Our study focused on the mature urban bushmeat market of Sekondi-Takoradi (hereafter Takoradi), Ghana's third largest city, located in the Upper Guinea Forest global biodiversity hotspot (Myers et al. 2000). Takoradi has several centuries of recorded settlement and has grown rapidly over the last 100 years through the development of gold mining, railways and harbour facilities. This economic growth has also transformed the surrounding hinterland into an agricultural farmbush matrix, consisting of a mosaic of plantations, mixed bush fallow (predominantly cocoa, coconut and oil palm) and remnant tropical forest (including secondary forest). Bushmeat has been documented as part of the local diet for centuries (Grubb et al. 1998). Data were collected in January–February 2000, a period representative of annual bushmeat trading: it avoided both the peak hunting season (May–July inclusive: Holbech 1998) and the closed season (August–November inclusive), and was described consistently as a typical month by the traders interviewed.
Data were collected using a combination of direct observation and semistructured interviews (following Magrath 1992), from a representative sample of all actors in the market (farmer hunters, commercial hunters, wholesalers, market traders and chopbars). These data describe a total of 2430 bushmeat transactions reported by 70 different actors encompassing 16 different taxa. Data collected for each transaction included: taxon identity, weight and condition (fresh/smoked), identity of purchaser (e.g. wholesaler, the public) and sale price (in Ghanaian cedis, ¢). For a subset of transactions, additional information included the supplier of the meat (n = 1745 transactions) and the capture location (n = 438 transactions between hunters and market traders). Distances between capture location and the city market were calculated on the basis of road distance (the distance travelled by the hunter). Transaction weight was either measured directly (n = 1094) or reported by the seller (n = 1362): both methods produced the same mean carcass weights and were combined for this analysis. Supplementary data were collected on actor perceptions of changes in bushmeat availability (n = 12 hunters) and the weight, price and type of fish and domestic meat sold in the market (n = 1138). All our informants were open and relaxed, and their responses showed a high degree of consistency when subjected to direct observation and when cross-referenced to other actors. None of the recorded trade was illegal, although illegal trade usually takes place openly where it does occur in Ghana (Ntiamoa-Baidu 1998). Further information on the Takoradi bushmeat trade and details of data collection are given in Cowlishaw et al. (2005) and Mendelson et al. (2003).
Our analyses focused on the 10 terrestrial mammals in the trade comprising 84% of the total biomass sold. All analyses involving sales price per kilogram were based on the market value of smoked meat (85% of all retail sales). Historical data describing 1963 Takoradi market prices are taken from Asibey (1966). To investigate changes in the real value of bushmeat (inflation-corrected), prices were adjusted according to the Consumer Price Index (CPI) taken from IMF (1980, 2001). The CPI measures the cost of a standardized basket of market goods and is the most widely used measure of inflation. The prices of bushmeat and other types of meat in 2000 were determined by averaging the mean sales price across traders. Our expectations about price changes (predictions 1·3 and 1·4) assume that bushmeat is a luxury, or superior, good. Otherwise, it would not become more expensive when scarce because consumers would switch to less expensive alternatives. Although we were unable to explore the precise elasticity of bushmeat consumption in Takoradi (data describing local household income and expenditure on bushmeat are unavailable), our research indicates that bushmeat is a luxury item. Bushmeat was more expensive per kg than fish, beef or mutton in 1963 (Asibey 1966) and continued to be in 2000 (see below), and during our study consumers in Takoradi preferred to eat bushmeat over these alternatives when they could afford it (Mendelson et al. 2003).
Determination of the current annual extraction of the 10 taxa across the Takoradi catchment took two steps. In the first step, we calculated species extraction for the city itself, in three discrete stages. (i) To determine the number of urban retailers, the number of market stalls, nm, was recorded through a complete census of market places while the number of chopbars, nc, was estimated by extrapolation from nm and the average per capita bushmeat biomass purchased by each chopbar from market traders (Bc,m) and sold by market traders to each chopbar (Bm,c) (calculated from 187 and 375 independently sampled transactions, respectively). This extrapolation was based on the assumption that nc · Bc,m = nm · Bm,c · (ii) To determine the annual urban sales by these traders, we multiplied nm and nc by the average number of species carcasses we observed each selling (per capita) in a typical 1-month period, and multiplied this by 12. (iii) To estimate annual bushmeat extraction for all urban sales, we accounted for the additional informal trade between hunters and consumers, on the basis that such sales may comprise up to 18% of urban sales (Ntiamoa-Baidu 1998). In the second step, we calculated total extraction from the entire catchment in two discrete stages. (i) We first accounted for rural sales of bushmeat, on the basis that only 17% of sales by farmer hunters are urban sales (Falconer 1992). (ii) We then accounted for bushmeat that hunters eat themselves or give away, on the basis that farmer hunters sell either a consistent fraction equal to 69% of all captures (Method 1) (Falconer 1992) or an inconsistent fraction of captures ranging from 17% to 100% (median 66%) depending on the species in question (Method 2) (Ntiamoa-Baidu 1998). These calculations thus produced two estimates of the annual extraction for each species. These estimates indicate that the formal urban bushmeat sales (160 000 kg fresh mass) are only 14% of total extraction (1130 tonnes: mean, Methods 1 and 2) for these taxa in the Takoradi catchment.
The two estimates of annual extraction for each species were compared to the estimated sustainable production for each species according to two sustainability indices (Milner-Gulland & Akçakaya 2001). These indices calculate a species sustainable production, P, according to its population density (either at its current population size, N, or at carrying capacity, K), its intrinsic rate of population increase, rmax, and its mortality or recovery factor, F. According to the Robinson and Redford algorithm:
In contrast, the US National Marine Fisheries Service algorithm states that:
- PNMFS = 0·5N(rmax − 1)FNMFS
To determine K and N, species population densities, d, were first taken from Fa & Purvis (1997) or estimated through standard allometric relationships (Rowcliffe, Cowlishaw & Long 2003). The catchment area for each species was then defined as the area around Takoradi between its minimum and maximum recorded capture distances. The landscape around Takoradi is characterized by a farmbush matrix: a representative habitat type for all species in the market (habitat preferences from Grubb et al. 1998). Nevertheless, species estimates for K and N within the catchment were set to 0·75d and 0·50d, to account for habitat heterogeneity and historical hunting in the matrix, respectively. To err on the side of caution, we also repeated the analysis with lower values (K = 0·50d, N = 0·25d), but similar results were obtained. Species values of rmax were also estimated by allometry (Rowcliffe et al. 2003). FRR was set at 0·6, 0·4 and 0·2 for very short-lived species (rodents, 0·5–4 kg), short-lived species (small ungulates, 4–14 kg) and long-lived species (bushbuck, 43 kg), respectively (Robinson 2000); FNMFS was set at 0·5 throughout (Milner-Gulland & Akçakaya 2001).
These algorithms are commonly used to determine the maximum sustainable production and thus whether an observed yield is unsustainable. However, they are less reliable at pinpointing a sustainable yield because production is not always maximal (e.g. Robinson 2000). We therefore made conservative estimates of production and sustainability by using modest values of K & N. Stephens et al. (2002) reported recently that PRR may overestimate sustainable yields in social species. However, their conclusions were based on models of marmots, Marmota marmota, which live in larger and more complex social groups than any of the species entering the Takoradi market. In addition, Milner-Gulland & Akçakaya (2001) found that PNMFS outperforms PRR in its ability to correctly predict sustainability. Here we use PRR because it provides an alternative estimate of sustainable production and because it is also the most widely used index for estimating sustainability in the field (thus providing comparability with previous studies). Nevertheless, it should be noted that the PNMFS figures are likely to be the more accurate of the two measures.
Parametric statistical tests were always employed where the data under investigation did not differ from a normal distribution (Kolmogorov–Smirnov tests: P > 0·05). Variables were loge-transformed where this improved the fit to a normal distribution (e.g. body mass). Where the data were skewed and could not be transformed satisfactorily, non-parametric tests were used. Given the inherent patterns of statistical non-independence in these data (e.g. one hunter will be responsible for several transactions, which in turn can involve multiple market traders), we used average species values across all transactions, rather than each individual transaction, as the unit of analysis. All statistical tests were two-tailed.