A synthesis of empirical plant dispersal kernels

• Published in: Journal of Ecology
• Main Authors: Bullock, James M., Mallada Gonzalez, Laura, Tamme, Riin, Götzenberger, Lars, White, Steven M., Pärtel, Meelis, Hooftman, Danny A.P.
• Format: Article in Journal/Newspaper
• Language: English
• Published: Wiley 2017
• Subjects: Ecology and Environment; Antarc*; Antarctica
• Online Access: http://nora.nerc.ac.uk/id/eprint/515569/; https://nora.nerc.ac.uk/id/eprint/515569/1/N515569PP.pdf; https://doi.org/10.1111/1365-2745.12666

1. Dispersal is fundamental to ecological processes at all scales and levels of organisation but progress is limited by a lack of information about the general shape and form of plant dispersal kernels. We addressed this gap by synthesising empirical data describing seed dispersal and fitting general dispersal kernels representing major plant types and dispersal modes. 2. A comprehensive literature search resulted in 107 papers describing 168 dispersal kernels for 144 vascular plant species. The data covered 63 families, all the continents except Antarctica, and the broad vegetation types of forest, grassland, shrubland, and more open habitats (e.g. deserts). We classified kernels in terms of dispersal mode (ant, ballistic, rodent, vertebrates other than rodents, vehicle or wind), plant growth form (climber, graminoid, herb, shrub or tree), seed mass and plant height. 3. We fitted 11 widely-used probability density functions to each of the 168 datasets to provide a statistical description of the dispersal kernel. The Exponential Power (ExP) and Log-sech (LogS) functions performed best. Other 2-parameter functions varied in performance. For example, the Lognormal and Weibull performed poorly, while the 2Dt and Power law performed moderately well. Of the single-parameter functions, the Gaussian performed very poorly, while the Exponential performed better. No function was among the best-fitting for all datasets. 4. For 10 plant growth form/dispersal mode combinations for which we had >3 datasets, we fitted ExP and LogS functions across multiple datasets to provide generalised dispersal kernels. We also fitted these functions to sub-divisions of these growth form/dispersal mode combinations in terms of seed mass (for animal-dispersed seeds) or plant height (wind-dispersed) classes. These functions provided generally good fits to the grouped datasets, despite variation in empirical methods, local conditions, vegetation type and the exact dispersal process. 5. Synthesis. We synthesise the rich empirical ...