At one time, this die-off was assumed to have been a gradual reduction over several million years. Now, however, it is commonly accepted that the event lasted less than a million years - from 252.3 to 251.4 mya (both numbers ±300,000 years) - geologically speaking, a very brief period of time. Organisms throughout the world, regardless of habitat, suffered similar rates of extinction, suggesting that the cause of the event was a global, not local, occurrence, and that it was sudden event, not a gradual change.
Many theories have been presented for the cause of the extinction, including an impact event, plate tectonics, a supernova, and extreme volcanism.
When large bolides (asteroids or comets) impact Earth, the aftermath weakens or kills much of the life that thrived previously. Release of debris and carbon dioxide into the atmosphere reduces the productivity of life and causes both global warming and ozone depletion. Evidence of increased levels of atmospheric carbon dioxide exist in the fossil record. Material from the Earth's mantle released during volcanic eruption has been shown to contain iridium, an element associated with meteorites. Geologist John Gorter of Agip Petroleum[?] has found evidence of a circular structure 200 kilometers in diameter called the Bedout[?] off the northwestern shore of Australia. The geology of the area dates to the end of the Permian; whether the Bedout is an impact crater is unclear.
It has also been proposed that such a collision might heat up ocean waters enough to produce "hypercanes," gigantic storms with winds possibly exceeding the speed of sound. Although not impossible, this theory has little supporting evidence.
At the time of the Permian extinction, the continents had recently joined to form the super-continent Pangea and the super-ocean Panthalassa. This configuration radically decreased the range of shallow aquatic environments; many marine ecosystems would not have survived those changes. Pangea's formation would alter both oceanic circulation and atmospheric weather patterns. However, very gradual changes like continental drift alone probably could not cause the sudden, simultaneous destruction of both terrestrial and oceanic life.
A supernova occurring within ten parsecs of Earth would produce enough gamma radiation to destroy the ozone layer for several years. The resulting direct ultra-violet radiation from the sun would weaken or kill nearly all existing species. Only those deep in the oceans would be unaffected. Statistical frequency of supernovae suggests that one at the P-T boundary would not be unlikely. While some sedimentary rock samples contain what may be records of short-term ozone destruction (large amounts of NOx gases and C14), this theory has little other evidence either for or against it.
The P-T boundary was marked with many volcanic eruptions. The acid rain, global cooling, and other weather effects associated with enormous eruptions could have globally threatened life. Volcanic activity affects the concentration of atmospheric gases directly, and, indirectly, the oceanic dissolved gases. Increases in carbon dioxide enhance the greenhouse effect and causes global warming, which would reduce the temperature gradient between the equator and the poles. As a result, thermo-haline circulation would slow and eventually stop. The oceans would stagnate, and nutrients would fail to disperse themselves. Many marine ecosystems rely on upwelling and circulation of nutrients, oxygen included; without the regular circulation, organisms would starve or suffocate. In addition, sulfur and particulates contribute to cooling, or volcanic winter, which usually lasts three to six months. Combinations of the two effects could produce a cooling cycle in which the climate alternatively warms then cools. Such temperature fluctuations could cause connective overturn of the oceans, bringing anoxic bottom waters to the surface; in an already oxygen-deprived envirnoment, this would be fatal to many forms of life.
Significant evidence supports this theory. Fluctuations in air and water temperature are evident in the fossil record, and the uranium/thorium ratios of late Permian sediments indicate that the oceans were severely anoxic around the time of the extinction. Numerous indicators of volcanic activity at the P-T boundary are present, though they are similar to bolide impact indicators, including iridium deposits. The volcanism theory has the advantage over the bolide theory, though, in that it is certain that an eruption of the Siberian Traps[?] -- the largest known eruption in the history of Earth -- occurred at this time, while no direct evidence of bolide impact has been located.
The Permian extinction is unequalled; it is obviously not easy to destroy almost all life on Earth. The difficulty in imagining a single cause of such an event has led to an explanation humorously termed the "Murder on the Orient Express" theory (Wikipedia contains spoilers): they all did it. Continental drift created a non-fatal but precariously balanced global environment, a supernova weakened the ozone layer, and then a large meteor impact triggered the eruption of the Siberian Traps. There is no way to calculate the odds of such a coincidence, but for it to have occurred once in the four thousand million year history of Earth is not unbelievable.