Habitat Compartmentation and EnvironmentalCorrelates of Food Chain Length Briand and Cohen (1) condude that "the primary decomposers (bacteria and saprodimensionality of the environment influ-phytic fungi) or do not have phytoplankton ences mean or maximal [food] chain length distinguished from zooplankton. more than environmental variability" but doWe find that the concept of habitat dinot offer an explanation. After examining mensionality lacks sufficient rigor to be used the first 40 food webs that Briand and in a standardized manner. In the study by Cohen present (1), we find that most of the Briand and Cohen, three-dimensional (soldifference in chain length between habitats id) habitats include lakes, oceans, and forests of different dimensions appears to be an (including kelp beds), whereas two dimenartifact of the completeness of the web sional (flat) habitats include creeks, rivers, descriptions. Our calculations indicate that intertidal zones, marshes, grasslands, deserts the first 40 webs are an adequate sample, as and tundra. Habitats with both two-and the range and median chain lengths of webs three-dimensional aspects are considered to 1 through 40 are similar to those of webs 1 have mixed dimensions. Habitats may apthrough 113 (Fig. 1).pear to us as solid or flat; however, we Many of the webs presented by Briand question whether organisms within the haband Cohen are truncated. In the first 40 itats make this distinction. For example, the webs, 17% of the 138 "producers" are actu-Long Island salt-marsh (estuary) includes an ally consumers. For example, the Aspen air column for birds, a water column large parkland community food web (2) produc-enough to support pelagic organisms and ers include primary producers, but also con-plankton, and a flat bottom for molluscs and sumers, for example, coots, ducks, mice, and water plants; yet We do not mean to criticize the original missing. In the New Zealand salt-meadow food web studies, since their objectives did (3) the low mean chain length (1.96) results not include having the webs subjected to from single-link chains that portray orga-structural analyses; however, the completenisms such as weevil larvae, Hemiptera, ness of Briand and Cohen's descriptions of harpacticoids, staphylinids, dipterous larvae, the food webs is confounded with the webs' haplotaxid worms, oribatid mites, bumble-dimensions. In the three-dimensional habibees, adult Hymenoptera, and redpolls as tats of Briand and Cohen, phytoplankton top predators. Jones (4) described a web for are differentiated from zooplankton and the River Clydach that included predatory generally include top predators [see webs fish, but Briand and Cohen and others (5, 6) 17, 19-21, 24, 25, 27, 29-32, and 40 (1, use a simplified web for this system, in 5)], whereas the two-dimensional habitats which predatory fish and some intermediate [webs 3, 10-13, 23, 34, and 35 (1, 5)] do consumers are deleted. Numerous webs are not have plankton differentiated and lack missing predatory birds and insects and top pred...