SPATIAL STATISTICS OF FOSSIL OCCURRENCES AND IMPLICATIONS FOR GEOGRAPHIC RANGE RECONSTRUCTION

The geographic ranges of organisms are abstractions based on observed occurrences (Gaston 2003) and/or predicted from environmental controls on their distribution (species distribution modeling). The recorded occurrences of fossil organisms are a complex product of their original time-averaged true ranges and of a plethora of overlaying taphonomic and geological processes acting at multiple spatial and temporal scales. Coordinates of fossil localities in the conterminous United States at epoch or period levels of resolution were downloaded from the Paleobiology Database. Their spatial distributions were analyzed in ArcMap and PASSaGE, using a variety of point pattern statistics. Lacunarity analysis, which measures the distribution of gaps among sites, indicated that the overall patterns of spatial distribution of sites are fractal for every studied interval, consistent with a hierarchical multi-scaled control on their occurrences. Ripley’s K (second order statistics) compares the average number of points within a given distance of each point compared to the number expected under a Poisson process. This analysis indicated clumping at all scales, with maximum deviation from random expectation at about 200 km for most intervals examined. We also determined the distribution of distances of localities from the nearest road. The greatest deviation from random expectation occurred at distances of under a kilometer, signifying the importance of roadcuts and accessibility in determining the location of collecting sites. Despite this, many sites were located far distant from road access, indicating that the geologic distribution of rich sites often outweighs accessibility. These results suggest that using reconstructed geographic ranges in interpretations of ecological and evolutionary processes first requires an understanding of the magnitude and causes of spatially related preservational biases.