The Late Mississippian and Pennsylvanian have been referred to as the Coal Age due to enormous paleotropical peat accumulations (coal beds). Numerous fossil floras have been collected from these coals, and their associated seat-earth paleosols and roof-shales, over more than two centuries, leading to the inference of vast swampy wetlands covering the Pangean tropics during the Pennsylvanian. In contrast, the Permian tropics are characterized as more arid, with sparser and more heterogeneous vegetation than inferred for the Pennsylvanian. In the tropics, the Pennsylvanian to Permian transition has been described as a changeover from a pteridophytedominated "Paleophytic flora", to a seed-plant dominated "Mesophytic flora. This view notwithstanding, floras dominated by xeromorphic seed plants also are well known from the Pennsylvanian tropics. Some authors have characterized these plants as being occupants of uplands, subsequently transported into basinal-lowland, preservational environments. In this model, uplands are well drained, causing areas of drought under otherwise everwet climates. In this paper, we present an alternative interpretation: that the apparent transition in Pennsylvanian-Permian tropical vegetation reflects two types of taphonomic megabias. First is a preservational megabias, strongly favoring the vegetation of humid climates over that of seasonally dry climates. Accordingly, tropical-plant preservational potential fluctuated in concert with Late Paleozoic Ice Age glacial-interglacial oscillations, and contemporaneous sea-level and climatic changes. Second is an analytical megabias, strongly favoring the discovery and collection of the wetland biome from Pennsylvanian strata, overlooking the less frequently and more poorly preserved drought-tolerant biome. By Permian times, vast wetlands, and their fossil record, had largely disappeared from central Pangea (although continuing in Cathaysia), making drought-tolerant vegetation more "visible" to searchers, without changing its preservational circumstances. We demonstrate that the upland model is untenable, being inconsistent with the principles of plant biogeography and with geological aspects of the fossil record.