Hawai‘i hosts over 1400 native plant species, 90% of which are endemic. Human land use, relentless non-native species introductions, and recent climate changes all threaten the conservation of unique island biota, especially in zones of montane wet forest that provide an essential ecosystem service for Hawaiians: fresh water. Over the last century, an indigenous peat moss (Sphagnum palustre) has spread unchecked through the Kohala forest on the Island of Hawaiʻi, overtaking large areas of forest floor and altering understory community dynamics. On Oahu, where the introduction of S. palustre was well-documented in the 1960s, optimal conditions have allowed for its colonization, domination, and suppression of native wetland plant communities. For this observational study, we surveyed plant communities across a gradient of S. palustre presence within its native range on the Big Island. In 2016, we visited 115 randomly-placed plots to assess overall S. palustre distribution and abundance, and sampled 32 of these plots at a fine scale for understory and canopy characteristics. Nonmetric multidimensional scaling of plant communities and environmental variables was used for predictor selection, and linear mixed models were built with understory and canopy diversity (richness, Simpson’s and Shannon-Weiner indices) as the primary response variables.
Ordination of understory species cover showed a distinct trend of homogenization in plots with high levels of S. palustre, and the fitting of environmental vectors to the ordinal space resulted in significant correlations with S. palustre presence (P = 0.01), longitude (P = 0.001), and mean annual rainfall (P = 0.002). At the canopy level, nearly all measured environmental variables were significantly correlated with community composition, but S. palustre presence did not exhibit any relationship. Our final linear models for understory diversity indices each suggested a strong negative influence of S. palustre presence (P < 0.001 in all cases). For measures of richness, location (longitude and elevation) and canopy closure also exhibited significant effects, while Simpson’s and Shannon’s indices were more sensitive to pig damage. Patterns in canopy diversity were less clear, suggesting a lag in response time at the canopy level, perhaps due to the recentness of S. palustre spread and the long average lifespan of dominant canopy species. Our results demonstrate the ecosystem engineering capacity of Sphagnum spp., even at sub-tropical latitudes, and they highlight the potentially negative role S. palustre plays in the understory species composition of Hawaiian montane wet forest.