Hypoxia, Global Warming, and Terrestrial Late Permian Extinctions

See allHide authors and affiliations

Science  15 Apr 2005:
Vol. 308, Issue 5720, pp. 398-401
DOI: 10.1126/science.1108019


A catastrophic extinction occurred at the end of the Permian Period. However, baseline extinction rates appear to have been elevated even before the final catastrophe, suggesting sustained environmental degradation. For terrestrial vertebrates during the Late Permian, the combination of a drop in atmospheric oxygen plus climate warming would have induced hypoxic stress and consequently compressed altitudinal ranges to near sea level. Our simulations suggest that the magnitude of altitudinal compression would have forced extinctions by reducing habitat diversity, fragmenting and isolating populations, and inducing a species-area effect. It also might have delayed ecosystem recovery after the mass extinction.

A catastrophic extinction marks the end of the Permian (1, 2) and is attributed to an acute climate crisis, among other causes (3-5). However, background extinction rates and ecosystem turnover were elevated throughout much of the Late Permian (6, 7), and recovery after extinction was slow (1, 2). Thus, environmental degradation likely occurred both before and after the final catastrophe, perhaps caused by major shifts in atmospheric chemistry (8). Indeed, modeling, isotope, and paleontological evidence (9-13) suggests that O2 levels plummeted in the Late Permian and Early Triassic (Fig. 1A) and would have restricted the supply of O2 to organisms. At the same time, CO2 levels were rising (Fig. 1A), and climate warming (14) would have increased metabolic demand for O2. Severe hypoxia was inevitable (9, 15, 16).

Fig. 1.

(A) Percent O2 over time (green) and the concentration of CO2 (gray) relative to present-day level (PAL) of each gas is indicated. Global hypoxia would have occurred in the Late Permian and Triassic because of dropping O2 combined with rising temperatures [data from (4, 37)]. The mass extinction at the Permian-Triassic boundary is indicated by a red dashed line. (B) Present-day altitude with PIO2 equivalent to that at sea level in the Phanerozoic. Thus, PIO2 at sea level at the Triassic O2 minimum would be found today at ∼5 km. (C) Predicted maximum altitude over time for hypothetical species having graded tolerances (2 to 8 km) to hypoxia. From the Late Permian through the Jurassic, however, PIO2 was sufficiently low that ranges would have been compressed to near sea level, and some species would have gone extinct. Period codes: Embedded Image, Cambrian; O, Ordovician; S, Silurian; D, Devonian; C, Carboniferous; P, Permian; Tr, Triassic; J, Jurassic; K, Cretaceous; T, Tertiary.

Here, we explore a biogeographic consequence of presumed low O2 levels during the Late Permian and Triassic: Terrestrial animals would have been restricted to low altitude, because even moderate altitudes would have insufficient O2. We simulate the magnitude of altitudinal compression and evaluate its contributions to the high background extinction rate, the catastrophic extinction, and the delayed recovery (1, 2).

Terrestrial animals attempting to live at moderate to high altitude are physiologically challenged by declining temperatures, food supply, habitat area, and O2 levels, along with increased respiratory water loss (17). The upper altitudinal limits of species likely reflect the influence of these factors as well as ecological, geological, and biogeographic factors (18).

The relative contribution of O2 in limiting altitudinal ranges has no doubt changed over time because O2 levels fluctuated so drastically (Fig. 1A). For example, when O2 levels appear to have been at their zenith during the cold Early Permian [∼300 million years ago (Ma), Fig. 1A], the partial pressure of inspired O2 (PIO2) even at an altitude of 6 km may have matched that at sea level in today's atmosphere (19). In this glaciated and O2-rich environment, altitudinal ranges would have been constrained by extreme cold or ice, not by O2 levels. But when O2 levels are calculated to be at their nadir during the hot Early Triassic (∼240 and ∼200 Ma), PIO2 even at sea level may have been equivalent to that at ≥5.3 km in today's atmosphere (Fig. 1B). If estimates of O2 (Fig. 1A) and climate are correct, the combination of low PIO2 and high temperature would have induced hypoxia and restricted terrestrial vertebrates to low altitudes.

We simulated the impact of O2 levels on maximum altitudinal ranges of vertebrates having a presumed mid-Permian physiology (20). The actual hypoxia tolerance of mid-Permian ectotherms is of course unknown but was probably lower than that of present-day vertebrates. Mid-Permian vertebrates not only had primitive respiratory systems (9) but also had been evolving in an O2-rich atmosphere for millions of years (Fig. 1A) and could not have become adapted to altitude (hence, to low O2 levels) because of icehouse temperatures during that time.

We start by deriving a reference set of hypothetical species living in the mid-Permian and having maximum altitudes ranging from 2.0 to 8.0 km, in 1-km intervals. We assume that the maximum altitude of each species was set by—or at least strongly influenced by—the critical PIO2 occurring at that altitude. Then, assuming that hypoxia tolerances were constant over time (i.e., always had the same critical PIO2), we calculate how maximum altitudes (Fig. 1C) changed with O2 levels (Fig. 1A). This involves estimating the average PIO2 at each altitude during the mid-Permian and then computing (at intervals of 10 million years) the altitude where each PIO2 would have occurred (20).

The presumed drop in O2 levels during the Late Permian (Fig. 1A) would have drastically compressed the altitudinal ranges of all hypothetical species (Fig. 1C). If O2 levels dropped to ∼16% at the end of the Permian (8), PIO2 at sea level would have been equivalent to that found today at 2.7 km. Animals with limited hypoxia tolerance would have gone extinct (Fig. 1C). To survive, a species would have needed to tolerate a PIO2 at least equivalent to that found at 6.0 km (red line, Fig. 1C) in the mid-Permian (21, 22).

O2 levels are thought to have continued to drop into the Early Triassic and to have stayed relatively low for the next 100 million years (Fig. 1A). In addition, CO2 levels (Fig. 1A) and global temperatures (14) remained high. Thus, hypoxia and the resultant altitudinal compression likely persisted into the early Cretaceous (Fig. 1, A and C) and may have contributed to a slow recovery after the mass extinction (1, 2).

Our simulations (Fig. 1, B and C) focus on O2 levels, but rising temperatures in the Late Permian and Early Triassic (14) would have exacerbated hypoxia. To reduce the impact, ectotherms could have dispersed along sea-level corridors to high latitudes, thereby lowering their body temperatures. Alternatively, they could have invaded relatively cool aquatic habitats. Indeed, air-breathing aquatic reptiles (e.g., ichthyosaurs, phytosaurs) diversified at this time.

Freshwater-breathing ectotherms were also probably restricted to low altitudes during O2 lows. Reduced atmospheric O2 levels plus warm water temperatures would lower aquatic O2 levels, but warm body temperatures increase O2 demand. Thus, water-breathing ectotherms would likely have been O2 challenged, even near sea level.

Insects, which suffered major extinctions during the Late Permian (23), might also have experienced some hypoxia and altitudinal compression. Insect development is slowed by low PO2 plus warm temperatures (24). Interestingly, giant dragonflies, which evolved during the Permian O2 high, are thought to have gone extinct in the Late Permian because of limitations on O2 diffusion (9). However, today's large adult insects rely on convective ventilation and are remarkably hypoxia tolerant (25); thus, the aquatic larvae of giant insects may have been the primary targets of hypoxia-induced extinction.

Altitudinal compression would have had important biogeographic consequences. Animals specialized for upland habitats could not have survived, as they would have been physiologically excluded from them. Moreover, mountain passes would have seemed physiologically higher to animals (26) than are passes of equivalent altitude today. Thus, populations would have become fragmented and isolated (fig. S2), and rates of local extinction may have increased (27).

Altitudinal compression plus rising sea levels at the very end of the Permian (14, 28) would have also reduced the land surface physiologically accessible to animals. This loss would have caused additional extinctions via a species-area effect (29) and also would have delayed the recovery.

Topographic maps for the mid-Permian to Late Permian (30) enable us to estimate (20), albeit with substantial uncertainty, the percentage of the land surface that was physiologically accessible to animals with specified hypoxia tolerance (Fig. 2). During the O2 high in the Permian, virtually the entire surface of Earth had sufficient O2 to sustain populations of our hypothetical species. By the Triassic, however, hypoxia would have reduced accessible land area for all but the most hypoxia-tolerant species (Fig. 1C). For example, a species physiologically capable of surviving up to 6.0 km in the mid-Permian would have been restricted to below 0.3 km by the Triassic; thus, it would have been excluded from more than half of the available land surface (gray dotted line in Fig. 2). By the early Triassic (Fig. 1A), even a species that could tolerate 8.0 km in the mid-Permian would have likely gone extinct (Fig. 1C) in the absence of compensatory adaptation. Thus, altitudinal compression could have had substantial effects on extinction and recovery rates, given the huge loss in land for all but the most hypoxia-resistant taxa (Fig. 2).

Fig. 2.

Plot of cumulative land area versus physical altitude during the mid-Permian to Late Permian [solid line (30)] and for the present day [dashed line (20)]. This relationship enables us to estimate the percentage of the land surface that was physiologically tolerable (20) for the hypothetical species shown in Fig. 1C. A species capable of surviving up to 6.0 km in the mid-Permian would been restricted to below 0.3 km by the end of the Permian (gray dotted line) and thus would have been able to occupy less than half of the available land area. Less hypoxia-tolerant species would have gone extinct; more tolerant ones would have suffered minimal area loss.

The reasonableness of our scenario of hypoxia-induced altitudinal compression depends fundamentally on the accuracy and timing of estimates of percent O2 (8). Those estimates, of course, have uncertainty (8, 20), and other models can yield somewhat different estimates (12, 13).

Our analyses also assume that O2 restricted the upper altitudinal limits during the Late Permian and Triassic, that terrestrial vertebrates then had modest hypoxia tolerance, and that adaptation to hypoxia was limited. None of these assumptions can be directly tested as yet, but our compression hypothesis leads to predictions and patterns that are subject to test:

  1. Terrestrial vertebrates that did survive the Late Permian should show respiratory adaptations for hypoxia. Indeed, morphological traits of Lystrosaurus, one of the few surviving therapsids, and of other Triassic dicynodonts have been interpreted as adaptations to hypoxia (16, 31).

  2. Hyperventilation in response to hypoxia would have elevated respiratory water loss during the Late Permian. Maxilloturbinates, which reduce respiratory water loss, first evolved at that time in mammal-like reptiles (32). Turbinates have been interpreted as indicating high respiratory rates associated with endothermy (32), but they might also be indicators of high respiratory rates induced by hypoxia.

  3. Fossil sites for terrestrial ectotherms by the Triassic should be located predominantly at high latitude, where relatively cooler temperatures would reduce metabolic requirements for O2 and ameliorate hypoxia. Distributional data for therapsids are consistent (33). However, therapsids might also have been restricted to high latitude simply because low-latitude sites were hot, dry, and had a depauperate flora (34).

  4. The compression-forced isolation of taxa into refugia could help to explain Late Permian endemism (33), which is surprising for a Pangaean supercontinent. It could also help to explain the “Lazarus effect” (i.e., the reappearance of taxa that disappeared from the fossil record) noted in this period (1, 35).

The extinctions during the Late Permian were the largest ever during the history of life on Earth. An abrupt climate change is thought to be responsible for the catastrophic extinction. Nonetheless, hypoxia may have contributed to the high background extinctions and high faunal turnover occurring before the mass extinction, to the final catastrophe itself, and to the delayed recovery. Low O2 levels would have promoted extinction and delayed recovery directly via physiological stress (16) and indirectly via altitudinal compression.

Note added in proof: A forthcoming paper (36) provides new and fine-scaled estimates of O2 levels during the Late Permian and Early Triassic. The drop in O2 during the Late Permian is slightly more recent than that in Fig. 1A, but still suggests that hypoxia and altitudinal compression were influential both before and after the mass extinction.

Supporting Online Material

Materials and Methods

Figs. S1 and S2


References and Notes

View Abstract

Navigate This Article