Introduction Willamette Valley wetland prairies are some of the most endangered ecosystems in the United States, and provide habitat for many federally listed species (Wilson et al. 1993, Noss et al. 1995, USFWS 2000, Schultz et al. 2003). A rigorous program of ecosystem restoration can protect and enhance wetland plants, animals, and services (Pywell and Putwain 1996). The restoration projects of the West Eugene Wetlands Program (WEWP) comprise one of the few large-scale and long-term integrated restoration programs in the world. Successful ecosystem restoration requires establishing and maintaining native plants. In turn, plant establishment hinges on having suitable environmental conditions, using species with adequate germination and growth rates, and reducing competitive pressure from non-native plants (Figure 1). In year one of this project, we synthesized the wealth of plant establishment data during wetland restoration in the West Eugene Wetlands Program (Wilson 2004). In year two, our objectives were to expand on these results in several important ways: • Generalize these results through the investigation of plant traits (Table 1) that consistently correspond to the patterns of establishment and vigor (Figure 1). • Systematically compile the results into a trait database. This database includes findings from similar ecosystems, both in the Willamette Valley and elsewhere. • Testing the effect of habitat variation on the relationship between traits and seedling establishment patterns. Our goal is to predict key aspects of prairie restoration performance, in this case establishment rates, based on species traits and habitat. These predictions can then be applied as management recommendations, such as which species to sow to maximize native plant abundance at a given site, even if the species have not been field tested. The two components of our project–plant traits and the database–are crucial to this goal. • Without the generalization that traits allow, understanding of wetland restoration increases slowly and expensively, one case study at a time. • The organization of the database will increase the power and efficiency of revealing the relationships between plant traits and plant performance. Perhaps even more important is the role of the database in developing a Web-based expert system for managers wishing to plan wetland restorations. Species and establishment rates For this project we expanded the Willamette Valley Prairie Plant Trait Database to include additional species important to the WEWP restoration projects and additional traits important to understanding and prediction establishment success. Traits, Final Report Page 2 Wilson, Clark and Roberts We selected 31 species (Table 2) that represent the range of establishment rates found among all species in the WEWP restoration sowings (Tables 3-5), a range of sowing rates, and a variety of traits. Establishment rate for this project is defined as cover in the second year after sowing per weight of seed sown per area (Wilson 2004). The units of establishment rate are cover (%) per g seed sown per cm2. The highly skewed establishment rates required transformation to meet the assumptions of statistical analyses. Establishment rates for all species within a data set (Vernal pool habitats, Deschampsia-dominated habitats, Emergent wetland habitats, all habitats combined) were rank- transformed. Higher ranks represent higher establishment rates. Because various number of species established in each data sets (Table 4), the value for the top rank varied. In practice, several species had zero establishment within a habitat, so tied for lowest establishment with a rank >1. The ranks for the 31 test species were extracted from this larger set of species ranks. Ranks for species with zero establishment were further differentiated into species that established in no habitats, which were given a rank of 0 (Table 6). The WEWP has its own Seeding Assessment rating system based on perceived success (E. Wold, pers. comm.). We updated these ratings of species with current propagation information (WEWP 2005) to create an additional ranking of the test species (Table 6). Traits The Willamette Valley Prairie Plant Trait Database is a compilation of ecological information on species important in upland prairies, wetland prairies, vernal pools, and emergent wetlands. The information is directly pertinent to prairie and wetland restoration. The Database currently contains more than 3800 data points on 134 species. The Database provided information on many of the traits of interest to this project (Table 1), such as clonal spread and growth form. We greatly expanded information in the Database for the 31 test species by measuring seed characteristics and plant growth under standardized growth chamber conditions. Methods for growth analysis Growth chamber procedures followed the general recommendations of Hendry and Grime (1993). Standardized conditions include specifications for germination media, transfer of germinants, pot size, growing media, nutrient solutions, growing illumination and temperatures, and dates of harvest. The use of standardized conditions allows us to integrate our results with those in the scientific literature. We used appropriate dormancy breaking strategies and optimal germination conditions to provide approximately 40 germinants for each test species. When the radical of the germinant was 1 mm long, we transferred the germinant to prepared pots filled with water-washed sand. Each pot was watered with 5 ml of Hoagland's solution three times per week. The growing Traits, Final Report Page 3 Wilson, Clark and Roberts regime in the growth chamber (Table 7) followed the general recommendations of Hendry and Grime (1993) and was monitored continuously with automatic sensors. Seedlings were harvested at 7 days or 21 days. Leaf area was measured from scanned images of fresh leaves taped flat. Root mass, shoot mass, and leaf-only mass were determined by weighing material that had been dried at 80 C for 48 hours. Measurement of seed mass, dimensions, and laboratory germination rate We measured seed mass in five replicate batches of 100 seeds, after drying at 80 C for at least 48 hr. Included in the mass measurements is the seed, proper, and any fruit structures that disperse with the seed, such as awns and perigynia. Thus mass measurements are of dispersules. Seed (dispersule) dimensions were measured so that we could calculate sphericity. Length, width, and thickness of the dispersule were measured with a caliper or micrometer. Sphericity was calculated as the variance of these three measurements, after each had been divided by the largest of the three values (Thompson et al. 1993). We determined laboratory germination rates under favorable conditions for each species. Seeds of each species were sowed into germination boxes that were lined with thick germination paper and filled half-way with moistened sand. For those species requiring stratification (Table 8), seeds were allowed to imbibe, then place at 2 C - 5 C for the required number of weeks. For those species requiring scarification, we removed small chips of testa by gently rubbing seeds between medium sand paper. Prepared seeds were added to germination boxes that were half-filled with washed sand. Temperature and light conditions were set to those favorable to the species being tested (Table 8). Trait values for species The new measurements and information from the Database provided for the 31 test species values in seven categorical traits, seven growth-analysis traits, and six other quantitative traits (Tables 9 and 10). Notice that these traits are characteristic of the species and independent of the particular field conditions at the WEWP restoration sites. Relationships between plant traits and field establishment rates Our prime objective is to determine which plant traits consistently correspond to the patterns of establishment, and to test the effect of habitat variation on these relationships. We examined these relationships with a series of statistical models. Traits, Final Report Page 4 Wilson, Clark and Roberts The standardized traits varied widely in their relationships with field establishment rates, both overall and in the three habitats (Table 11). The strongest relationships were with the categorical traits dealing with form (Life form and Growth form) and with the quantitative traits of Flowering peak month and Seed mass. Flowering peak month and Seed mass were also consistent in the direction of relationship across habitats: later flowering and smaller seeds were always associated with higher establishment rates. Growth analysis traits provided little explanatory power for establishment rates – a real puzzle considering that early growth should be crucial for establishment success.. Nature is multivariate, so we developed multivariable models of field establishment rates vs. standardized traits. Stepwise procedures and model comparisons were used to select the most parsimonious model with highest explanatory power, using a Mallows Cp procedure. The results models successfully explained 50%-70% of the variability in the four data sets (Table 12). This high level of explanatory power shows that standardized traits can explain patterns of field establishment. Moreover, the traits that explain variability were rather consistent across models. Flowering peak month Flowering peak month was significant as a single variable in all four data sets, and was part of the final model for three (All