Skip to main content
. 2020 May 21;11:2529. doi: 10.1038/s41467-020-16404-4

Table 2.

Statistics of warming and herbivory effects on NEE800, GPP800, and Re in 2017–2018.

NEE800 F NEE800 P GPP800 F GPP800 P Re F Re P
2017
 Soil OM content 0.2 0.687 6.8 0.026 41.0 <0.001
 Vascular plant cover 9.3 0.014 29.2 <0.001 42.6 <0.001
 Lichen cover 20.5 0.001 8.4 0.015 27.4 0.001
 Moss cover 1.8 0.211 0.1 0.712 7.6 0.024
 Date (D) 46.5 <0.001 121 <0.001 136 <0.001
 Warming (W) 2.0 0.190 4.6 0.062 3.1 0.119
 Herbivory reduction (H) 5.7 0.043 13.2 0.006 11.6 0.010
 W × H 1.0 0.352 3.5 0.094 7.0 0.030
 W × D 2.9 0.029 1.3 0.282 1.3 0.286
 H × D 1.1 0.355 1.6 0.188 0.8 0.541
 W × H × D 0.2 0.916 0.5 0.766 0.3 0.891
2018
 Soil OM content 0.02 0.895 2.4 0.153 10.2 0.011
 Vascular plant cover 0.7 0.411 8.8 0.014 26.6 0.001
 Lichen cover 8.2 0.016 3.5 0.085 2.0 0.185
 Moss cover 5.2 0.046 7.6 0.018 2.8 0.125
 Date (D) 31.5 <0.001 60.8 <0.001 87.4 <0.001
 Warming (W) 13.2 0.006 13.9 0.004 1.6 0.239
 Herbivory reduction (H) 8.3 0.020 13.1 0.006 7.3 0.033
 W × H 0.4 0.534 1.3 0.280 1.7 0.238
 W × D 1.5 0.129 1.2 0.264 2.4 0.008
 H × D 1.3 0.215 1.3 0.223 0.4 0.944
 W × H × D 1.5 0.154 1.4 0.158 0.7 0.775

Treatment effects on net ecosystem exchange (NEE800), gross primary production (GPP800), and ecosystem respiration (Re) were tested using repeated measures linear mixed models and Type I ANOVA (with two-sided significance tests), where the variance is allocated to explanatory variables in the order of their appearance. Soil organic matter (OM) content and cover of vascular plants, lichens, and mosses are continuous variables that describe the variation among the experimental plots prior to the establishment of the experiment. They were used in the models as covariates to remove plot-to-plot variation that might otherwise confound the treatment effects. Years were analyzed separately (N = 100 for 2017, N = 240 for 2018). Date was treated as a repeated measure, warming and herbivory reduction as fixed effects, and treatment block (not reported) as a random effect. F and P indicate F-statistics and P-values respectively; P < 0.05 are in bold.