Discussion
Our study encompassed a large timespan (60 weeks) and a variable spectrum of bushmeat site categories including small to large urban bushmeat markets and chop chop bars (maquis ). Although more accurate proxies of the trade dynamics were not investigated (e.g., trends in biomass and income), our unprecedented survey design allows, for the first time, to assess African bushmeat trade dynamics during and after a national ban.
Trade dynamics in Côte d’Ivoire as measured per number of sellers was strongly, negatively impacted by the COVID-19 lockdown and bushmeat ban, as all bushmeat sites went down to a null number of sellers right before or after the first governmental decisions were taken. This is in line with our original expectation on the efficiency of governmental measures. A significant reduction in carcass numbers was also observed in Nigeria after lockdown (Funk et al., 2021). The COVID-19 pandemic has shown that African states can be efficient in mitigating the bushmeat trade when resources are duly mobilized, a notable point considering the ongoing bushmeat crisis and the zoonotic outbreaks to come (D’Cruze et al., 2020; Reynolds et al., 2019).
However, we observed significant differences in the dynamics of bushmeat sites both during and after the strongest measures taken against the COVID-19 pandemic, deviating from our posited expectation on trends being similar between market types (as a measure of the effectiveness of government interventions). We speculate that bushmeat trade dynamics are affected by a combination of intrinsic market characteristics and dissimilar levels of governmental interventions. For instance, Toumodi restaurants were first to be fully closed (mid-March 2020, concomitantly with the announcement on the bushmeat ban) and first to start re-opening, at a time when all the bushmeat markets in Abidjan were closing (early June 2020). This suggests that predicting bushmeat trade dynamics under bans or restrictions is geographically and typologically (i.e., type of market) dependent. State authorities may have been more zealous in Toumodi by quickly applying the bushmeat ban before the decision on closing maquis one week later, but prompt to release pressure on sellers from a zone not affected by the lockdown (which was restricted to Abidjan).
The Ivorian state was permissive with the resumption of the bushmeat trade, as we observed a progressive return of sellers on the bushmeat sites before the sanitary measures were lifted. Although global trajectories among bushmeat sites were similar throughout the survey period, significant trend differences were observed during the constrained period. This reinforces our view that the dynamics of the bushmeat trade are site-specific and shaped by multiple factors. In our case, small (Abobo Grand Marché) or poorly accessible (Adjamé) markets could have been less controlled than the largest bushmeat market of Abidjan, Yopougon Siporex. The latter is clearly identified as the main hub of the bushmeat trade in Côte d’Ivoire (Gossé et al., 2022) and was the last to show sign of resumption, after the end of the sanitary measures against COVID-19 (October 2020).
The COVID-19 lockdown and bushmeat ban had a long-term impact on the bushmeat trade dynamics, as three months after the end of governmental measures all the bushmeat sites in Côte d’Ivoire exhibited lower numbers of sellers than before (c. 63 to 91% of the initial numbers). This effect was still prevailing in three of the bushmeat markets from Abidjan during our control survey eight months later (c. 56 to 91% of the initial numbers), despite stronger growth rate in large markets such as Yopougon Siporex. This clearly violates our initial assumption of market network resilience to the lockdown.
Forecasting predictions after 92 weeks since the start of sanitary measures in Côte d’Ivoire showed contrasting outputs relative to observed number of sellers. Most of the bushmeat sites were not optimally modelled due to their heterogeneity in growth, especially Toumodi restaurants. Random Forest algorithms performed better in the case of Yopougon and Abobo Mairie markets, where predicted values from both models converged closer to observed values. Convergence between constrained and unconstrained models may serve as an additional confidence estimate of model performance, coupled with the intrinsic estimates already available (RMSE and variance). Further investigations on Random Forest algorithms as applied to bushmeat trade dynamics over longer periods of time will have to be undertaken before the benefits of such predictive approach can be considered (e.g., as in finance research; Ghosh et al., 2022).
We conclude that neglecting the socio-geographical specificities of the different types of bushmeat sites could lead to erroneous projections of bushmeat trade dynamics. Eleven months after the end of governmental measures in Côte d’Ivoire, most of the bushmeat sites had not fully recovered in terms of number of sellers. Whether regular controls have slowed down the resumption of the trade requires further investigations. The social and economic implications behind the lack of the full recovery of certain markets at the time of our study, including that of the largest bushmeat market of Côte d’Ivoire, is unknown. Given the possible precariousness of certain bushmeat sellers (see Falola et al., 2015), it is likely that some could not economically support the consequences of a national ban for several months. However, with the data at hand, it remains hazardous to discuss the deleterious economic shockwave foreseen by some authors in relation to wildlife trade bans (e.g., McNamara et al., 2020), especially since future disease outbreaks and reduction in resources through over-harvesting will likely influence market dynamics under the current tolerated market scenario.
Understanding bushmeat trade dynamics in the context of mitigation measures will require a multi-dimensional approach where the characteristics of the different markets (e.g., type of markets, socio-demography of vendors, ban record, species hunted and places sourced) are clearly identified. Our study showed that wildlife trade bans can have a long-lasting impact on bushmeat trade dynamics, and that state power when guided by a clearly defined objective can efficiently setup trade mitigation, at least in the short-term. Actual mitigation of the bushmeat trade, whether conservation- or health-driven, will depend on a comprehensive understanding of its specific dynamics and the economic reliance of involved actors. To reach this objective, less talk, and more on-the-ground data with comprehensive modelling will be required.