Agroforests can play an important role in biodiversity conservation in complex landscapes. A key factor distinguishing among agroforests is land‐use history – whether agroforests are established inside forests or on historically forested but currently open lands. The disparity between land‐use histories means the appropriate biodiversity baselines may differ, which should be accounted for when assessing the conservation value of agroforests. Specifically, comparisons between multiple baselines in forest and open land could enrich understanding of species’ responses by contextualizing them. We made such comparisons based on data from a recently published meta‐analysis of the effects of cocoa (Theobroma cacao) agroforestry on bird diversity. We regrouped rustic, mixed shade cocoa, and low shade cocoa agroforests, based on land‐use history, into forest‐derived and open‐land‐derived agroforests and compared bird species diversity (species richness, abundance, and Shannon's index values) between forest and open land, which represented the 2 alternative baselines. Bird diversity was similar in forest‐derived agroforests and forests (Hedges’ g* estimate [SE] = ‐0.3144 [0.3416], p = 0.36). Open‐land‐derived agroforests were significantly less diverse than forests (g* = 1.4312 [0.6308], p = 0.023) and comparable to open lands (g* = ‐0.1529 [0.5035], p = 0.76). Our results highlight how land‐use history determined the conservation value of cocoa agroforests. Forest‐derived cocoa agroforests were comparable to the available – usually already degraded – forest baselines, but entail future degradation risks. In contrast, open‐land‐derived cocoa agroforestry may offer restoration opportunities. Our results showed that comparisons among multiple baselines may inform relative contributions of agroforestry systems to bird conservation on a landscape scale.