Does large-scale restoration work for biodiversity? Counterfactual evidence from Africa's Great Green Wall

The Great Green Wall (GGW) was launched in 2007 as a large-scale restoration program to combat land degradation across the African Sahel. While substantial progress has been made in vegetation restoration, its impacts on biodiversity remain poorly quantified. This study assesses the causal effects of the GGW on avian species richness in three representative countries: Senegal (West Africa), Nigeria (Central Africa), and Ethiopia (East Africa).We employed ensemble species distribution models (biomod2) to project habitat suitability for avian species in each country, producing predictions for baseline (2007–2015) and current (2016–2024) periods. Causal inference was established through 1:1 propensity score matching (PSM) based on pre-treatment environmental covariates, pairing GGW areas with comparable controls, followed by difference-in-differences (DID) estimation of the Average Treatment Effect on the Treated (ATT). To disentangle climate and vegetation contributions, we constructed factorial scenarios combining environmental layers from both periods, decomposing species richness changes into climate-driven, vegetation-driven, and interaction effects.Results reveal divergent GGW impacts. Nigeria demonstrated significant positive effects (ATT = +7.45; p < 0.001), with scenario decomposition indicating vegetation-driven effects dominated biodiversity gains—suggesting active restoration effectively enhanced habitat quality. Ethiopia showed no significant difference between GGW and control areas (ATT = −2.48; p = 0.13), with climate and vegetation effects comparable across treatments. Senegal exhibited limited benefits in GGW areas (ATT = −4.17; p < 0.001), where climate-driven changes dominated and vegetation effects remained constrained. These contrasting outcomes demonstrate that large-scale restoration does not uniformly deliver biodiversity co-benefits, as regional contexts and implementation intensity critically mediate effectiveness. Nigeria's success highlights the potential for well-implemented restoration to generate measurable biodiversity gains, while variable outcomes elsewhere underscore the need for adaptive management accounting for local conditions.Our findings provide policy-relevant evidence for optimizing pan-African restoration initiatives. We recommend prioritizing high-potential regions, integrating biodiversity monitoring into evaluation, and adopting locally tailored adaptive management. The PSM-DID-SDM-scenario decomposition framework offers a transferable methodology for evaluating large-scale conservation interventions globally.Keywords: avian biodiversity; species distribution models; causal inference; difference-in-differences; ecological restoration; Sahel