Table 2. Linear mixed model analysis of the effect of Mating System on floral traits of Arabis alpina.
| Trait (unit of measurement) | Transformation for analysis | Model estimate (transformed) of difference between outcrossing and selfing means a | t-value | p (df = 4) b |
|---|---|---|---|---|
| Flower area (mm2) | loge | -0.94 | -8.68 | 0.001 |
| Pistil length (mm) | square root | -0.14 | -2.15 | 0.0978 |
| Herkogamy short stamens (mm) | - | +0.58 | 6.54 | 0.0028 |
| Herkogamy long stamens (mm) | square root | -0.58 | -5.94 | 0.004 |
| Pollen number [x1000] (count) c | - | -61.6 | -10.7 | <0.001 |
| Ovule number (count) | - | -2.55 | -1.20 | 0.298 |
Cases where model estimates for selfing populations differ significantly from those of the outcrossing populations are indicated in bold.
a Model Fixed part: Mating system; Random part: Population (nested in Mating System) and Plant_ID (nested in Population), analysed with the lme function in the nlme package in R [43]
b df: degrees of freedom. For testing differences between mating system df = 4 because there were three outcrossing and three selfing populations
c This model included a separate variance for the two mating systems (varIdent option of lme), as there was a large variance heterogeneity among mating systems that could not be removed by transformation.