Ecological Archives E088-030-A5

Fernando T. Maestre and James F. Reynolds. 2007. Amount or pattern? Grassland responses to the heterogeneity and availability of two key resources. Ecology 88:501–511.

Appendix E. Comparisons between alternative statistical models for assemblage measurements.

TABLE E1. Comparisons between alternative statistical models for assemblage measurements. BAR = below: aboveground biomass ratio, RII = ratio of root biomass in the 0–15 soil depth layer to that in the 16–30 depth layer, M1 = model including all main effects, M2 = model including all main effects and interactions (full model), and M3 = model including all main effects and significant interactions, as revealed by MANOVA/ANOVA analyses (see Appendix B for full MANOVA/ANOVA results). P values below 0.05 are in bold, and indicate that the addition of interactions (present in models M2 and M3) significantly increases the amount of variation explained by the model. All the analyses were conducted using S-Plus for Windows version 6.0 (Insightful Co, Seattle, Washington, USA). All the models assume a Gaussian distribution of errors.

Variable

Comparison

Change in deviance

Change in df

F

P

Total biomass

M1 vs. M2

0.19

11

5.41

<0.001

M1 vs. M3

0.18

3

18.79

<0.001

BAR

M1 vs. M2

0.31

11

1.92

0.043

 

M1 vs. M3

0.265

3

5.99

0.001

RII

M1 vs. M2

0.12

11

1.65

0.092

 

M1 vs. M3

0.03

1

4.61

0.034

† Results are shown for log10-transformed data.

‡Log10-transformed total assemblage biomass was used as a covariate in this analysis.



[Back to E088-030]