Causes of the large variation in bryophyte species richness and composition among boreal streamside forests

Questions: Boreal forests along small streams are bryophyte diversity hotspots because they are moist, productive and relatively high pH. Do these factors also explain the large differences in species richness and species composition found among streamside sites? Do the species of species-poor sites represent nested subsets of the species of more species-rich sites? How do the results apply to conservation?Location: Forests along small streams in mid-boreal Sweden.Methods: Survey of the flora of liverworts and mosses and habitat properties, including calculation of a pH-index based on species indicator values, in 37 sites (1000-m² plots).Results: The number of bryophyte species per plot ranged from 34 to 125. Neither soil moisture nor basal area of trees (a proxy for productivity) correlated significantly with species richness and composition, whereas pH-index and cover of boulders did. Species richness and composition were more strongly correlated with pH-index for mosses than for liverworts. The richness and composition of bryophyte species most frequently found on moist ground, stream channel margins and, most unexpected, woody debris were all more strongly associated with the pH-index than with other habitat properties. Although species composition was significantly nested, there was still some turnover of species along the first ordination axis.Conclusions: To attain high numbers of species, streamside forests need to have boulders and at least pockets with higher soil and stream-water pH. The number of Red list species was weakly correlated with total species richness and the most species-rich sites contained many species found more in non-forest habitats. Hence, bryophyte conservation in streamside forests should not focus on species-rich sites but on the quality and quantity of substrate available for assemblages of forest species that are strongly disfavoured by forestry.Abbreviation: LOWESS == Locally Weighted Regression Scatterplot Smoothing.Nomenclature: Sööderströöm & Hedenääs (1998); Sööderströöm et al. (2000).