OIKOS 91: 233 – 244. Copenhagen 2000 Eelgrass, Zostera marina, growth along depth gradients: upper boundaries of the variation as a powerful predictive tool Dorte Krause-Jensen, Anne Lise Middelboe, Kaj Sand-Jensen and Peter Bondo Christensen Krause-Jensen, D., Middelboe, A. L., Sand-Jensen, K. and Christensen, P. B. 2000. Eelgrass, Zostera marina, growth along depth gradients: upper boundaries of the variation as a powerful predictive tool. – Oikos 91: 233–244. 1200 measurements of eelgrass (Zostera marina ) biomass, shoot density and cover along 19 depth gradients in Øresund, located between Denmark and Sweden, were analysed to characterise growth of eelgrass in relation to depth. The large data set allowed analyses of boundaries of distribution as well as of average trends. Natural variability is large in shallow water where populations are disturbed by wave action and other physical parameters. Models based on average values, therefore, did not adequately describe growth regulation by resources, and only explained a minor part (up to 30%) of the overall variation in data. In contrast, boundary functions, which describe the upper bounds of distributions, focus on the variation produced by the ultimately growth-regulating resource, and therefore provide models with high pre- dictive power. An exponential model explained up to 90% of the variation in upper bounds of eelgrass shoot density as a function of depth and indicated that shoot density was ultimately regulated by light availability. The boundary functions demon- strated that eelgrass shoot density, biomass and cover followed markedly different patterns as functions of depth and were affected differently by the factors governing their distribution. In addition, boundary functions revealed informative spatial structures in data and illustrated whether a given general trend was caused by changes in maximum values, minimum values or both. For example, upper and lower boundaries of biomass-shoot density relations changed markedly with depth, demon- strating depth-related changes in intraspecific succession and competition patterns. Boundary functions are therefore suggested as a promising tool for analysing ultimate regulating factors of distribution and growth of organisms when large data sets are available. D. Krause -Jensen, A. L. Middelboe, and P. B. Christensen, Dept of Lake and Estuarine Ecology, National Enironmental Research Inst., Vejlsøej 25, DK-8600 Silkeborg, Denmark (dkj@dmu.dk).– K. Sand -Jensen (and present address of ALM), Freshwater Biological Laboratory, Uni. of Copenhagen, Helsingørsgade 51, DK-3400 Hillerød, Denmark. Seagrasses are important structural components and primary producers of shallow, soft-bottom coastal wa- ters throughout the world (Den Hartog 1970, Mann 1982). The lower depth limit of seagrass meadows delimits the coastal area potentially covered by sea- grasses, and several studies have documented that vari- ations in seagrass depth limits among species and sites are mainly regulated by light attenuation in the water column (e.g. Ostenfeld 1908, Backman and Barilotti 1976, Duarte 1991, Sand-Jensen et al. 1994). In contrast to seagrass depth limits, seagrass abundance along depth gradients has not been intensively analysed. Pre- vious studies indicate that seagrasses display either a bell-shaped distribution along depth gradients, with maximum abundance at intermediate depths and lower abundance in shallow and deep water (Bay 1984, Sand- Jensen et al. 1997), or a general decline in seagrass abundance with increasing depth (Buesa 1975, Hulings 1979, West 1990). Exposure, desiccation and ice-scour may reduce seagrass abundance in shallow water Accepted 10 May 2000 Copyright © OIKOS 2000 ISSN 0030-1299 Printed in Ireland – all rights reserved OIKOS 91:2 (2000) 233