Comment: Critical examination of stable isotope analysis as a means for tracing carbon pathways in stream ecosystems

Abstract

On the basis of a review of current literature, France (1995) questioned the utility of stable isotope analysis (SIA) in describing food webs and understanding the effects of human perturbation in lotic ecosystems. His three main conclusions were (i) “autotrophic pathways within forested headwaters are much more important to lotic food webs than would be suggested by their particulate inputs alone,” (ii) “the great variability in attached algal δC may often preclude use of SIA for identifying carbon pathways in stream ecosystems,” and (iii) “the utility of carbon SIA in understanding anthropogenic alterations to the carbon budget of streams is presently minimal.” The first of these conclusions may well be correct, but points ii and iii appear to be based on a fundamental misunderstanding of the use of SIA. We have examined the same reports considered by France, as well as the wider literature concerning SIA, and feel that SIA can be a powerful technique for deciphering food webs and identifying impacts of alterations in land use. France’s attempt “to search for general patterns that transcend individual studies” by pooling δC data on carbon sources and animal consumers from freshwaters throughout the world is fundamentally flawed. While allochthonous organic matter is remarkably uniform in its isotopic signature, δC values of aquatic plants are widely recognized to be extremely variable between sites (Deines 1980; Osmond et al. 1981; Boutton 1991). It would be very surprising if there were any consistent pattern of C depletion or enrichment in the δC values of aquatic algae on a global scale. The fact that allochthonous δC values (≈–28‰) fall within the global range of reported autochthonous δC values (–46 to –22‰) in no way invalidates SIA as a means of assessing the importance of the two potential sources of energy. Rather, it means that such evaluations must be site specific (Rosenfeld and Roff 1992), and they are easily interpreted only when the potential carbon sources are isotopically distinct (Bunn et al. 1989). Since France bases his conclusions on pooled data, and not on site-specific algal and leaf litter δC values, no conclusions regarding the utility of SIA and the carbon dependencies of stream animals are possible. In addition, France fails to mention the importance of trophic fractionation in the interpretation of faunal δC values. Animals are usually enriched by ≈+1‰ relative to their diets (DeNiro and Epstein 1978), so that primary consumers are expected to be enriched by ≈+1‰, secondary consumers by ≈+2‰, and top predators by ≈+3‰ compared with their primary food sources (Rau et al. 1983). If one does not consider the trophic position of the organisms that are analyzed (France apparently expected the δC values of invertebrates and fish to be exactly the same as that of the primary carbon sources), no statement can be accurately made as to which animals are outside (or inside) the range of allochthonous dependence. There are similar problems with France’s critique of SIA in understanding anthropogenic alterations to the carbon flow of streams. After pooling data on “similar” species from forested and open sites, he found that (i) faunal δC could increase, decrease, or remain unchanged, and (ii) in almost 80% of the cases there was little or no change in faunal δC with landscape modification. The only surprising aspect of this is France’s expectation of consistent decreases in faunal δC values as evidence of increased utilization of autochthonous carbon following riparian deforestation. We agree that there have been too few analyses of autochthonous carbon sources, but it is precisely the “confounding by environmental co-determinants” that disallows France’s direct comparison between the δC values of similar species at forested and open sites: “although there is strong temptation to compare absolute