Maternal provisioning of animal embryos may be entirely through yolk deposited in the unfertilized egg (lecithotrophy) or may include post-fertilization nutrient transfer (matrotrophy) in varying degrees. Current theory suggests that the extent of post-fertilization provisioning is resource-dependent, with higher levels of matrotrophy being advantageous in more productive environments. In this study, we investigated post-fertilization embryo provisioning in a livebearing fish, Poecilia mexicana, from two different habitats (a toxic cave and a non-toxic surface habitat) that impose different energetic demands and therefore differ in resources available for reproduction. We predicted that fish in the benign habitat would be more matrotrophic than those from the toxic cave. We used two different techniques for this assay: (1) the matrotrophy-index analysis (MI) for field-collected fish and (2) both MI and radio-tracer assay for laboratory-reared females. According to the interpretation of the matrotrophy index, both populations are purely lecithotrophic, while the radio-tracer assay found females from both populations to actively transfer nutrients to developing embryos at approximately the same rate. Our results suggest that P. mexicana, which was traditionally classified as lecithotrophic, is capable of incipient matrotrophy, and that matrotrophy can contribute to embryo provisioning even in populations from resource-limited environments. Furthermore, the analysis of laboratory-reared animals provides evidence for a genetic component to the large offspring size in cave mollies, which had so far only been described from the field. Specifically, our results suggest matrotrophy occurs in species interpreted as lecithotrophic using the MI approach. Hence, to avoid misclassification, both techniques should ideally be employed in concert, rather than individually. Finally, our results provide further insights into the possible evolutionary pathway from lecithotrophic oviparity to matrotrophic viviparity.