Premium
Seed Inputs to Microsite Patch Recovery on Two Tropandean Landslides in Ecuador
Author(s) -
Myster Randall W.,
Sarmiento Fausto O.
Publication year - 1998
Publication title -
restoration ecology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.214
H-Index - 100
eISSN - 1526-100X
pISSN - 1061-2971
DOI - 10.1046/j.1526-100x.1998.00615.x
Subject(s) - melastomataceae , microsite , biology , vascular plant , asteraceae , canonical correspondence analysis , botany , fern , habitat , ecology , species richness , seedling
To understand landslide regeneration and provide information necessary for restoration, we sampled seed rain, seed pool, and plant cover on two Ecuadorian landslides. We trapped 1304 seeds and found that, while most seeds were in the family Asteraceae, there was substantial variation in seed rain among plant families. Four hundred and seventy‐five seedlings emerged from soil samples, including nonvascular and vascular families; again, species in Asteraceae dominated, with species in Piperaceae also very common. Plant cover, consisting of members of four fern families and 20 vascular plant families—with species in Asteraceae, Melastomataceae and Poaceae most common—was scored as a percentage of the total plot area. Principal components analysis (PCA) showed that, for all three of these plant life stages (seed rain, seed‐propagule pool, plant cover), spatial variation was dominated by differences between the two landslides rather than within‐landslide plot differences. PCA also showed that plots separated best on axes defined by the families Cecropiaceae, Urticaceae, Melastomataceae, Papilionaceae, Asteraceae, and Araceae with clumping of families in PCA space suggesting common successional strategies. Another multivariate technique, canonical correspondence analysis (CCA), showed that the combined seed rain and seed pool data could predict the percent cover of the family Verbenaceae and that the current plant cover families could predict Asteraceae seeds and seedlings. Finally, we use our past and present landslide data, along with multivariate modeling results, to suggest strategies for successful landslide restoration.