenya P O Box 40658 Nairobi Kenya

Glyn Davies Zoological Society of London Regents Park London NW1 4RY UK

Kim Howell Department of Zoology and Marine Biology University of Dar es Salaam P O Box 35064 Dar es Salaam Tanzania

Helen Newing Durrell Institute of Conservation and Ecology University of Kent at Canterbury Canterbury Kent CT2 7NS UK

Matthew Linkie Durrell Institute of Conservation and Ecology University of Kent at Canterbury Canterbury Kent CT2 7NS UK

Illustrations by John Clarke

*Current Address: Birdlife International, Wellbrook Court, Girton Road, Cambridge, CB3 0NA, UK Published in the UK in 2002 by Earthwatch Europe.

ISBN 0-9538179-4-6

Publisher’s reference: 141-04-02

© Copyright Earthwatch Institute (Europe) and contributors 2002. All rights reserved. The use and reproduction of any part of this publication is welcomed for non- commercial purposes only, provided that the source is acknowledged.

This publication was funded by the EC Tropical Forests budget line. The authors are solely responsible for all opinions expressed in this document, which do not necessarily reflect those of the European Union.

Printed by Seacourt Press, who hold ISO 14001 and EMAS environmental certifications, using waterless printing and vegetable-based inks on chlorine free part-recycled paper.

Earthwatch Institute (Europe) is a self-governing and self-financing charity (Registered Number 327017) operating under English law with an independent Board of Trustees, and is affiliated with a global organization led by Earthwatch Institute in the US from and through which Earthwatch Institute (Europe) obtains a variety of goods and services including the right to use the Earthwatch name and access to the Earthwatch international program of field projects. Acknowledgements This manual has benefitted from discussions and the reports from two forest survey workshops: Kakamega, Kenya (1995) and Limbe, Cameroon (1996). This restructured and rewritten document has been improved with helpful comments and advice from: Tom Butynski, Tim Davenport, Rob and Cheryl Fimbel, Frank Hawkins, Dwight Larsen, Martyn Murray, John Oates, Andy Plumptre, and Justina Ray. I am grateful to all of them for their time and assistance. I am indebted to all the authors, who have stuck at this through thick and thin, and to John Clarke whose plates have greatly enhanced the visual image of the document.

Mike Hoffmann came to our aid with the final compilation and editing, and Sylvia Howe assisted with the design, layout and proof reading. Julian Laird (Earthwatch Europe) has steadfastly supported the production of this document following a visit to Limbe in 1997, and we acknowledge the financial support of the EC Tropical Forests budget line and Rio Tinto plc.

Glyn Davies Editor

Main cover photograph by Glyn Davies. Side bar photographs by Glyn Davies, except bottom photograph courtesy of Marcus Rowcliffe.

Contents

1. Introduction 1.1 Background 1 1.2 Scope of the Manual 2 1.3 Structure and content 3

2. Forest Surveys 2.1 What is forest biodiversity? 5 2.2 Forest management 7 2.3 Research into forest biodiversity 9 2.4 Ethical and legal standards 11 2.5 Preparations 11 2.6 A note on market surveys and questionnaires/ interviews 14 2.7 Health and safety 15 2.8 References 16

3. Amphibians and reptiles: herptiles 3.1 Biology 17 3.2 Management issues 18 3.3 Methods 19 3.3.1 General surveys 21 3.3.2 Drift fences and pitfall traps 23 3.3.3 Canopy walkway trap 27 3.3.4 Snake trapping 29 3.3.5 Capture, mark, recapture 31 3.3.6 Forest litter plots 31 3.3.7 Time-constrained searches 33 3.3.8 Transect counts 34 3.3.9 Territory mapping 34 3.3.10 Sound recording surveys 35 3.4 Specimen handling 36 3.5 Health and safety 37 3.6 Conclusions 38 3.7 References 39 4. Small mammals: bats, rodents and insectivores 4.1 Biology 45 4.2 Management issues 47 4.3 Methods 48 4.3.1 General surveys 49 4.3.2 Bat roost surveys 49 4.3.3 Live-trapping: rodents and insectivores 50 4.3.4 Live-trapping: bats 54 4.3.5 Capture, mark, recapture 58 4.3.6 Removal or dead-trapping 59 4.4 Specimen handling 60 4.5 Health and safety 64 4.6 Conclusions 64 4.7 References 65

5. Large and medium-sized mammals 5.1 Biology 69 5.2 Management issues 70 5.3 Methods 72 5.3.1 Hunters’ calls, attractants and observation points 73 5.3.2 Net drives 75 5.3.3 Survey walks: reconnaissance surveys and transect 77 5.3.4 Indirect methods 82 A. Dung counts 82 B. Track (footprint) surveys 86 C. Photo-recording 90 5.4 Conclusions 92 5.5 References 92 6. Primates 6.1 Biology 99 6.2 Management issues 102 6.3 Methods 104 6.3.1 Distribution surveys 105 6.3.2 Line transects 106 A. Animal sightings 107 B. Nest counts 110 C. Mapping calls 111 6.3.3 Sweep surveys 114 6.4 Conclusions 116 6.5 References 116

7. Birds 7.1 Biology 121 7.2 Management issues 124 7.3 Methods 126 7.3.1 General surveys 130 7.3.2 Timed species-counts (TSCs) 131 7.3.3 MacKinnon lists and related methods 134 7.3.4 Timed transects (TTs) 136 7.3.5 Fixed-width transect counts 136 7.3.6 Fixed-width point counts 138 7.3.7 Distance sampling 140 7.3.8 Mist netting and ringing 141 7.3.9 Sound recording 147 7.3.10 Territory mapping 149 7.3.11 Special considerations 149 7.4 Specimen handling 151 7.5 Health and safety 152 7.6 Conclusions 152 7.7 References 154

1. Introduction Glyn Davies

1.1 Background

This manual is the product of many years of forest survey experience, and is based upon discussions between field workers about ways that surveys can be improved and standardised. The first steps towards producing this manual were taken at the 4th East African Regional Database workshop, held in Kampala in August 1993, when participants expressed an urgent need for guidelines that would allow standardisation of field methods for forest biodiversity surveys. In response, the regional Global Environment Facility (GEF) Project (Institutional Support for the Protection of East African Biodiversity) agreed to fund a training workshop, as part of its Conservation and Management of Closed Forests programme. This workshop was held in Kakamega Forest Reserve and Mount Elgon National Park, Kenya in November 1994, and led to the production of a workshop report/training manual titled Guidelines for Forest Biodiversity Inventories (1995:UHNO/RAF/006/GEF). Two years later, at another GEF-supported workshop in Limbe Botanic Garden, Cameroon, many of the same forest survey issues were discussed in the context of the Central African region (March 1996). Another workshop report was produced at this meeting: Protocols for Biological Surveys in Cameroonian Forests.

1 Although both workshop reports served their immediate purposes of recording conclusions of survey experience, there were frequent requests (often from isolated project managers and field staff) for copies of the documents long after the workshops had finished. Because the obvious conclusion from both of these documents was that many of the survey methods could be applied in the field in both forest regions, the requests prompted the current collaborative effort to produce a forest survey field manual that would be distributed widely.

1.2 Scope of the manual This manual concentrates only on forest vertebrates, excluding fish. The Kakamega and Limbe workshops focused on surveys of a much wider spectrum of forest fauna and flora, and attention was also given to socio-economic survey methods. It was beyond the resources of those involved in the production of this manual to cover this full range of subjects, but it is hoped that future survey manuals can be produced to cover them. Two excellent series that describe survey methods for single taxonomic groups are the comprehensive Measuring and Monitoring Biodiversity series, produced by the Smithsonian Institution (Washington, USA), and the less- detailed Expeditions Field Techniques series by the Royal Geographical Society (London, UK). Furthermore, an excellent technical handbook Conservation Research in African Rainforests (White & Edwards, 2000) has recently been published, which focuses on vegetation and large mammal surveys in central Africa. This manual differs in that it moves away from the single taxonomic group approach, and considers the full range of vertebrates found in African forests. By so doing, we hope to raise awareness about the possibilities of carrying out surveys of several taxonomic groups at a given forest site. This does not preclude surveys focusing on particuar groups; but does encourage data gathering on other species (see general surveys in each chapter). The target audience for the manual comprises four main groups: people carrying out short reconnaissance surveys and expeditions; undergraduate and graduate students carrying out project and thesis work; research departments of forest, wildlife and national parks departments; forest and wildlife managers and technicians with responsibility for monitoring biodiversity.

2 Enlightened forest management requires information about a broad range of species, and time is too short, and resources too limited, for all forest areas to be considered by separate specialist survey teams. By explaining the range of methods available to gather information on biodiversity issues, forest managers and planners will be aware of how information is gathered, and so feel better equipped to include biodiversity in their work. The primary aim of the manual, therefore, is to provide an overview of the methods that can be used to gather information needed for effective management of African forests, which takes full account of all vertebrates as a component of forest ecosystem biodiversity. An important extension of this aim is to encourage surveyors and researchers to use standardised methods so that survey results can be used to monitor change over time, whether changes are positive as a result of management interventions or negative as a result of unsustainable use or clearance. Long-term monitoring usually involves different surveyors, as people change jobs or move, and each set of new observers/surveyors should use the same methods if the results are to be comparable. While focusing attention on this need for standardised methods, it is understood that methods continue to be improved, and different forests, survey team resources, and management questions will all require adaptation of the standard techniques. Finally, the manual is intended as a field companion, and as a training tool for students, at college and university, and in forest and wildlife services. However, this manual is not a field identification guide and the relevant identification guides will be needed.

1.3 Structure and content Chapter 2, Forest Surveys, gives a brief introduction to forest biodiversity and management, and the need for research as a tool in managing forest biodiversity. In addition, it includes introductory notes on ethical and legal standards, preparations for carrying out surveys, and notes on health and safety. Chapters 3 to 7 cover the survey methods according to each respective group of animals. Each chapter includes sections on the biology and management issues of the relevant group, the various survey methods, followed by pointers on specimen handling and, in some cases, additional notes on health and safety. Each chapter concludes with a list of references. To ease reference, the survey methods discussed in these chapters have been organised to follow generally similar headings giving: additional/special equipment or personnel required (see 2.5); site selection (where pertinent); procedure; processes of recording; data analysis; and an assessment of the advantages and limitations of the methods. To avoid unnecessary repetition, certain

3 sections have not been duplicated in each chapter, in which case the reader will be referred to the relevant section in another chapter. However, readers are advised to read the introductory sections of all chapters because different authors have stressed different issues – all of which are important whatever species group is being surveyed. The manual can be carried into the field to guide survey work, in order to ensure that the right information is gathered for subsequent analysis and report writing. It is beyond the scope of this manual to provide details of the statistical tests or analyses required to analyse and interpret field survey results accurately. Instead, such texts are referenced in the chapters (including the Smithsonian series and White & Edwards (2000)), and these should be consulted in conjunction with this manual. Sample survey forms, which can be photocopied for use in the field, are included at the back of each chapter. Given the background, some chapters put stronger emphasis on forests in eastern Africa, and others on western and central Africa. However, the survey methods described, and the principles that need to be followed, apply in all forest surveys in Africa, in Madagascar, and, indeed, even on other continents.

4 2. Forest Surveys Glyn Davies

2.1 What is forest diversity?

Biodiversity is the wealth of all life on earth, which can be considered at three inter-linked levels: genetic, species and ecosystem (see Box 1). Biodiversity is ‘... the variability among living organisms from all sources, including, inter alia, terrestrial, marine and other aquatic ecosystems and the ecological complexes of which they are part; this includes diversity within species, between species and of ecosystems.’ (Article 2, Convention on Biological Diversity, 1992). Forest biodiversity can also be considered in terms of composition, structure and function, and generally is characterised by: very high species richness – 50% of all terrestrial species in the world are found in rain forests; multi-layered structure, with giant emergent trees, forest floor herbs, epiphytic herbs and woody lianas, and a correspondingly dark understorey; often infertile soils and rapid recycling of plant and soil nutrients; long timescales over which patterns of regeneration and reproduction take place.

5 Box 1: Levels of biodiversity

Genetic biodiversity refers to the frequency and variety of genes and/or genomes within, and between, populations of the same species, and the information contained within these genes provides the basis for evolution through adaptation. Examples of genetic biodiversity are reflected in the different coat colours of mona monkeys, Cercopithecus mona, or in the yields of a plantation tree species. Species biodiversity refers to the number and abundance of species in an area, and the extent to which species differ in their genetic make-up. It incorporates characteristics such as taxonomic unique- ness, size and structure, population dynamics, reproductive cycles and behaviour patterns. Ecosystem biodiversity is reflected in the definition of an ecosystem: ‘...a dynamic complex of plant, animal and micro-organism communities and their non-living environment, interacting as a functional unit.’ (Convention on Biological Diversity). The interplay between species includes pollination, predation, parasitism and symbiosis, while the interaction between species and their non-living environment includes soil formation, photosynthesis etc. Ecosystems and human culture have influenced each other over the millennia, giving rise to productive landscapes that combine biological and cultural diversity.

Measuring forest biodiversity must therefore take account of these characteristics, and pay attention to different species’ qualities, sometimes termed ‘bioqualities’– are they common, forest-dependent, rare, insectivorous, medicinal plants, marketable timber, and so on? With such a range of features to consider, and a lack of detailed ecological information on many forest species, it is inevitable that surveys have often focused on a number of indicator species in an area as a first approximation of forest biodiversity (Noss, 1990). But this raises the question of which species to select for surveys. One obvious approach is to focus on one, or a set of, key species in relation to a particular management question, such as over-exploitation of a medicinal plant, or the unsustainable trapping of large mammals. By monitoring these ‘threatened’ species, and ensuring that their use is managed in a sustainable way, other forest plant and animal groups may also benefit. This is not automatically the case, because species-specific threats may not apply to all groups; for example, small birds are unlikely to be affected by the hunting of large mammals, and timber trees will continue to stand long after medicinal herbs have been lost. But monitoring the impacts of threats is an important first step in improving forest management.

6 A similar approach applies when surveys focus on species that are known to be rare, on the basis that if conditions are suitable for the rarest forest species, then commoner forest species will probably have healthy populations. However, the current and past causes of rarity vary greatly, and therefore limit the usefulness of rare species as indicators. Nonetheless, if a number of nationally or globally rare species are present then the condition of the forest ecosystem is likely to be good. To assess the impacts of forest habitat change across a wider range of species, a set of indicator species are often selected from (a) particular taxonomic group(s) about which there is a good body of taxonomic and ecological knowledge. For example, changes to a forest habitat have been shown to influence the population densities of forest-specialist bird species (e.g. Newmark, 1991), and the factors that have caused their decline may also affect forest specialists in other taxonomic groups. But we need to research whether ‘sensitive’ species in different taxonomic groups all respond in the same way to the same changes in the forest ecosystem. Pending such an investigation, caution is necessary before extrapolating impacts of forest change from one indicator group to another. Besides being cautious about the use of indicator species, care is needed with regard to the seasons and timescale over which surveys are carried out. In some seasons, species may be easier or harder to locate because of particular behavioural traits: breeding, migration, food abundance and so on. To get year-round data, surveys need to be carried out in different seasons. Furthermore, long timescales need to be used if ecosystem functions are to be monitored, because the impact of species declines on other processes may take a considerable time before they are felt. For example, some trees whose seeds are dispersed by elephants may decline over decades as a result of the elimination of large mammals by hunters.

2.2 Forest management From the outset, it is important to stress that conservation and sustainable use of forests can only occur if forest habitat is maintained. Forest cover can include indigenous, naturally regenerating forest, and planted/managed forests of indigenous or exotic species. Trees planted on farms are also important for biodiversity, especially where they act as corridors connecting different forest patches. However, having emphasised this basic point, not all types of forest are equally important for the maintenance of biodiversity. Most of the natural forests in Africa face pressure from communities who derive their basic livelihood from forests, or the land on which they grow, and even greater pressure comes from commercial plantation companies and

7 extractors of timber and other products. Conflicts often occur as a result of competition for forest resources from local people’s livelihoods, commerce, wildlife and forestry, and the alarming rate of biodiversity loss in African forests poses an international concern. International and national discussions and processes, such as the national Tropical Forest Action Plans, the UN Forum on Forests and the International Timber Trade Organization, have all been developed to address this problem. These processes are mirrored by the conferences of the Convention on Biological Diversity, and its related Biodiversity Strategies and Action Plans, which include decisions on forest biodiversity. While these international policy processes evolve, there is a pressing need to address the conflicts on the ground, and inform policy debate with appropriate information on the range of uses that forests can fulfil. Also, it is at the field level that decisions need to be taken, by the owner/steward of a forest area, on the management aims for forest areas.

Box 2: Examples of forest management aims

Ensure high quality fresh water at acceptable flow-rates, minimise erosion and movement of soil, and stabilise hillsides, through forest management in watersheds. Conserve a representative sample of a biological region (province, biome or habitat) in a state relatively unaltered by modern man, and avoid the loss of species and erosion of genetic diversity. Maintain areas and features that are essential for ecological processes, such as migrations and biological cycles, and rehabilitate degraded areas. Protect sites of cultural or archaeological importance. Ensure the supply of wood and non-wood products to satisfy local/national/international demand. Provide facilities and opportunities for tourism, recreation, environmental education, research and monitoring. Retain a maximum choice of land-use options for the future.

Developing and implementing forest management plans through consul- tation processes, involving civil society (especially local communities), government and the private sector, allows different forest users’ needs to be taken into account. The management plan can then include actions to prevent damage to ecological services, and limit loss of genes, species and forest habitats, while forests continue to supply important goods and services. Drawing up

8 multiple-use management plans to address this range of issues requires a number of different types of information.

Box 3: Sets of information needed for forest management planning

Physical features: location; area; altitude; topography and drainage; infrastructure (including villages); climate. Biological resources: biodiversity; abundance and yields of commercial species. Social, policy and legal framework: the population density, with an assessment of the proportion and distribution of indigenous, local and recently arrived groups; patterns of forest use by different groups; and national and traditional laws relating to land ownership, forest use and management, and their effectiveness. Economic context: what economic policies and market forces are influencing the rates at which different forest resources (including forested land) are being used?

It is the responsibility of those carrying out biodiversity surveys to present their results in a form that can be understood by people interested in managing the forest (not just technicians and scientists), so that social, economic and biological information can be integrated, and the information understood by local people, government and private enterprise. The results also need to show the links between the details of forest resource availability, and the bigger picture of national or regional patterns of forest use and national development. Survey questions should be guided by information from household surveys on patterns of forest use, and market surveys of commercial patterns of use, road-side sales, bushmeat markets, and so on (see section 2.6). The links between in-forest and out-forest data collection need to be carefully considered.

2.3 Research into forest biodiversity Carrying out the forest surveys described in this manual will help identify research needs, and the results of these surveys should also provide the first steps in answering many research questions. Although time is one of the greatest restrictions on surveys, it is important to note that data becomes more robust as it accumulates over days, weeks, months and years. This manual focuses on getting started, so that the presence/absence of species can be assessed, and the relative abundance of some of the commoner and

9 more conspicuous species gauged. For more thorough ecological monitoring, poor visibility in forests requires that a great deal of time and effort be expend- ed before reliable results can be obtained (e.g. Walsh & White, 1999; Plumptre, 2000; White & Edwards, 2000). Perhaps one of the most important constraints to gathering useful information for multiple-use management is that inter-disciplinary research teams are few, and there has been little investment in developing research or survey methods that integrate biological, social and economic information. This is a key area of research that needs to build upon the foundations laid by ethno- biological and socio-economic studies, so that the interests of different stake- holders can be included in planning processes. This is especially true for those indigenous and local communities that already possess a wealth of knowledge about biodiversity and its management. There are precedents for carrying out participatory forest surveys, where local specialists plan and implement surveys, in collaboration with other parties, to agree on resource abundance, or to monitor patterns of forest use. Indeed, programmes for training ‘para-taxonomists’ have been developed, in which local experts are given training which enables them to integrate their knowledge with scientific and technical information. In this context, it is important that all fieldwork conforms to international standards on ethical and legal practice in the field (see below), respecting local knowledge. Immediate priorities for applied research include developing survey methods that can be used for rapid, problem-oriented inventories and monitoring. One of the subsidiary aims addressed in this manual is to focus attention on the need for more biodiversity information to be incorporated into Environmental Impact Assessment (EIA) checklists for forests in Africa. To date, little has been done to develop biodiversity criteria or indicators that can be used to assess or monitor impacts of road-building, agriculture and other developments on forest biodiversity in adjacent areas. While answering these pressing management questions, pure research into forest biodiversity continues to be vital, and many of the summary state- ments about ecology that are made in the chapters that follow are based on long-term and meticulous research efforts. Furthermore, the information accu- mulated at long-term sites, with well-studied populations, is essential for calibrating results from rapid surveys with known population figures. We still need research to understand what controls plant and animal population densities, and what elements of plant–animal interactions are central to maintaining healthy forest ecosystems. In addition, continued research in the field of taxonomy is necessary to ensure that accurate and consistent species names are attributed to field records.

10 2.4 Ethical and legal standards Whether field activities are short or long-term surveys, and whether carried out by national or visiting scientists, international standards of ethical and legal practice need to be followed (e.g. Fauna & Flora International, 2000). These have been compiled by a number of institutions, especially those concerned with anthropological work, and the reader should refer to the full texts if there is any uncertainty about planned actions. In general, care needs to be taken: to ensure that official research permits, including collecting permits and equipment import licences, have been provided, and that a sponsoring national institution has approved and supports the proposed survey work. Also ensure that any products that arise from the work (including reports, books, scientific papers, films, etc.) acknowledge the sponsoring institution, and provide copies to them and other government departments. to endeavour to work with and through local institutions, building from their capacity and taking their advice. Wherever possible con- tribute to building local capacity. When employing local field assistants ensure that local labour codes are respected. to collect animal specimens in a humane and ethical manner, with as few specimens collected as necessary to satisfy scientific needs, and with the absolute minimum amount of pain or suffering inflicted upon the animal. to take account of beliefs, customs and rights of local communities, and guard against the appropriation of their intellectual property.

2.5 Preparations Successful surveys require careful planning and preparation. In particular, you must think carefully about the purpose and objective of your survey, as this will determine the information that you need to obtain, and thus the methodology that is most appropriate. In addition, before you start surveying, you need to think how the data will be analysed. This is vital in order to develop an appropriate sample design. Although detailed discussion of data analysis is beyond the scope of this manual, some simple considerations are provided to help you ascertain whether the chosen method will prove useful for statistical analysis. Tips on identification are provided in each chapter. However, it is well worthwhile spending some time in museums inspecting skins, perusing field guides, and taking opportunities to visit field study sites. Through all of these

11 methods, and by speaking to knowledgeable people, it is a good idea to start compiling a species list for the forest sites, or general regions, that will be visited. Besides general reference books, atlas projects can provide very useful indications as to the possible occurrence of a species in a particular area. For example, bird atlas projects are underway or completed in a number of countries (for example, the whole of southern Africa, Kenya, Tanzania and Uganda), and a list of species recorded for a particular atlas square (or point, in the case of Uganda) can usually be produced on request to the coordinators. As well as being able to identify particular species, many of the survey methods described here rely on accurate estimation of distances. It is very important to practise distance estimation before you start your work. If you are using a cut-off point of 25m, for example, go into the forest and estimate this distance, then measure to see how accurate your estimation was. Continue practising until you can estimate this distance reliably in this habitat. Indeed, it is important to practise this in a similar vegetation type to the transects; distances appear very different in the open when compared to dense forest, and stride lengths tend to become much shorter when hopping over logs, running away from driver ants, or wading through a swamp. It is crucial that all those who are counting are accurate in their distance estimation. The latter point, namely the discrepancies that result in distance estima- tions as a result of people judging distances differently (see Mitani et al., 2000), argues well for the use of an optical range finder. The reliability and accuracy of optical range finders has made them an invaluable tool in the field, all the more so because all transect methods assume distances are exact. It takes little time to learn how to use an optical rangefinder properly, and they generally are inexpensive given the costs of a survey.

Survey equipment Although each chapter makes reference to special equipment and/or personnel necessary to conduct the individual survey methods, there is some basic equipment that is common to all surveys and may be considered as essential items to be carried into the field. It goes without saying that suitable clothing, footwear, field bags and camping equipment are basic necessities.

notebook (with plastic bag for rain protection): many people prefer to use a loose-leaf binder, so that only the notes for a particular field session are taken to the field. Previous notes can then be kept elsewhere for safety, and photocopied as soon as one returns from the field session. The importance of keeping duplicate records (either by using carbon paper in the field, or by photocopying), or of backing up information electronically, cannot be over-emphasised.

12 The advent of hand-held computers or personal data assistants (PDA’s) appears set to revolutionise data input/collection. data recording sheets or forms: these can be designed and photo-copied in advance, or simple formats can be reproduced daily in a notebook or binder. topographic maps of the survey area, on as large a scale as available, and map of trails, footpaths, etc. if available (you may have produced your own map from reconnaissance surveys). prismatic compass (in a protective case): essential, not only for making maps and determining survey routes, but also to help teams return to camp if they get lost. pencil/pen: propelling pencils, which need no sharpening, are most convenient, or pens with waterproof (India) ink. Ordinary ballpoint pens are NOT recommended for data recording: the ink is not waterproof, and your data sheet or notebook will be a mess if it gets wet. torches (preferably six-battery) and headlamp for night-time work (spare bulbs and batteries are essential). watch and/or stopwatch (should be easy to read in dim light conditions). field identification guides: these are discussed in more detail under the ‘Methods’ section of each chapter. Avoid the use of large, cumbersome reference works (which are best consulted back in the office/laboratory), and stick with lightweight, compact field guides. A species checklist for the area (if available) is advisable, or a preliminary list compiled from expected occurrences. binoculars: these are the most essential piece of equipment for surveys of larger mammals, and especially birds. Binoculars are normally labelled as 7x30 or 8x40, and so on. The first figure repre- sents the order of magnification, and the second the diameter of the objective lens measured in millimetres. The larger the second figure is, the greater the light-gathering potential of the lens. For forest work, a wide field-of-view and plenty of light-gathering capacity is best. The best magnifications are 7x and 8x; higher magnifications (10x) may allow you to identify birds in the treetops more easily, but will be less effective for more close-up work. The objective lens should be at least x40. A telescope (mounted on a light-weight tripod) is surprisingly useful for identifying treetop birds. Ideally, both binoculars and telescopes should be weather-proof; if they are not, then carry strong plastic bags for protection against rain (zip- lock bags are useful if available).

13 photographic equipment: a good camera is often useful for taking photographs of survey areas, different types of habitats, evidence of human activities, captured or surveyed specimens, etc. Equipment can range from inexpensive instamatic cameras that provide basic records of survey areas or details of animals and their signs, to expensive telephoto equipment for quality images that can be used in campaigns to raise awareness and as education materials. Different film speeds, an assortment of lenses, a flash (for photography in poor light conditions), and a protective bag are recommended. In addition, the increasing availability, resolution and affordability of digital cameras means that they are now a very valuable tool for specimen identification and for permanent recording of habitats trapped in, and so on. A quick digital image of each trap line (or specimen) is a cost-effective, or at least a valuable, addition to written descriptions of habitat (or specimens). optical (or laser) range finder (for estimating distances). (optional): Global Positioning System (GPS) for recording the start and end of transects, or positions of point counts. a first-aid kit (see below).

There is, as they say, no substitute for experience, and if you have not conducted many surveys it is important to try to have in your team experienced surveyors to help you.

2.6 A note on market surveys and questionnaires/interviews Although not discussed for each group, there are two additional survey methods which have great relevance for surveying African forest vertebrates: market surveys and questionnaires/interviews. Local markets can produce some interesting information about the local fauna, especially for general surveys. This is especially so for vertebrates in West and Central Africa, particularly in light of the boom in the bushmeat trade. In many markets there are stalls selling dead mammal species such as small antelope (particularly duikers), monkeys, chimpanzees and gorillas, pangolins and rodents, especially canerats (or grasscutters, as they are known in West Africa). However, determining the origin of market carcasses is often very difficult, especially for smoked meat that has been trucked a long distance and been through the hands of a number of intermediaries.

14 At the local level, even more important are interviews with local people ranging from local farmers, and particularly those that hunt, to forest, wildlife, national parks and other government officers. These people have considerable knowledge about animals in the area where they live and work, but care must be taken in verifying their verbal reports. For example, many villagers fail to recognise animals from pictures in guidebooks, and misunderstandings can arise through the misuse of local names, etc. Hunters, who tend to be most knowledgeable, are often reticent on the whereabouts of their future bag. Only clear descriptions and explanations, ideally from independent sources, should be recorded, and these should only be added to the dataset after further verifi- cation. Having noted the limitations of market surveys and interviews, which can only be overcome through several months of field work, much information on forest animals’ ecology, population status and levels of threat can be obtained. This has been done effectively for forest animals in: Sierra Leone (Davies & Richards, 1991); south-east Nigeria (Angelici et al., 1999); Democratic Republic of Congo (Dupain et al., 2000).

2.7 Health and safety Survey work involves many health and safety risks, including injury, infection, and disease. Safety precautions during observational surveys are largely common sense, and include wearing the correct clothing and using the right equipment, and, if possible, working in pairs rather than alone. When doing any kind of work off well-used trails, ensure that you have a GPS or compass, and a map, and that other members of the team know where you are working. Good survey management dictates that a basic first-aid kit is carried along on field work at all times, that everyone knows where the first aid equipment is, and that everyone knows how to use it. Discuss possible problems with medical experts before starting the expedition so that the first aid kit is correctly and appropriately equipped. Working with large mammal species, like primates, ungulates and carnivores, carries with it obvious risks. Be constantly on the alert and aware of your surroundings, being careful not to become so focused on a particular bird or reptile that you lose awareness of other, more dangerous wildlife. Handling small mammals, herptiles and birds also carries the risk of infection from disease or ectoparasites. Protective gloves and a surgical/face- mask may be necessary depending on the type of work being done. Always wash your hands thoroughly after handling any animals.

15 A wise precaution before undertaking any survey work is to get injections against tetanus and rabies. This may require more than one injection, so allow time for the full course before the fieldwork begins. Similarly, courses of anti- malarial tablets often need to be started several days before departure (and should be continued for several weeks after leaving the malaria area). Additional notes on health and safety, where relevant, are provided in the individual chapters.

2.8 References

Angelici, F.M., Grimod, I. & Politano, E. (1999). Mammals of the Eastern Niger Delta (Rivers and Bayelsa States, Nigeria): An environment affected by a gas-pipeline. Folia zool. 48(4): 249–264.

Davies, G. and Richards, P. (1991). The Rainforest in Mende Life. Unpubl report. Escor/ODA, London.

Dupain, J., van Krunkelsven, E., van Elsacker, L. & Verheyen, R.F. (2000). Current status of the bonobo (Pan paniscus) in the proposed Lomako Reserve (Democratic Republic of Congo). Biol. Cons. 94(3): 265–272.

Fauna & Flora International. (2000). Code of conduct for researchers. Oryx 35 (2): 99.

Mitani, J.C., Struhsaker, T.T. & Lwanga, J.S. (2000). Primate community dynamics in old growth forest over 23.5 years at Ngogo, Kibale national park, Uganda: implications for conservation and census methods. Int. J. Primatol. 21: 269–286.

Newmark, W.D. (1991). Tropical forest fragmentation and the local extinction of understorey birds in the East Usambara Mountains, Tanzania. Conserv. Biol. 5: 67–78.

Noss, R.R. (1990). Indicators for monitoring biodiversity: a hierarchical approach. Conserv. Biol. 4: 355–364.

Plumptre, A.J. (2000). Monitoring mammal populations with line transect techniques in African forests. J. Appl. Ecol. 37: 356–368.

Walsh, P.D. & White, L.J.T. (1999). What it will take to monitor forest elephant populations. Conserv. Biol. 13: 1194–1202.

White, L. & Edwards, A. (2000). Conservation Research in the African Rain Forests: a Technical Handbook. Wildlife Conservation Society, New York, USA.

16 3. Amphibians and reptiles: the herptiles Kim Howell twig snake (Thelotornis kirtlandii) (Thelotornis

3.1 Biology

Herpetology is the study of amphibians and reptiles, and these two groups will be collectively referred to as herptiles in this chapter. Amphibians (Class Amphibia) The best known amphibians are the anurans, the frogs and toads, of which there are about 161 species in East Africa, supplemented by about 10 species of apodans or caecilians (legless forms) found in forests in Kenya and Tanzania; there are about 600 anurans and 22 species of caecilians known from the continent. Apodans live mostly in moist soil, and emerge only after heavy rains, and, although the same can be said for many anurans, our discussion of survey methods focuses on frogs and toads, excluding the few species which are entirely aquatic. Furthermore, the assessment of larval forms is only briefly addressed; this topic is covered in more detail by Heyer et al. (1994) in their detailed review of amphibian survey methods. Most frogs and toads are extremely seasonal in their reproductive behaviour. In drier periods, many seem to simply disappear; they seek shelter where they will not be exposed to desiccating conditions, and are not seen or heard during the daytime or at night. However, during the rainy season(s) amphibians emerge and become much more active. They may still remain relatively hidden during the daytime, but at night male frogs and toads of many species produce loud vocalisations that serve to advertise their presence in order to attract females and also to defend their territories from other males.

17 Reptiles (Class Reptilia) It is mostly the snakes and lizards that occur in forests, though terrapins may be associated with wetlands in forests, and tortoises are occasionally found at the forest edge. These come from the following groups of reptiles: lizards (including geckoes and chameleons); snakes; amphisbaenians (or worm-lizards); chelonians (marine turtles, freshwater terrapins, and terrestrial tortoises); and crocodiles. In sub-Saharan Africa, the approximate numbers of species for these groups are as follows: chelonians (excluding sea turtles), 26; lizards, 680; amphisbaenians, 66; snakes, 466, and three species of crocodiles. Much of our knowledge of reptile distribution in forests is still rather limited, being restricted to preliminary species lists. For example, in Tanzania at least four species of lizard and two snakes new to science have been found within the past few years (Broadley, 1994, 1995a,b; Broadley & Wallach, 1996; Pasteur, 1995) and more surveys are needed. Reptiles are found from below the soil level to the tree canopy, so there are a variety of forms ranging from the fossorial to the arboreal.

3.2 Management issues Worldwide, amphibians seem to be declining for poorly understood reasons (Wyman, 1990). Recent research indicates that two cases of frog mass extinctions in rain forests are a consequence of fungal pathogen attacks (Berger et al., 1998), and elsewhere tadpole deaths from fungal attack have been linked with climate change and ultra-violet radiation (Kiesecker et al., 2001). We do not have data on the impacts of pesticides, but given the increasing use of agrochemicals, and the general increase in aquatic pollution, there is a need to monitor the levels of pollutants in the environment as well as in the tissues of amphibians. As in other groups of vertebrates, there seems to be a basic split between non-forest and forest species, and there is also a distinctly forest- dependent element that does not occur outside closed forest (e.g. Howell, 1993). The forest-dependent amphibians are vulnerable to forest alteration and/or clearance, and are under threat in many parts of Africa. We do not yet understand the physiological reasons for the dependence on forest, but the number of hiding or retreat sites is a possible critical factor in limiting tropical forest anuran populations (Stewart & Pough, 1983; Howell, 1993), especially if forest quality is altered, and/or forest patch size decreases. It is worth noting that this chapter focuses largely on amphibians which are forest dwellers, and which make use of temporary pools for breeding, or which are independent of free water for reproduction (e.g. Nectophrynoides

18 spp. toads, Arthroleptis spp. frogs, the bush squeakers, and some microhylid frogs). However, recent studies have shown that even savannah species that breed in seasonal wetlands rely on forest as a dry season refuge. Aside from local extinctions of forest-dependent amphibians and reptiles as a result of forest loss and degradation, the isolation of once continuous populations can be another problem. At present, no data exist as to the long-term effect of such isolations. A corollary to this is that areas of degraded forest may effectively become a means by which non-forest dependent species can invade; this already has occurred in many places along forest roads. In Africa, the paucity of information makes it difficult to develop predictive models for the abundance of amphibians and reptiles based upon capture data and measures of habitat quality. As many species of forest-dependent reptiles (and a few amphibians) are sought after by collectors for the live animal trade, there is a particular need to be aware of this pressure, especially on populations in already isolated forests, or near sites frequented by visitors. Commercial collecting, for example of chameleons, should be discouraged in such places. The larger reptiles would appear to be relatively long-lived, and intensive commercial collecting in a small area may have significant effects on populations. Amphibians and reptiles form an important part of the forest ecosystem, where they are significant predators on invertebrates as well as smaller vertebrates, and they themselves are important food items for birds and mammals. This also applies to large snakes (whether venomous or not), which eat many rodents, and can therefore also be beneficial to villagers. A management issue peculiar to snakes is that because some of the larger, more conspicuous species are venomous and potentially dangerous to man (see section 3.5), snakes in general are often regarded as harmful, and killed. In fact, relatively few species of snakes are dangerous to man, and it is important for managers to be aware of this, and also to educate others that it is not necessary to kill all snakes. This will undoubtedly encounter cultural resistance where snakes, and other herptiles, are part of local peoples’ belief systems.

3.3 Methods

General Our lack of knowledge about most forest herptiles means that current surveys deal largely with the building up of species lists, rather than indices of abundance or population studies (e.g. Broadley & Howell, 1991; Drewes & Vindum, 1994). Efforts have been made, in West Africa (Barbault, 1975) and in

19 eastern Africa (e.g. Western, 1974; Kreulen, 1979), to estimate populations of reptiles, but these have usually been of large and/or conspicuous species in open, drier habitats. There are, therefore, few standard methods that have been used to quantify amphibian and reptile populations. Those that have been used in Africa have dealt with forest, leaf-litter-dwelling anurans and reptiles in Cameroon (Scott, 1982), or anurans in open areas such as small seasonal breeding ponds (Bowker & Bowker, 1979). No satisfactory methods have yet been developed to sample the arboreal tree frogs, the fossorial apodans, or canopy-dwelling reptiles that are the focus of this chapter. Chameleons, for example, are hard to detect during the daytime and are best surveyed at night (e.g. Broadley & Blake, 1979; Jenkins et al., 1999), while snakes are very mobile and also difficult to detect. Studies of populations of anuran larval populations also appear to be lacking in East Africa. Identification A number of useful references and field guides are available for identifying amphibians. Frost (1985) edited a world list of amphibian species that serves as a basis for national and regional lists; in addition, African tree frogs are covered by Schiotz (1999). Other African national lists include: Stewart (1967), which provides a general introduction to some of the species found in eastern Africa; Rodel (2000), which covers many savannah species in West Africa; Fischer and Hinkel (1992), which describes Rwandan forms; Passmore & Carruthers (1995), covering South African frogs, many found in open habitats; Pitman (1974) which covers the snakes of Uganda, and Lambiris (1989), which gives useful general information on the biology of many species in Zimbabwe, including drawings of the tadpoles. At least two of the newer guidebooks feature CD-ROM recordings of frog vocalisations (Passmore & Carruthers, 1995; Rodel, 2000), which allow species identification without the need for specimen collection. Field guides currently available for reptiles are Branch (1998), which covers southern Africa and includes many common woodland (but not forest) species found in eastern Africa; MacKay & MacKay (1985), which gives details on how to identify venomous snakes in East Africa; and Broadley & Howell (1991), which provides a key and annotated list of species for Tanzania. Spawls et al. (2002) will provide coverage for East Africa, including Rwanda and Burundi. Older references include: Spawls (1978), which lists snakes of Kenya, and Schmidt & Noble’s (1919–1923) recently reprinted descriptions for West Africa. Glaw & Vences (1994) should prove useful to anyone conducting surveys of Madagascan herptiles.

20 3.3.1 General surveys When visiting an area for the first time, either to begin building up a species list, or to carry out a rapid assessment of sites for future studies, a general survey can be carried out to gather basic information. General surveys provide at least a minimum of information on species which may be present in an area. It is usually after a general survey has been conducted and unusual and/or interesting species found that more detailed studies are conducted. In most cases, a general survey will be done over a short period of time using qualitative rather than quantitative methods of sampling and with little or no strict sample design. Nevertheless, general surveys are a useful way to involve local residents in participating in activities and win- ning their good will and confidence; they often have a detailed local knowledge of particular species or habitats, and without their assistance a survey will usually fail to detect even common species which may be present. Equipment

cloth bags (various sizes – 80mm x 500mm to 140mm x 1000mm) plastic bags for specimen collection short handled rake or hoe for turning stones, logs, etc. gardening gloves snake tongs or grabbing stick (for picking up and handling snakes) a shorter type of grabbing instrument (like artery forceps) for grabbing small snakes, or controlling the heads of larger ones that have been grabbed or pinned down with a larger stick lizard noose (a loop of string held on a stick which permits you to slip it over the head of a lizard) catapult for collecting specimens from the canopy weighing scales Site selection and procedure i) It is necessary to take a number of different approaches during a general survey in order to establish whether amphibians and/or reptiles are present. During the daytime, surveying under relatively dry conditions, you should search hiding places, such as inside rotten logs, under bark, in leaf-litter at the base of trees (especially between tree buttresses), and in any tree cracks or holes. Old pit sawing sites in the forest, with the associated moist sawdust and rotting stumps and planks, have also proved especially productive. For microhylid anurans, crevices in road cuts and banks of soil should also be searched. ii) More and more emphasis has recently been placed on the need to sample not only adult amphibians, but also the larvae (e.g. frog and toad tadpoles). For many African species, these have not yet been described, and in

21 many cases the adults may no longer be in the area, leaving only the tadpoles as evidence of the species’ presence. There is thus a need to sample tadpoles, catching them with simple nets (mosquito netting sewn onto a small wooden or metal frame is fine for non-quantitative methods), seeking them in aquatic vegetation and under rocks and logs in pools where they like to hide. The eggs of amphibians can also be diagnostic of a species, and these should be collected. iii) It is useful to raise the tadpoles, from eggs if possible, in order to monitor their features as they develop, and eventually identify the species once the larvae have metamorphosed into adults. This involves keeping tadpoles in jars; many are filter-feeders and will survive on the water from the collecting site, as long as it is regularly changed. Each larval developmental stage should be collected and stored in 10% formalin solution, with careful labelling. An alternative is to collect from the site over a period of time that allows all stages of development, from egg to adult, to be collected. However, either method is time consuming, and may not be feasible during short survey visits. A more detailed and quantitative approach to estimating tadpole densities is given by Heyer et al. (1994). iv) Night-time surveys for amphibians, and perhaps some geckoes, involve listening for (and making tape recordings of) vocalisations, and visual searching of suitable resting sites with headlamps and torches. Streams that flow through forests can be sampled using both pit-fall traps, as well as audio transects (as long as the background noise of rushing water does not block out the frog vocalisations). Remember that while some frogs and geckoes live on the forest floor, others are found at varying heights, at least up to 5m, on trees. v) Reptiles can often be collected in similar situations to amphibians, namely under rocks, the bark of trees, and so on. They may also be seen basking on and above the ground. Indeed, it can be helpful to place sheets of metal, wood and cardboard besides tracks and roads to attract reptiles. Night- time collecting is also required for some snakes, geckoes (none of the east African species give loud vocalisations), and chameleons; the latter are often visible when asleep, clinging to vegetation at different levels above the ground. Be especially aware of the need to collect fossorial and burrowing forms, such as blind snakes and legless lizards – these are seldom sampled and poorly known. It is important to realise that general surveys are likely to under-represent the larger species, such as ridged or grass frogs (Ptychadena), and some tree frogs. vi) Indirect methods may be employed to detect the presence of a species, such as through information from local inhabitants (indigenous/local knowledge) and by examining the faeces/scats from some predators. The identification of reptile bones in owl pellets, and prey remains of large raptors, especially crowned eagles (see Msuya, 1993), as well as reptile and amphibian

22 remains in mammal faeces, may all provide information on the presence of a species (see Yalden, 1977). This is important in the case of apodans and legless lizards that are often not detected by conventional trapping but may be prey items of larger snakes. vii) Information gathered during these general surveys could be used to stratify a large area into separate zones with characteristic differences (e.g. marshy; woodland; near rivers) that influence herptile distribution and abundance. Thereafter, longer surveys and studies may be carried out in each of the various zones so that a full picture of the herptiles in a forest region is obtained. Recording i) Record each individual animal and its species name (Form 3.2). If you do not know its name, then call it ‘species a’, etc. ii) Take the standard measurements for each specimen along with any geographical and habitat information. General data required is similar to that collected for small mammals (section 4.4), with some obvious differences. In particular, some additional notes and measurements should be recorded including snout length (in the case of reptiles), iris colour/shape, and any other notes on anatomy, such as eyes (e.g. protruding) and feet. iii) Detailed notes need to be made of the colouration of the animal, supported, if possible, by colour or digital photographs (some people prefer to take these after the animal has been anaesthetised). Colour is a key feature in the identifications of amphibians, and once immersed in preservative the bright colours often fade to brown or white. iv) Collect and preserve voucher specimens (see section 3.4) Advantages/limitations The general survey technique continues to yield important information in East Africa, but is best used in conjunction with some of the following techniques such as pitfall traps and litter searching. Unfortunately, it does not provide information concerning populations, and it is difficult to quantify the results obtained from this type of collecting survey, especially given seasonal and annual variation.

3.3.2 Drift fences and pitfall traps This is a method that has been employed recently for sampling small mammals in forests, and has been shown to be especially effective in sampling leaf-litter frogs of the genus Arthroleptis, as well as Bufo toads and forest floor lizards.

23 The basic principle behind this trapping method is that animals on the forest floor encounter a barrier termed a ‘drift fence’ which causes them to drift into the trap. Rather than cross the fence, burrow under it, or break through it, they take the route of least resistance by moving either right or left and following the fence – which leads them to drop into a pitfall trap. A variety of patterns of arrangement for the bucket pitfall traps have been used, with varying degrees of success (see Bury & Corn, 1987). Presented here is the simplest arrangement: a drift fence in a straight line, termed here a ‘pitfall line’ (Fig. 3.1), using 20-litre plastic buckets for the pitfall traps and plastic sheeting for the drift fence.

Fig 3.1: Drift fence and pitfall trap

pitfall bucket wooden stake plastic sheeting

Equipment eleven plastic buckets (size: 20 litre), or any reasonable alternative which is sufficiently wide to prevent animals jumping across, and sufficiently deep to prevent them jumping out (for example, large empty tins). It is important to use containers that are readily available. The buckets should have covers so that they can be closed in rainy weather or when they cannot be monitored. plastic sheeting (length: 55m). This may be transparent or coloured; if the latter, black is likely to be a better choice, since bright colours may influence the catchability of some species. The exact height and thickness are not critical and may be determined by availability; plastic sheeting is often sold as a roll of open-ended tubing in approx. 0.5m widths, so this can be cut and used as a single thickness. Locally available alternative material may be equally effective (e.g. polypropylene gunny sack material, etc.). wooden stakes to support the plastic sheeting. staple gun and staples for attaching sheeting to supporting stakes (alternatively, you can punch holes in the sheeting and tie it to the stakes, but this requires much more time).

24 a hoe and pick suited to cutting through roots. measuring tape or string of measured length. Site selection Moist or low ground is usually a good place to trap, especially at the bases of valleys, but try to set pitfall lines in a variety of situations and habitats for comparison. Remember that altitude is an important variable in the distribution of amphibians. For comparisons to be made, the same method should be used at each sample site. Procedure i) Configuration of trap lines: each trap line is 55m long, with buckets placed every 5m. Thus, with one bucket at the beginning and one at the end, the total number of buckets is 11. There is nothing sacred about using a line 55m long, and you can always use a shorter line depending on local circumstances, but it is best to try to standardise the length and set-up so that you can compare catch rates between different sites. ii) Each bucket is sunk in the ground so that the upper rim is equal to, or slightly below the ground-surface level. This is very important; apparently, few animals will climb a mound of soil at the edge of a bucket, whereas if the bucket is level with or slightly below the soil surface, they readily fall in. iii) The plastic drift fence is erected so that the line is continuous from the first to the last bucket in each line. The fence should pass over the centre of each bucket. The best way to keep the plastic erect is to drive stakes into the ground along the drift fence and then to staple the plastic sheeting to the stakes. A stake is needed on either side of each bucket, and then at least four to support the area of plastic sheeting between consecutive buckets. The line does not have to be straight, and probably will have to curve in places in order to avoid trees, large rocks, and other obstacles. iv) It is important to clear away vegetation that gets in the way of the plastic drift fence. The plastic sheeting must rest flush with the ground; if vegetation props the bottom open, animals will move underneath the fence. Once the fence is in place and properly stapled, it is time to mound up a little soil on both sides of the fence (to further inhibit animals from pushing underneath). v) It is good practice to number each separate pitfall line, and then, within each line, number each bucket. This is best done with flagging tape attached to a nearby stick. Try to map the pitfall line exactly using a GPS (Global Positioning System) or use longitude and latitude coordinates to allow surveyors to come back to the exact spot and repeat the sampling. You also may wish to establish permanent markers indicating the sites of your pitfall line.

25 vi) Remove litter such as leaves and soil from buckets daily. Check buckets regularly for stones or branches that may have fallen in as these may allow animals to climb out. To ensure rain water drains from buckets, puncture the bottom of each, and the sides if necessary (this also reduces the attractive- ness of the buckets to local residents!). vii) Important note: When checking pitfall lines in early morning and late afternoon or at dusk, or any other stage when the light is deteriorating and visibility is poor, always use a torch when examining the contents of the pitfall buckets. Scorpions, centipedes and venomous snakes may be in the bucket in addition to the frog you are about to pick up. Look carefully before you put your hand in the bucket! viii) After you have finished your trapping regime and removed the pitfalls and drift fence, fill in the holes that were dug for the buckets. This ensures that no animals will accidentally be trapped and no larger animals will injure themselves when walking in the area. This will also enable the site to recover quickly, an important point if you wish to sample the same area at another time. Recording i) Detailed notes should be kept on local habitat, soil type, amount of leaf litter, ground cover, etc. for each pitfall line (Form 3.2); those describing the exact surroundings of each bucket may also be taken. ii) The number of animals captured per night in each bucket should be carefully recorded; these data can then be combined to calculate the trap success of each pitfall line. It is important to record from which bucket each animal came, rather than recording just catch per line. This permits later analysis, which may indicate which features of the microhabitat are related to the capture of particular species. iii) Information on captured species should be recorded on the standard form (Form 3.1). Data analysis i) Most general survey work involving pitfalls will simply record trap success, i.e. how many captures per number of (bucket) trap nights. This may then be compared with similar trap success for other types of traps, such as break- back traps in the cases of small mammals. ii) Trapping success of pitfall lines at different elevations, or in different habitats, may also be compared. iii) If adequate data are kept over a long period of time, it may also be possible to study microhabitat features by comparing trap success of buckets, for example, which are located near fallen logs and those which are not. iv) Simple graphs can be plotted of cumulative number of species (CNS) and cumulative number of individuals (CNI) against cumulative number of trap nights. If the CNS curve flattens out by the end of the sampling period, with few or no species added during the last nights of trapping, then sampling likely has detected most of the species which are detectable using that method. 26 Advantages/limitations i) The advantages of this technique are that it is easily repeatable, can easily be modified to suit local conditions, and can be used to sample mammals, amphibians and reptiles simultaneously. It is also one of the few techniques that can be used to sample apodans and burrowing reptiles that occasionally emerge on the forest floor. ii) It can be used as a non-destructive technique, and permits the mark- recapture method of population assessment (see below). However, this technique only samples members of the forest floor community, and so will not sample tree frogs, for example. iii) Occasionally, animals such as bushpigs or small antelopes may wander through the drift fence, necessitating repairs. It is good policy to check your pitfalls first thing in the morning, and then again in late afternoon. A quick check an hour or two after sunset will also allow you to detect early arrivals, and these can be removed so that they do not spend the night exposed in the bucket; in dry, cold forest, some amphibians may perish overnight from dehy- dration and/or exposure, thereby necessitating more frequent checks.

Fig 3.2: Drift net fence

3.3.3 Canopy walkway trap A method of setting traps on walkways in the canopy. It involves constructing a runway of mosquito mesh into which is sewed a funnel-shaped bag – the funnel trap. The entire construction is then raised by means of rope and pulleys into the canopy (see Vogt, 1987 for details and a photograph). This method works on the principle that reptiles will make use of walkways to travel through the forest canopies.

27 Equipment

plastic mosquito gauze (also called mosquito mesh or window screen) galvanised wire metal cutting scissors or tin snips pliers and wire-cutters steel rods (diam: 5mm; length: 1m), two for each walkway pulleys (x 4); several walkways can be tied at different heights to the same rope and pulley assembly) nylon or other non-rotting rope to fit pulleys Site selection This technique requires a site where the walkway will be touching as many trees and branches as possible, but will still permit the use of pulleys to raise and lower it. Procedure i) The basic feature of this method is a walkway made of plastic mosquito mesh (window screen) (Fig. 3.3). Simply use this material in its standard width (approx. 1m), and in 15-m lengths (or any other length you can conveniently handle in the field). Metal wire (galvanised if possible to prevent rusting) is threaded along the entire length of the plastic window screen on both edges; a length of the same wire is also inserted crosswise at 1-m intervals. These cross-wires serve to give support to the walkway. A 1-m length of steel-rod (diam. 5mm) is tied at each end of the screening to ensure a flat entrance onto the walkway. ii) Two funnel traps (1m x 0.8m) are then constructed of 8mm x 8mm galvanised wire mesh and are sewn onto the walkway using galvanised wire at 5-m intervals. The mouths of the funnel traps are as wide as the walkway, so that any animal that moves along it is directed into the trap. A guide line of string or rope is attached to both sides of each funnel trap; when the walkway is raised to its desired height these can be pulled tight and tied to trees or rocks to prevent it from inverting during winds and rains. iii) Ideally, walkways should be used at three different levels on the same set of ropes and pulleys: 3m, 10m, and 15m. The pulley system allows regular checking of the traps. This is accomplished by lowering both ends of the walkway at the same time. iv) The funnel traps should be checked regularly, ideally early morning, midday, and late afternoon. Depending on the trap success and the size of animals captured, it may be necessary to increase the depth of the funnel traps, especially if large snakes are encountered.

28 Recording Information on captured species should be recorded on the standard form (Form 3.1). Advantages/limitations Aside from climbing into trees and capturing by hand, this would appear to be the only effective method for sampling species which live in the canopy, or even just above ground level in the forest. However, it is labour-intensive and requires much experience to perfect. Furthermore, care should be used when setting and dismantling the trap; always have at least one and preferably two other people present in case of falls and related injuries.

Fig 3.3: Forest canopy walkway trap

3.3.4 Snake trapping Fritts (1988) developed a simple trap for snakes made of mosquito mesh wire. The trap is baited with bird droppings or feathers, and might be effective for arboreal species that feed on birds. Some snakes detect prey mainly by olfaction, whereas others respond to visual stimuli. For this reason, it would be necessary to experiment with different baits. This technique is aimed at assessing populations of a particular species rather than for general survey work (Fig. 3.4). However, it has proved useful even without bait and it can be used as a simple funnel trap for both lizards and snakes, and seems to work best when placed along a natural barrier, such as a log, large rock, and other obstacles. It might also be possible to use it with a drift fence, in an area with hard or rocky ground which would be unsuitable for pitfall traps.

29 Fig 3.4: Construction of a snake trap

A B

Steps in construction of a snake trap

A) make funnel from screening and cut 2.5cm – 5cm opening in apex B) make cylinder by rolling over the screening and stapling ends; C) insert funnel into cylinder and attach two cylinders together with plastic screening

Site selection Unless one is setting the traps in a particular grid or other arrangement, traps should be set in spots which look likely to have sufficient cover to attract the snake species which are to be trapped. For example, a trap set out in a completely or relatively open area would probably be less successful than one set along a natural barrier or hiding site, such as a fallen tree, a rock or a dead log. Small mammal tracks and paths may also be used by snakes when hunting prey, and these might also make good setting sites. Procedure and Recording Information on captured species should be recorded on the standard form (Form 3.1). Advantages/limitations This technique is labour-intensive and extremely time consuming.

30 3.3.5 Capture, mark, recapture When animals are captured (using one of the methods described above), marked, and released, the population can be sampled again using the same methods of original capture to estimate population size. The details of this method are discussed in Heyer et al. (1994), and summarised in the next chapter (section 4.3.5). Individual marking of amphibians and reptiles has been done traditionally by digital clipping: the number and position of digits clipped provides each animal with a unique number with which it can be identified if recaptured. It is important to be especially careful in the use of methods involving mutilation, and these should not damage the animal such that it is rendered incapacitated. Other workers have used tags and even simply tied thread or elastic bands around the waist of an amphibian to mark it (Muze, 1976), while for snakes, particular ventral scales may be notched or clipped (Ferner, 1979). The particular method of marking and numbering is carefully recorded in a notebook, and the animal released (Ferner, 1979; Waichman, 1992). Marking animals raises ethical issues, especially because much pain and suffering may be inflicted upon individuals if great care is not taken, and if appropriate methods are not used. Marking methods should therefore be discussed with experienced surveyors before carrying out fieldwork.

3.3.6 Forest litter plots In this method, a measured area of 2m x 2m (or any convenient size) is cleared of every bit of leaf-litter and the amphibians within the area identified and counted. A portable ‘fence’ of plastic or metal may be used to enclose the area for ease in sampling. Scott (1982) used this method to sample leaf-litter anurans in Cameroon. Equipment

portable plastic or metal fence to help enclose the area to be searched (made of corrugated sheeting or other material such as plastic sheeting used in drift fences (see above), c. 15cm high by 1m long, or varying lengths) short-handled rake and/or hoe cloth/plastic bags in which to hold sampled animals spring balances (50g, 100g, 500g, etc.) tape measure (30m) preservation material for voucher specimens

31 Site selection Select sites to be sampled; these will depend on your reasons for sampling. For a general survey, try several different habitat types within the forest, such as dry, hilly; moist, valley; disturbed versus undisturbed, and so on. It is important to adequately describe, using standard methods, the habitat and microhabitat of the area searched. If you are attempting to compare different sites or habitats, or to assess altitudinal differences or differences between disturbed and undisturbed areas, then you may wish to randomise your sample areas within a particular habitat type. Procedure i) Measure out the area to be searched and enclose with a suitable ‘fence’. Carefully search through the leaf-litter, using a small hoe or short-handled rake to move the leaf-litter away from a patch of ground in case snakes, scorpions, etc. are also present. Collect animals by hand and place in cloth bags. ii) It is best to sample as many sites as possible; it is likely at least 20 sites with animals present will be needed to meet the requirements of statistical tests. Recording Information on captured species should be recorded on the standard form (Form 3.1). Data analysis This method permits calculation of precise density figures. By measuring mass it is also possible to calculate biomass per unit area. Depending on how the sampling was done, it may be possible to compare counts made in different habitats, at different altitudes, etc. Measures of standard error must then be calculated to assess the reliability of population estimates. Advantages/limitations The method samples only small forest leaf-litter anurans; it is labour- intensive and will usually require more than one searcher. In some forests, herptile densities may be so low as to make assessment by this technique difficult. Microhabitat requirements and/or seasonality factors may result in situations arising in which an area sampled may yield no herptiles, while a plot immediately near that one might have high numbers. Thus, it is generally unsuitable for a species with extremely narrow microhabitat requirements.

32 3.3.7 Time-constrained searches In a time-constrained search, the observer attempts to exert a continuous sampling effort over a particular area or transect for a limited period of time. For most reptiles and amphibians, this method is difficult to use because some are extremely cryptic when not calling, and many hide under vegetation. Nevertheless, this technique may prove useful when animals are conspicuous, such as at breeding aggregations. Equipment

stopwatch (or watch which indicates seconds) plastic/cloth bags collecting and preserving material Site selection Unless you are using a randomised approach, then in practice the observer picks what would be regarded as a ‘typical’ situation, site or habitat. If you are sampling an entire study site, then there is a need to identify all of the major habitat types present. These need to be characterised using standard methods of habitat and vegetation description. Procedure i) Determine an area that is going to be surveyed, and then set a block of time (5–25 minutes), during which full concentration can be maintained for searching (in a standardised way). Take rests between search blocks (5 minutes or so). The observer should move slowly along survey transects and other paths, making every effort to look all around, up and down. ii) Heyer et al. (1994) describe a procedure for time-constrained searches of amphibians, and note that variables such as time spent on the survey, and using each collecting technique, number and experience of fieldworkers, topography and size (area) of site to be surveyed, local weather and climate, season, date and time of day all need to be considered and controlled for. iii) For amphibians, the most efficient time to survey is usually at night. It is useful, however, both in the interests of biology (it is possible to collect eggs, larvae and adults in the daytime) and safety (it is easier and safer to move over often difficult topography at night if you have seen it during the daylight hours) to make a preliminary survey of the area to be sampled during the day. Recording i) The observer carefully records the time spent searching a particular site; if more than one person is searching, it is important to record the number of searchers.

33 ii) General information on the site should also be recorded, such as heavy herb cover, any fallen logs present, etc. to give a picture of the situation and to allow for possible later comparisons. A detailed habitat description of the site is especially important. iii) The number and species of animals encountered and/or captured is recorded. Information on captured species should be recorded on the standard form (Form 3.1). iv) It is also important to note (in a notebook or on Form 3.2) the weather conditions, phase of moon (if field work is done at night), habitat, number of fieldworkers, time spent searching, etc. v) For both reptiles and amphibians, it is often necessary to search by turning over fallen logs, searching under bark, etc. The time spent using these different methods by each worker is carefully recorded. Data analysis The time spent searching is multiplied by the number of searchers to give the ‘total hours spent searching’; this, in turn, is related to the sightings per hour. Time species counts may be used either to detect number of species or the number of animals collected. Advantages/limitations This technique is probably best suited to sampling animals which are fairly visible from a distance, and therefore not applicable for many forest situations except when amphibians have congregated for breeding. It may be useful for reptiles that are ‘sit and wait’ predators, such as geckoes, which often occupy relatively conspicuous sites at night; it may also be applicable for night- time counts of chameleons.

3.3.8 Transect counts Such counts might be used for conspicuous species, such as skinks, which scuttle away from a path as an observer walks it. The transect methods are discussed in Chapters 5 and 6. Jenkins et al. (1999) describe the use of a modified line transect count method for surveying chameleons at night in Madagascar.

3.3.9 Territory mapping This technique can be used for lacertid lizards and for agamas, which are markedly territorial. The locations of individual territorial males are determined and then plotted on a map of the study site. However, it is very time- consuming and labour-intensive, and probably would be used only if you were concerned with a particular species or population. 34 3.3.10 Sound recording surveys Parker (1991) has argued convincingly for the use of tape recordings in avifaunal surveys, and his arguments also hold for surveys of amphibians. Tape recordings have been made of many frog vocalisations (Schiotz, 1999; Passmore & Carruthers, 1995; Rodel, 2000); Heyer et al. (1994) suggest protocols for recording of amphibian calls. This technique is not applicable to reptiles, because none of the African species vocalise with sufficient volume and regularity to be useful for such an approach. For additional information on using sound recording surveys see sections 6.3.2.c and 7.3.9. Equipment

high-quality portable tape recorder or mini-disc player microphone and batteries blank tapes or mini-discs recorded tapes/CDs of frog sounds plastic bag or similar for protecting equipment against moisture Site selection Tape recording may be conducted in the form of a general survey, or sampling might be randomised to meet a particular experimental set up. Remember that some anurans may be photophobic and thus may not call as much or with the same intensity on a night with a bright, large moon as on a dark, moonless night. Procedure and Recording i) Two approaches are possible. One is to record individuals as heard, and try to approach and capture these animals for identification; preferably a specimen with its field number noted on the recording of its call. A second approach is to place the tape recorder at a particular site and record 5–10 minutes of the general calling sounds of amphibians. When recording species which are sensitive to the approach of an observer/recorder, a useful approach is to have a long microphone cable; the microphone is left near the animal which has been vocalising, and the observer/recorder simply retreats to a certain distance, and records using the long microphone cable. ii) Give full habitat data, time of day, exact locality, name of surveyor, air and water temperature, and other relevant details. Effectively, the tape is a specimen, so it too must have as much precise data on it as possible. Long periods of natural anuran sounds are valuable and lacking in most sound libraries.

35 iii) A sample recording data sheet based on that of the Macauley Library of Natural Sounds at Cornell University is provided (Form 3.3). Data analysis i) By collecting detailed data on individual as well as community vocalisations, a record is built up which will enable most species to be identified by voice alone. Well-documented recordings also serve as valid records of a species. ii) Copies of these recordings should be deposited in a professionally- maintained sound collection where they will be available for researchers. Such institutions include national museums, the British Library of Wildlife Sounds, or the Macauley Library of Natural Sounds at Cornell University, Ithaca, New York, USA. Advantages/limitations Because sound is so important in anuran biology, this technique adds considerably to our knowledge of the species concerned. The human ear is very fallible; the tape recorder, however, records all sounds (within the limitations of its microphone and the tape response), thus creating a permanent record that can be assessed by others at a later date. It also permits identification of species that were not heard or seen by the recorder or other collectors, and detailed sonographic study is possible in the laboratory.

3.4 Specimen handling Given our rudimentary understanding about forest amphibians and reptiles, specimens are needed to allow accurate identification by experts, and for future reference in the event of taxonomic revisions. Specimens that are collected, therefore, need to be preserved in an appropriate fashion (Knudsen, 1966; Broadley, 1973; and Heyer et al., 1994, provide detailed instructions), and lodged in a suitable institution, museum or collection (see section 4.4). The collecting of voucher specimens, or larger series of specimens where necessary, also permits other data, such as reproductive condition, to be assessed, and one can also obtain information important to population biology (such as number of eggs per female). Careful labelling of specimens, with cross-reference to field notes and photographs, is essential. In addition to the whole specimen, many recommend that tissues of animals routinely be taken and fixed separately so that DNA may be analysed later. Similarly, it is extremely useful to collect tissues fixed in such a way that these may be used later for analysis of pesticides, metals, and other pollutants, as has been done to assess insecticide spraying in Zimbabwe (Lambert, 1993).

36 3.5 Health and safety Both amphibians and reptiles may harbour ectoparasites and endoparasites, and also may be infected by fungi and bacteria. While usually these have been of little concern to those working with them, recently in some countries there has been concern expressed that freshwater chelonians sold as pets may carry the Salmonella bacterium and cause disease. It is therefore advisable to take normal precautions when handling live animals as well as dead specimens, such as using gloves and/or carefully washing hands with soap and disinfectant. All amphibians have a glandular skin and some of the secretions are toxic and if ingested, potentially fatal. Most toads (family Bufonidae), for example, have specialised glandular areas on the skin, the secretions of which contain powerful toxins. Other anurans, such as members of the genus Phrynomantis (formerly Phrynomerus) also secrete extremely irritating substances from their skin glands. Although in both the toads and other frogs these substances are not usually a problem for those handling them, if the worker has a cut or abraded area of the skin, irritation will result. Similarly, if the secretion is transferred to the eye, nose or mouth, severe pain and irritation may result (Howell, 1978). Simply rinsing with water or other diluting liquids is the best first aid. As ably described by Cansdale (1962) in his book on West African snakes, and contrary to popular belief, very few people get bitten by snakes. Those that do seldom suffer severe consequences – it is more likely that some- one will be killed in a bus, or on a bicycle, than by a snake bite. The venom of only a relatively few species is potentially fatal to humans. Despite the low risks, however, it is still important to take precautions to avoid being bitten by snakes, since surveying them does greatly increase contact. Sensible, strong field boots (canvas, leather or rubber) are important, and can be supplemented with thick socks and denim trousers for the areas between ankle and knee (where most snakebites occur). Safety gloves (e.g. gardening gloves) as well as a snake stick are important when catching and handling. In the case of spitting cobras, which spray their venom at the eyes, safety gog- gles should be worn. All these precautions will reduce the chances of being bitten by snakes. In the unlikely event of a snake strike, every effort should be made, without risking further injury, to capture the snake (dead or alive) to show to first- aiders and medical staff. Snakes should be handled with extreme caution and readers are strongly advised to learn to recognise the local venomous forms and to avoid handling them. Details of venomous snakes and snakebite treatment can be found in Spawls & Branch (1995).

37 3.6 Conclusions Our knowledge of amphibian and reptile biology and population cycles in African forests is so rudimentary that substantially more survey work is needed before forest management actions can be taken to reduce the risks of herptile losses. General surveys and pitfall traps are probably the first survey methods to use at any site, supplemented with other methods outlined above where time, resources, and interest allows. Until more specialists are trained, and more funding is available for detailed studies on populations, biologists will continue to carry out general surveys that indicate presence of amphibians and reptiles, rather than the much-needed detailed population surveys and studies. Species lists are useful, especially those which are annotated and provide information on the conservation status of the fauna of an area. It is important for forest managers to recognise when they have herpetofaunal assemblages of high diversity and/or endemic, rare or endangered species of amphibians and reptiles in their area. They can then encourage individuals as well as organisa- tions with specialised training and experience to address the issues of amphibian and reptile populations in more detail than is currently the case.

38 3.7 References

Barbault, R. (1975). Les peuplements de lézards des savanes de Lamto (Côte d’Ivoire). Annal. Univ. Abidjan. Ser. B. Ecol. 8: 147–221.

Bennett, D. (1999). Reptiles and Amphibians: Expedition Field Techniques. Royal Geographical Society, London, UK.

Berger, L., Speare, R., Daszak, P., Greene, D.E., Cunningham, A.A., Goggin, C.R., Slocombe, R., Raga, M.A., Hyatt, A.D., MacDonald, K.R., Hinas, H.B., Lip, K.R., Marantelli, G. & Parkes, H. (1998). Chytriomycosis causes amphibian mortalities associated with population declines in the rainforests of Australia and Central America. Proc. Natl Acad. Sci. USA 95: 9031 – 9036.

Bowker, R.G. & Bowker, M.H. (1979). Abundance and distribution of anurans in a Kenyan pond. Copeia 1979: 278–285.

Branch, B. (1998). Field Guide to the Snakes and Other Reptiles of Southern Africa. 3rd edn. Struik, Cape Town, South Africa.

Broadley, D.G. (1973). Reptiles and Amphibians: Instructions for the Collection and Preservation. Trustees of the National Museums and Monuments of Rhodesia. 5pp.

Broadley, D.G. (1994). A review of Lygosoma Hardwicke and Gray 1827 (Reptilia: Scincidae) on the East African coast, with the description of a new species. Trop. Zool. 7: 217–222.

Broadley, D.G. (1995a). A new species of Scolecoseps (Reptilia: Scincidae) from southeastern Tanzania. Amphibia-Reptilia 16: 241–244.

Broadley, D.G. (1995b). A new species of Prosymna Gray (Serpentes: Colubridae) from Coastal Forest in northeastern Tanzania. Arnoldia Zimb. 10(4): 29–32.

Broadley, D.G. & Blake, D.K. (1979). A field study of Rhampholeon marshall marshalli on Vumba Mountain, Rhodesia (Sauria: Chamaeleonidae). Arnoldia Rhod. 8(34): 1–6.

Broadley, D.G. & Howell, K.M. (1991). A checklist of the reptiles of Tanzania, with synoptic keys. Syntarsus 1: 1–70.

Broadley, D.G. & Wallach, V. (1996). Remarkable new worm snake (Serpentes: Leptotyphlopidae) from the East African Coast. Copeia 1996: 162–166.

Bury, R.B. & Corn, P.S. (1987). Evaluation of pitfall trapping in north-western forest: trap arrays with drift fences. J. Wildl.mgmt 5: 112–119.

Cansdale, G.S. (1962). West African Snakes. Longmans, London.

Drewes, R.C. & Vindum, J. (1994). Amphibians of the Impenetrable Forest, south-west Uganda. J. Afr. Zool. 108: 55–70.

Ferner, J.W. (1979). A Review of Marking Techniques for Amphibians and Reptiles. Society for the Study of Amphibians and Reptiles Herpetological Circular No.9.

Fischer, E. & Hinkel, H. (1992). La Nature et l’Environment du Rwanda. Rheinmain Druck, Mainz, Germany.

Fritts, T.H. (1988). Instructions for making screen-wire snake traps. In: The Brown Tree Snake, Boiga irregularis, a Threat to Pacific Islands, pp. 28–30. US Fish Wildl. Serv., Biol Rep . 88 (31), 36 pp.

39 Frost, D.R. (Ed.) (1985). Amphibian Species of the World. Allen Press & Association of Systematics Collections, Lawrence, Kansas, USA.

Glaw, F. & Vences, M. (1994). A Field Guide to the Amphibians and Reptiles of Madagascar. 2nd edn. Moos Druck, Leverkusen, Germany.

Heyer, R.W., Donneley, M.A., McDiarmid, R.W., Hayek, L-A.C. & Foster, M.S. (1994). Measuring and Monitoring Biological Diversity. Standard methods for Amphibians. Smithsonian Institution Press, Washington, USA.

Howell, K.M. (1978). Ocular envenomation by a toad in the Bufo regularis species group; effects and first aid. EANHS Bull. July/Aug: 82–84.

Howell, K.M. (1993). Herpetofauna of the eastern African forests. Chapter 9. In: Biogeography and Ecology of the Rain Forests of Eastern Africa, pp. 173–210. (Eds. J.C. Lovett & S.K. Wasser). Cambridge University Press, Cambridge, UK.

Jenkins, R.K.B., Brady, L.D., Huston, K., Kauffmann, J.L.D., Rabearivony, J., Raveloson, G. & Rowcliffe, J.M. (1999). The population status of chameleons in Ranomafana National Park, Madagascar, and recommendations for future monitoring. Oryx 33(1): 38–46.

Kiesecker, M., Blaustein, A.R., Belden, L.K. (2001). Complex causes of amphibian population declines. Nature 410: 681–683.

Knudsen, J.W. (1966). Biological Techniques. Harper & Row, New York, USA.

Kreulen, D. (1979). Factors affecting reptile biomass in African grasslands. Am. Nat. 114: 157–165.

Lambert, M.R. (1993). Effects of DDT ground-spraying against tsetse flies on lizards in NW Zimbabwe. Environm. Poll. 82: 231–237.

Lambiris, A.J.L. (1989). The Frogs of Zimbabwe. Monografie X, Museo Regionale di Scienze Naturali, Torino.

Larsen, D.P. (1993). The Reptiles of Korup National Park, Cameroon. Herp. Nat. Hist. 1(2): 27–90.

MacKay, A. & MacKay, J. (1985). Poisonous Snakes of Eastern Africa and the Treatment of their Bites. Published by the authors, Nairobi, Kenya.

Msuya, C.A. (1993). Feeding habits of crowned eagles Stephaonaetus coronatus in Kiwengoma Forest Reserve, Matumbi Hills, Tanzania. Proc. VIII Pan-Afr. Orn. Congr. 118–120.

Muze, E.S. (1976). Studies on the sex ratio and polymorphism in Hyperolius puncticulatus (Rhacophoridae) at Amani, Tanzania. MSc thesis, University of Dar es Salaam, Tanzania.

O’Shea, M. (1992). Reptiles and Amphibians: Expedition Field Techniques. Royal Geographical Society, London, UK.

Parker, T.A. III. (1991). On the use of tape recorders in avifaunal surveys. Auk 108: 443–444.

Passmore, N. & Carruthers, V. (1995). South African Frogs: A Complete Guide. Revised edition. Southern Book Publishers and Witswatersrand University Press, Johannesburg, South Africa.

Pasteur, G. (1995). Biodiversité et reptiles: diagnoses de sept nouvelles espèces fossils et actuelles du genre de lizards Lygodactylus (Sauria, Gekkonidae). Dumerilia 2: 1–21.

Pitman, C.R.S. (1974). A Guide to the Snakes of Uganda. Revised edn. Wheldon and Wesley. 290pp.

40 Rodel, M.O. (2000). Herpetofauna of West Africa, Vol.1: Amphibians of the West African Savanna. 332pp, including CD.

Schiotz, A. (1999). Tree Frogs of Africa. Edition Chimaira, Frankfurt am Main, Germany.

Scott, N.J. Jr. (1982). The herpetofauna of forest litter plots from Cameroon, Africa. In: Herpetological Communities. (Ed. N.J. Scott Jr.). U.S. Dept. Int. Fish Wild Ser. Wild. Res. Rept. No. 13: 145–150.

Spawls, S. (1978). A checklist of the snakes of Kenya. J. E. Afr. Nat. Hist. Soc and Nat.Mus. 3 (67): 1–18.

Spawls, S. & Branch, B. (1995). The Dangerous Snakes of Africa. Blandford, Cassell Group, London.

Spawls, S., Howell, K.M., Drewes, R.C. & Ashe, J. (2002). A Field Guide to the Reptiles of East Africa. Academic Press, London.

Stewart, M. (1967). Amphibians of Malawi. State University of New York Press, Albany, USA.

Stewart, M. & Pough, F.H. (1983). Population density of tropical forest frogs: relation to retreat sites. Science 221: 570–572.

Vogt, R.C. (1987). Techniques. You can set drift fences in the canopy! SSAR Herp. Rev. 18: 13–14 .

Waichman, A.V. (1992). An alphanumeric code for toe clipping amphibians and reptiles. Herpetol. Rev. 23: 1992.

Western, D. (1974). The distribution, density and biomass density of lizards in a semi-arid environment of northern Kenya. E. Afr. Wildl. J. 12: 49–62.

Wyman, R.L. (1990). What’s happening to the amphibians? Conserv. Biol. 4: 350–352.

Yalden, D.W. (1977). The Identification of Remains in Owl Pellets. Occasional Publication of the Mammal Society. Reading, UK. 8pp.

41 Form 3.1: Herptile Catch Records (instructions see p52)

Surveyor: Field sheet ref: Date: (total observers): (dd/mm/yy)

Address:

Survey site: Altitude: Aspect:

Latitude: Longitude: UTM (if available):

Vegetation: Human disturbance:

Soil type: Leaf–litter/ground cover:

Season: Weather: Lunar phase: Temperature:

Other:

Trap line Microhabitat Water Topography Species and Other & no. association specimen sheet ref.

42 Form 3.2: Specimen Records: herptiles

Specimen sheet ref: Field sheet ref: Collector: Date: Time: (dd/mm/yy) Address: Collecting site: Altitude: Latitude: Longitude: Slope: Additional notes: Species: Field no.: Sex (if known): Age: Pregnant: Eggs: Breeding condition: Colour/markings: Wounds: Ectoparasites: Endoparasites: Measurements: HB TL TV Snout-vent Other W mm mm mm mm mm g Material Preserved: Skin Skull Skeleton Stomach Faeces Blood Liver Kidneys Stomach contents: Component: Percentage:

Remarks/Other

43 Form 3.3: Sound Recording Form

Species, Sound or Subject: Recordist(s) and Address: Date: Time: Weather: Place: Latitude: Longitude: Species/id No. of individuals Sex/Age Breeding status Sound Category Response to playback Notes:

For calls www.birds.cornell.edu www.bl.uk/collections/sound-archive/nsa.html Contacts: Curator Curator Library of Natural sounds NSA Wildlife Section Corne Laboratory of Ornithology The British Library 159 Sapsucker Wood road National Sound Archive Ithaca, New York 14850, USA 96 Euston Road London NW1 2DB, UK

44 4. Small mammals: bats, rodents and insectivores golden-rumped elephant shrew Glyn Davies and Kim Howell (Rhyncocyon chrysopygus) (Rhyncocyon

4.1 Biology Small mammals are a disparate collection of flying and non-flying species that have been grouped together because of their relatively small size, despite obvious anatomical and ecological differences. In this chapter, we consider three groups: rodents, bats, and insectivores (including elephant- shrews). All groups are elusive and difficult to survey because, in order to avoid predators, they have evolved dull colouration, secretive behaviour and, in many cases, nocturnal habits. These characteristics, along with their small size, make field identification difficult – a problem exacerbated by the very high diversity of African small mammal species (for example, there are about 190 bats and some 380 rodents in Africa). These problems are no less acute for insectivorous small mammals – there are about 165 African mainland species (many of them in the genus Crocidura alone). Mammal lists of the world (Corbet & Hill, 1991; Wilson & Reeder, 1993; Nowak, 1999), and regional checklists, may help to give a rough idea of the species present in an area, but only surveys and collecting carried out over different seasons will permit a more accurate assessment of the species present in a given forest.

45 Bats (Order Chiroptera) Bats are divided into two sub-orders: the fruit bats (Megachiroptera), sometimes termed ‘megabats’, which use their large eyes and relatively long noses to locate fruit and nectar/pollen food sources, and do not use high- frequency echolocation for navigation; and the insect-eating bats (Microchiroptera), or microbats, which make use of echolocation and hearing to find insects and small fruit foods, as well as nectar and pollen. Species from both groups roost in the daytime, sometimes in large congregations, and are most active soon after sunset, unless it is raining. Some are active during the day as well as at night, and during the night different species may show activity peaks at different times.

Rodents (Order Rodentia) The rodents are usually divided into two sub-orders: the Sciurognathi, which includes the squirrels, rats, and mice; and the Hystricognathi, represented in Africa by the porcupines, mole-rats, cane-rats and dassie-rat. Some members of this diverse order are large (e.g. cane rat: 7+ kg; crested porcu- pine: 15+ kg) and all have powerful front teeth for gnawing. The mice and rats mostly live in holes and forage for fruits, seeds, arthropods, etc. on the forest floor, or under/along fallen logs. Most rats and mice are terrestrial, and most squirrels are arboreal – however, there are some rare exceptions to this rule: dormice and climbing-mice climb quite extensively in under-storey trees, while some large squirrel species are ground-dwelling.

Insectivores (Orders Insectivora and Macroscelidea) The insectivores belong to two mammalian orders: Insectivora, which includes such diverse groups as the shrews, otter-shrews and hedgehogs, and the Macroscelidea – the distinctive elephant-shrews (although there have been recent taxonomic revisions). The shrews are distinguished from rodents by their protruding snout, usually tiny eyes and elongated lower incisors. Most forage in leaf-litter for live arthropods and other invertebrates. An important characteristic of small mammals is that many have the reproductive ability to undergo large population increases during favourable periods, and suffer substantial losses at other times. For species that show these boom-and-bust cycles, estimates of population sizes are difficult to extrapolate from year to year. There are also considerable differences between different seasons, so surveys need to take careful account of this potential bias; comparing dry season and wet season results will give a poor understanding of population differences between forests. There may also be differences in behaviour depending on weather conditions (e.g. mice sheltering during rainstorms) and lunar phases (e.g. dark nights affecting fruit bat activity), so these factors should be recorded during surveys.

46 4.2 Management issues Small mammals, especially the more abundant species, are important components of forest ecosystems. All small species are preyed upon, and therefore support populations of many groups of carnivorous and omnivorous mammals, birds and reptiles. Rodents that are not eaten are responsible for the destruction of many plants’ seeds, but they can also play a key role in seed dispersal; for example, when squirrels cache (hide) seeds in stores that they fail to relocate before the seeds have germinated. Bats also play a very important ecological role in the forest, through pollination of flowers and dispersal of seeds from fleshy fruits. Small mammals may also be good indicator species of habitat change, and some are pioneer species. Recent studies in southern Africa of small mammals colonising disturbed coastal sand dunes have indicated the usefulness of rodents and other small mammals, as indicators (for example in regen- rating coastal dune forests: Ferreira & van Aarde, 1997, and in central African forests along logging roads: Malcolm & Ray, 2000). In their relationships with humans, there are a number of rodent species that do considerable damage to crops and stored grains, and fruit bats that do substantial damage to soft fruits. As a result they are killed, often by trapping rather than shooting, to reduce crop losses. An important by-product of these pest control operations is bush-meat – including polythene-wrapped fruit bats for sale in supermarkets (e.g. Mickleburgh et al., 1992) and road-side carcasses of cane rats (grass-cutters). Many other species have a neutral impact on agriculture, and some are beneficial in pollinating fruit and vegetable crops. Recent conservation reviews indicate that rodents (Lidicker, 1989), bats (Mickleburgh et al., 1992; Hutson et al., 2001) and insectivores (Nicoll & Rathbun, 1990) are all declining in Africa. In the case of fruit bats, disturbance or destruction of roosting sites, over-exploitation of useful species and conflicts with fruit-growers have been cited as the main causes of declines (Mickleburgh et al., 1992). Schlitter (1989) listed some 67 species of African rodents (representative of eight families) as being of special conservation concern, while Nicoll & Rathbun (1990) listed 58 species of insectivores, including six of the 15 elephant-shrew species, which need special conservation attention. For all three groups, the most consistent cause of declines is modification, fragmentation and loss of habitats, especially forest environments (see the 2000 IUCN Red List of Threatened Species (Hilton-Taylor, 2000) www.redlist.org, for recent information). Another consistent comment in conservation reviews is that very little is known about these groups, either in terms of where they are found, or their ecology and population biology (most work having been done in temperate

47 zones). In this chapter, therefore, a summary of basic survey principles is given, along with an overview of the wide range of methods that have been developed, many of which need to be adapted for particular species, or particular forests.

4.3 Methods

General The methods used vary according to the particular group; obviously, bats must be surveyed using techniques which differ from those used for rodents. Yet the general approach is similar and many of the factors to consider are similar or the same. A first principle is to obtain the maximum amount of information about an individual detected or captured. This may be a relatively straightforward procedure when dealing with a specimen, but obtaining information on flying bats or rodents glimpsed only briefly is challenging to say the least.

Identification As mentioned earlier, many small mammal species can be very difficult to identify, and sometimes it is only possible to identify specimens to the level of genus. There are few field guides available to aid in the identification of African small mammals, although Kingdon (1997) is an exception; however, even this guide is limited in its discussion and representation of the smaller mammal species. Regional works, such as Rosevear (1965, 1969), Happold (1987) and Kingdon (1974), may be useful for identification, but they are all hefty tomes and cannot be carried into the field. In most cases, identification will need to rely on the use of identification keys, many of which are only available for individual families or genera and are not widely available. Furthermore, the identification of small mammals, especially shrews and rodents, to species level, usually requires a detailed examination of the skull and teeth. This effectively means that some animals must be sacrificed to serve as voucher specimens, and sent to museums outside of the region for examination and study by specialists. Thus, the preservation of voucher specimens (at least 10 individuals of each sex per species) is a necessary and vital part of any small mammal study. However, a new effort aimed at students of mammals in Tanzania is just bearing fruit. W. T. Stanley of the Field Museum of Natural History, Chicago, Illinois, U.S.A., with financial support from the MacArthur Foundation, has created a key to African mammals using either skulls or skins. While still in a preliminary stage, this can be accessed at: www.devdirection.com/tanzania/

48 4.3.1 General surveys General surveys can be used to start plotting the distribution of species, in different habitats and at different altitudes, and to select sites for more detailed investigation. All the information gathered during general surveys can be mapped to show species’ distributions (section 6.3.1). For bats, dusk-time walks near forest streams, potential roosting sites (e.g. caves) and fruiting/flowering trees provide an indication of bat numbers. Checking caves, hollow trees and fallen logs may also be rewarding, even in the daytime. For rodents and insectivores, searching under fallen logs for runs, where tiny feet have left a distinct path, as well as for signs of discarded food remains or faecal pellets, may help identify sites that could be sampled later with traps. In the case of elephant-shrews, spherical nests of grassy material in the leaf-litter, as well as runs, are indicative of their presence. Other indirect signs of small mammal presence include their teeth, skull and other skeletal remains in owl pellets (regurgitated by owls underneath their resting sites) and carnivore scats and skulls may also be found by searching rubbish tips near villages (Barnett, 1992). Hair analysis is another useful means of identifying small mammals indirectly; indeed, sampling of carnivore scats in order to identify the remains of small mammal species is a proven technique. In Central African Republic, Ray & Hutterer (1996) found that there were 16 species of sympatric shrews in one 35km2 study area just by analysing carnivore scats collected over a two-year period. They attributed this highly unusual diversity less to some incredibly feature of the site, but rather because carnivores, as a moving trap, represented a more efficient capture technique than conventional pitfalls. With detailed information on hair size, colour and structure (e.g. scale patterns), one can even design fur traps with sticky tape or tiny snags placed on tubes through which animals pass, to enable the sampling of hairs for subsequent identification to species. For bats, high-frequency bat detectors can be used to investigate the presence of insectivorous bats, and indicate where future trapping might focus. However, a reference collection of bat calls is needed to relate the calls to a particular species (see Wilson et al., 1996).

4.3.2 Bat roost surveys When a bat roost has been located, there are two approaches that can be taken to estimate the numbers present, namely emergence counts and roost counts. The emergence points from caves, for example, need to be located, and observers stationed at each in the late afternoon so that they can count how many bats emerge at dusk. Each observer should have a watch and

49 tally counter. All bats exiting and entering should be counted in convenient time units (e.g. five-minute intervals), and those that return to the cave (presumably to come out again) are deducted from the total. Care needs to be taken to distinguish between different species leaving the same roost. This may be feasible when only a few easily-distinguished species are present at a roost, but might be extremely difficult or impossible when closely related species, similar in size, shape and behaviour are present. If there are several hundred bats emerging then counts become less accurate, and more observers (coupled with photographic techniques) can be used to improve the accuracy of the counts (Barlow, 1999). Counting bats that are roosting in enclosed areas (e.g. in buildings) can be done using low lights and binoculars to make total counts, and tree-roosting fruit-bats can be counted directly in the daylight (Kunz, 1988). As the colony size gets larger, counting becomes more difficult, and sub-sampling of different sections of the roost (with different concentrations of bats) may be needed. For example, numbers of fruit bats in many trees may be estimated by counting a sample of trees and multiplying up by the number of occupied trees once a mean (plus standard error) number of bats/tree has been established. 4.3.3 Live-trapping: rodents and insectivores There are numerous reviews of this method of census (Delany, 1986; Barnett, 1992; Wilson et al., 1996) and this section summarises the main methods as they relate to forest survey work in Africa.

Equipment string and flagging tape specimen bags and polythene bags sedation materials gloves equipment for marking animals spring balances traps (see below) and bait

There is a wide range of live-traps to select from (Fig. 4.1): At the smaller end of the range, Longworth and Sherman live-traps are mostly made of aluminium, and measure approximately 230mm x 95mm x 80mm in size when set up. These traps are very lightweight, and Shermans have the added advantage that they can fold flat for storage and easy carrying in the field.

50 Fig 4.1: Live Traps

Havahart Sherman

Longworth

Havahart live-traps (available from international suppliers) are effective for sampling species such as African giant rat, and hyrax (dassie), and are convenient and easy to use. However, they are not collapsible and thus rather bulky. Pit-fall traps are important to catch mammal species that may not be caught in other types of live-traps. This includes species that do not like the baits on offer and/or species that forage widely and do not follow runs (including shrews and other insectivores). Shrews are probably better surveyed using pit-fall trapping (section 3.3.2). In the case of elephant-shrews, animals can be driven out of their nests into encircling nets (2m-wide fishing nets), once the nests have been located (see details on antelope drives – section 5.3.2). The presence of nests along transect surveys has been developed as an indirect index of abundance (Fitzgibbon and Rathburn, 1994), using the same principles as described in the next chapter (section 5.3.3).

Site selection i) Rodents tend to move around the edge of clearings, and beneath fallen logs and rocks. They also follow runs which may be visible, including along low branches and lianas, and their holes are often at the base of trees and rocks. They also use places to gather, store and consume foods, and to shelter. All these are potential trapping sites. ii) Once a trapping system has been developed, then every effort should be made to keep the same site selection procedures, and the number and types of traps consistent between different trapping periods (e.g. over consecutive years), or between sites in the same period. For example: 20% on lianas, 80% on the ground; 50% box traps, 50% break-back traps (see below).

51 Procedure i) Traps are generally placed in clusters, termed ‘trap stations’, spaced regularly (5–10m) along a transect or in a regular grid. Each trap station can have a number of traps, although three per trap station is probably a minimum. ii) The traps have to be baited, and particular attention needs to be paid to standardising the baits. There are a number of baits that have been successfully used: peanut butter works well, and can be mixed with other items (e.g. banana, maize meal, oats, raisins, forest fruits, chunks of manioc root, dried fish, etc.). In Tanzania, KH uses pieces of fried coconut, mixed with local peanut butter. This fits onto the trap bait-hook well, is attractive to rodents, and seems to survive the threats posed by rain and ants. The selection of bait will have a major impact on the species that will come to the trap, so the same bait needs to be used if trapping is to be standardised between sites. iii) If it is rainy, then some bedding (pieces of old newspaper) can be added to the box traps to reduce the risk of hypothermia. However, care must be taken in the process of preparing traps not to leave human scent which may deter animals from entering – rubbing other smells (e.g. meat fat) onto the trap is one option, but minimising handling and airing traps is always wise. If previous trapping has left urine or blood on traps, it is advisable to wipe/wash this off. Indeed, it is good practice to thoroughly wash traps to remove old bait, rodent urine, etc. prior to storage (see also the traps described in section 4.3.6). iv) The traps can now be placed in suitable trap sites, and need to be secured with strong string or stakes so that they are not moved either by trapped animals or, on rare occasions, by predators trying to get to the trapped animals. Trap entrances/surfaces should be flush with the substrate, so that animals do not have to go uphill to be caught, and they must not be easily flooded (or washed away) if there is sudden rain. v) Mark each trap site with flagging tape, and give each trap, trapping station and trap-line a unique number. vi) The traps must be inspected in the early morning, midday and late afternoon. If it is cold or wet, then more frequent inspections are advised. Ideally standardise the procedure so that all traps are baited and set between, say, 18:00 and 19:00, and are inspected between 06:00 and 07:00 the following morning.

Recording This section applies to all small mammals, including bats, caught in live-traps. i) The following should be recorded on a standard trapping record form (Form 4.1) at the beginning of each transect:

52 full name of surveyor; sheet reference, which could refer to the field notes recording system (e.g. 3 or 12); date (dd/mm/yyyy); and address of institution that has a copy of the field records and specimens collected; collector number. survey site: the name of the region/forest area, and site within the forest, where the survey was conducted (e.g. Kakamega Forest Reserve, Ischeno area); altitude (in metres above sea level); aspect (e.g. is the terrain steep or flat; valley-side, ridge-top or valley-bottom; facing north, south, east or west); latitude and longitude, in degrees, minutes and seconds (if available, using GPS), and the UTM (metric grid) can also be added here (two letters, followed by six numbers). season: wet or dry season. lunar phase: what quarter of the moon is it, and is it getting larger (waxing) or smaller (waning). vegetation: use terminology that is accepted internationally, in particular make use of White’s (1983) phytogeographic regions of Africa. Other national and regional categories can also be used. any indications/signs of human disturbance. weather: a statement about the weather during the trapping period (e.g. clear, clouds, rain, overcast, windy). temperature: typically the minimum night-time temperature. ii) Record all the traps that were set off during the night, making a note of the following: trap lines/trap no.: the trap location can be cross-referenced to a map of the survey site. trap type/bait: type of live-traps/nets or snap traps successful, and what bait was used. microhabitat: this refers to details of the trap location – beneath a log, at the base of a tree, on a low branch, in burnt land, tied to a liana, in leaf litter, etc. water association: whether the trap is situated near water bodies (e.g. 5m from stream edge) and the type of water association (e.g. stream, river, marsh, pond, dry river bed, etc.) topography: e.g. ridge top, halfway up hill, bottom of hill, valley, path, plain, and so on. species captured and corresponding specimen sheet no (see iii). Use the scientific name of the species where identification is certain and record the English name where possible. Follow a standard list when using Latin names. If the trap has been set off but no species has been captured, this should be recorded under ‘Other’. Note whether bait has been

53 removed or partly eaten and if traps have been moved. If a species has been captured, use the Other column to record information such as dominant plants in the immediate vicinity or any other important observations. iii) When an animal is caught, it is important to record as much information about the animal’s condition as possible on standard forms (Form 4.2; occasionally, specimens may be collected for preservation purposes, see section 4.4). iv) Care must be taken not to injure or traumatise the animals during this process, and to minimise risk of disease or infection spreading to surveyors (see section 4.5). Thick gloves are important, and surveyors should learn how to hold captured animals without injuring themselves or the animals. Captives can be sedated by carefully emptying the trap into a large polythene bag containing a small piece of cotton wool soaked in chloroform or ether. The animal should be drowsy, but not unconscious, when it is picked up for inspection.

Data analysis is discussed in the subsequent section.

Fig 4.2: Handling small rodents

4.3.4 Live-trapping: bats Although the same equipment will be required as for the live-trapping of rodents and insectivores, the trapping system for flying mammals obviously is entirely different. There are three basic bat-catching techniques:

Hand nets Hand nets (made with mosquito mesh if necessary), with long handles, a deep net, and firm rim can be used to catch bats. This can be done by holding it over a small hole through which they are emerging, or placing them over roosting bats on ceilings and cave walls (Barlow, 1999). Hand nets can also be used to catch flying bats by bringing the net quickly around the bat from behind (Wilson et al., 1996), although this method should be used infrequently because it runs the risk of damaging the bat’s wings.

54 Mist nets Mist nets for catching Microchiroptera should have a mesh size of about 36mm but stronger nets, with larger mesh sizes, are needed for Megachiroptera. Mist nets come in varying lengths of 6–18m. Monofilament nets should not be used for catching bats (Barlow, 1999). Discussions about mist netting are given in the chapter on bird surveys (section 7.3.8), including details of all the equipment needed. Mist nets tend to be most effective in catching medium and large bats, especially plant-visiting species travelling in the understorey of the forest. Smaller insectivorous species tend to evade mist nets, and quickly chew their way out when caught. The advantage of mist nests is that the nets are easily transported in bags to the field site, as long as large numbers of lightweight bamboo or aluminium poles are not needed. In addition, the surface area for catching can be enlarged by stringing a number of nets together. The main disadvantage of mist nets is that they are hard to move once set up. It is also a slow and tricky business removing bats from the nets, during which time both the bats and the nets can get damaged.

Harp traps Harp traps have been developed in the last 30 years, and operate on the principle that bats have difficulty in seeing – either visually or by means of echolocation – thin strands (Kunz et al., 1996). They are most effective in catching smaller, insectivorous bats. To make a harp trap, a rectangular frame (approx. 2m x 2m) is constructed, and vertical lines are attached at the top and bottom, 25mm apart (the harp). Monofilament fishing line (about 300g strength) is readily available for this purpose, although steel wires have been used. A second frame, with the same layout of lines, is fitted 70–100mm away from the first frame, with the vertical lines on the second frame corresponding to the gaps between the lines on the first frame (see Fig. 4.3). Bats fly into the trap and get blocked between the two sets of lines, causing them to fall or flutter into a canvas bag at the bottom of the trap. The canvas bag needs polythene flaps on either side of the entrance, leaving an open slit through which bats will fall, and under which the bats can rest in a dry place. Trapping efficiency can be increased by having three or four layers of lines on a single trap (Francis, 1989), and the traps can be made as large as construction materials and access to forest survey sites allows; a trap used to capture flying foxes was 15m high and 17m wide (Wilson et al., 1996). Some traps have legs to stand them up on the forest floor, but others are hung from tree branches using ropes and pulleys.

55 The advantage of harp traps is that they can easily be moved about to other catching sites within the survey area, and bats can be removed from them quickly. The disadvantages are that they generally offer a small surface area for trapping, and can be bulky to carry into the forests in the first place.

Fig. 4.3: A harp trap

Site selection i) For mist nets and harp traps, the entrance to roosts is an obvious place to survey, although the roost exit should not be completely blocked and care should be taken not to catch more bats than can be safely removed from the nets/traps. ii) Other suitable sites include beside, or stretched over, small pools and streams, or any flyways that bats appear to be making frequent use of. The best places are where there is a gap in the vegetation that funnels bats into a narrow area where nets/traps can be placed. Placing traps/nets at right angles to each other, or in a V-shape, may improve catches. iii) Catching bats in the upper strata of the forest, or above the tree canopy, obviously requires nets/traps to be hung from branches, or from aerial walkways that have already been constructed. Time is needed to fire strings into the canopy trees (with bows, cross-bows or sling-shots), and to haul up the nets/traps in such a way that they hang securely and do not get caught up in twigs and branches. Safety and specialist equipment are needed if climbers are clambering up the trees to set the nets/traps. iv) Bats learn to avoid places where traps/nets have been set, so traps/nets need to be moved periodically (every 2–3 days), or as soon as there is an obvious decline in the number of catches.

56 Procedure i) Nets and traps should be set up well before sunset, so that they are ready for the initial surge in bat activity at around dusk. In the case of mist nets this means that any birds caught before dusk need to be removed. ii) Nets and traps should be checked at least every 30 minutes. If too many bats are being caught then traps/nets should be closed so that bats don’t get damaged. They should also be closed if it starts to rain – bats die very quickly if they get cold and wet. iii) Although trapping efforts should be concentrated around dusk and early evening, it is important to keep going throughout the night, or at different periods on consecutive nights until dawn, in order to cover all periods of potential peak activity for different bat species. iv) Once a trapping/netting system has been established, it should be kept consistent between survey sites and periods (e.g. same number and size of nets/traps; same arrangement of traps/nets; same number of hours and periods of the night sampled).

Recording Follow the same procedures listed above for rodents and insectivores (section 4.3.3; Form 4.1) being particularly careful while handling live bats. Fig 4.4: Handling bats

Data analysis This section applies to all small mammals, including rodents and insectivores, caught in live-traps. At a most basic level, a list of the number of species caught can be used as an indication of biological richness for those species that can be trapped. Species lists can be built up over time, and supplemented with records from other trapping/survey procedures. Some rudimentary indices of abundance have been developed to make use of trap records. These indices present data in terms of the trapping success for a given trapping effort, often expressed as catches per trap-night (trap- nights = number of traps multiplied by the number of nights set), or catches per trap-hour, etc. In the case of bats, reasonably accurate population estimates can be obtained if a sample of 350–500 bats is caught (Barlow, 1999). 57 Advantages/limitations i) For all but the largest species, there is little option but to trap small mammals to carry out surveys. The constraints on this approach are the effectiveness of different types of traps to catch the full range of species present, and the ease/difficulty of getting the traps into remote forest areas. ii) At the level of generating a species list, the results reflect the presence of those species that are prepared to enter traps (for that bait). The results are therefore limited to giving information on the species richness of only those species for which the methods are suitable. iii) The same applies for population estimates. However, once an effective trapping system (including trap-siting, bait, etc.) has been developed for a species, then mark-recapture methods can be used to estimate absolute population densities, and make comparisons over time or between sites.

4.3.5 Capture, mark, recapture If a long-term study is planned, then mark-recapture techniques can be used to make better estimates of population size. In this approach, animals are captured (using one of the methods described above), marked, and released, and the population sampled again after some time, using the same trapping methods. The population estimates are based on the equation:

Total population = total first catch x total second catch number of recaptured marked animals

However, a number of key assumptions must be met for this to follow (after Kunz, 1988):

survival rate of marked individuals is representative of the population as a whole; the probability of survival between capture periods is equal for marked and unmarked individuals; the permanent loss of individuals from the population is a result of deaths, and not long-term emigration (or dispersal); marked individuals have an equal probability of being captured as unmarked animals; marks are not lost; the intensity of trapping (number of traps, number of days trapping, etc.) is the same in different surveys.

These requirements are often not met, especially during a general survey of many species in a short period.

58 More detailed ecological research, with reference to other survey books (e.g. Wilson et al., 1996), should be made before attempting this detailed level of analysis. The references give details of the mathematics of the survey approach, as well as different ways to mark animals before they are released after capture (e.g. clipping rodent toenails; fitting bands onto bat forewings; using luminous dyes and permanent markers on fur, etc.). Reference should also be made to computerised data analysis systems, such as CAPTURE (Pollock et al., 1990; obtainable from http//www.mbr.gov/software.html).

4.3.6 Removal or dead-trapping

Fig 4.5: Snap trap

Equipment In addition to the equipment necessary for live-trapping (section 4.3.3), the following should be borne in mind regarding the use of traps for dead-trapping: Important considerations when selecting traps include: whether their different parts are susceptible to rust, or rotting of wooden parts; whether the spring is too strong, and is therefore likely to destroy the specimens (loosening springs by one turn can resolve this problem); whether they have serrated edges which can severely damage specimens; whether they are too small to make a clean kill of medium to large species. Rat-size traps take up space and are heavy. If packed in such a way that they become bent or the triggers are damaged, they will not be effective. It is therefore important to pack them as com- pactly and as securely as possible. When storing traps, metal traps can be painted with red oxide primer to reduce rusting, and wooden traps can be dipped in linseed oil, which reduces their tendency to soak up water in the field (and increases trap life); as far as is known, the strong smell of this oil has no negative effect on capture rate.

59 There is a wealth of locally-made traps which can also be used, and which can improve the range of species being trapped in an area. Although there are also local trappers adept at operating these traps, the variability of trap construction usually precludes using these traps for systematic surveys in different forests. Try to select a type of trap which is available in large numbers and which is likely to be available for purchase in the coming years.

Procedure and recording The principles described above for live-trapping rodents and insectivores (section 4.3.3) all apply to trapping with break-back traps.

Advantages/limitations Break-back traps are much lighter and more compact to transport than live-traps. The standard break-back traps used to kill pest species such as rats (larger size) and mice (smaller size) are widely available, and can generally be purchased in the country where surveys are to be carried out. They generally catch more than live-traps, thereby providing specimens for identification and museum reference collections. However, they are indiscriminate in what they catch (i.e. they are not species-specific); usually, many individuals of one or two common species will predominate. Thus, many individuals are killed for a minimal amount of information that may be useful for management, a problem exacerbated when doing surveys in conservation areas where protection may be a focus of management. The catch is also strongly influenced by the baits that are used, and the ecology of the species in the forest site.

4.4 Specimen handling Specimens collected from break-back traps can be supplemented with those from live-traps. If an animal has been live-trapped, and examined, the animal may be killed by cervical vertebrae dislocation, thoracic compression or any other humane means. Barnett (1992) has offered some guidelines on when animals might be killed: when the animal is injured (physically or mentally) in live-traps and mist nets, including wet/cold animals that are hardly moving; to obtain reference specimens (voucher specimens) that can be used for the later identification of species; where specimens of identified species are needed for reference collections (for example, if it is found in a new region).

60 There is a wealth of information that needs to be collected from dead specimens. Some of this can be done in the field, but further research may be required back in the laboratory (Wilson et al., 1996). Since specimens are required for accurate species identification, it is very important that surveyors take time to visit museum or university specimen collections in order to become familiar with hair/skin colouration, and other diagnostic features for species identification, especially cranial/dental anatomy. Discussions should be held with curatorial staff to get advice on what needs collection, how to measure small mammals accurately, how to prepare specimens, and get copies of their standard specimen record sheets and identification guides.

Equipment dissecting kit: scissors (fine and thick), scalpel (including different size blades), forceps (fine and thick), syringes (with fine and thick needles), surgical mask and gloves ruler/callipers Pesola weighing scales waterproof specimen labels thick thread and sewing needles plastic screw-top storage jars muslin/paper towels alcohol (70% ethanol) and/or formalin

Recording i) Confirm the identity of the species and give the animal a field number. Unless these are supplied by your institution, it will be most convenient if you use a series of numbers that is preceded by your initials (e.g. CAM 305). Enter this on the data sheet (Form 4.2) and, if you are not using pre-numbered label tags, on a tag. A specimen without a label indicating the date, place of collection, and collector, has little value. ii) It is critically important to use label paper that will withstand field and storage conditions. If the specimen will be prepared as a study skin, standard dry label tags are available. If it is to be preserved in fluid (formalin or alcohol) then special water-resistant paper must be used for the labels. Furthermore, it is critical to use either a hard pencil or waterproof ink (use either Indian or Pigma felt-tip pen) on the label. If ink is used, make certain that it will not dis- solve in the fixative. Be certain that it is completely dry before immersing it in liquid. It may be necessary to dip the label in fluid first, and then dry it before immersing the specimen. iii) Examine whether the animal is male or female (if possible), try to assess age (e.g. infant, juvenile, subadult, adult, old) from size and/or tooth

61 wear, and note other features such as: pregnant or lactating females; breeding condition (i.e. is the vagina perforated); colour and markings (such as stripes and spots, including variations); presence of wounds on ears, tail, and elsewhere; dental formula; mammary formula; and parasitic infections (collect specimens if required); etc. iv) Take standard length measurements in millimetres: Head and Body (HB); Total Length (TL); Tail Vertebrae (TV), Ear (E) and Hindfoot (HF) (see Fig. 4.6). The hindfoot measurement usually includes the claw, i.e. HF/cu (cum unguis) but a few workers measure the hindfoot excluding the claw (sine unguis); we include the claw in our measurements. For bats, two additional measurements may be used: Length of Forearm (FA), and Length of Tragus (TR) – the latter is a prominence in front of the exterior opening of the ear. Body mass in grams (W) is measured using a spring balance of the appropriate scale. Make a note of any material that is preserved (e.g. skin, skull, blood, muscle tissue).

Fig. 4.6: Measurements for small mammals

Procedure Proper specimen preparation in the field is necessary to ensure that any mammal that dies, either as part of the process of collection for voucher specimens or incidentally by injury in a trap, being eaten by safari ants, or rotting in hot temperatures, is preserved, labelled and the data used. A specimen is of little value without a good label, so it is critically important to prepare your specimens carefully and label them well.

Wet specimens i) If the specimen is to be prepared as a fluid specimen (i.e. fixed in formalin and later stored in alcohol) then a small numbered tag is usually tied on the left hind foot.

62 ii) After the measurements have been taken and the number tag attached, the body cavity is cut open using scissors or a scalpel; this allows the fixative to enter the body cavity and gut as quickly as possible. If the animal has a very full stomach, it may be necessary to inject formalin or alcohol into muscle masses and into the stomach and/or intestine. Ideally, this is the time to carefully examine the reproductive tracts of females and record the condition of the ovaries and uterus, number of foetuses, and uterine scars. iii) Because the skull is so critical to the identification of many small mammals, some workers at this stage remove the skull from the carcass and store it in 70% ethyl alcohol. The skull should be tagged with a small label bearing the same number assigned to the animal, and kept with other such skulls. Later, when dermestid beetles are used to clean the skull, they will hap- pily devour the skin and muscle off such a specimen. Such is not the case for material preserved in formalin, which the beetles do not seem to like the taste of! iv) Information on what animals have been eating can be obtained from the stomach contents, and other parts of the intestine or the faeces. The stomach contents should be analysed one by one, and proportions of volume attributed to different seasons (e.g. 50% seeds; 20% insects; 30% plant fibres). v) Wrap specimens in muslin or paper towels before putting them into the storage jars to reduce the risk of damage during transport. Make sure that the preserving fluid is topped up and of sufficient strength; it is necessary to have a large volume compared to that of specimens. vi) If there is an opportunity to send tissue samples to laboratories for genetic and cellular analysis, then the laboratory will give clear instructions about how to prepare the samples.

Dry specimens i) Small mammal skin preparation is a standard procedure, and involves removing muscles and other tissues which are likely to rot, and stuffing the body with material (such as dry cotton wool) so that the body retains a normal shape. ii) Skins need to be pinned out and dry on both sides, but they should not be smoked (this will affect both skins and labels). Cover with cheesecloth to keep flies from laying their eggs on the skins. Beware of potential predators, such as insects, dormice, and other rodents, as well as birds of prey such as kites and ravens; the latter have even been known to remove covers of tins, etc. to get at specimens! Once skins have been dried (if they are sufficiently dry, the skin will crinkle and the ears will be stiff) then they can be packed for transport.

63 iii) Drying skins in forest conditions is very difficult as they are liable to insect attack and rot very quickly. If wet specimens cannot be prepared, then construct a drying oven using hurricane lamps and metal tins or buckets. This will ensure that specimens are quickly dried, but care must be taken not to char or smoke the skins. They can then be packed into plastic bags to prevent them absorbing the damp, along with some naphthalene crystals (mothballs) to reduce risks of insect attack. iv) It is usual to preserve some specimens of each species as skeletal; these have most of the large muscles removed, and are then dried. Or, if they are in a very wet situation, as may arise in a rain forest, the skeleton can be labelled and placed in 70% alcohol.

4.5 Health and safety The close proximity of surveyors to small mammals during trapping work means that care has to be taken to avoid the spread of infection or diseases. Rabies is a risk in Africa, and there are probably many arboviruses (arthropod- borne viruses) that might potentially be found in small mammals and their ectoparasites. Plague, a bacterial disease transmitted by fleas, is endemic in parts of Africa. For these reasons, field workers should take care to avoid being bitten by small mammals, and, if surveyors are bitten or scratched, lacerations should be treated immediately with antiseptic and bandages. Even with dead animals, care needs to be taken not to get scratches during specimen processing. Always wear a face or surgical mask and protective gloves when collecting trapped animals and preparing specimens (to avoid being bitten by any ectoparasites, such as fleas, ticks and lice), or when doing carnivore scat analysis (for indirect evidence of occurrence).

4.6 Conclusions There is a wealth of methods that can be used, and adapted, in order to survey small mammals. All require capturing the subjects of a survey, so a combination of methods should be selected to ensure that those mammals that avoid one method are captured with another.

64 4.7 References

Barlow, K. (1999). Expedition Field Techniques: Bats. Royal Geographical Society, London, UK.

Barnett, A. (1992). Expedition Field Techniques: Small Mammals (excluding Bats). Royal Geographical Society, London, UK.

Corbet, G.B. & Hill, J.E. (1991). A World List of Mammalian Species. 3rd edn. Oxford University Press, Oxford, UK.

Delany, M.J. (1986). Ecology of small rodents in Africa. Mammal Rev. 16: 1–41.

Ferreira, S.M. & van Aarde, R.J. (2000). Maintaining diversity through intermediate disturbances: evidence from rodents colonising rehabilitating coastal dunes. Afr. J. Ecol. 38(4): 286–294.

Fitzgibbon, C.D. and Rathburn, G.B (1994). Surveying Rhynhocyon Elephant-Shrews in tropical forest. Afr. J. Ecol. 32(1): 50–57.

Ferreira, S.M. & van Aarde, R.J. (1997). The chronosequence of rehabilitating stands of coastal dune forests: Do small mammals confirm it? S. Afr. J. Sci. 93(5): 211–214.

Francis, C.M. (1989). Comparison of mist nets and two designs of harp traps for capturing bats. J. Mammal. 70: 865–870.

Happold, D.C.D. (1987). The Mammals of Nigeria. Clarendon Press, Oxford, UK.

Hilton-Taylor, C. (compiler) (2000). 2000 IUCN Red List of Threatened Species. (Including CD- ROM). IUCN, Gland and Cambridge.

Hutson, A.M., Mickleburgh, S.P. & Racey, P.A. (Eds.) (2001). Microchiropteran Bats: Global Status Survey and Conservation Action Plan. IUCN, Gland, Switzerland.

Kingdon, J. (1974). East African Mammals: An Atlas of Evolution in Africa. (Vols 2A & 2B). Academic Press, London, UK.

Kingdon, J. (1997). The Kingdon Field Guide to African Mammals. Academic Press, London, UK.

Kunz, T.H. (1988). Ecological and Behavioural Methods for the Study of Bats. Smithsonian Institution Press, Washington, USA.

Kunz, T.H., Tideman, C.R. & Richards, G.C. (1996). Capturing mammals: small volant mammals. In: Measuring and Monitoring Biodiversity: Standard Methods for Mammals, pp 123–146. (Ed. by D.E. Wilson, F.R. Cole, J.D. Nichols, R. Rudran, & M.S. Foster). Smithsonian Institution Press, Washington, USA.

Lidicker, W.Z. (1989). Rodents: A World Survey of Species of Conservation Concern. IUCN SSC Occasional Paper no. 4. IUCN, Gland, Switzerland.

Malcolm, J.R. & Ray, J.C. (2000). Influence of timber extraction routes on central African small mammal communities, forest structure, and tree diversity. Cons. Biol. 14: 1623–1638.

Mickleburg, S., Hutson, A.M. & Racey, P.A. (Eds.) (1992). Old World Fruit Bats: An Action Plan for their Conservation. IUCN, Gland, Switzerland.

Nicoll, M.E. & Rathbun, G.B. (1990). African Insectivores and Elephant-shrews: An Action Plan for their Conservation. IUCN, Gland, Switzerland. 65 Nowak, R.M. (1999). Walkers Mammals of the World. 6th edn. John Hopkins University Press, Baltimore.

Pollock, K.H., Nichols, J.D., Brownie, C. & Hines, J.E. (1990). Statistical inference for capture- recapture experiments. Wildl. Monogr. 107: 1–97.

Ray, J.C. & Hutterer, R. (1996). Structure of a shrew community in Central African Republic based on the analysis of carnivore scats, with the description of a new Sylvisorex (Mammalia: Soricidae). Ecotropica 1: 85–97.

Rosevear, D.R. (1965). The Bats of West Africa. British Museum (Natural History), London, UK.

Rosevear, D.R. (1969). The Rodents of West Africa. British Museum (Natural History), London, UK.

Schlitter, D.A. (1989). African rodents of special concern: a preliminary assessment. In: Rodents: A World Survey of Species of Conservation Concern. (Ed. by W.Z. Lidicker). IUCN SSC Occasional paper No. 4. IUCN, Gland, Switzerland.

White, F. (1983). The Vegetation of Africa. UNESCO, Paris, France.

Wilson, D.E. & Reeder, D.M. (eds). (1993). Mammal Species of the World: A Taxonomic and Geographic Reference. 2nd Edition. Smithsonian Institution Press, Washington, USA.

Wilson, D.E., Cole, F.R., Nichols, J.D., Rudran, R. & Foster, M.S. (Eds.) (1996). Measuring and Monitoring Biological Diversity: Standard Methods for Mammals. Smithsonian Institution Press, Washington, USA.

66 Form 4.1: Small Mammal Catch Records (instructions see p52)

Surveyor: Field sheet ref: Date: (total observers): (dd/mm/yy)

Address:

Survey site: Altitude: Aspect:

Latitude: Longitude: UTM (if available):

Vegetation: Human disturbance:

Season: Weather: Lunar phase: Temperature:

Other:

Trap line Trap type Micro- Water Topography Species & Other & no. & bait habitat association specimen sheet ref.

67 Form 4.2: Specimen Records: bats, rodents and insectivores

Specimen sheet ref: Field sheet ref: Collector: Date: Time: (dd/mm/yy) Address: Collecting site: Altitude: Latitude: Longitude: Slope: Additional notes: Species: Field no.: Sex (if known): Age: Pregnant/lactating: Embryos: Breeding condition: Colour/markings: Wounds: Dental formulae: Mammary formulae: Ectoparasites: Endoparasites:

Measurements: Bats: FA TR mm mm

HB TL TV E HF W mm mm mm mm mm g Material Preserved: Skin Skull Skeleton Stomach Faeces Blood Liver Kidneys Stomach contents: Component: Percentage:

Remarks/Other

68 5. Large and medium mammals Helen Newing, Glyn Davies and Matthew Linkie zebra duiker ( Cephalophus zebra) Cephalophus

5.1 Biology This chapter will concentrate on the ungulates and the carnivores. Some species are sufficiently abundant in African forests to allow direct counting of animals to estimate popuation sizes, but many others are rarely seen, and therefore surveys rely on signs to assess their presence and gain a rough index of their abundance. The signs include conspicuous footprints in soft ground; large or persistent dung piles; diggings, and broken or trampled vegetation. This applies to many of the larger ungulates (see below), carnivores, pygmy hippo (Hexaprotodon liberiensis), and the pangolins or scaly anteaters, which live in burrows or tree-holes and feed on ants and termites. Ungulates (Orders Proboscidea, Perissodactyla, and Artiodactyla) Forest ungulates (mammals which walk on the tips of their toes) fall into two distinct size-groups. The small-bodied species (5-70kg) include duikers, chevrotain and bushbuck, which live in well-defined, stable home ranges. Larger species (>100kg) such as bushpig, giant forest hog, bongo, okapi, buf- falo, rhinoceros and elephant are often wide-ranging or even migratory. In general, ungulates have a keen sense of hearing and smell, but their eyesight is relatively poor. Duikers are the most commonly seen terrestrial mammals in many African forests, although daytime sightings all too often consist of a quick

69 movement and rustle in the bushes before you hear a whistle as they bound away. Indeed, their shy and reclusive habit is one of the reasons why they are so poorly studied. For many years they were thought to be a homogeneous group of solitary, nocturnal, monogamous fruit-eaters, but research since the 1980s has revealed that they vary considerably in ecological characteristics (Dubost, 1984; Feer, 1989). They have a varied vegetable diet including leafy browse as well as seeds and fruits that fall to the forest floor. Some species such as the blue duiker (Philantomba monticola) and the related Maxwell’s duiker (P. maxwelli) are active during the day. Others such as the bay duiker (Cephalophus dorsalis) are nocturnal, and the larger species (yellow-backed duiker, C. sylvicultor; Abbott’s duiker, C. spadix and Jentink’s duiker, C. jentinki) are active both by day and by night. Some species are solitary, some live in pairs, and others have been recorded occasionally in groups with one adult male and a number of adult females (e.g. Maxwell’s duiker: Newing, 1994; Peter’s duiker, C. callipygus: Feer, 1989). All appear to have territories that are marked with dung piles and musk from scent glands. Large-bodied species such as suids (pigs), bongos, okapis, buffaloes, elephants and rhinos have lower population densities than duikers. They may also engage in seasonal migrations. Surveys of these species are therefore based mainly on sign rather than direct sightings, as explained above.

Carnivores (Order Carnivora) All carnivores range widely relative to their body size – the larger the species the further they range – and they live at low population densities compared with the herbivorous mammals on which they prey. African forest carnivores range in size from the mongooses (300g to 5kg), genets and linsangs (500g to 3kg), otters, and the African palm civet (3kg) to the civet (c. 15kg), golden cat (c. 15kg) and leopard (c. 60–90 kg). Some open forest formations in East Africa are also home to striped hyenas (c. 40kg). All carnivores have good senses of smell, hearing and sight and are seldom seen as a result. Population surveys of smaller species depend on trapping animals using fish and meat baits (see previous chapter), whereas survey methods for larger cats include recording pug-marks (tracks), scats, scrapes and kills, and the use of photo-traps.

5.2 Management issues There are a number of important management reasons to survey larger mammals. From a conservation perspective, it is important to know how many animals are in different areas so that management plans can take account of migration routes and important locations for food or for refuge, as well as

70 identifying areas with concentrations of animals that have potential for ecotourism. Where animal populations are harvested, the effects of harvesting need to be monitored to ensure that harvesting is sustainable. Furthermore there are often conflicts between large terrestrial mammals and humans. Solutions to all of these management issues need to be guided by information from biological surveys. At a time when deforestation is accelerating across Africa, survey information is particularly important to assess and monitor the long-term effects of habitat changes. These range from complete loss of forest to minor vegetation changes due to intermittent use. Forest clearance leaves fragmented islands of forest containing small populations of ungulates that are often not viable in the long term. In contrast, selective logging can create patches of secondary vegetation that benefit a large proportion of forest ungulates that are grazers or mixed browser-frugivores (although a few species do appear truly dependent on old growth forest). Hunting and trapping heavily affect many forest species. Showy species, such as leopards and bongos, are hunted for their skins. Forest elephants, despite having smaller tusks than their savannah relatives, have long been hunted for ivory. Duikers have been hunted and trapped for food for centuries and are still the main source of fresh meat in many forested areas of Africa. Hunting and trapping have increased dramatically over the past few decades – partly because of rural population growth and partly because of increased trade to supply growing markets in urban centres, facilitated by improved access along logging roads. As a result, although the smaller duiker species reproduce quite quickly, their populations have been eliminated from many forest areas near large human settlements and roads (Wilkie & Finn, 1990; Muchaal & Ngandjui, 1999; Noss, 1999); larger-bodied mammals reproduce the slowest. In addition to human impacts on wildlife, large mammals can have a serious impact on humans. Elephants are major crop pests and the cause of frequent complaint from communities near forest areas. Forest pigs and buffaloes also cause problems for farmers, especially in terms of trampling and digging, and bushbucks eat vegetables in fields near villages. Leopards may prey on goats, sheep and cattle. There must be a balance between wildlife conservation and the control of animal pests, and to accomplish this we need effective monitoring of animal populations (Bell, 1984; Hill, 1998, 2000; Naughton et al., 1999). On a more positive note, large mammals are an important resource for tourism development, and forest populations of elephants, bongos, rhinos, and to a lesser extent buffaloes, can be attracted to waterholes and saltlicks near lodges for high-quality tourist viewing. However, surveys and monitoring are important to assess impacts of tourism on wildlife. A concentration of large

71 herbivores at waterholes can cause serious damage to the surrounding vegetation. Tourist trails can affect animal distributions in different ways – large animals, such as elephants and pygmy hippos will actually use these trails, because, like people, they find it easier than pushing their way through the undergrowth, but heavy tourist use and inappropriate behaviour by tourists (or by survey teams) may frighten animals away from the area. Conversely, species that are hunted may congregate in areas where there are tourists, because hunters are less likely to come there.

5.3 Methods

General Mammal surveys can provide three levels of data for managers. At the most basic level they can determine the presence or absence of different species at different sites in order to build up distribution maps. Information from such distribution surveys (see section 6.3.1) is most valuable for rare or endangered species or for species that can be used as indicators of forest condition. At the next level of detail, simple sampling can be used to determine the relative abundance of a species at different sites, or at a single site over time. At the third and final level, much more rigorous sampling, more extensive data collection, and thus use of robust statistical analyses makes it possible in some cases to arrive at a quantitative population density estimate. Surveys of terrestrial forest mammals are hampered because many animal species are shy and secretive, hiding in the undergrowth. However, several successful survey methods have been developed to overcome this problem for some groups. They can be divided into direct methods, which are based on sightings of the animals, and indirect methods, which are based on counting their signs. The rest of this chapter describes the various methods and indicates the strengths and weaknesses of each of them. Seasonal variations in climate can result in dramatic changes in animal behaviour, in visibility (according to density of undergrowth), and in the length of time that tracks and signs remain visible. Therefore basic surveys to determine presence or absence of species may benefit from short visits in different seasons, but any comparative studies should be restricted to a single season.

Identification A number of excellent field guides are available for identifying the larger mammals of Africa. Dorst & Dandelot (1983), Haltenorth & Diller (1984) and Kingdon (1997) all provide detailed information on identification, distribution and ecology. Other useful guides include Stuart & Stuart (1995, 1997) and

72 Estes (1991). Although not a field guide, Rosevear (1974) is a useful reference for identifying West African forest carnivores. 5.3.1 Hunters’ calls, attractants and observation points Hunters traditionally use a variety of snorts, calls and whistles to attract different species of animals. The best example is the nasal bleat used to attract duikers, and researchers have used hunters’ calls to check for the presence of different duiker species (e.g. Wilson, 1990). The results can be impressive: within a few minutes animals will often come running to within a few metres of the caller. This technique is most successful if experienced hunters are employed to do the calling, and they should not be accompanied by more than two surveyors. Calls should be made at distances of not less than 250m from each other. The surveyors should position themselves in an inconspicuous place, such as between the roots of a tree buttress, and remain still and quiet while the hunter calls. Other attractants include natural or artificial salt licks for herbivores and meat or scent stations for carnivores (see Blum & Escherich, 1979, for the latter). It may take a few weeks for animals to find a newly-sited attractant, so they are not suitable for quick, one-off surveys.

Equipment camouflage-netting and string machete, for construction of temporary hide

Site selection Strategic observation points include: natural salt-licks, waterholes and wallows, heavily fruiting trees, tree-fall gaps with a flush of new foliage, forest glades, logging roads, and areas with regular signs of tracks.

Procedure i) A simple hide can be built either with camouflage-netting hung between trees or buttress roots, or by cutting the fronds of a palm tree to form a see-through wall. In protected areas, check to make sure this is not prohibit- ed. The hide should be located downwind of the observation site. Make sure there is a comfortable sitting place so that you don’t fidget. Alternatively, rather than build a hide, one can sit on a low branch in a tree or on a ridge or rocky outcrop overlooking the forest floor below – few terrestrial mammals notice stationary objects above their heads.

73 ii) Once a hide has been constructed it should be left for at least a day before being used, so that animals become accustomed to it. iii) The best time to start watches for most species is just before dawn (so that you are settled down before the first light) or a couple of hours before dusk. Approach the hide quietly from the opposite direction to the observation site. Settle in a comfortable position so that you can keep still (use mosquito repellent!). iv) It is sometimes helpful to visit the ‘observation area’ to look on the ground for footprints and other signs (hairs, dung, spoor, etc.), especially if watches are not producing many sightings. This can be done in the midday hours so that dawn and dusk watches are not disrupted. v) A reasonable length of time for a first watch at a site is two to three hours. Leave the hide quietly, and in the opposite direction from the observation area.

Recording i) Fill in the survey data in the top section of the recording sheet (Form 5.1) before commencing the observation period. Give each survey site a name, and give observation points within each site numbers or codes (e.g. Survey site: Gouleako; observation point 3). Record the type of vegetation, the degree of human disturbance (which you may know from archival information or from direct observation), and any special features that may be relevant (e.g. riverine forest, mature lightly-logged, many-fruiting figs). Include altitude if you are working across an altitudinal gradient. ii) When one or more mammals are seen, note the time and watch quietly for a few minutes, even if you can identify the species immediately. If they remain at the site, begin to fill in the recording sheet (Form 5.1) very quietly noting the time at the start of the observation, the species and the total number of animals. Additional observations include the number of males, females and young animals, if known, and their behaviour. Make a note of any food eaten, and take a sample if necessary. If you cannot identify the animal species, write a detailed description in your notebook including an estimate of height, the shape (especially of the head and muzzle), horns (if present) and the coat pattern. Make a sketch of it. iii) Remember to note down the time at which the animals leave. When they have gone you can expand your observation notes and consult a mammal field guide to check any species identifications you’re not sure about. Also check the observation area for footprints, hairs, fallen fruits, etc.

74 Advantages/limitations This is a time-consuming method, and should be undertaken only at sites where there is evidence that animals are frequent visitors. Some skill is needed in identifying suitable sites, and consulting local hunters and others knowledgeable about local wildlife can save much time and effort. Within these limitations, successful watches are a valuable and rewarding way to record the presence of different species, to study behaviour, and to determine whether sites hold any potential for regular viewing by tourists. There are several potential negative impacts of longer-term feeding sites. Animals may come to rely on being provisioned at the site, or they may be placed in acute competition for the food at the feeding site. Alternatively, hunters may learn of the site.

5.3.2 Net drives Net drives are used both for traditional hunting, for example in the Central African Republic (Noss, 1999) and Zanzibar Island (Archer, 1994); and also by researchers to catch small deer and antelope (e.g. Bowland, 1990; Newing, 1994). There are inherent biases in the use of this method for determining population abundance. For example, radio-tracking has confirmed that Maxwell’s duikers and red duikers sometimes move away from the nets or drivers before they are observed (Bowland, 1990; Newing, 1994). Some duikers, especially infants, will be missed because they freeze in thickets and are not flushed out of their resting places. Nonetheless, net drives are sometimes the best option to survey secretive ungulates, especially where thick vegetation makes direct sightings difficult. The biases probably differ for different species, but can be assumed to cause an underestimate. Net drives can also be extremely noisy and disruptive, and there has been some concern about impacts on resident animal populations. However, evidence from radio-tracked animals has shown that they quickly return to their territories and resume their usual activities once the drivers have departed.

Equipment/personnel tape measure hunting nets (e.g. 50m x 2m; mesh 25–50mm; dark-coloured, elastic, ideally with a breaking strain of at least 100 kg) 60cm lengths of strong nylon cord attached to top of nets at intervals of about 4m machetes

75 large team (minimum of ten) of drivers/field assistants (with or without hunting dogs)

Site selection Within a survey site the region is stratified as described for transect surveys (see below) and sample units are selected to cover the full breadth and range of habitats.

Procedure and recording i) If experienced hunters are taking part in the survey, it is best to follow their normal method (see Noss, 1999), since it is likely that they are successful at finding animals. Hunters usually enclose a large area (typically 4–5ha) with nets linked together. ii) Where hunters are not involved the procedure will vary according to the thickness of vegetation, the amount of netting available, and the numbers and knowledge of field assistants/drivers. Each drive usually covers 0.5–1.0ha. If the forest is quite easy to walk through and visibility is good, it may be sufficient to set up a single net-line immediately before the drive takes place, on one side of the block to be searched. Where vegetation is very thick, a grid of trails (50m x 50m) should be cleared a few days before the drive takes place. This will allow the nets to be erected quickly and quietly and provide a sighting line for the monitors. In thick vegetation, it is advisable to net at least three sides of the drive area to make sure animals don’t escape unobserved. The more nets used, the greater the time needed per drive, but the smaller the number of people needed to monitor the edges of the block. iii) At the beginning of the drive, the field surveyors should surround the block to be surveyed as quietly as possible. The 2m-high x 50m-long nets should be erected by tying the nylon cords to trees at a height of 1.5–2 m, leaving enough slack on the ground to stop animals diving underneath. Once the nets are up, people should be stationed along three sides of the block with most people on the sides that are not netted, close enough to each other so that they can see clearly along the whole length of each side. This will be a distance of between 15 and 30m, depending on the thickness of vegetation (they usually need to be much closer together on the open sides). These people are the monitors. iv) The rest of the assistants form a line at one end of the block, opposite a netted side. These are the beaters. Once the nets are set up, the beaters enter the area and noisily dislodge animals from their hiding places by shouting and banging trees, bushes and fallen logs with sticks. Dogs may also be used for this purpose, but they must be kept under strict control at all times.

76 v) Some animals typically try to break back through the beaters, and so beaters should be positioned closer together than the monitors (10–15m apart). The more people involved, the larger the area that can be included in a single drive. Depending on team size and conditions, each drive (including setting up nets) can take anything from 30 minutes to two hours. vi) Whenever an animal is detected the species and, if possible, age and sex should be noted. The sighting is called out to the neighbouring monitors to avoid animals being counted more than once. Any animals that are caught should be restrained and examined, noting age/sex and condition. Specimens may also, under certain circumstances, be collected (see section 4.4). vii) At the end of the drive, the team comes together to compare observations and confirm the total number of animals of each species they have seen. Basic information on the survey site, vegetation, weather and time of drive should also be noted; Form 5.1 can be used, omitting the first and last columns.

Data analysis The population density is computed as the number of duikers counted divided by the area of forest sampled during the drives. During fieldwork, the cumulative density can be estimated at the end of each day of drives (by sum- ming results from all previous drives), and a graph drawn to get an idea of when the density estimate stabilises and sample size is sufficient.

Advantages/limitations Net drives are only feasible if a large labour force can be organised and transported for a few days to the survey sites. If this is possible, the method gives a large number of direct sightings and offers one of the most accurate ways of getting population information for management purposes. In areas of thick, secondary vegetation, it is sometimes the only option for surveys of terrestrial mammals. 5.3.3 Survey walks: reconnaissance and transect The use of observation points or net drives is most suitable for concentrated surveys in small areas of forest. To survey a larger area with relatively small survey teams (two or three surveyors) the only realistic option is to carry out survey walks through the area. Survey walks are used in two ways: either for a ‘quick and dirty’ first assessment (a reconnaissance survey), or for a more methodical evaluation of relative abundance or population density, through the use of carefully positioned transects (straight trails). They can be based on

77 either direct encounters with animals or indirect signs (footprints and dung – see section 5.3.4). It is not possible for a single surveyor to search thoroughly for both at the same time, but different members of the survey team can concentrate on different aspects in order to maximise data collection from each study site. Surveys should be carried out by night as well as by day, because a) nocturnal species will be seen that otherwise might remain undetected; and b) many diurnal species such as blue and Maxwell’s duikers freeze when caught in the beam of a strong torch, thereby allowing the surveyor to determine age and sex of individuals. To carry out transect surveys, new transects should be cut that are carefully located in order to sample different vegetation types and levels of human disturbance in proportion to their estimated occurrence in the study area. The perpendicular distance of each animal or group of animals (or each dung pile or group of piles) from the centre of the transect is measured, so that an estimate of population density (or dung density) can be made. For population estimates, a minimum of 40 sightings per species in each habitat is necessary, and ideally over 100 sightings should be used (Plumptre, 2000). Therefore, transect surveys can only generate population estimates for species that are seen relatively often. In practice, sightings of all species are recorded, and different techniques are used to analyse data for each species, depending on the amount of information gathered. The procedure for reconnaissance surveys is similar to that described below for transect surveys, but with two major differences – firstly in the sampling, and secondly in the lack of recording of perpendicular distances from the transect. Reconnaissance surveys differ from transect surveys in that they ‘follow the line of least resistance’ through the vegetation in order to cover as much ground as possible. They may use existing human or animal paths, follow streambeds or concentrate in areas of sparse undergrowth where it is possible to walk in a straight line without clearing vegetation (Walsh and White, 1999). They consist simply of recording all encounters with animals and sign for a given distance walked. Vegetation types and levels of human disturbance can be recorded during an initial reconnaissance survey.

Equipment/personnel To set up transects 30m tape-measure or topofil (hip-chain) fluorescent vinyl flagging tape & marker pens two machetes team of four or more people (including two or more line-cutters) maps, GPS, altimeter and clinometer if you are also undertaking mapping

78 To carry out transect surveys and reconnaissance surveys one or, at most, two people optional: optical range finder or survey laser binoculars

Site selection i) As a general rule, transects should cover the main habitats present in the same proportions as they occur in the study area. For studies in undisturbed forest, this is usually done by using the bottom of the valley as the main axis and cutting transects perpendicular to it. For surveys of very large areas, a series of baselines can be cut parallel to the valley bottom – for example, at intervals of 5, 10 or 50km – and transects can be cut perpendicular to each baseline. Transects should be a minimum of 2km apart (preferably more), and should not cross each other. (See White & Edwards, 2000, Chapter 3, for further discussion of stratified sampling). ii) Surveys that aim to determine the abundance of animal populations at the regional level must usually include areas with different levels of human activity and habitat disturbance (farmbush, secondary forest, logged forest, undisturbed forest). If disturbance is likely to have a greater effect on mammal populations than watersheds (for example, where hunting or farming is very intensive), it is more appropriate to use a road as the primary axis or to balance sampling according to the distance from human settlements (e.g. Lahm et al., 1998). iii) Once the approximate survey areas are decided, transects should be positioned at least 300–500m apart and at least 500m from the base camp, because camp noises and smells often deter mammals (unless they are searching your rubbish for food!). iv) Large animals may use existing paths, so new paths should be cut for surveying in order to ensure random sampling. However, cutting paths during surveys would frighten the animals away. As a compromise, new transects can be cut a day or so before beginning the survey. If the cutting of new transects is not felt to be justified and old paths are used, it is better to use narrow trails than wide tracks. v) Using existing trails and roads is an efficient and useful way to check for the presence of species. However, using long-established trails that hunters and trappers follow, or logging tracks which have very different vegetation to the rest of the area, will seriously bias survey results and influence population estimates.

Procedure — cutting transects i) Use a random number table to select a starting point and direction (left or right) from the baseline for each transect. Alternatively, select the

79 transect direction to pass through habitats of survey interest. The choice of options will depend upon the overall aims of the study and the degree of habitat specialisation of the animals you are surveying. Each transect should be of a length that can be traversed in a single session (typically 3–5km). ii) Each transect should be cut perpendicular to the baseline, using a compass to keep it straight and the 30m tape or hip-chain to record the distance cut. Slight detours around obstacles such as tree-falls are acceptable as long as the original direction of the transect is maintained. If the vegetation is thick, two people should take turns at cutting. Record special features (hunters’ camps, streams, etc.) as they are encountered, so that they can be mapped immediately. iii) Cut only the minimum vegetation necessary. If transects are going to be used only during the day for a period of weeks, it is necessary only to make the slightest clearance to allow movement through the area following a compass bearing (e.g. Walsh & White, 1999). If they are to be used at night, they must be visible enough so that they can be followed easily without distracting from the search for animals during night censuses. If a long-term study site is being created, trails should be cleared sufficiently so that they will need only minor maintenance. iv) Mark the distance along the transect every 50m. If transects are only going to be used for a few months, write the distances on pieces of fluorescent vinyl marker tape and tie to saplings by the side of the path. If permanent transects are being set up, paint the distances on trees, or nail numbered aluminium tags to tree trunks. v) Allow newly cleared transects to rest for at least one whole day before beginning surveys. This will let animals recover from the disturbance and return to their normal haunts and habits. This is also a good time to draw a sketch map of the transects. vi) As a precaution against opening up new areas of forest for hunters and trappers, the starting point of a transect may be disguised so as not to draw attention to the new path, or may be started 50m inside the forest (from a road or logging track). Some surveyors even cut transects with secateurs so that no path is left after them. If the transect is not going to be re-used, all distance markers should be removed at the end of the survey period.

Carrying out surveys i) Daytime surveys should preferably be carried out in the mornings from just after dawn to about 11:00, when animals are most active. If time is short, additional afternoon censuses can be carried out between 15:00 and 17:30, after which visibility becomes very poor. Night-time surveys can be conducted at any time during the night, but 20:00 is a common start time.

80 ii) Do not carry out surveys in the rain: it will be harder to pick up sounds, and will affect the behaviour of animals (and surveyors!). If it begins to rain, pause the survey until the dripping stops, or if it seems to be likely to continue, discontinue the survey. iii) Ideally, transects should be walked by a single observer in the daytime, or by two observers at night. Walk very slowly, looking from side to side for movements and listening intently for sounds in the undergrowth, animal calls or gnawing sounds. Aim for an average speed of about 1km per hour, including a pause every 100m or so to stop and listen. Additional time will be taken to record data when animals are sighted. Each transect should therefore take between three and five hours. iv) Accurate distance estimation is essential, and the use of an optical range finder is recommended (see also section 2.5). v) If possible, rotate observers between transect lines to cancel out idio- syncratic differences. It also makes it more interesting and gives everyone the chance to see unusual tracks and signs, which can be marked with fluorescent tape.

Recording i) At the start of each transect walk, fill out the top part of the form below (Form 5.2). When one or more mammals are sighted (and often this will be of animals in flight), complete the columns in the second part of the table. ii) The perpendicular distance is taken from the position at which the animal was first detected to the nearest point on the transect, thus it may be necessary to walk along the trail to reach the correct point. It also may be necessary to approach the animals quietly in order to get a clearer sighting. Additional observations can include the number of males and females, and presence of infants, activities and behaviour, and association or interactions with other species.

Data analysis i) Relative abundance: for species where sample sizes are too small to allow estimates of population density to be made, the number of animals encountered per kilometre walked gives a rough indicator of abundance. ii) Estimating population densities from line transect censuses involves complex mathematical modelling of the likelihood that animals are detected at different distances from the transect. However, these statistical analyses can be executed by the custom-made computer program DISTANCE, which is downloadable from the internet (www.ruwpa.st-and.ac.uk/distance) and should be used in conjunction with the accompanying book (Buckland et al., 1993).

81 Analysis rests on several assumptions: that transects are placed randomly with respect to the distribution of animals; that all animals on the line (strip width) are always detected; that animals are detected at their initial location, i.e. prior to any movement in response to the observer; that measurements of perpendicular distances (animal-transect, observer-transect) are exact. It is rarely the case that all of these assumptions are met perfectly, which means that population estimates must be treated with some caution. For a full review of the key issues on theoretical aspects of analysis, and how to deal with discrepancies, see Buckland et al. (1993).

Advantages/limitations The main advantage of reconnaissance surveys is that they are quick. Also, they can be carried out during routine activities such as patrolling by park guards, so lend themselves to monitoring. The main disadvantage is that samples are likely to be biased in terms of habitat and intensity of human use. However, some studies of elephant dung and gorilla nests have compared results of reconnaissance surveys and full transect surveys and have found a high level of correlation. The general rule is that recce surveys should be backed up by some formal transect surveys in order to evaluate any biases (Walsh & White, 1999; White & Edwards, 2000, Chapter 13). 5.3.4 Indirect methods When the survey subjects are not seen directly, or seen very rarely, then indirect survey methods are needed. This means that surveys are made for the signs left by the animals, and the population density of the animals producing the signs is estimated. Signs include faeces (e.g. pellet piles of duikers, dung piles of elephants and buffaloes, scats of cats), footprints or spoor, hairs, diggings and nests (for pigs), urine-marking sites (pygmy hippo, rhinoceros and carnivores). In this section, attention is focused on three approaches: dung counts, track counts, and photo-traps. A. Dung counts Animal population density can, in theory, be calculated from dung density on the forest floor according to two variables: The number of dung piles produced per animal per day (defecation rate). The length of time the dung takes to disappear (dung-decay rate). Unfortunately these two variables are affected by several different factors, and 82 this introduces many potential sources of error. In addition, for groups such as medium-sized duikers or small carnivores it is difficult to identify dung to species level. For example, only biochemical techniques will enable identification between the scats of golden cats and mongooses. Defecation rates vary with diet (e.g. White, 1995), and, in the case of big cats, with the oestrous cycle of females. Decay rates vary with the weather, microclimatic conditions, and with dung beetle activity. In Ugandan forests, for example, Nummelin (1990) showed that duiker pellets were encountered less frequently when rainfall was high prior to surveys. Any errors in estimation of decay rate and defecation rate will have a radical effect on the estimation of population density (Plumptre, 2000). Furthermore, territorial species such as duikers and carnivores use droppings to mark their territories, so distribution of dung is not random, presenting complex sampling problems. In the case of migratory or far-ranging species such as elephants, the survey area may not cover their whole range, so dung density will vary according to the passage of elephants through the survey area in the period prior to the survey. However, in spite of methodological difficulties, dung counts may be the best option available for surveys, since dung is the most frequently encountered sign of many larger forest mammals. In an effort to address the constraints, detailed techniques have been developed to use dung counts for surveying forest elephants (Barnes and Jensen, 1987), and similar methods have been used for species ranging from buffaloes to duikers. If dung density is very high, it may be possible to do away with the need to consider decay rates by surveying the same transects repeatedly and clearing them of dung after each survey. The number of piles produced for the unit area surveyed within a known time period (i.e between the two consecutive surveys) can then be recorded (Plumptre, 2000). For elephants, density estimates from dung counts have been shown to correlate well with those from other methods (Barnes, 2001). In general, however, the smaller the species the less useful the method, because decay rates are much more variable (see Plumptre & Harris, 1995, for discussion of methodological issues). For this reason, dung counts for smaller species should only be used with caution for a first indication of relative abundance.

Equipment/personnel 30m tape-measure or topofil (hip-chain) steel tape-measure (1 mm gradations) fluorescent vinyl marker tape and marker pens to mark distances, or more permanent numbered aluminium discs, hammer and nails team of four people

83 Site selection Most aspects of site selection are the same as those for transect surveys (section 5.3.3.). However, new transects should be used for dung transects. These may be along newly cut trails or may simply be unmarked survey routes following a compass bearing.

Procedure i) It is advisable for every study to begin with dung decay trials to determine the length of time dung piles remain on the forest floor (but see ii below). To do a decay rate trial, at least 50 fresh dung piles should be located and marked. For elephants, each pile should be visited once a week and its state of decay eval- uated according to the categories on Form 5.3. In north-east Gabon, an average decay rate for elephant dung was calculated at 2.4% per day, with individual dung piles lasting from a few days to many weeks (Barnes & Jensen, 1987). ii) Barnes et al. (1997) showed that decay times for elephant dung are inversely related to rainfall in the month of deposition, and dung densities are affected by rainfall in the previous two months. However, this varies between different geographical areas (Nchanji & Plumptre, 2001). Therefore, unless detailed baseline studies have been done in a given country or region, further surveys are necessary to estimate decay rates and calculate dung density per elephant (see Form 5.3 notes on elephant dung decomposition states). iii) The principles and procedures for dung surveys are similar to those explained for transect surveys in the previous section (section 5.3.3). iv) During dung surveys, one person should search the ground for dung piles while the others maintain the compass route, measure the distance walked, and cut a path. The searcher should advance slowly, scanning the ground from side to side. This takes a lot of concentration, and a different member of the team should take over as searcher every 250m or so. It may be worth surveying shorter distances (subsamples of the main transect) more slowly and carefully for the pellets of smaller species (e.g. duikers), at least for presence/absence data. v) For leopards and many other carnivores, cutting a transect through the forest will yield little information, and it may be best to survey along roads and trails. Distances can be measured by routing on a GPS.

Recording i) When a dung pile is detected, record the distance along the transect (with the 30m tape or hip-chain), the perpendicular distance from the centre of the transect to the centre of the dung pile (with the steel tape-measure), the state of decomposition of the boli (the individual spheres of dung), and the vegetation type at that location. See Fig. 5.1

84 Transect route ii) You can also record tracks encountered incidentally on the same sheet (Form 5.3), putting a track measurements and other comments in the general observations column. b1 Fig. 5.1: Dung counts Transect measures for pellet/pile counts. Measure the perpendicular dis- b2 c tance from the centre of a duiker pellet 1 pile to the transect (a). For elephants, piles may be more dispersed and need to be measured in two ways: i) pile clusters on one side of the transect should

be measured to the outer (b1) and inner (b2) limit, these added together and then divided by two; and ii) for a dispersed pile on either side of the trail, measure the outermost perpendicular distance on

either side (c1 & c2), then subtract these two distances and divide by two (after c2 White & Edwards, 2000).

Data analysis i) Calculating the width of the survey transect is done as for sighting transects (above), using the perpendicular distances from the trail to each pile detected (in the case of elephants, excluding E state bolus piles, Barnes & Jensen, 1987). The mathematical calculations used to assess the variable strip width are complex but can be done using the computer program DISTANCE (section 5.3.3.). If possible, a minimum of 100 dung piles should be recorded (Plumptre, 2000), and the absolute minimum for this method of analysis is generally taken to be 40. ii) The area of the transect is calculated as length multiplied by width, and the density of droppings as the number of dung piles divided by area. iii) Once dung density has been calculated, figures are required for defecation and decay rates. Unless you have been able to calculate a defecation rate, use a standard rate of 17 boli/day for elephants (based on Wing & Buss, 1970). For duikers the situation is more problematic, because it is unlikely that defecation or decay rates will be transferable between different species, forests or seasons, and few have been studied (Koster & Hart, 1988).

85 Short-cut method for elephant dung counts In an effort to reduce the time needed for detailed surveys, Barnes (1988) describes a short-cut method, whereby an observer follows a compass bearing without cutting a trail, and records the presence of all elephant dung piles (decomposition states A to E). At the most basic level of analysis, the proportion of 500m sections along the transect which contain dung is fitted to a calibration curve established during more thorough research efforts, and the dropping density (piles/km2) read from the graph. The same data analysis procedures as described above are then used to derive population estimates. This is a very coarse-grained method of estimating densities. However, as with reconnaissance surveys, the advantage is that a great distance can be covered relatively quickly. It provides distribution data and gives some indication of the relative importance of different areas for elephants in different seasons. Fuller use has been made of field data (the recce method) in elephant surveys in Gabon. Unlike in the full method described above, distance to each dung pile was not measured, but estimates were made from the number of piles/km. This method is obviously less accurate, but it can cover about four times more ground than the full method above (with the same effort). Walsh & White (1999) suggest using both methods, and calibrating different estimates.

B. Track (footprint) surveys Footprints or spoor (e.g. tracks of duikers, pug-marks of cats) give vital information on the presence of species, including those that are rare or hard to spot, and can be carried out alongside other types of survey work. However, this method is less robust than dung counts for estimates of relative abundance because track densities are affected by the type and dampness of the soil substrate, rainfall, and the movement patterns of animals through the survey area. Also, track size and shape change with the animal’s gait, the soil substrate, and the age of the track. Fresh tracks in an ideal soil type have well-defined, vertical edges, making it relatively easy to measure them accurately, but most tracks found will show some degree of spread as they fade. Edges become sloped and poorly defined, making measurements difficult. As a result of these variables, similar-sized species are hard to distinguish from their tracks. Also, young animals leave tracks that resemble those of adults from a smaller species (e.g. tracks of young red duiker resemble those of adult blue duiker). Some traditional hunters have been reported to be able to distinguish the tracks of all species (Koster & Hart, 1988), but, so far, biologists have been unable to come up with objective methods to do this, and most track surveys lump forest antelopes together into two or three size categories. Species identification is less of a problem for cats since only the leopard and the golden cat are present in forests, and they are very different in size.

86 If it is possible to distinguish individual animals by their tracks, an estimate of population density can be made. Stander (1998) succeeded in doing this for leopards in a semi-arid environment, with the assistance of experienced San hunters and the advantage of being able to survey large distances from vehicles (he only found an average of one spoor each 38.1km). However, as with distinguishing ungulate species, biologists have found it very difficult to pin down an objective methodology (see also Smallwood & Fitzhugh, 1993).

Equipment ruler (marked in mm) hand-rake for tracing prints: sheets of acetate or glass and marker pens for making casts of prints: plastic drinking straws, talcum powder, plaster of Paris powder, mixing container, stirrer, paper casting frames, water, scalpel/sharp knife, and fine brush. Vinegar can also be used to make casts set faster

Site selection Site selection can be: i) Opportunistic – wherever tracks are seen. ii) Strategic – search any area where there is a damp, soft or sandy damp substrate which will take an impression of a light footprint. Ideal places include damp, sandy or dusty areas on roads and paths and sandy stream beds (in the dry season) or river banks. iii) Systematic – set up track stations at regular intervals (e.g. every 50–100m) along a transect. Clear all leaves and debris and rake the ground so that it is smooth and soft enough to take animal footprints (e.g. Wilkie & Finn, 1990).

Procedure and Recording Site features and track details for different species/mammal groups should be recorded on Form 5.4 (see White & Edwards, 2000, Chapter 10, for more detail). These measurements should be taken for up to three footprints of the same animal if possible. If unsure of the mammal group, make a sketch indicating scale and dimensions measured. Tracing onto acetate or glass sheets makes identification of individual cats from pug-marks easier (Panwar, 1979).

87 Identification There are very few field guides that give details of footprint size and dimensions for larger terrestrial mammals. Walker (1991) is a noteworthy exception, but covers mostly savannah mammals. Stuart & Stuart (1995, 1997) may be helpful for people working in East Africa, while the books by Liebenberg (2000), although restricted to southern Africa, may prove useful to anyone trying to master the art of identifying mammal or other spoor.

Fig 5.2: Footprints The basic footprint measurements (in mm) are length and width – as shown on the figures. For carnivores, length of pad and claws should be distinguished, and for larger carnivores the length and width of the large heel should be measured (note if claws included). Length and width of toes can also be taken. Where possible, tracks of captive animals should be measured (e.g. in zoos) to get a clear idea of footprint shapes and sizes. One study of bottom: leopard & blue duiker, top: bushbuck antelopes in the West African & civet.

Table 5.1: Footprint lengths of species in West African forests

Max length (mm) Species

80–100 Bongo 50–75 Large duikers (yellow-backed, Jentink’s, Abbot’s) 18–45 Bushbuck (30–45mm), medium/small, duikers, chevrotain

Below 18 Dwarf antelopes (Neotragus spp.)

88 forest zone (Newing, 1990), distinguished four, clear footprint size classes (Table 5.1). In addition to the species listed below, tracks of the sitatunga are distinctive because of their very long, splayed hooves.

Taking permanent records of tracks When unusual tracks are found and cannot be identified immediately, it is worth taking a record for identification back at base camp for future work. This can consist of a photograph, a tracing of a footprint, and/or a plaster of Paris cast. The first step is to carefully clear any obvious debris obscuring the outline of the print with tweezers, making sure the edges of the print are not altered. Photographing prints: i) Place a ruler next to the print for reference and photograph. ii) Record the film exposure number, species name, collection date, location, identification number and collector’s name in a notebook. Tracing prints: i) Place a glass plate over the print, ensuring the plate is flat, which can be done using adjustable screws as legs in the four corners of the plate. ii) Trace the print outline, paying particular attention to its definition. iii) Record the species name, collection date, location, identification number and collector’s name, either on the corner of the plate or in a notebook if the print is then to be traced onto paper. Casting Prints (in addition to photographs): i) Using the straw, blow talcum powder over the print to prevent soil particles from sticking to the cast. ii) Place a casting frame around the print (e.g. paper, cardboard, stiff plastic). iii) Prepare the plaster of Paris – add water to the powder stirring con- tinuously until the mixture has a pancake batter texture. iv) Slowly pour the mixture into the mould evenly, gently tapping to remove any air bubbles, which may distort the impression. v) Engrave an identification number onto the cast before it solidifies. vi) Record the species name, collection date, location, identification number and collector’s name in a notebook. vii) Leave casts to harden for approximately two hours (cover with plastic if rain is imminent). viii) Remove casting frame and trim excessive edges, leaving a 10mm border around the print.

89 ix) Brush any loose debris from cast. x) Package in tissue or bubble wrap. xi) Store in a cool dry room.

Data analysis For most species, analysis will be limited to presence/absence, or the number of track sets and other signs recorded for each species per kilometre walked. For large carnivores it may be possible to carry out multivariate analyses to identify individuals (e.g. Smallwood & Fitzhugh, 1993) if there are many measurements.

C. Photo-recording Photo-recording is expensive and requires patience to overcome mal- functions of cameras, but it can be invaluable in recording the presence of hard-to-detect species. Setting cameras with automatic trigger mechanisms allows low-labour monitoring of natural attractions (e.g. salt licks), baited sites or commonly used thoroughfares. Camera-trapping can also be used to determine activity patterns (nocturnal, diurnal, crepuscular), reactions to disturbance (e.g. Griffiths & van Schaik, 1993), seasonal movements and breeding patterns, and social structure. If enough cameras are used, it can also provide some information on abundance. Seydack (1984) gives a good example in South African forest, and Griffiths (1994) documents successful use in rainforest conditions for carnivores in south-east Asia. A full discussion of camera-trapping is beyond the scope of this manual. Interested readers should refer to Karanth & Nichols (1998) and Carbone et al. (2001).

Equipment Photographic data-recording units, comprising: a) a camera with auto- winder (or Polaroid camera), enclosed in weather-proof casing, and mounted on a stand; b) a flash system with casing; c) a trigger device such as a trip-plate (300mm x 400mm), trip/bait wires, or a movement or heat sensor; and d) wires/other connectors between trigger and camera. Passive cameras have either a trip-plate or a movement/heat sensor. A problem with movement sensors is that most types are triggered too easily (for example, by falling leaves or fruit). However, a laser sensor is available that can be set to send a beam at different pulses, and when the beam is broken for a certain length of time it activates the camera; thus, the pulse rate can be set for a specific species. If a trip-plate is used, it is placed in a narrow place along a path. When an animal stands on it an electrical connection is made, causing the

90 camera to expose a frame when the flash goes off, and the whole system automatically reloads for the next passing animal. If bait is being used to attract carnivores, then a mechanical trigger attached to the bait will set off the film and flash when tugged.

Site selection The camera needs to be facing a path, track or road along which mammals commonly walk. A suitable position can be determined by finding animal trails crossing paths and by the presence of dung, scrapes and tracks. It is also important to find places where the track is narrow, and animals pass near the camera. A set of camera lines can be established to cover the whole survey area, or in the centre of several sampling units (e.g. Seydack, 1984). A suitable alternative is to use a baiting station and set the camera so that it will record any animals that come to the station. This would be preferable in cases where trails get too much human activity, or to sample small carnivores and other animals that do not habitually travel along roads/trails.

Procedure and Recording i) Each camera is positioned at a strategic point and can be systematically rotated to maximise the area sampled. If a single camera is used, it can be moved to different sites at intervals of a few days. Also record weather. ii) Some trials are necessary to get the best pictures. Film speed, shutter speed, aperture, distance from the plate, etc. all need to be adjusted to suit local conditions (e.g. Seydack, 1984). iii) The majority of camera traps automatically record time and date upon activation. If not, a board giving the camera number, location and date can be placed within the field of the photograph, but without interfering with the subjects.

Data analysis i) The photographs are developed and the species identified, along with age and sex where possible. For rare species this gives good information on presence, and sometimes information on population structure. However, be prepared that the majority of photos will be of empty trails or individuals of a large group of a single species, such as mangabeys. A lot of film will be required to obtain a few photos of rare species. ii) For species where individuals can be distinguished by markings (such as coat patterns), photos from camera lines or cameras placed in sampling blocks can be used to count individuals, describe ranging patterns, and calculate population densities. Plotting the number of new individuals caught on film against the cumulative photographic effort will give an indication of when 91 the majority of individuals have been photographed in the sample area, and will enable you to use mark-recapture analysis techniques to estimate densities.

5.4 Conclusions All observations of animals or their signs should be recorded to build up a broad picture of their distribution and abundance. For most species this is all that rapid surveys can accomplish, especially where relatively few animals or their signs are encountered. However, relative abundance can be determined more accurately for elephants (with dung transect surveys) and duikers (with a mixture of day and night transects and net drives). Long-term studies are essential for reliable population density estimates.

5.5 References

Archer, A.L. (1994). A survey of hunting techniques and the results thereof on two species of duiker and the suni on Zanzibar Island. Unpublished report to Zanzibar Forestry Development Project.

Barnes, R.F.W. (1988). A Short-cut Method for Obtaining Preliminary Estimates of Elephant Abundance in Forests. IUCN/WCMC, Cambridge, UK. 9pp.

Barnes, R.F.W. (2001). How reliable are dung counts for estimating elephant numbers? Afr. J. Ecol. 39: 1–9.

Barnes, R.F.W. & Jensen, K.L. (1987). How to Count Elephants in Forests. IUCN African Elephant and Rhino Specialist Group Technical Bulletin 1: 1–6. IUCN, Gland, Switzerland.

Barnes, R.F.W., Asamoah-Boateng, B., Naada Majam, J. & Agyei-Ohemeng, J. (1997). Rainfall and the population dynamics of elephant dung-piles in the forests of southern Ghana. Afr. J. Ecol. 35: 39–52.

Bell, R.H.V. (1984). The man–animal interface: an assessment of crop damage and wildlife control. In: Conservation and Wildlife Management in Africa, pp. 387–416. (Eds. R.H.V. Bell & E. McShane- Caluzi. US Peace Corps Training and Program Support.

Blum, L.G. & Escherich, P.C. (Eds.) (1979). Bobcat Research Conference Proceedings. National Wildlife Federation Technical Report Series 6.

Bowland, A.E. (1990). The response of red duikers Cephalophus natalensis to drive counts. Koedoe 33(1): 47–53.

Buckland, S.T., Anderson, D.R., Burnham, K.P. & Laake, J.L. (1993). Distance Sampling: Estimating Abundance of Biological Populations. Chapman & Hall, London, UK.

Carbone, C., Christie, S., Conforti, K., Coulson, T., Franklin, N., Ginsberg, J.R., Griffiths, M., Holden, J., Kawanishi, K., Kinnaird, M., Laidlaw, R., Lynam, A., Macdonald, D.W., Martyr, D., McDougal, C., Nath, L., O’Brien, T., Seidensticker, J., Smith, D.J.L., Sunquist, M., Tilson, R. & Wan Shahruddin, W.N. (2001). The use of photographic rates to estimate densities of tigers and other cryptic mammals. Anim. Cons. 4: 75–79.

92 Dorst, J. & Dandelot, P. (1983). A Field Guide to the Larger Mammals of Africa. Collins, London, UK.

Dubost, G. (1984). Comparison of diets of frugivorous ruminants of Gabon. J. Mammol. 65(2): 298–316.

Estes, R.D. (1991). The Behavior Guide to African Mammals: including Hoofed Mammals, Carnivores, Primates. University of California Press, Berkeley and Los Angeles, California, USA.

Feer, F. (1989). Comparaison des régimes alimentaires de Cephalophus callipygus et C.dorsalis, bovidés sympatriques de la forêt sempervirente africaine. Mammalia 53(4): 563–604.

Griffiths, M. (1994). Population density of Sumatran tigers in Gunung Leuser National Park. In: Sumatran Tiger population and Habitat Viability Analysis Report. (Eds. R. Tilson, K. Soemarna, W. Ramono, S. Lusli, K. Traylor-Holzer & U. Seal). Indonesian Directorate of Forest Protection and Nature Conservation and IUCN/SSC Conservation Breeding Specialist Group. Apple Valley, Minnesota, USA.

Griffiths, M. & van Schaik, C.P. (1993). The impact of human traffic on the abundance and activity periods of Sumatran rain forest wildlife. Conserv. Biol. 7: 623–626.

Haltenorth, T. & Diller, H. (1984). A Field Guide to the Mammals of Africa, Including Madagascar. Collins, London, UK.

Hill, C.M. (1998). Conflicting attitudes towards elephants around the Budongo Forest Reserve, Uganda. Environm. Conserv. 25: 244–250.

Hill, C.M. (2000). Conflict of interest between people and baboons: crop raiding in Uganda. Int. J. Primatol. 21(2): 299–315.

Karanth, K.U. & Nichols, J.D. (1998). Estimation of tiger densities in India using photographic captures and recaptures. Ecology 79: 2852–2862.

Kingdon, J. (1997). The Kingdon Field Guide to African Mammals. Academic Press, London, UK.

Koster, S.H. & Hart, J.A. (1988). Methods of estimating ungulate populations in tropical forests. Afr. J. Ecol. 26: 117–126.

Lahm, S.A., Barnes, R.F.W., Beardsley, K. & Cervinka, P. (1998). A method for censusing the greater white-nosed monkey in north-eastern Gabon using the population density gradient in relation to roads. J. Trop. Ecol. 14: 629–643.

Liebenberg, L. (2000). Photographic Guide to Tracks and Tracking in Southern Africa. New Holland, South Africa.

Muchaal, P.K & Ngandjui, G. (1999). Impact of village hunting on wildlife populations in the western Dja Reserve, Cameroon. Conserv. Biol. 13: 385–396.

Naughton, L., Rose, R. & Treves, A. (1999). The social dimensions of human–elephant conflict in Africa: A literature review and case studies from Uganda and Cameroon. Report to IUCN African Elephant Specialist Group, Human Elephant Task Force. Gland, Switzerland.

Nchanji, A.C. & Plumptre, A.J. (2001). Seasonality in elephant dung decay and implications for censusing and population monitoring in south-western Cameroon. Afr. J. Ecol. 39: 24–32.

Newing, H.S. (1990). Distinguishing antelope dung and tracks - a zoo study of Upper Guinean forest species, West Africa. Unpublished report. 93 Newing, H.S. (1994). Behavioural ecology of duikers (Cephalophus spp.) in forest and secondary growth, Taï, Côte d’Ivoire. PhD Thesis, University of Stirling, Scotland.

Noss, A.J. (1999). Censusing rainforest game species with communal net hunts. Afr. J. Ecol. 37(1): 1–11.

Nummelin, M. (1990). Relative habitat use of duikers, bush pigs, and elephants in virgin and selec- tively logged areas of the Kibale Forest, Uganda. Trop. Zool. 3: 111–120.

Panwar, H.S. (1979). A note on tiger census technique based on pugmark tracings. Tigerpaper 6: 16–18.

Peres, C. (1999). General guidelines for standardising line transect surveys of tropical forest primates. Neotropical Primates 7(1): 11–16.

Plumptre, A.J. (2000). Monitoring mammal populations with line transect techniques in African forests. J. Appl. Ecol. 37: 356–368.

Plumptre, A.J. & Harris, S. (1995). Estimating the biomass of large mammalian herbivores in a tropical montane forest: a method of faecal counting that avoids assuming a steady state system. J. Appl. Ecol. 32: 111–120.

Rosevear, D.R. (1974). The Carnivores of West Africa. Trustees of the British Museum (Nat. Hist.), London, UK.

Seydack, A.H.W. (1984). Application of a photo-recording device in the census of larger rain-forest mammals. S. Afr. J. Wildl. Res. 14: 10–14.

Smallwood, K.S. & Fitzhugh, E.L. (1993). A rigorous technique for identifying individual mountain lions Felis concolor by their tracks. Biol. Cons. 65(1): 51–59.

Stander, P.E. (1998). Spoor counts as indices of large carnivore populations: the relationship between spoor frequency, sampling effort and true density. J. Appl. Ecol. 35: 378–385.

Stuart, C. & Stuart, T. (1995). Southern, Central and East African Mammals. Struik Publishers, Cape Town, South Africa.

Stuart, C. & Stuart, T. (1997). Field Guide to the Larger Mammals of Africa. Struik Publishers, Cape Town, South Africa.

Walker, C. (1991). Signs of the Wild: Field Guide to the Spoor and Signs of the Mammals of Southern Africa. Struik, Cape Town, South Africa.

Walsh, P.D. & White, L.J.T. (1999). What it will take to monitor forest elephant populations. Conserv. Biol. 13(5): 1194–1202.

White, L.J.T. (1995). Factors affecting the duration of elephant dung piles in rain forest in the Lope Reserve, Gabon. Afr. J. Ecol. 33: 142–150.

White, L. & Edwards, A. (Eds.) (2000). Conservation Research in the African Rain Forests: A Technical Handbook. Wildlife Conservation Society, New York, USA.

Wilkie, D.S. & Finn, J.T. (1990). Slash-burn cultivation and mammal abundance in the Ituri Forest, Zaïre. Biotropica 22(1): 90–99.

Wilson, V. (1990). Duiker Survey in Central African Republic. Chipangali Wildlife Trust, Zimbabwe.

Wing, L.D. & Buss, I.O. (1970). Elephants and forests. Wildl. Monogr. 19: 1–92.

94 uvyr drs:Dt:Field sheet ref: observation Date: Altitude: Time at end of watch: Vegetation: UTM (if available): End time of Additional observations (behaviour, food eaten etc.) Time at start of watch: No. of animals Observation point: Longitude: Species of obs. Address: Start time Other: Weather: Latitude: Survey site: (total observers) Surveyor: Points Observation from Sightings for Sheet Recording 5.1: Form Female adults inf Male/ Sub Juv/ (dd/mm/yy)

95 st detected (in metres) viour, association with other viour, (dd/mm/yy) adult inf Male/ Sub- Juv/ Female alongtransect tation distance dicular spread H (heard) or S (seen) Perpendicular distance = from nearest point on transect to position at which animal was fir – Cue Form 5.2: Recording Sheet for Sighting Transects Surveyor:(total observers) Address: Date: Field sheet ref: Survey site:Latitude: length:Transect Other: Time Distance Vege- Longitude: Species Vegetation: Start time: No. of animals Cue UTM (if available): Perpen- Group Weather: Additional observations Altitude: End time: Group spread = of group animals a single species, recorded in sighting (in metres) Observations beha species; any other comments

96 uvyr drs:Dt:Field sheet ref: Date: End time: tion and max. width length (mm) for tracks) Altitude: foot (adapted from Barnes & Jensen, 1987) Make measurements of the diameter intact elephant boli (in cm) Weather: C2 = < 50% boli intact; D formless, flat mass; E Decayed to a stage that it cannot be detected at range of two metres, an transect (cm) Notes: Distance from start = of transect; T/D = tracks/dung; Bolus state: A & diam. UTM (if available): Vegetation type and general observations (including descrip- Distance from Bolus. state Start time: T/D Vegetation: Longitude: (km) Address: Species Distance Other: Transect length: Latitude: Survey site: (total observers) Surveyor: Transects Dung for Sheet Recording 5.3: Form – fresh, whole, moist and smelly; B = whole odourless; C1 > 50% boli intact; d would not be seen on a transect unless under- (dd/mm/yy)

97 Form 5.4: Recording Sheet for Tracks

Surveyor: Date: Field sheet ref: (total observers): (dd/mm/yy)

Address:

Survey site: Vegetation: Weather:

Latitude: Longitude: Altitude:

Transect length: Start time: End time:

Other: Location Species Soil Soil Vegeta- Clarity of Measurements (transect type mois- tion type print (mm) marker ture number)

Soil type: S = mostly sandy; C = mostly clay; St = stones; Si = mostly silt. Use combinations where necessary, e.g. C+St = mostly clay + stones Soil moisture: dry; damp; wet Clarity of print: distinct = well-defined with clear edges; fair = mostly well-defined but some edges ‘spread’ or confused by other tracks or debris; indistinct = clear enough for identification but measurement difficult because of spread, other tracks or debris.

98 6. Primates Glyn Davies

6.1 Biology

African primates are divided into three taxonomic groups (Oates, 1996): the small, nocturnal prosimians (20+ species); the monkeys (45+ species); and the apes (3+ species). Like humans, non- human primates that are active in the daytime generally have a poor sense of smell, moderate hearing and excellent eyesight (nocturnal species are obviously very different, having weak distance-vision and acute hearing). Savannah species (e.g. baboons) and the lemurs of Madagascar are not discussed in this chapter, but the for- Cercopithecus mona) est survey procedures still apply to these species

where they do mona monkey ( occur in forests.

Prosimians These primitive primates are separated into two families: the galagos (Galagonidae) and the lorisids (Loridae). All species are primarily nocturnal (Charles-Dominique, 1977). Forest galagos are generally small (18+ species; 50–300g), although three species exceed 1,500g in weight. They live in small family units, travelling and foraging in the understorey and middle canopy of

99 the forest, and make distinctive calls between group members to maintain contact. Loud calls are made when alarmed, and to signal to other groups/individuals. Recent taxonomic work has shown how these calls are important in distinguishing different species (Bearder et al., 1995). The potto (genus Perodicticus) and two angwantibo species (genus Arctocebus) are larger (400g–1,500g), mostly solitary, and move slowly along branches or through liana tangles, often in the middle and upper canopies of the forest. Angwantibos will also use shrubby undergrowth in clearings. When not in direct contact, they communicate by scent-marking twigs and branches, and do not make loud calls.

Guenons Forest guenons (genera Cercopithecus, Miopithecus and Allenopithecus) are relatively small-bodied monkeys (1–8+kg), which live in groups of about 10–30 individuals occupying stable home ranges that generally cover 20–100ha (Gautier-Hion et al., 1988). All species can be distinguished on the basis of their species-specific facial colour patterns, and the loud calls given by the adult males. Calls are made at various times during the day, especially early morning and late afternoon, and when one group calls this often sets off a sequence of calls and replies between neighbouring groups. Many guenons have species’ ranges with a wide altitudinal distribution and habitat use. Exceptions to this generalisation are: the l’Hoest/Preuss species group, most of which do not occur in the lowlands (excluding the sun- tailed monkey, C. solatus, in the lowlands of Gabon), and species which concentrate in riverine/swamp forest environments (e.g. Allen’s swamp monkey, talapoins), or in forest-edge vegetation (e.g. Sykes monkeys). Most breeding groups have a single adult male, several adult females and young. Young adult males tend to leave the group of their birth and travel substantial distances, sometimes in the company of other adult males, sometimes as lone males.

Colobus monkeys There are three taxonomic groups of colobus monkeys: olive (one species), red (14+ species/subspecies) and black-and-white (five species) (Davies & Oates, 1997). The black-and-white colobus species (9.5–13kg) generally live in small groups (c. 5–15) and occupy small home ranges (20–50ha). Each group has one, or sometimes two, adult males that engage in daily choruses of territorial loud calls, often pre-dawn and in the early morning. Despite this noisy behaviour, however, they tend to remain silent and hide when alarmed by humans. Two exceptions to this ecological model are Angolan pied colobus (Colobus

100 angolensis) in Ituri (Zaire) and Nyungwe (Rwanda) and Black colobus (C. satanus) in Forêt des Abeilles (Gabon), which live in much larger groups (sometimes hundreds of animals) in larger home ranges (not territories). Red colobus (8.5–10kg) adult males lack loud calls, but can be detected from their occasional noisy chatter and alarm calls on seeing humans. Red colobus (Procolobus spp) live in groups of 10–50 individuals (sometimes even larger) occupying home ranges of 50–100ha that are not defended as territories; there is considerable overlap between neighbouring groups. The diminu- tive olive colobus monkeys, Procolobus verus, (4kg) are very secretive, live in small groups with one or sometimes two adult males, have a quiet, shrill call, and often travel in the company of other monkey groups. They are difficult to detect.

Mangabeys, mandrills and drills The mangabeys (females: 6kg; males: 10kg), and drills/mandrills (10–20kg) are large-bodied and live in large groups (drills up to 100; mangabeys: 15–30+) that travel rapidly over wide areas of forest, often splitting into smaller groups when foraging. Groups contain several adult males that make loud calls which may be audible from distances greater than 1km, as well as the noisy squabbles within groups which indicate a group’s presence within 100m or so. In rare instances, mandrills, Mandrillus sphinx, have been recorded forming hordes of over 500 animals, reflecting a dynamic social system that can be maintained in large tracts of forest. Most time is spent travelling and foraging on the forest floor, often noisily searching leaf-litter for insects and fallen fruits, but they also feed and sleep in large trees. Grey-cheeked mangabeys, Lophocebus albigena, are exceptional in being largely arboreal throughout the day.

Apes There are four ape species: chimpanzee, Pan troglodytes, (30–40kg), gracile (or bonobo or pygmy) chimpanzee, Pan paniscus (25–35kg), and two species of gorilla (90–200kg), the western lowland gorilla, Gorilla gorilla, and the mountain gorilla, G. beringei; all are large-bodied and travel long distances along the ground (McGrew et al., 1998). Chimpanzees live in fission-fusion communities, so group sizes vary from small units comprising a mother and offspring to large congregations in excess of 20 individuals around large food sources (e.g. fruiting fig trees). They are very vocal, making noisy displays with hoots, shouts and drumming on tree buttresses, all of which aid detection during surveys. Bonobos have a society that is more cohesive and influenced by strong female-female bonds. They are restricted to low-lying forest formations in the Congo Basin south of the Congo

101 River. Gorillas live in stable family groups of 5–30 animals, with a single, or sometimes two fully adult males. They are less noisy than chimps, but males do have clearly audible chest-beating displays. All four ape species leave hindfoot and front knuckle prints when they travel over damp/soft ground, make nests to sleep in each night (chimps and bonobos in trees but sometimes on the ground, gorillas in the understorey and middlestorey and on the ground), and produce large, long-lasting characteristi- cally shaped faeces at sleeping and feeding sites. These signs are commonly used during surveys, supplemented with evidence from areas damaged during feeding and play, and records of sightings, loud calls, tree drumming, and chest-beating.

6.2 Management issues Primates are important components of forest ecosystems: gorillas can have a major impact on plant regeneration; guenons pollinate flowers and dis- perse seeds; colobus monkeys commonly destroy seeds; and the combined numbers of all primates account for the bulk of medium-sized mammalian biomass in many forests. Their unseen and often unrecorded influence on ecosystem function is an important consideration for forest managers. Moreover, the loss of forest habitat is a key reason for primate species extinctions (e.g. Cowlishaw, 1999; Oates et al., 2000). Probably the most important use of forest primates is exploitative: hunting and trapping for food – or bushmeat. Bushmeat is important in rural communities both in terms of subsistence consumption, providing much needed animal protein, and for trading activities. In West and Central Africa, where economic hardship has resulted in greater dependence on forest products, there has been a marked increase in commercial exploitation of primates in recent years. Large tracts of forest have been heavily hunted to supply primates, and other species, to ever demanding urban centres. Hunting and trapping of primates is less common in the drier southern and eastern African regions where cattle do relatively well and consumption of primate meat is often disdained on religious or cultural grounds; however, it does still occur. Larger species tend to be the main target of the bushmeat trade, because these animals supply more food for each cartridge spent, and traditional taboos on hunting apes are increasingly being ignored, with the result that these large-bodied animals are now used in trade. Smaller species are taken when encountered, and when larger species have been eliminated. In addition to hunting for food, some primates are consumed for other reasons that can have significant localised impacts. These include: providing ceremonial skins (e.g. black-and-white colobus); selling trinkets for the tourist trade (e.g.

102 gorilla hands); supplying animals for the pet trade; supplying biomedical research centres (e.g. chimpanzees). All these activities obviously lead to declines in the target species. The size of forest primates is a less useful factor for predicting timber extraction impacts. For all forest species, there is a risk that as forest habitat is altered it will become less suitable, thereby eliminating the least adaptable species first. Gathering of forest products has long affected forests throughout Africa, and primates have continued to survive, but the scale and rate of recent commercial timber extraction has radically altered forest quality. A recent review of the effects of logging on primates (Plumptre & Johns, 2001) indicates that very few African primates are actually lost as a result of habitat changes following logging, with red colobus at some sites being an exception (e.g. Skorupa, 1986). However, long-term effects of vegetation changes are hard to predict (e.g. Chapman et al., 2000), and improved access for hunters and trappers along logging roads is a major secondary impact of logging. A common problem that arises at the forest-farm boundary is that of crop raiding by primates which refuge in the forest; this is a major management issue. Guenons and mangabeys are common pests of cash and food crops, and even galagos are considered as pests of cashew nuts on the Kenya coast. Chimpanzees and gorillas are keen on bananas and plantains and can do substantial damage in a short time to these flimsy plants. The consequence is that farmers kill the primates in order to protect their crops, which has resulted in losses of even protected species such as gorillas on the periphery of the Bwindi Impenetrable National Park in Uganda. A positive attribute of primates is that tourists find them attractive and interesting. In areas with a large number of primate species, ideally at high population densities, there is ample scope for ecotourism development, even where thick forest vegetation hampers viewing. In Uganda, this has been developed for the two largest, and most tourist-attractive, species: the chimpanzee and the gorilla. In all cases, the animals have been habituated to human visits, and this usually takes a long time. There is an associated risk that animals that lose their wariness of humans become easy targets for hunters, or become pests around farms, villages and tourist hotels. They also become vulnerable to contracting human diseases from visiting tourists (Butynski & Kalina, 1998). Ecotourism and the maintenance of primate populations are obvious complementary activities. In this context, the IUCN conservation action plan for African primates (Oates, 1996) makes the point that dry forest and savannah zone primates in Africa have wide distributions and occur in many protected areas, whereas the populations of the majority of primates from the six lowland forest and four upland/highland forest communities are under much greater

103 threat. Surveys of the distribution and abundance of primates are therefore needed to guide conservation planning and action, as well as obtain facts to inform forest management.

6.3Methods

General Survey methods for primates have been thoroughly reviewed in a 1981 publication by a subcommittee of the US National Research Council (NRC, 1981) on the Conservation of Natural Populations, to which reference should be made. The interpretation of survey field data relies on an understanding of primates’ reproductive biology, ecology and behaviour, which has been collected from long-term studies of primates. Examples of long-term study sites in Africa include: Kibale and Budongo in Uganda; Tana River and Kakamega in Kenya; Gombe and Mahale Mountains in Tanzania; Virunga Volcanoes in Rwanda; Tiwai Island in Sierra Leone; Taï in Côte d’Ivoire; and Makokou and Lopé in Gabon. Information from these and other sites should be consulted before embarking on a survey. Selecting a survey method involves a compromise between time/personnel available and the information needed for management planning. It is therefore important to write down clearly what questions the survey is to answer, and then decide if there are sufficient resources to gather this information. Before making a final decision on which methods to choose, a quick walk through an area is important to make a general assessment of the populations to be surveyed, and identify future transect survey sites.

Identification The use of field guides, such as those discussed in section 5.3 for large mammals, should be supplemented with information on the colour of animals, based on examination of skin collections in museums, on live specimens in zoos, and on an appreciation of the vocal repertoire of different species. There are few primate call sound libraries, so it is helpful to visit an existing study area where primate calls can be learned. Such visits also provide an opportunity to make inquiries about local names and develop a general understanding about species’ ecology, abundance and interaction with human populations. The survey team’s ability to identify primate species correctly is central to obtaining useful results; time spent in training will pay good dividends during surveys.

104 6.3.1 Distribution surveys One of the first things that should be determined is what primate species occur in the area. This can be done by checking a range of sources, including records from the literature and museum specimens, and by spending time with local farmers and hunters, as well as with forest, wildlife, national parks and other government officers. The interview information needs to be verified carefully (see section 2.6), and then recorded onto standard checklists (Form 6.1), which can then be computerised and marked onto maps. The information on distribution maps, which show where different species occur, or have occurred in the past, can be presented on a range of scales from local to global. To facilitate transposing information from one scale to the next, it is important to use a standard mapping unit from the outset, such as degrees (or quarter degrees) of latitude and longitude or the UTM global metric grid. This allows countrywide maps to be drawn, onto which changes in species distribution can be mapped over time (see Fig. 6.1). At the local level, an accurate (1: 50 000 scale) topographical map is needed, supported with recent aerial or satellite photographs if available. These should be used to assess information on vegetation and altitude, and proximity of villages, roads, urban centres, etc. Thereafter, visits can be made to different areas, and by mapping all reliable sightings and other records a preliminary assessment can be made of which areas are suitable for primates. This information is very important to guide forest management, but it is insuffi- cient to estimate population densities.

Fig. 6.1: Distribution of four primate species in Sierra Leone (Grubb et al., 1998)

105 6.3.2 Line transects Line transect methods for surveying vertebrate populations (review in Buckland et al., 1993) have been adapted for forest primate surveys (see reviews: NRC, 1981; Whitesides et al., 1988; Peres, 1999; Plumptre, 2000). The main aim of these surveys is to determine the population density of primates in the area and to ascertain what factors play a role in affecting their numbers.

Equipment/personnel The details of general survey equipment, personnel and site selection have been elaborated in the previous chapter (section 5.3.3). The same principles should be applied for primates, although direct surveys of mobile, group- dwelling species present their own problems that will be discussed in the section on data analysis.

Site selection i) On an accurate map, select the places where transects should begin, and the direction in which they should run. This can be done using random number tables to select transect start points and bearings. Alternatively, transects can be laid out to sample particular areas of interest on a stratified sampling basis depending on the area and spread of forest habitats. Transects can be cut to point in different directions and should be at least 1km apart at all points, to gain an idea of population densities over a wide (and therefore representative) area (see section 5.3.3. for further discussion). Straight line transects should be cut through the forest, following a compass bearing. ii) Local hunters and guides can help in locating suitable campsites and transect start points. As noted for ungulates, transects should not begin too close to the camp noises and smells (over 300m away). iii) The use of established roads, tracks and paths greatly reduces the time spent in establishing a survey transect. However, this tends to introduce serious bias to the survey, because: a) the roadside vegetation differs signifi- cantly from the rest of the forest, and/or b) primates change their behaviour in relation to the road, track or path (e.g. if hunters commonly use them). iv)Once the transects have been cut, ensure that the survey area has a distinctive name, and that each transect is clearly distinguished from the others by a unique number. After cutting the transect, wait at least 24 hours before proceeding with the survey so that primates can recover from the disturbance to the area.

106 A. Animal sightings Procedure i) Early in the morning (soon after dawn) or in the afternoon (after about 15:00), a surveyor (or small group of surveyors) should walk slowly and quietly along the transect. For example, Butynski (1984) walked at less than 1km/hr, stopping every 60m for 30–60 seconds to look around and listen, and others (e.g. Peres, 1999) have described similar speeds of 1.25km/hr. ii) Transects should not be crossed by other survey teams during the survey period, since this will disturb groups which may leave the area to be surveyed. iii) The ability to detect primates may be affected by the weather: some species call less on windy days and all are more difficult to see in windy and rainy conditions. As a result, it is important to avoid surveys during rain, and immediately afterwards, when the sound of dripping water obscures other sounds. iv)Other factors affecting survey results include: a) time of day – primates tend to be more active in the early morning and late afternoon; b) human activities – primates call less and are more wary in areas where there is hunting; and c) the experience of the observer. These parameters all need to be recorded (see top section of Form 6.1).

Recording i) At the first encounter with a primate, the means of ‘detection’ (sighting, branch movement, falling fruit, alarm calls, fleeing animals, etc.) should be noted immediately, and the recording sheet filled out as fully as possible (Form 6.1). It is very important to make accurate measurements of the distance from the transect to all individuals that are seen (see Fig. 6.2). The perpendicular distance from the transect to each primate should be determined by direct measurement, using a tape measure, marked (non-stretch ropes), or optical (or laser) range-finder. As mentioned in section 2.5, estimating distance by eye is fraught with inaccuracy and variability between observers, and, over time, the calculation of perpendicular distance from the observer to animal and the angle of sighting introduces two potential sources of error. ii) It is often useful to put a field bag on the transect at the point from which the first animal was seen, as a reference point from which other measurements can be made. The places where animals were first seen need to be clearly recognised, and the transect marker nearest the observer should be recorded.

107 Fig. 6.2: Line transect Transect iii) Information should be gath- route ered about the primate group itself, including: how many different individuals were seen (adult males, females, monkey infants, any special marks like bent group tails, etc.); estimated group size

perpendicular (including those heard but not seen); distance the area over which sighted animals are spread, as well as the overall group spread. Time should be allowed observer/animal to move up and down the transect, distance and left and right, to gather this infor- sighting mation, although it is important not to angle move away from the transect for more than 10 minutes. iv) Other behavioural details to record include: what the group was doing when encountered (if feeding, what were the animals feeding on); the group’s reaction upon seeing the observer (e.g. wary curiosity, panicked flight, indiffer- ence, etc.); at what height, and in what type of vegetation, they were found.

Data analysis To calculate the population density in groups/km2, three sets of data are required: i) The length of transect surveyed, which is determined by multiplying the length of the transect by the number of times it was surveyed. ii) The number of groups encountered, which should include all those groups containing both males and females, with a separate note of groups which appeared to contain only adult males; solitary animals should also be noted separately. Using these two pieces of information the relative abundance of a species (e.g. groups per kilometre walked) can be calculated. This was done in Kibale, where Skorupa (1986) provided statistically robust data that showed differences in primate numbers between logged and unlogged forests. But this can only be done if it is possible to determine differences in the probability of detecting primates at either site (Skorupa, 1987). iii) The width of the survey strip multiplied by the distance surveyed will give an area of forest in which a certain number of primate groups were recorded; this, in turn, provides an indication of population density (groups/unit area). However, the detectability of groups on either side of the transect line is not fixed, i.e. it is not possible to say that all groups within 40m on either side of the transect will be seen, and all those more than 40m away will not be seen. 108 The transect width is influenced by: a) the detectability of the species (e.g. shy versus conspicuous behaviour); b) vegetation type; c) terrain; and d) the spread of a group (Whitesides et al., 1988). All these parameters can vary between sites, seasons and species, and to be able to estimate the transect width it is necessary to gather enough information on species-specific and habitat-specific observer–animal sighting distances (preferably 40–100 independent field records for each species at each survey site). With these data, the effective strip width for sighted individuals can be calculated using the DISTANCE programme (Buckland et al., 1993; see section 5.3.3). A complicating factor in judging the width of a transect is the group spread (the area occupied by a group of monkeys). This will influence how many monkeys you see at a given group encounter, since the presence of one group affects the locality of other groups, as well as the concentration of monkeys within an area (Fig. 6.3). Some surveyors have tried to address this problem by estimating average group spread for each species (from long-term studies), and adjusting transect width estimates accordingly to the anticipated centre of the group from the transect (Whitesides et al., 1988). However, others advise against this approach, emphasising that it is necessary to use only sighting data for all individuals seen (Plumptre, 2000), and calculating the centre of the group spread from the centre of the sightings.

Fig. 6.3

The cumulative number of sightings of groups of two species of primates: spot-nosed monkey [s], Cercopithecus petaurista, 100 and Diana monkey [d], Cercopithecus diana, with 80 increasing distance on either side of the transect. 60 The two triangles show the distance from the transect 40 within which 80% of all group sightings were made 20 (adapted from Whitesides s d et al., 1988). cumulative number of group sightings (%) 10 20 30 40 50 60 70 80 90 100

perpendicular distance from the transect (m)

109 Advantages/limitations i) Given the poor visibility in forest conditions, and the short time over which surveys are conducted, obtaining detailed estimates of population densities is very difficult. However, clear estimates of the relative abundance of primates at different sites is often adequate for management purposes (for example, knowing that there are more primates of species X in forest A than in forest B). Certainly it can give a clear indication of areas that need special management attention. ii) The use of DISTANCE software to analyse survey results is based on a number of basic assumptions (see section 5.3.3). Some of these factors are outside the control of the observer, but the quality of field records is greatly improved if time is taken to develop an understanding of the primates being surveyed, and to do practice surveys (Peres, 1999). iii) The number of times that surveys must be repeated depends upon the length of the transect and the number of encounters per kilometre. As a rough guide, 100 records were determined as a minimum sample size for robust statistical analysis in Budongo forest (Uganda), which required over 200 km of surveys (Plumptre, 2000). Smaller sample sizes can be used, but statistical confidence limits associated with density estimates need to be shown clearly. iv)The line transect survey method can be used by a single observer (or small survey team), which is a major advantage when manpower constraints limit survey work. However, the observer needs to be experienced with species behaviour for this method to be effective.

B. Nest counts The nests that are made by gorillas and chimpanzees, both for sleeping at night and also for resting during the day, persist on the ground or in trees well after they were first made. These conspicuous signs of species’ presence, which generally occur at low population densities, have long been used in surveys (e.g. Tutin & Fernandez, 1983). The same principles described for indirect surveys of ungulate signs along transects and recce tracks can be applied (section 5.3.4), and are described in more detail by White & Edwards (2000) – using Form 6.2. As with dung surveys, the critical information needed to translate counts of nests/km surveyed are: the rate at which nests are made/individual (taking account of age differences), and the rate at which nests disintegrate until they are hardly discernible any longer (which will vary with season, altitude, species and so on). At sites where there are many nests, the need to calculate decay rates can be avoided by doing repeat surveys (e.g. every one to three months) and marking all nests seen on a map (or with flagging tape) – see

110 section 5.3.4A. It is then possible to record the number of new nests produced in a given period of time, within a known survey area (km2), and then to convert a nest density figure into an estimate of individuals/km2 using a calibration for the number of nests made per day by different types of animals (males, females, etc). A number of studies have been carried out to look at these problems in different African forests, and they should be referred to for more details (Hashimoto, 1995; Tutin et al., 1995; Plumptre & Reynolds, 1997; Hall et al., 1998; Blom et al., 2001). In summary, the analysis of the results depends on the amount of background information available for the survey area during the season when the surveys were carried out. If there is good evidence that the nests are from a particular species, then species distribution maps can be drawn up. As information on nest building and decay rates is determined (by site and season), so it will become possible to give reasonable estimates of gorilla and/or chimpanzee populations’ decay rates (which are often variable).

C. Mapping calls Loud calls can be used to detect groups from greater distances than is possible with sightings. In areas where species are very vocal this provides a useful survey method, but even quieter chatters and squeals, and movements in the branches, can be used to detect primates over shorter distances. See also sections 3.3.10 and 7.3.9.

Equipment/personnel map-making equipment: 360o protractor, graph paper, ruler, pencils, etc. high-quality portable tape recorder or mini-disc player, microphone and batteries, blank tapes or mini-discs (to record unidentified and unusual calls, or to analyse calls and compare calls between areas and seasons) recorded tapes/CDs for playback experiments (to get resident groups to call) plastic bag or like for protecting equipment against moisture surveyors with good knowledge of primate calls

Procedure i) For diurnal species, the surveyor should start early (just before dawn for many species) and walk slowly (1km/hr) and wait at marked points to listen for calls. On hearing a call the surveyor should wait for between 15 and 30 minutes until the calling group, and any that are replying, seem to have

111 stopped. For nocturnal species, the start time of the survey should obviously change: galagos commonly give loud calls around dusk and during the hour or so before dawn. ii) The survey should be repeated for a number of days, depending on the species, site, season and weather conditions. For vocal species (e.g. black- and-white colobus), as little as three days may be sufficient to record all calling groups in the survey area, but longer survey periods will be needed at sites where fewer or quieter calls are made. In general, surveys need to continue until the number of groups mapped in a given area becomes consistent.

Recording On hearing a call, the following information must be collected (using Form 6.3, or a separate sheet for calls): date, time, weather, etc.; species; detection (type of call); bearing (of the call from the surveyor); observer–animal (estimated distance to calling animal); map (the trail marker from where the call was heard). Whenever the same group calls again, after the surveyor has moved to a new position, new bearings should be taken again so that the group’s position can be mapped more accurately through triangulation.

Data analysis i) An accurate map of the survey area should be drawn (useful scale is 10mm: 100m) onto graph paper, including key topographical features (e.g. streams, ridge tops, etc). Copies of the master map should then be made for mapping each species separately (Fig. 6.4). ii) Date and time are marked on a pencil line showing the bearing of the call, starting from the observer’s position on the map. The distance from observer to calling animal is estimated and also marked to scale, and special attention should be given to triangulating calls of the same group that were given at different times. iii) Great care must be taken not to double-count groups, especially for species which travel rapidly over wide areas in a short time and call from different places (e.g. mangabeys). Information on species’ home-range sizes, home- range overlap, travel and foraging patterns (gathered during long-term studies) are important in determining how to reduce this type of error. iv)Supplementary information can be added to the species’ maps, recording those animals seen, but not recorded calling, to give a fuller picture of all the groups present.

112 Fig. 6.4: Mapping calls and sightings

5/12

6/12

3/12 Map of sightings and calls of Campbell’s monkey,

3/12 Cercopithecus campbelli, during a four-day survey (3–6 December). Numbers refer to dates. Groups sighted are squared, and group calls are 6/12 4/12 shown by arrows indicating the direction from the survey- 3/12 or and the arrow-head indicates the estimated location of the group. 5/12 For species with home ranges of 20–50ha, surveys by small teams for 20 days, over 6–12 months, give good spread indications of group densities.

6/12

Survey trail with 100m markers Advantages/limitations Since most primate groups call at some time during the day, and the call can be heard over distances greater than are detectable by sight, inclusion of calling records greatly increases the sample size of encounters for a unit survey effort. For example, on three morning surveys by two observers in Kakamega forest (Kenya), there were 86 encounters with primates (sightings and calls) of which 63% were calls of animals that were not seen (Davies, pers. obs.). This method gives accurate information on the numbers of groups in a given area for those species that are vocal during the survey period. Quieter groups, species, and times of the year will all give lower quality results.

113 6.3.3 Sweep surveys In practice, field workers make the most of all their field records by combining information on sightings and calls. This can be done during transect walks, but even more information can be gained when the surveys are carried out along parallel transects on a survey grid by a number of surveyors working simultaneously. These are called sweep surveys (Whitesides et al., 1998).

Equipment/personnel survey grid cut and mapped (see below) at least three experienced field observers

Procedure i) A rectangular survey area has to be prepared, with three or more survey transects running parallel to each other, spaced at 100-m intervals, and extending for at least 1km. They should be linked at either end by perpendicular trails, to get surveyors to the start points quickly and quietly. All paths are marked at 50-m intervals. ii) Soon after dawn, the surveyors should assemble at the start of their survey trails and travel slowly and quietly along the path, beginning at a pre- arranged time (having synchronised watches beforehand). The surveyors move forwards in a single front, stopping occasionally to make field records. To keep the line of surveyors coordinated, several pre-determined restart points should be marked along the survey routes.

Recording i) When a primate is sighted, its position and the group-spread are marked on a scale map (e.g. 10mm: 50m). All primate group sightings and calls are recorded (as noted in sections 6.3.2 and this section above). ii) By noting the time of all records (sightings and calls), simultaneous records of the same group by different observers can be mapped, thereby avoiding double-counting of the same groups.

Data analysis i) At the end of the survey, all surveyors reassemble and a map is compiled for each species from all records collected. Once the map has been completed for a number of repeat surveys, separate primate home ranges can be marked, and the number of groups within the survey area and its vicinity can be counted.

114 ii) Some groups’ home ranges will fall wholly within the survey grid, but others will be partly outside. Whitesides et al. (1988) considered groups having home ranges in excess of 80% inside the survey grid as being within the area surveyed, groups having 80–30% inside the survey grid as half a group, and groups with 30% or less of their home ranges inside the grid were regarded as falling outside the survey area. By adding whole groups and half groups, the number of groups/km2 can be calculated. iii) Detailed information on the size of primate groups, and their age/sex composition, combined with information on the body weights of different types of animals (eg adult males, juveniles, females, etc.), can then be used to calculate population densities and biomass.

Advantages/limitations This method provides very reliable and accurate estimates of absolute group densities, plus solitary individuals, within a known survey area. It can be verified by repeated sweep surveys over time to take account of seasonal biases. If the sample area is representative of the forest as a whole, then the results from these surveys can be extrapolated to wider areas. The disadvantages of this method include the time taken to build up an accurate picture of group densities, the large number of surveys required, the need for a team of three or more experienced surveyors (although a team of just two can allow triangulation on calling groups), and clearing and maintenance of the transect grid.

Rapid sweeps Using the same principle as the sweep surveys, but in circumstances where there is only one surveyor, the system can be modified to a rectangular survey area (1km x 500m, depending on the species being surveyed), in which all sightings and calls of primates are recorded onto accurate maps. Over time, a picture of the number of groups in the area builds up, although the possibility of missing groups in the centre of the survey rectangle are higher than for the sweeps, and groups on the edge may also be missed. In Sierra Leone, data on relative primate abundance was gathered by a single observer using survey rectangles of 1km x 500m (e.g. Davies, 1987). Rapid sweeps have also been used in the small patches of forest along the Tana River in Kenya (Butynski & Mwangi, 1994), where pairs of observers searched in a zigzag fashion through parallel swathes of forest with each survey team’s route separated by about 100–150m. The survey routes were carefully mapped, and plenty of time taken to search for primates. At the end of the survey, all teams met and discussed results before group encounters were mapped and the numbers of groups in each patch calculated.

115 6.4 Conclusions There are a range of primate survey methods that have been adapted for particular conditions and resource constraints, many of which depend on combining records of primate calls and sightings. Selection of any one method should be determined by the question that has to be answered, and care must be taken to allocate the necessary resources required (e.g. time, personnel, transport, cash, etc). Given the problem of resource and personnel constraints, line transects are a useful and common survey method. They are practical for determining the relative density of primates at different sites as long as there are sufficient encounters to determine transect width. Sweep surveys are most useful for determining primate group density in areas of about 1km2. Whatever survey method is being used, a rule of thumb is to gather as much field information as possible, during transect walks, when in the camp, when meeting with local hunters and villagers, and so on. All of this will help improve the interpretation of field survey results.

6.5 References

Bearder, S., Honess, P.E. & Ambrose, L. (1995). Species diversity among galagos, with special reference to mate recognition. In: Creatures of the Dark: The Nocturnal Prosimians, pp 331–352. (Eds. L. Alterman, G.A. Doyle & M.K. Izard). Plenum Press, New York, USA. Blom, A., Almasi, A. & Heitkonig, I.M.A. (2001). A survey of the apes in the Dzanga-Ndoki National Park, Central African Republic: a comparison between the census and survey methods of estimating the gorilla (Gorilla gorilla gorilla) and chimpanzee (Pan troglodytes) nest group density. Afr. J. Ecol. 39(1): 98–105. Buckland, S.T., Anderson, D.R., Burnham, K.P. & Laake, J.L. (1993). Distance Sampling: Estimating Abundance of Biological Populations. Chapman & Hall, London, UK. Butynski, T.M. (1984). Ecological survey of the Impenetrable (Bwindi) Forest, Uganda, and recommendations for its conservation and management. Unpublished report to the Government of Uganda. pp. 150. Butynski, T.M. & Kalina, J. (1998). Gorilla tourism: a critical look. In: E.J. Milner-Gulland & R. Mace (eds), Conservation of Biological Resources, pp 280–300. Blackwell Science, Oxford, UK. Butynski, T.M. & Mwangi, G. (1994). Conservation status and distribution of the Tana River red colobus and crested mangabey. Unpublished report to the Kenya Wildlife Service. pp. 67. Chapman, C.A., Balcomb, S.R., Gillespie, T.R., Skorupa, J.P. & Struhsaker, T.T. (2000). Long-term effects of logging on African primate communities: a 28-year comparison from Kibale National Park, Uganda. Conserv. Biol. 14(1): 207–217. Charles-Dominique, P. (1977). Ecology and Behaviour of Nocturnal Primates: Prosimians of Equatorial West Africa. Duckworth, London, UK. Cowlishaw, G. (1999). Predicting the pattern of decline of African primate diversity: an extinction debt from historical deforestation. Conserv. Biol. 13(5): 1183–1193. 116 Davies, A.G. (1987). The Gola Forest Reserves, Sierra Leone: Wildlife Conservation and Forest Management. IUCN, Gland, Switzerland. Davies, A.G. & Oates, J.F. (1997). The Colobine Monkeys: their Ecology, Behaviour and Conservation. Cambridge University Press, Cambridge, UK. Gautier-Hion, A., Bourliere, F. & Gautier, J-P. (1988). A Primate Radiation: Evolutionary Biology of African Guenons. Cambridge University Press, Cambridge, UK. Grubb, P.A., Jones, T.S., Davies, A.G., Edberg, E., Starin, E.D., Hill, J.E. (1998). Mammals of Ghana, Sierra Leone and The Gambia. Trendrine Press, UK. Hall, J.S., White, L.T.J., Inogwabini, B.I., Ilambu, O., Morland, H.S., Williamson, E.A., Saltonstall, K., Walsh, P., Sikubabuo, C., Dumbo, B., Kaleme, P.K., Vedder, A. & Freeman, K. (1998). A survey of Grauer gorillas (Gorilla gorilla graueri) and chimpanzees (Pan troglodytes schweinfurthii) in the Kahuzi Biega National Park lowland sector and adjacent forest in eastern Congo. Int. J. Primatol. 19: 207–235. Hashimoto, C. (1995). Population census of the chimpanzees in the Kalinzu Forest, Uganda: comparison between methods with nest counts. Primates 36: 477–488. McGrew, W.C., Marchant, L.F. & Nishida, T. (1998). Great Ape Societies. Cambridge University Press, Cambridge, UK. National Research Council. (1981). Techniques for the Study of Primate Population Ecology. National Academy Press, Washington D.C., USA. Oates, J.F. (1996). African Primates: Status Survey and Conservation Action Plan. IUCN, Gland, Switzerland. Oates, J.F., Abedi-Lartey, M., McGraw, W.S., Struhsaker, T.T. & Whitesides, G.H. (2000). The possible extinction of a West African Red Colobus monkey. Conserv. Biol. 14(5): 1526–1532. Peres, C. (1999). General guidelines for standardising line-transect surveys of tropical forest primates. Neotrop. Primates 7(1): 11–16. Plumptre, A.J. (2000). Monitoring mammal populations with line transect techniques in African forests. J. Appl. Ecol. 37: 356–368. Plumptre, A.J. & Reynolds, V. (1997). Nesting behavior of chimpanzees: Implications for censuses. Int. J. Primatol. 18: 475–485. Plumptre, A.J. & Johns, A.D. (2001). Primate populations. In: Wildlife-logging Interactions in Tropical Forests. (Ed. by R.A. Fimbel, A.Grajal, & J. Robinson). Colombia University Press, New York, USA. Skorupa, J.P. (1986). Responses of rainforest primates to selective logging in Kibale forest, Uganda: a summary report. In: Primates: the Road to Self-sustaining Populations, pp 57–70. (Ed. K. Benirschke). Springer-Verlag, New York, USA. Skorupa, J.P. (1987). Do line-transect surveys systematically underestimate primate densities in logged forests? Am. J. Primatol. 13: 1–9. Tutin, C. & Fernandez, M. (1983). Recensement des gorilles et des chimpanzés du Gabon. CIRMF, Gabon. Tutin, C.E.G., Parnell, R.J., White, L.J.T. & Fernandez, M. (1995). Nest-building by lowland gorillas in the Lope-Reserve, Gabon – environmental-influences and implications for censusing. Int. J. Primatol. 16: 53–76. White, L. & Edwards, A. (2000). Conservation Research in the African Rain Forests: a Technical Handbook. Wildlife Conservation Society, New York, USA. Whitesides, G.H., Oates, J.F., Green, S.M. & Kluberdanz, R.B. (1988). Estimating primate densities from transects in a West African rain forest: a comparison of techniques. J. Anim. Ecol. 57: 345–367.

117 Form 6.1: Primate Recording Sheet for Line Transects Surveyor: Field sheet ref: Date: (total observers): (dd/mm/yy)

Address:

Survey site: Altitude: Aspect:

Latitude: Longitude: UTM (if available):

Vegetation: Human disturbance:

Season: Weather: Lunar phase: Temperature:

Transect length: Start time: End length:

Other:

Time

Species

Detection

Bearing

Transect-Animal Distances (m) Transect-Estimated Group Centre (m)

MAP (m from start)

Estimated & Sighted Number

Age/Sex

Group Spread

Activity

Reaction

Notes

118 uvyr drs:Dt:Field sheet ref: Date: End time: Altitude: Additional observations Weather: D = very old and disintegrating (after White Edwards, 2000); distance from transect = fresh, still with animal odour; B recent, green leaves sions (widest and narrowest diameter in cms); state\age: A Notes: Distance from start = of transect; species (based on evidence faeces, hair other evidence); nest Distance UTM (if available): State/age Start time: Dimensions Nest Vegetation: Longitude: Species (km) from start Address: Distance Other: Transect length: Latitude: Survey site: (total observers) Surveyor: Counts Nest for Sheet Recording 6.2: Form – perpendicular distance from the transect (m) from transect (m) – in tree or on ground; woody herbaceous, etc; dimen- , but no odour; C = old, intact green foliage; (dd/mm/yy)

119 Form 6.3: Primate recording sheet for mapping calls

Surveyor: Field sheet ref: Date: (total observers): (dd/mm/yy)

Address:

Survey site: Altitude: Aspect:

Latitude: Longitude: UTM (if available):

Vegetation: Human disturbance:

Season: Weather: Lunar phase: Temperature:

Transect length: Start time: End length:

Other:

Time

Species

Detection

Bearing

Observer–Animal (m)

MAP (m from start)

Age/Sex

Group Spread

Activity

Reaction

Notes

120 7. Birds Leon Bennun and Kim Howell Platysteria jamesoni) Jameson’s wattle-eye ( Jameson’s

7.1 Biology Birds are the best-known group of vertebrates. There have been numerous studies on forest birds of eastern Africa, and most species are readily iden- tifiable using field guides and standard reference works. Birds play an important role in forests as pollinators of flowers and dispersers of seeds. Many of the smaller species also eat large numbers of insects and other arthropods. Birds, in turn, are preyed upon by reptiles, mammals, and other birds. The African crowned eagle, Stephanoaetus coronatus, is an example of a top predator in some forests, and may take prey as large as colobus monkeys. Birds are often considered as a useful indicator group, either for monitoring environmental change (see Furness et al., 1993) or for assessing biodiversity importance (Thirgood & Heath, 1994; Stattersfield et al., 1998). Birds as a group have many characteristics that make them good indicators: they are well-studied, taxonomically stable, easily surveyed, widely-distributed across almost all habitats, and include both generalised and specialised species. There are enough bird species (more than 1,300 in East Africa; more than 2,170 in Africa and Madagascar) to make meaningful comparisons between sites, but few enough that taxonomic and identification problems are rarely an issue. However, there are few precise details and guidelines on how birds can be used as indicators. Bennun & Fanshawe (1998) discuss bird surveys to

121 evaluate the effects of forest management, while Howard et al. (1998) demon- strate that information on birds can be used to select a set of priority sites for biodiversity conservation, even if the distributions of birds and other animals and plants are poorly correlated.

Categories of forest-dependence About one-third of the bird species in East Africa are found in forest. However, the extent to which they depend on forest differs. Bennun et al. (1996) list forest birds in Kenya and Uganda, in three categories: FF species (forest specialists) are the true forest birds, characteristic of the interior of little-disturbed forest. They may persist in secondary forest and forest patches if their particular ecological requirements are met. Where they do occur away from the interior, they are usually less common. They are rarely seen in non-forest habitats. Breeding is almost invariably within forest. F species (forest generalists) may occur in undisturbed forest but are also regularly found in forest strips, edges and gaps. They are likely to be more common there and in secondary forest than in the interior of closed-canopy forest. Breeding is typically within forest. f species are birds which are often recorded in forest, but are not dependent upon it. They are almost always more common in non-forest habitats, where they are most likely to breed. These categories can be applied to forests elsewhere in Africa, although the same species may fall into different categories in different parts of its range (Bennun et al., 1996). The forest-specialist birds tend to have smaller distribution ranges than the other categories, and are more likely to be threatened with extinction (Bennun et al., 1996). This is not surprising, because they are less tolerant of habitat disturbance than other species. The exact ecological requirements of most forest-specialist species are still poorly known, but many seem to use a narrow range of habitats within the forest undergrowth or the canopy. Changes in the forest structure, brought about, for example, by selective logging, may make it difficult for them to survive and reproduce successfully. Removal of forest causes them to disappear entirely. In most cases, conversion to plantations will have a similar effect: very few forest birds survive in plantation forest, except where indigenous tree species are planted. Some forest birds are long-distance migrants (for example, the African pitta, Pitta angolensis), while others (such as some parrots, hornbills and barbets) may make long movements in search of patchy supplies of food. Many, however, spend their whole lives within a small area of forest and may be reluctant to cross even small gaps between forest patches (Newmark, 1991). Forest species that migrate altitudinally between montane and lower-altitude

122 forests present a particularly complex conservation problem, as forests at both altitudes must be maintained.

Feeding guilds If comparisons are to be made across space (i.e. between sites) or time (i.e. monitoring at a particular site) then it is often useful to be able to subdivide the data according to forest dependence and guild categories. Guilds are groups of birds, not necessarily taxonomically related, that feed or behave in a similar way – for example bark-gleaning insectivores or just insectivores are both guilds. Bennun & Fanshawe (1998) show that these classifications can be useful for understanding the effects of forest management, since different guilds respond differently to particular structural changes. One advantage of using guilds or forest-dependence categories is that they average out the idio- syncratic responses of individual species, so that a more general pattern emerges. Plumptre & Owiunji (unpubl.) have developed the following set of feeding guilds for Budongo forest in Uganda (with codes for computerising data):

Feeding strategy Code

Frugivore FR Frugivore-insectivore FRIN Insectivore a. Sallying – from perch to flying insect INsa b. Ground – feeds on insects in leaf litter INgr c. Gleaning c1: Gleans from leaves INglL c2: Gleans from bark INglB Gramnivore (seed-eater) GR Gramnivore-insectivore GRIN Nectarivore-insectivore NEC Omnivore (eats many types of food) OM Raptor a: catches below canopy RAPb b: catches at canopy mainly RAPc

123 Feeding height Different bird species make use of different levels in the forest, so it is important to record species that are being seen. Three simple categories can be used:

1. feeds at or within three metres of ground level GRD 2. feeds in the middle strata of the forest or in understorey tree canopies MID 3. feeds at forest canopy height (in tallest trees) CAN

This classification provides a useful level of detail (Bennun & Fanshawe, 1998) and is recommended for application in other African forests. If survey data are to be lumped into forest-dependence or guild categories, this may affect the survey methods that are chosen. Not all survey methods produce data that can be pooled in this way (see section 7.3 below).

7.2 Management issues Intact natural forest is a diminishing habitat everywhere. This puts forest- specialist bird species at risk. In places where large blocks of forest have become fragmented, bird populations that were once continuous are now split into isolated units that may have limited interchange with each other (e.g. Lens et al., 1999). Fragmentation has other negative effects: fragments have relatively more edge and less interior than large blocks; they are especially vulnerable to habitat degradation; and they may be easier for predators or parasites to penetrate. Local extinctions have already been demonstrated in small forest fragments across East Africa (e.g. Tanzania: Newmark (1991); Kenya: Brooks et al. (1998); and Uganda: Dranzoa (1993)). In the Taita Hills, the asymmetry of birds’ plumage features (fluctuating asymmetry, a sign of stress) increases in smaller forest fragments (Lens & van Dongen, 1999), and there may be con- comitant effects on population sex-ratios (Lens et al., 1998). Management strategies to mitigate the effects of fragmentation may include restoration or maintenance of habitat corridors (e.g. forest strips along river valleys) and more effective protection from human disturbance. Where large forest blocks remain, zoning and control of forest exploitation should be designed to prevent habitat fragmentation occurring in the first place. Habitat degradation also has negative effects on forest birds. The populations of forest-specialist species decline as the structure of the habitat is modified (see Bennun & Fanshawe, 1998 for a review). Mechanised logging is the most conspicuous form of habitat degradation, but severe damage may

124 also be done by non-mechanical methods such as pit sawing. On a small scale, logging opens up gaps that mimic natural tree-falls and increase the diversity of species – but at the expense of the sensitive forest-interior birds. Exploitation for poles and fuelwood, and grazing by livestock, can cause serious problems for forest-specialist species. Collection of fallen deadwood affects insect populations, and thus birds too. Hole-nesting birds such as wood- peckers, barbets and hornbills rely heavily on standing deadwood or old, over- mature trees where nest-sites can be found or excavated (Newton, 1994; Du Plessis, 1995). A forest that is well managed for timber, with dead or dying trees carefully removed, may be very badly managed for birds (and other biodiversity). Tye (1993) reviews many of the relevant conservation issues with examples from Tanzania. Generally, arguments for the sustainability of these forms of forest use must be critically assessed; in practice, sustainability usually means a particular trade-off between economic benefits and biodiversity loss. Deciding whether such a trade-off is acceptable or not requires detailed ecological knowledge about the species of conservation concern and a monitoring programme to assess the effects of forest-use, both local (e.g. Hall & Rodgers, 1986) and commercial. Some bird species are hunted for food. In East Africa at least, this is not usually a concern for most forest species. However, it may be a problem for a few, such as forest francolins; for example, hunting already poses a threat to the Udzungwa partridge, Xenoperdix udzungwensis, which has a restricted distribution range in Tanzania. Capture of birds for the live-bird export trade affects a small number of forest species, such as the grey parrot, Psittacus erithacus. Any legal offtake of birds, for local use or export, must be controlled properly and based on detailed distribution and population studies for the species concerned. Such control measures do not currently exist. Birds can themselves be used as a management and conservation tool (Bennun, 1999; Bennun & Njoroge, 1999). At one level, they are likely to be the easiest group to monitor if changes in forest biodiversity need to be assessed. At another level, they provide an excellent focus for conservation education and action. For example, site-support groups with a birdwatching emphasis are already active around several forest Important Bird Areas in Kenya. Birdwatching has great tourism potential in East African forests, and can provide a source of local employment and revenue generation. Although birds still lack the public profile of large mammals, in East Africa they are receiving increasing attention and increasingly high priority in conservation and management. For example, in Tanzania, the finding of a new species of bird (as well as of other vertebrates) helped catalyse the establishment of that country’s first forest National Park.

125 7.3 Methods

General Surveys of forest birds may be undertaken for a variety of reasons, including: characterising the avifauna of a little-explored site; comparing the bird communities of different forests, in order to set conservation priorities; tracking changes in bird communities in relation to forest management (a form of monitoring: see below); investigating the distribution and status of particular birds of interest within a forest, or among a set of fragments. The different kinds of surveys differ in whether they require simple lists, measures of relative abundance, or measures of absolute abundance. Simple lists are the easiest kind of information to collect. However, with very little extra effort it is possible to collect information on relative abundance. This allows comparison of sites within and between forests and is generally much more useful. It is at this level that most survey work is carried out. Assessing the actual population densities of birds requires considerably more work, and should only be undertaken if the extra information really justifies the effort. This chapter provides only a brief outline of the survey methods most useful for forest birds, many of which have been field-tested in Kenya and Uganda. For more information and discussion, see Pomeroy (1992) and Bibby et al. (1998, 2000).

Inventory versus monitoring Inventory (finding out what species are in a particular site) and monitoring (tracking changes over time) are at opposite ends of a spectrum of survey types. It is important to be clear about what kind of work you are doing, as otherwise much valuable effort can be wasted. Strictly, monitoring implies assessing changes against some target value or threshold. With forest birds, we are more often involved with surveillance – a series of surveys over time. In either case, it is important that data are collected in a highly standardised way, on a regular (though not necessarily frequent) basis. You need to be able to repeat the same kind of data collection at the same place at the same time. Some useful background and guidelines on surveillance and monitoring can be found in Goldsmith (1991), Stork & Samways (1995), Tomas Vives (1996) and Bennun (2000, 2001). The key feature of monitoring (including surveillance) is consistency. Often, this means that you will only be able to survey a small portion of a

126 particular site. Entire bird communities can be monitored, but monitoring often concentrates on one or a few key species or species-groups that are particularly significant. For example, depending on your interest or the threats and changes facing a particular site, monitoring surveys might focus on a threatened species, or a particular feeding guild like large frugivores. Because monitoring aims to detect changes, it is important to minimise the sampling errors in estimates (see below). This means careful sample design and high-effort, low- coverage surveys. Suitable techniques for monitoring are those that can be exactly repeated in the same places time after time, and that give accurate estimates – for example, fixed- and variable-width transects and point counts. With inventory, the concern is usually comparisons in space (with other sites) rather than comparisons over time. The aim is to build up as complete a picture as possible of a site’s avifauna. Because bird distributions are often patchy, it is important to cover as much area as possible. Since inventory often has to be rapid, this means that surveys at a particular location may be relatively superficial. To make sure the species list is complete, different techniques (such as mist netting, playback and timed species-counts) may need to be used together, and all available habitats investigated. Absolute abundance measures are usually unnecessary for inventory work – relative abundance is enough, so long as similar approaches have been used at the comparison sites. Good inventory techniques will thus cover large areas and produce long species-lists quickly. Examples below include timed species-counts and MacKinnon lists (and their variants).

Identification To survey birds you need to be able to identify them, both by sight and sound. This is not easy in forests. Building identification skills takes time and effort, although going to the field with more experienced birders who are willing to teach and encourage you is probably the best way to improve identification skills rapidly. You can also build your knowledge of species by examining specimens in the collection of a museum and/or university, and through careful study of reference books and other literature. The number of useful references is increasing, including: illustrated field guides or checklists. For East Africa, the books of choice are Stevenson & Fanshawe (2001), for the whole region (except Ethiopia), and Zimmerman et al. (1996), an essential handbook for Kenya and northern Tanzania that includes many birds of Ugandan forests. Many good guides are available for southern Africa, such as Sinclair et al. (1997) and Newman (2000). Van Perlo (1999) covers Zambia, Angola and Mozambique, omitted by most other

127 guides. In West Africa the choice is more limited: Serle & Morel (1992) covers the whole region but it is out of date (and out of print in English). There are a few field guides for particular countries, such as SeneGambia (Barlow et al., 1997) and Sao Tomé and Príncipe (Christy & Clarke, 1998). Birds of Africa, a multi-author, multi-volume handbook published by Academic Press, and now into its sixth volume (Fry et al., 2000), with one more volume due. Certainly not one for the field, since each volume weighs several kilograms, but an essential reference work. The African Handbook of Birds by Mackworth-Praed & Grant (1955– 1973; six volumes) provides detailed information and is very useful for identifying birds in the hand, especially where no modern field guide exists. The nomenclature is somewhat out of date, and the illustrations are scanty and inadequate, but these volumes remain invaluable. Notes on particular difficult groups, in periodicals such as the African Bird Club Bulletin and Africa: Birds and Birding, or regional publica- tions like Scopus and Kenya Birds (e.g. Allport et al. (1996) on illadopses, Turner & Zimmerman (1979) and Bennun (1994) on Kenyan greenbuls). There are now good commercial compilations of sound recordings for each major region of Africa: West (by Claude Chappuis); East (by Brian Finch, published alongside Stevenson & Fanshawe, 2001), and South (by Guy Gibbon and others). These are very valuable reference sources for forest bird survey work, and will amply repay study. Bear in mind that there is substantial individual variation in calls, and that bird species do not always sound the same across their entire range. Many researchers and birdwatchers also keep their own sets of recordings and may be willing to loan them; other useful compilations are listed in Pomeroy (1992). Learning the birds that you are likely to encounter in a specific area, by sight and sound, will save much time and effort in the field. If you can reliably detect differences between birds, so that the number of different species you see can be recorded fairly accurately, this will enable you to collect meaningful data even if you cannot be sure of all identifications. Indeed, if you find yourself unable to identify a particular species in the field, even after having consulted a field guide or birding companion, then it is important to note down important features of the bird (e.g. general size/colour, beak colour/shape, eye colour, etc.) or to make a rough sketch. Do not spend too much time doing this, because the distraction may cause you to miss other sightings. With practice, your ability to record features of birds in your notes will improve, allowing you to identify more birds using the reference works.

128 Especially when you are starting off, it may be wise to note which birds you identify on the basis of their calls alone. This will enable you to go back to your data and correct your records (in the event of allocating the wrong name to a particular bird call). Ideally, you should identify all the birds you record reliably to the level of species. If you are unable to do so, then try and identify them reliably to the level of the group or genus (e.g. greenbul or flycatcher). Although birds are much easier to identify than many other groups of animals, they are still challenging – especially when you are using calls. Always be cautious, and take your level of experience into account. Never be tempted to name a bird unless you are really certain of the identification and can justify it from your notes if necessary. Write down doubtful cases as ‘Unidentified greenbul’ (or whatever), or in the worst case, ‘Unidentified bird’. Failing to make an identification is much preferable to making the wrong one.

A note on sampling A detailed discussion of sample design is beyond the scope of this chapter, and more information on sampling for bird surveys can be found in Pomeroy (1992) and Bibby et al. (1998). However, some general points are discussed by Bennun & Fanshawe (1998): i) Forest bird survey data tend to be noisy. If forests or forest blocks are to be compared statistically, there must be an adequate number of sampling units. ii) This means trying to balance the number of replicates, the time and effort needed to carry them out, and the size of the sample that can be collected in each case. For instance, if time allows you to run a total of 10km of transects in each block of forest, how should you divide these up? One 10km transect is probably not a good idea, but a hundred 100m transects might not be useful either – the number of birds you would record along each one might be very low. Most probably, something in between would be more suitable, with transect length determined by the number of birds you expect to record and the number of these transects determined by how much time you have. iii) Survey results can be strongly influenced by season, time of day and local habitat variation (including elevation). It is essential to minimise bias in your data by taking these sources of variation into account (e.g. by conducting counts at different sites during the same seasons, randomising count order across the day, and stratifying your sample to take habitat and altitudinal variation into account). If you intend to do transect work, or carry out mist netting, you will need to think about the location of your lines. Usually these should be randomised, as far as possible, within each stratum of your sample (but if you are mainly

129 interested in building up a species list then you will want to place mist nets in a variety of micro-habitats – also see section 5.3.3). It is often desirable to use existing small trails and paths, rather than destroying more vegetation by cutting your own. Be aware, however, that trails are themselves often aligned in a highly non-random way. Transect lines do not necessarily have to be entirely straight. Some reconnaissance to map out trails may be necessary (and is almost always useful) before survey work begins. When trails are mapped they can be partitioned into convenient sections of equal length. By giving each section a number, you can select sections for sampling using random number tables. Alternatively, especially if you do have to cut your own trails, you can lay out transects or mist net lines systematically. For example, you might decide to sample at every 500m alternately to left and right along a line bisecting the forest. Systematic sampling has the advantage of simplicity and of covering the whole study site (by definition, random sampling does not always do this). However, where there is a regular pattern in habitat variation, such as evenly spaced ridges and valleys for instance, it can potentially lead to biased results. See also section 7.3.11 for tips on suiting the method to the bird.

7.3.1 General surveys A general survey consists of recording the species, and sometimes number, of birds seen and heard in an area. After a period of observation, a broad indication of abundance, such as ‘regular’, ‘common’‘rare’, and so on, can be given to each species.

Site selection, procedure and recording A regular record is kept of bird species seen and heard during walks (or at any other time, e.g. around a base camp or while watching a fruiting or flowering tree). Usually, observers try to cover as much of an area as possible and investigate all major habitats and micro-habitats, in order to maximise the number of species recorded. Observers might also use play-back and mist netting to try and detect inconspicuous undergrowth species. Preferably, habitat and numbers (at least for flocking species) should be noted for each record.

Advantages/limitations This is the least useful approach with respect to quantitative results and comparative work, but it does allow an initial species list to be drawn up. It is difficult to standardise observer effort, and impossible to make any but the broadest comparisons with other sites.

130 If a survey is being planned specifically for birds, then it is much better to use a more systematic approach (such as timed species-counts). If the bird work is being done as a side-line, perhaps while conducting botanical surveys, then a general survey may be all that is possible. In any survey it is important to record ad hoc observations of birds (i.e. casual records in addition to those on scheduled counts) as these may add substantially to the species list.

7.3.2 Timed species-counts (TSCs) The timed species-count technique provides a quick and simple method for gaining a measure of relative abundance of canopy and mid-level bird species in a fairly large, defined area. It has the advantage of covering a bigger area than point counts or transects (see below), and it is not tied down to particular localities or lines. Thus it is possible to build up an overall species list much faster. TSCs are essentially repeated species lists, on which are indicated the first time when each species is first positively identified by sight or sound (Pomeroy & Tengecho, 1986; Pomeroy & Dranzoa, 1997). Species receive a cumulative score according to when they were first recorded on each count: species that are observed more frequently receive higher mean scores, as they tend to occur early within a count as well as in a high proportion of counts. TSCs were developed initially for use in open habitats, but have been modified in Kenya for application in forests.

Site selection As far as possible, TSCs should be well spread over the whole study area being surveyed. They should also take in all the different micro-habitats (such as different forest types on ridges and along streams). During the count, there is no need to keep to a set path, and you can wander off to investigate sounds of bird activity, fruiting trees, and so on. If you are surveying forest birds, it is usually a good idea to avoid going into areas of entirely different habitat, such as grassland or cultivation, as this could obviously give you misleading results. If you have a GPS, it is useful to record the start and end points of each TSC. If you work in more than one study site within a forest, you should carry out the same number of TSCs in each.

Procedure and recording i) The observer walks slowly and quietly along a path in the forest for a fixed time period: 40 minutes has been used in Kenya, 60 minutes in Uganda.

131 ii) In its simplest version, when the observer positively identifies a particular species for the first time during that count, the species is recorded along with the time. In the modified version for forests, a note is also made, using a simple code, as to whether the birds are above or below 3m from the ground, and whether they are more or less than 25m from the observer’s present position or intended route. In this modified version, it is important to note the first time the bird is recorded within these limits: i.e. above 3m from the ground and within 25m from the trail. Only this information is used in calculating the TSC index. Data may be entered on Form 7.1. iii) The 3m limit excludes hard-to-detect understorey birds, which this technique samples inefficiently. The 25m limit removes some of the bias due to noisy or conspicuous species, which tend to be identified early in a count and thus receive high scores (Bennun & Waiyaki, 1993). Records outside these limits are still valuable in compiling the overall species list. They can also be used, if so desired, to calculate an overall occurrence index, i.e. the proportion of counts on which the species is detected. The 3m limit was developed for use where undergrowth birds are being sampled by mist netting (see below) and may be ignored when mist netting is not being used. iv) In Kenya, the typical procedure has been to carry out at least 20 TSCs for each study area (a particular sector of a forest). Several TSCs may be carried out in a morning, separated by intervals of at least 10 minutes (in time) and 100m (in space). Timed species-counts are undertaken by a single observer, usually after the early peak of morning bird activity has decreased; this makes the results more consistent, and also allows workers to deal with the early activity expected at the mist nets (see below). For standardisation purposes, the suggested count period is between 08:30 and 12:00, but this may need to be adjusted for different sites. General records of weather conditions (including wind) should be made at the beginning of each count. v) This exercise should also be repeated as many times as possible, along different routes. Several people can conduct TSCs along different routes at the same time. vi) If using the modified method, you will need to use a simple system of symbols, or a column on a form, to indicate whether or not the bird is above or below 3m and within or outside the 25m limit. It may also be a good idea, as suggested above, to indicate whether the bird was seen (‘s’) or only heard (‘h’).

Data analysis In the modified method, the TSC index is based on only the records above 3m and within 25m. To calculate this index, we need to know only the first time that each species is recorded within these limits; once a species has

132 been recorded above 3m and within 25m, there is no need to record it when it appears again during the same count. For birds recorded outside the limits, each species need only be recorded once and the time is unimportant. For each count, each species is then assigned an index ranging from 0 to 4, depending on whether it was recorded during the first 10 minutes (= 4), second ten minutes (= 3), down to 0 for a species not recorded during that count. An average score is then taken over all the counts. For an hour-long TSC, the scores go from 0 to 6 in the same way. The scores for the two systems can be made comparable by adding 2.0 to the scores for the 40-minute counts.

Advantages/limitations The TSC is a good method for assessing relative abundance. It is not necessary to record the number of birds detected, just the species, and you can concentrate on detecting new species. This is an advantage for relatively inexperienced observers who may need to spend more time on identification, and also for more experienced observers visiting a new site for the first time and still familiarising themselves with the local birdlife. It is also useful when trying to cope with mixed-feeding parties, when many species may pass through in a very short time. The TSC index is a useful comparative measure of different sites and forests. It is not as good a method as the point count for detecting shy birds of the forest interior, but appears to be just as efficient (perhaps more so) for sampling canopy species. Pomeroy & Dranzoa (1997) show that species richness can be assessed from TSCs either from species accumulation curves or regression estimates. Measures of relative abundance for individual species from TSCs also correlate well with those from transect counts (Pomeroy & Dranzoa, 1997) or timed transects (see below: Bennun & Waiyaki, 1993). TSCs produce data on more species and in less time than traditional transects, thus making them more efficient. Because the TSC does not have to follow a set path over a set distance, it is much easier to carry out in forests than is a standard line transect. If the aim is to assess species richness (or the richness of a sub-group of species, such as forest specialists), or the relative abundance of particular species in different forests or compartments, then TSCs are a good method to use. In the unmodified version of TSCs, scores strongly reflect detectability as well as abundance. They thus cannot be used to compare even the relative abundance of species that differ widely in their detectability (Pomeroy & Dranzoa, 1997). This problem is partially overcome by the modified version with a 25m distance cut-off (Bennun & Waiyaki, 1993). The relative abundance (compared to other species) of flocking birds is underestimated by TSCs (Bennun & Waiyaki, 1993), and TSCs do not adequately sample shy understorey birds.

133 However, the main problem with TSCs is that the indices are not very easy to handle mathematically. Each TSC index is a measure of relative abundance for a particular species on a scale of 0–4 or 0–6. It is not clear whether TSC scores can legitimately be summed to create a cumulative index for forest-dependence categories or for guilds. If summed abundance measures for these sub-groups are needed, then it may be better to use another method – perhaps a timed transect (see below). For this reason, in particular, TSCs may not be very useful for monitoring purposes (see Bennun & Fanshawe, 1998). 7.3.3 MacKinnon lists and related methods MacKinnon lists (MacKinnon & Phillips, 1993; Bibby et al., 1998) also allow calculation of an index of relative abundance. In essence, you build up a picture of the richness of the avifauna and species’ relative abundance by compiling a series of species lists – each with the same number of species. The faster the total number of species rises as you add lists, the greater the overall richness. The more lists a particular species occurs on, the more abundant it is. Fjeldså (1999) developed this method further for use in rapid assessment of forest avifaunas, using a 20-species list.

Site selection As for timed species-counts. Ideally, your effort should be well spread across your study site and cover all micro-habitats. Simple MacKinnon listing requires searching for birds in whatever may be the most efficient manner. Different ground should be covered from one list to another. The minimum number of lists needed for each study site is around 15. The Fjeldså technique involves listing within a defined study site – in his case, of area 1.5 km2 in a particular forest type. Within this, you walk randomly, searching for and recording birds.

Procedure and Recording In the simple version, you search for and record each new bird species until you have a pre-set number of species on your list – usually between 8 and 20 (the more species-rich your forest, the higher the number on the list should be). When the species total reaches the pre-set number, start another list, on which the same species can appear again if you see or hear them. Fjeldså’s method is similar, but you record every bird seen and heard throughout the count. Fjeldså walked randomly within the study site from dawn to dusk, recording throughout. However, the method can just as well be applied

134 for shorter periods – the minimum on one day being the time taken to acquire a list of 20 species.

Data analysis Plotting the total number of species recorded against the number of lists included gives a curve that rises rapidly at first and then levels out, finally reaching a plateau. This plateau level gives the observed species richness, which can be compared across sites (using the same number of lists, with the same number of species in each list, for each site). This value is not the same as the real, total species richness, but can be used to estimate it: Fjeldså (1999) gives a formula for this, based on Colwell & Coddington (1994). Relative abundance of species can be expressed as the fraction of lists on which a species occurs (Bibby et al., 1998). However, since each species can only occur once on each list, this will severely underestimate the abundance of common species. Fjeldså (1999) dealt with this problem by recording every bird seen or heard, whether a new species for the list or not. The total number of records for each species can then be expressed as a percentage of the overall total. This is a similar approach to the timed transect method (see below). Relative abundances calculated this way correlated very strongly with measures derived from intensive point count observation in the same forests (Fjeldså, 1999).

Advantages/ limitations The simple MacKinnon list method does not produce reliable relative abundances, which is a disadvantage for most studies. Fjeldså (1999) lists the following advantages of the adapted method: time is used efficiently – all the time available is devoted to data collection, rather than, for example, moving between transect lines or point count points; more information is gathered than in a total list of species seen: data can be standardised and total species richness extrapolated; it produces reliable information on relative abundances (provided that all the birds seen or heard are recorded); it is less influenced by the relative skills of observers, compared with timed species-counts. This is because you can take as much time as you need to identify a particular bird, and it does not matter whether you complete your list of 20 species in an hour or in a whole morning. To obtain relative abundances with this method requires more effort than with timed species-counts, since all birds seen or heard have to be recorded, rather than just each new species. The reduced influence of relative skills is only real if all observers can eventually identify all the birds they see or

135 hear – no survey method will work well with very inexperienced observers. With an experienced survey team, this approach differs very little from timed transects (see below).

7.3.4 Timed transects (TTs) Timed transects (TTs) were developed for use in Kenyan forests as a simple method of assessing relative abundance that did not suffer from some of the problems of the TSC. Timed transects measure the number of birds seen in a set time, rather than along a set distance.

Site selection, procedure and recording The timed transect technique is similar to the TSC, except that the observer also records the numbers of birds seen or heard within the limits described for the modified TSC (i.e. above 3m and within 25m), each time a bird is identified (not just the time a species is first detected). Species detected outside the limits are recorded separately in order to build up the species total.

Advantages/limitations Timed transect and TSC scores are strongly correlated (Bennun & Waiyaki, 1993), and the two methods share the same advantages of being simple and quick to perform. However, because the actual number of birds is recorded, the timed transect method is less susceptible to the biases of the TSC index in favour of conspicuous species and against flocking species. Because the TT index is the actual number of birds detected for each species, it is straightforward to produce cumulative scores for forest-dependence and guild categories. However, TTs require more effort than TSCs. All birds within the limits must be identified and counted. It can be especially difficult to estimate the numbers of birds that are only heard – some experience of the forest and the species may be necessary before this can be done with accuracy. Timed transects could, in principle, be used for monitoring. However, it is difficult to repeat exactly the same series of counts in the same places, and to tie these down to other aspects being monitored (such as vegetation). Point counts or fixed-length transects are recommended for monitoring instead.

7.3.5 Fixed-width transect counts Transect counts have been used extensively in open habitats, but, in closed forest, visibility is poor and the viable transect width tends to be narrow. It can also be difficult to lay down fixed-length transects in forest habitat. In the- 136 ory, line transects provide a measure of absolute abundance; in practice, this is probably not the case, as many birds are probably missed – although the results of fixed-width transects are expressed as a density. See the discussion of line transects in the mammals chapters (sections 5.3.3 and 6.3.2).

Site selection Because transects start and stop at specific points, it is usually possible, and desirable, to randomise their locations, or to use systematic sampling (e.g. at fixed distances along a grid). See the section on sampling above.

Procedure and recording i) You can either use existing trails or grids, or cut your own. Along the route you have selected, measure out and mark a trail of known length (a useful length might be 1km). ii) The procedure is to walk the measured route quietly and slowly. The best time to do this is in the early morning or late afternoon (when birds are most active), but, in any case, counts should be made at consistent times across different sites. iii) Record all the birds you see within a fixed distance of each side of the selected route: distances between 10m and 25m are feasible depending on the nature of the habitat (use Form 7.2). If birds are in groups note the number seen. Record the time and species of each sighting (making notes/sketches of any birds that you cannot immediately identify). Continue walking the transect and make sure that you look at all levels, from the ground to the tops of the trees above the route. iv) After completing the transect, go back over your notes and attempt to identify any species you were unable to identify in the field using field guides and other literature.

Data analysis Transect data can be used to give a measure of species richness, to calculate diversity indices (see Magurran, 1988 and Pomeroy, 1992 for more information), and to give a density of individual species or categories/guilds within a defined area of forest. This area is the length of all transects walked multiplied by the width covered.

Advantages/limitations Fixed-width transects provide a measure of density (although this will usually be an underestimate, as many birds may be missed). They also allow

137 various diversity indices to be calculated, as there is an abundance associated with each species. However, this is rarely very useful in practical terms, and the best measure of diversity is usually simply species richness. Fixed-width transects are time-consuming and cover a relatively small area. They are, therefore, not a very efficient way of building up species lists. They take no account of differences in detectability of species between different habitat types (e.g. disturbed and undisturbed forest), and so can potentially give misleading density estimates. If reliable density measures of particular species are required, variable-width transects using distance sampling may be more useful (see below). Fixed-width transects can also be difficult to lay down, requiring at least mapping and measuring, if not extensive cutting of trails. This takes a good deal of effort and may lead to unnecessary forest disturbance.

7.3.6 Fixed-width point counts Although excellent for surveys, TSCs and TTs are not really suitable for monitoring purposes, and point counts have been used for this purpose (among others) in Kenya. A major problem with this technique, however, is that very few birds tend to be recorded, since point counts sample a relatively small area.

Site selection It is usually easiest to lay out point counts on a grid in a systematic way. Random locations are also possible, but may be difficult and time-consuming to find (and re-find) on the ground. You should attempt to make at least 50 point counts in each study site. In Kenya, a simple method of spacing out counts was devised. Counts were made along a cut transect every fifteen minutes, i.e. the observer spent eight minutes walking (fast) along the transect, followed by seven minutes waiting and counting at the point (see below).

Procedure and recording i) The observer stands at a pre-determined point that forms the centre of a count cylinder that extends from forest floor to tree tops. After a two-minute settling-in period, the next five minutes are spent recording all the birds seen and heard within a radius of 25m (Form 7.2 may be used). These times can be varied if necessary. Theoretically, a point count is supposed to detect all the birds around a point at the moment the count starts. Therefore, there must be a balance between the time taken to search the area thoroughly, and the likelihood of birds not recorded in the count area moving into it during the count.

138 ii) The species and numbers of individuals are recorded for all birds within the limits. Species detected outside the limits can be recorded separately to build up the species list. It can be useful to record whether birds are above or below 3m height (regardless of whether mist netting is also being used) so that undergrowth and higher-level birds can be distinguished. Bird community changes in response to habitat change can be different at these levels (Bennun & Fanshawe, 1998). iii) Point counts should be made over a standard period; for monitoring in Kenya this has been defined as 09:00 to 11:15, after the main mist netting activity of the morning (this would mean a maximum of eight counts per observer per day, following the routine above). Usually, point counts are carried out by a single observer, but it is possible to have an experienced observer accompanied by a trainee. Data analysis Absolute densities for birds detected within the limits can be calculated as the total number of birds recorded, divided by the total area covered by the counts. It is also possible to calculate an occurrence index (the proportion of counts where a species was recorded, for all birds inside or outside the limits). Advantages/limitations Point counts are useful for monitoring because they can be replicated precisely: counts can be made in more-or-less exactly the same places, in the same defined area of forest, on a future occasion. Point counts also have the great advantage that habitat parameters can be measured around each point, and related to the presence or density of the birds. For example, Oyugi (1998) used point counts to assess bird abundance in Kakamega Forest, Kenya. After measuring habitat parameters at each point, he was able to develop predictive models for the density of various bird species, and for particular guilds and forest-dependence categories. Similarly, Fanshawe (1995) used point counts to compare birds in more- and less-disturbed habitats in Arabuko-Sokoke Forest, Kenya. Point counts sample a relatively small area and the number of individuals recorded for most species on most counts will be zero. This makes it difficult to compare density estimates for particular species statistically. One approach is to compare the proportions of counts where particular birds were recorded. This loses important information, however. Another possibility is to combine density estimates for guilds or forest-dependence categories, or to sub-sample blocks so that a set of point counts are combined into a single data point. The small area covered by point counts makes them unsuitable for most rapid survey work, at least where the main objective is to draw up species lists and obtain a general idea of the avifaunal composition.

139 7.3.7 Distance sampling Distance sampling provides a powerful set of methods for estimating absolute abundance of particular species of interest (see Buckland et al., 1993). Distance methods take into account the fact that many birds that should be detected during a transect or point count will actually be missed. They also take into account the fact that birds may not be equally easy to detect in different forest types. Owiunji (1996, 2000) used point counts with distance sampling to measure abundance of birds in different forest types in Budongo Forest. Kosgey (1998) used transects with distance sampling to assess the densities of Turner’s eremomela, Eremomela turneri, in South Nandi Forest. Similar methods were used by Musila (2001) to census Sokoke pipits, Anthus sokokensis, and Mulwa (2001) to study Taita white-eye, Zosterops poliogaster. In the Uluguru Mountains, Tanzania, Tom Romdal (unpubl.) played back calls of the Uluguru Bush-shrike, Malaconotus alienus, at fixed points, and estimated the distance of responses.

Site selection, procedure and recording Distance methods may be used in conjunction with either transects or point counts. The basic procedure is as described above for fixed-width methods. However, in this case, there is no cut-off point at a particular distance. Rather, all records (with group sizes) of the species of interest are noted, together with their perpendicular distance from the transect line, or the point count observer.

Data analysis Data are analysed using the programme DISTANCE (the software and manual are available, free of charge, on the internet at www.ruwpa.st-and.ac.uk/distance). DISTANCE uses the set of observed distances to model detectability functions (the way in which numbers of records decline with distance from the observer) and to give estimates of actual densities. Different functions can be obtained for each species and for different habitat types (see also section 5.3.3).

Advantages/limitations Distance sampling is a very effective and powerful method for a specific purpose: obtaining reliable estimates of the actual abundance of focal species (e.g. a particular threatened bird whose population size is unknown). It is unlikely to be a method of choice for most biodiversity surveys. Obtaining distances for each observation is time-consuming and difficult. For species that 140 are rarely encountered, there may not be enough data to model the detectability functions. Densities can only be obtained for the more frequently recorded species, and estimating densities for forest-dependence categories and for guilds will not usually be possible. Using playback of calls combined with distance sampling can be very effective for birds (like the aforementioned Uluguru Bush-shrike) that are skulk- ing, scarce, or hard to detect. This does assume, though, that the birds call in response as soon as they hear the recording. If the birds come closer to you before responding, the technique may provide biased estimates of abundance. 7.3.8 Mist netting and ringing All of the above techniques will tend to miss or under-represent forest understorey birds, which often tend to be difficult to detect and identify visually. Knowledge of vocalisations can help considerably, but some species are largely silent and others sing only at certain times of the year (see section 4.3.5). The only means of overcoming this difficulty is to use mist nets. Mist netting and ringing are powerful techniques for surveying and studying birds, and have been used to follow weight changes, moult, breeding seasons and movement in individual birds. Employing capture-mark-recapture techniques, it can also be used to estimate population size. However, mist netting and ringing require considerable expertise and extreme attention to detail, which can be acquired only through lengthy training. If you will be using mist nets to capture, identify and possibly ring birds, special permission is required. You should contact the appropriate ringing scheme (e.g. the Ringing Organiser, East Africa Natural History Society, P.O. Box 44486, Nairobi) for advice. Mist netting and ringing should be carried out only by competent and experienced persons and qualified ringers, who can handle the nets and the birds ethically and safely. If birds are not handled properly, you may affect both the results of your study and, more importantly, the long-term survival of the population. If you are not a qualified ringer, it is your responsibility to ensure that at least one and preferably two qualified ringers are members of your survey team, and that they train others in the proper techniques of netting, handling, and ringing birds. The nets are called mist nets because they are made of extremely fine nylon thread and therefore are almost impossible to see when stretched out. They come in a variety of heights, lengths and mesh sizes, but all feature various numbers of shelves – a pocket of mesh suspended from a strong thread, which runs along the net. If the nets are set correctly, birds flying across the net line do not detect them. When they fly into the net, they drop into one of the mesh pockets, and become entangled (Fig. 7.1). They can then be carefully removed, identified, banded and measured.

141 Fig. 7.1: Examples of mist nets

Equipment mist nets poles (straight bamboo poles about 40mm in diameter are ideal – light but strong) string bird bags (cloth bags with a draw-string, in which to hold netted birds) rings (variety of sizes)1 ringing pliers ringing book (printed books2, or use a hardbound accounting ledger and label the columns: these data must later be transcribed into the schedules for the ringing scheme under which you operate; see also the Ringing Form at the end of this chapter). A waterproof bag is always advisable. stop-end ruler spring balances (sizes 50g max. and 100g max.) tarpaulin or fly sheet to protect birds and ringers from the elements at the ringing station flagging tape (for marking net sites for location before dawn, etc.)

1 Available from: Ringing Organiser, Nature Kenya, PO Box 44486, Nairobi, Kenya ([email protected]); Ghana Wildlife Society ([email protected]); Cameroon Ornithology Club ([email protected]). 2 Contact: Ornithology Department, National Museums of Kenya, P.O. Box 40658, Nairobi; Kenya.

142 Site selection Mist nets work best when they are set against a background of vegetation. That makes them less visible to birds. Forests usually provide a good habitat for netting, but open glades and canopy gaps should be avoided. Areas with a very open understorey usually produce low capture rates too, while nets close to streams or other water sources are often especially productive. If you are trying to obtain a representative sample, however, you will need to lay out your nets along randomly or systematically located lines, stratified according to major habitat divisions, as explained above.

Procedure i) To set mist nets, you could use existing (reasonably straight) trails, or cut runs about 1m wide, if the vegetation will not permit the net to be set as it is. The aim is not to clear a wide path, or to alter the vegetation, but simply to let the net be set so that it is not in contact with branches or other vegetation. The ground over which the net will be placed should also be cleared of any vegetation or obstacles (such as rocks) that will entangle the net. For a well- illustrated explanation on how to set mist nets, see Howes & Bakewell (1989). ii) A commonly used size of mist net is 12m in length and 3m high, with four shelves and a mesh size of about 30mm (1.25in). Longer and shorter lengths, and various mesh sizes, are also available. It is important to have the bottom shelf of the net resting close to ground level. Many forest birds are taken only in the bottom shelf, or the one above it. However, beware of ant swarms and small mammal predators, which might attack birds trapped close to the ground – frequent patrolling is important. iii) The nets must be checked frequently at not less than half-hour intervals (preferably more often), and the birds removed (by trained extractors only) and placed in cloth bags. The birds are then taken to a ringing station a little way from the nets (close enough to be convenient, but far enough not to inter- fere with netting) where they are processed (identified, measured, weighed, ringed, examined for moult, brood patch, etc.) before release. iv) Metal rings (termed ‘bird bands’ in North American literature), come in several different sizes, and the correct size must be fitted to the tarsus of an individual. Each ring has its own number and instructions on it (e.g. in East Africa, the message ‘Send Museum Nairobi’). Ringing is not an essential part of bird surveys using mist nets. However, it is important that captured birds are marked in some way, so that you can tell if you retrap them. This is usually the most convenient (and by far the most informative) way of doing so.

143 Suggested surveying procedure i) This procedure has been used successfully for survey by mist nets (with simultaneous timed species-counts) in Kenya. ii) The exact number of net sites used and the length of net set up will vary according to the time available, i.e. the survey schedule, and the particular forest. The aim is to catch at least 40 birds per net site in order to have a representative sample. Since capture rates vary between forests, so will the netting effort required. For each defined study block, try to net over two sites, each with between 120–200m of net. Each site is netted for two consecutive mornings. The nets are operated for four hours after dawn each day. This standardisation allows comparison of capture rates per net metre hour. The importance of using the same time relative to dawn is that bird activity patterns vary greatly between dawn, mid-day and evening. Experience has shown that capture rate usually drops off dramatically in the late morning, and that evening capture rates are usually lower than those of the early morning too. iii) Nets are set in a straight line with no breaks, along transects cut through the forest. When possible, it is useful to net along the same transects as those cut for other studies, such as botanical work. This saves much time and effort, but it will still probably be necessary to clear the transect further (especially at ground level) to make it suitable for placing the nets. Mist nets are set as close to the ground as possible. Nets 18m in length are preferred (the fewer nets, the less effort), but 9m and 12m long nets can also be used as needed. These nets have four shelves and are a small mesh size, suitable for capturing passerine birds. iv) The number of nets operated could always be increased to give a bigger sample, but this would mean more effort spent in clearing lines, setting up and moving nets, and would not leave much time to collect important biological data on the captured birds. The time spent handling each netted bird is obviously a critical factor. Processing very large numbers is usually not feasible for a survey with limited time and people-power.

Suggested monitoring procedure i) This procedure has been used for forest bird monitoring by mist nets (with simultaneous point counts) in Kenya. ii) For monitoring, one needs to obtain data that can be compared statistically between monitoring sessions. As with the survey, a combination of mist netting and observation is employed. iii) The sample unit is the net-line. Six to eight net-lines are operated in each study area; the more the better, but time is usually a real constraint. Six net-lines is probably the minimum for a study area. iv) Each net-line consists of a straight series of nets, continuous along a

144 line cut perpendicular to a survey transect (Fig. 7.2). The length of net is adjusted to ensure a sample of at least 30 birds per net-line. Because of differences in understorey bird density, different lengths of nets will be required at different localities. In Kakamega, 66m of net were found to be sufficient; in the Mau forests, 102m were needed. The most suitable length requires some previous experience of the area. In any case, all net lines in a study area should preferably be of identical length. Two lines are operated at once, which means that they cannot be too far apart. In Kenya, lines have been cut alternately to left and right, with around 200m between their starting points. The ringing station is then based halfway between the two lines. v) Nets are operated on a strict timetable, again for four hours from dawn onwards. Records are kept of which site each netted bird comes from (via a small piece of paper folded into each bird bag). In Kenya, netters have found no records of birds moving between sites, suggesting that each line is indeed an independent sample.

Fig. 7.2: Mist net set up line

net line

transect

200m

Recording i) For each net site, the layout of the nets should be sketched and notes made of the site location and habitat type and condition. ii) For each netting session, record the date, the start and end time (when nets were unfurled and furled), and the weather conditions (see Form 7.3). iii) For each bird caught during a netting session, it is essential to record the species, ring number and age and sex (if determined). A number of standard biometric measures are normally taken, including wing length, weight and the status of moult in the primary wing feathers. Additional biometrics, such as head and tarsus length, secondary feather, tail and body moult, are also often recorded. For detailed studies, it can be useful to record in which net, and in which shelf, particular birds were trapped. The easiest way to do this is to write the information on a small piece of paper when extracting the bird, then to fold this and place it in the bird bag. When the bird is removed from the bag so is the paper, and the details are recorded in the ringing book.

145 iv) Some studies involve taking blood for DNA analysis, putting colour bands on the birds’ legs to allow individual identification when resighted, or taking repeated measurements to assess fluctuating asymmetry. All this information can be recorded in the ringing book as well. v) Recaptured birds need to be indicated, usually by means of an ‘R’ in a circle by the ring number.

Data analysis As far as survey work goes, the data from mist netting essentially consists of a list of species and the numbers trapped. These can be used as a means of assessing relative abundance. Catch rates for individual species can be compared between species and sites if they are expressed as individuals per 100-metre-net-hours; see Pomeroy (1992) for examples. Within a study, using a standard length of net for the same hours each day simplifies comparisons. Totals can readily be summed across feeding or nesting guilds. Species accumulation curves can be drawn using mist net data, plotting the cumulative number of species against a measure of effort (such as the number of mist net sessions, or cumulative metre-net hours).

Advantages/limitations Mist nets are important because currently they are the only method available to sample birds of the understorey adequately. Mostly using mist nets, Baker and colleagues (unpubl. data) added some 14 species of forest birds to the Tanzanian list during fieldwork in Kagera Region, Tanzania. Almost none of these species had been identified using binoculars, and the presence of many probably would have gone undetected without the use of mist nets. The main disadvantages of using mist nets are that they are expensive to purchase, demand a high degree of training to use properly, and are very labour-intensive to use requiring substantial time and effort, especially in cutting net-lines and shifting nets. Although mist nets allow a standardised approach to surveying understorey birds, they do not sample birds of the mid- and upper canopy. They also have a number of biases. Capture rates are strongly affected by the time of day, as mentioned above, so comparisons should be made cautiously if the approaches have not been standardised. The type of nets and the way they are set also affects capture rates. For monitoring, or collecting comparative survey data, nets should be of high quality with plenty of pocket, must be in good condition (holes dramatically reduce the number of captures), and set in a consistent manner (stretched tight between poles, and just off-taut between shelves). The condition of nets should be checked regularly (at least every half hour, preferably more often) and any damage repaired (mist nets require

146 considerable maintenance). Nets do not sample all birds (even small ones) equally. The standard mist nets also are not suitable for very large, heavy birds, or fast fliers, such as large parrots, pigeons, and birds of prey. Any ringer will be aware that some species are more catchable than others. Remsen & Good (1996) showed that small changes in bird behaviour, e.g. alterations in foraging height following forest disturbance, could potentially cause large changes in capture rates. This does not invalidate the use of mist nets, but means that results must be interpreted cautiously. It is another good reason for combining mist netting with observational work, such as timed transects or point counts. Bear in mind, however, that a team of at least four is needed if mist netting is to be combined with counting of canopy birds using point counts.

7.3.9. Sound recording The use of tape recorders in bird surveys has not yet been developed as a standard technique in East Africa, but it has many positive aspects (see Parker, 1991, for a discussion). The human ear is notoriously selective and often screens out sounds. However, the tape recorder documents all sounds detected by the microphone. Recorded sounds can then be later used to identify birds, just as field notes and diagrams can be compared in field guides. Recorded sounds can also be amplified and played back in order to attract birds close enough to enable identification. This can be extremely useful for surveying and monitoring shy and hard-to-see species, especially when these are scarce and call infrequently. It can be used in an ad hoc way (recording and playing back calls of birds you have detected but do not recognise) or more systematically (targeted at difficult-to-locate key species). Small, inexpensive recorders are readily available and can be linked up with a video-recorder microphone to form a handy, portable system. It is important to use a directional microphone, and to ensure that the system produces loud enough playback. Though more expensive than tape recorders, mini-disc systems work very well for sound recording. They are light and portable, store sound digitally, and offer the great advantage that you can edit and copy your material almost effortlessly.

Equipment high-quality portable tape recorder or mini-disc player microphone and batteries blank tapes or mini-discs recorded tapes/CDs of bird sounds plastic bag for protecting equipment against moisture

147 Site selection, procedure and recording For inventory of key species, playback of recordings can be applied in a systematic way, for instance by playing a recording for a set length of time at a set volume at a particular point. Any responses by the birds within a set time period (and, if appropriate, their distance: see also above) can then be noted. Virani (2000) applied this method to survey Sokoke scops owl, Scops ireneae, though using a small flute rather than a recorder to imitate their whistled calls. Systematic playback can also be used for monitoring of one or more target species: the exact points where it is carried out (and the exact time and volume of playback) then remain fixed between surveys. For inventories of whole bird communities, Parker (1991) makes the following suggestions: i) Get up well before dawn and be out in the area to be surveyed at least 15 minutes before first light. ii) Choose a different spot each morning from which to record, preferably at areas at least 500m apart, and let the recorder run for 15 minutes or more (depending on the amount of vocal activity). Point the microphone in the direction of louder sounds for at least 60 seconds. Try to record in all directions and from the undergrowth up to the canopy. Cover as many types of forest and microhabitats as possible. iii) Find areas where mixed-species flocks are forming at dawn and record them for at least 10–15 minutes. Get 5–10 minutes of sounds from any large flock found at any time of day. (Note: mixed-species flocks are especially conspicuous in South American forests. Not all African forests have many mixed-species flocks, nor are these flocks always very vocal.) iv) Once a comprehensive collection of recordings has been assembled for a locality, attempt to obtain additional recordings of different individuals and each species. Data analysis Just as for specimens, recordings need to be accurately documented, and originals or copies deposited in institutions equipped to curate them. See Form 3.3 for an example of a data sheet used by the Macauley Library of Natural Sounds at Cornell University, Ithaca, USA. The recordings can be analysed with respect to numbers of species, time of calling, season of calling etc. Advantages/limitations Recordings (within the physical limits of microphones and tapes/mini- discs) represent a unique way to document avian species diversity. They can easily be copied and sent to others for research purposes, and also make an excellent teaching tool. 148 The main limitation is how to keep the equipment from being exposed to excess heat and humidity. Sophisticated recorders and microphones are also relatively expensive, but the costs are outweighed by their versatility and the type of data they are able to collect. Playback targeted at key species needs to be used sensibly and in a limited way. Birds responding to the tape (or imitated call) do so because they think they have detected an intruder in their territory. This disrupts their normal activities. While it is probably not damaging to carry out single surveys, or repeated monitoring at long intervals, subjecting particular territorial individuals to frequent playback is unethical and to be avoided. Bird species respond differently to playback. Some call back but stay where they are; others come toward the observer and either call or investigate silently; yet others show no obvious response at all. It is important that you test the technique on the species you want to target before embarking on a systematic playback survey. Using tape recordings for bird community surveys, as done by Parker, is very time-consuming: you need to spend at least as long analysing the recording as making it in the first place. Perhaps for this reason the technique has rarely been used in Africa.

7.3.10 Territory mapping For detailed population studies, it is, in theory, possible, but very time- consuming, to map the territory of individual singing males. These are then plotted on a map of the study site, and give precise information as to density (e.g. Terborgh et al.,1990). However, there are many practical difficulties involved. In eastern African forests, it is sometimes difficult to detect singing males. Furthermore, breeding seasons are not always as clear cut as in the northern hemisphere, where there is strong, clear, seasonal breeding and where males are frequently more visually and vocally conspicuous. Moyer (1993) provides a comparative study in which densities of forest birds were estimated using territory-mapping, mist netting and direct counts of non-territorial birds. He concludes that territory mapping is the most promising method when estimates of territory size and absolute density of breeding pairs are needed.

7.3.11 Special considerations Forest birds present a number of challenges to surveyors. Remember to plan carefully and think clearly about the purpose and objective of your survey, as this will determine the information that you need to obtain, and thus the

149 methodology that is most appropriate. If you are mainly interested in, for example, shifts of guilds between different forest types, then it will not matter much if you fail to detect certain cryptic or reclusive species; however, if your focus is on population sizes of those species, then that is a different matter. Ideally, you need to know something about the birds you are looking for before you start. For example, many fruit-eating birds congregate at particular fruiting trees. If there are only one or two such trees in your area, a standard transect-based survey might under- or over-estimate abundance. A better method might be to locate the trees and do a total count of all the frugivores that are using them. Very cryptic and silent birds can sometimes only be censused by mist netting. Some reclusive species show strong seasonal patterns of singing (e.g. East Coast Akalat, Sheppardia gunningi: Nemeth & Bennun, 2000). Unless you find out when these times are, you might easily make erroneous comparisons between sites. Nocturnal birds can often be censused by calls at night, but again frequently show variations during the night, and within and between months in their calling activity. This is one of the reasons why surveys need to be standardised; for instance, all census work might be done during the early hours of the evening, around full moon, during the dry season. In tall forests, canopy species can be extremely hard to identify. A telescope can often help, but many birds might still be missed. You need to budget more time for survey work in tall forests. You should also be careful of comparing population density estimates for canopy birds between low and tall forests. Species that fly above the canopy are even more of a problem: they might need to be censused using species-specific techniques, such as counts from raised points. These are especially useful for birds that fly to and from roosts each day, like parrots and hornbills. In some forests, birds form mixed-species flocks – feeding aggregations of several, often many, different species that move through the vegetation together. Flocks may move very fast and range over large tracts of forest. Where mixed-species flocking is common, survey techniques that cover relatively small areas, like point counts, may not work well. With transects or timed species-counts, moving rapidly through the forest until you locate a flock may be the best approach. Distance methods can be difficult to apply for mixed- species flocks. One approach is to try to estimate the distance to the flock centre and calculate the density of flocks, rather than of individual species. For some scarce species, such as large forest raptors, conventional census methods may be hard to apply. Where these species have big, conspicuous nests, one approach is to locate and count active nests rather than individual birds.

150 7.4 Specimen handling Casualties among mist-netted birds should be very rare if the nets are properly handled, but they do sometimes occur. Although specimen collection is not usually a goal of surveys, the value of the casualty is maximised if it can be collected for a museum. Sometimes it may be necessary to take voucher specimens (for instance, to confirm species identification or range extension, or if a suspected new taxon is suspected). Specimen preservation is a topic in itself, and cannot be covered properly here.

Equipment If voucher specimens or net casualties are to be collected, specimen collecting permits and material are necessary. For wet specimens: 10% formalin or 70% alcohol (c.5 litres) container for holding preserved specimens hypodermic syringe and needle For collecting dry specimens, more elaborate equipment (e.g. dissecting kit, needles and thread, borax powder) is required. Consult museum personnel or a specialist text on specimen preparation (see Section 4.4).

Procedure and recording Museums have traditionally favoured skins of birds, but more are now realising the value of specimens preserved in fluid (10% formalin or 70% alcohol), which is a technique much more suited to the non-specialist. After washing the bird with soapy water to reduce the water repellency of the feathers, it is immersed in 10% formalin of at least three times the volume of the bird. A label on waterproof paper, with a numbered code for the specimen, should be attached to the left tarsus. See Chapter 4 (on small mammals) for additional information. Before preserving the specimen, record biometrics and other information, including soft-part colours (see Fig. 7.4), on the specimen record form (Form 7.4), or in a hardback book. If the specimen is dissected before preservation, it is important to note the sex and the stomach contents. Specimens are very valuable and it is worth recording as much information on each one as possible.

151 Fig. 7.3: Standard measurements for birds

7.5 Health and safety Very little work seems to have been done on the potential risk to investi- gators of birds as regards disease transmission, either direct or by arthropod transmission. Birds do carry ectoparasites such as fleas, mites, and ticks, and the latter are known to harbour tick-borne diseases that affect humans. It is prudent for those handling birds to avoid their ectoparasites and to wear surgical gloves and a mask when dissecting specimens and/or preparing study skins. After handling birds during mist netting and ringing, make sure you wash your hands thoroughly before eating or drinking. Bats sometimes become caught in mist nets and need to be carefully freed. Follow the advice given in Chapter 4 on handling small mammals.

7.6 Conclusions The following section is adapted from Bennun & Fanshawe (1998, p.16) and summarises the advice given above on choosing the most appropriate method: Choose your methodology carefully. For most survey and monitoring work, only relative abundance measures are required. Distance-sampling using variable-width methods (Buckland et al., 1993) is difficult, demanding, and generally only useful for specialised purposes. Because density estimates can be calculated for only the most common species, compiling data for guilds or forest-dependence categories is difficult. These methods do have the great advantage that they take into account the relative ease or difficulty of detecting birds in particular vegetation types. If bird calls are used in conjunction with sightings, the problem of visual detectability is diminished. Also, if there are big enough changes in vegetation between sites or monitoring visits to greatly affect detectability, it is likely that changes in the bird community will be sufficiently large to be picked up with a simple method.

152 Traditional transects can be difficult to place in forests. Timed species- counts, which do not depend on a straight-line route, are one solution, but they have substantial disadvantages too. Some of these can be overcome by simple modification. For comparative survey and monitoring, however, the biggest drawback concerns the calculated abundance indices, which are not arithmeti- cally tractable – therefore, one cannot simply calculate overall indices for, say, feeding guilds. Timed transects have been tested in several Kenyan forests, and combine the flexibility of timed species-counts with the additive indices of a standard, fixed-width transect. All of these visual/aural methods work best for canopy and mid-level birds. For undergrowth species, standardised mist netting is appropriate for survey and monitoring work, although it is labour-intensive and time-consuming. By confining point counts and timed transects to birds above a certain level (3m has been used in Kenyan forests), it is possible to differentiate undergrowth and higher-level birds. As the results of these studies show, bird community changes can be different at these two levels. This is to be expected, given that structural change after logging can affect high- and low-level vegetation in very dissimilar ways.

153 7.7 References

Allport, G.A., Ausden, M.J., Fishpool, L.D.C., Hayman, P.V., Robertson, P.A. & Wood, P. (1996). Identification of Illadopsis spp. in the Upper Guinea forest. Bull. Afr. Bird Club 3: 26–30.

Barlow, C., Wacher, T. & Disley, T. (1997). A Field Guide to the Birds of the Gambia and Senegal. Pica Press, London, UK.

Bennun, L.A. (1994). Identification first aid... Kakamega greenbuls. Kenya Birds 2(2): 48–52.

Bennun, L.A. (1999). Threatened birds and rural communities: Balancing the equation. In: Proc. 22 Int. Ornithol. Congr., Durban: pp 1546–1555 (Ed. by N.J. Adams & R.H. Slotow). BirdLife South Africa, Johannesburg.

Bennun, L.A. (2000). Monitoring bird populations in Africa: an overview. Ostrich 71: 214–215.

Bennun, L.A. (2001). Long-term monitoring and conservation of tropical wetlands: high ideals and harsh realities. Hydrobiologia. 458: 9–19.

Bennun, L. & Fanshawe, J. (1998). Using forest birds to evaluate forest management: an East African perspective. In: African Rainforests and the Conservation of Biodiversity, pp. 10–22 (Ed. by S. Doolan). Earthwatch Europe, Oxford, UK.

Bennun, L. & Njoroge, P. (1999). Important Bird Areas in Kenya. East Africa Natural History Society, Nairobi.

Bennun, L.A. & Waiyaki, E.M. (1993). Using timed species-counts to compare avifaunas in the Mau Forests, south-west Kenya (summary only). Proc. Pan-Afr. Ornith. Congr. 8: 366.

Bennun, L., Dranzoa, C. & Pomeroy, D. (1996). The forest birds of Kenya and Uganda. J. E. Afr. Nat. Hist. 85: 23–48.

Bibby, C., Jones, M. & Marsden, S. (1998). Expedition Field Techniques: Bird Surveys . Expedition Advisory Centre, London, UK.

Bibby, C.J., Hill, D.A., Burgess, N.D. & Mustoe, S. (2000). Bird Census Techniques. 2nd edn. Academic Press, London, UK.

Brooks, T., Lens, L., Barnes, J., Barnes, R., Kihuria, J.K. & Wilder, C. (1998). The conservation status of the forest birds of the Taita Hills, Kenya. Bird Cons. Intl 8: 119–139.

Buckland, S.T., Anderson, D.R., Burnham, K.P. & Laake, J.L. (1993). Distance Sampling: Estimating Abundance of Biological Populations. Chapman & Hall, London, UK.

Christy, P. & Clarke, W.V. (1998). Guide des Oiseaux de Sao Tomé et Principe . Ecofac, Gabon.

Colwell, R.K. & Coddington, J.A. (1994). Estimating terrestrial biodiversity through extrapolation. Phil. Trans. Roy. Soc. Lond ., Ser. B. 345: 101–118.

Dranzoa, C. (1993). Birds of fragmented forest areas in southern Uganda. Proc. Pan-Afr. Ornith. Congr. 8: 334–340.

Du Plessis, M.A. (1995). The effects of fuelwood removal on the diversity of some cavity-using birds and mammals in South Africa. Biol. Cons. 74: 77–82.

154 Fanshawe, J.H. (1995). The Effects of Selective Logging on the Bird Community of Arabuko- Sokoke Forest, Kenya. DPhil thesis, University of Oxford, Oxford, UK.

Fjeldså, J. (1999). The impact of human forest disturbance on the endemic avifauna of the Udzungwa Mountains, Tanzania. Bird Conservation International 9: 47–62.

Fry, C.H., Keith, S. & Urban, E.K. (Eds.) (2000). The Birds of Africa. Vol. VI. Academic Press, London, UK.

Furness, R.W. & Greenwood, J.J.D. (Eds.) (1993). Birds as Monitors of Environmental Change. Chapman & Hall, London, UK.

Goldsmith, F.B. (Ed.) (1991). Monitoring for Conservation and Ecology. Chapman & Hall, London, UK.

Hall, J.B. & Rodgers, W.A. (1986). Pole cutting pressure in Tanzanian forests. For. Ecol. Mgmt 14: 133–140.

Howard, P.C., Viskanic, P., Davenport, T.R.B., Kigenyi, F.W., Baltzer, M., Dickinson, C.J., Lwanga, J.S., Matthews, R.A. & Balmford, A. (1998). Complementarity and the use of indicator groups for reserve selection in Uganda. Nature 394: 472–475.

Howes, J. & Bakewell, D. (1989). Shorebird Studies Manual. AWB Publication No. 55. Asia Waterfowl Bureau, Kuala Lumpur, Malaysia.

Kosgey, D.K. (1998). Status and Habitat Choice of Turner’s Eremomela Eremomela turneri (van Someren 1920) in South Nandi Forest Reserve, Kenya. MPhil thesis, Moi University, Eldoret, Kenya.

Lens, L. & van Dongen, S. (1999). Evidence for organism-wide asymmetry in five bird species of a fragmented afrotropical forest. Proc. R. Soc. Lond., Ser. B 266: 1055–1060.

Lens, L., Galbusera, P., Brooks, T., Waiyaki, E. & Schenk, T. (1998). Highly skewed sex ratios in the critically endangered Taita Thrush as revealed by CHD genes. Biodiversity & Conservation 7: 869–873.

MacKinnon, J. & Phillips, K. (1993). A Field Guide to the Birds of Sumatra, Java and Bali. Oxford University Press, Oxford, UK.

Mackworth-Praed, C.W. & Grant, C.H.B. (1955, 1957). African Handbook of Birds. Series I: Volumes I & II. Birds of Eastern and North Eastern Africa. Longman, London, UK.

Mackworth-Praed, C.W. & Grant, C.H.B. (1962, 1963). African Handbook of Birds. Series II: Volumes I & II. Birds of the Southern Third of Africa. Longman, London, UK.

Mackworth-Praed, C.W. & Grant, C.H.B. (1970, 1973). African Handbook of Birds. Series III: Volumes I & II. Birds of West Central and Western Africa. Longman, London, UK.

Magurran, A. (1988). Ecological Diversity and its Measurement. Croom Helm, London, UK.

Moyer, D.C. (1993). A preliminary trial of territory mapping for estimating bird densities in Afromontane forest. Proc. Pan-Afr. Ornith. Congr. 8: 302–311.

Mulwa, R.K. (2001). The population status and ecology of Taita White-eye Zosterops Poliogaster silvanus (Peters and Loveridge, 1935) in the fragmented forests of Taita Hills, Kenya . MSc thesis, Kenyatta University.

155 Musila, F. (2001). Status, ecology and conservation of the Sokoke Pipit, Anthus sokokensis, in Arabuko-Sokoke Forest, Malindi, Kenya. MSc thesis, Nairobi University.

Nemeth, E. & Bennun, L. (2000). Distribution, habitat selection and behaviour of the East Coast Akalat (Sheppardia gunningi sokokensis) in Kenya and Tanzania. Bird Cons. Intl 10: 115–130.

Newman, K. (2000). Newman’s Birds of Southern Africa. Green Edition. Southern Book Publishers, Halfway House, South Africa.

Newmark, W.D. (1991). Tropical forest fragmentation and the local extinction of understorey birds in the eastern Usambara Mountains, Tanzania. Conserv. Biol. 5: 67–78.

Newton, I. (1994). The role of nest-sites in limiting the numbers of hole-nesting birds: a review. Biol. Cons. 70: 265–276.

Owiunji, I. (1996). The Long-term Effects of Forest Management on the Bird Community of Budongo Forest Reserve, Uganda. MSc thesis, Makerere University, Kampala, Uganda.

Owiunji, I. (2000). Changes in avian communities of Budongo Forest Reserve after 70 years of selective logging. Ostrich 71(1-2): 216–219.

Oyugi, J.O. (1998). Tropical Forest Fragmentation and Avian Population Changes in Kakamega Forest, Kenya. MPhil thesis, Moi University, Eldoret, Kenya.

Parker, T.A. III. (1991). On the use of tape recorders in avifaunal surveys. Auk 108: 443–444.

Pomeroy, D. (1992). Counting Birds: a Guide to Assessing Numbers, Biomass and Diversity of Afrotropical Birds. AWF Technical Handbook Series no. 6. African Wildlife Foundation, Nairobi, Kenya.

Pomeroy, D. & Dranzoa, C. (1997). Methods of studying the distribution, diversity and abundance of birds in East Africa – some quantitative approaches. Afr. J. Ecol. 35: 110–123.

Pomeroy, D.E. & Tengecho, B. (1986). Studies of birds in a semi-arid area of Kenya. III – The use of ‘timed species-counts’ for studying regional avifaunas. J. Trop. Ecol. 2: 231–247.

Remsen J.V. Jr. & Good, D.A. (1996). Misuse of data from mist net captures to assess relative abundance in bird populations. Auk 113: 381–398.

Serle, W. & Morel, G. (1992). A Field Guide to the Birds of West Africa. Collins, London, UK.

Sinclair, I., Hockey, P. & Tarboton, W. (1997). SASOL Birds of Southern Africa. 2nd edition. Struik New Holland, Cape Town, South Africa.

Stattersfield, A.J., Crosby, M.J., Long, A.J. & Wege, D.C. (1998). Endemic Bird Areas of the World: Priorities for Biodiversity Conservation. BirdLife Conservation Series no. 7. BirdLife International, Cambridge, UK.

Stevenson, T. & Fanshawe, J. (2001). A Field Guide to Birds in East Africa. T & AD Poyser, London, UK.

Stork, N.E. & Samways, M.J. (1995). Inventorying and monitoring. In: V.E. Heywood (Ed.) Global Biodiversity Assessment, pp. 457–543. UNEP and Cambridge University Press, Cambridge, UK.

Terborgh, J., Robinson, S.K., Parker, III, T.A., Munn, C.A. & Pierpont, N. (1990). Structure and organisation of an Amazonian forest bird community. Ecol. Monogr. 60: 213–238.

156 Thirgood, S.J. & Heath, M.F. (1994). Global patterns of endemism and the conservation of biodiversity. In: Systematics and Conservation Evaluation, pp, 207–227. (Eds. P.L. Forey, C.J. Humphries and R.L. Vane-Wright). Clarendon Press, Oxford, UK.

Tomas Vives, P. (Ed.) 1996. Monitoring Mediterranean Wetlands: A Methodological Guide. MedWet Publication. Wetlands International, Slimbridge, UK and ICN, Lisbon, Portugal.

Turner, D.A. & Zimmerman, D.A. (1979). Field identification of Kenya greenbuls. Scopus 3: 33–47.

Tye, A. (1993). Forest and bird conservation in the East Usambara Mountains, north-east Tanzania. Proc. Pan-Afr. Ornith. Congr. 8: 287–292.

Van Perlo, B. (1996). Collins Illustrated Checklist: Birds of Eastern Africa. Harper Collins, London, UK.

Van Perlo, B. (1999). Collins Illustrated Checklist: Birds of Southern Africa. Harper Collins, London, UK.

Virani, M. (2000). Distribution and population size of the Sokoke Scops Owl Otus ireneae in the Arabuko-Sokoke Forest, Kenya. In R.D. Chancellor & B.-U. Myberg (eds), Raptors at Risk. Proceedings of the Fifth World Conference on Birds of Prey and Owls. (Midrand, South Africa, 4–11 August 1998), pp. 795–801. Hancock House, Surrey, Canada.

Williams, J.G. & Arlott, N. (1980). A Field Guide to the Birds of East Africa. Collins, London, UK.

Zimmerman, D.A., Turner, D.A. & Pearson, D.J. (1996). Birds of Kenya and Northern Tanzania . Princeton University Press, Princeton, NJ, USA.

157 Form 7.1: Timed Species-count Records for Forest Birds

Surveyor: Field sheet ref: Date: (total observers): (dd/mm/yy)

Address:

Survey site: Altitude: Aspect:

Latitude: Longitude: UTM (if available):

Vegetation: Human disturbance:

Season: Weather: Lunar phase: Temperature:

Start time: End time: Other:

Time Species Within 25m? Above 3m? Cue Score

Cue = H (heard) or S (seen) 158 Form 7.2: Bird Recording Sheet for Transect Counts

Surveyor: Field sheet ref: Date: (total observers): (dd/mm/yy)

Address:

Survey site: Altitude: Aspect:

Latitude: Longitude: UTM (if available):

Vegetation: Human disturbance:

Season: Weather: Lunar phase: Temperature:

Transect length: Start time: End length: Other:

Time Species No. of birds Additional observations: in groups

Write a description on the back of the sheet (noting things such as general size and colour, beak colour and shape, etc.) for any species you cannot identify with confidence. This can be used later for comparison with illustrations and descriptions in standard reference works.

159 e before this to ease the (dd/mm/yy) total list covering the schedules submitted. ’ do not send photocopies. The ringing year runs from 01 July to 30 June. All schedules, whether The ringing year runs from 01 July to 30 June. do not send photocopies. – arsus Weight Other Fat bp Primary Secndry Tail Body Net Time Init Notes Head T complete or not, should be sent to the ringing organizer in early July each year but completed schedules may at any tim end-of-the-year paper work. It is most useful if you can prepare and send in a species # SexIf you re-ring a bird, record the original ring number in Notes column. Return this original schedule to the ringing organiser biometrics MLT MLT MLT MLT Location:Other: Latitude: Longitude: Weather: Nets open: nets closed: Time Form 7.3: Bird Mist Netting and Ringing Sheet Collector:Ring Species Age Address: Wing Date: nets open: Time

160 Form 7.4: Specimen Records: birds Specimen reference number: Collection accession number1:

Collector: Date: Time: (dd/mm/yy)

Address:

Collecting site: Altitude:

Latitude: Longitude: Slope:

Additional notes:

Species: Sex: Age:

Ectoparasites: Endoparasites:

Measurements: Soft part colours Wing (mm) Tarsus (mm) Bill (mm) Iris Bill Tarsus Foot

Other:

Moult:

Tissue sample(s) preserved: Blood sample(s) preserved:

Stomach contents: Component: Percentage:

Remarks/Other

1 To be filled in by Museum

161 Responsible forest management requires accurate information about a broad range of species. However, time is too short, and resources too few, for all forest areas to be considered by specialist survey teams. This manual provides an overview of the methods which can be used to gather information. It is designed to be carried into the field to guide survey work, and enable the user to consider the full range of vertebrates, excluding fish, found in African forests. It also explains the basic techniques and basic standards needed for the development of essential inventory and monitoring programmes.

The manual is particularly aimed at: • people carrying out short reconnaissance surveys and expeditions; • undergraduate and graduate students carrying out project and thesis work; • research departments of forest, wildlife and national parks departments; • forest and wildlife managers and technicians with responsibility for monitoring biodiversity.

Published by Earthwatch 57 Woodstock Road, Oxford, OX2 6HJ, UK Tel: +44 (0)1865 318825, Fax: +44 (0)1865 311383, Email: [email protected] www.earthwatch.org/europe Registered Charity 327017

ISBN No. 0-9538179-4-6

Publication of this manual has been made possible through a generous donation from Rio Tinto plc