Those who work with the past, however, do tend to ignore the very similar problems facing biogeographers whose target organisms are still extant. This morning, though, I came across a paper reporting a piece of research into the use of museum collections to fill gaps in scientific knowledge of biogeography and hence to improve both conservation efforts and ecological understanding.
The research, by Newbold (2010), focuses on one particular technique, called ecological niche modelling, which I have always assumed would be particularly useful to palaeontologists. Essentially, ecological niche modelling was developed to "fill the gaps" in our knowledge of a species' distribution. If we plot every known occurrence of a particular species, for example, the resulting distribution will be incomplete, because we are unlikely to have sampled every possible site where that species might occur. Some of the apparently empty sites on our distribution map, then, will actually represent sites that are not sampled. There are a number of ways we can deal with this. Most simply, we can ignore unsampled sites by assuming they are empty, although this is unreliable (Newbold 2010). Instead, then, models can be developed to assign each unsampled site a value (presence/absence or occupied/empty). This assignment can be random, or it can employ an ecological niche model, which analyses the distribution of known presences in light of their environmental conditions to identify a set of rules that describe an organisms' distribution in terms of its context. So, for example, an ecological niche model might determine that all occurences of species X are in woodland and within 20km of a water body, and then can use these rules to decide which unsampled cells are likely to be occupied.
So far, so good. For palaeontologists, this technique holds potential - it would allow us to patch some of the gaps in species' distributions that are the inevitable result of using fossil data. However, Newbold then goes on to discuss the limitations of museum data in ecological niche modelling, which I had not yet thought much about. Museum collections are exactly what palaeoanthropologists would be working with: our fossils are kept in collections, with location data and environmental reconstructions published in the associated literature. However, as Newbold quite rightly notes, the records kept by curators and museums, particularly where the fossils were discovered a long time ago, may be both biased and even incorrect. For example, those fossils which were found are those which eroded from rock faces, but only where there were people to find them. Certain areas of Africa, for example, are likely to be poorly sampled by palaeoanthropologists because they are politically unsettled or hostile to Western nations, so any fossils that have emerged are unlikely to have been recognised. This is only a problem where the bias favours certain palaeoenvironments and hence affects the rules produced by the model(Newbold 2010), but, as of yet, we cannot know whether this is the case in palaeoanthropology.
In addition, small errors in the location records of fossil finds may also affect our models (Newbold 2010). The Taung child, the famous first fossil of Australopithecus africanus, for example, was famously found in a limestone quarry in South Africa by workers - it's exact location was never noted. In addtion, prior to the use of GPS, many fossil findspots were difficult to locate with the accuracy possible using modern technology. This georeferencing problem is particularly common (Newbold 2010).
Now, I would argue that these problems in their own right do not invalidate ecological niche models, particularly where - as in palaeoanthropology - there are limited opportunities for obtaining better sampling of distributions. But just after finishing this paper, I encountered a second, this time in the Journal of Biogeography, which highlights the dangers of niche modelling in a very different way. The authors, Lozier et al. (2009) have constructed an ecological niche model based on sightings of the sasquatch - bigfoot - to explore whether reasonable distribution models can be constructed from questionable observational data. The distribution their model produced, in fact, was very successful in tests (proving to be capable of producing a set of ecological rules which matched the conditons of all but one of over 500 sightings), and was very similar to that of black bear, despite being based only on uncertain sightings of a creature that has never been proven to exist. The paper, overall, not only gives grounds for serious thought about the use of uncertain data in ecological niche modelling, but also enables its authors to propose that bigfoot may, in fact, be a misidentified black bear....
Clearly, it is very important to critically assess the nature and quality of data used in ecological niche modelling if the technique is to be useful and produce reliable results. This may particularly be the case in palaeoanthropology, where taphonomic processes (which are inherently biased towards certain palaeoenvironments) have been involved, even where we can be pretty sure that the subjects of the model actually existed!
Reference
NEWBOLD, T. (2010). Applications and limitations of museum data for conservation and ecology, with particular attention to species distribution models Progress in Physical Geography, 34 (1), 3-22 DOI: 10.1177/0309133309355630
LOZIER, J., ANIELLO, P., & HICKERSON, M. (2009). Predicting the distribution of Sasquatch in western North America: anything goes with ecological niche modelling Journal of Biogeography, 36 (9), 1623-1627 DOI: 10.1111/j.1365-2699.2009.02152.x
No comments:
Post a Comment