Effects of herbivory on N-cycling and distribution of added 15NH4+ in N-limited low-alpine grasslands. Plant Soil.Vegard Martinsen*, Gunnar Austrheim, Atle Mysterud & Jan Mulder.
*Department of Plant and Environmental Sciences, Norwegian University of Life Sciences,P.O. Box 5003, NO-1432 Ås, Norway.e-mail: .
Online resource 4Parameter estimates of N-content (% total-N, ln-transformed), 15N (‰) natural abundance,proportional tracer recovery (ln-transformed) and 15N atom% excess (percentage increase in atom% 15N of enriched samples relative to the background atom%, ln-transformed) of different plant and soil components (Alchemilla alpina, Vaccinium myrtillus, Avenella flexuosa, Nardus stricta, surface layer, soil from the O-horizon and bulked plant roots >2 mm) within grassland habitats, Hol, Norway. The estimates derive from linear mixed effect models (REML-estimation; minimum adequate models) with block and enclosure (see material and methods) as random factors. The variance components of the random effects are included in the table. The full models (i.e. prior to backward selection; see material and methods) are presented in square brackets.The fixed effects (i.e. explanatory variables) of the N-content models include month [three levels for A.alpina, A.flexuosa, N.stricta, soil and roots: June 2008 (J08), August 2008 (A08) and August 2009 (A09); and two levels for V.myrtillus and surface layer: September 2009 (S09) and August 2009 (A09)] and treatment (3 levels of grazing; control, low and high). The 15N natural abundanceinclude component (7 levels; A. alpina, V. myrtillus, A. flexuosa, N. stricta, surface layer, O-horizon and bulked plant roots >2 mm) as the only fixed effect factor, as grazing treatment (full model) was not included in the minimum adequate model. The proportional tracer recovery models include treatment and a continuous variable; N-pool of the individual plant species and of the roots [ln(N-pool/g m-2)]. The fixed effects for the 15N atom% excess models are the same as those for the N-content models (Vaccinium myrtillus and surface layer are not included because not sampled A08). N-pools were not included due to lack of data in 2008.The table shows se and adjusted p-values (difference from 0). Bolded values indicate significant differences (significantly different from 0). Number of observations (n) is shown.Differences between factor levels are based on “treatment contrasts”.Treatment contrasts is the default in R, with factor levels arranged in alphabetical sequence [Crawley ( 2007), pg. 377]. Thus the intercept will be the level that comes first in the alphabet and the parameter estimates of the different factor levels are reported as differences between means relative to the intercept. The only exceptions are in the N-content models and 15N atom% excess models, where differences between months for V.myrtillus and surface layer are based differences to September (intercept) and not June.
aThree plots (2 from enclosure D; low and 1 from enclosure H; high) with N-contents of 3.36, 2.45 and 2.53 % were omitted from the analysis in order to fulfil the model criteria.This significantly affected the model. When including the 3 plots, treatment was not significantly affecting the N-content of N. stricta.bOne plot from enclosure H (high grazing pressure) was omitted from the analysis but did not affect the fixed effect structure or the difference between factor levels in the model. cBD missing. dTwo outliers were omitted to fulfil the model criteria. This did not affect the model structure.