African Biodiversity: Molecules, Organisms, Ecosystems. Proc. 5th Intern. Symp. Trop. Biol., Museum Koenig, Bonn (BA Huber, BJ Sinclair, K-H Lampe, eds). Springer Verlag. 2005.

FRUIT-FEEDING BUTTERFLY COMMUNITIES OF FOREST ‘ISLANDS’ IN GHANA: SURVEY COMPLETENESS AND CORRELATES OF SPATIAL DIVERSITY

Janice L. Bossart1, Emmanuel Opuni-Frimpong2, Sylvestor Kuudaar3 and Elvis Nkrumah3 1 Rose-Hulman Institute of Technology, ABBE, Terre Haute, IN 47803, USA; E-mail: [email protected] 2 School of Forest Resources and Environmental Science, Michigan Technological University, Houghton, MI 49931, USA 3 Forestry Research Institute of Ghana, Kumasi, Ghana

Abstract: We conducted a year long survey of the fruit-feeding butterfly fauna of sacred forest groves and forest reserves of Ghana. About one-third of all fruit-feeding species recorded for Ghana were trapped, but our total sample had not yet reached the point of species saturation. Rarefied species richness was higher in the larger forest reserves than in the small sacred groves, as predicted by species-area relationship theory. Bebearia and showed the greatest decline in species richness across sites, demonstrating that common, wide- spread species can be vulnerable to fragmentation.

Key words: ; indigenous reserves; species’ vulnerability

1. INTRODUCTION

Estimates are that anywhere from 70 to 90% of Ghana’s original high canopy climax forest has been eliminated (Hawthorne and Abu-Jaum, 1995; Weber et al., 2001). Very little original forest-cover exists outside of demarcated forest reserves and much of this relict habitat is sacred lands that were set aside by indigenous peoples hundreds of years ago and strictly protected via religious sanctions and taboos (Lebbie and Freudenberger, 152 Janice L. Bossart et al.

1996; Ntiamoa-Baidu, 2001). Ghana’s long-protected sacred forest groves were originally embedded within continuous forest cover, but now exist as isolated habitat ‘islands’. These indigenous reserves likely stand as biodiver- sity repositories of forest communities, and stepping stones that help link discrete forest blocks. In highly degraded regions, sacred forest groves constitute the only examples of old growth forest that remain in the country and some harbor the only surviving specimens of plant species associated with disappearing forest subtypes (Hall and Swain, 1981). The long term goal of this project is to identify key predictors of community composition and species persistence and extinction in highly modified landscapes. Butterflies are excellent model systems for this task. They show a wide diversity of relative sensitivities to environmental change, are tightly linked to ecological systems as both primary consumers (herbivores) and food items, are easily collected and identified, and additionally elicit the emotional concern necessary to bring about conservation action in the face of the conflicting socioeconomic priorities that take precedence over nature conservation in developing countries.

2. METHODS

The fruit-feeding butterfly species (which account for approximately 1/3 of the ~730 forest-endemic butterfly species in Ghana; (Larsen, 2001)) were inventoried at two forest reserves and four sacred grove sites located in the moist semi-deciduous forest zone using typical fruit-baited traps. These remnant forests differ greatly in size. The forest reserves, Bobiri and Owabi, are 5000 and 1200 ha., respectively. The sacred groves, Asantemanso, Gyakye, Bonwire, and Kajease, are 259, 11.5, 8, and 6 ha., respectively. Fruit-bait traps were hung in the forest understory, approximately 10 cm above the ground. Traps were baited with mashed, fermented banana, and butterflies were collected approximately 24h later. A total of 61 traps were installed across all sites. Conscious effort was taken to install all traps in similar microhabitats within areas of closed canopy forest. Five traps were hung in each of three discrete areas at Bobiri and Owabi, and in two at Asantemanso, in order to gain a more representative sample from these larger forests. Four traps were hung in each of two separate areas at Kajease and Bonwire, but the distance between sub-areas was small and restricted by the size and irregular shape of these forests. Five traps total were hung at Gyakye. Individual traps within sub-areas were separated from each other by at least 50 m and by no more than 200 m. Trap sampling occurred at regular intervals for one year. Sites were sampled from 18-22 times during the course of the study. Butterflies in Relict Forests of Ghana 153

Figure 1. Rarified species richness curves showing numbers of species accumulated at each site as a function of increasing numbers of individuals trapped through time. Sub-samples (and 95% confidence intervals) from larger total samples were calculated relative to the total sample at each site with a smaller total sample. For example, five sub-samples are shown for Owabi, the site with the largest total sample, each of which corresponds to the sample size at a site with a smaller total sample.

Trap data were pooled for each site-date combination to minimize variance associated with different traps, e.g., due to differential trap attractiveness, annihilation of trap collections by driver ants, etc. Collections from each site were rarefied using EcoSim700 (Gotelli and Entsminger, 2003) to allow for comparison of richness among sites that had total samples of different sizes (Gotelli and Colwell, 2001). Rarefaction is a robust statistical technique that calculates estimates of species richness for sub- samples of a specified size drawn at random from the total sample. Comparable sized sub-samples can then be compared across sites. Rarefaction curves are similar to species accumulation curves but cannot be used to extrapolate estimates of total species richness.

3. RESULTS

Nearly 7000 specimens were collected from our baseline surveys (Fig. 1), representing 78 species and 34% of the total fruit-feeding butterfly fauna known for Ghana (Table 1). Traps were most successful at collecting species with an affinity for the moist forest subtype and species found in all forest subtypes (34 and 51% of those species possible, respectively). Only five species were collected that are viewed as non-specialized on forest habitats. Traps were also more successful at collecting Satyrinae (54% of fruit- 154 Janice L. Bossart et al.

Table1. Sampling completeness. Species are categorized based on habitat affinities and scarcity rankings1 and grouped by subfamily (from Larsen, 2001). The first number in each cell is the number of species trapped during this survey; the second number is the actual number of species in that category. Dashes indicate non-existent habitat-scarcity combinations. Rows were not included in cases where all cells were empty. Numbers in parentheses are species trapped/actual species of each habitat subtype and percentage represented. VC = very common, C = common, NR = not rare, R = rare, and VR = very rare.

Argynninae Limenitinae Charaxinae Nymphalinae Satyrinae Wet Forests (11/54 = 20%) C -- 0/2 ------NR -- 5/11 1/2 0/2 3/9 R -- 2/16 0/7 -- 0/3 VR -- 0/2 ------

Moist Forests (22/64 = 34%) C 0/2 7/12 0/1 -- 4/4 NR -- 3/16 3/6 0/2 3/5 R -- 2/11 0/4 -- -- VR -- 0/1 ------

Dry Forests (5/18 = 26%) C 0/1 ------1/1 NR -- 4/6 0/3 0/3 0/1 R -- 0/1 0/2 -- --

All Forests (35/68 = 51%) VC -- 1/3 -- 2/2 3/3 C -- 16/22 6/10 3/7 1/1 NR -- 1/7 1/5 0/1 1/3 R -- 0/2 0/2 -- --

Ubiquitous & Savannah (5/25 = 20%) VC ------0/1 -- C -- 0/3 1/4 0/4 1/4 NR -- -- 0/1 0/3 3/3 R -- -- 0/2 -- --

Total (78/229) 0/3 41/115 12/49 5/25 20/37 feeding species) relative to other subfamilies (36, 24 and 20% of Limenitinae, Charaxinae, and Nymphalinae fruit-feeders, respectively). Total samples collected from each site had not reached species saturation when trapping was terminated. New species continued to accumulate as ad