Community-wide germination strategies in a temperate rainforest of Southern Chile: ecological and evolutionary correlates

Delayed seed germination ('dispersal in time'), as a component of a plant's germination strategy, was studied in dicotyledoneous species of a temperat e rainforest flora in Chiloe Island (42 degrees 30' S), southern Chile. The objective of this investigation was to assess, for this temperate rainfore st flora, what proportion of interspecific variation in the time of seed ge rmination-measured in days since the onset of seed dispersal in space-could be attributed to the plants' historical and phylogenetic background and wh at proportion was associated with life history and ecological attributes su ch as seed mass, life form, dispersal syndromes and dispersal periods. To c haracterise germination times for 44 species from Chiloe forests (n = 150 s eeds sowed per species in laboratory assays), we computed the mean germinat ion time (GT), in days since sowing, for all seeds germinated of each speci es. Seeds were taken from the plants at the onset of dispersal and germinat ed in Petri dishes at 10/20 degrees C. Considering all species, GTs varied between 3 and 385 days and presented an L-shaped frequency distribution. On e-way ANOVAs measured the effects of each factor across all other variables . Two-way ANOVAs were used to assess significant interactions between facto rs. Multifactorial ANOVAs were used to evaluate the independent effects of each of six historical, phylogenetic and ecological factors on GT and to de tect associations between factors. In one-way ANOVAs, phylogenetic grouping (at or above order) explained 12% of the variance in GT; dispersal period (summer v. mainly autumn dissemination of ripe seeds), biogeographic elemen t (endemic, austral, neotropical or cosmopolitan) and dispersal syndrome (f leshy v. dry propagules) explained 7, 6 and 5% of the variance in GT, respe ctively. The factors life form (trees, shrub and woody vines combined, herb s and non-woody epiphytes) and seed mass (light v. heavy) explained the 4 a nd 2% of the variance in GT, respectively. Taxa related to Ranunculales pre sented the longest mean GT (148 days). Endozoochorous species had a more de layed germination than species with other dispersal syndromes. Herbs and no n-woody epiphyte species showed mean GT (41 days) significantly shorter tha n trees and shrubs plus woody vines combined (86 and 85 days, respectively) . All interactions in two-way ANOVAs were significant. Multifactorial ANOVA s revealed that the three major factors contributing to differences in GT i n this temperate rainforest flora were phylogenetic relatedness, dispersal syndromes and life form (7, 6 and 6% of the interspecific variation, respec tively). In this analysis, biogeographic element, dispersal period and seed mass were not significantly related to GT. For the factors examined, failu re-time analysis, which takes into account all viable seeds not germinating in laboratory assays, confirmed results from multifactorial ANOVAs.