CARBONATE COLLAPSE AND THE LATE PALEOZOIC ICE AGE BIODIVERSITY CRISIS

The Late Paleozoic ice age was characterized by persistently low diversity of marine invertebrates which followed a second-order mass extinction. We demonstrate that low diversity was caused by the collapse of carbonate environments when the LPIA began. Our data set includes North American (paleotropical) fossil occurrences of well-preserved marine genera from the Paleobiology Database, combined with lithologic units from Macrostrat. After dividing the data by lithology, the decline in diversity observed in the total data set was only evident in carbonate environments. Diversity within siliciclastic environments increased during the LPIA, following a brief decline in the Serpukhovian (late Mississippian). Both diversity patterns closely matched respective changes in the volume of carbonate and siliciclastic rocks. The contrasting patterns observed in the two environments suggest that habitat loss was a direct cause of changes in diversity, because other factors, such as temperature, would have affected genera in both environments. A causal relationship is also supported by the finding that diversity remained high in carbonate refugia (carbonate beds within majority-siliciclastic formations) until the Bashkirian, postdating the onset of icehouse conditions by ~8 m.y. We further explored the connection between substrate and diversity by constructing a model of Late Paleozoic diversity based on whole-interval genus accumulation curves. The modeling results confirmed that changes in rock volume almost completely account for the observed changes in diversity. The accumulation curves also demonstrated that carbonate rocks are better sampled than siliciclastic rocks, and that the supposedly low diversity during the LPIA was a consequence of carbonate oversampling. A model based on equivalent sampling intensity among lithologies showed that total diversity was suppressed only during the two main pulses of glaciation in the late Mississippian (Serpukhovian) and early Permian (Asselian-Sakmarian). The discrepancy between this result and the prolonged biotic impact observed in global-scale data may indicate that the global data are likewise affected by carbonate oversampling, or that the siliciclastic expansion that helped to offset diversity loss in North America did not occur in other regions.