Spatial Factors Affect Primary Succession on the Muddy River Lahar, Mount St. Helens, Washington

We surveyed vegetation 28 summers after the 1980 eruption of Mount St. Helens, Washington spawned a massive lahar. Our goals were to describe vegetation patterns early in primary succession; determine factors associated with these patterns; and seek evidence for assembly rules. We determined species composition, plot position and surface features in 151 plots. Multiple regression, redundancy analysis (RDA) of the plant to explanatory variables and Mantel tests assessed relationships between habitat and vegetation. We classified plots into nine community types (CTs). The CTs overlapped in both habitat and multivariate space. RDA accounted for only 24% of the species variation, indicating that the vegetation was not fully determined by the measured variables. Elevation and location had the strongest correlations with vegetation composition. The distribution of most species could not be predicted well with our data and appeared stochastic. Priority effects appear to have generated alternative succession trajectories. Plots with distinct composition can occupy similar environments and plots with similar vegetation occur under similar conditions. Species composition cannot be predicted closely from the data available, suggesting that assembly rules are either weak or come into play only as the vegetation matures. DEL MORAL, SANDLER & MUERDTER— 2 Fig. 1. Topographic map of the Muddy River lahar study area), showing locations of 151 sample plots. Dark grey = intact forest, light gray and stipples = open forest damaged by lahar. Contour intervals = 20 ft (6.09 m). Study conducted between 900 and 1320 m a.s.l. Introduction Mount St. Helens remains a unique setting to explore primary succession and to unravel mechanisms that affect community assembly. This paper describes vegetation patterns on a large lahar formed in 1980. Our goals were to provide a detailed description of vegetation patterns; determine factors associated with these patterns; and seek evidence for assembly rules. While a chronosequence study would be ideal to demonstrate floristic and vegetation change (Clarkson 1998; Smits et al. 2002; del Moral 2007), a single survey of vegetation structure can also provide insights into vegetation dynamics. There are surprisingly few detailed studies of early primary succession on volcanoes. This deficiency has hampered understanding of how species assemble (del Moral et al. 2007; Walker and del Moral 2008). Appreciating these processes will streamline the restoration of devastated landscapes (e.g. mine spoils) by illuminating factors that retard invasion, establishment or development (Walker and del Moral 2003; del Moral and Walker 2007; Parsons et al. 2007; Walker et al. 2007). Studies on Mount St. Helens showed that barren sites need physical amelioration (Wood and del Moral 1987), good dispersal (del Moral & Eckert 2005) and establishment (Wood & Morris 1990). Establishment improves in favorable microsites (Tsuyuzaki & Titus 1996; Titus & del Moral 1998). Even early in vegetation development, biotic interactions such as facilitation (Bellingham et al. 2001; Gosling 2005; Herriquez & Lusk 2005), competition (Fraser & Keddy 2005) and herbivory (Bishop 2002; Fagan et al. 2005) can guide and deflect successional pathways. Species colonization is affected by chance, which permits different species to establish in similar habitats. The often-random order of species invasion (i.e. priority effects; D’Antonio et al. 2001) can affect succession (Wiegleb & Felinks 2001). If alien species invade preemptively, trajectories can be fundamentally altered (Eriksson & Eriksson 1998; Seabloom & van der Valk 2003; Corban & D’Antonio 2004). While random events can dominate early species assembly (cf. del Moral & Grishin 1999), more deterministic patterns can develop as a system matures (Hodkinson et al., 2003; Puyravaud et al. 2003). Are vegetation patterns linked to spatial or to environmental factors? If species distributions during early succession were related to landscape effects, we could conclude that dispersal, which is largely stochastic, remains important. If environmental factors better explain vegetation patterns, then deterministic processes may have overcome priority effects, and assembly rules would be strong (Holdaway & Sparrow 2006).