Geographic information science has advanced passed the point of mere field survey and photointerpretation. Now, there is predictive vegetation modeling where the researchers are able to predict the distribution of vegetation across a landscape based on the relationship between the spatial distribution of vegetation and certain environmental variables (Franklin, 1995; Guisan and Zimmermann, 2000). These relationships can be predicted and we can find correlations based on observation and physiological limitations of certain plant species. In order to better understand this new way to study vegetation patterns, we must consider Tobler's Law. Things in close proximity are closer in relation than those further away. Vegetation that grows near to each other will be more closely related and therefore more easily. This is different from older schools of thought where each observation was individually significant and not to the group. Now, we examine everything in relation to each other and take everything into account including the plants' abilities to grow in certain environments.
The study area in this article is the Mojave Desert, the smallest desert in the United States. Due to the lack of precipitation, there is a high concentration of alkaline soil even with the diversity of soils containing high or low amounts of organic material.
The acronym, CORA, represents a shrubland alliance; it is the most
widely spread in the area on rocky soils, upper bajadas, pediments, and
hill slopes. This alliance is what the researchers focused on the most.
In order to make the most accurate predictions, the sample size must be large enough to account for a lot of variability. This is especially true in cases where it is difficult to control external forces affecting the subject. Relying too much on a small sample can lead to extrapolation. Generalizing too much can often be inaccurate and unhelpful as to a solution to a problem. Also, over-simplification of spatial patterns can lead to a similar fate. The more realistic the maps are, the better chance there is of accuracy. The larger the sample size, the better chance there is of precision. If there is enough mental capacity and usage of software combined, we can accomplish both of these goals and be both accurate and precise in our predictions.
Predictive Vegetation modeling and Speices Distribution modeling is a field where statitics, GISc, and biology all intersect. As you can see with this study, it is deeply founded in GISc taking into account spatial relationships. Many of the parameters for the model are based on the biophysical characteristics of the landscape. Finally the stats, or more specification the regression analysis is where you see the math work. I like this field because it is very integrative and many findings here can apply to other fields. Of course there is a major conservation and planning application of the results, but the results are interesting to statistics and GIS people as well who may be modeling something completely different then vegetation in the desert.
ReplyDelete