Tetrapods currently comprise over 30,000 species distributed globally and occupying a stunning diversity of bodyplans and ecologies. After many years of fruitful work, the early evolutionary history of the group can be considered well-understood. We have a detailed sequence of anatomical change across the fish-tetrapod transition in the Devonian period and a rich fossil record from the late Early Carboniferous (Mississippian) and Late Carboniferous (Pennsylvanian). These two datasets reveal the initial assembly of the tetrapod bodyplan at one end and the proliferation of early members of crown group lineages at the other. However, there remains an unbridged divide between the apparently low-diversity, obligately aquatic, fishlike Devonian tetrapod assemblage and the speciose, ecologically diverse post-Devonian assemblage. This divide is inflected by the end-Devonian mass extinction (EDME), which is itself followed by a ~15 million year hiatus in the tetrapod fossil record (Romer’s Gap) that is just now beginning to be populated. In this dissertation I attempt to understand the evolution of early tetrapods through the end-Devonian mass extinction and its aftermath by integrating data from comparative morphology, phylogenetics, and community ecology. In Chapter 2, I redescribe the postcranial skeleton of the middle Mississippian stem tetrapod Whatcheeria deltae to generate new anatomical and character information and better understand its relationships. In Chapter 3, I analyze a new early tetrapod phylogenetic dataset to more evaluate the effect of new Whatcheeria data on the structure of the apical tetrapod stem group, hypotheses of tetrapod crown group membership, and inferred patterns and timing of branching events during the Late Devonian and Mississippian. Finally, in Chapter 4 I use comparative ecological methods to evaluate effect of the end-Devonian mass extinction on richness of guilds (=functional groups) and community-level resistance to perturbation across environments, with special attention paid to the middle Devonian-middle Mississippian origin of terrestrial communities. In Chapter 2 I redescribe the postcranial anatomy of Whatcheeria on the basis of hundreds of previously unstudied specimens. Whatcheeria is revealed to be an unusual large-bodied form with an elongate neck, robust appendicular skeleton, and regionalized trunk ribs. Limb proportions resemble those of terrestrial crown tetrapods from the Permian such as Eryops, but the presence of a well-developed cranial sensory canal system indicate that Whatcheeria was an aquatic animal, albeit one adapted for walking rather than swimming in the water column. Using a diagnosis improved by new character data, the family Whatcheeriidae can be restricted to the earliest Mississippian Pederpes and middle Mississippian Whatcheeria. Whatcheeriid autapomorphies can now be recognized in several Mississippian specimens as well. Whatcheeria indicates that Mississippian stem tetrapods were capable of much more morphological, physiological, and likely functional complexity than previously appreciated. Moreover, such complexity was not necessarily tied to terrestrialization. In Chapter 3 I analyze a new phylogenetic dataset incorporating new postcranial character data from Whatcheeria as well as recent discoveries from Romer’s Gap. The dataset combines new characters with those of multiple dataset ‘lineages’. Parsimony analysis finds a well-supported, monophyletic Whatcheeriidae composed of Whatcheeria and Pederpes located deep on the tetrapod stem. Contra recent hypotheses, large aquatic embolomeres and the limbless aistopods and adelospondyls are found within the amniote total group. The colosteids have an ambiguous relationship to the tetrapod crown node, alternating between the sister group of crown tetrapods to the sister group of the temnospondyls within the lissamphibian total group. Analysis of anatomical partitions of characters indicates similar levels of signal in cranial and postcranial data but divergent patterns of evolution across partitions, particularly between the anterior and posterior appendicular skeletons. Estimation of node ages supports a Devonian origin for most stem tetrapod lineages but not terrestriality. Node age estimation and anatomical partition analyses both support at minimum one independent origin of terrestriality in each of the crown tetrapod lineages. In Chapter 4 I analyze a dataset of 16 paleocommunities from the middle Devonian (Givetian) through the end of the Mississippian (Serpukhovian). Non-metric multidimensional scaling (NMDS) analysis finds a strong time-environment separation between axes when the diversity of high-level taxonomic groups is used, but this distinction breaks down substantially when diversity of ecological guilds is used. These results suggest substantially greater functional continuity than taxonomic continuity through the EDME. Simulation of paleocommunity response to perturbation using the Cascading Extinction on Graphs (CEG) model finds no clear difference in response between communities before and after the Frasnian-Famennian invertebrate extinction or the EDME. The response of paleocommunities is bimodal with low variance, shifting extremely rapidly from low to high levels of secondary extinction at approximately 50% perturbation. Curiously, variance in secondary extinction is low throughout. I propose that this response pattern is due to a combination of low guild richness, high guild evenness, broad prey profiles among predators, and top-down pressure from high-trophic-level predators. At low levels of perturbation, the generalist nature of predators results in low per-species predation pressure. However, at high levels of perturbation, increasing secondary extinction, in combination with low guild richness, begins to rapidly eliminate entire guilds. Top-down predation pressure is still being widely applied, and the entire food web collapses. Gilboa, a middle Devonian terrestrial paleocommunity, and East Kirkton, the oldest terrestrial tetrapod paleocommunity (a diverse arthropod assemblage is also present), both share this response pattern with the aquatic localities, though variance in secondary extinction values is greatly increased by the lower species diversity. Terrestrial communities appear to have developed through the diversification and proliferation of plants and arthropods; tetrapods later fit into guilds which had been previously defined and occupied solely by arthropods. I propose the origin of more fully terrestrial vertebrate communities may lie somewhere in the Mississippian, and previous hypotheses that late Pennsylvanian/early Permian assemblages represent the initial organizational structure of the first terrestrial tetrapod communities are not supported. Datafiles for the Chapter 4 paleocommunity food web analyses are listed in the Supplementary Files section. These files contain faunal lists, guild assignments, and model parameter information in Excel (.xlsx) format.