Trapper profiles and strategies: insights into sustainability from hunter behaviour

Abstract

Hunters are the critical link between demand and supply of bushmeat. An understanding of the incentives that drive hunter behaviour might thus help to predict the impacts of hunting and inform management of bushmeat hunting systems. However, hunter behaviour has been generally under-represented in studies of exploitation, in particular trapper behaviour, despite the fact that trapping is the most common form of hunting in central Africa. We collected data on hunter profiles and measures of catch and effort over 15 months in the Monte Mitra area of continental Equatorial Guinea, through interviews, hunter follows and an offtake survey. Younger trappers, and those not born in the village, were found to expend the greatest trapping effort. Trappers operated under three distinct strategies, reflecting different levels of effort and impact: low-impact village trappers, medium-impact forest trappers and high-impact forest trappers. Among different measures of effort, time expended and distance travelled were found to be less important in predicting trapping success than the number of effective traps, a measure that incorporates trap age. Regular checking of traps was found to be important in reducing wastage and therefore increases trapping success. Trapping is currently the main hunting method in Monte Mitra, due to lower barriers to entry and higher profits compared with gun hunting, but increasing affordability and availability of guns and cartridges warns of a possible future switch to gun hunting in the area, which is likely to have adverse impacts on vulnerable species, particularly arboreal primates. An understanding of the influence of a hunter's profile on hunting effort and success enables a prediction of the impacts of socioeconomic changes on wildlife populations and management actions to improve hunting sustainability.

Introduction

Hunting and trade of bushmeat (wild meat) is believed to be increasingly unsustainable throughout tropical forests (e.g. Robinson & Bennett, 2000), including central Africa (e.g. Bowen-Jones & Pendry, 1999). As hunters are the critical link between demand and supply (Bowen-Jones, Brown & Robinson, 2002; Cowlishaw, Mendelson & Rowcliffe, 2005), an understanding of hunter incentives, strategies and decision making is needed to ensure sustainable levels of hunting, for two reasons. First, a hunter's personal profile (e.g. his background, skills, income security, number of dependants and physical ability) exerts an influence on the type of hunting he practises and the overall effort he expends. Second, the form this effort takes, such as how far to go, for how long and what sort of prey to target, together with the volume and type of prey he catches, can help to predict under what conditions hunting effort may change and hence the overall impacts on a harvesting system (Milner-Gulland, 2006). Unfortunately, the role of hunter behaviour in determining patterns of sustainability has been largely neglected in studies of exploitation (Sutherland & Gill, 2001).

Hunting can be broadly categorized as active pursuit hunting, or more passive trapping or snaring (Rowcliffe, Cowlishaw & Long, 2003). Most studies of hunting behaviour have concentrated on pursuit hunting, either by traditional methods such as bow and arrow or blowpipe (e.g. Kuchikura, 1988) or by modern shotguns (e.g. Alvard, 1993). Pursuit hunting allows a much greater degree of prey choice, with studies demonstrating that hunters make decisions that are consistent with the predictions of optimal foraging theory, switching to alternative prey species as the preferred species becomes scarce (Alvard, 1993). However, there has been very little work on trapping behaviour, despite it being the most widespread method of hunting in central African forests, and almost nothing is known about the determinants of trapper decisions (but see work by Fitzgibbon, Mogaka & Fanshawe, 1995; Noss, 1998, 2000).

With this in mind, we carried out a study of trapping in the Monte Mitra area of continental Equatorial Guinea, where the bushmeat trade is substantial and increasing, and trapping is still the most widely practiced method of hunting (Kümpel, 2006). We addressed three questions: (1) what type of trapper puts in most trapping effort, and how; (2) how do different components of trapping effort predict trapping success; (3) why do hunters choose trapping over other hunting methods? We end by considering the current and potential impacts of hunters on wildlife populations, and the implications for the sustainability of bushmeat hunting in the area.

Methods

Study area

This study focuses on the village of Sendje in continental Equatorial Guinea. Sendje (population 317 in 2003: Kümpel, 2006) is 41 km south of the regional capital, Bata (population 132 235 in 2001: Ministério de Planificación y Desarrollo Económico, 2002), and a major source of bushmeat for the city. Data were collected from November 2002 to January 2004. The hunting catchment of Sendje is largely mature secondary evergreen humid closed forest (Senterre & Lejoly, 2001) and spans either side of the village, extending eastwards into the central third of the 2000 km² Monte Alén National Park, the Monte Mitra forest, about 10 km away.

Small villages and logging camps once occurred throughout the surrounding area, but these are now abandoned (the last logging operations ceasing in 1997) and the only permanent settlements are along the main road. Many of these old settlements are now used as bases for trapping and, to some extent, gun hunting. Ten of these hunter camps were in use during the study period (Fig. 1). Hunters visited the closest camps, such as Bisun and Bingungun (8.5 and 9.5 km from Sendje, respectively), on day trips, spent Monday to Friday in those camps at intermediate distance, and remained in Mabumom camp, the furthest camp at 30 km from Sendje, for up to 2 weeks at a time, utilizing porters to transport food to camp and meat back to the village.

Figure 1.  Map of the study area showing trapping zones around camps and increasing trapping intensity with distance from camps. Sendje (N001°32.001′/E009°49.485′) and hunt camps are marked with large black dots. Minimum convex polygons show the extent of traps around each camp during the study period. Grey dots represent individual trap groups, with larger dots signifying more traps in the group (see key on map). The dark grey area is Monte Alén National Park, thin grey lines are roads and thick grey lines rivers. Five-kilometre grid squares are marked.

Trapping with wire snares is the main type of hunting practised in continental Equatorial Guinea. Nga leg snares are the most common type of snare; the standard (small) nga is set with a single thickness of wire to catch small to medium prey such as blue duiker Cephalophus monticola or brush-tailed porcupine Atherurus africanus, but also larger prey when available. In areas where the trapper believes there is a higher density of larger terrestrial prey such as red river hog Potamochoerus porcus or sitatunga Tragelaphus spekei, heavy-duty large nga snares are often set, using quadruple thickness of wire. Ebeneñong neck snares, the more common type of neck snare, are typically set outside burrows or around fields, sometimes in conjunction with fencing, primarily to catch rodents like the giant pouched rat Cricetomys emini or brush-tailed porcupine. Abenqua neck/body snares are rarer, being complicated and time consuming to set, and are usually positioned on top of fallen logs, aimed at preferred species such as pangolins and large rodents.

Gun hunting (with shotguns) has been minimal in the country since a ban on firearms in the 1970s that reduced their availability and affordability (Butynski & Koster, 1994). Officially a permit is now required to own a shotgun, and a separate licence to hunt; rifles are still illegal throughout the country and rarely used. Of 83 hunters registered in the Monte Mitra area during this study, only 14 used shotguns (hereafter ‘guns’), and this was usually in conjunction with trapping. Net hunting is not practiced in the area. The bushmeat trade is in theory regulated by official protection of particular species from hunting, as well as all hunting being prohibited in protected areas (Law No. 8/1988: República de Guinea Ecuatorial, 1988), but there was little active enforcement of these laws during the study period.

Data collection

We interviewed 72 of the 83 hunters active during the study period. We asked hunters about their personal background, previous and current alternative livelihoods, reasons for hunting, species targeted and why, favoured gear types and why and costs incurred for different gear types. The presence of other wage earners in the hunter's household or ownership of a bar/shop by the household was taken from a census of all households in Sendje (Kümpel, 2006).

We accompanied trappers during their hunting trips to record spatially and temporally explicit information on trap configuration and trapping success. Specific data-entry templates for trapper follows were written for the CyberTracker program (http://www.cybertracker.co.za/) and downloaded onto handheld computers (Handspring Visor Neo/Deluxe; Handspring Inc., Mountain View, CA, USA) with attached GPS unit (Magellan GPS Companion; MiTAC Digital Corporation, Santa Clara, CA, USA). Time and geographical position were thus automatically logged for all data points, including start and end points, and the program was also set to log the hunter's position every 15 min. The use of CyberTrackers to record data served the dual purpose of ensuring that all fields required were systematically filled in while conducting a check on the research assistants' temporal and spatial activities. It also allowed for subsequent analysis using GIS software (ArcView 3.3).

Follows were conducted from February 2003 to January 2004. A ‘trap group’ was defined as a spatially distinct grouping of traps of the same age (i.e. those that had been set on the same day by the trapper). Each trap group was logged, with the number of each type of trap, the age in days of the trap group and other details such as the presence of fencing between traps and habitat type. For each animal caught in or escaped from a trap, the species (when known), its age class, sex and state, whether it was collected, discarded or had escaped, the trap type and trap age were recorded. An attempt was made to follow as many trappers active during the study period as possible, and the majority of trappers were followed at least once. Seventy-seven follows were carried out with 48 trappers, 18 around Sendje and 59 around eight hunter camps.

All hunter offtake passing through Sendje was recorded from November 2002 to January 2004. Information was recorded on the hunter (name, household code), hunting trip (location, hunting methods used, number of hours/days) and each animal caught (species, weight, method of capture, whether eaten or sold, carcass price). A total of 9374 animals was recorded. Finally, bushmeat consumption within the majority of the households (42 out of 59) was recorded every 8 days through a 24-h food recall interview from January 2003 to January 2004.

Local research assistants from the village were trained and supervised by the first author to assist in interviews and conduct follows. Interview/follow participants were given small monetary or other gifts which were sufficient to encourage participation in the study but not to change their behaviour (see Kümpel, 2006 for details).

Data analysis

The analyses were run for 34 trappers who had been followed and for whom the complete complement of traps had been counted, using SPSS 15.0 for Windows software. Individuals who combined trapping with gun hunting during the same follow were excluded, because this might compromise trapping effort and affect the catch rate (Kümpel, 2006). We included mean data per trapper from a randomly selected maximum of two follows per camp to avoid over-representation of individual trappers.

Seven socioeconomic characteristics of hunters were assessed as explanatory variables of different measures of trapping effort in generalized linear models (GLMs). These were age, marital status, number of children, education level (illiterate, primary or secondary), whether or not he was native to Sendje, whether or not he had been employed during the last decade and whether or not his household had an alternative source of income (i.e. owned a bar/shop or contained a wage earner). Variables were entered or removed from the model using a stepwise procedure according to the following criteria: probability of F to enter ≥0.050 and to remove ≤0.100.

The different trapping effort variables were: distance of start of trap circuit from village according to the path walked by hunters to a particular camp (‘trap circuit distance’), days spent trapping per month (proportion of days, arcsine root transformed; for trappers making day trips from the village, 8 h was assumed to be equivalent to 1 day) and number of traps. In addition, two further measures of trapping effort were derived: an ‘effective trap index’ and a ‘total trapping effort index’. The effective trap index was the number of traps indexed by the mean trap age (square root-transformed in order to give a normal distribution of the index), as a surrogate for frequency of moving traps and therefore effort expended, and was found to be predicted better by the GLM explanatory variables than trap number alone (Table 1). The ‘total trapping effort index’ was the principal component extracted from a factor analysis combining the two effort response variables that had been predicted best by the GLM explanatory variables, days trapping per month and effective trap index (Table 1). Together they formed a single component with an eigenvalue <1 (0.393), which explained 80.4% of the variance. The inclusion of trap circuit distance as a third factor did not increase the proportion of variance explained.

Table 1.   Results of generalised linear models of significant hunter profile variables against different measures of hunter effort (d.f.=34)

Dependent variables (hunter effort)	Explanatory variables (hunter profile)
	Age		Non-native to Sendje		Overall
	t	P	t	P	P	R ^2
The overall model is the minimum adequate model after removal of all non-significant explanatory variables originally included in the GLM (number of children, marital status, education level, the presence of an alternative source of income in the household and whether the hunter had had a job in the past decade). a Non-equal errors (Levene's test of homogeneity of variance) mean that results of this model are not meaningful, but are given here for comparison.
Trap base distance from village (km)	−3.74	0.001	NS	NS	0.001	0.282
Number of traps	NS	NS	3.837	0.001	0.001	0.294
Days trapping per month^a	−4.976	< 0.001	NS	NS	< 0.001	0.419
Effective trap index	−3.015	0.005	3.464	0.002	< 0.001	0.488
Total trapping effort index	−3.013	0.005	3.582	0.001	< 0.001	0.498

A K-means cluster analysis was then performed on the total trapping effort index to group trappers into three cluster types according to their trapping effort (Table 2). Convergence was achieved after four iterations (with a maximum absolute coordinate change for any centre of 0.00).

Table 2.   Trapping strategies and descriptions according to the three cluster types defined by the K-means cluster analysis from the total trapping effort index

	Strategy 1	Strategy 2	Strategy 3
	Low-impact village trappers	Medium- impact forest trappers	High-impact forest trappers
No trappers in cluster (n=34)	11	15	8
Final cluster centre	−1.19	0.20	1.26
Mean age of trappers (years)	64	43	26
Mean proportion of neck snares (%)	62	4	0.4
Mean days trapping per month	3.3	13.5	16.4
Mean trap circuit distance from village (km)	5.5	24.8	32.4
Mean effective trap index	5.8	14.0	27.3
Trapping strategy summary	Few days per month, close to village, low number of effective traps	Average days per month, medium distance, medium number of effective traps	Average-many days per month, medium-long distance, high number of effective traps

Another similar set of GLMs was run to identify which trapping effort variables were the best predictors of two measures of individual trapper success (average number of carcasses and biomass caught per month). Carcass number and biomass were calculated from the offtake survey, except where trappers were mainly village based and tended to consume the few animals they caught, where capture rates were estimated from the household consumption survey.

Results

Predictors of trapping effort

Of all measures of trapping effort tested, the total trapping effort index was found to be the effort variable predicted best overall by trapper profile variables, closely followed by the effective trap index (Table 1). Trapper age was a highly significant predictor of nearly every measure of trapping effort (effort decreasing with age). Whether or not the trapper was native to Sendje was also a good predictor of effort variables that included a measure of trap number (non-natives expending more effort). Although these two hunter characteristics are related (the age of Sendje non-natives is much lower than that of Sendje natives: mean±se=30.0±3.4 vs. 47.2±3.7 years, respectively), they exert independent effects in the GLM. In contrast, the remaining socioeconomic variables had no influence on trapping effort.

The cluster analysis of the total trapping effort index separated trappers into three distinct clusters representing different trapping strategies (low, medium and high effort). Each trapping strategy can be broken down by time expended, trap circuit distance from village and effective trap index, showing the influence of different effort variables on overall effort (Table 2). There is a greater distinction between the number of effective traps in each group (splitting clearly into low, medium and high number of effective traps) than days spent trapping and trap circuit distance, which do not differ greatly between strategies 2 and 3. Trapping strategy 1 describes the village trapper group, which stays close to the village, going on only short trips and having only a few traps. Strategies 2 and 3 describe two different types of forest trappers, which both tend to stay overnight in the forest and spend a similar amount of time trapping. The majority of these forest trappers move camps on a regular basis: every few months to a year. However, strategy 2 trappers have either many older traps, or a fewer younger ones, while strategy 3 trappers have many more effective traps. The three strategies correlate with age, strategy 1 trappers being the eldest and strategy 3 the youngest (Table 2).

Predictors of trapping success

Days spent trapping per month and the effective trap index were highly significant predictors of trapping success, measured both as carcasses and biomass per month in the second GLM (Table 3). Trap circuit distance from village was found to be a non-significant explanatory variable. As the total trapping effort index was a composite of days per month and the effective trapping index, it could not be included in the GLM, but in a univariate linear regression it was found to be a highly significant predictor of trapping success (Table 3 and Fig. 2a), and more so than the GLM overall or if either days per month or the effective trap index were regressed alone (Table 3). As expected given that trapper age was a significant predictor of trapping effort, in a linear regression of trapper age against offtake, trapper age significantly predicted trapping success (adjusted R²=0.355, F=19.165, t=−4.378, d.f.=33, P<0.001), with younger trappers catching more animals (Fig. 2b).

Table 3.   Results of generalised linear models and linear regressions to identify which measures of trapping effort are better predictors of trapping success

Explanatory variables (trapping effort)	Response variables (trapping success)
	Carcasses per month	Biomass per month (kg)
Trap circuit distance from village was found to be a non-significant explanatory variable in the GLM so was removed from the minimum adequate model.
GLM
Days per month (arcsin root-transformed)	t=7.530, P<0.001	t=5.802, P<0.001
Effective trap index	t=6.038, P<0.001	t=3.837, P=0.001
Overall model	Adjusted R^2=0.868, F=106.325, d.f.=33, P<0.001	Adjusted R^2=0.767, F=48.836, d.f.=30, P<0.001
Simple linear regression
Trapping effort index (principal factor from days per month and effective trap index combined)	Adjusted R^2=0.890, F=258.718, t=16.085, d.f.=32, P<0.001	Adjusted R^2=0.789, F=109.372, t=10.458, d.f.=29, P<0.001

Figure 2.  (a) Trapping effort index and (b) trapper age against offtake, measured as number of carcasses per month (n=33 followed trappers with sufficient data to estimate carcasses trapped per month), separated by different trapping strategies.

The fact that the effective trap index is one of the best predictors of trapping success demonstrates that the age, as well as simply the number, of traps is important in defining both trapping effort and success. At every hunter camp (bar one) where follows were conducted, the mean age of traps that caught animals was lower than the mean age of all traps recorded. Overall, the mean age of all traps recorded during the course of 74 follows (n=6024 traps) was 125 days, whereas the mean age of successful traps (n=53 animals) was 71 days.

On average, 9% of animals were discarded as they were found rotten in traps. The likelihood of decomposition increases with time left between checking of traps. For traps checked 1–2 days previously, only 5% of animals caught were discarded, but this proportion increased to 13% for traps checked 3–4 days previously and 22% for traps left 5 days or more. Trappers who check their traps more frequently therefore collect a greater proportion of animals caught.

The predictions of our model, that trappers with more effective traps are likely to show greater trapping success, are supported by consideration of the trapping strategy of the most prolific trapper in Sendje, who was not included in the analyses as he also gun hunted. His greater catch per day trapping (mean 2.6 carcasses day⁻¹ trapping compared with 1.8 for all trappers) resulted in a very high trapped offtake (mean 48 animals trapped per month in 2003, out of a total of 71 animals hunted by any method). This was not a result of having an unusually high number of traps (he had a mean of 124 traps over four follows, compared with 116 for the average forest trapper), although the mean age of his traps (44 days) was substantially lower than for all traps in Monte Mitra. However, his mean trap age did not differ greatly from that of all traps in his camp (54 days), but his rate of capture per trap was 2.5 times higher than the other trappers at his camp. His exceptional trapping success rate may be explained by a combination of unusual care, skill and planning in setting his traps. He operated no special type of trap, but moved them frequently, appeared to set them sensitively enough to retain animals, and checked them regularly, dismantling them when returning to the village for extended periods. His rate of collection of animals caught in traps was thus far above average (he collected 91% of animals caught compared with 56% for all other trappers), as fewer animals were wasted (5%) or recorded as escaping (5%, compared with 36% overall).

Choice of trapping over other hunting methods

Trapping is the most common method of hunting in Sendje, accounting for 83.5% of captures in 2003 (with gun hunting accounting for only 10.2%; n=8267). When asked what gear type they preferred, 65% of respondents stated trapping, compared with 33% gun hunting (n=62). Hunters cited various barriers to entry to gun hunting as their main reasons for trapping, such as prohibitive cost, lack of availability of guns and cartridges and lack of skill or experience in gun hunting, some of the older trappers stating they were too old (Fig. 3a). In contrast, most hunters who expressed a preference for gun hunting gave positive reasons, such as that it was an easier or quicker method of hunting, or even that they enjoyed it (Fig. 3b). This suggests that if guns and ammunition became cheaper and more accessible, there may well be a switch from trapping to gun hunting.

Figure 3.  Reasons cited for (a) traps (n respondents=39) and (b) guns (n respondents=17) being the preferred hunting method, taken from hunter interviews.

However, a handful of respondents did mention that they trapped because it reaped preferred prey and higher profits. Different hunting methods (including different trap types) target very different types of prey (Fig. 4). The main distinction between trap types is between neck and foot snares. The mean distance from Sendje for neck snares is 11 km, whereas for leg snares it is 21 km; older, village-based trappers adopting trapping strategy 1 use a greater proportion of neck snares (Table 2). Shotguns are used primarily to hunt diurnal arboreal monkeys (Fig. 4); 95% of primates in the offtake were gun hunted.

Figure 4.  Proportion of different taxa recorded in Sendje offtake survey caught in foot snares, neck snares and with shotguns (data from 2003 only).

The mean biomass of all carcasses caught and mean price obtained for all traded carcasses, as well as the range of weights and prices, varies between these three hunting methods (Table 4). Animals hunted by gun have the greatest biomass, but only a slightly higher price, and when the cost of the cartridge plus wastage due to duds and mis-hits is taken into account, the mean profit is actually less for trapping (that has minimal running costs). The wider range of weights and corresponding prices of trapped animals also shows that more valuable prey can be trapped (usually by the large nga snares) than shot, even if only occasionally. Although most animals trapped by neck snares are consumed rather than sold, those that are sold (mainly pangolins) are of high value by weight. The majority of animals shot and trapped with leg snares are sold, confirming that these two methods are the main gear types used by commercial hunters.

Table 4.   Relative offtake and profit for the three main hunting methods, taken from offtake data (whole, fresh carcasses only), November 2002–January 2004

Gear type	Number of carcasses caught	Mean biomass per carcass caught (kg)	Proportion carcasses sold	Mean biomass per carcassvsold (kg)	Mean price per carcass sold (CFA)	Mean maximum profit per carcass (CFA)	Carcass weight range (kg)	Carcass price range (CFA)
All guns used by Sendje hunters were shotguns. Exchange rate on 15 June 2003 (mid-way through study) was 1000 CFA:US$1.82. ^a Minus 850 CFA per carcass to take into account cartridges used. ^b Taking into account estimated costs of wire of 150 CFA per carcass (see Kümpel, 2006).
Gun	623	4.97	0.97	5.05	3746	2896^a	0.9–34	1000–35 000
Leg snare	5512	2.77	0.91	2.88	3595	3445^b	0.2–50+	300–70 000
Neck snare	274	1.36	0.34	2.08	3427	3277^b	0.1–26	500–45 000

Discussion

Trappers in Monte Mitra operate under three strategies, reflecting different levels of effort and impact: low-impact village trappers (strategy 1), medium-impact forest trappers (strategy 2) and high-impact forest trappers (strategy 3). As not all measures of trapping effort have equivalent impacts on prey, it is important to distinguish between them. For example, a trapper could travel a considerable distance, but set few or ineffective snares; thus, an apparently high level of effort in terms of travel costs would have a minimal impact. Number of traps per se was not found to be a good measure of effort, but the age of the traps was found to be important in modifying their overall effectiveness, and a derived variable, the ‘effective trap index’, was a good predictor of trapping success. Trapping success was best predicted by the derived variable, the ‘total trapping index’, a combination of the effective trap index and days trapping per month.

Effort depends on trappers' age, in that younger trappers go further, for longer and set a greater number of effective traps. Whether or not a trapper was born in Sendje was also an important predictor of effort and success, in that non-natives tended to adopt strategy 3, but as they were a younger subset of the trapper population as a whole this was hard to separate from the effect of age. The fact that none of the other socioeconomic hunter profile variables predicted trapping effort shows that trapping in Sendje is not limited to particular social groups, ‘unemployable’ individuals with lower status or education, or those with no other household income. The majority of males hunted to some extent (with hunting being the sole livelihood for 55/93 adult males in the village), if not as their sole livelihood then as a fall-back activity when they had no alternative source of income (Kümpel, 2006). The presence of high-impact commercial hunters from outside the village demonstrates that hunting in the Monte Mitra area is a major income-generating livelihood for these itinerant young men.

We also found that younger trappers not only put in more effort, they had greater trapping success. In contrast, Walker et al. (2002) reported that hunting success in the Ache of Paraguay peaked at a later age than physical strength, because experience was required to acquire gun-hunting skills. Our results suggest that this is generally not the case for trapping, which appears to be easier to learn and to achieve reasonable ability than pursuit hunting. Nevertheless, there may be some additional skills required to become an exceptional trapper. The most prolific trapper in Sendje was 33 at the time of the study (so older than most of the strategy 3 trappers), had never had a job, and had spent at least the last 8 years trapping and gun hunting in Monte Mitra (he was also non-native to the village); in contrast, many of the other fit, young, relatively successful trappers were fresh out of school or trapped only as a fall-back activity when they were out of work.

However, the fact that much of the trapping success of this hunter can be attributed to the fact that he was simply collecting a higher proportion of animals trapped compared with other trappers, shows that trapping efficiency is also clearly important. An increasing proportion of animals are wasted due to decomposition the longer traps are left unchecked (also shown in other studies, e.g. Fimbel, Curran & Usongo, 2000). Overall, we recorded 9% wastage, compared with 9% for a previous study in the same area (Fa & Garcia Yuste, 2001) and 27% carcasses recorded decomposed/scavenged in the Central African Republic (Noss, 2000). This underlines the need to take wastage into account when evaluating the impact of hunting. Reducing wastage would increase productivity and rates of returns to hunters without increasing the level of exploitation (K. Homewood, unpubl. data). Other researchers have called for limits on the hunting radius from villages and/or enforced regular checking of traps to reduce wastage (Muchaal & Ngandjui, 1999; Fimbel et al., 2000; Noss, 2000). It has also been suggested that injuries to escaped animals could be reduced if traps were set so that the noose is released from the leg if the cable broke, rather than remaining tightly attached (Mossman, 1989). Villagers reported that some of the hunters who had come to Sendje from other villages for a few months of intensive commercial trapping had left leaving hundreds of traps each still in place, implying that animals continued to be trapped – and wasted – for as long as the traps remained viable. Not being permanent members of the village, these trappers would not have had the vested interest of Sendje residents in maintaining sustainable yields.

Even though different trap types are designed to catch different types of prey due to their action, placing, size or weight (Noss, 1998), trapping remains relatively unselective amongst species with similar body size and behaviour. Gun hunting currently predominately targets arboreal primates and is a less important activity than trapping in terms of number of participants, proportion of overall offtake and relative profit made. The current preference for trapping is largely due to the greater barriers to entry (cost, availability and skill) of gun hunting compared with trapping, and the higher returns from trapping. However, gun hunting is apparently becoming more affordable as a result of Equatorial Guinea's recent oil boom, at least for those hunters whose household economy is tied to salaried employment (9/34 hunters in our sample). The price of a cartridge in Sendje remained fairly constant when adjusted for inflation from 1990 to 2003, but reduced in cost in October 2003 to only 600 CFA each (Kümpel, 2006). This, coupled with an over threefold increase in the country's inflation-adjusted official minimum wage in less than 4 years (Kümpel et al., 2007), as well as increasing openness of the country to its neighbours, means that guns and ammunition are becoming increasingly affordable and available. A model by Damania, Milner-Gulland & Crookes (2005) also warns that increases in hunter income are likely to lead to a switch from cheaper, less effective traps to more efficient and expensive guns (with gun hunting likely to become an additive rather than alternative source of income). This already appears to have happened elsewhere in continental Equatorial Guinea where more bushmeat is hunted with guns (Puit, 2003; Kümpel, 2006). The resulting impacts on prey populations depend on whether any hunting controls are in place; in theory more selective gun hunting allows rare species to be avoided and may increase the likelihood of sustainable hunting (e.g. Elkan, 2000), but in the absence of suitable controls, which is currently the case in Equatorial Guinea, the larger bodied and generally more vulnerable species will be preferentially targeted (Kümpel et al., 2008).

This study, one of the few conducted on hunter behaviour in central Africa, shows that an understanding of incentives for hunting, the impact of different types of hunters and of the conditions that may encourage a switch from traps to guns is crucial in gauging future sustainability of hunting in the area. In particular, it is useful in underlining the subtleties of different measures of trapping effort. We have shown that younger trappers, and those from outside the village, put in more trapping effort, with more effective traps, and subsequently have greater impact in terms of animals trapped. Currently, trapping is the main method of hunting, but if barriers to entry to gun hunting are removed, and if the gradual depletion of terrestrial prey is taken into account (Kümpel et al., in press), a switch by many trappers to more efficient gun hunting in Monte Mitra seems likely, with devastating impacts on primates. While primates are not preferred by bushmeat consumers in Bata, as bushmeat consumption and preference is linked to availability in the market they are acceptable alternatives to trapped terrestrial taxa (East et al., 2005).

There is currently little management or control of bushmeat hunting in the Monte Mitra area other than a tax imposed by the village president on foreigners to the village to exploit local forest resources, resulting in overexploitation by immigrant commercial hunters at the expense of Sendje village trappers (as also recorded in south-east Cameroon by Fimbel et al., 2000). The sustainability of Sendje-based hunting could be improved by: (1) reducing incentives for young men, in particular those from outside the village, to hunt in the first place, by developing alternative livelihoods – ideally located away from Sendje to avoid gun hunting becoming an additive livelihood (Kümpel, 2006); (2) reducing overall hunting effort, for example, by setting limits to the number and geographic extent of immigrant hunters and their traps and only allowing hunting within 10 km of Sendje (meaning traps could be checked on day trips and would no longer be inside the park); (3) minimizing wastage from individual traps, by encouraging regular checking of more sensitive traps. These last two measures should be set, enforced and ideally monitored at the community level, with appropriate technical assistance and political oversight and assuming access and resource use rights, in order to ensure local responsibility and buy-in.

Acknowledgements

We thank the Ministry of Forestry and Infrastructure, INDEFOR and ECOFAC in Equatorial Guinea for supporting our research, and Brigid Barry, Guy Hills-Spedding, Bienvenido Ondo Ndong, Santiago Eñsen, Antonio Ayong Nguema and Teresa Akeng for assistance in data collection. The study was funded by the UK's ESRC/NERC (awards R42200134017 and PTA-026-27-1416) and Conservation International through the CARPE programme. This study is a contribution to the ZSL Institute of Zoology Bushmeat Research Programme.