Tuesday, June 23, 2009

Good news for paleontologists?

Paleontologists, as most folks know, study fossils (or, more generally, the evidence of past life of any kind). By examining the types and distributions of fossils in rocks of various ages, paleontologists can give us insight into how life on Earth has evolved. Thanks to the study of fossils, we know, for example, that Cambrian oceans were full of trilobites, that the Mesozoic Era was dominated by giant reptiles, and that giant "terror birds" once roamed South America.

Yes, fossils are undoubtedly vital to our understanding of life on Earth. However, although fossils are the only evidence we have for the existence of past life, they have--like all evidence--limitations. Foremost among these is the preservation bias. There's a reason nearly all the fossils you'll see in a museum or private collection are fossils of shells, bones, and teeth: hard parts are much more likely to fossilize than are soft parts.

This means that critters like the sea squirt and the cuttlefish, cute though they may be, are unlikely to appear in the fossil record. Their bodies are entirely (or almost entirely) made of soft tissue, which decays rapidly once they die. About the only way soft tissue can be preserved is through mummification or other direct preservation methods; and they are pretty darned uncommon.

Size and depositional environment also play a role in preservation bias. Larger body parts may be more likely to be preserved and fossilized than are smaller body parts, because it takes large parts longer to break down (thus allowing them more time to be buried and mineralized--although this isn't a hard-and-fast rule). Similarly, critters that die in the water are much more likely to be preserved, because they're more likely to be buried before they decay completely.

Ultimately, preservation bias means that our understandings of life on Earth are inevitably biased toward largeish, ocean-dwelling animals with shells, bones, and/or teeth. This is why we know so much more about trilobites than we do about, say, ancient jellyfish.

Of course, paleontologists acknowledge this problem, and make attempts to compensate for it. One way to try to compensate for preservation bias is to use so-called "live:dead" ratios. For example, suppose in a particular ocean ecosystem 30% of the animals are bivalves, 25% are bony fish, 35% are crustaceans, and 10% are "squishies" such as anemones and jellyfish. That critter composition is known as a "live assemblage" or a "life assemblage" for that ecosystem. (I made up those numbers. They probably bear almost no relation to realistic numbers--and those particular types of critters may not occur together. Just bear with me for the sake of demonstration.) To try to correct for preservation bias, a scientist might count the number of dead critters in each category. (I should note that this type of analysis would be based on numbers of individuals, not numbers of remains--so two clam shells would count as one clam, for example.) This "dead assemblage" or "death assemblage" can then be compared to the life assemblage to figure out relative preservation rates. If, for example, 30% of the living critters are bivalves, but 40% of the remains are bivalve remains, then bivalves would have a higher preservation rate than other critters in the ecosystem.

Potential problems with this method are probably obvious: How do you know which types of modern environments to compare ancient remains to? How do you know that preservation rates in remains are the same as fossilization rates? How do you know preservation rates for different types of critters are the same today as they were then? What happens if the ecosystem changes rapidly--do the death assemblages still accurately reflect the life assemblages?

In the May 22 issue of Science, Western and Behrensmeyer present data that may help to address the last two of these questions. They used a 40-year record from the Amboseli ecosystem in Kenya to study the relative preservation rates for large mammal (15 kg-4000 kg) bones. Previous studies have shown that the life and death assemblages for these mammals are similar at specific points in time; that is, at a given time, the proportions of different species in the life assemblage are similar to those in the death assemblage.

A variety of factors have caused the Amboseli environment to change quite rapidly since the 1960s. Woodlands have shrunk, grasslands have expanded, and swamps have doubled in size. These environmental changes, in addition to direct human actions, have substantially affected the mammal populations in Amboseli during that time. The ratios of different types of organisms--grazers vs. browsers, for example--have changed as a result, and overall species diversity has declined.

Bone and live animal surveys were conducted during two time periods: 1975-1976 and 2002-2004. The bones studied during those times could be separated into subintervals based on how long ago the animal died; this allowed the researchers to divide the samples into four subintervals (1964-1969, 1970-1976, 1993-1998, and 1999-2004). They also used census data to determine the numbers of live animals in various groups during those same time intervals.

For each of the time periods, the researchers compared the proportions of different organisms in the life assemblages with those in the death assemblages. They used these data to determine how well the death assemblages "track" or represent the life assemblages. What they found is pretty interesting:

Statistically significant correlations between live populations and bone counts for the different time intervals indicate that organisms that make up a larger fraction of a living community also make up a proportionally larger fraction of the bone assemblage for that community. In other words, at least for this ecosystem, you can use the death assemblage as a pretty direct proxy for the life assemblage--if 50% of the individuals represented by the death assemblage are medium-sized grazers, then you can infer that about 50% of the organisms in the ecosystem (on average) over the time period you're looking at were medium-sized grazers. You can also use the death assemblages to study how populations in the ecosystem changed over time; the ratios of grazers to browsers in the death assemblages roughly paralleled those in the life assemblages for the same time period. They were able to distinguish changes in population composition over time scales as small as 5 years; they were even able to "predict" ecological structure from the death assemblages (and those predictions were largely confirmed by the life assemblages).

Western & Behrensmeyer's data could be very useful for paleontologists, particularly large-vertebrate paleontologists; the data suggest that bone distributions in death assemblages can be used to infer population and community structures for ancient ecosystems. With some assumptions about ecolosystem properties, bone assemblages can also be used to infer other properties of ancient ecosystems, such as species richness and productivity.

Obviously, these data have limitations; Amboseli is a relatively dry terrestrial ecosystem populated by relatively large mammals, so it's not clear whether the same correlations apply to marine ecosystems, wetter (or drier) terrestrial ecosystems, or to those inhabited primarily by smaller organisms or invertebrates. Additionally, because all of the remains studied were relatively recent (40 years isn't long enough to produce fossilization or even significant burial in most terrestrial ecosystems), it's not clear how the processes of preservation, burial, and fossilization might affect the death assemblages. (Although they do note that partially buried bones--a "pre-fossil" assemblage--seem to show the same correlations as unburied remains.) But studies like these are still very important in determining the error bars (accuracy) of ecosystem studies based on fossil assemblages.

Their data also suggest that studies of death assemblages in modern ecosystems can be of use to scientists studying the effects of human actions and other phenomena, as well as to those wishing to confirm (or obtain) estimates of vertebrate population sizes and compositions.

Western, D., & Behrensmeyer, A. (2009). Bone Assemblages Track Animal Community Structure over 40 Years in an African Savanna Ecosystem Science, 324 (5930), 1061-1064 DOI: 10.1126/science.1171155

No comments: