Habitat loss and fragmentation have often been viewed as major threats to species interaction and global biodiversity conservation. However, habitat degradation can also give rise to positive ecological and behavioral responses, challenging the notion that its consequences are entirely detrimental. While controlling for the degree of total habitat loss, we studied the influence of habitat fragmentation and phenotypic diversity on the evolution of tag-based cooperation in structured populations with multiple strategies. We developed a spatially explicit agent-based model with empty sites in which phenotypically diverse artificial decision makers engaged into pairwise Snowdrift-game interactions and imitated strategies of their opponent co-players. We systematically varied the number of phenotypic features in the population, the clustering degree of empty sites unsuitable for habitation, as well as the cost-to-benefit ratio $r$, and we measured the resulting equilibrium densities of conditional and unconditional strategies. Our Monte Carlo simulations revealed a complex interplay between the three investigated factors, such that higher phenotypic diversity in combination with lower $r$ and low to intermediate clustering degrees of empty sites markedly suppressed ethnocentric cooperation but simultaneously boosted unconditional, pure altruism. This dominance of unconditional cooperation was remarkably robust to variation in the initial conditions, suggesting that heterogeneous fragmentation of empty sites in moderately degraded habitats can function as a potent cooperation-promoting mechanism even in the presence of initially more favorable strategies. Our study showcases anti-fragility of cooperators in spatially fragmented but phenotypically diverse populations, as they were also able to benefit from harsh environmental conditions emerging in sparsely connected habitat remnants.