Running Head: Soil organic matter and N addition

Article Type: Original research

Title: Nitrogen addition changes grassland soil organic matter decomposition

Charlotte E. Riggs1*, Sarah E. Hobbie1, Elizabeth M. Bach2,3, Kirsten S. Hofmockel2, Clare E. Kazanski1

1 Department of Ecology, Evolution, and Behavior, University of Minnesota, Saint Paul, Minnesota 55108

2 Department of Ecology, Evolution, and Organismal Biology, Iowa State University, Ames, Iowa 50011

3 Present affiliation: Illinois Natural History Survey, Champaign, Illinois 61820

*Corresponding Author: ; Phone: 612-625-5700; Fax: 612-624-6777

Appendices

Contents

Appendix A: Soil sampling and analysis

Appendix B: Evaluation of equifinality from maximum-likelihood estimation (MLE)

Appendix C: ANOVA tables

1

Appendix A: Soil sampling and analysis

Table 1 Experimental plots sampled for each analysis performed in this study

Analysis / All plots a / Control and +N plots / Control plots only
Microbial respiration / X
Total soil C and N / X
Microbial biomass C and N / X
POM C and N b / X
Soil pH / X
Net N mineralization / X
Water-stable soil aggregates / X
Root biomass / X
Mycorrhizal colonization of root biomass / X
Soil texture / X

a Full factorial nutrient experiment: control, +N, +P, +K, +NP, +NK, +PK, and +NPK plots.

b POM: particulate organic matter.

1

Table 2 Number of experimental plots included in the statistical analysis of each variable measured a

Cedar Creek, Minnesota / Cedar Point, Nebraska / Chichaqua Bottoms, Iowa / Konza Prairie, Kansas / Shortgrass Steppe, Colorado
Analysis / Ambient N b / Added N b / Ambient N b / Added N b / Ambient N b / Added N b / Ambient N b / Added N b / Ambient N b / Added N b
Microbial respiration / 20 / 20 / 12 / 12 / 24 / 24 / 12 / 12 / 12 / 12
Total soil C and N / 20 / 20 / 12 / 12 / 24 / 24 / 12 / 12 / 12 / 12
Microbial biomass C and N / 20 / 20 / 12 / 12 / 24 / 24 / 12 / 12 / 12 / 12
POM C and N c / 17 / 20 / 12 / 12 / 24 / 24 / 12 / 12 / 11 / 12
Soil pH / 20 / 20 / 12 / 12 / 24 / 24 / 12 / 12 / 12 / 12
Net N mineralization / 20 / 20 / 12 / 12 / 24 / 24 / 12 / 12 / 12 / 12
Water-stable soil aggregates / 5 / 5 / 3 / 3 / 6 / 6 / 3 / 3 / 3 / 3
Root biomass / 5 / 5 / 3 / 3 / 6 / 6 / 3 / 3 / 3 / 3
Mycorrhizal colonization of root biomass / 3 / 5 / 3 / 0 / 6 / 5 / 3 / 3 / 2 / 3

a Plots were excluded from statistical analyses because the sample was missing or there was sample contamination during lab analyses.

b Treatment codes: For the analyses where the full nutrient factorial was sampled, ambient N includes all plots where N was not added (control, +P, +K, +PK plots); added N includes all N addition plots (+N, +NP, +NK, +NPK). For the analyses where only control and +N plots were sampled, ambient N = control plots and added N = +N plots.

c POM: particulate organic matter.

1

Appendix B: Evaluation of equifinality from maximum-likelihood estimation (MLE)

Parameter estimates fit using maximum-likelihood estimation (MLE) can result in equifinality: multiple combinations of parameters that produce equally good model fits (Beven 2006). We evaluated whether equifinality in parameter estimates was possible in the parameter space of our decomposition parameter estimates by randomly generating 50,000 parameter combinations for each sample. The parameters were randomly selected from a defined parameter space that spanned from 0.33x to 3x of the parameter values fit with MLE (using the bbmle package in R). We fit both the one-pool and two-pool models (see Methods for model details) with these randomly generated parameters for each sample. We then compared the predicted C respiration rate (mg C g soil-1 day-1) to the actual C respiration rate of each sample to generate an R2 value for each randomly generated parameter combination.

In all cases, the best R2 of the randomly generated parameter combinations were no better than, but similar to, those selected using MLE. Furthermore, the best-fit models from the randomly generated parameter set converged on one area of parameter space, indicating that there are not multiple combinations of parameters that result in equally good model fits. See Appendix B, Figure 1 for illustrative examples of three samples for which we randomly generated 1,000,000 parameter combinations and evaluated model fit (R2) against our MLE parameter values.

Figure Legend

Figure 1 Model fits (R2) of randomly generated parameter combinations versus model fit (R2) of MLE parameter values from three soil samples. In each panel, the graphs show R2 (color coded) of 1,000,000 randomly generated combinations of the two-pool model parameters (from left to right): fast pool decay (kf) versus fast pool size (Cf); slow pool decay (ks) versus fast pool decay (kf); and slow pool decay (ks) versus fast pool size (Cf). Grey triangles are the top ten best parameter combinations (based on R2) from the randomly generated parameters. The black triangle shows the parameter values selected with MLE. The average R2 from the top ten best parameter combinations (“Manual Pred”) was always less than the R2 from the MLE parameters (“Fit Pred”). Panel a: Chichaqua Bottoms (Iowa) control plot, sample number 24. Panel b: Cedar Creek (Minnesota) +K plot, sample number 57. Panel c: Cedar Point (Nebraska) +PK plot, sample number 14.

1

Figure 1

1

Appendix C: ANOVA tables

Table 1 ANOVA table: decomposition parameters and cumulative respiration from the microbial respiration incubation

Effect / kf / ks / Cf / Cs / Cumulative C respired
Site / *** / **** / * / **** / ****
N / * / * / † / ***
P
K
N x P
N x K
P x K / *
N x P x K
Site x N / ** / * / NA / NA / *
Marginal R2 a / 0.3614 / 0.3865 / 0.1723 / 0.7887 / 0.6226
Conditional R2 b / 0.3614 / 0.3958 / 0.2006 / 0.8023 / 0.6593

† p ≤0.10, * p ≤0.05, ** p ≤0.01, *** p ≤0.001, **** p ≤0.0001, NA = non-significant interaction term removed from model.

a Marginal R2 represents the variance that is explained by fixed effects only.

b Conditional R2 represents the variance that is explained by both fixed and random effects.

1

Table 2 ANOVA table: aggregate-occluded and mineral-associated soil fractions

Effect / Large macro-aggregate fraction / Small macro-aggregate fraction / Micro-aggregate fraction / Mineral-associated fraction
Site / **** / † / **** / ****
N / †
Site x N / NA / NA / NA / NA
Marginal R2 a / 0.7862 / 0.4900 / 0.8784 / 0.8315
Conditional R2 b / 0.7862 / 0.7663 / 0.8842 / 0.9130

† p ≤0.10, * p ≤0.05, ** p ≤0.01, *** p ≤0.001, **** p ≤0.0001, NA = non-significant interaction term removed from model.

a Marginal R2 represents the variance that is explained by fixed effects only.

b Conditional R2 represents the variance that is explained by both fixed and random effects.

1

Table 3 ANOVA table: additional soil variables

Effect / Total soil C / Total soil N / Soil C:N ratio
Site / **** / **** / ****
N / * / ***
P
K
N x P
N x K
P x K / * / *
N x P x K
Site x N / NA / NA / NA
Marginal R2 a / 0.7902 / 0.7966 / 0.6829
Conditional R2 b / 0.8042 / 0.8143 / 0.7081
Effect / Microbial C / Microbial N / Microbial C:N ratio
Site / **** / **** / *
N
P / †
K
N x P
N x K
P x K
N x P x K
Site x N / NA / * / *
Marginal R2 a / 0.7003 / 0.6057 / 0.2023
Conditional R2 b / 0.7115 / 0.6057 / 0.2023
Effect / POM C c / POM N c / POM C:N ratio c
Site / ** / *** / **
N / * / **
P
K
N x P / ** / **
N x K
P x K
N x P x K
Site x N / NA / NA / NA
Marginal R2 a / 0.3586 / 0.3954 / 0.2879
Conditional R2 b / 0.4607 / 0.5113 / 0.3921
Effect / Soil pH / Net N mineralization
Site / **** / ****
N / **** / ****
P
K
N x P / *
N x K / *
P x K
N x P x K
Site x N / NA / ****
Marginal R2 a / 0.6439 / 0.6495
Conditional R2 b / 0.7318 / 0.6952

† p ≤0.10, * p ≤0.05, ** p ≤0.01, *** p ≤0.001, **** p ≤0.0001, NA = non-significant interaction term removed from model.

a Marginal R2 represents the variance that is explained by fixed effects only.

b Conditional R2 represents the variance that is explained by both fixed and random effects.

c POM: particulate organic matter.

1

Table 4 ANOVA table: root variables

Effect / Root biomass / Mycorrhizal colonized root biomass (absolute) / Mycorrhizal colonized root biomass (%)
Site / **** / **** / †
N / * / †
Site x N / NA / NA / NA
Marginal R2 a / 0.7338 / 0.8057 / 0.3759
Conditional R2 a / 0.7338 / 0.8133 / 0.3759

† p ≤0.10, * p ≤0.05, ** p ≤0.01, *** p ≤0.001, **** p ≤0.0001, NA = non-significant interaction term removed from model.

a Marginal R2 represents the variance that is explained by fixed effects only.

b Conditional R2 represents the variance that is explained by both fixed and random effects.

1