such lava is accompanied by a massive release of carbon dioxide into the atmosphere, it was hypothesized that a sudden spike in global temperatures was somehow involved in the extinction. The oxygen drop was also invoked but not the precipitous drop advocated by Retallack. Instead, the heat spike overlay a long-term drop in oxygen, which began long before the extrusion of the Siberian trap basalts. The extinction pattern under such a scenario would show long-term extinction due to the oxygen drop, followed by (and greatly exacerbated by) a sudden rise in global temperatures. This is the idea that I promoted in two papers published in Science in 2005. One showed the extinction pattern of mammal-like reptiles across the Permian boundary in South Africa, the second (published with Ray Huey of the University of Washington) invoked a loss of terrestrial habitat due to the drop in oxygen. As atmospheric oxygen dropped, even moderate altitudes would have had even lower-oxygen content. If oxygen dropped below about 12 percent (the value found by an early version of Berner’s GEOCARB modeling), the only place on Earth where terrestrial animals could live would be at sea-level. By using reconstructions of the Permian world, we then showed that sea-level elevations made up about 50 percent of the Permian land surface. Thus, half the land area would be unavailable for animal life, and the areas that were habitable would be cut off from other habitable areas by even modest altitude. The problem with this hypothesis was that while it readily explained the observed pattern of long-term, elevated extinction during the late Permian, the heat spike did not seem severe enough to cause the observed short-term bump in extinction rates at the boundary itself. The effects of a heat spike could be imagined but not modeled.
Hydrogen sulfide poisoning. The final entry into the Permian extinction sweepstakes was put forward in 2004 by Lee Kump and colleagues at Pennsylvania State University, and is considered here to be the winner of the Permian extinction sweepstakes. Kump and his colleagues suggested that the long period of low oxygen seen both in models and by direct evidence in marine sediments themselves would have created conditions in the sea favoring the massive growth of hydrogen sulfide–releasing bacteria, which overwhelm the surface water’s oxygen supply and result in colorful bacterial blooms (hence the sugges-