Species Distribution - Statistical Determination of Distribution Patterns

Statistical Determination of Distribution Patterns

There are various ways to determine the distribution pattern of species. The Clark-Evans nearest neighbor method can be used to determine if a distribution is clumped, uniform or random. To utilize the Clark-Evans nearest neighbor method, researchers examine a population of a single species. The distance of an individual to its nearest neighbor is recorded for each individual in the sample. For two individuals that are each other's nearest neighbor, the distance is recorded twice, once for each individual. To receive accurate results, it is suggested that the number of distance measurements is at least 50. The average distance between nearest neighbors is compared to the expected distance in the case of random distribution to give the ratio:

If this ratio (R) is equal to 1, then the population is randomly dispersed. If R is significantly greater than 1, the population is evenly dispersed. Lastly, if R is significantly less than 1, the population is clumped. Statistical tests (such as t-test, chi squared, etc.) can then be used to determine whether R is significantly different from 1.

The Variance/Mean ratio method focuses mainly on determining whether a species fits a randomly spaced distribution, but can also be used as evidence for either an even or clumped distribution. To utilize the Variance/Mean ratio method, data is collected from several random samples of a given population. In this analysis, it is imperative that data from at least 50 sample plots is considered. The number of individuals present in each sample is compared to the expected counts in the case of random distribution. The expected distribution can be found using Poisson distribution. If the variance/mean ratio is equal to 1, the population is found to be randomly distributed. If it is significantly greater than 1, the population is found to be clumped distribution. Finally, if the ratio is significantly less than 1, the population is found to be evenly distributed. Typical statistical tests used to find the significance of the variance/mean ratio include Student's t-test and chi squared.

However, many researchers believe that species distribution models based on statistical analysis, without including ecological models and theories, are too incomplete for prediction. Instead of conclusions based on presence-absence data, probabilities that convey the likelihood a species will occupy a given area are more preferred because these models include an estimate of confidence in the likelihood of the species being present/absent. Additionally, they are also more valuable than data collected based on simple presence or absence because models based on probability allow the formation of spatial maps that indicates how likely a species is to be found in a particular area. Similar areas can then be compared to see how likely it is that a species will occur there also; this leads to a relationship between habitat suitability and species occurrence.