Ecological Archives --A3

Jessy Loranger, Sebastian T. Meyer, Bill Shipley, Jens Kattge, Hannah Loranger, Christiane Roscher, and Wolfgang W. Weisser. 2012. Predicting invertebrate herbivory from plant traits: evidence from 51 grassland species in experimental monocultures. Ecology 93:2674–2682. http://dx.doi.org/10.1890/12-0328.1

Appendix C. Details on the imputation analyses used in this manuscript.

Table C1. List of the traits for which some values necessitated imputation before those variables could be use to predict leaf standing herbivore damage observed in monocultures at the field site of the Jena Experiment (Germany).

Variables Nb Species missing values Method Reference variables or species
RGR 15 Cirsium oleraceum; Crepis biennis; Galium mollugo; Geranium pratense; Knautia arvensis; Onobrychis viciifolia; Pastinaca sativa; Pimpinella major; Plantago media; Primula veris; Tragopogon pratensis; Trifolium fragiferum; Trifolium hybridum; Veronica chamaedrys Multiple imputation Seed mass; Beginning of seed shedding; Leaf lifespan; Shoot nitrogen concentration; Trampling tolerance
Leaf phosphorus concentration 6 Alopecurus pratensis; Cirsium oleraceum; Pastinaca sativa; Plantago media; Primula veris; Trifolium fragiferum Multiple imputation Leaf dry matter content; Leaf hemicellulose content; Leaf water-soluble content
Stem growth form 1 Taraxacum officinale Comparison Leontodon hispidus; L. autumnalis
Seed shedding height 3 Phleum pratense; Poa trivialis; Taraxacum officinale Multiple regression 7.51997 + 0.3019 × Maximum height + 0.45872 × Minimum height + 0.05004 × Typical height + 0.52805 × Vegetative height
R2 = 0.8525
Beginning of seed shedding 6 Centaurea jacea; Galium mollugo; Geranium pratense; Medicago x varia; Taraxacum officinale; Tragopogon pratensis Comparison Beginning of flowering + 1
Period of seed shedding 6 Centaurea jacea; Galium mollugo; Geranium pratense; Medicago x varia; Taraxacum officinale; Tragopogon pratensis Comparison Period of flowering + 1,5

Notes: Nb: Number of missing values for a given trait; Method: "Comparison" means that the imputed variable has simply been compared with a similar complete variable or a similar species to deduce the missing values. "Multiple regression" means that the imputed variable has been regressed against several correlated variables to get an equation predicting the missing values. "Multiple imputation" means that the imputation has been done following Su et al. (2011); The fourth column lists the variables (or species) that have been used to impute the missing values of a given trait.


Literature Cited

Su, Y.-S., A. Gelman, J. Hill, and M. Yajima. 2011. Multiple Imputation with diagnostics (mi) in R: opening windows into the black box. Journal of statistical software 45:1–31.


[Back to E093-248]