Spatial and matrix influences on the biogeography of taxa in forest fragments in central

Perpetra Akite

Dissertation for a cotutelle award of Doctor of Philosophy Degree of Makerere University, Uganda and University of Bergen, Norway

Makerere University University of Bergen

2016

Department of Biological Sciences, Makerere University

Department of Biology, University of Bergen

ii

DECLARATION OF ORIGINALITY

This is my own work and it has never been submitted for any degree award in any University

iii

TABLE OF CONTENTS

DECLARATION OF ORIGINALITY...... iii

LIST OF CONTENTS...... iv

ACKNOWLEDGEMENTS...... vi

LIST OF PAPERS...... vii

Declaration of authors’ contributions…………………….…...……………...……...viii

ABSTRACT...... x

BACKGROUND...... 1

Problem statement...... ……….2

Objectives...... 3

Research questions...... 4

INTRODUCTION...... 5

Forest fragmentation and Biodiversity...... 6

Does the Matrix matter?……………………………………………………………...... 9

Roles of in ecosystem dynamics……………………………………..……… 12

How much is known about insects in Uganda?………………………………………14

STUDY AREAS AND SITES...... 15

MATERIALS AND METHODS…………………………………………………….………17

Study protocols…………………………………………………………………………….…17

Paper I…………………………………………………..……………………….……17

Paper II…………………………………………………………………………..……19

Paper III……………………………………………………………………………….21

Paper IV…………………………………………………………………………….…23

RESULTS AND DISCUSSION...... 24

Temporal dynamics of forest biodiversity...... 26

iv

Species distribution along the vertical continuum...... 29

Habitat use by insects in a patch-matrix landscape...... 30

The need for focused sampling in face of limited inventory………………….………33

Challenges of biodiversity conservation in modified forest landscapes in Uganda…. 34

The social and political realities and their effects on forest protection in Uganda…...36

CONCLUSIONS...... 39

RECOMMENDATIONS...... 40

REFERENCES...... …….…..42

PLATES SHOWING REPRESENTATIVE SPECIES OF DIFFERENT GROUPS

INDIVIDUAL PAPERS I – IV

APPENDICES OF LISTS OF SPECIES OF DIFFERENT INSECT TAXA

v

ACKNOWLEDGEMENTS

This study was supported by the Norwegian Research Council (FRIMUF programme) funded the study through the MATRIX project (# 184912) and supported by the University of

Bergen–Makerere University Collaboration. I am grateful to my supervisors for guiding me through my doctoral study. You have been inspiring and supportive to my research decisions.

Your invaluable expertises in the different aspects of this study are highly regarded. I am indebted to Vigdis Vandvik PI-MATRIX project) for her enthusiastic, encouraging and valuable discussions throughout this work. Your fast feedback on manuscripts, open minded face to face discussions and very constructive criticisms were instrumental in making this study end a reality. Amy Eycott, it’s said better late than never and your coming on board the

Matrix project was an honour for me. Thank you for your tireless efforts, endless reading of manuscripts, mapping and all other help you extended to me. I am especially grateful to Paul

Waring and Hugh Rowell for all the expert field trainings, Ian Kitching and Alessandro Giusti from the Natural History Museum London, Torben Larsen (R.I.P) for identifying difficult specimens, research associates Therese Kronstad and Jenny Reiniö and our local assistants, colleagues at Ecology and Environmental Research Group (University of Bergen) and the Department of Biological Sciences, Makerere University for a friendly and inspiring work atmosphere, Cathy Jenks; my welfare hero in Bergen, Josephine Esaete for all the ‘big sister’ counsel and fantastic coordination, Ulf Bjelke for financial help following gruesome road accident and for sharing my passion for insects and photography, Janet McCrae, Joseph

Chipperfield, Mikail Erdogan and Jan Skogvang. Above all, I am grateful to God for holding everything together to the very end.

This thesis is dedicated to a dear friend, Petter Hansen (the cruel hand of death did not let you see the end of this journey), and to all who appreciate nature.

vi

LIST OF PAPERS

The thesis is based on the following four papers which will be referred to by their Roman numerals hereafter.

Paper I: Akite, P., Telford, R.J., Waring, P., Akol, A.M. & Vandvik, V. (2015) Temporal

patterns in Saturnidae (silk ) and (hawk moth) assemblages in

protected forests of Central Uganda. Ecology and Evolution 5 (8): 1746–1757.

Paper II: Akite, P., Akol, M.A., Kronstad, T., Vandvik, V & Telford, R.J. Vertical

distribution of fruit-feeding in three protected forests in Central Uganda.

Manuscript under revision.

Paper III: Akite, P., Akol, M.A., Eycott, A.E., Vandvik, V. & Telford, R.J. The use of matrix

habitats by forest insect species. Manuscript.

Paper IV: Akite, P. & Rowell, C.H.F. (2013) Oshwea dubiosa rediscovered in Uganda.

Journal of Orthoptera Research 22 (1): 45–49.

vii

Declaration of authors’ contributions

Paper I: P. Akite, R.J. Telford, P.Waring, A.M. Akol & V. Vandvik. Temporal patterns in

Saturnidae (silk moth) and Sphingidae (hawk moth) assemblages in protected forests of

Central Uganda.

Akite, P.: Compiling historical data, survey design, field work, laboratory work, data processing, statistical analyses, writing

Telford, R.J.: Project design, analytical design, data processing, statistical analyses, editing

Waring, P.: Survey design, field work, editing

Akol, A.M.: Analytical design, statistical analyses, editing

Vandvik, V.: Project design, analytical design, statistical analyses, editing

Paper II: Akite, P., Akol, M.A., Kronstad, T., Vandvik, V & Telford, R.J. Vertical distribution of fruit-feeding butterflies in three protected forests in Central Uganda.

Akite, P.: Survey design, field work, laboratory work, data processing, statistical analyses, writing

Akol, M.A.: Analytical design, editing

Kronstad, T.: Field work, laboratory work, data processing, editing

Vandvik, V.: Project design, analytical design, statistical analyses, editing

Telford, R.J.: Project design, analytical design, data processing, statistical analyses, editing

viii

Paper III: Akite, P., Akol, M.A., Eycott, A.E., Vandvik, V. & Telford, R.J. The use of matrix habitats by forest insect species.

Akite, P.: Survey design, field work, laboratory work, data processing, statistical analyses, writing

Akol, M.A.: Editing

Eycott, A.E.: Data processing, statistical analyses, editing

Vandvik, V.: Project design, analytical design, statistical analyses, editing

Telford, R.J.: Project design, analytical design, statistical analyses, editing

Paper IV: Perpetra Akite & C.H.F. Rowell (2013). Oshwea dubiosa rediscovered in Uganda.

Akite, P.: Field work, data processing, editing

Rowell, C.H.F.: Field work, laboratory work, data processing, writing

ix

ABSTRACT

How best to manage forest patches, mitigate the consequences of forest fragmentation, and enable landscape permeability are key questions facing conservation scientists and managers. In Uganda, most protected forests have undergone considerable changes from a range of human activities in recent decades. As such once continuous forests now exist as ‘islands’ (patches) in a matrix of non-forest habitats. In fragmented forest landscapes, the capacity of matrix habitats to support forest species varies. However, very little is known of how faunal communities and in particular insect assemblages utilize resources within the patch-matrix landscape. This study explored spatial and temporal patterns of diversity and distribution of butterflies, and grasshoppers in the patch-matrix landscapes of three protected forests: Mabira (largely stable and recovering from past encroachment, albeit with minor illegal logging, but disturbed at the edges), Zika and Mpanga (relatively undisturbed and unchanged internally, but with substantially altered matrix), all in central Uganda. All together, 25156 individuals (326 species) of butterflies, 1131 individuals (41 species) of silkmoths, 1564 individuals (44 species) of hawkmoths and 2173 individuals (49 species) of grasshoppers were recorded in the forests and surrounding matrix habitats. There was a marked decline in species richness and abundance over time and across the land-use gradient: from mature forests to mixed gardens. The decline was more evident for forest specialist species, while generalist species mostly showed the reverse trend. This decline was observed even in relatively stable Zika and Mpanga forests. Along the vertical stratum, there were more individuals in the understorey compared to the canopy, but higher proportions of specialists were in the canopy. One strategically directed survey in Mabira resulted in the re-discovery of Oshwea dubiosa only previously known from a single holotype female from West-Central Democratic Republic of Congo. This has added to the scientific knowledge of the species, its range and habitat affiliation. Although results of this study underscore the value of protected forests in maintaining biodiversity, it also acknowledges the importance of some of the altered habitats such as cardamom and shade coffee plantations that maintain relic forest species. Therefore detailed knowledge of species and threats within patch-matrix landscape is critical to site prioritization and conservation planning. Since protected forests are ecologically linked to their surrounding habitats, failure to stem broad-scale loss and degradation of the matrix habitats will consequently increase the likelihood of serious biodiversity declines and subsequent extinction.

x

1.0 BACKGROUND

Although tropical forests support vast amounts of biodiversity, much of that biodiversity is now imperiled by past and on-going forest loss and degradation (MEA 2005, FAO 2011).

Throughout the world forests have been severely reduced in extent and increasingly fragmented, with continued rapid human population growth, irresponsible logging, and conversion of land for agricultural purposes remaining major threats (MEA 2005). In many parts of the developing world, deforestation continues to accelerate in tandem with poverty and high levels of population growth posing a much greater challenge to transition to sustainable forest management (Rodrigues et al. 2003). This results in a decrease in the possibility of faunal interchange between increasingly isolated forest patches, ultimately reducing the overall population size and probability of persistence (Fahrig 2003).

The widespread destruction of tropical forests is widely thought to be precipitating a global extinction crisis and a mass extinction is widely anticipated if these losses should continue unabated (Pimm & Brooks 2000, Dirzo & Raven 2003, Sodhi et al. 2004, MEA

2005). There is therefore a need to understand the consequences of such environmental changes for a given taxonomic group especially in areas such as: (i) documentation of response of different taxa to anthropogenic disturbances, (ii) unravelling and understanding mechanisms driving change in biodiversity, (iii) understanding the consequences of biodiversity loss on critical ecological processes, and (iv) how to translate biodiversity information into effective and practical policies for decision makers. These challenges are increasingly being met through the use of ecological indicator assemblages – groups of species whose presence and abundance in a given area provide a useful gauge for measuring and interpreting responses to changing environmental situations (e.g. Beccaloni & Gaston

1994, Tscharntke et al. 1998, Schulze & Fiedler 1999).

1

Despite their enormous contribution to tropical diversity and ecosystem functions, knowledge of endangered insects lags behind that of vertebrates and vascular (Thomas et al. 2004, Lewinsohn et al. 2005), and attempts to quantify losses among insects have often been hampered by lack of data (New 2004, Thomas 2005). Insects are highly susceptible to the adverse effects of forest fragmentation (Didham et al. 1996), exceptionally sensitive to environmental change (e.g. Kremen 1994, Koh & Sodhi 2004) and widely responsible for a wide array of ecosystem process (e.g. Yang & Gratton 2014). The study of insects in fragmented forests is still in its infancy and there are only a few known definite trends for some well-studied groups. Many tropical species are locally endemic or are rare and with patchy distribution which predisposes them to increased extinction risk when habitats are modified (e.g. Terborgh 1992, Koh 2007). Consequently, conservation of many such species will depend on the capacity of fragmented forests to support their populations.

1.1 Problem statement

Human dominated landscapes cover a majority of land area in most parts of the world

(Ricketts et al. 2001) and fragmented habitats under human land-use regimes have become increasingly important for the conservation of biodiversity (Hilt et al. 2006). Despite great concern about the effects of fragmentation on biodiversity, quantitative studies are still scarce with respect to many major groups and important environments. Well–studied natural reference sites are few and the distribution of diversity both vertically within the forest profile and amongst forest habitats remains poorly understood.

Insects represent more than half of global biodiversity (e.g. Heywood 1995, Dunn

2005) and are very sensitive to habitat changes, offering good models for understanding biogeographical processes and dynamics, and specifically in continuous areas where the matrix is structurally more similar to the remnant patches (e.g. Franklin & Lindernmayer 2

2009, Prevedello & Vieira 2010). There are only a few and inconclusive studies on the spatial and matrix effects on insect taxa in the forest fragments and their surrounding habitats (e.g.

Jules & Shahani 2003, Campbell et al. 2011). This presents an important gap that this study was designed to address; to show the potential outcomes of habitat fragmentation and patch- matrix dynamics on the conservation and long-term survival of species. Butterflies, moths and grasshoppers were chosen for this study to explore the patch-matrix dynamics.

1.2 Objectives

The main objective of this study was to assess the patterns of diversity of three insect taxa in different forest fragments/patches and surrounding habitats and their relation to habitat heterogeneity, disturbance gradients and how anthropogenic impacts may be altering these patterns.

Specific objectives of the study were;

1. To assess changes in patterns of diversity, distribution and composition of

target taxa over time within the forest fragments.

2. To assess vertical stratification as basic niche dimension in forest ecology and

how it influences diversity, distribution and composition of target taxa within

the forests fragments.

3. To assess influence of different land-use types on the diversity, distribution and

composition of target taxa; from forest to home gardens.

4. To evaluate how different land-use types contribute to the maintenance of

forest dependent species of the three insect taxa.

3

1.3 Key questions that guided the study included:

 What changes have taken place over time in the diversity, distribution and

composition of insect taxa within the selected forest fragments?

 Do the local insect populations or assemblages in the forest fragments show consistent

responses to disturbance?

 Do the surrounding habitats‘ quality/ type affect species diversity, distribution and

composition of the selected insect groups?

 Can particular group of insects/ species be identified as reliable indicators of

ecological disturbance?

 To what extent do the responses of the selected insect groups reflect the responses of

vegetation?

4

2.0 INTRODUCTION

Habitat loss is commonly identified as a major threat to global biodiversity (Summerville &

Crist 2004). According to Ehrlich (1988) and Fahrig (2001), it is regarded as the single greatest threat facing forest plants and communities worldwide. Throughout the world forests have been severely reduced in extent and increasingly fragmented. Continued rapid human population growth, irresponsible logging, and conversion of land for agricultural purposes remain threats even to forests in good condition that are in the public domain.

In ecology, traditional habitat fragmentation studies envision habitats as discrete fragments surrounded by a matrix of very different, inhospitable habitat whereby patch equals habitat; matrix equals wasteland (e.g. Driscoll et al. 2013). However, real-world landscapes often do not fit into such distinct categories. This raises a number of questions (i) what happens to the predictions about species distribution patterns when the patch and the matrix are similar in habitat structure or habitat type?, (ii) how does this change the expectations of traditional island biogeographic models or habitat fragmentation effects?, (iii) how does it change expectations regarding dispersal between patches or survival and reproduction within the matrix?. The understanding of the processes and patterns in fragmented landscapes requires taking into account the suitability of altered habitats around remnants (i.e. the matrix habitats) for the occurrence or dispersal of organisms (Umetsu et al. 2008).

A variety of investigators have examined the effects of landscape context on local species distribution patterns (e.g. Fahrig & Merriam 1994, Kupfer et al. 2006). Therefore, efforts to understand the spatial and ecological dynamics that underpin responses of populations to fragmentations have been a key focus for ecologists and conservationists worldwide (e.g. Turner 2005, Collinge 2009). The concept of using certain species as bio- indicators for identifying specific environmental disturbance is relatively well established

(e.g. Forman & Gordon 1986, Noss 1990). According to Lindenmayer et al. (2000), a basic 5

requirement for the use of many of the proposed biodiversity indicators is the knowledge of what species are in a given habitat, where they are, their autecology and how they might respond to disturbance. However, limited scientific data on indicator species abundance, distribution and habitat utilization exists for many areas. Since no single indicator will possess all of the desirable properties, a set of complementary indicators is required (e.g. Tscharntke et al. 1998, Schulze & Fiedler 1999).

2.1 Forest fragmentation and Biodiversity

Tropical forests harbor the greatest wealth of biodiversity of any terrestrial community

(Sutton 1983). However, this biodiversity is threatened by continued habitat loss and fragmentation (e.g. Franklin 1993), and environmental changes such as logging and conversion of natural habitats for agriculture, human settlement, recreation, amenity or industry, greatly affect organism diversity (McNeely et al. 1995).

Deforestation in the tropics often involves conversion of landscapes with continuous forests to ones with remnant forest patches set in matrix of non-forest vegetation. This has dire consequences on biodiversity both at landscape and forest-fragment levels (e.g. Turner

1996). Factors like fragment size, degree of isolation and time since excision from continuous forest may directly influence biodiversity of the fragment, and subsequently the landscape

(Figure 1)

6

Forest isolation Reduced forest size Newly created edges

Population Increased Micro- Intrusion of human climatic non-forest sub-division pressure changes species

Reduction of Changes in species

population sizes interaction

Loss of species diversity Loss in genetic diversity

Loss of biodiversity

Figure 1. A schematic representation of the theoretical ways in which four forest fragmentation factors (sample effect, forest isolation, reduced forest size and newly—created forest edges) may cause reduction of population sizes which may result ultimately in loss of biodiversity. Arrows indicate causal relations (Adopted and modified from Zuidema et al. 1996). 7

The forests of present-day Eastern Africa are relatively small and numerous, and have been subject to fragmentation and isolation since before the Pliocene (Kingdon 1971). In

Uganda, closed canopy tropical forest has been reduced from 20% of the country‘s land area to just 3%, while 18% of its remaining forest was lost between 1990 and 2000 (Howard et al.

2000). Recent estimates suggest an annual loss rate of 7% for tropical high forest, 5% for woodland and 4% for bushland (Pomeroy & Tushabe 2004). Uganda is exceptionally rich in both flora and fauna (Sayer et al. 1992) which is attributed to the fact that it has seven of the mainland Africa‘s phytochoria. For example, before the introduction of agriculture, regions around are believed to have been covered by forests (Hamilton 1984).

However, in recent years, a lot of forests in Uganda have been lost due to a number of human factors (Obua et al. 2010); including agricultural encroachment, logging (both legal and illegal), and settlements. At present these forests are now fragmented, forming ecological islands surrounded by an agro-ecosystem matrix (for examples, see Plate 1). This destruction is ongoing in both the protected forest reserves as well as in private forests (Obua et al. 2010,

Bulafu et al. 2013). The ultimate result of these anthropogenic influences is the loss of biodiversity in the species rich tropical forests.

The negative consequences of habitat loss and fragmentation to different aspects of biodiversity have been shown by a large number of theoretical and empirical studies, in different environments, and for a large array of taxa (Fahrig 2003). The ultimate effect of habitat fragmentation is extinction of species due to reduction in total habitat area (e.g.

Didham et al. 1996). However, simple empirical studies do not answer the questions we most want to answer about fragmented systems. Therefore a more focused, functional approach to the study of forest fragmentation is required to move beyond the description of pattern and to determine how changes in species assemblages affect ecosystem processes in fragmented forests. 8

2.2 Does the Matrix matter?

According to Gascon et al. (1999), the vulnerability of species in fragmented landscapes may affect their abilities to use the matrix of modified habitats surrounding forest fragments.

Matrix habitats vary considerably in their capacity to support forest species, and this is largely determined by the history and intensity of land use (Lawton et al. 1998), with more disturbed sites generally having lower species richness (Tocher 1998). Studies like Gascon et al. (2000) suggest that the matrix of modified habitats surrounding fragments is likely to have pervasive effects on species communities in forest fragments.

The matrix often acts as a selective filter not an absolute barrier for the movements of species across the landscape where the movement of individuals is determined by the type of vegetation in the matrix. Thus the matrix may exert a strong influence on within-remnant community dynamics (Janzen 1986, Laurance 1990). In addition, species associated with the matrix may invade forest remnants altering the species composition of some taxonomic groups (Tocher et al. 1997). For many forest dependent species, less disturbed or primary forest is clearly of critical importance for long-term survival, either because they avoid modified habitats altogether or because these habitats are clearly suboptimal (e.g. Usher &

Keiller 1998, Slade et al. 2013).

According to Antongiovanni and Metzger (2005), the importance of the matrix in species responses to fragmentation varies depending on the structural features of the matrix and the biological characteristics of the species. Depending on its structure and composition, the matrix can offer food resources and even breeding areas of inferior quality but passable for use as territory. Because some species are able to exploit a matrix, or at least tolerate its effects, populations may remain stable or even increase in size throughout a fragmentation process (Pearson 1993). Species that are the most vulnerable to habitat fragmentation are

9

those that do not tolerate changes in the structure of their habitat and that rarely use the inter- habitat matrix.

10

Plate 1: Selected land-use types found in the survey areas.

Mature forest Secondary forest Cardamom plantation

Swamp forest Coffee plantation Mixed garden

11

2.3 Roles of insects in ecosystem dynamics

Whether measured by species, individuals or biomass (Wilson 1985, Stork 1988, Gaston

1991), invertebrates dominate terrestrial ecosystems (e.g. Dunn 2005). Invertebrates also occupy a wide variety of niches and perform many important ecological functions (e.g. pollination) but receive relatively little attention no doubt partly due to taxonomic difficulties of identifying most invertebrate taxa (e.g. Dunn 2005). Their temporal and spatial distributions span the ranges occupied by many vertebrate and species, including finer- grained patch sizes and geographical distributions, have more complex seasonal and successional sequences and patch dynamics with more rapid turnover (Gaston & Lawton

1988).

The concept of functional biodiversity links species diversity to ecosystem functioning through resource-use patterns (e.g. Tscharntke et al. 2008). In dynamic human-dominated landscapes, only a diversity of species can insure resilience (i.e. the capacity to recover after disturbance). This is because many insects are likely to respond quickly in terms of population declines or losses to habitat changes due to their short generation times and rapid growth rates

(e.g. Sodhi et al. 2010).

Three insect groups – butterflies, moths and grasshoppers – were the focus for this study. These insect taxa were chosen as model organisms for study in three protected forests in central Uganda located in rapidly changing forest-matrix landscapes. Species of these taxa differ in their life histories, habitat preferences, and their habitat fidelity and therefore are expected to respond differently to land-use change.

Butterflies (for examples, see plate 3), number 20000 species worldwide, 4000 in

Africa, and about 1245 in Uganda (Davenport 1996). Taxonomically, they are relatively well known, with most of the species having already been described and their habitat preferences studied. Their present distributions provide a current snapshot of complex biogeographical

12

processes and in a rough and ready way butterflies can be considered indicators of overall biodiversity for entire ecosystems (e.g. Thomas 2005). Butterflies are sensitive indicators of environmental change associated with natural and human-induced disturbances

(e.g. Hamer et al. 1997, Warren & Bourn 2011). Their populations are influenced by changes in local climatic conditions and the availability of host plants for larval and adult stages

(Ehrlich et al. 1972, Thomas et al. 1998). According to Blair and Launer (1997), changes in the abundance, structure, and diversity of butterfly assemblages have been linked to gradients of human-generated disturbances. Notable local declines in the distribution and abundance of butterflies in Europe, North America, and South America have resulted from the conversion of land for human purposes, including residential, commercial, and agricultural development

(Pollard & Yates 1993, Warren 1993). Frequently disturbed environments are considered unstable and unpredictable and as a result have low species diversity, whereas less disturbed, more stable environments are expected to promote high species diversity (Odum 1985).

Moths (for examples, see plates 4 & 5) are a useful group that comprises a diverse and abundant taxon in many forest systems. They play important roles as herbivores, pollinators and prey (Janzen 1987, Barlow & Woiwod 1989). They are host specific (Janzen 1988) and thus serve as indicators of native plant diversity and local land management (Erhardt &

Thomas 1991, Luff & Woiwod 1995). More than 90% of the Lepidopteran species are moths, a majority of which are nocturnal (Young 1997). They have been described as a very useful group for biogeographical and conservation research (e.g. Ricketts et al. 2001).

Grasshoppers (for examples, see plate 6) are a major if not dominant group of herbivorous insects throughout the world, and they often contribute half or more of the total arthropod biomass in the grass layer (Gillon 1983). Grasshoppers can be the dominant herbivores even in ecosystems supporting a high biomass of herbivorous , including domestic stocks (Gandar 1982). Grasshopper assemblages are particularly sensitive to land 13

management (Samways & Moore 1991, Kemp 1992). Their high diversity, functional importance and sensitivity, combined with the ease with which they can be sampled makes grasshoppers excellent bioindicators for use in assessments of ecological change associated with land use (Armstrong & van Hensbergen 1997, Samways 1997).

2.4 How much is known about the selected insect taxa in Uganda?

In Uganda, except for crop pests and vectors such as mosquitoes and biting flies, only a few limited ecological and taxonomic studies have been done on invertebrate groups, and especially on those considered as ecological indicators. These include studies on few insect groups such as dung beetles, butterflies, large moths and grasshoppers.

For butterflies (e.g. Plate 3), some of the studies include; Rogers and van Someren

(1925b), van Someren and Rogers (1932), van Someren (1939), Carcasson (1961, 1963,

1975), D'Abrera (1980), Henning (1988), Bwanika (1995), Kaggwa (1995), Howard and

Davenport (1996), Byaruhanga et al. (2001), Asasira (2003), Molleman et al. (2006) and

Akite (2008), Nyafwono et al. (2014a,b).

For moths, only two large families (Saturnidae and Sphingidae, Plate 4 & 5) have been studied and these studies include A. McCrae (unpublished), Carcasson (1976), Birungi (1995) and Howard and Davenport (1996).

Similarly, very few studies in relation to the ecology of the Orthoptera fauna (e.g.

Plate 6) in Uganda are known. Some studies only focused on the locust groups as part of the agricultural pests and according to Hugh Rowell (pers. comm), not much has been done as far as the forest habitats are concerned, apart from strictly taxonomic work. A total of some 350 species have been recorded from Uganda and about 800 for E. Africa (Rowell pers. database), but their ecology is less studied. In general, tropical forest grasshoppers are food plant

14

specialists, eating a very restricted number of the available plant species (e.g. Rowell 1978).

This makes them unlikely to be able to exploit a matrix of a different vegetation type.

3.0 STUDY AREAS AND SITES

The study was conducted in three forest reserves in Central Uganda (Mabira, Zika and

Mpanga forests) and their surrounding habitats (Figure 2).

Figure 2. Geographical location of the study forest reserves and their matrix habitats in central Uganda

These three forests were chosen because they are surrounded by agricultural landscapes and are situated amidst high population areas. The forests also have baseline species data of the three insect groups (e.g. Howard & Davenport, 1996 for butterflies and moths, Rowell pers.

15

database for grasshoppers), thus providing a unique opportunity to assess changes that could be attributable to impacts on habitats and the surrounding matrix over the years.

Mean annual rainfall in the forests ranges from 1200 to 1600 mm and mean annual temperature is 28°C (MWLE 2002). Rainfall patterns in central Uganda have changed in recent decades, resulting in either less precipitation or alteration in timing of the rainy season

(e.g. MWLE 2002, Williams and Funk 201, Michaelsen and Marshall 2012). This could have a direct effect on food availability and adult emergence.

Transects were located in different land-use types in and around each forest. In

Mabira, five land-use types were considered: the mature forest characterized as a sub-climax forest but influenced by human activities in the past and with fewer disturbances at present

(MWLE, 2008); secondary forests with unregulated exploitation and encroachment and lacking large old trees that characterize the mature forest, and mostly comprised of invasive paper mulberry (MWLE, 2008); cardamom plantation characteristic of agro-forestry but have relatively intact canopy while the understorey is cleared and planted with cardamom

(Elettaria cardamomum); coffee plantation mostly of Arabica coffee intercropped with banana and shade of remnant forest trees like Maesopsis and Albisia sp besides natalensis; and mixed gardens that formed a mosaic of crops close to households. Common crops were cassava, beans, sweet potatoes, maize and yam.

In Zika, sites included the heterogeneous mature forest dominated by Lovoa brownie and Maesopsis eminii, permanent swamp forest dominated by Mitragyna stipulosa and

Erythrina excelsa (Howard & Davenport, 1996), and mixed garden that comprised mostly of maize, beans, sweet potatoes and cassava.

In Mpanga, fours sites were considered including mature forest dominated by Celtis–

Aningeria associations and other trees like Entandrophragma angolense and Pseudospondias microcarpa, the secondary forest which is more open, but have similar and younger tree (e.g. 16

Buxton 1952), and the mixed garden that comprised mostly of beans, cassava, sweet potatoes, and yams.

3.1 MATERIALS AND METHODS

3.1.1 Study protocols

This thesis focuses on how diversity, distribution and composition of insect assemblages within forests changed over time (Paper I); how the assemblages are distributed within the forest stratum (Paper II), and how distribution and diversity vary along a land-use gradient in a forest-matrix landscape (Paper III). It also focuses on identifying groups of insects that can reliably be used as indicators of ecological disturbance (Papers I, II & III) and on exploring the conservation potential of altered habitats in maintaining forest-dependent biodiversity

(Paper III). It finally includes a note on the discovery of new species for Uganda in a well- studied system (Paper IV). All the forests are situated in central Uganda, are surrounded by agricultural landscapes and densely populated areas (Figure 2). Key aspects of sampling strategies and analytical approaches are presented below. Detailed field protocols and sampling of insect variables and statistical methods are described in the respective papers.

3.1.2 Paper I

This study resampled historical data on assemblage to give insights into the long-term ecological integrity of these protected forests in rapidly changing matrix landscapes. This paper uses historical data on two large moth families (Saturnidae & Sphingidae, See plates 4

& 5) that were previously investigated in the selected forests and compared them with present study data collected in the same forests. The study forests are in central Uganda (Figure 1,

Paper I). All investigated sites were located within the mature forest and secondary forest fragments. Given that most Ugandan forests and their matrix landscape have undergone

17

considerable changes in recent decades (Obua et al. 2010); many forests have been lost, some are well protected and have experienced little structural and tree compositional change in recent decades (e.g. Bulafu et al. 2013). The close proximity of these forests to human activities makes them relevant reference points for understanding forest biodiversity dynamics in human dominated landscapes. Saturnidae and Sphingidae were chosen because past data were available and also given their role in ecological processes (e.g. Janzen 1987, Barlow &

Woiwod 1989), their differing life histories and feeding habits (e.g. Haber & Frankie 1989), and the ease of sampling and identification.

The different data sets (i.e. 1970s, 1990s and 2010s) comprise light trap data. Traps were run from dusk to dawn in all survey periods. Although stronger light sources were used in the earlier surveys, the portable 15-watt actinic light source used in the 2010 period is highly effective with an advantage of minimizing cross-habitats attraction (e.g. Muirhead-

Thomson 1991, Schulze & Fiedler 2003). To attain data representative of the species present in the different forests, sampling was done during the wet and dry seasons. Zika forest was sampled in the 1970s, 1990s and 2010s for Saturnidae whereas Saturnidae in Mabira and

Mpanga were only sampled in the 1990s and 2010s. Sphingidae were only sampled in the

1990s and 2010s across all the three forests.

Trap data were pooled for both moth families for each forest per sampling period. The exponent of the bias-corrected Shannon index was used to evaluate species diversity across the sampling periods (Chao & Shen 2003). Individual-based rarefaction curves (Gotelli &

Colwell 2001) were used to evaluate the effectiveness of sampling. For the statistical comparison of the accumulation curves, the rarefied number of species and the 95% confidence interval using bootstrap resampling with replacement was calculated (Colwell

2006). All moth species recorded (Appendices A2 & A3) were categorized into their respective ecological habitat preferences, herein referred to as ecotypes following Howard and 18

Davenport (1996) and Carcasson (1976). The ecotypes include: F) forest-dependent species, f.) forest non-dependent species, G) non-forest species characteristic of open habitats, and W) widespread species. The proportion of the total moth fauna belonging to each ecological habitat preference based on abundance and presence/absence data was calculated, and a Chi- square test for homogeneity performed to test whether the observed differences in richness and abundances of the different ecotypes over time within the two moth families were significant.

3.1.3 Paper II

This study focused on the distribution of fruit-feeding butterflies between the canopy and understorey in three protected forests in central Uganda (Figure 1, Paper II). Butterfly sampling was done using standard baited traps similar to ones used by other studies (e.g.

DeVries et al. 1997, Molleman et al. 2006, Figure 3).

Canopy trap

Understorey trap

Figure 3. Sampling design for fruit-feeding butterflies. Each trap station was fitted with an understorey trap (1–1.5 m height) and canopy trap (15–30 m height). 19

Canopy traps were suspended using a thin nylon rope over branches to create a pulley system that allowed the traps to be lowered and raised from the ground and positioned to sample from within the crown of the trap tree. Traps were baited prior to the first day of sampling with over-ripe bananas which had been mashed, mixed well, and fermented overnight in a container. Traps were checked twice a day but not during or immediately after the rains to avoid damaging the specimens. All observed species were described in terms of their known ecological habitat preferences based on existing knowledge of each species‘ ecology (Davenport 1996). Species were also partitioned into their respective subfamilies to evaluate the vertical structure of the different taxonomic groups within the fruit-feeding guilds. Data from two publications (DeVries 1988, Molleman et al. 2006), with their individual abundances in the canopy and understorey were extracted for some comparisons.

Species data were pooled by forest type per forest and stratum (Appendix 1, paper II).

Members of families that are not fruit-feeders but caught in the traps by chance were not included in the analyses. Rank abundance plots were used to compare species richness and abundance of butterfly assemblages between canopy and understorey in the different forests and forest types. To assess the stratification of fruit feeding butterfly species, the multinomial model (Chazdon et al. 2011), based on relative abundance of individual species in the canopy and understorey habitats was used. A super-majority specialization threshold (K = 2/3 and P =

0.05) was used to evaluate habitat specialisation of individual species and (K = 2/3 and P =

0.005) was used to assess the overall community pattern. To check for trade-off between sensitivity of the model to specialists and classification of higher fractions of total species in the samples, a simple-majority specialization threshold (K = 1/2) was instead used. The vertical distribution of ecotypes and subfamilies (higher taxonomic level) as model indicators to delineate stratification was similarly assessed. A two-tailed Chi-square test for homogeneity (df = 1, p-critical = 0.05), based on the 1-sample proportions test with continuity 20

correction for the sample estimates was performed to assess how standard statistical methods compare to the multinomial model. These frequencies were tested at species level for the different forests, forest types, combined forests data set and the selected publications.

3.1.4 Paper III

To understand how different land-use types affect insect communities, selected insect groups were surveyed in different habitats in the forest-matrix landscape. Transects were established within each of the selected sites in the forests and matrix habitats. In Mabira, transects were established in mature forest (4 transects), secondary forest (3 sites), cardamom plantation (1 site), coffee plantations (2 sites) and mixed gardens (2 sites). In Zika, one transect in each site was established and these included the heterogeneous dry forest, permanent swamp forest and mixed garden. In Mpanga forest, transects were located in mature forest (2 sites), secondary forest (1 site) and one transect in mixed garden.

Baited traps have been used to target specific butterfly subfamilies such as

Nymphalinae, Acraeinae and (Pinheiro & Ortiz 1992). Banana-baited traps were set along transect at intervals. In this paper, data sets include those obtained not only in the forest sites but also those from selected modified habitats surrounding the forest patches where the canopy traps height varied depending on the availability of trees. Butterflies and grasshoppers were sampled using sweep netting on transect at each site walked at a constant pace for two hours. For butterflies, transect walks followed Pollard‘s transect walking technique (Pollard 1977). Netting was carried out only when weather conditions were favourable (dry and warm); attempts were made to identify and record all the individuals seen within the prescribed limits of five metres on either side of the transect. For the moths, low- power light source that minimizes cross attraction of moths between sampled habitats (e.g.

Muirhead-Thomson 1991, Schulze & Fiedler 2003) was used to sample the two moth families 21

considered. The only difference was that the light trap was operated in both forest habitats as well as modified habitats surrounding the forest patches. Grasshopper community structure was assessed through intensive sweep netting to determine the abundance of the grasshoppers, and only adults were recorded (Appendix A4).

Species data across the different taxonomic groups were pooled for each land-use type.

The exponent of the bias-corrected Shannon index (Chao & Shen 2003) was calculated for each land-use type. Rarefied number of species was used to evaluate the effectiveness of sampling and a non-parametric abundance-based richness estimator Chao1 (Colwell &

Coddington 1994) was used to assess within habitat diversity by extrapolation. Faunal composition for different habitat types with corresponding error bars (where there is more than one transect of a particular habitat type) was measured using the mean richness and abundance of species within each of the five functional groups. The proportion of forest dependent species richness and abundance, with associated standard error within each habitat types across the three forests was added to these plots.

Models for richness and abundance per site were explored to test for differences in species richness and abundances of the different functional groups between the three forests and forest types. Due to the sampling structure of the data, a negative binomial model with log link function was used. Spearman's rank correlation followed by a t-test was performed to check whether matrix occurrence of forest dependent species is an effect of their population size within the forest habitats.

Ordination (i.e. non-metric multidimensional scaling) (Clarke 1993) was used to specifically visualize patterns of community structure and composition among different transects and between forests. Species composition was tested for significance using the

Monte Carlo permutation test with 999 runs. The ordination scores were extracted and an analysis performed using a permutation test with variables Forest, forest type and transect in 22

the Non-metric Multidimensional Scaling analysis to test their effects on the different functional group assemblages.

3.1.5 Paper IV

In this paper, one strategically directed field survey was carried out to assess status of grasshoppers in one habitat that was previously surveyed in the 1970s. This was important given that this section of forest has since the 1970s been logged and presently subjected to heavy human influence. Currently much of the area is being colonized by the invasive and exotic paper mulberry (Broussonetia papyrifera). All photographs taken were reviewed back at camp. Dirsh (1965) generic key was used to identify unfamiliar species. Discovery of one unique photograph led to subsequent focused systematic search of the entire area close to where the species was photographed. One individual was captured (Plate 6, Appendix A4), pinned and standard taxonomic methods were employed to verify identify. The pinned specimen was relaxed in water for examination and dissection of the phallic complex.

23

4.0 RESULTS AND DISCUSSION

When forest habitats are disturbed, biodiversity changes and these changes can be positive

(e.g. Winfree et al. 2007), negative (e.g. Hamer et al. 1997, Cordeiro & Howe 2003) and at times neutral although rarely. These changes can only be realised after decades, centuries, or even millennia after the initial habitat reduction or change. This study summarizes how diversity, distributions and composition of the three insect taxa changed in space and time in the light of past and present land-use alterations within three protected forests and their surrounding matrix habitats (Papers I, II, III & IV).

The results highlight changes in insect assemblages over time with remarkable shifts in species abundance and composition of moth assemblages (Paper I); over time widespread generalists species showed increases while forest dependent species showed marked decline over the years (Paper I). The moth assemblages revealed strong temporal patterns; temporal effects on species richness, abundances and composition were far more pronounced amongst

Saturnidae than in the Sphingidae and among the forest dependent species. These differences are indicative of the variation in the life histories of the two families where Saturnidae are short-lived and their larvae are mostly dependent on mature tree resources, whereas the majority of the Sphingidae are long-lived and their larvae feed mainly on younger vegetation

(Paper I).

There was also variation in species distribution across the vertical continuum, with particular species categorized as specialists in either canopy or understorey, while others were identified as generalists (Paper II). This study observed varied distribution of ecotypes and the higher taxonomic groups (subfamilies) within the vertical stratum and this is important in understanding niche structure along the vertical gradient. Also the vertical distribution of different subfamilies indicates how strongly stratification patterns are affected by taxonomic

24

relatedness, and presumably by phylogenetic constraints (e.g. Schulze et al. 2001). The results of this study on the vertical distribution of subfamilies of fruit-feeding butterflies are consistent with theories of niche conservation (e.g. Chase & Leibold 2003). These findings are important in understanding the large-scale geographic distribution of species since the niche determines conditions under which a species can persist (e.g. Holt 2009).

Marked variation in insect distributions along the horizontal land-use gradient was observed with mature forests generally having higher richness and abundances compared to other modified habitats (Paper III). The community-level responses of organisms to land-use change are ultimately the consequence of how each species is adapted to its natural environment and how it responds to changes in biotic and abiotic factors following forest modification (e.g. Koh 2007). The observed decline in forest dependent species richness and individual abundances along the land-use gradient is an indication of resource availability and provides evidence to show that resource use within the different land-use types by the individual species is not uniform across different habitats.

Patterns in temporal, vertical and spatial variations of insect population abundances or assemblage composition and their underlying processes are important in ecological studies. It has been observed that insect abundance changes both in short-term cycles (e.g. weeks) as well as in long-term (e.g. years) (e.g. Cook & Graham 1996). Such data on population size fluctuations and associated changes at assemblages are needed especially in humid or seasonal tropics where very little is known (e.g. Wolda 1978, Ehrlich 1984, DeVries et al. 1997,

Schulze & Fiedler 2003). This study is the first attempt to use groups of insects as indicators of habitat change in Uganda for understanding long-term patterns of biodiversity over the temporal scale and across land-use gradients, in a rapidly changing forest landscape. Studies such as Nyafwono et al. (2014a, b) assessed indicator properties of butterflies in a well protected forest, where restoration of previously logged sections are being made. The 25

conservation implications of these studies are limited since restoration does not provide a clear cut gradient of land-use, rather same land-use with differing age structures.

There is a growing need for baseline data against which to judge the effort to reduce biodiversity loss and this highlights the importance of long-term datasets (Magurran et al.

2010). For example, the Saturnidae and Sphingidae communities experienced temporal turnover but the biggest challenge was to distinguish change that can be attributed to external factors like anthropogenic activities from underlying natural factors. Habitat destruction reduces or eliminates insect populations because smaller habitats support fewer species over long time periods. As such, local populations of a species may become extirpated through stochastic events. Through the sampling periods (Paper I), some species (e.g. widespread generalists) appear to benefit from land-use change while forest dependent species were heavily impacted (Figures 3 & 4, Paper I).

4.1Temporal dynamics of forest biodiversity

In Uganda, there have been immense changes in forest habitats over the last decades, most of which have been converted to agricultural landscapes (e.g. Obua et al. 2010). This study shows that the selected insect taxa richness has declined with a notable fall in individual abundances within the study forests. Overall, forest dependent saturniids and sphingids showed strong temporal changes in abundance (Fig 3, Paper I).

Temporal dynamics was far stronger in Saturnidae than in Sphingidae. This confirms the role of ecological traits in determining species distributions, for example varied life histories where Sphingidae are less specialized compared to Saturnidae in terms of their feeding requirements (e.g. Haber & Frankie 1989). The much lower diversity of Saturnidae in the modified habitats can be explained by their preference for woody plants as larval hosts

26

(Janzen 1984). Furthermore, specialised saturniids species such as Imbrasia anthina frequently depend on larval food plants that only occur in mature forests.

Growing knowledge about ecosystem change underlines the importance of long term datasets but there is still relatively few biodiversity time-series that spans decades (e.g. Wolfe et al. 1987). In particular, insect conservation efforts are often hampered by lack of information about trends in distribution and population densities of species, which are important in assessing their threat status (e.g. Schulz & Hammond 2003). Globally, there are only few long term datasets in which biodiversity data have been collected using consistent methods over a number of decades (e.g. Conrad et al. 2004). For the majority of studies, there are often variation in sampling methodology, intensity and interval, often as a result of the waxing and waning of research priorities. In Uganda, only the two large moth families sampled have long-term data sets although data have been collected sporadically and with methodological differences. This study therefore provides evidence of the fate of forest biodiversity over time, and highlights the need to repeatedly monitor biodiversity even within protected and relatively intact forests.

Similarly, Bulafu et al. (2013) found little evidence for increased disturbance or degradation within two of the study forests, suggesting that the observed changes are related to the changes in the matrix landscape. Findings of this study on moths together with similar patterns reported for trees in similar habitat settings (Bulafu et al. 2013) indicate a worrying reduction in the capacity of protected forests in central Uganda to maintain biodiversity. As their habitat and landscape requirements differ, forest specialist, edge and widespread species are expected to respond differently to the overall landscape changes. Whereas the forest specialists respond primarily to the decline and fragmentation of intact mature forests, the edge species can utilise a much broader range of habitats, including degraded forests and other wooded areas. Negative impacts on metapopulations of the landscape-scale forest loss and 27

degradation will therefore be expected to occur at different intensities of disturbance, as illustrated in figure 4. Sampling times may reflect different points along this continuum.

Figure 4. An hypothetical distribution of moth assemblages over time. F= Forest dependent species (specialists), f.= Woodland/forest edge species (forest generalists), W= Widespread species (open habitat generalists).

In addition to significant temporal dynamics in both moth families, specific individual species showed stronger temporal changes in abundance, for example the absence of forest dependent Imbrasia anthina from recent sampling periods of 1990s and 2010s compared to the sudden hyper-abundance of widespread Imbrasia anna in the same periods. These observed changes in the two moth families and constituent species qualify them as suitable and reliable indicators of community diversity and change along temporal scale. Moreover samples were collated over various sampling periods and similar times of the year allowing

28

for valid statements about moth diversity and species compositions in relation to change over time following habitat disturbance both within forests and at forest-matrix landscape.

4.2 Species distribution along the vertical continuum

This study found clear patterns of species distribution between the canopy and understorey with a marked stratum-specific difference in species richness response by the different species, subfamilies and ecological types (Paper II). Preference for the canopy is more pronounced for migrant and forest edge species whereas the majority of shade-loving forest dependent species are restricted to the dark understorey (Paper II). There were also striking differences in the proportions of species classified as understorey or canopy specialists between the multinomial model and the Chi-square test for homogeneity even when exactly the same data are used (Table 1, paper II). This can partly be explained by different assumptions in the two models (Paper II): the multinomial model allows generalist species to be classified and distinguished from species that are too rare to classify (Chazdon et al. 2011), whereas the Chi-square classifies species into ‗too rare‘ based on a rule-of-thumb criterion

(Paper II). Although the Chi-square test for homogeneity generally classified more species as specialists compared to the multinomial model, results from multinomial model versus the classical chi-square test of homogeneity showed that proportions of species classified as too rare by the multinomial model varied greatly between datasets, suggesting that statistically based multinomial model that takes data structure and distribution into account is preferable to the Chi-square test that relies on predetermined assumptions made in defining rare species

(Paper II).

Vertical distribution of fruit-feeding butterflies was also found by other studies (e.g.

DeVries et al. 2012, Molleman et al. 2006). Studies have shown that disturbance disrupts the vertical stratification of insect communities in tropical forests (e.g. DeVries 1988, Schulze et 29

al. 2001, Fermon et al. 2005). Therefore any changes in community composition and diversity are clearly related to conditions along vertical strata. Factors such as resource availability, microclimate preferences and predator avoidance are known to determine patterns of species distribution between canopy and understorey (e.g. Schultze et al. 2001, Basset et al. 2003).

Results of this study show that within the study forests, the dense understorey vegetation and a reduced wind velocity induced dark and still environments that favoured shade-loving species and weak flyers like members of the subfamily Satyrinae while making it unfavourable for many light-loving and fast-flight species such as members of subfamily

Charaxinae (Paper II). These results of vertical distribution therefore show that the degree of specialization to resources and physiological tolerances to microclimate are important in structuring insect assemblage between the canopy and the ground (e.g. Stork & Grimbacher

2006).

Whereas there was only minimal disturbance in some of the study forests, the results herein demonstrate that fruit-feeding butterflies are sensitive indicators of even low-intensity human disturbance. Contrasting results of vertical distribution of fruit-feeding butterflies at the level of species, subfamilies and ecotypes imply that both the canopy and the understorey are rich in resources, but also suggest that fine scale resource partitioning exists. These findings make the use of fruit-feeding butterflies as focal taxonomic group for monitoring forest structural changes evident and could be adopted.

4.3 Habitat use by insects in a patch-matrix landscape

In line with expectations, this study found the highest richness and abundance of the different insect groups in mature forest sites compared to other land-use types in the study areas (Paper

III). With increasing anthropogenic pressure from forests to regular home gardens, insect diversity and abundances decreased. These observed patterns are similar to results of other 30

studies where species richness reduced along land-use gradients (e.g. Beck et al. 2002).

Despite the differences in the local diversity between the functional groups considered by this study, there was an unexpected similarity in their pattern of distribution along the land-use gradient (Paper III).

In Uganda, there has been widespread decline in the quality and quantity of forest habitats in the recent past (e.g. Obua et al. 2010, Bulafu et al. 2013). In Mabira forest, severe deterioration has mostly been due to rapid intensification of monoculture agricultural systems especially large plantations of sugarcane, tea, Cardamom and oil palm that have led to decrease in pristine forest areas (Winterbottom & Eilu 2006). In Zika and Mpanga forests that have remained relatively undisturbed and unchanged internally (Bulafu et al. 2013), their surrounding matrix landscape has been substantially altered; the majority of neighbouring forest fragments in the greater Kampala–Entebbe landscape and the Mpigi archipelago have either been cleared or greatly reduced in extent and/or quality (Bulafu et al. 2013). These forms of land-uses have been implicated in the decline of species of other taxa in Uganda (e.g.

Bolwig et al. 2006) and similar decline of species from other parts of the world (e.g. Ewers &

Didham 2006, Fuller et al. 2008, Potts et al. 2010, Meng et al. 2011, Inger et al. 2015).

Land use change has consistently been reported to have a negative impact on species richness (e.g. DeVries et al. 1997, Perfecto et al. 2003, Dumbrell & Hill 2005). Sudden removal of a dense tree layer imposes drastic changes in abiotic and biotic conditions, destabilising the ecosystem. Although species richness and abundance of all five functional groups was relatively high even in human modified habitats, low numbers of forest specialists were found outside mature forests (Paper III). This is seen mainly as a measure of how altered habitats play a role in supporting these species in terms of resource availability (e.g. Driscoll et al. 2013). As forest habitats are altered, characteristic forest dependent species become replaced by disturbance-tolerant widespread species within the modified habitats. Forest 31

biodiversity has generally been shown to decline along a coarse gradient from old-growth forest (mature forest) to secondary forest, agroforestry, plantations, arable crops and pasture

(e.g. Bobo et al. 2006, Schulze et al. 2004, Basset et al. 2008). Due to their specific response to forest land-use change, forests dependent species that have narrow dietary preferences, poor dispersal ability and specific habitat affinities have emerged as good indicators for the study of impacts of forest disturbance (e.g. Holloway et al. 1992, Summerville et al. 2004,

Slade et al. 2013).

The significance of modified habitats such as secondary forests and other plantation forests in maintaining forest biodiversity has been previously reported (e.g. Bhagwat et al.

2008). The importance of the quality of surrounding landscape structure has been well- documented (e.g. Driscoll et al. 2013). However, the relative conservation values of these habitats are highly disputed (e.g. Holloway et al. 1992). This study reports a general reduction in richness and abundance of insect fauna from mature forests to mixed gardens but only to a smaller degree in secondary forests and Cardamom plantation (Paper III). These results correspond to findings from other studies (e.g. Barlow et al. 2007). Considering the species richness of forest specialists that are poor dispersers in modified habitats, land-use types like cardamom plantation and secondary forests do have remnant vegetation where at least some of these species are still expected to occur. Nevertheless, low density specialists and poor dispersers that are highly associated with undisturbed habitats are unlikely to occur in these modified habitats. This explains the trend observed in this study where the majority of forest species were recorded predominantly in mature forests, with an opposite pattern observed across the other land use types. This study also confirmed that only those forest dependent species that are rather abundant within the mature forests were regularly encountered in the matrix habitats (Paper III). These results suggest that any forms of land-use change that converts forest habitats into homogeneous landscapes will only benefit generalists and mobile 32

species. Consequently the retention or management of structurally and floristically complex habitats like secondary forests, cardamom plantations and some well-managed shade coffee can often ensure the persistence of some forest species in managed landscapes (e.g. Lamb et al. 2005, Bhagwat et al. 2008, Scales & Marsden 2008).

4.4 The need for focused sampling in face of limited inventory

The discovery of Oshwea dubiosa in Mabira shows that best results can often be achieved by using smaller sampling units within a small area. This is where focused sampling is advantageous over general inventory. The result of this focused survey has greatly improved the general knowledge of the Oshwea species and this information can be valuable in the management of the species persistence within the forest habitats. Ground-level information about biodiversity provides basis for informed decisions about the protection, sustainable use or development of specific areas or species and for the rehabilitation or restoration of degraded environments (e.g. Luff & Woiwod 1995). However, knowledge of biodiversity including geographical and habitat distributions of most organisms ranges from precise and incomplete to nonexistent yet many habitats and biota are disappearing or being degraded faster than their biodiversity can be inventoried (MEA 2005). Lack of funds coupled with lack of trained personnel and taxonomic limitations make this even much dire for the least explored areas of the developing countries (e.g. Smith et al. 2003, Grieneisen et al. 2014). It is therefore logistically infeasible to completely survey large areas, so rare species are missed out. Most studies often use species extrapolations to infer on species richness of a given area

(e.g. Colwell & Coddington 1994, Chao & Shen 2003). Only limitation with this is that such extrapolations show that sampling was incomplete but do not tell what other species could be recorded from that area.

33

4.5 Challenges of biodiversity conservation in modified forest landscapes in Uganda

Different organisms display diverse and individual responses to habitat modification depending on their life history traits (e.g. Barbaro & van Halder 2009, Öckinger et al. 2010,

Slade et al. 2013) and the spatial scale at which they perceive the environment (e.g. Gehring

& Swihart 2003, Stoks & McPeek 2003, Holland et al. 2005). This governs access to resources of various kinds and affect species turnover among habitats, along environmental gradients or between regions (e.g. Mason et al. 2007, Williams et al. 2010). These traits are likely to determine the sensitivity of species to anthropogenic disturbance (e.g. Larsen et al.

2005, Jauker et al. 2009).

The ability to utilise highly modified habitats, in addition to native forest habitat, has enabled some generalist species to prosper and expand their ranges (e.g. Öckinger, et al.

2010). Typically, generalist species have both a wider range of available resources within a focal patch and are more likely to find alternative resources outside the habitat patch, which makes them less susceptible to decreased patch size (Öckinger et al. 2010). In contrast, many forest species have specific habitat requirements that restrict them to certain elements of the landscape (e.g. mature forest areas), their ranges contract within modified landscapes often resulting in local extinction (e.g. Kotiaho et al. 2005, Öckinger et al. 2010). Species with narrow feeding niches are able to fulfil their resource requirements in only the largest and most well connected habitat patches, where the probability of co-occurrence of essential resources is largest (Öckinger et al. 2010). In addition, resource specialists are less likely to utilize resources in the matrix outside the primary habitat patches (e.g. Brotons et al. 2003).

Consequence this increases the sensitivity of specialists to habitat loss and fragmentation (e.g.

Ewers & Didham 2006). As patches become smaller and more isolated from other patches, the progression of habitat loss eventually leads to species loss.

34

In many cases, the spatial isolation of patches and the nature of the intervening matrix restrict the capacity of an organism to move through the landscape (e.g. Brotons et al. 2003,

Kupfer & Franklin 2009). In the absence of landscape connectivity, populations in small patches are vulnerable to extinction. For species dependent on native vegetation, the clearance of vegetation and replacement with alternative land-uses represents habitat loss.

Consequently, smaller patches become unsuitable for many species because they do not provide sufficient resources e.g. suitable food (e.g. Öckinger et al. 2010). Smaller patches support smaller populations that are vulnerable to extinction because of natural fluctuations in resources, random changes in population demographics, adverse genetic processes and disturbance events (e.g. selective logging, clear felling).

The presence of a species within a patch does not necessarily equate to a locally viable population (e.g. Hanski et al. 1996, Akçakaya & Sjögren-Gulve 2000). Species may persist within vegetation patches because of immigration of individuals from resource-rich areas outside the patch or locality. These populations are considered ‗sink‘ populations as they are unable to sustain their numbers in the absence of immigration. Such populations are likely to be prevalent throughout highly modified landscapes and mask the true status of populations or species (e.g. Ewers & Didham 2006).

Evidence suggests that species loss may take many decades to manifest following fragmentation (e.g. Tilman et al. 1994). Local extinction of a species can occur after a substantial delay following habitat loss, fragmentation or disturbance. That a species can initially survive environmental perturbations, but later become extinct despite no additional change to the habitat, is known as ‗extinction debt‘. As long as species still persist following environmental perturbations, then there is time to implement counter measures, such as habitat restoration, to prevent extinction. However, extinction debt poses a significant but under-recognised challenge for biodiversity conservation, because the delay between the 35

perturbation event and the demise of a species may take many decades and go unrecognised until near extinction (e.g. Tilman et al. 1994). Some ecological consequences of past landscape modification have yet to occur.

The ability of a species to disperse between isolated fragments depends on the intrinsic behaviour of the species, its habitat requirements and the nature of the intervening matrix (e.g.

Perfecto & Vandermeer 2002). From the perspective of an individual species, this response varies. Whereas highly mobile species can traverse open areas of modified habitats to feed on patches of suitable habitats, the same landscape may be impermeable to species that requires dense vegetation. This distribution pattern is very well exhibited by the forest dependent species in the study systems.

4.6 The social and political realities and their effects on forest protection in Uganda

The current size and distribution of the world‘s human population makes the influence of human communities on protected areas a significant challenge. As in many parts of the developing world, deforestation in Uganda continues to accelerate in tandem with poverty and high levels of population growth. This has led to forests becoming severely reduced in extent and become increasingly fragmented. For these regions, the transition to sustainable forest management is a much greater challenge and as such, many species will have to survive, if at all in human-modified areas (e.g. Gascon et al. 1999).

Human communities surrounding protected forests influence the ecological effectiveness of such areas by triggering edge effects such as increased species mortality and habitat loss on the edges of the protected forest (for example the case of Mabira) or through unabated land-use change within the matrix habitats that may also cause density gradients or increased extinction risk inside the protected forest even when they remain stable (for example the case of Zika and Mpanga forests). 36

The establishment of protected areas is known to generate several types of conflict among local residents and among stakeholders (e.g. Williams et al. 2003, Christie 2004). The nature of these conflicts is varied and may include power struggles, heavy-handed enforcement methods, competing management goals (e.g. land- and resource-use displacement (e.g. West et al. 2006). One such conflict was evident during the much hyped

Mabira give away project in 2009 that led to riots following its pronouncement by the government of Uganda (See plate 2).

Plate 2. A snapshot of media report on the then government give-away plan for sections of

Mabira forest for sugarcane growing.

Therefore there is need to develop approaches that enable communities to understand and embrace protected area programs. This may involve education to build social as well as political support and encourage local participation in the design and management of protected areas (e.g. Christie et al. 2003a, Thomas & Middleton 2003).

37

In Uganda, a strategy of the National Forestry Authority to help resolve some of these conflicts has been to establish Collaborative Forest Management agreements (CFM) which define rights, roles and responsibilities of communities living on forest-adjacent land. In practice, the effectiveness of such arrangements has been hampered by issues such as a lack of guidelines for benefit sharing, limited implementation capacity, lack of information for communities, and the limited duration of agreements (EMPAFORM 2006). One recent approach to improve understanding between forest management and local communities is through implementation of the Copenhagen accord- known as Reducing Emissions from

Deforestation and Degradation (REDD+). This is a nonbinding political statement that outlines principles to curb global warming to 2°C (www.un-redd.org/). The accord agreed on inclusion of payments for habitat protection .i.e. REED+ projects. Three major projects are now in Uganda and include; a) The Nile Basin Reforestation project that began in 2006 and consists of five small-scale Clean Development Mechanism (CDM) projects and is being implemented in the Rwoho Central Forest, b) the Kikonda forest reserve project being implemented by German-based company, Global- Woods AG through a local subsidiary called ‗Sustainable Use of Biomass‘, and c) the Trees for Global Benefits project being implemented by ECOTRUST in three districts in western Uganda (ECOTRUST 2007, 2008).

These projects not only have explicit objective of poverty reduction but also to enhance environmental protection (e.g. Peskett et al. 2011). Furthermore, in the event that biodiversity value is incorporated into the carbon market (e.g. Bekessy & Wintle 2008, Miles & Kapos

2008, Putz & Redford 2009), well-managed REED+ projects in degraded forest would likely represent a biodiversity-friendly strategy.

38

5.0 CONCLUSIONS

This study used a combination of insect taxa as model indicators to describe how communities respond to forest habitat alterations across temporal, vertical and spatial scales. The temporal patterns of moth communities between study periods is an indication that forest disturbance and alteration has negative impacts on its native fauna although these impacts may not be realized immediately following the disturbance events. This underlines the need for repeated monitoring of the forest systems and their associated fauna. Fruit-feeding butterflies (i.e. species, subfamilies and ecotypes) showed marked variation in distribution between the canopy and understorey, highlighting the role of niche conservatism in mediating ecological patterns that help individuals/group to choose resources (e.g. microhabitats) that they are best adapted to utilize. Different land-use types registered diverse communities of butterflies, moths and grasshoppers. Species richness, abundance and diversity however declined along a land-use gradient from mature forest to mixed gardens, with a more drastic trend for forest specialist species. These results not only underscore the value of protected forests in maintaining biodiversity, but also emphasize the management roles of other land-use types like secondary forests and cardamom plantations especially in areas that have undergone deforestation.

From this study, three plausible factors are important in driving biodiversity change within protected forests in central Uganda. These include habitat disturbance arising from activities like illegal logging and forest fragmentation that was highly visible through reclamation of forest for monoculture agriculture and plantation forests ( in Mabira), as well as matrix intensification observed through edge reclamation and clearing of adjacent forests

(in Zika and Mpanga forests). Sufficient data, predictive analyses and scenario building explored in this study can make important contributions towards devising appropriate conservation policy strategies to safeguard forests and their associated taxa. 39

This study can conclude that butterflies, moths and grasshoppers communities are good indicators of forest structure, habitat complexity and conservation status. The response patterns of the five functional groups covered by this study, most especially the change in diversity and abundance of forest specialist species shows that any group can be reasonably used prima facie as biodiversity indicator group for another, or even for other organisms. The ever increasing human-modified landscapes present a daunting challenge to conservation of tropical forest biodiversity. Therefore, alternative policies targeting maintenance of forest ecological integrity, while ensuring human welfare should be prioritized and brought to the forefront of our national conservation agenda.

5.1 RECOMMENDATIONS

 Given the observed change in insect assemblages over time even in relatively stable

forests, and across land-use gradient, there is need to regularly monitor biodiversity in

protected areas, and a combination of indicators should be used since different groups

occupy various niches in the forests and use resources differently.

 Although this study setup provided some insights into how forest land-use change

influence insect diversity, additional studies are needed to provide a more complete

picture of this relationship. For example, insect-plant interactions like pollination,

herbivory amongst others.

 My results showed that forest specialists are the most vulnerable species group to

forest modification. This study therefore recommend that in order to promote the long-

term persistence of such biodiversity in human modified forest landscapes, particular

attention should be given to other land-uses that maintain forest structure and

minimize edge effects, as well as control of illegal logging across the entire landscape.

40

 Although forest management should prioritize maintenance of protected forests,

surrounding matrix habitats that maintain heterogeneity across the landscape, for

example cardamom and shade coffee plantations (these were shown to be contributing

to survival of forest species although minimally) should be integrated in conservation

management.

 Since protected forests are linked ecologically to their surrounding habitats, there is

need to stem broad-scale loss and degradation even in matrix habitats to reduce the

likelihood of serious biodiversity declines/loss and subsequent extinction.

 Each of the forests had their own range of species and do not substitute for one another in terms of conservation value. Therefore efforts should be made to maintain the integrity of these forests.

41

6.0 REFERENCES Akçakaya H.R. and Sjögren-Gulve, P. (2000) Population viability analysis in conservation

planning: an overview. Ecological Bulletins 48, 9–21.

Akite, P. (2008) Biodiversity Monitoring for Conservation and Ecology: A study of butterflies

of selected sites in Sango Bay and Iriiri Areas. MSc thesis, MUIENR.

Antongiovanni, M. and Metzger, J.P. (2005) Influence of matrix habitats on the occurrence of

insectivorous bird species in Amazonian forest fragments. Biological Conservation

122, 441–451.

Armstrong, A.J. and van Hensbergen, H.J. (1997) Evaluation of afforestable montane

grasslands for wildlife conservation in the north-eastern Cape, .

Biological Conservation 81, 179–90.

Asasira, J. (2003) Butterfly diversity and Conservation in eleven Important Bird Areas of

Uganda. MSc thesis, MUIENR.

Barbaro, L. and van Halder, I. (2009) Linking bird, carabid beetle and butterfly life-history

traits to habitat fragmentation in mosaic landscapes. Ecology 32, 321–333.

Barlow, H.S. and Woiwod, P.I. (1989) Moth diversity of tropical forest in peninsular

Malaysia. Journal of Tropical Ecology 5, 37–50.

Barlow, J., Gardner, T.A., Araujo, I.S., Avila-Pires, T.C., Bonaldo, A.B., Costa, J.E.,

Esposito, M.C., Ferreira, L.V., Hawes, J., Hernandez, M.I.M., Hoogmoed, M.S., Leite,

R.N., Lo-Man-Hung, N.F., Malcolm, J.R., Martins, M.B., Mestre, L.A.M., Miranda-

Santos, R., Nunes-Gutjahr, A.L., Overal, W.L., Parry, L., Peters, S.L., Ribeiro-Junior,

M.A., da Silva, M.N.F., da Silva Motta, C. and Peres, C.A. (2007) Quantifying the

biodiversity value of tropical primary, secondary, and plantation forests. Proceedings

of the National Academy of Science 10, 18555–18560.

42

Basset, Y., Hammond, P.M., Barrios, H., Holloway, J.D. and Miller, S.E. (2003) Vertical

stratification of arthropod assemblages. Pp. 17–27 in Basset, Y., Novotný, V., Miller,

S.E. and Kitching, R.L. (Eds). of tropical forests–spatio–temporal

dynamics & resource use in the canopy. Cambridge University Press, Cambridge.

Basset, Y., Missa, O., Alonso, A., Miller, S.E., Curletti, G., De Meyer, M., Eardley, C.,

Lewis, O.T., Mansell, M.W., Novotny, V. and Wagner, T. (2008) Changes in

arthropod assemblages along a wide gradient of disturbance in . Conservation

Biology 22, 1552–1563.

Beccaloni, G.W. and Gaston, K.J. (1994) Predicting the species richness of Neotropical forest

butterflies: Ithomiinae (: ) as indicators. Biological

Conservation 71, 77–86.

Beck, J., Schulze, C.H., Linsenmair, K.E. and Fiedler, K. (2002) From forest to farmland:

diversity of geometrid moths along two habitat gradients on Borneo. Journal of

Tropical Ecology 17, 33–51.

Bekessy, S.A. and Wintle, B.A. (2008) Using carbon investment to grow the biodiversity

bank. Conservation Biology 22, 510–513.

Bhagwat, S.A., Willis, K.J., Birks, H.J.B. and Whittaker, R.J. (2008) Agroforestry: a refuge

for tropical biodiversity? Trends in Ecology and Evolution 23, 261–264.

Birungi, J. (1995) Status and vertical distribution of Saturniid and Sphingid moths

(Lepidoptera) in Zika Forest, Uganda. MSc Thesis, Makerere University.

Blair, R.B. and Launer, A.E. (1997) Butterfly diversity and human land use: species

assemblages along an urban gradient. Biological Conservation 80, 113–125.

Bobo, K.S., Waltert, M., Fermon, H., Njokagbor, J. and Mühlenberg, M. (2006) From forest

to farmland: butterfly diversity and habitat associations along a gradient of forest

conversion in Southwestern . Journal of Insect Conservation 10, 29–42. 43

Bolwig, S., Pomeroy, D., Tushabe, H. and Mushabe, D. (2006) Crops, trees, and birds:

biodiversity change under agricultural intensification in Uganda‘s farmed landscapes.

Danish Journal of Geography 106, 115–130.

Brotons, L., Mönkkönen, M. and Martin, J.L. (2003) Are fragments islands? Landscape

context and density-area relationships in boreal forest birds. The American Naturalist

162, 343–357.

Bulafu, C., Baranga, D., Mucunguzi, P., Telford, R.J. and Vandvik, V. (2013) Massive

structural and compositional changes over two decades in forest fragments near

Kampala, Uganda. Ecology and Evolution 3, 3803–3823.

Bwanika, J. (1995) A study of the butterflies of Mabira Forest. MSc Thesis, Makerere

University, Kampala.

Byaruhanga, A., Kasoma, P. and Pomeroy, D. (2001) Important Bird Areas in Uganda.

Nature Uganda, The East Africa Natural History Society, Kampala.

Campbell, R.E., Harding, J.S., Ewers, R.M., Thorpe, S. and Didham, R.K. (2011) Production

land use alters edge response functions in remnant forest invertebrate communities.

Ecological Applications 21, 3147–3161.

Carcasson, R.H. (1961) The Butterflies of East Africa (Lepidoptera, Acraeidae).

Journal of East Africa Natural History Society. Special Supplement. No 8.

Carcasson, R.H. (1963) The Milkweed Butterflies of East Africa (Lepidoptera, Danaidae).

Journal of East Africa Natural History Society 24, 19–32.

Carcasson, R.H. (1975) The Swallowtail Butterflies of East Africa (Lepidoptera,

Papilionidae). E W Classey Ltd, Faringdon, Oxon, England.

Carcasson, R.H. (1976) Revised catalogue of the African Sphingidae, Lepidoptera with

descriptions of the East African species. E W Classey Ltd, Oxon, England.

44

Chao, A. and Shen, T-J. (2003) Nonparametric estimation of Shannon‘s index of diversity

when there are unseen species in sample. Environmental and Ecological Statistics 10,

429–443.

Chase, J.M. and Leibold, M.A. (2003) Ecological Niches: Linking Classical and

Contemporary Approaches. Chicago University Press, Chicago.

Chazdon, R.L., Chao, A., Colwell, R.K., Lin, S.Y., Norden, N., Letcher, S.G., Clark, D.B.,

Finegan, B. and Arroyo, J.P. (2011) A novel statistical method for classifying habitat

generalists and specialists. Ecology 92, 1332–1343.

Christie, P. (2004) MPAs as biological successes and social failures in Southeast Asia.

Pp.155–164 in Shipley, J.B. (Ed). Aquatic Protected Areas as Fisheries Management

Tools: Design, Use, and Evaluation of These Fully Protected Areas. American

Fisheries Society, Bethesda, Maryland.

Christie, P., Buhat, D. Garces, L. R. and White. A. T. (2003a) The challenges and rewards of

community-based coastal resources management. Pp. 231–249 in Brechin, S.R.,

Wilshusen, P.R., Fortwangler. C.L. and West, P.C. (Eds). Contested nature, promoting

international biodiversity with social justice in the twenty-first century. State

University of New York Press, Albany.

Clarke, K.R. (1993) Non-parametric multivariate analyses of changes in community structure.

Australian Journal of Ecology 18, 117–143.

Collinge, S.K. (2009) Ecology of Fragmented Landscapes. Johns Hopkins University Press.

Colwell, R.K. (2006) EstimateS: statistical estimation of species richness and shared species

from samples. Version 9.1.0. http://viceroy.eeb.uconn.edu/estimates/.

Colwell, R.K. and Coddington, J.A. (1994) Estimating the extent of terrestrial biodiversity

through extrapolation. Philosophical Transactions Royal Society, Series B 345, 101–

118. 45

Conrad, K., Woiwod, I., Parsons, M., Fox, R. and Warren, M. (2004) Long-term population

trends in widespread British moths. Journal of Insect Conservation 8, 119–136.

Cook, L.M. and Graham, C.S. (1996) Evenness and species number in some moth

populations. Biological Journal of the Linnean Society 58, 75–84.

Cordeiro, N.J. and Howe, H.F. (2003) Forest fragmentation severs mutualism between seed

dispersers and an endemic African tree. Proceedings of the National Academy of

Sciences of the United States of America 100, 14052–14056.

D'Abrera, B. (1980) Butterflies of the Afrotropical Region. Lansdowne Press, Melbourne.

Davenport, T.R.B. (1996) The Butterflies of Uganda - An Annotated Checklist. Uganda

Forest Department, Kampala, Uganda. 48 pp.

DeVries, P.J. (1988) Stratification of fruit-feeding Nymphalid butterflies in Costa Rican

rainforest. Journal of Research on the Lepidoptera 26, 98–108.

DeVries, P.J., Alexander, L.G., Chacon, I.A. and Fordyce, J.A. (2012) Similarity and

difference among rainforest fruit-feeding butterfly communities in Central and South

America. Journal of Animal Ecology 81, 472–482.

DeVries, P.J., Murray, D. and Lande, R. (1997) Species diversity in vertical, horizontal, and

temporal dimensions of a fruitfeeding butterfly community in an Ecuadorian

rainforest. Biological Journal of the Linnean Society 62, 343–364.

Didham, R.K., Ghazoul, J., Stork, N.E and Davis, A.J. (1996) Insects in fragmented forests: a

functional approach. Trends in Ecology and evolution 11, 255–260.

Dirsh, V.M. (1965) The African genera of Acridoidea. Anti-Locust Research Centre and

Cambridge University Press, London, 579 pp.

Dirzo, R. and Raven, P. (2003) Global state of biodiversity loss. Annual Reviews in

Environment and Resources 28, 137–167.

46

Driscoll, D.A., Banks, S.C., Barton, P.S., Lindenmayer, D.B. and Smith, A.L. (2013)

Conceptual domain of the matrix in fragmented landscapes. Trends in Ecology and

Evolution 28, 605–613.

Dumbrell, A.J. and Hill, J.K. (2005) Impacts of selective logging on canopy and ground

assemblages of tropical forest butterflies: implications for sampling. Biological

Conservation 125, 123–131.

Dunn, R.R. (2005) Modern insect extinctions, the neglected majority. Conservation Biology

19, 1030–1036.

ECOTRUST. (2007) Trees for Global Benefits (TGB) Program in Uganda: a Plan Vivo

Project annual report. ECOTRUST, Kampala.

ECOTRUST. (2008) Trees for Global Benefits (TGB) Program in Uganda: a Plan Vivo

Project annual report. ECOTRUST, Kampala.

Ehrlich, P.R., Breedlove, D.E., Brussard, P.E. and Sharp, M.A. (1972) Weather and the

"regulation" of subalpine butterfly populations. Ecology 53, 243–247.

Ehrlich, P.R. (1984) The structure and dynamics of butterfly populations. Pp. 25–58 in Vane-

Wright, R.I. and Ackery P.R. (Eds).The biology of butterflies. Symposium of the

Royal Entomological Society of London, Number 11, Academic Press.

Ehrlich, P.R. (1988) The loss of diversity: causes and consequences. Pp. 21–27 in Wilson, E.

O. (Ed). Biodiversity. The National Academies Press.

EMPAFORM. (2006) Participatory Forest Management initiatives in Uganda: key

implementation concerns and recommendations for policy actions. EMPAFORM

Policy Briefing Paper No. 1. EMPAFORM, Kampala.

Erhardt, A. and Thomas, J.A. (1991) Lepidoptera as indicators of change in the semi-natural

grasslands of lowland and upland Europe. Pp. 213–236 in Collins, N.M. and Thomas,

J.A. (Eds). The Conservation of Insects and Their Habitats. Academic Press, London. 47

Ewers, R.M. and Didham, R.K. (2006) Confounding factors in the detection of species

responses to habitat fragmentation. Biological Reviews 81, 117–142.

Fahrig, L. (2001) How much habitat is enough? Biological Conservation 100, 65–74.

Fahrig, L. (2003) Effects of habitat fragmentation on biodiversity. Annual Review in Ecology,

Evolution and Systematic 34, 487–515.

Fahrig, L. and Merriam, G. (1994) Conservation of fragmented populations. Conservation

Biology 8, 50–59.

FAO (2011) State of the world's forests 2011. Food and Agricultural Organisation (United

Nations), Rome, .

Fermon, H., Waltert, M. and Mühlenberg, M. (2003) Movement and vertical stratification of

fruit-feeding butterflies in a managed West African rainforest. Journal of Insect

Conservation 7, 7–19.

Fermon, H., Waltert, M., Vane-Wright, R.I. and Mühlenberg, M. (2005) Forest use and

vertical stratification in fruit-feeding butterflies of Sulawesi, Indonesia: impacts for

conservation. Biodiversity and Conservation 14, 333–350.

Forman, R.T.T. and Godron, M. (1986) Landscape Ecology. John Wiley & Sons, Inc., New

York.

Franklin, J.F. (1993) Preserving Biodiversity: Species, Ecosystems, or Landscapes?

Ecological Applications 3, 202–205.

Franklin, J.F. and Lindenmayer, D.B. (2009) Importance of matrix habitats in maintaining

biological diversity. Proceedings of the national Academy of Sciences 106, 349–350.

Fuller, R.A., Warren, P.H., Armsworth, P.R., Barbosa, O. and Gaston, K.J. (2008) Garden

bird feeding predicts the structure of urban avian assemblages. Diversity and

Distributions 14, 131–137.

48

Gandar, M.V. (1982) The dynamics and trophic ecology of grasshoppers (Acridoidea) in a

South African savanna. Oecologia 54, 370–378.

Gascon, C., Lovejoy, T.E., Bierregaard Jr, R.O., Malcolm, J.R., Stouffer P.C., Vasconcelos,

H.L., Laurance, W.F., Zimmerman, B., Tocher, M. and Borges, S. (1999) Matrix

habitat and species richness in tropical forest remnants. Biological Conservation

91, 223–229.

Gascon, C., Williamson, B. and Fonseca, G.A.B. (2000) Receding forest edges and vanishing

reserves. Science 288, 1356–1358.

Gaston, K.J. (1991) The magnitude of global insect species richness. Conservation Biology 5,

283–296.

Gaston, K.J. and Lawton, J.H. (1988) Patterns in distribution and abundance of insect

populations. Nature 331, 709–712.

Gehring, T.M. and Swihart, R.K. (2003) Body size, niche breadth, and ecologically scaled

responses to habitat fragmentation: mammalian predators in an agricultural landscape.

Biological Conservation 109, 283–295.

Gillon, Y. (1983) The invertebrates of the grass layer. Pp. 289–311 in Boulière, F. (Ed).

Ecosystems of the World 13: Tropical Savannas. Elsevier, Amsterdam.

Grieneisen, M.L., Zhan, Y., Potter, D. and Zhang, M. (2014) Biodiversity, Taxonomic

Infrastructure, International Collaboration, and New Species Discovery. BioScience

64, 322–332.

Haber, W. and Frankie, G. (1989) A tropical hawkmoth community: Costa Rican dry forest

Sphingidae. Biotropica 21, 155–172.

Hamer, K.C., Hill, J.K., Lace, L.A. and Langan, A.M. (1997) Ecological and biogeographical

effects of forest disturbance on tropical butterflies of Sumba, Indonesia. Journal of

Biogeography 24, 67–75. 49

Hamilton, A.C. (1984) Deforestation in Uganda. Oxford University Press, Nairobi.

Hanski, I., Moilanen, A. and Gyllenberg, M. (1996) Minimum viable metapopulation size.

The American Naturalist 147, 527–541.

Henning, S.F. (1988) The Charaxinae Butterflies of Africa. Aloe Books, Johannesburg, South

Africa.

Heywood, V.H. (Ed). (1995) Global biodiversity assessment. United Nations Environment

Programme. Cambridge (UK): Cambridge University Press.

Hilt, N., Brehm, G. and Fiedler, K (2006) Diversity and ensemble composition of geometrid

moths along a successional gradient in the Ecuadorian Andes. Journal of Tropical

Ecology 22, 155–166.

Holland, J.D., Fahrig, L. and Cappuccino, N. (2005) Body size affects the spatial scale of

habitat-beetle interactions. Oikos 110, 101–108.

Holloway, J.D., Kirk-Spriggs, A.H. and Chey, V.K. (1992) The response of some rain forest

insect groups to logging and conversion to plantation. Philosophical Transactions of

the Royal Society of London B 335, 425–436.

Holt, R.D. (2009) Bringing Hutchinsonian niche into the 21st century: Ecological and

evolutionary perspectives. Proceedings of the National Academy of Science 106,

19659–19665.

Howard, P.C. and Davenport, T.R.B. (Eds). (1996) Forest Biodiversity Reports. Vols. 1-33.

Uganda Forest Department, Kampala, Uganda.

Howard, P.C., Davenport, T.R.B., Kigenyi, F.W., Viskanic, P., Balzer, M.C., Dickinson, C.J.,

Lwanga, J.S., Matthews, R.A. and Mupada, E. (2000) Protected area planning in the

tropics: Uganda‘s national system of forest nature reserves. Conservation Biology 14,

858–875.

50

Inger, R., Gregory, R., Duffy, J.P., Stott, I., Voříśek, P. and Gaston, K.J. (2015) Common

European birds are declining rapidly while less abundant species‘ numbers are rising

Ecology Letters 18, 28–36.

Janzen, D.H. (1984) Two ways to be a tropical big moth: Santa Rosa saturniids and sphingids.

Oxford surveys in Evolutionary Biology 1, 85–140.

Janzen, D.H. (1986) The eternal external threat. Pp. 286–302 in Soule, M.E. (Ed).

Conservation Biology: The Science of Scarcity and Diversity. Sinauer Associates Inc,

Sunderland, Massachusetts.

Janzen, D.H. (1987) Insect diversity of a Costa Rican dry forest: Why keep, and how?

Biological Journal of the Linnean Society 30, 343–356.

Janzen, D.H. (1988) Ecological characterizations of a Costa Rican dry forest caterpillar fauna.

Biotropica 20, 120–135.

Jauker, F., Deikötter, T., Schwarzbach, F. and Wolters, V. ( 2009) Pollinator dispersal in an

agricultural matrix: opposing responses of wild bees and hoverflies to landscape

structure and distance from main habitat. Landscape Ecology 24, 547–555.

Jules, E.S. and Shahani, P. (2003) A broader ecological context to habitat fragmentation: why

matrix habitat is more important than we thought. Journal of Vegetation Science 14,

459–464.

Kaggwa, D.K. (1995) The abundance, diversity and distribution of butterflies in the Sango

Bay project area. MSc thesis, Makerere University, Kampala.

Kemp, W.P. (1992) Rangeland grasshopper (Orthoptera: Acrididae) community structure: A

working hypothesis. Environmental Entomology 21, 461–70.

Kingdon, J. (1971) East African Mammals: An atlas of evolution in Africa, Vol. 1. Academic

Press, London.

51

Koh L.P (2007) Impacts of land use change on South-east Asian forest butterflies: a review.

Journal of Applied Ecology 44, 703–713.

Koh, L.P. and Sodhi, N.S. (2004) Importance of reserves, fragments and parks for butterfly

conservation in a tropical urban landscape. Ecological Applications 14, 1695–1708.

Kotiaho, J.S., Kaitala, V., Komonen, A. and Päivinen, J. (2005) Predicting the risk of

extinction from shared ecological characteristics. Proceedings of the National

Academy of Sciences, USA 102, 1963–1967.

Kremen, C. (1994) Biological inventory using target taxa: a case study of the butterflies of

Madagascar. Ecological Applications 4, 407–422.

Kupfer, J.A. and Franklin, S.B. (2009) Linking Spatial Pattern and Ecological Responses in

Human-Modified Landscapes: The Effects of Deforestation and Forest Fragmentation

on Biodiversity. Geography Compass 3, 1331–1355.

Kupfer, J.A., Malanson, G.P. and Franklin, S.B. (2006) Not seeing the ocean for the islands:

the mediating influence of matrix-based processes on forest fragmentation effects.

Global Ecology and Biogeography 15, 8–20.

Lamb, D., Erskine, P. and Parrotta, J.A. (2005) Restoration of Degraded Tropical Forest

Landscapes. Science 310, 1628–1632.

Larsen, T.H., Williams, N.M. and Kremen, C. (2005) Extinction order and altered community

structure rapidly disrupt ecosystem functioning. Ecology Letters 8, 538–547.

Laurance, W.F. (1990) Comparative responses of five arboreal marsupials to tropical forest

fragmentation. Journal of Mammalogy 71, 641–653.

Lawton, J.H., Bignell, D.E., Bolton, B., Bloemers, G.F., Eggleton, P., Hammond, P.M.,

Hodda, M., Holt, R.D., Larsen, T.B., Mawdsleu, N.A., Stork, N.E., Srivastava, D.S.

and Watt, A.D. (1998) Biodiversity inventories, indicator taxa and effects of habitat

modification in tropical forest. Nature 391, 72–76. 52

Lewinsohn, T.M., Freitas, A.V.L. and Prado, P.I. (2005) Conservation of terrestrial

invertebrates and their habitats in Brazil. Conservation Biology 19, 640–645.

Lindenmayer, D.B., Margules, C.R. and Botkin, D.B. (2000) Indicators of Biodiversity for

Ecologically Sustainable Forest Management. Conservation Biology 14, 941–950.

Luff, M.L. and Woiwod, P.I. (1995) Insects as indicators of land-use change: a European

perspective, focusing on moths and ground beetles. Pp. 399–422 in Harrington, R. and

Stork, N.E. (Eds). Insects in a changing environment. Academic Press, London.

Magurran, A.E., Baillie, S.R., Buckland, S.T., Dick, J.M., Scott, E.M., Smith, R.I.,

Somerfield, P.J. and Watt, A.D. (2010) Long-term datasets in biodiversity research

and monitoring: assessing change in ecological communities through time. Trends in

Ecology and Evolution 25, 574–582.

Mason, N.W.H., Lanoiselee, C., Mouillot, D., Irz, P. and Argillier, C. (2007) Functional

characters combined with null models reveal inconsistency in mechanisms of species

turnover in lacustrine fish communities. Oecologia 153, 441–452

McNeely, J.A., Gadgil, M., Leveque, C., Padoch, C. and Redford, K. (1995) Human

influences on biodiversity. Pp. 711–822 in Heywood, V.H. and Watson, R.T. (Eds).

Global biodiversity assessment. UNEP, Cambridge, Cambridge University Press.

Meng, L.Z., Martin, K., Weigel, A. and Liu, J-X. (2011) Impact of rubber plantation on

carabid beetle communities and species distribution in a changing tropical landscape

(southern Yunnan, China). Journal of Insect Conservation 16, 423–432.

Michaelsen, F.C.J., and Marshall, M. (2012) Mapping recent decadal climate variations in

precipitation and temperature across eastern Africa and the Sahel, chap. 14 of Remote

Sensing of Drought: Innovative Monitoring Approaches. ftp://chg.geog.

ucsb.edu/pub/pubs/mapping_decadal_variations.pdf.

53

Miles, L. and Kapos, V. (2008) Reducing greenhouse gas emissions from deforestation and

forest degradation: global land-use implications. Science 320, 1454–1455.

Millennium Ecosystem Assessment. (2005) Ecosystems and Human Well-being: Biodiversity

Synthesis. World Resources Institute, Washington, DC.

Ministry of Water Lands and Environment. (2002) Department of meteorology: ―Uganda

Initial National Communication to the United Nations Framework Convention on

Climate Change‖. Ministry of Water, Lands and Environment, Kampala, Uganda.

Molleman, F., Kop, A., Brakefield, P.M., De Vries, P.J. and Zwaan, B.J. (2006) Vertical and

temporal patterns of biodiversity of fruit feeding butterflies in a tropical forest in

Uganda. Biodiversity and Conservation 15, 107–121.

Muirhead-Thomson, R.C. (1991) Trap responses of flying insects. Academic Press, London.

287 pp.

New, T.R. (2004) Moths (Insecta: Lepidoptera) and conservation: background and

perspective. Journal of Insect Conservation 8, 79–94.

Noss, R.F. (1990) Indicators for monitoring biodiversity: a hierarchical approach.

Conservation Biology 4, 355–364.

Nyafwono, M., Valtonen, A., Nyeko, P. and Roininen, H. (2014b) Butterfly community

composition across a successional gradient in a human-disturbed Afro-tropical rain

forest. Biotropica 46, 210–218.

Nyafwono, M., Valtonen, A., Nyeko, P. and Roininen, H. (2014a) Fruit-feeding butterfly

communities as indicators of forest restoration in an Afro-tropical rainforest.

Biological Conservation 174, 75–83.

Obua, J., Agea, J.G. and Ogwal, J.J. (2010) Status of forests in Uganda. African Journal of

Ecology 48, 853–859.

54

Öckinger, E., Schweiger, O., Crist, T.O., Debinski, D.M., Krauss, J., Kuussaari, M., Petersen,

J.D., Pöyry, J., Settele, J., Summerville, K.S. and Bommarco, R. (2010) Life-history

traits predict species responses to habitat area and isolation: a cross-continental

synthesis. Ecology Letters 13, 969–979.

Odum, E.P. (1985) Trends expected in stressed ecosystems. Bioscience 35, 419–422.

Pearson, S.M. (1993) The spatial extent and relative influence of landscape-level factors on

wintering birds populations. Landscape Ecology 8, 3–18.

Perfecto, I. and Vandermeer, J. (2002) Quality of Agroecological Matrix in a Tropical

Montane Landscape: Ants in Coffee Plantations in Southern Mexico. Conservation

Biology 16, 174–182.

Perfecto, I., Mas, A., Dietsch, T. and Vandermeer, J. (2003) Conservation of biodiversity in

coffee agroecosystems: a tri-taxa comparison in southern Mexico. Biodiversity and

Conservation 12, 1239–1252.

Peskett, L., Schreckenberg, K. and Brown, J. (2011) Institutional approaches for carbon

financing in the forest sector: learning lessons for REDD+ from forest carbon projects

in Uganda. Environmental Science and policy 14, 216–229.

Pimm, S.L. and Brooks, T.M. (2000) The Sixth Extinction: How large, how soon, and where?

Pp. 46 – 62 in Raven, P. (Ed). Nature and Human Society: the quest for a sustainable

world. National Academy Press, Washington, DC.

Pinheiro, C.E.G. and Ortiz, J.V.C. (1992) Communities of fruit feeding butterflies along a

vegetation gradient in central Brazil. Journal of Biogeography 19, 505–511.

Pollard, E. (1977) A method for Assessing changes in abundance of butterflies. Biological

conservation 12, 115–134.

Pollard, E. and Yates, T.J. (1993) Monitoring butterflies for ecology and conservation. The

British butterfly monitoring scheme. Chapman and Hall, London. 274 pp. 55

Pomeroy, D. and Tushabe, H. (2004) The State of Uganda’s Biodiversity 2004. Makerere

University, Kampala.

Potts, S.G., Biesmeijer, J.C., Kremen, C., Neumann, P., Schweiger, O. and Kunin, W.E.

(2010) Global pollinator declines: trends, impacts and drivers. Trends in Ecology and

Evolution 25, 345–353.

Prevedello, J.A. and Vieira, M.V. (2010) Does the type of matrix matter? A quantitative

review of the evidence. Biodiversity and Conservation 19, 1205–1223.

Putz, F.E. and Redford, K.H. (2009) Dangers of carbon-based conservation. Global

Environmental Change 19, 400–401.

Ricketts, T.H., Daily, G.C., Ehrlich, P.R. and Fay, J.P. (2001) Countryside biogeography of

moths in a fragmented landscape: biodiversity in native and agricultural habitats.

Conservation Biology 15, 378–388.

Rodrigues, A.S.L., Andelman, S.J., Bakarr, M.I., Boitani, L., Brooks, T.M., Cowling, R.M.,

Fishpool, L.D.C., Fonseca, G.A.B.da., Gaston, K.J., Hoffmann, M., Long, J., Marquet,

P.A., Pilgrim, J.D., Pressey, R.L., Schipper, J., Sechrest, W., Stuart, S.N., Underhill,

L.G., Waller, R.W., Watts, M.E.J. and Yan, X. (2003) Global Gap Analysis: Towards

a Representative Network of Protected Areas.Washington DC: Center for Applied

Biodiversity Science, Conservation International. Washington, DC.

Rogers, K. St.A. and van Someren, V.G.L. (1925b) The butterflies of and Uganda, part

2. Journal of East Africa Natural History Society 23,105–145.

Rowell, C.H.F. (1978) Food plant specificity in Neotropical rain-forest acridids. Entomologia

Experimentalis et Applicata 24, 651–662.

Samways, M.J. (1997) Conservation biology of Orthoptera. Pp. 481–96 in Gangwere, S.K.,

Muralirangan, M.C. and Muralirangan, M. (Eds). The Binomics of Grasshoppers,

Katydids and Their Kin. CAB International, Wallingford. 56

Samways, M.J. and Moore, S.D. (1991) Influence of exotic conifer patches on grasshopper

(Orthoptera) assemblages in a grassland matrix at a recreational resort, Natal, South

Africa. Biological Conservation 57, 117–37.

Sayer, J.A., Harcourt, C.S. and Collins, N.M. (1992) The conservation Atlas of tropical

forests of Africa. Macmillan publishers Ltd, London.

Scales, B.R. and Marsden, S.J. (2008) Biodiversity in small-scale tropical agroforests: a

review of species richness and abundance shifts and the factors influencing them.

Environmental Conservation 35, 160–172.

Schultz, C.B. and Hammond, P.C. (2003) Using population viability analysis to develop

recovery criteria for endangered insects: Case study of the Fender's Blue butterfly.

Conservation Biology 17, 1372–1385.

Schulze, C.H. and Fiedler, K. (1999) Species richness of South East Asian butterflies-How

can it be estimated using faunal lists at different geographic scales? Pp. 63–70 in

Mohamed, M. and Bernard, H. (Eds). Tropical ecosystem research in Sabah. For

whom and for what? Proceedings of 3rd SITE Seminar. Universiti Malaysia Sabah,

Kota Kinabalu, Malaysia.

Schulze, C., Waltert, M, Kessler, P.J.A, Pitopang, R., Shahabuddin., Veddeler, D.,

Muhlenberg, M., Gradstein, S.R., Leuschner, C., Steffan-Dewenter, I. and Tscharntke,

T. (2004) Biodiversity indicator groups of tropical landuse systems: comparing plants,

birds and insects. Ecological Application 14, 1321–1333.

Schulze, C.H. and Fiedler, K. (2003) Vertical and temporal diversity of a species-rich moth

taxon in Borneo. Pp. 69–85 in Basset, Y., Novotný, V., Miller, S. and Kitching, R.L.

(Eds). Arthropods of tropical forests: spatio-temporal dynamics and resource use in

the canopy. Cambridge University Press, Cambridge.

57

Schulze, C.H., Linsenmair, K.E. and Fiedler, K. (2001) Understorey versus canopy: patterns

of vertical stratification and diversity among Lepidoptera in a Bornean rain forest.

Plant Ecology 153, 133–152.

Slade, E.M., Merckx, T., Riutta, T., Bebber, D.P., Redhead, D., Riordan, P. and Macdonald,

D.W. (2013) Life-history traits and landscape characteristics predict macro-moth

responses to forest fragmentation. Ecology 94, 1519–1530.

Smith, R.J., Muir, R.D.J., Walpole, M.J., Balmford, A. and Leader-Williams, N. (2003)

Governance and the loss of biodiversity. Nature 426, 67–70.

Sodhi, N.S., Koh, L.P., Brook, B.W. and Ng, P.K.L. (2004) Southeast Asian biodiversity: an

impending disaster. Trends in Ecology and Evolution 19, 654–660.

Sodhi, N.S., Posa, M.R.C., Lee, T.M., Bickford, D., Koh, L.P. and Brook, B.W. (2010) The

state and conservation of Southeast Asian biodiversity. Biodiversity and Conservation

19, 317–328.

Stoks, R. and McPeek, M.A. (2003) Predators and life histories shape Lestes damselfly

assemblages along a freshwater habitat gradient. Ecology 84, 1576–1587.

Stork, N.E. and Grimbacher, P.S. (2006) Beetles assemblages from an Australian tropical

rainforest show that the canopy and the ground strata contribute equally to

biodiversity. Proceedings of the Royal Society B 273, 1969–1975.

Stork, N.E. (1988) Insect diversity: Facts, fiction and speculation. Biological journal of the

Linnean Society 35, 321–337.

Summerville, K.S. and Crist, T.O. (2004) Contrasting effects of habitat quantity and quality

on moth communities in fragmented landscapes. Ecography 27, 3–12.

Summerville, K.S., Ritter, L.M. and Crist, T.O. (2004) Forest moth taxa as indicators of

lepidopteran richness and habitat disturbance: a preliminary assessment. Biological

Conservation 116, 9–18. 58

Sutton, S.L. (1983) The spatial distribution of flying insects in tropical rain forests. Pp 77–91

in Sutton, S.L., Whitemore, T.C. and Chadwick A.C. (Eds). Tropical rain forests:

Ecology and management. Blackwell Scientific Publications, Oxford.

Terborgh, J. (1992) Maintenance of diversity in tropical forests. Biotropica 24, 283–292.

Thomas, C.D., Jordana, D., Lewis, O.T., Hill, J.K., Sutcliffe, O.L. and Thomas, J.A. (1998)

Butterfly distributional patterns, processes and conservation. Pp. 107–138 in Mace,

G.M., Balmford, A. and Ginsberg, J.R. (Eds). Conservation in a changing world.

Cambridge University Press.

Thomas, J.A. (2005) Monitoring change in the abundance and distribution of insects using

butterflies and other indicator groups. Philosophical Transactions of the Royal Society

B 360, 339–357.

Thomas, J.A., Telfer, M.G., Roy, D.B., Preston, C.D., Greenwood, J.J.D., Asher, J., Fox, R.,

Clarke, R.T. and Lawton, J.H. (2004) Comparative losses of British butterflies, birds,

and plants and the global extinction crisis. Science 303, 1879–1881.

Thomas, L. and Middleton, J. (2003) Guidelines for Management Planning for Protected

Areas. Gland, Switzerland and Cambridge, UK: IUCN.

Tilman, D., May, R.M., Lehman, C.L. and Nowak, M.A. (1994) Habitat destruction and the

extinction debt. Nature 371, 65–66.

Tocher, M., Gascon, C. and Zimmerman, B. (1997) Fragmentation effects on a central

Amazonian frog community: a ten-year study. Pp. 124–137 in Laurance, W.F. and

Bierregaard Jr., R.O. (Eds). Tropical Forest Remnants: Ecology, Management, and

Conservation of Fragmented Communities. University of Chicago Press, Chicago,

Illinois.

Tocher, M.D. (1998) A comunidades de anfíbios da Amazônia central: diferenças na

composição específica entre a mata primária e pastagens. Pp. 219–232 in Gascon, C. 59

and Moutinho, P. (Eds). Floresta Amazônica: dinâmica, regeneração e manejo .

Instituto Nacional de Pesquisas da Amazônia. Manaus, Brazil.

Tscharntke, T., Gathmann, A. and Steffan-Dewenter, I. (1998) Bioindication using trap-

nesting bees and wasps and their natural enemies: community structure and

interactions. Journal of Applied Ecology 35, 708–719.

Tscharntke, T., Sekercioglu, C.H., Dietsch, T.V., Sodhi, N.S., Hoehn, P. and Tylianakis, J.M.

(2008) Landscape constraints on functional diversity of birds and insects in tropical

agroecosystems. Ecology 89, 944–951

Turner, I.M. (1996) Species Loss in Fragments of Tropical Rain Forest: A Review of the

Evidence. Journal of Applied Ecology 33, 200–209.

Turner, M.G. (2005) Landscape ecology: what is the state of the science? Annual Review of

Ecology, Evolution and Systematics 36, 319–344.

Umetsu, F., Metzger, J.P. and Pardini, R. (2008) Importance of estimating matrix quality for

modeling species distribution in complex tropical landscapes: a test with Atlantic

forest small mammals. Ecography 31, 359–370.

UN-REDD Programme, www.un-redd.org/.

Usher, M.B. and Keiller, S.W.J. (1998) The macrolepidoptera of farm woodlands:

determinants of diversity and community structure. Biodiversity and Conservation 7,

725–748.

Van Someren, V.G.L. (1939) Butterflies of Kenya and Uganda vol. II, part II ().

Journal of East Africa Natural History Society 14, 15–100.

Van Someren, V.G.L. and Rogers, K.St.A. (1932) The butterflies of Kenya and Uganda, part

10. Journal of East Africa Natural History Society 42, 141–172.

Warren, A. (1993) Naturalness: a geomorphological approach. Pp. 15–24 in Goldsmith, F.B.

and Warren, A. (Eds). Conservation in Progress. John Wiley and Sons, New York. 60

Warren, M.S. and Bourn, N.A.D. (2011) Ten challenges for 2010 and beyond to conserve

Lepidoptera in Europe. Journal of Insect Conservation 15, 321–326.

West, P., Igoe, J. and Brockington, D. (2006) Parks and Peoples: The Social Impact of

Protected Areas. The Annual Review of Anthropology 35, 251–277.

Williams, A.P., and Funk, C. (2011) A westward extension of the warm pool leads to a

westward extension of the Walker circulation, drying eastern Africa. Climate

Dynamics 37, 2417–2435.

Williams, N.M., Crone, E.E., Roulston, T.A.H., Minckley, R.L., Packer, L. and Potts, S.G.

(2010) Ecological and life-history traits predict bee species responses to environmental

disturbances. Biological Conservation 143, 2280–2291.

Williams, P.H., Moore, J.L., Toham, A.K., Brooks, T.M., Strand, H., D‘Amico, J., Wisz, M.,

Burgess, N.D., Balmford, A. and Rahbek, C. (2003) Integrating biodiversity priorities

with conflicting socio-economic values in the Guinean–Congolian forest region.

Biodiversity and Conservation 12, 1297–1320.

Wilson, E. O. (1985) The biological diversity crisis. Bioscience 35, 700–706.

Winfree, R., Griswold, T. and Kremen, C. (2007) Effect of Human Disturbance on Bee

Communities in a Forested Ecosystem. Conservation Biology 21, 213–223.

Winterbottom, B. and Eilu, G. (2006) Uganda Biodiversity and Tropical Forest Assessment,

Final Report. A Publication for the United States Agency for International

Development.

Wolda, H. (1978) Fluctuations in abundance of tropical insects. American Naturalist 112,

1017–1045.

Wolfe, D.A., Champ, M.A., Flemer, D.A. and Mearns, A.J. (1987) Long-term biological data

sets: Their role in research, monitoring, and management of estuarine and coastal

marine systems. Estuaries 10, 181–193. 61

Yang, L.H. and Gratton, C. (2014) Insects as drivers of ecosystem processes. Current Opinion

in Insect Science 2, 26–32.

Young, M. (1997) The natural history of moths. Poyser, London. 271 pp.

Zuidema, P.A., Sayer, J.A. and D, W. (1996) Forest fragmentation and biodiversity: the case

for intermediate-sized conservation areas. Environmental conservation 23, 290–297.

62

Plates showing representatives of different insect groups sampled.

Plate 3. Photographs of some butterflies recorded by this study

Plate 4. Photographs of some silk moths (Emperors) recorded by this study

Plate 5. Photographs of some hawk moths recorded by this study

Plate 6. Photographs of some grasshoppers recorded by this study

63

Plate 3. Representative butterfly species recorded during the study

Papilio lormieri Papilio dardanus Papilio demodocus

Papilio dardanus

Larinopoda terea protoclea Protogoniomorpha parhassus

Euphaedra medon Cymothoe herminia Bicyclus graueri

Pseudoneptis bugandensis Hypolycaena antifaunus64 Celaenorrhinus galenus

Plate 4. Representative emperor moths species recorded during the study

Tagoropsis genoviefae Epiphora rectifascia Imbrasia dione

Imbrasia epimethea Aurivillius aratus Goodia falcata

Bunaea alcinoe Pseudobunaea tyrrhena Pseudimbrasia deyrollei

Goodia lunata Pselaphelia flavivitta Ludia delegorguei 65

Plate 5. Representative hawk moths species recorded during the study

Nephele Polyptychus Polyptychus carteri Daphnis nerii accentifera orthographus

Macropoliana Euchloron megaera Nephele accentifera Nephele rosae natalensis

Dovania poecila Pseudoclanis postica Acheronita atropos 66

Plate 6. Representative grasshopper species recorded during the study

Oshwea dubiosa Afromastax rubricosta Pterotiltus hollisi Auloserpusia poecila

Abisares viridipennis Auloserpusia Abisares viridipennis Usambilla sagonai Taphronota c. dimidiata synpicta

Paracoptacra cauta Odontomelus Heteracris Humbe tenuicornis Paracoptacra cauta kwidschwianus67 guineensis

Papers I- IV

Paper I

Akite, P., Telford, R.J., Waring, P., Akol, A.M. & Vandvik, V. (2015)

Temporal patterns in Saturnidae (silk moth) and Sphingidae (hawk

moth) assemblages in protected forests of central Uganda.

Ecology and Evolution 5 (8): 1746–1757. Temporal patterns in Saturnidae (silk moth) and Sphingidae (hawk moth) assemblages in protected forests of central Uganda Perpetra Akite1,2, Richard J. Telford2, Paul Waring3, Anne M. Akol1 & Vigdis Vandvik2 1Department of Biological Sciences, Makerere University, Kampala, Uganda 2Department of Biology, University of Bergen, Bergen, Norway 3Windall View, Werrington, Peterborough, UK

Keywords Abstract Compositional change, extinction debt, forest degradation, Lepidoptera, matrix Forest-dependent biodiversity is threatened throughout the tropics by habitat intensification, resampling, species decline. loss and land-use intensification of the matrix habitats. We resampled historic data on two moth families, known to play central roles in many ecosystem pro- Correspondence cesses, to evaluate temporal changes in species richness and community struc- Perpetra Akite, Department of Biological ture in three protected forests in central Uganda in a rapidly changing matrix. Sciences, Makerere University, P.O. Box 7062 Our results show some significant declines in the moth species richness and the Kampala, Uganda. relative abundance and richness of forest-dependent species over the last 20– Tel: +256 772 902633; Fax: +256 414 531061; 40 years. The observed changes in species richness and composition among dif- E-mail: [email protected] ferent forests, ecological types, and moth groups highlight the need to repeat- edly monitor biodiversity even within protected and relatively intact forests. Funding Information The research was funded by the Norwegian Research Council FRIMUF programme through the MATRIX project (# 184912) and supported by the University of BergenMakerere University Collaboration.

Received: 20 December 2014; Revised: 2 March 2015; Accepted: 5 March 2015

Ecology and Evolution 2015; 5(8): 1746– 1757 doi: 10.1002/ece3.1477

Introduction matrix landscapes (Gascon et al. 1999). Different species perceive the matrix differently: what is inhospitable to For terrestrial ecosystems, the most important driver for one may be habitable to another and what is a barrier to biodiversity change in the last 50 years has been land cover one may be easily traversed by another (Bowler and Ben- change (e.g., Sala et al. 2000; Fahrig 2003; Foley et al. 2005; ton 2005; Eycott et al. 2012). Consequently, ecological MEA 2005). In some regions, less than 10% of the original patterns and processes within patches may also be influ- vegetation remains after clearance for agriculture and other enced by the nature of the surrounding matrix landscape purposes (Saunders et al. 1991). Tropical forests are among (Ricketts 2001; Vandermeer and Carvajal 2001; Prugh the habitats experiencing the highest loss rates. In addition et al. 2008) particularly in species exhibiting a metapopu- to the area loss per se, deforestation has resulted in frag- lation structure (Hanski 1998). mentation of once-continuous forests; the resulting frag- Species response to landscape changes is influenced by ments are now surrounded by a matrix of other land-uses key ecological and life-history attributes such as longevity, (Fahrig and Merriam 1994). reproductive rates, body size, trophic specialization, and Species vulnerability to forest area loss and fragmenta- dispersal ability. These traits directly determine changes tion is strongly affected by their ability to use these in abundance and mediate extinction risk (Pimm et al.

1746 ª 2015 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. P. Akite et al. Moths in Protected Forests of Uganda

1993). Identifying how species with different traits are dif- well protected and have experienced little structural and ferentially affected by landscape change allows insight into tree compositional change in recent decades (e.g., Bulafu shifts in communities or guild composition beyond sim- et al. 2013). Our study aims to resample historical data ple changes in species richness (Williams et al. 2010). on moth communities to give insights on the long-term One way to understand the mechanisms that determine ecological integrity of these protected forests in rapidly the structure of communities is through the comparison changing matrix landscapes. Based on the literature sum- of species diversity at different spatial and temporal scales marized above, we test the predictions that (1) moth spe- in different ecological and biogeographical settings (Fuka- cies richness has declined over time; (2) forest-dependent mi and Wardle 2005). Temporal patterns of biodiversity species are more affected than the generalist species; and have received much less attention than spatial ones (Ma- (3) Saturnidae will have declined more rapidly than Sph- gurran et al. 2010). Such comparisons are important in ingidae. developing and planning conservation program (e.g., Kre- men et al. 2007) and provide a vital tool for management Materials and Methods of wildlife (Turner et al. 2003). Insects make an enormous contribution to diversity Study areas and ecosystem functions (Lewinsohn et al. 2005) but knowledge of population changes in insects lags behind The study was conducted in three protected forests in that of vertebrates and vascular plants (Thomas et al. central Uganda: Zika, Mpanga, and Mabira (Fig. 1). 2004). Despite their importance for many critical eco- These forests are near Lake Victoria, in the wettest dis- logical functions, unparalleled contribution to biodiver- tricts of central Uganda where mean annual rainfall sity, and their potential use in conservation planning, ranges from 1200 to 1600 mm and mean annual tempera- long-term ecological studies of invertebrates are extre- ture is 28°C (MWLE 2002). They lie within an elevational mely scarce (Kremen et al. 1993). For Lepidoptera, data range of 1070 – 1340 m asl. on population size fluctuations and associated changes at community level are available for many temperate Study sites ecosystems but much less is known about their dynam- ics in the humid and seasonal tropics (e.g., Schulze and Zika (0°070N, 32°310E) covers an area of 0.13 km2. The Fiedler 2003). Moths play a central role in many ecosys- forest is part of a narrow sinuous strip of lakeside forests tem processes as prey, herbivores, and pollinators (Jan- skirting the extensive grass and papyrus swamps of Waiya zen 1987; Barlow and Woiwod 1989). Saturnidae (Silk Bay, a sheltered inlet of Lake Victoria. Buxton (1952) rec- moths) and Sphingidae (Hawk moths) are two of the ognized three zones in the forest: a) permanent swamp most species-rich families of moths in the tropics (Jan- forest dominated by Mitragyna stipulosa, Erythrina excelsa, zen 1984). They can be rapidly surveyed, identified and and Voacanga obtusa; b) raised wet forest dominated by are relatively well documented, thus well placed to act Pseudospondias microcarpa, Parkia filicoidea, and Maca- as indicator groups. Members of the two families have ranga monandra; c) raised seasonal forest dominated by differing life histories and feeding habits. Tropical Sph- Lovoa brownie, Maesopsis eminii, and Piptadenia africana. ingidae are long-lived with a few exceptions, mate The forest has been under the jurisdiction of the Uganda repeatedly, lay a few eggs per host plant and oviposit Virus Research Institute since 1960. through adult life (e.g., Haber and Frankie 1989). They Mpanga (0°150N, 32°180E) covers an area of 4.53 km2. are accomplished fliers with migratory tendencies It is a remnant tropical, medium altitude, moist ever- in some taxa. In contrast, Saturnidae have short non green, and swamp forest comprised of a) swamp – per- feeding flight periods, lasting less than 10 days, mate manently flooded or water logged Mitragyna–Phoenix once and lay many eggs. They are often associated with associations; b) the slopes with Celtis–Aningeria associa- undisturbed forest habitats. Caterpillars of Saturnidae tions; and c) main forest dominated by Pseudospondias often select older leaves and are usually found in microcarpa, Erythrina excelsa, Canarium schweinfurthii, crowns of adult trees or woody vines, while Sphingidae and Entandrophragma angolense (Buxton 1952). The for- are less particular about plant age and commonly feed est was gazetted as a nature reserve in 1950 and is under on young leaves (Janzen 1984). These factors make it the jurisdiction of the National Forestry Authority (NFA). likely that Sphingidae are typically better dispersers and Mabira (0°240–0°350N, 32°520–33°070E) covers an area less habitat specific than Saturnidae. of 306 km2, the largest block of moist semi-deciduous Most Ugandan forests and their matrix landscape have forest remaining in central Uganda (Carswell 1986). The undergone considerable changes in recent decades (Obua reserve is considered to be a secondary forest and Howard et al. 2010). Many forests have been lost, but some are (1991) described four major types within this a) younger

ª 2015 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. 1747 Moths in Protected Forests of Uganda P. Akite et al.

Figure 1. Location of the surveyed forests including major habitat types in the matrix. secondary forests dominated by colonizing Maesopsis emi- the forests. In 2000, there was 3% forest around Zika, nii; b) valley bottom forest dominated by insig- 17% around Mpanga, and 5% around Mabira. Loss nis; c) the Celtis–Holoptelea dominated forest; and d) 2000–2012 around Zika was 2.3%, Mpanga 4.7%, and mixed communities. As noted by Winterbottom and Eilu Mabira 1.7%. (2006) and from field observations, vast areas of the for- est are now covered by the exotic species Broussonetia pa- Field methods pyrifera. The forest is protected and managed as a Central Nature Reserve by the NFA. The historic data have been sampled in two periods. In Zika and Mpanga forests have remained relatively Zika forest, the Saturnidae were surveyed by Angus McC- undisturbed and unchanged internally (Bulafu et al. 2013), rae between February 1969 and April 1971 (unpublished) but their surrounding matrix landscape is substantially and by the Uganda Forest Department (FD) between altered; the majority of neighboring forest fragments in the March 1993 and January 1995 (Howard and Davenport greater Kampala–Entebbe landscape (Bulafu et al. 2013) 1996). The FD surveys included the Sphingidae. In and the Mpigi archipelago have either been cleared or Mpanga, the FD carried out moth surveys between Sep- greatly reduced in extent and/or quality. Parts of Mabira tember 1993 and February 1995 and in Mabira, the sur- are recovering from the encroachments of the 1970s–1980s veys were carried out between October 1992 and February (MWLE 2008), and large sections of the nature reserve are 1995 (Howard and Davenport 1996). relatively stable, albeit with minor disturbances from illegal McCrae sampled the moths using a single Robinson logging, but the forest area has declined as edges have been trap powered by a portable generator and operated from lost to sugarcane, tea, cardamom or oil palm plantations, dusk to dawn. Short-term trapping was done in the per- plus non-native monoculture forestry and small- to med- iod February to March 1969, followed by an intensive ium-scale agro-ecosystems (MWLE 2008). study from April 1969 to April 1971 (unpublished). The To assess recent changes in the matrix, we calculated moths were identified by McCrae, and all voucher speci- the percentage of the forest area in a 5-km buffer around mens from this study are deposited at the Hope Museum each forest, and the loss of forest over the period 2000- in Oxford. 2012 based on the remote sensing analysis of Hansen The 1990s data come from short-term surveys and a et al. (2013). We reclassify Hansen et al. (2013) so that more intensive sampling program run by the FD. The cells with 60% forest cover or above are classified as moths were sampled using a 125-watt choked mercury forest to reduce misclassification of papyrus swamps near vapor lamp mounted in a Skinner box trap. The trap was

1748 ª 2015 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. P. Akite et al. Moths in Protected Forests of Uganda powered by a portable generator and operated from dusk Howard and Davenport (1996) describe the ecological to dawn. The FD surveys were conducted to be as similar habitat preferences of all species they observed based as possible to the earlier study by McCrae. All moths were on prior knowledge of their ecology (Carcasson 1976). identified by Peter Howard in consultation with McCrae, Habitat preferences include: F) forest-dependent species and voucher specimens are deposited at the Makerere restricted to closed-canopy forest habitats; f.) forest University Zoology Museum. Howard and Davenport nondependent species not infrequently recorded in (1996) give detailed findings of sampling in each forest, closed-canopy forest, but also encountered in a variety of including dates, captures per night, and total trap nights. forest edge, degraded forest, and woodland habitats; G) The resampling data were collected for the two moth nonforest species characteristic of open habitats such as families in the three forests from September 2010 through grassland, open savannah, and arid habitats; and W) March 2011. Traps were placed away from the edge by at widespread species, generalist that occur in a variety of least (≥100 m) except in the much smaller Zika forest forest and nonforest habitats. where traps were placed in the center of the forest. We surveyed the same localities as Howard and Davenport Data analysis (1996) and trap locations remained the same throughout the entire survey. In Zika, sampling was carried out in Trap data were pooled for each of the two moth families the periods 3–11 January then 13–23 February 2011; in in each forest per sampling period. We calculated the Mpanga, sampling was carried out in the periods 15–29 exponent of the bias-corrected Shannon index (Chao and December 2010 then 13–23 February 2011; and in Mabira, Shen 2003) for each sample period, called the “effective sampling was carried out in the period 9 September – 10 number of species” of the community (Hill 1973). This December 2010. Sampling was carried out using a index converges rapidly with little bias even for small portable light-trap consisting of a 15-watt actinic tube samples (Magurran 2004). (Sylvania blacklight F 15 w/BLB–TB) run on a portable The observed number of species is a misleading indica- car battery (32 amps) and a net and was run from dusk tion of species richness because of the difficulty of obtain- to dawn. Actinic traps are known to be highly effective ing a complete inventory of species-rich communities and minimize cross attraction of moths between sampled (Price et al. 1995). Individual-based rarefaction curves habitats (Muirhead-Thomson 1991; Schulze and Fiedler were therefore used to evaluate the effectiveness of sam- 2003). Each morning, special attention was given to pling and for comparison of species accumulation curves searching the area around the trap for any moths that following Gotelli and Colwell (2001). For the statistical had been attracted to the light, but had not entered comparison of the accumulation curves, we calculated the the trap; individuals of the target families were hand- rarefied number of species and the 95% confidence inter- collected. val using bootstrap resampling with replacement using Specimens of the two families were first photographed EstimateS Version 9.1.0 (Colwell 2006). at the trap, and then, representative samples collected and The proportion of the total moth fauna belonging to dried in envelopes. Voucher specimens were examined each ecological habitat preference types was calculated and identified (where possible) with reference to Pinhey based on abundance and presence/absence data. We per- (1972) and D’Abrera (1986), and the reference collections formed a chi-square test for homogeneity to check from the FD surveys. Clarification was sought for some whether the observed differences in richness and abun- species by checking the extensive collections of A. McCrae dances of the different ecotypes over time within the two at The Hope Museum (Oxford, UK) and the collections moth families are significant. Data were analyzed using at The Natural History Museum London (NHM). Some the statistical program R (v. 2.13.1, R Core Team 2013). of the Sphingidae were identified by Ian Kitching. Results Data combination and species classification The combined dataset consists of 3687 individuals (54 The sampling effort differed between the sampling peri- species) of Saturnidae and 1041 individuals (49 species) ods. The 1970s and 1990s species lists are extensive – a of Sphingidae across the three forests. The number of large fraction of the species that occurred in (and possibly moths differed with effort between sampling periods, but around) these forests were probably recorded. To achieve the mean numbers of individuals per trap night are gen- meaningful comparisons, subsets of data from the 1970s erally comparable across the sampling periods (Table 1). and 1990s were extracted which covered only the same There was no overall pattern in the effective number of portion of the year as the 2010s surveys (Table 1). These species over time (Table 1). Visual assessment of the rare- are referred to as the 1970, 1990, and 2010 data hereafter. faction curves reveals that many of the curves do not

ª 2015 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. 1749 Moths in Protected Forests of Uganda P. Akite et al.

Table 1. Moth total abundance, mean number of individuals per trap night, number of species, and the exponent of the Shannon in the three forests.

Total trap Number Number Mean number of Moth family Forest sampled Sample period Year nights of moths of species individuals per night expH’

Saturnidae Zika MaCrae’s subset 1969–1971 161 1765 54 11.0 25.0 Forest Dept subset 1993–1995 82 474 33 5.8 17.4 Resample 2011 12 131 20 10.9 15.2 Mpanga Forest Dept subset 1993–1995 27 161 32 6.0 24.7 Resample 2010–2011 18 151 26 8.4 23.0 Mabira Forest Dept subset 1992–1995 98 528 40 5.4 24.0 Resample 2010 42 427 35 10.2 28.1 Sphingidae Zika Forest Dept subset 1993–1995 82 1143 46 13.9 23.2 Resample 2011 12 196 34 16.3 29.3 Mpanga Forest Dept subset 1993–1995 27 500 46 18.5 20.7 Resample 2010–2011 18 321 29 17.8 23.4 Mabira Forest Dept subset 1992–1995 98 867 39 8.8 16.5 Resample 2010 42 524 29 12.5 17.6

Zika Mpanga Mabira Saturnidae

1970 1990 0 102030405060 0 102030405060 2010 0 102030405060 0 500 1000 1500 0 500 1000 1500 0 500 1000 1500 Number of species Sphingidae 0 1020304050 0 1020304050 0 1020304050 0 200 400 600 800 0 200 400 600 800 0 200 400 600 800 Number of Individuals

Figure 2. Individual-based rarefaction curves for the moths ( CI 95%; Green = 1970, Blue = 1990, Red = 2010), for three different forest reserves in central Uganda. approach asymptotes (Fig. 2). Generally, the curves in the The proportions of individuals belonging to forest- 2010 resampling were lower than that in the earlier peri- dependent species are greatly reduced in the 2010 sam- ods (Fig. 2). Comparing species accumulation curves (CI: pling period compared with the earlier sampling peri- 95%) at the lowest abundance value (2010 resample per- ods (Fig. 3). Proportion varied significantly with iod) show clear decreases for the Sphingidae in Mpanga sampling period among the Saturnidae in Zika and Mabira, less conclusive patterns for Saturnidae in (v2 = 297.81, df = 6, P < 0.0001, Fig. 3) and Mpanga Zika and Mpanga, and no support for change for Saturni- (v2 = 56.37, df = 3, P < 0.0001, Fig. 3), and also among dae in Mabira and Sphingidae in Zika (Fig. 2). the Sphingidae in Mpanga (v2 = 26.55, df = 3,

1750 ª 2015 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. P. Akite et al. Moths in Protected Forests of Uganda

Zika Mpanga Mabira 1.0

0.8

0.6 Saturnidae 0.4

0.2

0.0 1970 1990 2010 1990 2010 1990 2010

1.0 Proportion of individuals

0.8

0.6 Sphingidae Figure 3. Proportional changes in composition 0.4 based on individual abundances of each ecotype over time. Black = forest-dependent species; dark gray = forest nondependent 0.2 species; white = open habitat species, and light gray = widespread species. The width of the bars is proportional to the total number of individuals. The small black line on the right 0.0 represents eight individuals per trap night. 1990 2010 1990 2010 1990 2010

P < 0.0001, Fig. 3) and Mabira (v2 = 100.73, df = 2, There was turnover in species over the years P < 0.0001, Fig. 3). Saturnidae in Mabira show a mar- (Table 2). Nine species of Saturnidae from the 1970s in ginally significant decline (v2 = 6.94, df = 3, P = 0.07, Zika are absent from the 1990s resampling with one Fig. 3), and there was no statistical difference among rediscovered in the 2010 sampling period. Orthogoniopti- the Sphingidae in Zika (v2 = 4.50, df = 3, P = 0.212, lum luminosum, relatively abundant in Zika forest in the Fig. 3). 1970s, has not been recorded in the subsequent 1990 The same pattern is reflected in the species richness, and 2010 resample periods. Two species, Temnora hol- with an increase in the proportions of forest edge and landi and Imbrasia oyemensis, known only from the widespread species and a decline in the forest-dependent Kampala–Entebbe area in the East African parts of their species (Fig. 4). This is only statistically significant for ranges and only recorded in Zika in the 1970s (Angus Sphingidae in Mabira (v2 = 7.35, df = 2, P = 0.025, McCrae, personal notes) were absent from the 1990 and Fig. 4), whereas there was no statistical difference for Sa- 2010 resampling. turnidae (Zika: v2 = 8.78, df = 6, P = 0.175; Mpanga: v2 = = = v2 = = 4.40, df 3, P 0.221; Mabira: 2.20, df 3, Discussion P = 0.531, Fig. 4) and for Sphingidae in Zika (v2 = 2.41, df = 3, P = 0.492, Fig. 4) and Mpanga (v2 = 3.81, df = 3, Despite only minor changes in forest structure and vege- P = 0.283, Fig. 4). tation of our forests in recent decades (Obua et al. 2010;

ª 2015 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. 1751 Moths in Protected Forests of Uganda P. Akite et al.

Zika Mpanga Mabira 1.0

0.8

0.6 Saturnidae 0.4

0.2

0.0 1970 1990 2010 1990 2010 1990 2010

1.0 Proportion of species

0.8

0.6 Sphingidae 0.4

Figure 4. Proportional changes in composition 0.2 based on total number of species of each ecotype over time. See Fig. 3 for legend. The width of the bars is proportional to the total 0.0 number of species; the black lines on the right 1990 2010 1990 2010 1990 2010 represent 12 species.

Bulafu et al. 2013), we observe large changes in moth mately similar efficiencies. Summerville and Crist (2005) communities. Although we do not have good estimates of found that differences among trap types contributed less the total moth richness, rarefaction curves suggest declin- than 10% to differences in richness. A similar study on ing diversity. The most striking pattern is the change in moths in a lowland dipterocarp forest in Peninsular the ecotype composition of the forests with consistent Malaysia using two different trap types (Intachat and Wo- decline in the relative abundance of forest-dependent spe- iwod 1999) found no significant differences between over- cies and an associated increase in widespread species. In all diversity for geometroidea between the trap types and line with our predictions, Saturnidae were more affected. that total catches for nongeometroidea were remarkably These declines in moths reported here are in line with similar. The reduced cross-habit attraction of actinic traps patterns of moth declines recorded elsewhere around the (Muirhead-Thomson 1991; Schulze and Fiedler 2003) will world (e.g., Great Britain, Conrad et al. 2004; Finland, bias our results, but the direction of bias is opposite to Hulden et al. 2000). Our results give some insights into the patterns we observe. Regarding sampling effort, a the ecological processes operating in these forests and the comparison between long-term sampling and short-term surrounding landscapes. but intensive sampling yielded a 76% overlap in species Although sampling protocols and sampling efforts var- recorded (Landau et al. 1999). Similarly, Summerville and ied between the sampling periods, the number of individ- Crist (2005) reported that increased sampling effort only uals per trap night was comparable across the sample reduced the proportion of singletons and unique species periods, suggesting that different trap types have approxi- and this peaked off after 10 trap nights. Therefore, we

1752 ª 2015 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. P. Akite et al. Moths in Protected Forests of Uganda

Table 2. Number of Saturnidae and Sphingidae species in the different sampling periods.

Zika Mpanga Mabira

Categories Saturnidae Sphingidae Saturnidae Sphingidae Saturnidae Sphingidae

1970 only 19 1970 + 1990 15 All three samples 18 1970 + 2010 2 1990 only 0 17 10 22 12 16 1990 + 2010 0 29 20 24 28 23 2010 only 0 5 6 5 7 7

believe that the observed changes in rarefied species rich- Rapid expansion and intensification of agriculture, cou- ness and species turnover among the Saturnidae and Sph- pled with loss and deterioration of suitable habitats, have ingidae within these protected forests are real. been implicated in the decline of moths elsewhere (e.g., The weak patterns in overall species richness and diver- Conrad et al. 2004; Fox 2012) and several other insect sity reflect the replacement of forest-dependent species groups (e.g., dung beetles, Nichols et al. 2007; butterflies, with widespread species. Thus, the dramatic decline of Ekroos et al. 2010). We found significant declines in the forest-dependent species relative abundance and richness moths’ richness and relative abundance of forest-depen- is masked by the rise of the widespread and generalist dent species especially in the poorly dispersed Saturnidae. taxa. This highlights the limited utility of diversity metrics In central Uganda, there has been massive intensification for conservation; indeed, they can be misleading. in the use of the matrix surrounding protected forests; Changes in moth diversity and abundance have often previously, forested areas have been replaced by exotic been correlated with or assumed to be caused by environ- plantations (e.g., oil palm, cardamom, or eucalyptus), mental changes within the study sites (Summerville et al. agro-ecosystems (e.g., shade coffee and home gardens), or 2004). Despite past encroachment, these forests have cleared for settlements and other human developments either remained relatively stable in size and structure (Obua et al. 2010). Matrix intensification can lead to a (Zika and Mpanga; Bulafu et al. 2013) or have been breakdown in metapopulation dynamics, making land- recovering from disturbance (Mabira) over the resampling use-driven environmental changes outside reserves just as period (Winterbottom and Eilu 2006; Obua et al. 2010). important as those within reserves in determining the fate It is therefore unlikely that the observed declines in forest of regional-scale biodiversity (e.g., Hanski 1998; Perfecto specialist species are only driven by environmental change and Vandermeer 2002). For example, a study on birds in within the forests. the farmed landscapes of Central and southwestern parts Alternative explanations for the observed patterns in of Uganda reported declines with increased land-use moth community within our study forests include the intensification especially among forest specialist species effect of increased isolation due to matrix transformation, (Bolwig et al. 2006). extinction debt, and climate change. Habitat isolation in In tropical forests where historical forest area loss and both space and time disrupts species distribution patterns, landscape change are high, extinction debt might drive consequently affecting metapopulation dynamics of species loss long after forest reserve boundaries have sta- patch-dwelling populations. This makes matrix habitats bilized (e.g., Kuussaari et al. 2009). The magnitude of strong determinants of fragmentation effects within extinction debt that can be expected is largely dependent remnants through regulating dispersal, dispersal-related on spatiotemporal configuration of habitat patches, the mortality, and mediating edge-related microclimatic gra- time since the habitat was altered and the nature of the dients (Ewers and Didham 2006). Such consequences of alteration (Kuussaari et al. 2009) but also on the life-his- isolation may be heightened in Zika forest where Bulafu tory traits of the assemblages. We expect that extinction et al. (2013) reported a 50% loss of all its neighboring debt would be repaid fastest in the small Zika forest, with forests over the last 20 years despite the forest itself its consequently small populations, leading to high rates remaining stable with very low levels of disturbance. This of species loss. is supported by the remote sensing forest loss analysis Rainfall patterns in central Uganda have changed in where over 60% of the forest area in a 5-km buffer recent decades, resulting in either less precipitation or around Zika has been lost, the highest proportion of any alteration in timing of the rainy season (e.g., MWLE of our forests (Hansen et al. 2013). 2002; Williams and Funk 2011; Michaelsen and Marshall

ª 2015 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. 1753 Moths in Protected Forests of Uganda P. Akite et al.

2012). This would affect adult moth emergence, often tainable harvesting for making drums (Omeja et al. 2005; triggered by rainfall signaling larval food availability. Pro- Were 2010), and this species of moth was missing in the nounced fluctuations in the abundance of individual spe- 2010 resample period. Imbrasia anna which feeds on sev- cies or entire guilds of moths over seasons are reported to eral members of the Arecaceae family had larger popula- be frequent (e.g., Fiedler and Schulze 2004). Climate tions in our dataset and has a wide geographic range effects have been reported in Britain and northwestern compared to Imbrasia oyemensis which is a forest-depen- Europe, where substantial decreases in the overall abun- dent species that only feeds on Entandrophragma ango- dance of macro-moths and populations of many wide- lense – an IUCN red listed tree species. However, a spread species have been attributed to habitat loss in shortage of trait data for most species of tropical Africa, combination with climate change (Conrad et al. 2006; especially larval host plants, hinders our inferences in this Fox 2012). regard.

Compositional change in ecological types Saturnidae and Sphingidae In general, habitat specialists are more susceptible to hab- We found steeper declines among the Saturnidae than the itat loss and degradation than generalists (e.g., Ockinger€ Sphingidae. This is in accordance with our expectations et al. 2010). In the earlier sampling periods, forest-depen- based on the lifestyle and dispersal ability of the two fami- dent species especially those associated with woody plants lies. Saturnidae caterpillars tend to feed on older leaves and were prevalent and abundant (McCrae unpublished, are often found in the crowns of trees, whereas the Sphingi- Howard and Davenport 1996). Our 2010 data are charac- dae tend not to be particular about plant age and com- terized by species that have wider ranges and do not need monly feed on younger leaves (Holloway and Hebert 1979; good quality forests to survive. Bernays and Janzen 1988). This predisposes Saturnidae to The observed declines in forest-dependent species vs. greater impacts from habitat disturbance and especially increases in forest edge and widespread species within our when mature trees and woody vines are continuously taken study forests are indicative of the ability of nonspecialist out of their ecosystems (Basset 1992; Kitching et al. 2000). species to utilize a much broader range of habitats across the landscape compared to the specialist species that Conclusions require intact and less disturbed habitats for their sur- vival. In a connected landscape, forest-dependent species Our results show some significant change in the moth may benefit from reinforcement between fragments as a communities in the target forests in the last 20–40 years. result of metapopulation dynamics rescuing species from This highlights the need to repeatedly monitor biodiver- imminent extinction (Nee and May 1992; Hanski 1998). sity even within protected and relatively intact forests. The severity of deforestation in our study area will clearly Our findings together with similar patterns reported for reduce the opportunities for forest-dependent species to trees (Bulafu et al. 2013) in similar habitat settings indi- cope with the changes in their environment. cate a worrying reduction in the capacity of protected for- Kitching et al. (2000) and Usher and Keiller (1998) ests in central Uganda to maintain biodiversity. Matrix both note that forest specialist moths tend to be monoph- intensification around our forests appears to have reduced agous and feed on woody plants, trees, and vines, whereas the capacity of the landscape to buffer and support popu- moths that favor disturbed sites are often polyphagous, lations in protected forests. Removal of any functional feeding on herbaceous and weedy food plants. This could groups will alter the ecological integrity of these forests. potentially account for some of the variability in our data. In our case, if an important fraction or entire guilds of Individual species’ responses can thus be interpreted species are lost (e.g., forest-dependent species as our data through their guild membership – as forests are lost or suggest), detrimental effects on ecological services medi- altered, monophagous (forest specialist) species are more ated by these moths could become apparent. Protected likely to decline or go locally extinct than those that are forests are linked ecologically to their surrounding habi- polyphagous (i.e., generalist, Holloway and Hebert 1979). tats, and failure to stem broad-scale loss and degradation Several previously common forest-dependent species were of such habitats could sharply increase the likelihood of absent from subsequent resampling periods (e.g., Imbrasia serious biodiversity declines. anthina), while forest edge and widespread species (e.g., Cirina forda and Imbrasia anna) became more common. Acknowledgments The Parasol tree, (Polyscias fulva) which is a larval food plant for Imbrasia anthina, is now scarcely found in its We are grateful to Hugh Rowell for reading the manu- natural habitat within our study forests as result of unsus- script drafts; Darren Mann and James Hogan at the Hope

1754 ª 2015 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. P. Akite et al. Moths in Protected Forests of Uganda

Museum, Oxford, for help accessing Angus McCrae’s field Conrad, K., I. Woiwod, M. Parsons, R. Fox, and M. Warren. notes and moth collections; Ian Kitching of Natural His- 2004. Long-term population trends in widespread British tory Museum, London (NHM), for identifying difficult moths. J. Insect Conserv. 8:119–136. Sphingidae; Alessandro Giusti of NHM for locating all Conrad, K. F., M. S. Warren, R. Fox, M. S. Parsons, and I. P. the relevant collections; and Janet McCrae for hosting PA Woiwood. 2006. Rapid declines of common, widespread in Oxford, Amy Eycott for processing remote sensing data British moths provide evidence of an insect biodiversity and commenting on the manuscript, and four anonymous crisis. Biol. Conserv. 132:279–291. reviewers for their comments on earlier versions of this D’Abrera, B. 1986. Sphingidae Mundi: hawk moths of the manuscript. The research was funded by the Norwegian world. E.W Classey, Oxon, UK. Research Council FRIMUF programme through the Ekroos, J., J. Heliol€ €a, and M. Kuussaari. 2010. Homogenization MATRIX project (# 184912) and supported by the Uni- of lepidopteran communities in intensively cultivated – versity of Bergen–Makerere University Collaboration. agricultural landscapes. J. Appl. Ecol. 47:459 467. Ewers, R. M., and R. K. Didham. 2006. Confounding factors in the detection of species responses to habitat Conflict of Interest fragmentation. Biol. Rev. 81:117–142. None declared. Eycott, A., G. Stewart, L. Buyung-Ali, D. Bowler, K. Watts, and A. Pullin. 2012. A meta-analysis on the impact of References different matrix structures on species movement rates. – Barlow, H. S., and I. P. Woiwod. 1989. Moth diversity of a Landscape Ecol. 27:1263 1278. tropical forest in Peninsular Malaysia. J. Trop. Ecol. 5:37– Fahrig, L. 2003. Effects of Habitat Fragmentation on – 50. Biodiversity. Annu. Rev. Ecol. Evol. Syst. 34:487 515. Basset, Y. 1992. Host specificity of arboreal and free-living Fahrig, L., and G. Merriam. 1994. Conservation of Fragmented – insect herbivores in rain forests. Biol. J. Linn. Soc. 47:115– Population. Conserv. Biol. 8:50 59. 133. Fiedler, K., and C. H. Schulze. 2004. Forest modification Bernays, E. A., and D. H. Janzen. 1988. Saturniid and affects diversity (but not dynamics) of speciose tropical – Sphingid caterpillars: two ways to eat leaves. Ecology pyraloid moth communities. Biotropica 36:615 627. 69:1153–1160. Foley, J. A., R. DeFries, G. P. Asner, C. Barford, G. Bonan, S. Bolwig, S., D. Pomeroy, H. Tushabe, and D. Mushabe. 2006. R. Carpenter, et al. 2005. Global consequences of land use. – Crops, trees, and birds: biodiversity change under Science 309:570 574. agricultural intensification in Uganda’s farmed landscapes. Fox, R. 2012. The decline of moths in Great Britain: a review – Danish J. Geogr. 106:115–130. of possible causes. Insect Conserv. Divers. 6:5 19. Bowler, D. E., and T. G. Benton. 2005. Causes and consequences Fukami, T., and D. A. Wardle. 2005. Long-term ecological of animal dispersal strategies: relating individual behaviour to dynamics: reciprocal insights from natural and – spatial dynamics. Biol. Rev. 80:205–225. anthropogenic gradients. Proc. R. Soc. B Biol. Sci. 272:2105 Bulafu, C., D. Baranga, P. Mucunguzi, R. J. Telford, and V. 2115. Vandvik. 2013. Massive structural and compositional Gascon, C., T. E. Lovejoy, R. O. Bierregaard, J. R. Malcolm, P. changes over two decades in forest fragments near Kampala, C. Stouffer, H. L. Vasconcelos, et al. 1999. Matrix habitat Uganda. Ecol. Evol. 3:3804–3823. and species richness in tropical forest remnants. Biol. – Buxton, A. P. 1952. Observations on the diurnal behaviour of Conserv. 91:223 229. the redtail monkey (Cercopithecus ascanius schmidti Matschie) Gotelli, N. J., and R. K. Colwell. 2001. Quantifying in a small forest in Uganda. J. Anim. Ecol. 21:25–58. biodiversity: procedures and pitfalls in the Carcasson, R. H. 1976. Revised catalogue of the African measurement and comparison of species richness. Ecol. – Sphingidae (Lepidoptera) with descriptions of the East Lett. 4:379 391. African species, pp. 148. E.W. Classey, Oxon, UK. Haber, W., and G. Frankie. 1989. A tropical hawkmoth Carswell, M.. 1986. Birds of the Kampala area (Scopus special community: Costa Rican dry forest Sphingidae. Biotropica – supplement), pp 89. Ornithological Sub-committee, EANHS, 21:155 172. Nairobi, Kenya. Hansen, M. C., P. V. Potapov, R. Moore, M. Hancher, S. A. Chao, A., and T.-J. Shen. 2003. Nonparametric estimation of Turubanova, A. Tyukavina, et al. 2013. High-Resolution Shannon’s index of diversity when there are unseen species Global Maps of 21st-Century Forest Cover Change. Science – in sample. Environ. Ecol. Stat. 10:429–443. 342:850 853. – Colwell, R. K. 2006. EstimateS: statistical estimation of species Hanski, I. 1998. Metapopulation dynamics. Nature 396:41 49. richness and shared species from samples. Version 9.1.0. Hill, M. O. 1973. Diversity and evenness: a unifying notation – http://viceroy.eeb.uconn.edu/estimates/. and its consequences. Ecology 54:427 432.

ª 2015 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. 1755 Moths in Protected Forests of Uganda P. Akite et al.

Holloway, J. D., and P. D. N. Hebert. 1979. Ecological and Sensing of Drought: Innovative Monitoring Approaches. taxonomic trends in macrolepidopteran host plant selection. ftp://chg.geog. ucsb.edu/pub/pubs/ Biol. J. Linn. Soc. 11:229–251. mapping_decadal_variations.pdf. Howard, P. 1991. Nature conservation in Uganda’s tropical Millennium Ecosystem Assessment. 2005. Ecosystems and forest reserves. IUCN, Gland, Switzerland and Cambridge, human well–being: biodiversity synthesis. World Resources UK. Institute, Washington. Howard, P. C., and T. R. B. Davenport. 1996. Forest Ministry of Water Lands and Environment. 2002. Department biodiversity reports. Uganda Forest Department, Kampala, of meteorology: “Uganda Initial National Communication Uganda. to the United Nations Framework Convention on Climate Hulden, L., A. Albrecht, J. It€amies, P. Malinen, and J. Change”. Ministry of Water, Lands and Environment, Wettenhovi. 2000. Atlas of Finnish macrolepidoptera. Kampala, Uganda. Lepidopterological Society of Finland, Finnish Museum of Ministry of Water Lands and Environment. 2008. Mabira Natural History, Helsinki. forest management plan. Draft report Ministry of Water Intachat, J., and I. P. Woiwod. 1999. Trap design for Lands and Environment, Kampala, Uganda. monitoring moth biodiversity in tropical rainforests. Bull. Muirhead-Thomson, R. C. 1991. Trap responses of flying Entomol. Res. 89:153–163. insects: the influence of trap design on capture efficiency. Janzen, D. H. 1984. Two ways to be a tropical big moth: Santa Academic Press, London. Rosa Saturnidae and Sphingidae. Oxf. Surv. Evol. Biol. 1:85– Nee, S., and R. M. May. 1992. Dynamics of metapopulations: 140. habitat destruction and competitive coexistence. J. Anim. Janzen, D. H. 1987. Insect diversity of a Costa Rican dry Ecol. 61:37–40. forest: why keep it, and how? Biol. J. Linn. Soc. 30:343–356. Nichols, E., T. Larsen, S. Spector, A. L. Davis, F. Escobar, Kitching, R. L., A. G. Orr, L. Thalib, H. Mitchell, M. S. M. Favila, et al. 2007. Global dung beetle response to Hopkins, and A. W. Graham. 2000. Moth assemblages as tropical forest modification and fragmentation: a indicators of environmental quality in remnants of upland quantitative literature review and meta-analysis. Biol. Australian rain forest. J. Appl. Ecol. 37:284–297. Conserv. 137:1–19. Kremen, C., R. K. Colwell, T. L. Erwin, D. D. Murphy, R. F. Obua, J., J. G. Agea, and J. J. Ogwal. 2010. Status of forests in Noss, and M. A. Sanjayan. 1993. Terrestrial arthropod Uganda. Afr. J. Ecol. 48:853–859. assemblages: their use in conservation planning. Conserv. O¨ ckinger, E., O. Schweiger, T. O. Crist, D. M. Debinski, J. Biol. 7:796–808. Krauss, M. Kuussaari, et al. 2010. Life-history traits predict Kremen, C., N. M. Williams, M. A. Aizen, B. Gemmill-Herren, species responses to habitat area and isolation – A cross G. LeBuhn, R. Minckley, et al. 2007. Pollination and other continental synthesis. Ecol. Lett. 13:969–979. ecosystem services produced by mobile organisms: a Omeja, P., J. Obua, and A. B. Cunningham. 2005. Demand conceptual framework for the effects of land-use change. and Supply of Wood for Drum Making in central Uganda. Ecol. Lett. 10:299–314. Int. Forest Rev. 7:21–26. Kuussaari, M., R. Bommarco, R. K. Heikkinen, A. Helm, J. Perfecto, I., and J. Vandermeer. 2002. Quality of Krauss, R. Lindborg, et al. 2009. Extinction debt: a challenge Agroecological Matrix in a Tropical Montane Landscape: for biodiversity conservation. Trends Ecol. Evol. 24:564–571. ants in Coffee Plantations in Southern Mexico. Conserv. Landau, D., D. Prowell, and C. E. Carlton. 1999. Intensive Biol. 16:174–182. versus long-term sampling to assess lepidopteran diversity in Pimm, S. L., J. Diamond, T. M. Reed, G. J. Russell, and J. a southern mixed mesophytic forest. Ann. Entomol. Soc. Verner. 1993. Times to extinction for small populations of Am. 92:435–441. large birds. Proc. Natl Acad. Sci. 90:10871–10875. Lewinsohn, T. M., A. V. L. Freitas, and P. I. Prado. 2005. Pinhey, E. C. G. 1972. Emperor moths of South and South- Conservation of terrestrial invertebrates and their habitats in Central Africa. C. Struik, Cape Town, South Africa. Brazil. Conserv. Biol. 19:640–645. Price, P. W., I. R. Diniz, H. C. Morais, and E. S. A. Marques. Magurran, A. E.. 2004. Measuring biological diversity. 1995. The abundance of insect herbivore species in the Blackwell Publishing, Oxford, UK. tropics: the high local richness of rare species. Biotropica Magurran, A. E., S. R. Baillie, S. T. Buckland, J. M. Dick, E. 27:468–478. M. Scott, R. I. Smith, et al. 2010. Long-term datasets in Prugh, L. R., K. E. Hodges, A. R. E. Sinclair, and J. S. biodiversity research and monitoring: assessing change in Brashares. 2008. Effect of habitat area and isolation on ecological communities through time. Trends Ecol. Evol., fragmented animal populations. Proc. Natl Acad. Sci. 25:574–582. 105:20770–20775. Michaelsen, F. C. J., and M. Marshall. 2012. Mapping recent R Development Core Team. 2013. R: A language and decadal climate variations in precipitation and temperature environment for statistical computing. R Foundation for across eastern Africa and the Sahel, chap. 14 of Remote Statistical Computing, Vienna, Austria.

1756 ª 2015 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. P. Akite et al. Moths in Protected Forests of Uganda

Ricketts, T. H. 2001. The matrix matters: effective isolation in ecological response: the contribution of long-term ecological fragmented landscapes. Am. Nat. 158:87–99. research. Bioscience 53:46–56. Sala, O. E., F. S. III Chapin, J. J. Armesto, E. Berlow, J. Usher, M. B., and S. W. J. Keiller. 1998. The macrolepidoptera Bloomfield, R. Dirzo, et al. 2000. Global biodiversity scenarios of farm woodlands: determinants of diversity and for the year 2100. Science 287:1770–1774. community structure. Biodivers. Conserv. 7:725–748. Saunders, D. A., R. J. Hobbs, and C. R. Margules. 1991. Vandermeer, J., and R. Carvajal. 2001. Metapopulation Biological consequences of ecosystem fragmentation: a dynamics and the quality of the matrix. Am. Nat. 158:211– review. Conserv. Biol. 5:18–32. 220. Schulze, C. H., and K. Fiedler. 2003. Vertical and temporal Were, L. M., 2010. Population structure of drum making tree diversity of a species rich moth taxon in Borneo. Pp. 69–85 in species: The population structure of drum making tree V. N. Y. Basset, S. Miller and R. L. Kitching, eds. Arthropods species: A case study of Mpanga Forest Reserve, VDM of tropical forests: spatio-temporal dynamics and resource use publishing, Uganda. pp 52. in the canopy. Cambridge University Press, UK. Williams, A. P., and C. Funk. 2011. A westward extension of Summerville, K. S., and T. O. Crist. 2005. Temporal patterns the warm pool leads to a westward extension of the of species accumulation in a survey of Lepidoptera in a Walker circulation, drying eastern Africa. Clim. Dyn. beech-maple forest. Biodivers. Conserv. 14:3393–3406. 37:2417–2435. Summerville, K. S., L. M. Ritter, and T. O. Crist. 2004. Forest Williams, N. M., E. E. Crone, T. A. H. Roulston, R. L. moth taxa as indicators of lepidopteran richness and habitat Minckley, L. Packer, and S. G. Potts. 2010. Ecological and disturbance: a preliminary assessment. Biol. Conserv. 116:9–18. life-history traits predict bee species responses to Thomas, J. A., M. G. Telfer, D. B. Roy, C. D. Preston, J. J. D. environmental disturbances. Biol. Conserv. 143:2280–2291. GreenwooD, J. Asher, et al. 2004. Comparative losses of Winterbottom, B., and G. Eilu, 2006. Uganda biodiversity and British butterflies, birds, and plants and the global tropical forest assessment. Final report. http:// extinction crisis. Science 303:1879–1881. www.vub.ac.be/klimostoolkit/sites/default/files/documents/ Turner, M. G., S. T. Collins, A. E. Lugo, J. J. Magnuson, T. S. uganda_biodiversity_assessment_usaid.pdf. Rupp, and F. J. Swanson. 2003. Disturbance dynamics and

ª 2015 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. 1757 Paper II

Akite, P., Akol, M.A., Kronstad, T., Vandvik, V & Telford, R.J. Vertical

distribution of fruit-feeding butterflies in three protected forests in

Central Uganda.

Manuscript under revision (Insect Conservation and Diversity Journal).

1 Vertical distribution of fruit-feeding butterflies in three protected forests in Central

2 Uganda

3

4 Running title: Vertical distribution of forest butterflies

5

6 Akite, Pa,b*., Akol, M.Aa., Kronstad, Tb., Vandvik, Vb and Telford, R.Jb.

7

8 aDepartment of Biological Sciences, School of Natural Sciences, Makerere University, P. O.

9 Box 7062 Kampala, Uganda.

10 bDepartment of Biology, University of Bergen, Thormøhlensgate 53A, N-5006 Bergen,

11 Norway

12 *Corresponding Author: Perpetra Akite: E-mail: [email protected] ; Phone:

13 +256772902633; Fax: +256 414 531061

14

15

16

17

18

19

20

21

22

23

24

25 1

26 Abstract.

27 1. Vertical stratification of plant and animal assemblages is considered a key

28 characteristic of intact forest ecosystems and is recognized as a basic niche dimension in

29 forest ecology. Several studies describe vertical distribution among fruit-feeding butterflies,

30 but the significance of this pattern is difficult to test with classical statistical approaches due

31 to high species richness and sparsity of butterfly assemblages.

32 2. Fruit-feeding butterflies from three forest reserves in central Uganda were sampled

33 using standard baited traps. The multinomial model was used to categorize species on the

34 basis of their vertical affinity. The vertical distribution of different species groups and

35 subfamilies was also examined using the same method. Results were compared with the

36 classical Chi-square test for homogeneity for our data and two published datasets.

37 3. The understorey was more speciose and had higher butterfly densities than the

38 canopy. However, the canopy had higher proportions of specialist species. The canopy was

39 dominated by forest edge species and species with migratory tendencies, while the

40 understorey was dominated by forest dependent species. Six subfamilies were represented and

41 their distributions differed between strata; the observed trend supports what is known about

42 fruit distribution and adaptation of the proboscis to different feeding techniques amongst

43 different subfamilies.

44 4. The observed pattern of distribution for different species, subfamilies and species

45 groups are supportive of how biotic and abiotic factors plus species evolutionary constraints

46 influence stratification. In addition, direct comparison with data from other studies greatly

47 improved our understanding of how different models could be useful in identifying priority

48 species for local conservation and monitoring.

49 Key-words: Tropical forest insects, niche differentiation, specialization, species 50 richness, ecological habitat preferences

2

51 Introduction

52 Niche differentiation has been put forward as one of the main explanations for species

53 coexistence (e.g. Rosenzweig, 1981; Gravel et al., 2011) but this paradigm has recently been

54 challenged by neutral models of coexistence (e.g. Rosindell et al., 2012). Empirically testing

55 niche differentiation is challenging as we rarely have direct empirical data on species’ niches

56 or niche dimensions due to low niche differentiation between species over evolutionary time

57 scale. Therefore species distributions along environmental gradients and in time and/or space

58 along with traits and/or behaviours are often used as proxies, based on the assumptions that

59 species whose occurrence overlaps in time and space presumably share important aspects of

60 their habitat niches (e.g. Lushai et al., 2003; Turlure et al., 2009). A species' niche includes all

61 of its interactions with the biotic and abiotic factors of its environment that ensure its

62 population persistence in that environment. Species differentiate their niches in many ways,

63 for example by consuming different foods, or using different parts of the environment. This

64 interaction may bring about niche shifts or changes in competitive hierarchies that allow

65 species to coexist (Bruno et al., 2003).

66 Vertical stratification of plant and animal assemblages is considered a major

67 characteristic of intact forest ecosystems and it is recognized as a basic niche dimension in

68 forests (e.g. Smith, 1973 for trees; Terborgh, 1980 for birds; Bourlière, 1989 for mammals;

69 Basset, 2001 for insects). Several studies have investigated vertical stratification of insects

70 within tropical forests and demonstrated that there are distinctive canopy and understorey

71 fauna (e.g. DeVries et al., 1997, 2012; Beck & Schulze, 2000; Hill et al., 2001 for butterflies,

72 Hammond et al., 1997; Chung, 2004; Stork & Grimbacher, 2006; Stork et al., 2008 for

73 beetles). This has been linked to differences in light levels (e.g. DeVries, 1988; Davis &

74 Sutton, 1998; DeVries et al., 1999) and other biotic and abiotic factors such as food

75 availability (e.g. Rogers & Kitching, 1998; Basset et al., 2003). 3

76 Quantifying species overlap in time and space can be challenging, especially in cases

77 where data are sparse (as they often are, especially in field data from high-diversity species

78 groups and regions). Previous studies have used classical statistical approaches such as

79 frequency distributions of presence records (e.g. DeVries, 1988; Brühl et al., 1998; Hill et al.,

80 2001; Fermon et al., 2003), the Chi-square test of the null hypothesis of homogeneity (e.g.

81 Molleman et al., 2006; DeVries et al., 1999, 2012) and ordinations (e.g. Stork & Grimbacher,

82 2006; DeVries et al., 2012) to visualize and describe vertical stratification of forest

83 arthropods. These analytical statistical approaches differ in their ability to detect

84 specialisation, and often require data manipulation before analysis (Chazdon et al., 2011).

85 Identifying appropriate statistical methods for sparse data is a key step towards examining the

86 underlying biological and ecological factors that lead to differential distribution of species

87 among habitats. The multinomial model, developed by Chazdon et al. (2011), for classifying

88 species into generalist or specialist groups, is a two-habitat species classification approach

89 that classifies species into habitat generalists and specialists based on estimated relative

90 abundance of species in two distinguishable habitats. Advantages of the multinomial model

91 include the ability to classify species into generalists and specialists without excluding rare

92 species a priori, the ability to set user-defined specialization thresholds relevant for the

93 particular research question at hand, and to distinguish habitat generalists from those species

94 that are simply too rare to assess. This method is robust to differences in sampling intensities

95 between two habitat types and to insufficient sampling within each habitat, two common

96 problems in ecological data which can bias estimates based on the classical tests (Chazdon et

97 al., 2011).

98 Fruit-feeding butterflies, an important insect group in tropical forests (e.g. Bonebrake

99 et al., 2010), gain most of their nutritional requirements from rotting fruits, plant sap and

100 decaying materials and have been used to explore the vertical dimension of distribution within 4

101 tropical forests (Schulz et al., 2001; Fermon et al., 2005). Comparing the vertical distribution

102 of different ecological guilds (e.g., nectar and fruit feeding guilds) and ecotypes (e.g.,

103 different habitat preferences) can yield knowledge of niche structure along the vertical

104 gradient, whereas the distribution of different subfamilies can indicate how strongly vertical

105 distribution patterns are affected by taxonomic relatedness, and presumably by phylogenetic

106 constraints (e.g. Schulz et al., 2001). However, the vertical distribution remains to be tested

107 with robust methods for a broad range of taxa and guilds to understand the role of vertical

108 distribution in niche structure of these forest insect assemblages (Schulz et al., 2001).

109 This study focused on the distribution of fruit-feeding butterflies between the canopy

110 and understorey in three protected forests in central Uganda. Davenport (1996) described the

111 ecotypes for the different butterfly species recorded in Ugandan forests based on their habitat

112 affinities but did not take the vertical patterns of species distributions into account. We use the

113 multinomial model to delimit habitat strata specialists and generalists, and based on the

114 literature reviewed above, we tested the hypotheses that (i) Species richness and abundance of

115 fruit-feeding butterflies are equally distributed between the canopy and understorey (ii)

116 Ugandan forest butterfly assemblages contain strata-specific species, ecotypes, and

117 subfamilies (iii) specific parts of the vertical stratum are dominated by similar species,

118 ecotypes, and sub-families across different forests and forest types, and (iv) the multinomial

119 model has advantages over the Chi-square test in assessing and comparing stratification across

120 butterfly assemblages.

121

122

123

124

125 5

126 Materials and Methods

127 Study sites

128 The study was conducted in three Ugandan forest reserves within the catchment of the Lake

129 Victoria basin (Figure 1); Mabira, located between 0°24´− 0°35´ N and 32°52´− 33°07´E

130 covering an area of 306 km2; Mpanga, located at 0°15´ N and 32°18´ E covering an area of

131 4.53 km2 and Zika located at 0°07´ N and 32°31´ E covering an area of 0.1 km2. These forests

132 are situated amidst densely populated agricultural landscapes, reflecting factors implicated in

133 the fragmentation and loss of forest habitats globally.

134 In Mabira, four sites in mature forest and three sites in secondary forest were sampled.

135 The mature forest is classified as sub-climax forest as it has been influenced by low-intensity

136 human impacts for a long time and is dominated by tree species such as Celtis sp, Albizia sp,

137 Antiaris sp, and Chrysophyllum sp. The secondary forest comprise forests regenerating after

138 felling that lack the large old trees that characterize the mature forest but with similar species

139 composition, though some areas are dominated by the invasive paper mulberry Broussonetia

140 papyrifera (MWLE, 2008). In Zika, two sites were sampled: including the heterogeneous

141 mature forest dominated by Lovoa brownie and Maesopsis eminii, and permanent swamp

142 forest dominated by Mitragyna stipulosa and Erythrina excelsa (Howard & Davenport, 1996).

143 In Mpanga three sample sites were established in mature forest dominated by Mitragyna-

144 Phoenix and Celtis-Aningeria associations (Howard & Davenport, 1996).

145

146 Sampling method

147 Butterfly sampling was done using standard baited traps similar to ones used by other studies

148 (e.g. DeVries et al., 1997; Molleman et al., 2006). In each of the sites in Mabira and Mpanga,

149 a 10 x 500m transect directed away from the forest edge (≥ 100m), was established. In the

150 Zika sites, transect was located in the middle of each forest type because of small forest size. 6

151 Transects were marked out using a GPS. Ten trap stations were then positioned at 50m

152 intervals along transects. Each trap station was fitted with one understorey trap and one

153 canopy trap with a total of twenty traps per transect. The position of the understorey traps

154 varied between 1 and 1.5m above ground while that of the canopy traps varied between 15

155 and 30m. Canopy traps were suspended using a thin nylon rope over branches to create a

156 pulley system that allowed the traps to be lowered and raised from the ground. They were

157 positioned to sample from within the crown of the trap tree. Traps were baited prior to the first

158 day of sampling with over-ripe bananas which had been mashed, mixed well, and fermented

159 overnight in a container. Bait was placed in a small plastic plate inside each trap, and fresh

160 bait was added the following day especially in traps where they had either been lost or eaten

161 by monkeys or other forest dwellers. During each trapping period, traps were emptied twice

162 daily between 0900 and 1600 hours for three successive days. Trap checking was not done

163 when it was raining or immediately after rainfall as the specimens would then be wet and their

164 wings easily damaged. In Mabira, sampling lasted from February 17th to June 4th 2009; in

165 Zika from November 14th 2009 to July 25th 2010; in Mpanga from January 16th to April 20th

166 2010. A total of 840, 240 and 360 trap-days were achieved in Mabira, Zika and Mpanga,

167 respectively.

168 Trapped butterflies were either identified in the field or voucher specimens collected

169 and placed in labelled glassine envelopes. These specimens were subsequently identified with

170 help of available museum collections at the Makerere University Zoology Museum and

171 identification field guides (e.g. Kielland, 1990; Larsen, 1991, 2005). For most of the familiar

172 and abundant species, individuals were checked and released back into the habitat, and details

173 of numbers and sex recorded. Confirmation for some species identifications were obtained

174 from Torben Larsen (pers.comm).

175 7

176 Species classification

177 Species were described in terms of their known ecological habitat preferences based on

178 existing knowledge of each species’ ecology (Davenport, 1996). These ecotypes include: F)

179 forest-dependent species restricted to closed canopy forest; f.) forest non-dependent species

180 recorded in closed-canopy forest, but also encountered in a variety of forest-edge, degraded

181 forest and woodlands; O) non-forest species characteristic of open habitats such as grassland,

182 open savannah and arid habitats; M) species with migratory tendencies , and W) widespread

183 generalists. We also partitioned the species into their respective subfamilies to evaluate the

184 vertical structure of the different taxonomic groups within the fruit-feeding guilds.

185

186 Data from the literature

187 We extracted species data from two publications, with their individual abundances in the

188 canopy and understorey. These were the DeVries (1988) study carried out at Finca La Selva,

189 Costa Rica and the Molleman et al. (2006) study that was carried out at the Makerere

190 University field station in Kibale National Park, Uganda. The DeVries (1988) study used

191 frequency distributions of presence records in canopy and understorey to describe

192 stratification while the Molleman et al. (2006) study used the Chi-square test for homogeneity

193 to describe vertical stratification of the fruit-feeding butterflies.

194

195 Data analysis

196 Species data for the canopy or understorey were pooled for each forest type to avoid

197 pseudoreplication . Two species (Abisara neavei and lemolea) belonging to family

198 Riodinidae and respectively were trapped but were excluded from analyses as

199 this family are not fruit-feeders but are often caught in baited traps by chance.

8

200 We used rank abundance plots for general description of our datasets and compared

201 species richness and abundance of butterfly assemblages between canopy and understorey in

202 the different forests and forest types. In order to visualize species stratification, we used the

203 multinomial test. Here, two user-defined specialization thresholds (K) are considered; a

204 conservative threshold (K = 2/3) in which the relative abundance of a species in one stratum is

205 >2/3 in the other stratum, and the liberal threshold (K = 1/2) where the relative abundance of a

206 species in one stratum is >1/2 in the other stratum. We used a super-majority specialization

207 threshold (K = 2/3 and P = 0.05) to evaluate habitat specialisation of individual species and (K

208 = 2/3 and P = 0.005) to assess the overall community pattern (Chazdon et al., 2011), and the

209 simple-majority specialization threshold (K = 1/2) to check for trade-off between sensitivity of

210 the model to specialists and classification of higher fractions of total species in the samples.

211 These analyses were done for each forest type within each forest and for the combined data

212 from all three forests. Similarly, using the super-majority specialization threshold (K =2/3)

213 and assemblage-wide classification using P = 0.005, we assessed the vertical distribution of

214 higher taxonomic levels using habitat preferences and subfamilies as model indicators to

215 delineate which species belonging to particular species groups and subfamily can be

216 categorically classified as canopy specialist, understorey specialist or generalist.

217 To assess how standard statistical methods compare to the multinomial model, the

218 Chi-square test; a widely used standard test, was selected. The differences in species

219 abundance frequencies between the canopy and understorey were tested using the two-tailed

220 Chi-square test for homogeneity (df = 1, p-critical = 0.05), based on the 1-sample proportions

221 test with continuity correction for the sample estimates. Species classified as generalist,

222 canopy specialist or understorey specialist included only those that had at least 10 individuals,

223 and hence with an expected frequency count of at least five in each stratum. Species which do

224 not meet this criterion were considered too rare to classify. These frequencies were tested at 9

225 species level for the different forests, forest types, combined forests data set and the selected

226 publications. All the above analyses were done using the statistical programme R (v. 2.13.1).

227

228 Results

229 In total, 10362 individual butterflies representing 115 species and six subfamilies of fruit-

230 feeding butterflies were recorded in the three forests (See Appendix 1). A large proportion of

231 the butterflies belonged to relatively uncommon species, a majority had less than 5 individuals

232 recorded (Figure 2). Species richness varied between forests, and was highest in Mpanga and

233 lowest in Zika (Figure 2). Almost 75% of the total abundance was found in the understorey,

234 and rarefied species richness was consistently higher in the understorey than the canopy

235 across all forests and habitat types (Figure 2).

236 A higher proportion of species were classified as canopy specialists than as

237 understorey specialists, and this pattern was consistent across all forests and forest types, and

238 in the combined data (K = 2/3, p = 0.05, Table 1, Figure 3). The proportion of species too rare

239 to classify was generally higher in smaller datasets, but even in the combined dataset 53.0 %

240 were too rare to classify, whereas 30.4 % were classified as canopy or understorey specialists,

241 and 16.5 % as generalists (Table 1). Some species were consistently categorized into one of

242 the three specialisation classes (generalist, canopy or understorey) across forests and forest

243 types and a number of species were also consistently too rare to classify (Appendix 1).

244 The classification was consistent across ecotypes: high proportions of the forest-

245 dependent species were classified as understorey specialists, forest edge species tended to be

246 canopy specialists (K = 2/3, p = 0.005, Table 2). Widespread species were mostly classified as

247 generalists (Table 2). Around 40 % of the forest-dependent, forest edge and widespread

248 species were too rare to classify, despite the relatively lax thresholds used in this test (Table

10

249 2). Few migratory and open habitat species were encountered in the data, with more migratory

250 individuals found in the canopy (Table 2).

251 Subfamilies differed more in distribution within forests than ecotypes. Species

252 belonging to the Nymphalinae and Charaxinae were typically classified as canopy specialists,

253 the Satyrinae as understorey specialists, and the Acraeinae as generalists (K = 2/3, p = 0.005,

254 Table 2). The Libytheinae and Apaturinae with only one species each in the entire

255 Afrotropical region; these species were classified as canopy specialists (Table 2). The genera

256 Bicyclus, Gnophodes, Henotesia and Euphaedra were typically classified as understorey

257 specialists; whereas Apaturopsis, Charaxes, Euptera, Neptis, Pseudacraea and Sallya were

258 canopy specialists (Appendix 1). The other genera recorded were either categorized as

259 generalists or were often too rare to classify (Appendix 1).

260 The Chi-square test for homogeneity generally classified more species as specialists

261 compared to the multinomial model (Table 1). For example, in the DeVries- data set, only

262 4.4% of total species were categorised as specialist using the multinomial model compared to

263 13.1% using the Chi-square test. The corresponding number for this study and the Molleman

264 data was 30.4% vs. 46.9% and 46.8% vs. 62.7%, respectively. These differences could to a

265 large extent be explained by differences in the proportions of species judged as too rare to

266 classify, which was much higher (often twice as high) in the multinomial model compared

267 with the Chi-square test. Varying the K (1/2 vs 3/2) and P (0.05, 0.005) criteria in the

268 multinomial test led to variation in the number of species classified as specialists vs.

269 generalists, and as too rare to classify, respectively (Appendix 2).

270

271 Discussion

272 Our results show that both the understorey and canopy are species rich, with overall species

273 richness distributed unequally between the strata across the different forests and forest types. 11

274 We found more individuals in the understorey compared to the canopy. There were

275 differences in patterns of abundance, richness, and specialisation of different ecotypes and

276 subfamilies across the vertical strata. Migrantory and forest edge species tended to be canopy

277 specialists, forest dependent and open habitat species tended to be understorey specialists.

278 Widespread species showed no preference for either the canopy or understorey and were

279 typically classified as generalists. We also found differences in vertical distribution for

280 species belonging to the different subfamilies; Satyrinae was dominated by understorey

281 specialists, Apaturinae, Charaxinae, Libytheinae and Nymphalinae by canopy specialists.

282 Members of the subfamily Acraeinae were largely classified as generalists.

283 Previous studies have shown marked variation in the way particular species select the

284 vertical distribution; some species are entirely associated with either the canopy or

285 understorey and some species are equally split between the strata (e.g. DeVries et al., 1997,

286 2012; Beck & Schulze, 2000; Hill et al., 2001; Molleman et al., 2006). Although we found

287 relatively small variation in total species richness between the canopy and understorey across

288 the different forests and forests types, total abundance was consistently higher in the

289 understorey indicating that the fruit-feeding butterfly assemblages in our study forests are

290 indeed vertically structured. Both the canopy and understorey samples contained only a subset

291 of the total assemblage, indicating heterogeneity in the vertical distribution. Such patterns

292 have also been reported in other insect taxa (e.g. Stork & Grimbacher, 2006 for beetles), and

293 show that the degree of specialization to resources and physiological tolerances to

294 microclimate are likely important in structuring arthropod assemblage differences between the

295 canopy and the ground (Stork & Grimbacher, 2006).

296 The observed stratification of the butterfly assemblages supports the notion that any

297 tropical insect study that does not take the vertical dimension into account would likely

298 underestimate total richness (DeVries et al., 1997). For most of the arthropods, such 12

299 differences between the canopy and understorey are likely to be determined by factors such as

300 resource availability, microclimate preferences and predator avoidance (e.g. Schultze et al.,

301 2001; Basset et al., 2003). The canopy and understorey shared many species although some

302 species dominated the butterfly assemblages of either the canopy (e.g. Charaxes spp) or the

303 understorey (e.g. Gnophodes spp). A higher proportion of species were classified as canopy

304 specialists. This parallels findings from other forests in Uganda (e.g. Molleman et al., 2006),

305 other Africa countries (e.g. Kwaku et al., 2012), and/or other continents (e.g. DeVries et al.,

306 1997, 1999, 2012).

307 Unlike most previous studies on vertical distribution of fruit-feeding butterflies, we

308 also assessed the response of functional groups – the butterfly ecotypes as described by

309 Davenport (1996) – to vertical structure. We found that migrantory and forest edge species

310 tended to be classified as canopy specialists whereas forest-dependent species were mostly in

311 the understorey. Several explanations are possible; for example, higher wind velocity may

312 result in the canopy capturing more ‘tourists’ (e.g. Moran & Southwood, 1982; Szarzynski &

313 Anhuf, 2001), or the more heterogeneous light environment in the canopy may be favourable

314 for these species (e.g. DeVries, 1988; DeVries et al., 1999). The presence of migrant species

315 and forest edge species in the canopy also means these species are able to find suitable

316 habitats such as recently formed clearings that would otherwise be inaccessible to them if they

317 were restricted to travel through intact forests. This was apparent when comparing the more

318 open and recently disturbed secondary forest in Mabira and Mpanga to the mature forest in

319 Mabira and Zika. In the more disturbed forests, the proportions of migrant species were higher

320 in the canopy compared to the understorey.

321 Although the differences between the canopy and understorey are subtle, they can

322 easily be detected by using fine-scale taxonomic data such as subfamilies. The contrasting

323 vertical stratification results at the subfamily level may also suggests that fine scale resource 13

324 portioning at higher taxonomic level exist. This may contribute to the patterns of vertical

325 distribution of the different ecotypes observed within our study forests.

326 The striking differences in vertical distribution between subfamilies are consistent

327 with niche conservation within the fruit-feeding butterfly guilds (e.g. Chase & Leibold, 2003).

328 The understorey-specialised Satyrinae are weak flyers and shun bright light, often keeping to

329 shade while the canopy-specialised Charaxinae are generally strong flyers, often seen perched

330 on high tree trunks (e.g. Hill et al., 2001; Hamer et al., 2003; personal field observations).

331 These patterns indicate the role of conserved life-history traits (for example dispersal abilities,

332 egg-laying strategy, feeding strategy) in determining species distribution in space and across

333 the vertical strata (e.g. Hamer et al., 2003). This would suggest that for the fruit-feeding

334 butterflies, vertical location of members of different subfamilies is dependent on their specific

335 needs that may include abundance and quality of host plant (e.g. Schulze et al., 2001), biotic

336 factors such as presence of competitors and abiotic factors such as light levels (e.g. Davis &

337 Sutton, 1998).

338 Looking at the results from multinomial model versus the classical chi-square test of

339 homogeneity, it was evident that the proportions of species classified as too rare by the

340 multinomial model vary greatly between datasets, suggesting that a statistically based method

341 for classifying species which takes data structure and distribution into account is preferable to

342 those methods that use the ‘rule of thumb’, for example the predetermined assumptions made

343 in defining rare species in the Chi-square test.

344 Species rarity presents challenges for any model of classification (e.g. Chazdon et al.,

345 2011); often making it impossible to determine habitat affinities for such species with

346 statistical confidence based on relative abundance data. From our study, more than 50% of

347 species recorded in every forest and forest type were considered too rare to classify. There

348 were also striking differences in the proportions of species classified as understorey or canopy 14

349 specialists between the two methods even when exactly the same number of traps and trap

350 nights are used. This can partly be explained by different assumptions in the two models; as

351 the multinomial model explicitly controls for unequal sampling (and by implication, unequal

352 total butterfly abundances and/or habitat availability in the two strata) (Chazdon et al., 2011).

353 The multinomial model thus classifies each species as specialist or generalist relative to the

354 total population under study (which may or may not be equally distributed and/or sampled),

355 whereas the Chi-square method tests for unequal distribution between habitats on a species-

356 by-species basis, without considering variation in total densities, habitat availability, or

357 sampling effort.

358 In addition, the multinomial model also allows generalist species to be classified and

359 distinguished from species that are too rare to classify (Chazdon et al., 2011), whereas the

360 Chi-square classifies species into ‘too rare’ based on a rule-of-thumb criterion. Since niche

361 relationships between species are important, it is critically important to use a method that is

362 robust to unequal sampling. In allowing four possible classes for species categorization

363 (specialist of either habitat, generalist, or too rare), the multinomial model outperforms most

364 classical statistical models, vis-à-vis the Chi-square test by eliminating the reliance on equal

365 sampling effort and on often subjective pre-determination of species that are too rare to

366 classify.

367 In conclusion the canopy and understorey both hosted species that were characteristic

368 specialised inhabitants of each stratum, implying that the vertical dimension is a major

369 structural component in tropical forest fruit-feeding butterfly assemblages. The canopy and

370 understorey shared many species, and the specialist/generalist concept is effective and useful

371 for distinguishing true generalists from taxa too rare to classify. Species habitat preferences

372 are important in understanding factors that maybe influencing the overall observed vertical

373 distribution patterns. Nonetheless, we need to continually monitor and manage the butterfly 15

374 assemblages especially specialist and rare species identified through such analytical

375 approaches in order to ensure their survival within the habitats.

376

377 Acknowledgements

378 We thank Amy Eycott, Hugh Rowell and three anonymous reviewers for comments on earlier

379 versions of this manuscript. We thank the Norwegian Research Council FRIMUF programme

380 for funding the MATRIX project (# 184912), and all the project members for their

381 encouragement.

382

383 References

384 Basset, Y., Hammond, P.M., Barrios, H., Holloway, J.D. & Miller, S.E. (2003) Vertical

385 stratification of arthropod assemblages. Arthropods of tropical forests – spatio-

386 temporal dynamics & resource use in the canopy (ed. by Basset, Y., Novotný, V.,

387 Miller, S.E. & Kitching, R.L.), pp. 17–27. Cambridge University Press, Cambridge.

388 Basset, Y. (2001) Invertebrates in the canopy of tropical rain forests: how much do we really

389 know? Plant Ecology, 153, 87–107.

390 Beck, J. & Schulze, C.H. (2000) Diversity of fruit-feeding butterflies (Nymphalidae) along a

391 gradient of tropical rain forest succession in Borneo with some remarks on the

392 problem of ‘pseudoreplicates’. Transactions of the Lepidopterological Society of

393 Japan, 51, 89–98.

394 Bonebrake, T.C., Ponisio, L.C., Boggs, C.L. & Ehrlich, P.R. (2010) More than just indicators:

395 A review of tropical butterfly ecology and conservation. Biological Conservation, 143,

396 1831–1841.

16

397 Bourlière, F. (1989) Mammalian species richness in tropical rainforests. Vertebrates in

398 Complex Tropical Ecosystems (ed. by Harnelin-Vivien, M.J. & Bourlière, F.).

399 Springer-Verlag, New York, Berlin, Heidelberg, London, Paris, Tokyo.

400 Brühl, C.A., Gunsalam, G. & Linsenmair, K.E. (1998) Stratification of ants (Hymenoptera,

401 Formicidae) in a primary rain forest in Sabah, Borneo. Journal of Tropical Ecology,

402 14, 285–297.

403 Bruno, J.F., Stachowicz, J.J. & Bertness, M.D. (2003) Inclusion of facilitation into ecological

404 theory. Trends in Ecology and Evolution, 18, 119–125.

405 Chase, J.M. & Leibold, M.A. (2003) Ecological Niches: Linking Classical and Contemporary

406 Approaches. Chicago University Press, Chicago.

407 Chazdon, R.L., Chao, A., Colwell, R.K., Lin, S.Y., Norden, N., Letcher, S.G., Clark, D.B.,

408 Finegan, B. & Arroyo, J.P. (2011) A novel statistical method for classifying habitat

409 generalists and specialists. Ecology, 92, 1332–1343.

410 Chung, A.Y.C. (2004) Vertical stratification of beetles (Coleoptera) using flight interception

411 traps in lowland rainforest of Sabah, Malaysia. Sepilok Bulletin, 1, 29–41.

412 Davenport, T.R.B. (1996) The Butterflies of Uganda - An Annotated Checklist. Uganda Forest

413 Department, Kampala, Uganda.

414 Davis, A.J. & Sutton, S.L. (1998) The effects of rainforest canopy loss on arboreal dung

415 beetles in Borneo: implications for the measurement of biodiversity in derived tropical

416 ecosystems. Diversity and Distributions, 4, 167–173.

417 DeVries, P.J. (1988) Stratification of fruit-feeding Nymphalid butterflies in Costa Rican

418 rainforest. Journal of Research on the Lepidoptera, 26, 98–108.

419 DeVries, P.J., Alexander, L.G., Chacon, I.A. & Fordyce, J.A. (2012) Similarity and difference

420 among rainforest fruit-feeding butterfly communities in Central and South America.

421 Journal of Animal Ecology, 81, 472–482. 17

422 DeVries, P.J., Murray, D. & Lande, R. (1997) Species diversity in vertical, horizontal, and

423 temporal dimensions of a fruit-feeding butterfly community in an Ecuadorian

424 rainforest. Biological Journal of the Linnaean Society, 62, 343–364.

425 DeVries, P.J., Walla, T.R. & Greeney, H.F. (1999) Species diversity in spatial and temporal

426 dimensions of fruit-feeding butterflies from two Ecuadorian rainforests. Biological

427 Journal of the Linnaean Society, 68, 333–353.

428 Fermon, H., Waltert, M. & Mühlenberg, M. (2003) Movement and vertical stratification of

429 fruit-feeding butterflies in a managed West African rainforest. Journal of Insect

430 Conservation, 7, 7–19.

431 Fermon, H., Waltert, M., Vane-Wright, R.I. & Mühlenberg, M. (2005) Forest use and vertical

432 stratification in fruit-feeding butterflies of Sulawesi, Indonesia: impacts for

433 conservation. Biodiversity and Conservation, 14, 333–350.

434 Gravel, D., Guichard, F. & Hochberg, M.E. (2011) Species coexistence in a variable world.

435 Ecology Letters, 14, 828–839.

436 Hamer, K.C., Hill, J. K., Benedick, S., Mustaffa, N., Sherratt, T. N., Maryati, M. & Chey, V.

437 K. (2003) Ecology of butterflies in natural and selectively-logged forests of northern

438 Borneo: the importance of habitat heterogeneity. Journal of Applied Ecology, 40,150–

439 162.

440 Hammond, P.M., Stork, N.E. & Brendell, M.J.D. (1997) Tree-crown beetles in context: a

441 comparison of canopy and other ecotone assemblages in a lowland tropical forest in

442 Sulawesi. Canopy Arthropods (ed. by Stork, N.E., Adis, J. & Didham, R.K.), pp. 184–

443 223. Chapman & Hall, London.

444 Hill, J.K., Hamer, K.C., Tangah, J. & Dawood, M. (2001) Ecology of tropical butterflies in

445 rainforest gaps. Oecologia, 128, 294–302.

18

446 Howard, P.C. & Davenport, T.R.B. (1996) Forest biodiversity reports. Uganda Forest

447 Department, Kampala, Uganda.

448 Kielland, J. (1990) Butterflies of . Hill House, Melbourne and London.

449 Kwaku, A-P., Oduro, W., Oppong, S.K., Larsen, T., Ofori-Boateng, C. & Molleman, F.

450 (2012) Spatial and temporal variation in butterfly biodiversity in a West African forest:

451 lessons for establishing efficient rapid monitoring programmes. African Journal of

452 Ecology, 50, 326–334.

453 Larsen, T. L. (1991) Butterflies of Kenya and Their Natural History, Oxford University Press.

454 Larsen, T.B. (2005) Butterflies of West Africa 2 vols., 596 pp. Apollo Books.

455 Lushai, G., Smith, D.A.S., Gordon, I.J, Goulson, D., Allen, J.A. & Maclean, N. (2003)

456 Incomplete sexual isolation in sympatry between subspecies of the butterfly Danaus

457 chrysippus (L.) and the creation of a hybrid zone. Heredity, 90, 236–246.

458 Ministry of Water Lands & Environment. (2008) Mabira Forest management plan. Draft

459 report, Kampala, Uganda.

460 Molleman, F., Kop, A., Brakefield, P.M., DeVries, P.J. & Zwaan, B.J. (2006) Vertical and

461 temporal patterns of biodiversity of fruit-feeding butterflies in a tropical forest in

462 Uganda. Biodiversity and Conservation, 15, 107–121.

463 Moran, V. C. & Southwood, T. R. E. (1982) The guild composition of arthropod communities

464 in trees. Journal of Animal Ecology, 51, 289–306.

465 Rogers, D.J. & Kitching, R.L. (1998) Vertical stratification of rainforest collembolan

466 (Collembola: Insecta) assemblages: description of ecological patterns and hypotheses

467 concerning their generation. Ecography, 21, 392–400.

468 Rosenzweig, M.L. (1981) A theory of habitat selection. Ecology, 62, 327–335.

469 Rosindell, J., Hubbell, S.P., He, F., Harmon, L.J. &. Etienne, R.S. (2012) The case for

470 ecological neutral theory. Trends in Ecology and Evolution, 27, 203–208. 19

471 Schulze, C.H., Linsenmaier, K.E. & Fiedler, K. (2001) Understorey versus canopy: patterns of

472 vertical stratification and diversity among Lepidoptera in a Bornean rain forest. Plant

473 Ecology, 153,133–152.

474 Smith, A.P. (1973) Stratification of Temperature and Tropical Forests. The American

475 Naturalist, 107, 671–683.

476 Stork, N. E. & Grimbacher, P.S. (2006) Beetles assemblages from an Australian tropical

477 rainforest show that the canopy and the ground strata contribute equally to

478 biodiversity. Proceedings of the Royal Society B, 273, 1969–75.

479 Stork, N.E., Grimbacher, P.S., Storey, R., Oberprieler, R.G., Reid, C. & Slipinski, S.A. (2008)

480 What determines whether a species of insect is described? Evidence from a study of

481 tropical forest beetles. Insect Conservation and Diversity, 1,114–119.

482 Szarzynski, J. & Anhuf, D. (2001) Micrometeorological conditions and canopy energy

483 exchanges of a neotropical rain forest (Surumoni-crane project, Venezuela). Plant

484 Ecology, 153, 231–239.

485 Terborgh, J. (1980) Vertical stratification of a neotropical forest bird community. Proceeding

486 of the 17th International Ornithological Congress, 1005–1012.

487 Turlure, C., Van-Dyck, H., Schtickzelle, N. & Baguette, M. (2009) Resource-based habitat

488 definition, niche overlap and conservation of two sympatric glacial relict butterflies.

489 Oikos, 118, 950–960.

490

491

492

20

493 Table 1. Summary of the species classification using two statistical approaches; Multinomial model with specialization thresholds K= 2/3 and P

494 = 0.05 and the Chi-squared test for homogeneity. The Chi-squared analyses from DeVries (1988) and Molleman et al. (2006) were taken from the

495 original papers; all others were calculated based on raw data.

A) Multinomial model B) Chi-square test (Null model) Generalist Canopy Understorey Too rare to Generalist Canopy Understorey Too rare to species (%) specialist (%) specialist (%) classify (%) species (%) specialist (%) specialist (%) classify (%) This study Dry forest (Zika) 9.5 14.3 9.5 66.7 14.6 9.8 26.8 48.8 Swamp forest (Zika) 18.5 11.1 3.7 66.7 23.1 3.8 26.9 46.2 Mature forest (Mabira) 6.0 16.4 9.0 68.7 11.9 7.5 25.4 55.2 Secondary forest (Mabira) 6.1 12.1 12.1 69.7 10.8 10.8 23.1 55.4 Mature forest (Mpanga) 11.7 14.3 11.7 62.3 22.1 13.0 27.3 37.7 All forests combined 16.5 17.4 13.0 53.0 24.3 13.9 33.0 28.7

Other studies DeVries (1988) 2.2 2.2 2.2 93.4 10.9 10.9 2.2 76.1 Molleman et al.(2006) 18.1 17.0 30.9 34.0 18.1 8.5 54.3 19.1 496

497 498 499 500 501

21

502 Table 2. Summary of the species classification for the combined forests’ data using simple-

503 majority specialization threshold (K= 2/3. P = 0.005) to classify higher taxonomic levels

504 (Ecotypes and subfamilies). GE = Generalist. CS = Canopy specialist. US = Understorey

505 specialist. TR = Too rare to classify.

ECOTYPE Species numbers Species numbers in Multinomial model (K =2/3, P = 0.005) in canopy understorey GE (%) CS (%) US (%) TR (%) Forest dependent 59 66 10.8 21.6 27.0 40.5 Forest edge 18 22 8.3 25.0 20.8 45.8 Migrant 4 5 0.0 40.0 20.0 40.0 Open habitat 2 2 0.0 0.0 33.3 66.7 Widespread 9 7 11.1 22.2 22.2 44.4

SUBFAMILY Acraeinae 2 6 33.3 0.0 0.0 66.7 Apaturinae 1 1 0.0 100.0 0.0 0.0 Nymphalinae 34 42 10.6 23.4 19.1 46.8 Charaxinae 28 21 3.4 20.7 10.3 65.5 Libytheinae 1 1 0.0 100.0 0.0 0.0 Satyrinae 26 31 25.8 9.7 48.4 16.1 506

507

508

509

510

511

512

513

514

515

516

517

22

518

519 520 521 Figure 1. The location of forests and sampled sites for fruit-feeding butterflies in protected

522 forests. Inset: location of study region within Uganda.

523

23

Mabira mature forest Mabira secondary forest

Understorey 50

Canopy

200

40

50

30

20

20

5

10

2

1 0 0 10 20 30 40 50 0 10 20 30 40 50

Zika dry forest Zika swamp forest

200

200

50

50

10

10

Abundance

5

5

2

2

1 1 0 10 20 30 40 50 0 10 20 30 40 50

Mpanga mature forest Combined forests

500

500

50

50

5

5

1 1 0 20 40 60 80 100 0 20 40 60 80 100

Number of species 524

525 Figure 2. Rank abundance distribution of species of fruit-feeding butterflies sampled in bait

526 traps in canopy and understorey within the different forests and forest types. Abundances are

527 given as total number of individuals captured in the traps.

528

529

530

24

531 532 Figure 3. Classification of butterfly species within the different forests and forest types using 533 the super-majority specialization threshold (K=2/3. P= 0.05). Red = Canopy specialist. Green 534 = Understorey specialist. Black = Generalist. Blue = Too rare to classify. 535 536

25

537 Appendix 1. Species of fruit-feeding butterflies encountered in the different forests, forest types, and strata. C= Canopy. U = Understorey.

538 Entries are abundances given as total number of individuals captured. Fonts distinguish the classification category for the vertical distribution

539 each species in each forest according to the multinomial model using super-majority specialization threshold (K=2/3. P= 0.05). Bold = Specialist.

540 Italic = Too rare to classify. Plain text = generalist or individuals of a specialist found in the other stratum.

Zika dry Zika swamp Mabira Mabira Mpanga All forests forest forest mature secondary forest forest forest Species Ecotype C U C U C U C U C U C U Acraea alcinoe F 1 1 1 1 Actinote aurivilli F 0 1 0 1 Actinote jodutta F 0 1 0 1 Actinote lycoa F 1 0 1 0 Actinote penelope F 0 3 0 3 Actinote semivitrea F 1 1 1 1 delius F 1 1 1 1 Apaturopsis cleochares F 8 1 7 0 27 0 42 1 Ariadne albifascia F 0 1 0 1 Ariadne enotrea F 2 0 1 0 5 0 4 1 12 1 galene F 0 3 0 1 0 4 Bebearia carshena F 0 4 0 4 Bebearia cocalia f. 1 4 0 16 0 3 0 3 0 13 1 39 Bicyclus angulosus O 1 14 1 14 Bicyclus auricrudus F 1 3 0 1 0 3 0 1 13 60 14 68 Bicyclus buea F 0 1 0 2 9 31 9 34 Bicyclus campinus f. 0 2 0 2 3 5 3 9 Bicyclus funebris F 3 37 1 11 0 3 0 3 1 57 5 111 26

Bicyclus golo F 0 1 0 1 8 146 8 148 Bicyclus graueri F 0 7 1 7 0 29 1 14 2 57 Bicyclus iccius F 1 149 1 149 Bicyclus istaris f. 0 11 0 11 Bicyclus jefferyi f. 0 1 0 1 Bicyclus mandanes F 1 1 1 10 0 2 4 27 6 40 Bicyclus mesogena F 0 1 1 7 0 1 14 74 15 83 Bicyclus mollitia F 50 106 103 497 0 19 889 1904 992 2420 Bicyclus safitza W 20 250 0 1 70 357 Bicyclus sambulos F 0 1 2 25 0 13 19 135 21 174 Bicyclus sanaos F 0 11 0 8 0 1 0 20 Bicyclus sandace F 0 2 2 14 1 11 3 17 Bicyclus saussurei F 51 465 7 350 58 815 Bicyclus sebetus F 0 7 0 7 Bicyclus smithi F 3 69 1 40 19 221 23 330 Bicyclus sophrosyne f. 0 6 6 15 0 1 0 4 0 30 6 56 Bicyclus uniformis F 8 614 0 18 8 632 Bicyclus vulgaris W 129 283 28 176 0 5 0 17 157 481 Bicyclus xenoides F 1 1 8 25 9 26 Bicyclus zinebi F 1 56 1 56 Catuna crithea F 0 2 0 1 0 3 Charaxes ameliae f. 1 0 1 0 Charaxes anticlea f. 4 1 4 1 Charaxes bipunctatus F 22 5 12 1 17 2 51 8 f. 1 0 1 0 0 1 3 1 Charaxes candiope W 1 1 0 2 1 3 Charaxes catachrous F 1 1 1 1 Charaxes cedreatis F 1 0 1 0 Charaxes cynthia F 0 1 56 27 40 17 0 1 96 46 Charaxes etesipe f. 1 0 1 0

27

Charaxes etheocles F 4 1 4 1 Charaxes eudoxus f. 2 0 1 1 3 1 Charaxes fulvescens F 5 21 5 20 21 115 31 156 Charaxes lactetinctus O 1 0 1 0 2 0 Charaxes lucretius F 2 0 2 0 Charaxes numenes f. 4 2 6 3 7 1 3 1 31 12 51 19 Charaxes paphianus F 3 0 3 0 Charaxes pleione f. 1 0 1 0 3 1 5 1 f. 3 1 3 1 Charaxes porthos F 1 0 0 1 1 0 Charaxes protoclea f. 1 0 0 1 2 5 3 6 Charaxes pythodoris f. 0 1 0 1 Charaxes tiridates F 22 7 21 3 49 18 92 18 Charaxes varanes W 0 3 1 2 1 5 Charaxes virilis F 7 0 6 1 5 0 18 1 Charaxes zingha F 7 1 1 2 2 0 48 22 58 25 Cymothoe caenis F 1 1 1 1 Cymothoe herminia F 0 1 78 61 6 2 36 15 120 79 Cynandra opis F 0 1 0 1 Elymnias bammakoo F 4 5 1 0 1 2 6 7 Euphaedra alacris F 1 0 1 0 Euphaedra eleus F 0 6 0 6 Euphaedra harpalyce F 1 1 1 1 Euphaedra medon F 0 5 0 6 0 1 0 5 0 17 Euphaedra preussi F 2 86 0 16 0 2 2 104 Euphaedra rex F 0 10 0 6 0 16 Euphaedra ruspina F 0 1 0 1 Euphaedra uganda F 0 1 6 0 6 1 Euptera elabontas F 23 5 1 6 1 6 1 0 26 17 Euryphura chalcis F 1 0 1 0

28

Eurytela dryope W 2 0 3 0 1 0 0 1 6 1 Eurytela hiarbas f. 9 2 5 8 5 0 1 0 14 5 30 15 Euxanthe crossleyi F 3 1 2 1 4 2 Euxanthe eurinome F 1 0 1 0 Euxanthe trajanus F 5 41 5 41 Gnophodes betsimena F 0 14 2 15 0 9 1 34 3 72 Gnophodes chelys F 0 13 0 2 0 36 1 11 6 236 7 298 Harma theobene F 10 55 5 27 3 3 18 85 Henotesia peitho W 4 195 0 28 4 223 Henotesia perspicua O 0 7 0 7 Hypolimnas anthedon F 0 1 1 0 1 1 Hypolimnas dinarcha F 3 2 1 0 4 2 Hypolimnas monteironis F 0 3 0 2 1 0 2 3 3 8 Hypolimnas salmacis F 2 5 0 1 1 1 3 7 Junonia stygia f. 0 6 0 1 0 7 Junonia westermanni F 1 0 1 1 2 1 Lachnoptera anticlia f. 0 4 0 4 Libythea labdaca M 0 1 2 0 1 0 3 1 Melanitis leda W 1 11 0 2 0 6 0 2 1 4 2 25 Mesoxantha ethosea F 1 0 0 1 1 1 ophione f. 1 0 3 1 4 1 Neptis conspicua F 1 0 1 0 Neptis melicerta F 0 1 1 2 12 1 2 1 15 5 Neptis metella f. 0 3 17 2 1 0 18 5 Neptis nemetes f. 0 2 3 4 2 1 5 7 Neptis nicobule W 1 0 1 0 Neptis nicomedes f. 4 1 25 3 29 4 Neptis saclava W 1 0 1 0 2 0 ussheri F 2 1 1 0 3 1 Phalanta eurytis M 0 1 0 1 29

Pseudacraea dolomena F 12 38 12 38 Pseudacraea eurytus F 10 10 0 2 5 0 2 2 17 14 Pseudacraea lucretia f. 18 2 1 2 1 0 2 1 3 0 25 5 Protogoniomorpha f. 0 2 0 2 parhassus Sevenia boisduvali M 186 44 5 0 1 0 0 1 15 5 207 50 Sevenia garega M 3 0 5 25 1 0 20 2 29 27 Sevenia occidentalium M 41 9 17 6 3 1 41 4 102 20 541

30

542 Appendix 2. Summary of the species classification using the Multinomial model with different settings: specialization thresholds K= 2/3 and K

543 =1/2 using P = 0.05 and P = 0.005. GE = Generalist. CS = Canopy specialist. US = Understorey specialist. TR = Too rare to classify

Multinomial model K = 2/3. P = 0.05 K = 2/3. P = 0.005 K = 1/2. P = 0.05 K = 1/2. P = 0.005 GE CS US TR GE CS US TR GE CS US TR GE CS US TR Dry forest (Zika) 9.5 14.3 9.5 67.0 9.5 9.5 4.8 76.2 14.3 14.3 14.3 57.1 9.5 11.9 9.5 69.0 Swamp forest (Zika) 18.5 11.1 3.7 67.0 14.8 3.7 3.7 77.8 11.1 14.8 7.4 66.7 14.8 3.7 3.7 77.8 Mature forest (Mabira) 6.0 16.4 9.0 69.0 6.0 7.5 4.5 82.1 6.0 16.4 10.4 67.2 6.0 9.0 9.0 76.1 Secondary forest 6.1 12.1 12.1 70.0 3.0 9.1 6.1 81.8 3.0 13.6 19.7 63.6 1.5 9.1 15.2 74.2 (Mabira) Mature forest 11.7 14.3 11.7 62.0 13.0 11.7 10.4 64.9 2.6 19.5 23.4 54.5 5.2 14.3 16.9 63.9 (Mpanga) All forests combined 16.5 17.4 13.0 53.0 16.5 13.9 11.3 58.3 10.4 22.6 25.2 41.7 7.8 17.4 19.1 55.7

DeVries (1988) 2.2 2.2 2.2 94.0 0.0 2.2 0.0 97.8 2.2 8.7 8.7 80.4 0.0 4.3 2.2 93.5 Molleman et al.(2006) 19.1 17.0 29.8 34.0 20.2 13.8 26.6 39.4 8.5 21.3 38.3 31.9 12.8 18.1 36.2 33.0 544

545 546

547

31

Paper III

Akite, P., Akol, M.A., Eycott, A.E., Vandvik, V. & Telford, R.J. The use of

matrix habitats by forest insect species.

Manuscript

The use of matrix habitats by forest insect species

Akite, P., Akol, M.A., Eycott, A.E., Vandvik, V. and Telford, R.J.

Abstract

Tropical forests are among the most species-rich terrestrial habitats. Little is known about their insect communities and how these respond to anthropogenic habitat alteration. We investigated speciose insects groups (butterflies, silkmoths, hawkmoths and grasshoppers) in three forests in central Uganda and their surrounding matrix habitats. Sites span a gradient of land-use from mature forest to mixed gardens.

In total, 25156 individuals of 327 species of butterflies, 1131 individuals of 41 species of silkmoths , 1564 individuals of 44 species of hawkmoths and 2173 individuals of 49 species of grasshoppers were recorded. Species diversity was generally highest in mature forests compared to other land-use type across the three forests. Both rarefied and Chao1 estimated species richness were generally higher in mature forest sites whereas assemblages in heavily disturbed mixed gardens were significantly less diverse. Compared to mature forest, all other land-use types had a negative effect on raw values for species richness and abundance of the different species groups, though significance of these relationships were inconsistent. The exceptiion was the cardamom plantation which had a positive effect on trapped butterfly richness (Est = 0.027, P> 0.05) and abundance (Est = 0.607, P = 0.001), and secondary forest which had a positive effect on grasshoppers’ abundance (Est = 0.023, P>

0.05). The forest dependent species showed a marked decline from mature forest to mixed gardens across the five species groups.

The occurrence of forest dependent species in matrix habitats is significantly dependent on species population size within the forest habitats for netted butterflies (rho =

1

0.28, p < 0.001) and trapped butterflies (rho = 0.33, p = 0.003). There was no significant difference between population size within the forest habitats and that in the matrix for silkmoths (rho = 0.23, p > 0.05), hawkmoths (rho = 0.28, p = >0.05), and grasshoppers (rho =

0.12, p >0.05) respectively. NMDS ordination showed that distinct segregation of species assemblages consistently emerged between the different land-use types. Segregation between assemblages of the different forests was also clear.

Our results suggest that the use of different species groups provide the opportunity to make novel insights into how we interpret and understand the conservation values of human modified habitats. Although we strongly urge measures that prioritize the conservation of protected forests in central Uganda and within the country as a whole, we acknowledge the notion that surrounding matrix habitats that maintain landscape heterogeneity, especially those that have elements similar to those within mature forests, should be maintained.

Keywords: Tropical forest, central Uganda, habitat disturbance, land-use gradient, insects

2

Introduction

A key strategy for preventing biodiversity loss is through creation of protected areas (Pimm et al. 2001, Gaines et al. 2010, Joppa & Pfaff 2010). Nonetheless, even those areas that are thought to have a relatively limited human footprint have experienced substantial biodiversity change (Mora & Sale 2011). The majority of tropical landscapes are in a state of extensive forest fragmentation resulting from severe deforestation (Laurance & Bierregaard 1997), posing a major threat to biodiversity (e.g. Rands et al. 2010). With only small patches of forest remaining, protected areas have become increasingly insufficient to maintain biodiversity across the wider landscape (Green et al. 2005), and the limit to expansion of protected area systems underlines the importance of seeking new strategies to conserve biodiversity.

There has been growing recognition of the importance of the areas between habitat patches (matrix) in mitigating species responses to, and behavior in fragmented landscapes

(e.g. Vandermeer & Carvajal 2001, Franklin & Lindenmayer 2009, Shreeve & Dennis 2011,

Sweaney et al. 2014). Although most studies focus on conserving the undisturbed forests and forest patches (e.g. Posa & Sodhi 2006, Barlow et al. 2007,Vu 2009, Driscoll et al. 2013), several studies stress the significance of human-modified habitats (e.g. Bhagwat et al. 2008,

Hamer et al. 2003, Chazdon et al. 2009). This is because the quality and extent of the matrix in terrestrial fragmented landscapes may influence the persistence and behaviour of patch- associated fauna (e.g. Sweaney et al. 2014, Hanski 1999, McIntyre & Hobbs 1999).

Roles of altered habitats

The matrix and habitat patches are often defined in terms of species resources requirements

(e.g. Dennis et al. 2013), and different species perceive matrix habitats differently (e.g.

Driscoll et al. 2013). In terrestrial landscapes, matrix habitats consist of a complex of different

3

land cover types (e.g. Ricketts 2001, Driscoll et al. 2013). This complexity has influence on some aspects of resident species and communities since fragmented forest landscapes maintain some degree of connectivity via modified habitats surrounding these fragments

(Gascon et al. 1999).

In forested landscapes, the capacity of matrix habitats to support forest species varies considerably and this is largely determined by the history and intensity of land use (Lawton et al. 1998). For example forest specialist species that occur in very low densities are not often able to establish viable populations within the altered habitats due to their life history requirements such as poor dispersal abilities, host plant specificity, amongst others. As such, sharp ecotonal boundaries between the patch and matrix would cause these species to cluster inside the remnant forests (e.g. Schtickzelle & Baguette 2003). On the other hand, the vagrant specialists are able to venture into and successfully exploit some aspects of the matrix while maintaining their forest affinities (e.g. Ewers & Didham 2006), but their movement between disjunct patches may decline due to altered behaviour or increased mortality (e.g. Didham et al. 2012, Schooley & Wiens 2004). The least specialized species meanwhile can survive and establish viable populations even in altered habitats that maintain some key forest structures

(e.g. Kuefler et al. 2010).

Including matrix habitats in survey design (e.g. Driscoll et al. 2013, Sweany et al.

2014) is therefore key to understanding how forest dependent species are affected by forest fragmentation (e.g. Didham et al. 2012, Kupfer et al. 2006). This generates information on how different species respond to different components of the landscape, e.g. forest patches, remaining continuous forest, and the intervening matrix (e.g. Campbell et al. 2011, Prevedello

& Vieira 2010). This is very important when considering conservation actions in human dominated landscapes (e.g. Driscoll et al. 2013, Sweany et al. 2014) as a way of turning

4

around current forest biodiversity declines and bridging the knowledge gap in ecological studies.

Role of insects in understanding forest-matrix dynamics:

Insects are diverse with a range of functional roles important for many ecosystem processes.

They respond rapidly and dramatically to changes in environmental conditions (e.g. Gerlach et al. 2013) and in particular have been found to be good indicators of various forest conditions (e.g. Niemelä 2001, Maleque et al. 2009). Many tropical species are locally endemic or are rare and with patchy distribution that predisposes them to increased extinction risk when habitats are modified (e.g. Terborgh 1992). Consequently, conservation of many such species will depend on the capacity of fragmented forests and the quality and extent of the matrix to support their populations.

Changes in the abundance, structure, and diversity of butterfly assemblages have been linked to gradients of human-generated disturbances (Blair & Launer 1997); making butterflies the most popular taxon for study within fragmented landscapes (Warren & Bourn

2011). Moths comprise a diverse and abundant taxon in many forest systems and play important roles as herbivores, pollinators and prey (Janzen 1987, Barlow & Woiwod 1989).

They are host specific (e.g. Janzen 1988) and serve as indicators of native plant diversity and local land management (Erhardt & Thomas 1991, Luff & Woiwod 1995, Ricketts et al. 2001).

Grasshoppers are a major if not dominant group of herbivorous insects throughout the world; their high diversity, functional importance and sensitivity, along with the ease with which they can be sampled makes grasshoppers excellent bioindicators for use in assessments of ecological change associated with land-use (Saha & Haldar 2009, Samways 1997, Bazelet &

Samwaus 2011).

5

A key question for this study is how much altered habitats maintain forest biodiversity. We aim to provide quantitative species-level datasets on the distribution of selected insect fauna in modified habitats for comparison with data from adjoining forests.

We specifically evaluate the presence of forest specialist species, and test the hypotheses that:

(a) modified habitats have significantly lower species richness and individual abundances of selected insect groups than less disturbed forest habitats; (b) species richness and abundance of forest specialist species within modified habitats is dependent on habitat type; (c) strongly disturbed habitats are more similar to each other due to biotic homogenization; d) different species groups will show contrasting responses to the land-use gradient.

Study areas and Methods

Study sites

Surveys were carried out between February 2009 and March 2011 in three protected forests of contrasting size in central Uganda (Mabira, Mpanga and Zika central forest reserves) and their surrounding habitats (Figure 1).

Mabira (0º24´– 0º35´N, 32º52´– 33º07´E) covers an area of 306 km2, the largest block of moist semi-deciduous forest remaining in central Uganda (Carswell 1986). Howard (1991) described four major habitat types: a) younger secondary forests dominated by colonizing

Maesopsis eminii; b) valley bottom forest dominated by Baikiaea insignis; c) the Celtis–

Holoptelea dominated forest and d) mixed communities. Large sections of the forest are now covered by the exotic species Broussonetia papyrifera (Winterbottom and Eilu (2006, field observations). The forest is protected and managed as a Central Nature Reserve by the

National Forestry Authority (NFA).

Mpanga (0º15´N, 32º18´E) covers an area of 4.53 km2. It is a remnant tropical, medium altitude, moist evergreen and swamp forest comprised of: a) swamp – permanently

6

flooded or water logged Mitragyna–Phoenix associations; b) the slopes with Celtis–Aningeria associations; and c) main forest dominated by Pseudospondias microcarpa, Erythrina excelsa, Canarium schweinfurthii and Entandrophragma angolense (Buxton 1952). The forest was gazetted as a nature reserve in 1950 and is under the jurisdiction of NFA.

Zika (0º07´N, 32º31´E) covers an area of 0.13 km2. The forest is part of a narrow strip of lakeside forests skirting the extensive grass and papyrus swamps of Waiya Bay, a sheltered inlet of Lake Victoria. Buxton (1952) recognized three zones in the forest: a) permanent swamp forest dominated by Mitragyna stipulosa, Erythrina excelsa and Voacanga obtusa; b) raised wet forest dominated by Pseudospondias microcarpa, Parkia filicoidea and

Macaranga monandra; c) raised seasonal forest dominated by Lovoa brownie, Maesopsis eminii and Piptadenia africana. The forest has been under the jurisdiction of the Uganda

Virus Research Institute since 1960.

Insect sampling strategy

Transects (10 x 500 m) were established within the selected sites in the forests and matrix habitats. In forest sites, transects were located away from the edge (≥ 100 m), except in Zika where transect was placed through the centre of the forest because of its small size. In Mabira, transects were in mature forest (4), secondary forest (3), cardamom plantation (1), coffee plantations (2) and mixed gardens (2). In Zika, sites considered were dry heterogeneous forest, permanent swamp forest and mixed gardens. In Mpanga, transects were in mature forest (2), secondary forest (1) andmixed gardens (1). Each transect was sampled for three consecutive days in both wet and dry seasons.

7

Baited traps for fruit-feeding butterflies (TB)

Standard baited traps similar to ones used by Molleman et al. (2006) and DeVries et al.

(1997) were used. Ten trap stations were positioned at 50m intervals along each transect, with each trap station fitted with one understorey trap and one canopy trap. The understorey traps were hung between 1 and 1.5 meters above the ground while the canopy trap heights varied between 15 and 30 meters. For the matrix habitats, the canopy height varied depending on the availability of trees. Traps were positioned in such a way as to sample from within the crown of the trap tree. Traps were baited prior to the first day of sampling with fermented banana prepared prior to use. Bait was placed in a small plastic plate inside each trap; fresh bait was added the following day, especially in traps where they had either been lost or eaten by other forest dwellers. During each trapping period, traps were emptied twice daily between 0900 and 1600 hours for three successive days. Checking was not done during or immediately after the rains to avoid damaging the specimens. Trapping was done for three consecutive trap nights per transect in both wet and dry seasons.

Transect sweep netting for butterflies (NB) and grasshoppers (GH)

The established transect at each site was walked at a constant pace lasting two hours. For butterflies, transect walks followed Pollard’s transect walking technique (Pollard 1977).

Sweep netting was done using a standard net with a conical bottom as described by Williams

(1971). Netting was carried out only when it was dry and warm, between 0900 hours and

1700 hours. Grasshopper assemblage was assessed through intensive sweep netting undertaken along transects at each of the study sites. Grasshoppers were seen as they dispersed away from the collector or rested on the vegetation, and collected with the net.

Sweep netting was done for three consecutive days per transect in both wet and dry seasons.

8

Light traps for Silk moths (SM) and Hawk moths (HM)

Two moth families: Saturnidae and Sphingidae were sampled. Sampling was done using a portable light-trap consisting of a 15 watt actinic tube (Sylvania blacklight F 15 w / BLB-TB) run on a 32 amps car battery and a net and was run from dusk to dawn. Trap points were located at beginning, middle and end of transect. Special attention was given to searching the area around the trap each morning for any moths that had been attracted to the light but had not entered the trap, and individuals of the target families were hand-collected.

Species identification and classification

For butterflies, familiar species were identified in the field, marked and released; those with unconfirmed identification were taken as voucher species and identified using guides (e.g.

Larsen, 1992 & 2005) and collections at Zoology museum (Makerere). Confirmations of other species were made by Torben Larsen. All moths of the two families were photographed; familiar species were identified in the field, marked and released. Voucher specimens were identified using collections at Zoology museum (Makerere), the Natural History Museum,

London and the Hope Museum, Oxford and web-based resource (www.Africanmoths.com).

Confirmations for difficult hawk moths were made by Ian Kitching. All grasshoppers were identified by Hugh Rowell.

All species recorded were assigned their ecological habitat preferences (ecotypes). For butterflies, the categorization followed Davenport (1996) and ecotypes included: a) Forest- dependent species (F) restricted to closed canopy forest habitats; b) Forest non-dependent species (f.), recorded in closed-canopy forest, but that are also encountered in a variety of forest-edge, degraded forest and woodland habitats; c) Open habitat species (O) characteristic of open habitats such as grassland, open savannah and arid habitats; d) Swamp/wetland

9

species (S) that are typical of forest riverine and swamps/ wetland habitats; e) Widespread generalist (W), ; f) species with migratory tendencies (M).

The status of all observed moth species was described in terms of their known ecotypes (Howard & Davenport 1996, Carcasson 1976) that included: a) Forest-dependent species (F) restricted to closed canopy forest habitats; b) Forest non-dependent species (f.) recorded in closed-canopy forest, but that are also encountered in a variety of forest-edge, degraded forest and woodland habitats; c) Non-forest species (G) characteristic of open habitats such as grassland, open savannah and arid habitats; d) Widespread generalists (W).

For grasshoppers, the ecotypes followed descriptions by Hugh Rowell

(pers.comm)and they included; a) typical forest species (F) found under closed forest canopies; b) forest edge species (f.) that typically need more light, either for themselves or their preferred food plants; c) open habitat species; (G) characteristic of open habitats.

Statistical analysis

Estimating within habitat diversity.

Species data across the different species groups were pooled for each land-use type. We calculated the effective number of species with Shannon index (Chao & Shen 2003, Hill

1973) for each land-use type. This index converges rapidly with little bias even for small samples (Magurran 2004). Because of the difficulty of obtaining a complete inventory of species rich communities (Price et al. 1995), the observed number of species is a misleading indication of species richness. Rarefied richness was therefore used to evaluate the effectiveness of sampling. Alternatively, richness was quantified using a non-parametric abundance-based, richness estimator Chao1 (Colwell & Coddington 1994). This estimator yields robust estimates and was used to assess within habitat diversity by extrapolation.

10

Estimating between habitat diversity

Assemblage composition has commonly been used for comparing the proportions of families of Macrolepidoptera (Kitching et al. 2000, Ricketts et al. 2001), but it has rarely been used for comparisons across different species groups within same localities. We used bar graphs to visualized assemblage composition using the mean richness and abundance of species within each of the five species groups: netted butterflies, trapped butterflies, silkmoths, hawkmoths and grasshoppers. The proportion of forest dependent species richness and abundance, in each habitat type across the three forests was added to the plot. Comparisons of the number of species observed and number of individuals per habitat type were performed with ANOVA.

Models for richness and abundance per site were explored to test for differences in species richness and abundances of the different species groups between the three forests and forest types. A negative binomial model with log link function was then used and due to the sampling structure of the data, variable Transect was nested within variable Forest to avoid pseudo-replication. Spearman's rank correlation of forest species abundance in forest vs. matrix habitats was performed to find out whether matrix occurrence of forest dependent species is an effect of their population size within the forest habitats. A t-test was performed to check if those forest specialist species found in the matrix habitats are more abundant in the forest compared to those that do not make it to the matrix.

Community analyses

Patterns of community structure and composition among different transects and between forests were visualized using non-metric multidimensional scaling (NMDS: Clarke, 1993) and species composition tested for significance using the Monte Carlo permutation test (999 runs, full model). The ordination scores were extracted and an NMDS analysis with three variables

(Forest, forest type and transect) was performed to test their effects on the different species

11

group assemblages. All the above data analyses were done using the statistical program R (v.

2.13.1, R Core team 2013).

Results

All togetheIn total, 25156 individuals of 327 species of butterflies, 1131 individuals of 41 species of Silkmoths , 1564 individuals of 44 species of Hawkmoths and 2173 individuals of

49 species of Grasshoppers were recorded across the three forests and their surrounding matrix habitats. Species diversity was generally highest in mature forests compared to other land-use types across the three forests; the effective number of species ranged between 16.9 and 53.4 (netted butterflies), 6.1 and 34.9 (trapped butterflies), 8.3 and 26.8 (silkmoths), 10.0 and 27.1 (hawkmoths), and 4.5 and 11.3 (grasshoppers) respectively (Table 1). Both rarefied and Chao1 estimated species richness were generally higher in mature forest sites whereas communities in heavily disturbed mixed gardens were significantly less diverse (Table 1). The

Chao1 estimates show that all the species groups except the butterrflies had been sampled nearly completely (Table 1).

Compared to mature forest, all other land-use types sampled had a negative effect on raw values for species richness and abundance of the different species groups, though significance of these relationships were inconsistent. Exceptions were the cardamom plantation which showed positive effect on trapped butterfly richness and abundance and secondary forest which showed positive effect on grasshoppers abundance (Table 2).

Secondary forests had significantly lower richness and abundance for netted and trapped butterflies compared to mature forest (Table 2, Figures 2 & 3). The other species groups were not significantly different compared to mature forest. Cardamom plantation had significantly lower richness for silkmoths, and significantly lower abundances for trapped butterflies,

12

silkmoths and grasshoppers compared to mature forest (Table 2, Figures 2 & 3). Swamp forest had significantly lower richness for netted butterflies and trapped butterflies (Table 2,

Figure 2), and significantly lower abundances for all species groups except grasshoppers compared to mature forest (Table 2, Figure 3). Coffee gardens had significantly lower richness for netted butterflies, hawkmoths and grasshoppers compared to mature forest (Table

2, Figure 2), and also significantly lower abundances for all the species groups except for silkmoths compared to mature forest (Table 2, Figure 3). Mixed gardens had significantly lower richnes for all species groups except grasshoppers, and significantly lower abundances for all species groups except netted butterflies (Table 2, Figures 2 & 3).

Across the three forests, sites in Zika forest had significantly lower richness for netted butterflies, trapped butterflies and hawkmoths compared to sites in Mabira, and significantly lower abundances for all species groups except grasshoppers compared to sites in Mabira

(Table 2). Sites in Mpanga forest had significantly lower richness for all species groups except silkmoths compared to sites in Mabira, and also significantly lower abundances for trapped butterflies, hawkmoths and grasshoppers compared to sites in Mabira (Table 2).

The NMDS reflects consistent patterns across the different species groups and geography (Figure 4). Differences in composition of the five species groups varied across forests and habitat types; the three forests form separate groups (Figure 4) meaning that there exists unique fauna within each forest due to their ecological and climatic conditions. There is also consistent pattern of distribution across the different species groups within the different land-use types hereby interpreted as a measure of habitat disturbance.

Considering site:transect relationships, for netted butterfly richness, only three site:transect combinations were significant (MA6: Est =0.60, P = 0.014; MA8: Est =0.32, P =

0.028; MA10: Est =0.42, P = 0.045), while four site: transect combinations were only marginally significant (MA5: Est = -0.26, P = 0.095; MA9: Est =0.44, P = 0.079; MA11: Est

13

=0.41, P = 0.095; MA12: Est = -0.31, P = 0.068). For netted butterfly abundances, only three site:transect combinations were significant (MA8: Est =0.67, P < 0.001; MP2: Est = -0.49, P

= 0.044; MP3: Est = -0.56, P = 0.02), and one site:transect combination marginally significant

(MA6: Est =0.61, P = 0.05). For trapped butterfly richness, three site:transect combinations were significant (MA5: Est = 0.42, P = 0.007; MA8: Est = 0.63, P < 0.001; MA10: Est = 0.52,

P = 0.009), with one site:transect combination only marginally significant (MA12: Est = 0.21,

P = 0.097). Abundance of trapped butterflies varied significantly within seven site:transect combinations (MA5: Est = 0.87, P < 0.001; MA6: Est = 1.89, P <0.001; MA8: Est = 0.80, P <

0.001; MA9: Est = 1.58, P < 0.001; MA10: Est = 1.25, P < 0.001; MA11: Est = 1.12, P =

0.001; MA12: Est = 0.86, P < 0.001) and marginally significant within three site:transect combinations (MA4: Est = 0.54, P = 0.079; MP2: Est = 0.47, P = 0.082; MP3: Est = 0.51, P =

0.061). For silkmoths, site:transect combinations were only significant for abundance data within three site:transect combinations (MA3: Est = -7.635e-01, P = 0.019; MP2: Est = -

5.908e-01, P = 0.022; MP3: Est = -8.285e-01, P = 0.002). Only one site:trasnect combination was marginally significant for hawkmoths abundance (MP3: Est = -0.45, P = 0.06) and no output was possbile for the grasshopper data with the specified model.

Reaction of forest dependent species to land-use gradient

The occurrence of forest dependent species in matrix habitats is significantly dependent on species population size within the forest habitats for netted butterflies (rho = 0.28, P < 0.001) and trapped butterflies (rho = 0.33, P = 0.003). The occurrence of forest dependent species in matrix habitats was not significantly dependent on species population size within the forest habitats for silkmoths (rho = 0.23, P = 0.42); hawkmoths (rho = 0.28, P = 0.27) and grasshoppers (rho = 0.12, P = 0.55) respectively. Subsequent test show that forest dependent species found in the matrix habitats were significantly more abundant in the forest amongst

14

netted butterflies (t = 3.78, df = 81, P < 0.001), marginally significant for trapped butterflies (t

= 1.87, df = 40, P = 0.07), not significant for silkmoths (t = 0.92, df = 13, P = 0.38), hawkmoths (t = 1.40, df = 15, P = 0.18) and grasshoppers (t = 1.54, df = 20, P = 0.14) respectively.

15

Discussion

Insect diversity and abundance along land-use gradient: from mature forest to mixed gardens

Findings of this study indicate that species richness and abundance of different insect groups vary considerably along land-use gradient. The predictions that modified habitats would have significantly lower species richness and individual abundances of selected insect groups than less disturbed forest habitats was therefore confirmed. Although our gradients are not true replicates of each other, patterns which emerged for these land-use types show some general trends. i) the mature forests even though moderately disturbed in the past have very high diversity of the five species groups, ii) continuously disturbed habitats like mixed gardens have much lower diversity of the five species groups, iii) Cardamom plantation and secondary forests showed a more variable picture.

Overall, diversity declined steadily with increasing human influence on the habitats.

Several studies have indicated similar patterns of species distributions (e.g. Barlow et al.

2007, Kitching et al. 2000, Prevedello & Vieira 2010). In herbivorous insects, different mechanisms are responsible for the impoverishments of communities with increasing disturbance. For example, change in microclimatic conditions due to forest modification is known to impact greatly on shade-adapted butterfly species (e.g. Hamer & Hill 2000). Also there is a shift in resource availability from mature forest to mixed gardens, and this subsequently impact on species along such gradients (e.g. Dennis et al. 2013).

NMDS plots revealed that communities of the different species groups from the three forests were not particularly similar to each other, although they share broadly similar community responses to land-use change. Assemblages in Zika and Mpanga forests were much similar to each other compared to communities in Mabira forest irrespective of the degree of habitat disturbance, providing a measure of geographical differentiation. Similarly,

16

swamp forest in Zika had unique arrays of grasshoppers compared to those in other land-use types. These findings provide evidence for a well-defined habitat specific fauna.

Assemblages of the different species groups converged along gradient of land-use within each forest. This provides evidence for biotic homogenization (i.e. decline of diversity along land- use gradient) in patch-matrix landscapes and is more in agreement with a notion that communities and species interactions are frequently determined by locally acting ecological and environmental processes (e.g. MacArthur 1972), rather than by the often presumed regional species pool (e.g. Cornell & Lawton 1992). Such biodiversity results are valuable in identifying ways of managing the rapidly changing tropical forest landscapes by providing evidence that would help minimize ecological impacts that result from forest habitat destruction and alteration (e.g. Shreeve & Dennis 2011, Sweaney et al. 2014).

Spatial heterogeneity and the landscape configuration of a particular land-use type are critical in determining the total number of species such a land-use can support. The lack of species response data from some of the most dominant and rapidly expanding land-use systems around protected forests especially in Uganda (e.g. sugarcane, tea and palm oil amongst others) is a major drawback to our understanding of biodiversity prospects in tropical forest landscapes (e.g. Fitzherbert et al. 2008). For example, some forest species may not be able to establish self-sustaining populations in the matrix because of the complex nature of matrix habitats and their subsequent influences on patch associated fauna (e.g. Kraus et al.

2003, Dennis 2012). This is potentially because most of such species require the presence of specific plant species for their survival (e.g. Dennis et al. 2004). However, some matrix types may successfully contribute to the persistence and survival of populations within patches by:

(i) facilitating the dispersal of species between patches (e.g. Gascon et al. 1999, Kuefler et al.

2010) and influencing the outcome of movement into patches (e.g. Schwab & Zandbergen

2011), (ii) providing food resources to fauna inside patches, and/or (iii) influencing conditions

17

experienced at the edges of patches which in turn favours certain species over others (e.g.

Driscoll & Donovan 2004, Lindenmayer et al. 2009, Ries & Sisk 2010). This was evident from our results in the cardamom plantation that maintains some structural aspects of mature forests with intact canopy and understorey in the form of planted cardamom, enabling the plantation to host high proportions of forest dependent species across the different species groups.

Role of matrix habitats in maintaining forest dependent biodiversity

There was a remarkable decline in richness and abundance of forest dependent species from mature forest to regularly maintained mixed gardens. The capacity of the forest species to utilize matrix habitats is dependent not only on the structural resistance of different matrix habitats (e.g. Stevens et al. 2004) but also on food availability (e.g. Dennis et al. 2013), edge permeability (e.g. Campbell et al. 2011) and increased mortality (e.g. Russel et al. 2003) within these altered habitats. Since these factors are distributed heterogeneously in the forest- matrix landscape, species-specific dispersal capacity would also vary (e.g. Ricketts 2001).

The ability of species to disperse through and use the space between habitat fragments is affected by matrix quality (e.g. Sweaney et al. 2014). Recent research has shown that within- patch dynamics are affected by the surrounding landscape on quite large spatial scales (e.g.

Bergman et al. 2004), and are not just limited to the area immediately surrounding a patch edge (e.g. Didham & Ewers 2012, Kennedy et al. 2010).

Forest specialists’ species from all the five species groups we studied showed very similar response to gradient of land-use. The interpretation of our results is supported by the findings from other studies (e.g. Roland et al. 2000, Ricketts 2001). The presence of fewer forest dependent species in the matrix habitats could be due to: a) suitability of matrix habitats for these species, or (b) their effective dispersal from adjacent mature forest. For many

18

organisms, movement rates within fragmented landscapes are known to vary with environmental factors such as resource availability and population density (e.g. Bowler &

Benton 2005). Our findings show that the occurrence of forest dependent species in matrix habitats is significantly dependent on species population size within the forest habitats, although the levels of significance varied with different species groups that we studied.

Studies have shown that factors like intensity of land-use within the intervening matrix habitat and patch isolation do affect species distribution (e.g. Prugh et al. 2008).

The permeability of matrix habitats is therefore important in determining species movement as well as effecting dispersal success within the patch-matrix landscape. For example, some disturbed habitats in our study such as degraded secondary forest remnants and cardamom plantation had plant resources and environment similar to mature forests and were able to effectively attract forest dependent species from adjacent mature forests. The ability of these species to utilize both forest and the modified habitats is a key factor in determining their diversity and abundance within the wider forest-matrix landscape. This is because altered habitats near forest fragments represent sink habitats for some species, offering sufficient resources to allow reproduction, but making them dependent on an influx of individuals from the forest to avoid local extinction. As such, dispersal rates and distance between patch and the matrix are therefore key factors that determine colonisation rates of suitable matrix habitats. Consequently, they influence patch extinction rates and ultimately determine the fate of local populations and the survival of metapopulation within the forest- matrix landscape (e.g. Hansson 1991, Wiens 2001).

Different species do not perceive human-modified landscapes as black and white mosaics of habitat and non-habitat (Fischer & Lindenmayer 2007). Despite taxon-specific differences in response patterns, forest biodiversity have generally been shown to decline along a coarse gradient from old-growth forest (mature forest) to secondary forest, agro-

19

forestry, plantations, arable crops and pasture (Schulze et al. 2004, Basset et al. 2008). This was exhibited by our results on species richness and abundance distribution of all species groups and even more especially for the forest dependent species, broadly reflecting the decline in floristic and structural diversity along land-use gradient. Consequently the retention or management of structurally and floristically complex habitats like secondary forests, cardamom plantations and shade coffee can often ensure the persistence of some forest species in managed landscapes (e.g. Lamb et al. 2005, Bhagwat et al. 2008, Scales &

Marsden 2008). Nevertheless, studies of regenerating forests demonstrate that biotic recovery occurs over considerably longer time scales than structural recovery, and that reestablishment of certain species and species group composition can take centuries or millennia (DeWalt et al. 2003, Liebsch et al. 2008). For example, in Kibale National Park, studies showed differences in fruit-feeding butterflies among successional stages of forest recovery, and only a marginal directional change along the successional gradient (Nyafwono et al. 2014b). It was estimated that fruit-feeding butterfly assemblages of restored forests would be similar that of primary forests within 40 years, provided that primary forests are nearby (Nyafwono et al.

2014a)

Conclusion

The use of different species groups provided the opportunity to make insights into how we interpret and understand the conservation values of human modified habitats. Results of this study highlight the fact that matrix resource use by the different species groups and individual species is not uniform across different types of matrix and is not necessarily different from resource use within suitable habitat. Although we strongly urge measures that prioritize the conservation of protected forests in central Uganda and within the country as a whole, we acknowledge that surrounding matrix habitats that maintain heterogeneity across the

20

landscape especially those that have elements similar to those within mature forests (in our case cardamom and some shade coffee plantations) should be prioritized. For effective forest conservation within a wider landscape context and alternative biodiversity friendly forest use, such habitats should be prioritized. Nevertheless, their long-term conservation value as alternative conservation areas will be determined by the proportions of forest dependent species whose distribution is restricted to mature forest within these modified habitats.

21

Acknowledgements

We are grateful to Hugh Rowell reading the manuscript drafts; Torben Larsen (R.I.P) for identifying difficult butterflies; Darren Mann and James Hogan at the Hope Museum, Oxford for help accessing Angus McCrae’s field notes and moth collections; Ian Kitching of Natural

History Museum, London (NHM) for identifying difficult Sphingidae; Alessandro Giusti of

NHM for locating relevant collections; and Janet McCrae for hosting me in Oxford. We are also grateful to Joseph Chipperfield of UiB for help with some analyses. The research was funded by the Norwegian Research Council FRIMUF programme MATRIX project (#

184912) and supported by the University of Bergen–Makerere University Collaboration.

References

Barlow, H.S. and Woiwod, I.P. (1989) Moth diversity of a tropical forest in peninsular

Malaysia. Journal of Tropical Ecology 5: 37–50.

Barlow, J., Gardner, T.A., Araujo, I.S., Avila-Pires, T.C.S., Bonaldo, A.B., Costa, J.E.,

Esposito, M.C., Ferreira, L.V., Hawes, J., Hernandez, M.I.M., Hoogmoed, M.S.,

Leite, R.N., Lo-Man-Hung, N.F., Malcolm, J.R., Martins, M.B., Mestre, L.A.M.,

Miranda-Santos, R., Nunes-Gutjahr, A.L., Overal, W.L., Parry, L., Peters, S.L.,

Ribeiro-Junior, M.A., da Silva, M.N.F., , da Silva Motta, C. and Peres, C.A. (2007)

Quantifying the biodiversity value of tropical primary, secondary and plantation

forests. Proceeding of the National Academy of Science USA 104: 18555–18560.

Basset, Y., Missa, O., Alonso A., Miller, S.E., Curletti, G., De Meyer, M., Eardley, C., Lewis,

O.T., Mansell, M.W., Novotny, V. and Wagner, T. (2008) Faunal turnover of

arthropod assemblages along a wide gradient of disturbance in Gabon. African

Entomology 16: 47–59.

22

Bazelet, C.S. and Samwaus, M.J. (2011) Identifying grasshopper bioindicators for habitat

quality assessment of ecological networks. Ecological Indicators 11:1259–

1269.

Bergman, K.O., Askling, J., Ekberg, O., Ignell, H., Wahlman, H. and Milberg, P. (2004)

Landscape effects on butterfly assemblages in an agricultural region. Ecography 27:

619–628.

Bhagwat, S.A., Willis, K.J., Birks, H.J.B. and Whittaker, R.J. (2008) Agroforestry: a refuge

for tropical biodiversity? Trends in Ecology and Evolution 23: 261–264.

Blair, R.B. and Launer, A.E. (1997) Butterfly diversity and human land use: Species

assemblages along an urban gradient. Biological Conservation 80: 113–125.

Bowler, D.E and Benton, T.G. (2005) Causes and consequences of animal dispersal strategies:

relating individual behaviour to spatial dynamics. Biological Reviews 80: 205–225.

Buxton, A.P. (1952) Observations on the Diurnal Behaviour of the Redtail Monkey

(Cercopithecus ascanius schmidti Matschie) in a Small Forest in Uganda. Journal of

Animal Ecology 21: 25–58.

Campbell, R.E., Harding, J. S., Ewers, R.M., Thorpe, S. and Didham, R.K. (2011) Production

land use alters edge response functions in remnant forest invertebrate communities.

Ecological Applications 21: 3147–3161.

Carcasson, R.H. (1976) Revised catalogue of the African Sphingidae (Lepidoptera) with

descriptions of the East African species. 2nd edn. E.W. Classey, Oxon, England.

Carswell, M. (1986) Birds of the Kampala area. Scopus special supplement Number 2 of the

Ornithological Subcommittee of the East Africa Natural History Society

Ornithological Sub-committee, EANHS. An annotated checklist No. 4232. Nairobi,

Kenya, pp: 86.

23

Chao, A. and Shen, T-J. (2003) Nonparametric estimation of Shannon’s index of diversity

when there are unseen species in sample. Environmental and Ecological Statistics 10:

429–443.

Chazdon, R.L., Harvey, C.A., Oliver, K., Griffith, D., Ferguson, B., Martinez-Ramos, M.,

Morales, H., Nigh, R., Soto-Pinto, L., Van Breugel, M. and Philpott, S.M. (2009)

Beyond reserves: a research agenda for conserving biodiversity in human-modified

tropical landscapes. Biotropica 41: 142–153.

Clarke, K.R. (1993) Non-parametric multivariate analyses of changes in community structure.

Australian Journal of Ecology 18: 117–143.

Colwell, R.K. and Coddington, J.A. (1994) Estimating the extent of terrestrial biodiversity

through extrapolation. Philosophical Transactions Royal Society, Series B 345: 101–

118.

Cornell, H.V. and Lawton, J.H. (1992) Species interactions, local, and regional processes, and

limits to the richness of ecological communities: a theoretical perspective. Journal of

Animal Ecology 61:1–12.

Davenport, T.R.B. (1996) The Butterflies of Uganda - An Annotated Checklist. Uganda Forest

Department, Kampala, Uganda.

Dennis, R.H., Dapporto. L., Dover. J. and Shreeve, T. (2013) Corridors and barriers in

biodiversity conservation: a novel resource-based habitat perspective for butterflies.

Biodiversity Conservation 22: 2709–2734.

Dennis, R.L. (2012) A resource-based habitat view for conservation: butterflies in the British

landscape. Wiley, London.

Dennis, R.L.H., Hodgson, J.G., Grenyer, R., Shreeve, T.G. and Roy, D.B. (2004) Host plants

and butterfly biology. Do host plant strategies drive butterfly status? Ecological

Entomology 29:12–26.

24

DeVries, P.J., Murray, D. and Lande, R. (1997) Species diversity in vertical, horizontal, and

temporal dimensions of a fruit-feeding butterfly community in an Ecuadorian

rainforest. Biological Journal of the Linnaean Society 62: 343–364.

DeWalt, S.J., Maliakal, S.K. and Denslow, J.S. (2003) Changes in vegetation structure and

composition along a tropical forest chronosequence: implications for wildlife. Forest

Ecology and Management 182: 139–151.

Didham, R.K. and Ewers, R.M. (2012) Predicting the impacts of edge effects in fragmented

habitats: Laurance and Yensen’s core area model revisited. Biological Conservation

155: 104–110.

Didham, R.K., Kapos, V. and Ewers, R.M. (2012) Rethinking the conceptual foundations of

habitat fragmentation research. Oikos 121:161–170.

Driscoll, D.A., Banks, S.C., Barton, P.S., Lindenmayer, D.B. and Smith, A.L. (2013)

Conceptual domain of the matrix in fragmented landscapes. Trends in Ecology and

Evolution 28: 605–613.

Driscoll, M.J.L. and Donovan, T.M. (2004) Landscape context moderates edge effects:

nesting success of wood thrushes in central New York. Conservation Biology

18:1330–1338.

Erhardt, A. and Thomas, J.A. (1991) Lepidoptera as indicators of change in the semi-natural

grasslands of lowland and upland Europe. In: The Conservation of Insects and their

Habitats (eds. Collins, M, and Thomas, J.A.), pp 213–236. Academic Press, London.

Ewers, R.M. and Didham, R.K. (2006) Confounding factors in the detection of species

responses to habitat fragmentation. Biological Reviews 81:117–142.

Fischer, J. and Lindenmayer, D.B. (2007) Landscape modification and habitat fragmentation:

a synthesis. Global Ecology and Biogeography 16: 265–280.

25

Fitzherbert, E.B., Struebig, M.J., Morel, A., Danielson, F., Brühl, C.A., Donald, P.F. and

Phalan, B. (2008) How will oil palm expansion affect biodiversity? Trends in Ecology

and Evolution 23: 538–545.

Franklin, J.F. and Lindenmayer, D.B. (2009) Importance of matrix habitats in maintaining

biological diversity. Proceedings of the National Academy of Sciences U.S.A. 106:

349–350.

Gaines, S.D., Lester, S.E., Grorud-Colvert, K., Costello, C., Pollnac, R. (2010) Evolving

science of marine reserves: new developments and emerging research frontiers.

Proceedings of the National Academy of Sciences U.S.A. 107: 18251–18255.

Gascon, C., Lovejoy, T. E., Bierregaard Jr, R. O., Malcolm, J.R., Stouffer P. C.,. Vasconcelos,

H. L., Laurance, W. F., Zimmerman, B., Tocher, M. and Borges, S. (1999) Matrix

habitat and species richness in tropical forest remnants. Biological Conservation

91: 223-229.

Gerlach, J., Samways, M. and Pryke, J. (2013) Terrestrial invertebrates as bioindicators: an

overview of available taxonomic groups. Journal of Insect Conservation 17: 831–850.

Green, R.E., Cornell, S.J., Scharlemann, J.P.W. and Balmford, A. (2005) Farming and the

Fate of Wild Nature. Science 307: 550–555.

Hamer, K.C. & Hill, J.K. (2000) Scale-dependent consequences of habitat modification for

species diversity in tropical forests. Conservation Biology 14: 1435–1440.

Hamer, K.C., Hill, J.K., Benedick, S., Mustaffa, N., Sherratt, T.N., Maryati, M. and Chey,

V.K. (2003) Ecology of butterflies in natural and selectively-logged forests of northern

Borneo: the importance of habitat heterogeneity. Journal of Applied Ecology 40: 150–

162.

Hanski, I. (1999) Habitat connectivity, habitat continuity, and metapopulations in dynamic

landscapes. Oikos 87: 209–219.

26

Hansson, L. (1991) Dispersal and connectivity in metapopulations. Biological Journal of the

Linnean Society 42: 89–103.

Hill, M.O. (1973) Diversity and evenness: a unifying notation and its consequences. Ecology

54: 427–432.

Howard, P.C. and Davenport, T.R.B. (1996) Forest biodiversity reports. Uganda Forest

Department, Kampala, Uganda.

Howard, P.C. (1991) Nature conservation in Uganda's tropical reserves. IUCN. Gland,

Switzerland.

Janzen, D.H. (1987) Insect diversity of a Costa Rican dry forest – why keep it, and how.

Biological Journal of the Linnean Society 30: 343–356.

Janzen, D.H. (1988) Ecological characterizations of a Costa Rican dry forest caterpillar fauna.

Biotropica 20: 120–135.

Joppa, L.N. and Pfaff, A. (2011) Global protected area impacts. Proceeding of the Royal

Society London, Series B 278: 1633–1638.

Kennedy, R.E., Yang, Z. and Cohen, W.B. (2010) Detecting trends in forest disturbance and

recovery using yearly Landsat time series: 1. LandTrendr–Temporal segmentation

algorithms. Remote Sensing of Environment 114: 2897–2910.

Kitching, R.L., A.G. Orr, L.Thalib, H. Mitchell, M.S. Hopkins, and A.W. Graham. (2000)

Moth assemblages as indicators of environmental quality in remnants of upland

Australian rain forest. Journal of Applied Ecology 37: 284–297.

Krauss, J., Steffan-Dewenter, I. and Tscharntke, T. (2003) How does landscape context

contribute to effects of habitat fragmentation on diversity and population density of

butterflies? Journal of Biogeography 30: 889–900.

27

Kuefler, D., Hudgens, B., Haddad, N.M., Morris, W.F. and Thurgate, N. (2010) The

conflicting role of matrix habitats as conduits and barriers for dispersal. Ecology 91:

944–950.

Kupfer, J.A., Malanson, G.P. and Franklin, S.B. (2006) Not seeing the ocean for the islands:

the mediating influence of matrix-based processes on forest fragmentation effects.

Global Ecology and Biogeography 15: 8–20.

Lamb, D., Erskine, P. and Parrotta, J.A. (2005) Restoration of Degraded Tropical Forest

Landscapes. Science 310: 1628–1632.

Laurance, W.F. and Bierregaard Jr, R.O. (eds). (1997) Tropical Forest Remnants: Ecology,

Conservation, and Management of Fragmented Communities. University of Chicago

Press, Chicago.

Lawton, J.H., Bignell, D.E., Bolton, B., Bloemers, G.F., Eggleton, P., Hammond, P.M.,

Hodda, M., Holt, R.D., Larsen, T.B., Mawdsleu, N.A., Stork, N.E., Srivastava, D.S.

and Watt, A.D. (1998) Biodiversity inventories, indicator taxa and effects of habitat

modification in tropical forest. Nature 391: 72–76.

Liebsch, D., Marques, M.C.M. and Goldenberg, Renato. (2008) How long does the Atlantic

Rain Forest take to recover after a disturbance? Changes in species composition and

ecological features during secondary succession. Biological Conservation 141: 1717–

1725.

Lindenmayer, D.B., Wood, J.T., Cunningham, R.B., Crane, M., Macgregor, C., Michael, D.

and Montague-Drake, R. (2009) Experimental evidence of the effects of a changed

matrix on conserving biodiversity within patches of native forest in an industrial

plantation landscape. Landscape Ecology 24: 1091–1103.

Luff, M.L. and Woiwod, I.P. (1995) Insects as indicators of land-use change: a European

perspective, focusing on moths and ground beetles. In: Insects in a Changing

28

Environment (eds. Harrington, R. and Stork, N.E.), pp 399–422. Academic Press,

London.

MacArthur, R.H. (1972) Geographical ecology. Harper and Row, New York

Magurran, A. E. (2004). Measuring Biological Diversity. Blackwell. Oxford.

Maleque, M.A., Maeto, K. and Ishii, H.T. (2009) Arthropods as bioindicators of sustainable

forest management, with a focus on plantation forests. Applied entomology and

Zoology 44: 1–11.

McIntyre, S. and Hobbs, R.J. (1999) A framework for conceptualising human effects on

landscapes and its relevance to management and research models. Conservation

Biology 13: 1282–1292.

Molleman, F., Kop, A., Brakefield, P.M., DeVries, P.J. and Zwaan, B.J. (2006) Vertical and

temporal patterns of biodiversity of fruit-feeding butterflies in a tropical forest in

Uganda. Biodiversity and Conservation 15: 107–121.

Mora, C. and Sale, P. (2011) Ongoing global biodiversity loss and the need to move beyond

protected areas: A review of the technical and practical shortcoming of protected areas

on land and sea. Marine Ecology Progress Series 434: 251–266.

Niemelä, J. (2001) The utility of movement corridors in forested landscapes. Scandinavian

Journal of Forest Research Supplementum 3: 70–78.

Pimm, S., Ayres, M., Balmford, A., Branch, G., Brandon, K., Brooks, T., Bustamante, R.,

Costanza, R., Cowling, R., Curran, L., Dobson, A., Farber, S., da Fonseca, G., Gascon,

C., Kitching, R., McNeely, J., Lovejoy, T., Mittermeier, R., Myers, N., Patz, J., Raffle,

B., Rapport, D., Raven, P., Roberts, C., Rodrıguez, J., Rylands, A., Tucker, C., Safina,

C., Samper, C., Stiassny, M., Supriatna, J., Wall, D. and Wilcove, D. (2001) Can we

defy nature’s end? Science 293: 2207–2208.

29

Pollard, E. (1977) A method for Assessing changes in abundance of butterflies. Biological

conservation 12:115–134.

Posa, M.R.C. and Sodhi, N.S. (2006) Effects of anthropogenic land use on forest birds and

butterflies in Subic Bay, Philippines. Biological Conservation 129: 256–270.

Prevedello, J.A. and Vieira, M.V. (2010) Does the type of matrix matter? A quantitative

review of the evidence. Biodiversity and Conservation 19: 1205–1223.

Price, P.W., Diniz, I.R., Morais, H.C. and Marques, E.S.A. (1995) The abundance of insect

herbivore species in the tropics: the high local richness of rare species. Biotropica 27:

468–478.

Prugh, L.R., Hodges, K.E., Sinclair, A.R.E. and Brashares, J.S. (2008) Effect of habitat area

and isolation on fragmented animal populations. Proceedings of the National Academy

of Sciences 105: 20770–20775.

Rands, R.W., Adams, W.M., Bennun, L., Butchart, S.H.M., Clements, A., Coomes, D.,

Entwistle, A., Hodge, I., Kapos, V., Scharlemann, J. P.W., Sutherland, W.J. and Vira,

B. (2010) Biodiversity conservation: Challenges beyond 2010. Science 329: 1298–

1303.

Ricketts, T. H. (2001) The Matrix Matters: Effective Isolation in Fragmented Landscapes. The

American Naturalist 158: 87–99.

Ricketts, T.H., Daily, G.C., Ehrlich, P.R. and Fay, J.P. (2001) Countryside Biogeography of

Moths in a Fragmented Landscape: Biodiversity in Native and Agricultural Habitats

Conservation Biology 15: 378–388.

Ries, L. and Sisk, T.D. (2010) What is an edge species? The implications of sensitivity to

habitat edges. Oikos 119: 1636–1642.

Roland, J., Keyghobadi, N. and Fownes, S. (2000) Alpine The Matrix Matters 99 Parnassius

butterfly dispersal: effects of landscape and population size. Ecology 81: 1642–1653.

30

Russell, R.E., Swihart, R.K. and Feng, Z.L. (2003) Population consequences of movement

decisions in a patchy landscape. Oikos 103: 142–152.

Saha, H.K. and Haldar, P. (2009) Acridids as indicators of disturbance in dry deciduous forest

of West Bengal in India. Biodiversity and Conservation 18: 2343–2350.

Samways, M.J. (1997) Conservation biology of Orthoptera. In: The Binomics of

Grasshoppers, Katydids and Their Kin (eds. Gangwere, S.K., Muralirangan, M.C. and

Muralirangan, M.), pp. 481–496. Wallingford, UK: CAB International.

Scales, B.R. and Marsden, S.J. (2008) Biodiversity in small-scale tropical agroforests: a

review of species richness and abundance shifts and the factors influencing them.

Environmental Conservation 35: 160–172.

Schooley, R.L. and Wiens, J.A. (2004) Movements of cactus bugs: patch transfers, matrix

resistance, and edge permeability. Landscape Ecology 19: 801–810.

Schtickzelle, N. and Baguette, M. (2003) Behavioural responses to habitat patch boundaries

restrict dispersal and generate emigration-patch area relationships in fragmented

landscapes. Journal of Animal Ecology 72: 533–545.

Schulze, C. H., Waltert, M., Kessler, P. J. A., Pitopang, R., Shahabuddin, R., Veddeler, D.,

Mühlenberg, M., Gradstein, S.R., Leuschner, C., Steffan-Dewenter, I. and Tscharntke,

T. (2004) Biodiversity indicator groups of tropical land-use systems: comparing

plants, birds, and insects. Ecological Applications 14: 1321–1333.

Schwab, A.C. and Zandbergen, P.A. (2011) Vehicle-related mortality and road crossing

behavior of the Florida panther. Applied Geography 31: 859–870.

Shreeve, T.G. and Dennis, R.L.H. (2011) Landscape scale conservation: resources, behaviour,

the matrix and opportunities. Journal of Insect Conservation 15: 179–188.

31

Stevens, V.M., Polus, E., Wesselingh, R.A., Schtickzelle, N. and Baguette, M. (2004)

Quantifying functional connectivity: experimental evidence for patch-specific

resistance in the Natterjack toad (Bufo calamita). Landscape Ecology 19: 829–842.

Sweaney, N., Lindenmayer, D.B. and Driscoll, D.A. (2014) Is the matrix important to

butterflies in fragmented landscapes? Journal of Insect Conservation 18: 283–294.

Terborgh, J. (1992) Maintenance of diversity in tropical forests. Biotropica 24: 283–292.

Vandermeer, J. and Carvajal, R. (2001) Metapopulation Dynamics and the Quality of the

Matrix. The American Naturalist 158: 211–220.

Vu, L.V. (2009) Diversity and similarity of butterfly communities in five different habitat

types at Tam Dao National Park, Vietnam. Journal of Zoology 277: 15–22.

Warren, M. and Bourn, N. (2011) Ten challenges for 2010 and beyond to conserve

Lepidoptera in Europe. Journal of Insect Conservation 15: 321–326.

Wiens, J.A. (2001) The landscape context of dispersal. In: Dispersal (eds. Clobert, J.,

Danchin, E., Dhondt, A.A. and Nichols, J.D.), pp 96–109. Oxford University Press,

Oxford, UK.

Williams, J.G. (1971) A field guide to the Butterflies of Africa. Collins, London.

Winterbottom, B. and Eilu, G. (2006) Uganda Biodiversity and Tropical Forest Assessment,

Final Report. A Publication for the United States Agency for International

Development.

32

Table 1. Number of individuals captured, observed species, and total species richness and diversity estimates for each of the forest areas

surveyed. . NB= Netted butterflies, TB = Trapped butterflies, SM = Silk moths, HM = Hawk moths, GH = Grasshoppers

Forest Habitat type Number of individuals expH’ Rarefied species richness Chao1 Name

NB TB SM HM GH NB TB SM HM GH NB TB SM HM GH NB TB SM HM GH Mabira Mature forest 3381 2246 255 309 595 53.2 11.5 26.8 15.4 7.1 145.0 94.0 32.0 26.0 19.0 185±18.0 80±8.7 32±0.1 31±5.5 20±1.8

Secondary forest 1676 538 172 215 169 42.6 34.9 20.7 11.6 8.9 88.0 88.0 28.0 16.0 15.0 137±10.5 99±19.3 28±0.9 16±0.2 15±0.2 Cardamom plantation 586 981 40 59 69 36.8 7.9 10.5 12.0 4.5 87.0 35.0 12.0 14.0 10.0 122±16.3 44±7.3 13±1.3 14±0.9 13±4.1

Coffee gardens 863 578 117 86 43 31.7 22.7 14.1 10.0 5.5 85.0 49.0 17.0 14.0 8.0 147±30.3 58±6.4 18±2.3 14±0.1 9±2.2 Mixed gardens 947 230 82 94 64 25.1 31.0 12.7 12.2 6.8 98.0 48.0 15.0 17.0 9.0 144±19.0 68±13.5 16±1.3 19±2.3 9±0.0

Zika Mature forest 665 1669 131 196 47 33.8 8.7 14.3 27.1 8.1 77.0 42.0 20.0 34.0 11.0 127±27.3 46±4.1 20±0.7 34±0.4 11±0.7

Swamp forest 185 1002 80 108 45 22.8 6.1 12.6 15.1 5.3 40.0 27.0 17.0 20.0 7.0 55±9.7 31±4.2 17±0.5 20±0.1 7±0.2 Mixed garden 425 522 60 96 66 29.5 7.4 8.3 11.9 7.9 54.0 28.0 9.0 15.0 12.0 65±7.2 30±2.5 9±0.0 16±1.3 12±0.9

Mpanga Mature forest 1348 4069 93 218 576 35.0 8.2 21.3 21.7 11.3 94.0 74.0 26.0 27.0 28.0 115±12.6 85±7.3 26±0.8 27±0.2 29±1.0

Secondary forest 659 847 58 103 425 30.9 6.9 17.0 20.3 10.8 66.0 39.0 21.0 28.0 15.0 73±5.3 50±8.5 24±2.9 30±2.0 15±0.0 Mixed garden 1348 391 43 80 75 16.9 13.4 11.4 11.5 9.6 48.0 42.0 17.0 16.0 12.0 53±4.6 66±16.4 24±6.4 16±0.7 12±0.0

33

Table 2. Summary of effect of different habitat types on species richness and abundance of selected insect groups. Also included are relations of species richness and abundance in Zika and Mpanga forests compared to Mabira forest. Significance codes: 0 ‘***’, 0.001 ‘**’, 0.01 ‘*’, 0.05

‘.’. Ma = Mature forest, Se = Secondary forest, Ca = Cardamom plantation, Sf = swamp forest, Co = Coffee garden, Mg = Mixed garden

Habitat types Other forests Species Ma Se Ca Sf Co Mg Zika Mpanga groups Richness NB 3.221 -0.292*** -0.130 -0.714*** -0.547*** -0.275*** -0.310*** -0.198** TB 2.598 -0.350*** 0.027 -0.425*** -0.096 -0.566*** 0.2161** 0.054*** SM 2.621 -0.088 -0.423* -0.237 -0.223 -0.356** -0.225 -0.176. HM 2.386 -0.141 -0.226 -0.310 -0.489** -0.339** 0.409** 0.370*** GH 2.396 -0.125 -0.317 -0.616 -0.787* -0.256 0.166 0.497**

Abundance NB 4.205 -0.311** -0.153 -0.858*** -0.604*** -0.077 -0.639*** -0.032 TB 3.979 -1.032*** 0.607** -0.695** -0.616*** -1.512*** 1.323*** 1.347*** SM 3.043 -0.055 -0.453* -0.378* -0.073 -0.468*** 0.618*** -0.145 HM 3.286 -0.105 -0.307. -0.458** -0.624*** -0.386*** 0.756*** 0.336*** GH 4.699 0.023 -1.754** -0.711 -1.703*** -0.743* -0.181 0.964***

34

Figure 1. The location of forests. Black dots are locations of sample points along individual transects.

35

Mabira forest 100 Mature forest

80 Secondary forest Cardamom plantation

Coffee plantation 60

Mixed garden

40

20 0 Netted Trapped Silk Haw k Hopper

Mpanga forest 100

Mature forest

80 Secondary forest

Mixed garden

60

40

20

Species richnesstransect per 0 Netted Trapped Silk Haw k Hopper

Zika forest 100 Mature forest

80 Sw amp forest

Mixed garden

60

40

20 0 Netted Trapped Silk Haw k Hopper Functional group

Figure 2. Richness (mean ± SE) of netted butterflies, trapped butterflies, silk moths, hawk moths and grasshoppers recorded in different land-use types from forest to mixed gardens per transect standardized by transect surveyed per transect walk (NB, GH) or trap night (TB, SM,

HM). White dots represent the richness of only the forest dependent species, superimposed over the total richness.

36

Mabira forest 1000 Mature forest

800 Secondary forest Cardamom plantation Coffee plantation

600 Mixed garden

400

200 0 Netted Trapped Silk Haw k Hopper

Mpanga forest 2500 Mature forest

Secondary forest 2000

Mixed garden

1500

1000

500 0 Species abundance per transect per Species abundance Netted Trapped Silk Haw k Hopper

Zika forest 2000 Mature forest Sw amp forest

1500 Mixed garden

1000

500 0 Netted Trapped Silk Haw k Hopper Functional group

Figure 3. Abundance (mean ± SE) of netted butterflies, trapped butterflies, silk moths, hawk moths and grasshoppers recorded in different land-use types from forest to mixed gardens per transect standardized by transect surveyed per transect walk (NB, GH) or trap night (TB, SM,

HM). White dots represent forest dependent species.

37

Trapped butterflies Netted butterflies

1.0

1.0

0.0

0.0

NMDS2 NMDS2

-1.0

-1.0 -2.0 -2 -1 0 1 2 -2 -1 0 1 2 NMDS1 NMDS1

Silkmoths Hawkmoths

1.0

1.0

0.5 0.5

0.0

0.0

NMDS2 NMDS2

-0.5 -1.0

-1.5 -0.5 0.0 0.5 1.0 1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 NMDS1 NMDS1

Grasshoppers 2.0 Mature Forest

1.0 Secondary Forest

Swamp Forest 0.0

NMDS2 Cardamon Plantation Coffee Plantation

-1.0 Mixed Garden

-2 -1 0 1 2 NMDS1 Figure 4. NMDS ordination for species abundances of five different species groups from 12 sites in Mabira forest (Black), three sites in Zika forest (Red) and four sites in Mpanga forest

(Green). Sampling was done in different habitat types from mature forest to mixed gardens.

38

Paper IV

Akite, P. & Rowell, C.H.F. (2013) Oshwea dubiosa rediscovered in

Uganda. Journal of Orthoptera Research 22 (1): 45–49.

Journal of Orthoptera Research 2013, 22(1): 45-49

Oshwea dubiosa rediscovered in Uganda

Perpetra Akite and C.H.F. Rowell

Dept. of Biological Sciences, Makerere University, Kampala, Uganda. Email: [email protected]; [email protected]

Abstract

The second known specimen of the genus Oshwea Ramme 1929 was nearly 100 years), neither is very likely to be forthcoming in the captured in Southern Uganda; it appears to be the unknown male of O. near future. dubiosa, previously recorded only from West-Central Congo. At present it seems best to retain it in the subfamily Catantopinae, though it differs from Methods other members of that taxon in several anatomical features.

Standard taxonomic methods were employed. The pinned Introduction specimen was relaxed in water for examination and dissection of the phallic complex. The latter was extracted, macerated in 8% In early March 2013 the authors spent a morning in the field NaOH solution, cleared and neutralised in very dilute acetic acid with the aim of members together, primary photographing typical solution, stained with acid fuchsin, differentiated in water, and of the local grasshopper fauna in their natural environment. The preserved in glycerine. Dimensions were measured with a digital locality was the overgrown verges of an unsurfaced road in the * stage micrometer used under 25 magnification. Drawings were Mabira forest, Uganda, connecting the hamlet of Bwola to the prepared using a drawing tube on a Wild M5 stereo microscope, major Kampala/Jinja highway which traverses the forest from East digitized, and elaborated using "Photoshop" (Adobe Systems Inc.). to West. The area and its fauna are well known to us, so we were not for as we no unknown equipped collecting, anticipated spe Description cies. Later, however, while reviewing photographs taken by Akite, we realised that some showed a completely unfamiliar insect. We Male paratype: UGANDA: Buganda: Mukono District: Mabira therefore returned to the same the and locality following morning, Central Forest Reserve: 1 km S.W.of Bwola, about 2.5 km ENE of were lucky enough to be able to find what appeared to be the same Najjembe, coordinates 0° 25'5"N, 33°3'33"E. 10.03.2013 (Akite individual and to capture it; regretably, no other specimens were P., Rowell CHF.) Specimen number 2013012. (BMNH London). found. Using Dirsh's 1965 generic key, it was easily identified as a member of the Catantopine genus Oshwea. Habitus, Figs 1 & 5. Oshwea is one of the least known members of the African for For details of dimensions, see Table 1. est fauna. It was described (Ramme 1929) from a single holotype female specimen of O. dubiosa, from the surroundings of the town = Medium in size, L (fastigium to subgenital plate) 18.99 mm. of Oshwe, in Bandundu Province, West-Central * Congo (3°40'S, Antennae filamentous, 1.73 as long as head and prono ° long, 19 with no other data, and to our knowledge has never been 50'E), tum together. since. Ramme's is meagre, and is illustrated x reported description Interocular space narrow, 0.76 as wide as antennal scape. a very inadequate photograph. The holotype, in the Berlin = only by Pronotum (P) in midline 3.78 mm. Museum, was pinned from alcohol, and has little coloration, as = base Elytra (E) 11.26 mm, shorter than abdomen, extending to noted Ramme. Only its black elytra with paler striping already by of supra-anal plate. to reflect the state of the living animal. Fortunately Dirsh x its appear Hind femur (F) long, 12.05 mm, and slender, length 4.55 and lateral of the had excellent dorsal drawings holotype prepared maximum width. G.W. for his 1965 and 1970 books, which allow one to by Dalby Hind foot fairly long, 43% of hind tibia. form a of the tarsal picture species. Foot formula (i.e., the lengths of the first, second and third Our is a male; as the unique holotype of O. of the Ugandan specimen joints, each expressed as a percentage of the total length foot) dubiosa is a female, we have no that the two are of - proof specimens 34:20:46 second tarsal joint rather short, third tarsal joint almost the same However, the to date is monospecific, the species. genus as long as first and second joints together. well, and the male shows the same striped specimens correspond Integument of head and thorax rugose, densely pitted. Integu black as Dirsh's of the female. We presume, therefore, elytra figures ment of legs and abdomen smooth, polished. in the absence of other evidence, that our specimen is the unknown male of O. dubiosa Ramme, and its here. Cer x present description Antennae filiform, long (1.73 as long as head and pronotum would either a female corresponding to the downward tainty require Ugandan together), 21 flagellar segments. Fastigium sloping, or, more a male from Oshwe corresponding to towards its holotype critically, slightly concave, with weak lateral carinulae base, merg our In view of the rarity of the taxon (2 specimens in above medial ocellus specimen. ing smoothly into frontal ridge. Frontal ridge

Journal of Orthoptera Research 2013, 22(1)

This content downloaded from 196.43.134.183 on Fri, 22 May 2015 09:53:50 UTC All use subject to JSTOR Terms and Conditions 46 p. akite and c.h.f. rowell

Fig. 1. OshiveaOshwea dubiosa. Male habitus. For color version, see Plate IV.

wide, not sulcate, pitted; below medial ocellus, increasingly obsolete, of the elytron. Wings cycloid, opaque blackish brown, somewhat Compound eyes protuberant, interocular space narrow, less than paler at the base. width of antennal scape. Pro- and mesothoracic femora somewhat laterally compressed, and widened dorsoventrally. Hind femora long and slender, exceed Pronotum subcylindrical, medial and lateral carinae absent. Disc ing both the abdomen and the elytra: medial dorsal carina almost of pronotum crossed by three fairly strong sulci. Anterior margin of completely smooth, ending at the knee in a minute tooth. Upper PN slightly convex, very weakly indented in the midline; posterior knee lobe rounded, lower knee lobe acute angular with a slightly margin obtuse-angulate, the apex truncate. Metazona shorter than rounded tip. Outer face of hind femur with a fairly prominent prozona. The posterior angles of the lateral lobes are slightly flared chevron pattern. Hind tibia with 6 external and 7 internal spines, outwards. Prosternai process short and conical. Mesosternal inter- including external and internal apical spines (Fig. 2). Hind foot space wide and open at the posterior edge, metasternal interspace fairly long (43% as long as tibia), foot formula 34:20:46. narrow and laterally compressed, almost closed posteriorly. Tenth abdominal tergite completely divided, furcula absent. Male Elytra slightly brachypterous, extending only to base of the cerci long and slender, extending to the tip of the subgenital plate, supra-anal plate. They are widest at about one third their length, inward curving, acutely (and in side view assymetrically) pointed, thereafter narrowing towards the tip. They are black in color, with a Supra anal plate triangular, with a weak medial depression basally, prominentvenationandreticulum;thereisanarrowtaperingyellow the tip rounded. Apex of subgenital plate smoothly rounded in stripe centred on the first anal vein, which does not reach the tip lateral view, with a weak medial carina in apical view (Fig. 3).

(See Figs 1 & 5) Antennae, reddish brown, basal seg (Coloration.— ments greenish. Head and thorax dark olive green, eyes black, palps 1 mm green. Elytra black with a short yellow longitudinal stripe basally. Mesothoracic episternum and epimeron, and metathoracic epister num, light yellow-green, forming a light patch contrasting with the rest of the head and thorax. All femora, leaf green, hind knee black. • Hind tibia blue green suffused with black, the distal tip carmine red. Hind foot carmine red. Tibial spines and spurs, mostly black.

Phallic complex.—(Fig. 4.) Epiphallus bridge-shaped, undivided, quite

B

Fig. 2. Oshwea dubiosa. Tip of hind tibia. The external apical spine Fig. 3. Oshwea dubiosa. Male terminaba.terminalia. A, lateral; B, dorsal,dorsal. is arrowed. Hatched area is pallium.

Journal of Orthoptera Research 2013, 22(1)

This content downloaded from 196.43.134.183 on Fri, 22 May 2015 09:53:50 UTC All use subject to JSTOR Terms and Conditions P. AKITE AND C.H.F. ROWELL 47

B

1 mm

Fig. 4. Oshwea dubiosa. Phallic complex. A-C, epiphallus. A, epiphallus, axial view. B, epiphallus, dorsal view. C, epiphallus. Lateral view. Sparsely stippled areas are membrane, heavier dots are sensilla in the membrane. D, phallic complex without epiphallus, dorsal view. The posterior projection of the zygoma is arrowed. E, phallic complex, lateral view. The posterior projection of the zygoma is arrowed. The heavy arrow indicates the ventral retractor apódeme (striated in the Figure), inserting at the tips of the rami. To conserve the unique specimen, the endophallus was not dissected out. large in relation to the phallus, about as wide as the endophallus genera. Oshwea has an external apical spine on the hind tibia (Mg. is long. Lophi large, erect, triangular, with an outer strip of clear, 2), a character which sets it apart from most catantopine genera, non-staining cuticle. Ancorae large, pointed, inwardly curved, the The other 9 African catantopine genera possessing this feature are membrane near their base densely provided with sensilla. Lateral mostly large apterous forest floor leaf-litter dwellers from W. Africa, plates of epiphallus rather small, pointed laterally. Oval sclerites such as Mazaea and Barombia; Oshwea, however, isa typical light-gap - large, roughly triangular, arranged in a "south cone" attitude, with species in its morphology smallish in size, brightly colored, alate, an apex pointing ventrally. with long antennae and protuberant eyes. The structure of the hind Ectophallus with robust diverging anterior apodemes, which foot, with its short second tarsal joint, does not suggest a primarily continue caudally beyond the zygoma and join together to form a arboreal life style. Its phallic structures are nothing like those of the prominent V-shaped process overhanging the aedeagus (arrowed in Mazaea group, which in turn differ greatly from those of the "core" Fig. 4D, E). Rami well developed, encircling the endophallus, but Catantopinae such as the genera around Catantops. The dramatic not fused ventrally. Arch not dissected, but apparently large (Fig. triangular posterior extension of the zygoma of Oshwea resembles 4E) and supporting the dorsal aedeagal valves. no other acridid we know, and certainly no other Catantopine that Endophallus not dissected (to preserve the unique specimen). we have examined. On the other hand, no other subfamily currently Anterior apodemes laterally flattened, sharply divergent in dorsal seems more appropriate to contain Oshwea. view (Fig. 4D).Gonopore processes long and spatulate with rounded The distance from Oshwe to the Mabira is approximately 1750 tips. Details of flexure not visible, ventral aedeagal valves upwardly km in a straight line. However, the intervening space was almost ° slanted at nearly 45 to the horizontal. entirely occupied by equatorial wet forest until recently, so dispersal of the species over this distance need not surprise, and there are Discussion many other acridid genera among the African forest fauna (e.g., Abisares, Chirista, Pterotiltus) which are similarly distributed. The Both Dirsh (1965,1970) and the Orthoptera Species File online previous lack of records, however, seems to show that it is very (OSF on-line) ( Fades et al. 2013) place Oshwea in the Catantopinae. uncommon throughout its large range. One of us (Rowell) has col There is however no exclusive diagnosis of this subfamily, which lected in Ugandan forests, including the forests on the Congolese historically has been used as a depository for forms that do not fit border, intermittently since the 1960s, while Akite has worked the readily into other established categories (see e.g., remarks in Dirsh Mabira and some other Bugandan forests intensively over recent 1961: 409) and despite the modern removal of its former New years. Neither of us has ever encountered Oshwea before. World components it still contains a great variety of rather disparate

Journal of Orthoptera Research 2013, 22(1)

This content downloaded from 196.43.134.183 on Fri, 22 May 2015 09:53:50 UTC All use subject to JSTOR Terms and Conditions 48 p. akite and c.h.f. rowell

Table 1. Dimensions of male Oshivea dubiosa. Abbreviations of characters: L (length from fastigium to tip of subgenital plate). F (length of hind femur). FD ( Depth of femur: the maximum width of the hind femur). Tib ( length of femur, measured in the folded position). Tl, T2, T3 (lengths of the three tarsal segments, with T1 being the most proximal). T1-T3 (total length of the three tarsal segments). E (length of elytron), Ant (length of antenna), IOS (interocular space: the minimal distance between the compound eyes when seen in dorsal view). P (length of the pronotum in the dorsal midline). H + PN (head plus pronotum: their combined lengths).

Specimen no. 2013012 character mm ratios L 18.99 F 12.05 FD 2.65 Tib 11.02 T1 1.62 Tl/Tl-3 0.34 foot formula T2 0.95 T2/T1-3 0.20 foot formula T3 2.19 T3/T1-3 0.46 foot formula Tl-3 4.76 T1-3/T1-3 1.00 foot formula E 11.26 Ant 11.11 IOS 0.34 P 3.78

prozona 2.1 metazona 1.63

Scape width 0.45 H+PN 6.43

IOS/scape 0.76 T1-3/F 0.40 Size of hind foot Tl-3/tib 0.43 Size of hind foot F/FD 4.55 Femoral proportions ANT/H+PN 1.73 Relative length of antenna

References

Dirsh V.M. 1961. A preliminary revision of the families and subfamilies of Acridoidea (Orthoptera, Insecta). Bulletin of the British Museum of Natural History (Entomology) 10: 349-419. Dirsh V.M. 1965. The African genera of Acridoidea. Anti-Locust Research Centre and Cambridge University Press, London, 579 pp. Dirsh V.M. 1970. Acridoidea ofthe Congo (Orthoptera). Ann. Mus. roy. Afr. centr., Tervuren, Serie IN-8", sci. zool. 182: 605 pp. Eades D.C., Otte D., Cigliano M.M., Braun H. 2013. (Accessed). Orthoptera Species File. Version 5.0/5.0. http://Orthoptera.SpeciesFile.org. Ramme W. 1929. Afrikanische Acrididae. Revisionen und Beschreibungen wenig bekannter und neuer Gattungen und Arten. Mitt. Zool. Mus. Berlin 15: 247-492, 14 plates, 106 Figs.

Journal of Orthoptera Research 2013, 22(1)

This content downloaded from 196.43.134.183 on Fri, 22 May 2015 09:53:50 UTC All use subject to JSTOR Terms and Conditions P. AKITE AND C.H.F. ROWELL 49

Fig. 5. Oshwea dubiosa Ramme 1929, in the wild. Photo P. Akite. For color version, see Plate IV.

Journal of Orthoptera Research 2013, 22(1)

This content downloaded from 196.43.134.183 on Fri, 22 May 2015 09:53:50 UTC All use subject to JSTOR Terms and Conditions PLATE IV

P. AKITE AND C.H.F. ROWELL, Oshwea dubiosa rediscovered in Uganda, p. 45.

Fig. 1. Oshwea dubiosa Ramme 1929. Male habitus.

Fig. 5. Oshwea dubiosa Ramme 1929, in the wild. Photo P. Akite.

Journal of Orthoptera Research 2013, 22(1)

This content downloaded from 196.43.134.183 on Fri, 22 May 2015 09:53:50 UTC All use subject to JSTOR Terms and Conditions

Powered by TCPDF (www.tcpdf.org) Appendix A1. Butterflies recorded across the three forests (Numbers represent abundance, DRF=DRY FOREST, SWF= SWAMP FOREST, MIG = MIXED GARDEN, MAF = MATURE FOREST, SEF = SECONDARY FOREST, CAP = CARDAMOM PLANTATION, COP = COFFEE PLANTION) Zika Mabira Mpanga

Species Ecotype DRF SWF MIG MAF SEF CAP COP MIG MAF SEF MIG Nymphalidae Acraea alcinoe F 2 2 Acraea egina W 2 Acraea epaea F 1 29 14 1 4 Acraea insignis f. 4 Acraea leucographa F 5 5 Acraea macaria F 2 Acraea macarista F 4 2 4 Acraea oncaea f. 1 Acraea poggei F 1 Acraea pseudegina W 30 55 Acraea tellus F 1 Acraea zetes W 1 3 2 3 2 Actinote acerata W 21 50 152 126 Actinote alcipioides F 9 2 Actinote Alicia W 1 1 3 Actinote althoffi F 9 14 2 Actinote aurivilli F 22 13 1 2 8 7 Actinote encedon W 2 6 Actinote iturina F 1 Actinote jodutta F 39 18 1 20 3 8 Actinote johnstoni f. 5 1 Actinote lycoa F 3 5 6 6 3 Actinote orinata F 4 Actinote peneleos F 5 1 5 Actinote Penelope F 8 2 37 16 3 2 Actinote pharsalus f. 12 1 11 Actinote quirina F 12 14 1 1

Actinote quirinalis F 1 42 8 Actinote parrhasia F 12 8 1 2 Actinote semivitrea F 1 1 Actinote serena W 8 53 3 6 153 Actinote sotikensis F 1 1 albimaculata F 3 1 3 4 1 Amauris Hecate F 1 Amauris niavius W 3 6 2 2 10 16 5 3 20 1 Amauris oscarus F 1 1 Amauris tartarea f. 11 22 11 3 12 3 11 2 8 Antanartia delius F 7 2 Apaturopsis cleochares F 9 7 3 2 4 21 8 1 Ariadne albifascia F 5 1 Ariadne enotrea F 6 1 1 5 5 11 9 14 2 3 Ariadne pagenstecheri F 1 Aterica galena F 26 25 8 3 2 4 Bebearia absolon F 6 Bebearia carshena F 3 27 Bebearia cocalia f. 11 18 8 4 1 39 14 Bicyclus angulosa O 15 3 1 Bicyclus anynana O 144 7 1 Bicyclus auricrudus F 5 1 4 4 77 1 61 18 10 Bicyclus buea F 1 2 81 24 16 Bicyclus campinus f. 2 2 8 Bicyclus ena O 1 Bicyclus funebris F 40 12 11 3 3 24 3 57 2 11 Bicyclus golo F 1 1 253 129 25 Bicyclus graueri F 7 8 34 17 1 Bicyclus iccius F 137 17 Bicyclus ignobilis F 1 1 1 Bicyclus istaris f. 11 1 Bicyclus jefferyi f. 1 21 4 19 Bicyclus mandanes F 2 15 15 43 34 5 2

Bicyclus mesogena F 1 8 1 45 43 Bicyclus mollitia F 789 22 5 2381 517 3 Bicyclus safitza W 156 270 86 1 9 12 1 34 Bicyclus sambulos F 1 30 17 6 112 42 Bicyclus sanaos F 11 8 12 2 3 1 Bicyclus sandace F 2 34 1 1 3 11 1 Bicyclus saussurei F 538 383 120 Bicyclus sebetus F 7 Bicyclus smithi F 3 88 84 372 206 37 1 Bicyclus sophrosyne f. 6 21 4 1 6 101 26 4 5 Bicyclus uniformis F 622 21 10 1 1 Bicyclus vulgaris W 433 205 192 6 15 64 19 16 1 147 Bicyclus xenoides F 2 1 34 Bicyclus zinebi F 51 6 Byblia anvatara M 12 20 22 Byblia ilithya O 6 Catuna crithea F 120 242 96 4 204 85 Charaxes ameliae f. 1 2 2 Charaxes anticlea f. 8 1 1 Charaxes bipunctatus F 27 13 1 19 11 16 3 2 Charaxes boueti W 36 3 Charaxes brutus f. 4 4 1 22 6 1 20 Charaxes candiope W 2 2 2 2 12 6 1 W 18 6 1 Charaxes catachrous F 1 23 2 2 Charaxes cedreatis F 1 2 1 Charaxes Cynthia F 1 86 58 40 1 1 Charaxes etesipe f. 1 1 5 1 3 Charaxes ethalion W 4 Charaxes etheocles F 5 Charaxes eudoxus f. 2 2 Charaxes eupale F 12 Charaxes fulvescens F 30 27 2 124 12 2 Charaxes jasius O 1 3

Charaxes lactetinctus O 1 1 1 Charaxes lucretius F 1 6 2 27 Charaxes numenes f. 7 9 2 8 5 1 1 32 11 1 Charaxes paphianus F 4 3 1 Charaxes pleione f. 1 1 9 6 4 1 Charaxes pollux f. 2 4 1 Charaxes porthos F 1 Charaxes protoclea f. 1 1 1 1 1 6 1 3 Charaxes pythodoris f. 1 Charaxes tiridates F 30 24 7 24 14 61 6 7 Charaxes varanes W 3 3 7 1 27 20 4 Charaxes virilise F 7 7 2 6 3 4 1 5 Charaxes zelica F 1 Charaxes zingha F 9 3 2 3 59 11 1 Cymothoe caenis F 3 Cymothoe herminia F 4 207 9 1 39 26 Cynandra opis F 34 Camillus F 1 8 5 1 1 9 9 3 Danaus chrysippus M 9 29 26 19 Elymnias bammakoo F 9 1 1 5 1 Euphaedra alacris F 11 5 Euphaedra eleus F 19 1 2 Euphaedra harpalyce F 18 1 1 1 Euphaedra medon F 45 76 44 1 1 90 57 Euphaedra paradoxa F 1 Euphaedra preussi F 201 24 4 15 Euphaedra rex F 63 13 3 Euphaedra ruspina F 1 1 9 12 Euphaedra uganda F 1 4 7 Euphaedra zaddachii F 1 Euptera elabontas F 28 7 7 1 Euriphene ribensis F 6 4 5 4 Euryphura chalcis F 1 Eurytela dryope W 2 3 2 1 2 11 9 1 1 9 4

Eurytela hiarbas f. 29 21 1 9 4 10 6 5 47 2 2 Euxanthe crossleyi F 4 Euxanthe crossleyi F 1 1 1 Euxanthe eurinome F 4 1 1 2 Euxanthe trajanus F 50 5 Gnophodes betsimena F 14 24 9 5 34 17 33 2 9 Gnophodes chelys F 13 2 1 44 15 1 183 74 Graphium leonidas M 2 2 Harma theobene F 152 61 2 1 1 9 2 Henotesia peitho W 298 36 1 Henotesia perspicua O 7 6 Hypolimnas anthedon F 4 4 5 1 1 2 Hypolimnas dinarcha F 7 1 3 1 Hypolimnas misippus M 1 9 4 5 Hypolimnas monteironis F 1 5 4 2 6 2 Hypolimnas salmacis F 13 5 1 1 2 Junonia chorimene O 7 36 10 4 33 Junonia hierta M 1 Junonia oenone W 12 34 17 10 Junonia orithya M 3 Junonia Sophia W 1 31 10 7 109 Junonia stygia f. 1 10 44 1 1 Junonia terea W 1 1 35 1 27 15 13 19 Junonia westermanni F 1 2 3 2 2 2 Lachnoptera anticlia f. 91 3 3 1 Lachnoptera ayresii f. 20 Libythea labdaca M 10 6 5 4 3 2 5 Melanitis leda W 12 2 7 6 2 3 50 11 5 4 Mesoxantha ethosea F 1 18 1 1 6 17 Neptidopsis ophione f. 1 1 2 3 3 4 10 7 7 Neptis conspicua F 2 1 1 2 Neptis melicerta F 7 3 1 55 55 6 2 21 3 Neptis metella f. 1 73 85 23 1 1 29 19 5

Neptis nemetes f. 7 2 24 9 7 1 1 10 6 2 Neptis nicobule W 1 Neptis nicomedes f. 1 31 45 3 1 Neptis nicoteles F 1 Neptis saclava W 10 6 4 2 13 8 6 3 1 Neptis serena W 2 9 1 2 Palla ussheri F 4 1 Phalanta eurytis M 12 2 16 15 20 2 4 Phalanta phalantha M 7 2 1 Precis ceryne S 12 Precis Octavia W 1 2 1 Precis tegula f. 1 Pseudacraea boisduvali f. 1 1 Pseudacraea dolomena F 59 Pseudacraea eurytus F 30 2 10 11 Pseudacraea lucretia f. 46 3 16 31 5 18 8 Pseudargynnis hegemone f. 2 Pseudoneptis bugandensis F 104 70 12 2 1 23 8 Protogoniomorpha cacti F 1 Protogoniomorpha parhassus f. 2 1 1 3 1 4 Sevenia boisduvali M 230 5 22 1 1 31 5 20 1 Sevenia garega M 3 30 17 1 103 16 22 48 Sevenia occidentalium M 50 23 5 1 4 29 3 45 4 Tirumala formosa f. 3 1 Tirumala petiverana M 1 Vanessula milca f. 1 1 10 1 Ypthima albida f. 2 2 20 6 3 22 Ypthima asterope O 9 16 1 1 13

6

Ypthima doleta W 1 1

Riodinidae Abisara neavei F 15 162 4 3 4 67 7 7

Lycaenidae Anthene indefinita O 1 30 Anthene larydas F 6 3 1 16 11 Anthene ligures F 1 Anthene lunulata W 1 Anthene rubricinctus F 1 Anthene schoutedeni F 5 7 6 2 Argyrocheila inundifera F 1 Aslauga marshalli f. 1 1 Azanus jesous M 1 2 Azanus natalensis W 2 Cacyreus lingeus f. 5 2 3 1 Citrinophila erastus F 1 Cupidestes ysobelae f. 1 Eicochrysops Hippocrates W 2 1 Epitola orientalis F 1 1 Eresina pseudofusca F 1 Euchrysops malathana O 2 10 11 7 Euchrysops osiris W 2 Falcuna orientalis F 2 Hewitsonia intermedia F 1 1 Hypolycaena antifaunus F 5 8 1 Hypolycaena hatita F 10 3 1 2 2 4 Hypolycaena philippus W 1 1 2 Iolaus silanus W 2 Larinopoda tera F 5 26 15 1 4 14 7

Leptotes pirithous M 3 7 1 9 Liptena xanthostola F 6 3 Megalopalpus zymna F 189 128 2 34 7 Micropentila victoriae F 1 Mimacraea krausei F 1 Oboronia punctatus F 5 4 Ornipholidotos overlaeti F 13 Pentila pauli f. 4 17 3 2 Pentila tachyroides F 2 43 53 9 1 Ptelina carnuta O 1 4 4 1 2 Spalgis lemolea f. 1 1 Telipna sanguinea F 11 8 Tetrarhanis ilma F 6 24 9 4 1 Thermoniphas plurilimbata F 1 Triclema nigeriae f. 1 Uranothauma falkensteini W 1 3 1 1 Uranothauma heritsia F 1 Zizeeria knysna W 11 12 59 83 37 Zizina antanossa W 6 11 3 15 11 Zizula hylax W 3 3 36

Hesperiidae Acleros ploetzi f. 1 1 2 1 7 2 2 Andronymus neander M 1 1 Ankola fan F 8 4 Borbo borbonica M 1 1 Borbo ferniginea f. 1 Borbo gemella W 1 Borbo kaka F 2 Borbo micans S 1 Caenides xychus F 6 8

Celaenorrhinus galenus F 4 6 6 2 11 Celaenorrhinus proxima F 1 1 1 Ceratrichia flava F 5 4 1 Ceratrichia hollandi F 27 12 2 5 Ceratrichia mabirensis F 4 Coeliades chalybe F 1 1 Coeliades forestan W 3 1 1 1 Coeliades libeon M 1 Eagris leucetia f. 1 1 lugens W 2 12 7 9 4 10 Eretis melania W 1 Eretis umbra O 4 Gamia shelleyi F 1 3 Gegenes hottentota O 15 Gegenes pumilio W 1 2 Gorgyra aretina f. 4 Gorgyra bibulus W 1 Gorgyra bina f. 1 Gorgyra minima F 1 2 Metisella midas S 2 Monza alberti F 4 Osmodes thora F 16 6 6 4 3 Paracleros biguttulus f. 2 2 2 1 incerta F 13 1 1 6 1 8 Pardaleodes sator F 2 Pardaleodes tibullus F 1 Pelopidas mathias M 1 Pteroteinon caucaenira O 1 Sarangesa bouvieri F 1 1 Sarangesa maculata O 1 2 Spialia spio O 1 1 1 4 2 9

Tagiades flesus F 10 6 1 7 8 Xanthodisca vibius F 1 Zenonia zeno f. 1

Pieridae Appias epaphia M 2 7 6 2 Appias sabina F 14 Appias Sylvia F 6 1 Belenois aurota M 2 1 3 2 Belenois calypso F 19 1 Belenois creona M 2 3 1 15 1 Belenois solilucis f. 5 2 1 3 1 2 Belenois subeida f. 2 Belenois theora f. 70 2 2 1 Belenois thysa f. 3 3 Catopsilia florella M 3 22 15 78 Colotis elgonensis F 6 orbona W 1 2 Dixeia pigea W 3 1 Eronia cleodora O 2 Eurema brigitta M 7 3 28 Eurema desjardinsi W 2 Eurema floricola F 6 Eurema hapale S 1 Eurema hecabe M 11 15 13 1 3 25 69 22 14 15 Eurema regularis W 1 1 Eurema senegalensis F 53 17 1 1 28 7 4 Leptosia alcesta W 1 2 Leptosia hybrida F 99 96 4 1 Leptosia nupta F 13 1 14 2 14 72 160 164 124 34 204 Leptosia wigginsi F 35 64 142 16 1 13 6 Mylothris chloris W 1 1 1 Mylothris continua F 34 1 3 8 12 3 9 Mylothris hilaria F 6 1 10

Mylothris kiwuensis F 17 1 3 Mylothris rubricosta S 1 1 Mylothris yulei f. 5 Nepheronia argia F 97 22 15 21 1 48 17 5 Nepheronia pharis F 2 1 1 3 Nepheronia thalassina f. 112 7 15 15 10 55 Pontia helice M 4 93 F 35

Papilionidae Papilio chrapkowskoides f. 12 13 7 1 20 16 9 7 5 Papilio cynorta F 6 50 25 10 1 2 13 7 Papilio dardanus W 39 12 9 129 55 40 8 4 58 35 1 Papilio demodocus M 2 1 7 31 17 4 Papilio hesperus F 1 Papilio lormieri F 8 10 5 3 1 3 3 Papilio mechowi F 1 Papilio nireus f. 1 1 Papilio phorcas F 15 166 70 12 18 9 95 49 Papilio zoroastres f. 1 1

11

Appendix A2. Silk moths (Emperors) recorded across the three forests (Numbers represent abundance, DRF=DRY FOREST, SWF= SWAMP FOREST, MIG = MIXED GARDEN, MAF = MATURE FOREST, SEF = SECONDARY FOREST, CAP = CARDAMOM PLANTATION, COP = COFFEE PLANTION)

Zika Mabira Mpanga

2010s 2010s 2010s Ecotype 1970s 1990s 1990s 1990s

Species DRF SWF MIG MAF SEF CAP COP MIG MAF SEF MIG Athletes ethra F 4 12 Athletes nyanzae f. 77 44 14 9 48 8 11 3 2 4 Aurivillius aratus f. 52 15 11 3 10 8 Aurivillius triramis f. 81 11 1 11 68 12 7 8 6 4 2 Bunaea alcinoe f. 64 26 3 6 11 7 14 4 11 10 2 3 4 1 Bunaeopsis jacksoni G 2 2 4 3 1 Bunaeopsis licharbas f. 23 1 6 4 8 16 6 3 5 1 3 4 2 3 Bunaeopsis oubie zaddachi G 4 4 12 12 3 5 3 1 Cirina forda f. 12 5 10 14 3 865 8 18 4 13 11 6 14 3 10 Decachorda rosea G 4 8 Epiphora albida F 48 10 2 Epiphora rectifascia F 9 11 2 2 Epiphora vacuna f. 3 92 9 1 Goodia lunata F 4 5 12 4 6 1 Goodia nubilata/oxytela F 88 4 31 7 2 2 Gynanisa festa f. 3 6 3 7 7 Gynanisa jama F 1 Holocerina angulata f. 8 6 5 5 25 6 6 7 6 3 2 Holocerina smilax f. 4 3 5 7 6 5 9 12

Imbrasia alopia F 6 1 29 1 Imbrasia anna W 28 10 29 1 5 26 28 21 19 9 4 4 3 Imbrasia anthina F 102 8 324 8 Imbrasia conradsi F 92 33 Imbrasia eblis F 12 5 4 Imbrasia emini f. 35 12 4 11 7 3 1 1 1 Imbrasia epimethea F 287 107 3 3 177 8 3 3 11 2 3 Imbrasia godarti F 2 Imbrasia hecate f. 2 Imbrasia jamesoni F 35 2 26 7 Imbrasia oyemensis F 7 2 1 Imbrasia petiveri/dione f. 164 70 4 2 248 14 11 7 9 3 5 Imbrasia rectilineata f. 4 Imbrasia staudingeri F 3 3 8 3 Imbrasia truncata F 39 88 2 1 1 Lobobunaea acetes F 5 1 64 5 2 3 6 1 Lobobunaea acetes F 77 Lobobunaea ansorgei F 3 Lobobunaea christyi f. 1 7 5 5 3 4 Lobobunaea goodii f. 6 2 7 6 5 5 2 2 6 1 Lobobunaea phaedusa f. 89 15 2 14 6 5 4 1 1 1 Ludia delegorguei F 2 Ludia hansali f. 3 1 2 7 2 Ludia orinoptena f. 17 9 5 Ludia orinoptena f. 48 59 26 Micragone .sp F 1 Micragone agathylla F 39 23 8 1 Micragone ansorgei F 1 Micragone nubifera f. 12 3 2 1 3 2 1 7 5 13

Orthogonioptilum adiegatum F 6 2 5 Orthogonioptilum luminosum F 1 1 7 2 1 Orthogonioptilum sp F 3 Orthogonioptilum sp. C. F 1 Orthogonioptilum tristis F 1 13 9 10 1 Orthogonioptilum vestigiatum F 3 5 Pselaphelia flavivitta F 55 8 2 2 17 10 2 9 3 2 Pselaphilia gammifera F 6 3 Pseudaphelia apollinaris/simplex f. 3 1 2 2 1 1 Pseudimbrasia deyrollei f. 4 1 5 7 34 1 8 1 2 5 Pseudobunaea cleopatra F 6 105 5 Pseudobunaea epithyrena f. 27 6 29 2 2 1 2 2 Pseudobunaea tyrrhena f. 67 16 6 1 100 5 6 5 4 2 1 Tagaropsis flavinata f. 9 5 10 3 36 10 11 3 8 1 1 2 Tagaropsis genoviefae/rougeoti F 42 6 7 1 Urota sinope f. 14 7 10 8 4 1 3 6 2 1 3 3 1

14

Appendix A3. Hawk moths recorded across the three forests (Numbers represent abundance, DRF=DRY FOREST, SWF= SWAMP FOREST, MIG = MIXED GARDEN, MAF = MATURE FOREST, SEF = SECONDARY FOREST, CAP = CARDAMOM PLANTATION, COP = COFFEE PLANTION) Zika Mabira Mpanga Ecotype 1990s 2010s 1990s 2010s 1990s 2010s

Species DRF SWF MIG MAF SEF CAP COP MIG MAF SEF MIG Acanthosphinx guessfeldti F 11 2 27 8 1 7 2 Acherontia atropos W 13 12 14 11 11 28 38 7 13 16 7 27 16 19 Andriasa contraria W 1 13 7 28 8 3 10 1 2 5 Antinephele achlora F 4 Antinephele anomala F 1 1 2 Antinephele maculifera f. 1 11 3 Antinephele marcida F 2 5 9 3 Atemnora westermanni f. 4 1 Basiothia aureata f. 14 6 4 1 1 3 Basiothia charis W 4 7 1 1 1 Basiothia medea W 1 9 4 1 7 1 Celerio lineata W 1 4 9 11 1 18 31 7 23 3 1 19 3 2 Centroctena rutherfordi F 10 5 5 5 15 2 4 2 3 2 4 3 Cephonodes hylas W 3 Chloroclanis virescens f. 4 2 2 3 210 7 1 6 8 4 Coelonia mauritii W 126 14 5 189 28 10 30 12 12 9 Deilephila nerii W 34 4 2 7 43 1 5 7 5 20 7 3 3 Dovania poecila F 78 12 4 6 Euchloron megaera W 60 9 2 5 127 11 4 3 6 6 3 15

Falcatula cymatodes F 1 Falcatula falcata W 2 4 2 3 3 6 3 4 Herse convolvuli W 76 14 18 15 307 8 110 10 5 Hippotion aporodes F 1 4 157 19 3 7 8 47 Hippotion balsaminae W 46 7 8 1 3 Hippotion celerio W 25 9 3 2 53 6 31 2 6 5 6 7 2 4 Hippotion eson W 156 3 34 20 2 25 5 6 2 Hippotion irregularis F 24 6 5 1 Hippotion osiris W 8 4 8 19 17 10 2 Lophostethus demolini W 4 105 Macropoliana natalensis f. 1 1 40 9 11 5 Neopolyptychus serrator f. 61 90 22 5 9 2 7 3 Nephele accentifera W 61 2 2 1 6 2 4 2 3 4 2 Nephele aequivalens f. 45 2 93 17 40 8 4 Nephele bipartita f. 15 2 3 7 2 Nephele comma W 9 70 3 10 5 2 Nephele discifera F 27 Nephele funebris W 23 122 1 9 Nephele maculosa F 1 Nephele oenopion F 1 11 1 Nephele peneus f. 5 6 8 7 4 1 1 4 1 Nephele rectangulata F 8 9 4 Nephele rosae f. 55 7 2 8 2 2 2 Nephele vau W 1 Platysphinx constrigilis F 9 1 6 Poliana buchholzi F 1 5 5 2

16

Polyptychoides digitatus F 7 Polyptychus affinis F 211 3 Polyptychus carteri F 2 5 295 6 7 21 Polyptychus orthographus F 9 5 Polyptychus retusus F 26 5 Pseudoclanis postica W 61 11 579 3 7 2 54 7 1 Temnora crenulata F 32 4 1 1 1 3 Temnora elisabethae F 16 2 22 2 Temnora eranga F 1 Temnora fumosa W 58 9 13 7 60 29 9 4 18 3 19 2 11 Temnora funebris F 1 1 Temnora hollandi F 1 Temnora iapygoides F 6 2 3 1 1 10 1 Temnora sardanus f. 45 4 1 1 Temnora scitula F 2 32 5 Temnora spiritus F 3 Temnora subapicalis F 1 Theretra capensis G 2 5 7 2 13 1 4 Theretra jugurtha f. 10 2 31 3 8 2 1 7 8 Theretra orpheus f. 1 1 Xanthopan morgani f. 17 4 3 4 2 2

17

Appendix A4. Grasshoppers recorded across the three forests (Numbers represent abundance, DRF=DRY FOREST, SWF= SWAMP FOREST, MIG = MIXED GARDEN, MAF = MATURE FOREST, SEF = SECONDARY FOREST, CAP = CARDAMOM PLANTATION, COP = COFFEE PLANTION)

Zika Mabira Mpanga Species Ecotype DRF SWF MIG MAF SEF CAP COP MIG MAF SEF MIG Abisares viridipennis F 4 9 6 3 4 Acanthacris ruficornis ruficornis f. 6 4 3 13 Acorypha modesta F 4 Acrida bicolor G 24 Afromastax rubricosta f. 18 4 Afroxyrrhepes procera F 2 Atractomorpha acutipennis acutipennis F 13 Atractomorpha acutipennis gerstaeckeri F 3 Auloserpusia poecila F 118 28 5 18 103 80 Auloserpusia synpicta F 29 55 Calephorus compressicornis F 1 Catantopsilus grammicus f. 2 Chirista compta f. 1 3 10 8 Chrotogonus sp f. 2 Coryphosima stenoptera F 4 3 Cyphocerastis falcifera F 3 3 3 31 4 Cyrtacanthacris aeruginosa aeruginosa F 1 Euschmidtia congana F 4 2 Eyprepocnemis plorans ibandana f. 3 Eyprepocnemis plorans plorans f. 4 Gemeneta terrea F 8 14 2 32 37 Gowdeya picta f. 14 4 Gymnobothrus lineaalba G 2 2 4 18

Gymnobothrus temporalis G 5 1 8 Hadrolecocatantops kissenjianus F 4 4 2 3 1 5 5 Hadrolecocatantops quadratus F 6 4 1 6 5 Heteracris guineensis f. 1 5 3 3 5 8 Heteracris vinacea f. 2 Heteropternis thoracica G 3 Humbe tenuicornis G 2 1 10 3 Kassongia flavovittata F 1 Odontomelus kwidschwianus f. 115 2 1 6 Oshwea dubiosa 1 Oxya hyla f. 2 Oxyaeida carli F 1 3 1 37 19 Paracoptacra cauta f. 14 9 4 162 39 37 14 14 74 15 9 Parapropacris rhodoptera f. 1 2 2 Pezocatantops lobipennis F 1 Phryganomelus phalangidus F 2 3 52 9 Pteroperina steini F 5 Pterotiltus hollisi F 2 2 30 17 13 13 3 10 30 Serpusia nr. lemarineli F 7 101 36 3 132 57 Spathosternum pygmaeum F 3 Stenocrobylus diversicornis F 1 1 1 Taphronota calliparea dimidiata f. 11 8 17 Tetrigidae sp f. 7 7 5 5 3 Thamithericles birunga F 1 6 Trichocatantops digitatus f. 4 Trilophidia conturbata G 3 3 2 Usambilla sagonai sagonai F 45 56

19