Small-scale gold mining is important to rural livelihoods in the developing world but also a source of environmental externalities. Incentives for individual producers are the classic policy response for a socially efficient balance between livelihoods and the environment. Yet monitoring individual miners is ineffective, or it is very costly, especially on frontiers with scattered small-scale miners. We ask whether monitoring at a group level effectively incentivizes cleaner artisanal mining by combining lower-cost external monitoring with local collective action.We employ a mining-framed, threshold-public-goods experiment in Colombia’s Pacific region, with 640 participants from frontier mining communities. To study compliance with collective environmental targets, we vary the target stringency, including to compare increases over time in the stringency versus decreases. We find that collective incentives can induce efficient equilibria, with group compliance — and even inefficient overcompliance — despite the existence of equilibria with zero contributions. Yet, for demanding targets in which the reward for compliance barely outweighs the cost, compliance can collapse. Those outcomes improve with past successes for easier targets, however, so our results suggest gain from building coordination via graduated stringency.