Mycorrhizal responsiveness trends in annual crop plants and their wild relatives- A meta-analysis on studies from 1981 to 2010

Plant and Soil

Anika Lehmann1, E. Kathryn Barto1, Jeff R. Powell2 and Matthias C. Rillig1, *

1Freie Universität Berlin, Institut für Biologie, Dahlem Center of Plant Sciences, Altensteinstr. 6, D-14195 Berlin, Germany

2Hawkesbury Institute for the Environment, University of Western Sydney, Hawkesbury Campus, Bourke St, Richmond 2753, NSW, Australia

* Author for correspondence


Matthias C. Rillig

Institut für Biologie, Dahlem Center of Plant Sciences

Ökologie der Pflanzen

Altensteinstr. 6

D- 14195 Berlin

+49 (0)30 838-53165; fax -53886


Online Resource 2: Evaluation of mathematical behavior of mycorrhizal responsiveness indices

We tested two different indices to calculate mycorrhizal responsiveness:

1)  The response ratio: MR = ln(M/NC); a simple ratio of mycorrhizal and non-mycorrhizal plant biomass. Ratios were widely used throughout the literature.

2)  Absolute mycorrhizal responsiveness: AR = M-NC. This is a modified version of the model presented by Janos (2007) for a specific P level.

For our (non-parametric regression) analysis, we needed an index that is linearly related to non-mycorrhizal plant biomass (NC), to achive a statistically robust rendition of the mycorrhiza effect.

A non-linear relationship of an index and NC would be characterized by extreme values of NC (outliers) biasing the slope and thus the interpretation of the mycorrhiza effect in the given population.

To test the indices for their behavior, we used the regression method proposed by Sawers et al. (2010) and Galvan et al. (2011).

Fig. OR2: Regression of indices (for measuring mycorrhizal effects) against non-mycorrhizal plant biomass (NC) and mycorrhizal plant biomass (M) for both indices mycorrhizal responsiveness (MR) and absolute responsiveness (AR). (a) Natural logarithm of AR (lnAR) against lnM. (b) lnAR against lnNC. (c) Natural logarithm of MR against lnM. (d) lnMR against lnNC. The graphical parameters are given in Table OR2


Additionally, we tested our dataset for common and specific variation as suggested by Sawers et al. (2010). Therefore, we used their proposed method.

Table OR2: Measurements for evaluation of behavior of mycorrhizal effect indices. Correlation calculated with Kendall’s Tau, P< 0.05 (*), P < 0.01 (**) and P 0.001 (***)

x / Correlation( x, lnNC) / Correlation( x, lnM) / Common variation / Specific variation
lnAR / 0.09** / 0.38*** / 0.09 / 0.91
lnMR / -0.23*** / 0.07* / 0.16 / 0.84

The correlation between the two indices lnAR and lnMR and NC was weak due to high variability in the dataset (Table OR2). For lnAR, the correlation was the lowest and followed a logistic model, thus this index would over-estimate the mycorrhiza effect, especially in plants showing a high dependence upon mycorrhiza.

Dependence is defined as “the inability of a plant to grow without mycorrhizas below a particular level of soil phosphorus” (Janos 2007).

Both indices suggested a high specific variation in the population of this dataset, i.e. variation in plant growth response to AMF of mycorrhizal or non-mycorrhizal plants alone.

The response ratio lnMR was more appropriate for the statistical analysis of our dataset than lnAR. In a study on AMF responsiveness in onion (Galvan et al. 2011), absolute responsiveness reflected the mycorrhizal effect best and showed a linear relationship with NC. Onions are highly mycorrhizal responsive plants. For less responsive plants this index might be inappropriate. Therefore, it is not surprising that absolute responsiveness is strongly linked to mycorrhizal biomass (Fig. OR2) and there, the non-linear character was most clearly pronounced.

The response ratio lnMR is more obviously influenced by NC and thus showed for this variable the highest correlation and the best linear fit. So, we decided to use the response ratio lnMR as our effect size.