Neighbor proximity analysis; a technique for assessing spatial patterns in the fossil record

Time-averaging can be a major obstacle in reconstructing fine-scale ecological processes in the fossil record. This study presents a technique, Neighbor Proximity Analysis that attempts, despite time-averaging, to elucidate fine-scale ecological information on a single bedding plane. The technique takes advantage of the tendency for most sessile organisms to cluster in conspecific clumps, and assumes that time-averaging will tend to overprint patterns of conspecific clustering given sufficient time. The fossil species of interest is mapped onto an appropriate grid, point-counted, and the number of conspecific neighbors for each individual is counted. The community distribution is then permuted randomly for multiple iterations; and for each iteration, the number of randomly generated conspecific neighbors is compared to the observed number of conspecific neighbors. Taxa that exhibit clustering greater than expected under the random model probably clustered in life, revealing features of the local spatial structure. An example of the method is presented, using bedding planes from the Rapid Member of the Little Cedar Formation at the Devonian Fossil Gorge, Coralville, Iowa. The results indicate that of the common brachiopod present, Spinatrypa bellula clustered conspecifically during life.