Logistic regression was used to model the response curves of seven aquatic bryophyte species from calcareous lowland streams of two hydrographic networks. Almost all the species exhibited log-normal response curves and significant optima when the data coming from the two hydrographic networks were simultaneously considered, but monotonic response curves within each of the hydrographic networks. The response curves of Amblystegium riparium in the Walloon hydrographic network, in the Alsatian hydrographic network, and in both networks considered simultaneously, showed a strong overlap. Conversely, Amblystegium fluviatile showed distinctly different response curves between the studied hydrographic networks. This was also the case for A. tenax, Fissidens crassipes, and Fontinalis antipyretica. Amblystegium tenax has a typical oligotrophic status in the Alsatian floodplain, but in the Walloon hydrographic network, is still present in the most polluted waters whose annual mean concentrations of N-NH4+ can reach up to 9,000 pIg 1-'. Amblystegium fluviatile and Fontinalis antipyretica presented similar response patterns, exhibiting optima in eutrophic waters when the data from the two hydrographic networks were simultaneously considered, but reaching maximal occurrence in the most oligotrophic waters in the Walloon hydrographic network. It is suggested that these species might include several ecotypes with different trophic requirements. Concern over the impact of nutrients on rivers in Europe has increased since the publication of the European Community's Urban Wastewater Treatment Directive (Kelly 1998). It was demonstrated that the natural geochemical content of P-PO43in the large European rivers is of ca 10 pxg 1-', but that it is commonly 10 to 100 times greater in most of the rivers due to anthropogenic eutrophication (Carbiener et al. 1995). Despite efficient urban and industrial discharge sanitation policies, and a great deal of campaign information on the dangers of abusive mineral fertilizer use, the waters of most of the rivers have not as yet recovered the chemical quality they had at the end of the last century. This has been shown by long term monitoring studies in rivers such as the Seine or the Marne (France), where the oxidation of ammonium into nitrates causes nitrogen concentrations to increase further (Cun et al. 1997). The great interest that developed with problems due to nutrient enrichment focused research on eutrophication and has involved a rapid increase in biological methods for monitoring nutrients (see Kelly & Whitton 1998 for review). Weighted averaging is an easy quantitative method for predicting values of an environmental variable from species composition, by averaging the indicator values of species that are present (Ellenberg et al. 1991). The estimation of the species indicator values can be empirically derived or averaged over the values of the environmental variable for the samples in which a species occurs (e.g., Empain 1977; Haury et al. 1996; Kelly 1998). This may, however, give misleading results if the investigated species present a broad ecological range and if the distribution of the environmental variable is not homogeneous over the whole range of occurrence of the species (Ter Braak & Looman 1986). Another alternative to define species indicator values is to estimate the species optimum by fitting a curve to the species data by regression. In the case of presence-absence data, the logistic regression fits a curve giving the probability of occurrence of a species as a function of a given environmental factor. Such a technique was already employed to determine diatoms ecological optima and tolerances used in calibration methods for inferring past chemistry of lake waters (Birks et al. 1990; Oksanen et al. 1988). In aquatic bryophytes, the species response curves have not yet been investigated. The ecophysiological processes explaining their reaction to nutrients remain poorly understood (Bates 1992) and, except for some experimental work (Frahm 1975, 1976; Steinman 1994) and correlative sur0007-2745/99/720-728$1.05/0 This content downloaded from 207.46.13.75 on Fri, 08 Jul 2016 05:28:12 UTC All use subject to http://about.jstor.org/terms 1999] VANDERPOORTEN & DURWAEL: RESPONSE CURVES 721 veys on the relationships between aquatic bryophytes and water trophic level (Vanderpoorten 1999a; Vanderpoorten et al. 1999), few studies have yet been conducted. Hence, Duill (1991) gave no information concerning the trophic requirements of aquatic mosses in the ecological index system of central Europe. The aim of this paper is to 1) model the response curves of aquatic bryophytes from hard lowland running waters along gradients of N-NHjand PPO4-, covering the entire range of the species using logistic regression to provide precise autoecological data on their trophic requirements, 2) calculate species optima and tolerances to these factors in order to provide a basis for simple methods of survey of water trophic level, and 3) compare the species response curves between two different hydrographic networks to investigate the extent to which the species exhibit similar response curves between the basins, and therefore to assess whether methods of biomonitoring based on species optima are transposable from an area to another.
Read full abstract