Herbivory is thought to be nutritionally inefficient relative to carnivory and omnivory. But, herbivory evolved from carnivory in many lineages, suggesting that there are advantages to eating plants. To understand the adaptive significance of the transition from carnivory to herbivory, I proposed five hypotheses for the adaptive evolution of herbivory and reviewed the current freshwater literature to identify conditions where eating plants might be adaptive over eating animals. I tested three of these ideas (Suboptimal Habitat, Heterotroph Facilitation, and Lipid Allocation) using the herbivorous Sailfin Molly (Poecilia latipinna)and identified each as a potential mechanism for the evolution of herbivory. To understand the origins of herbivory in Sailfin Mollies, I reconstructed ancestral habitats and dietsacross a phylogeny of the genus Poeciliaand then used phylogenetically independent contrasts to identify patterns of diet evolution. I found that the degree of herbivory increases with increasing salinity affiliation, suggesting that in this genus, herbivory evolved as an adaptation for invading less productive saline habitats from freshwaters. This result is consistent with the Suboptimal Habitat hypothesis, which states that herbivory allows organisms to invade and persist in ‘suboptimal’ habitats. To understand how herbivory is maintained in extant populations, I raised juvenile Sailfin Mollies in mesocosms and enclosure cages placed in the Everglades to document that dietary autotrophic lipids play a role in early life history by supporting rapid growth (Lipid Allocation). However, dietary bacterial fatty acids promoted fish survival, consistent with the Heterotroph Facilitation hypothesis, which states that indirect detritivory supplements the herbivorous diet. Finally, I quantified periphyton quality/availability and consumer density across the Everglades landscape to examine the correlates of trophic dynamics in nature. Results revealed that herbivores can persist in diverse habitats and survive on varying resources when habitats are unfavorable, supporting the Suboptimal Habitat hypothesis.