Effect of immersion on growth

When the plants were 67 d old (early December 2014), 12 individuals from each cultivar (OSR growth stage 1.3–1.6; i.e. emergence of third–sixth true leaves) were allocated at random to one of five treatment groups. These were 24 or 96 h immersion in seawater (Sw) (collected from Plymouth Sound – electrical conductivity = 45.51 mS cm^–1 at 16.2 °C); 24 or 96 h immersion in deionized water (Dw); or a no-immersion control treatment. Although river floodplains can experience much longer periods of immersion, sometimes extending to several months (Van Eck et al., 2004; Muchan et al., 2015), in this way we simulated the average 1 d long seawater flooding event reported for low-lying UK coastline habitats and extended the period to the maximum reported flood duration of 4 d (Environment Agency, 2014).

We recognize that inundation following coastal storm surge or fluvial flooding events would be likely to result in full shoot submergence but, by immersing to pot level (in large plastic tubs), our approach allowed us to separate the effect of ionic imbalance in the root zone rather than the impact of oxygen deficiency caused by full immersion that both treatments would impose. Immediately after immersion, the pots were allowed to drain fully before being arranged randomly on a wire mesh-topped bench inside the greenhouse; the wire mesh allowed free drainage and prevented cross-contamination between treatment groups. The pots were watered to capacity (with tap water) 48 h after seawater immersion.

Eight plants per treatment/cultivar combination were retained inside the greenhouse; the remainder were positioned outside, on adjacent elevated mesh. For both sets of plants, each individual was positioned at random, 20 cm apart from its nearest neighbour in 1 m long rows separated by 30 cm to simulate the recommended field density for OSR cropping systems in the UK (HGCA, 2014).

Greenhouse plants were watered twice weekly for a further 56 d; temperatures during this phase of the experiment were 4.9 °C (± 0.5) minimum and 18.0 °C (± 0.6) maximum. Surviving plants (growth stages 2.0–2.3; i.e. the plant has a rosette growth form and an extended stem with internodes) from each treatment/cultivar group were harvested at 113 d old (late January 2015), cleaned of any adhering compost and oven-dried at 50 °C for 24 h. A Levene’s test for homogeneity of variance across treatment and cultivar levels was negative (F_19,189 = 2.896, P < 0.001) and biomass data were log₁₀ transformed, resulting in a positive Levene’s test (F_19,189 = 1.335, P = 0.165). Univariate general linear model (GLM) analyses were used to compare the results of experimental flooding on transformed biomass. Rather than classifying plants in different groups into two separate factors (water, Dw vs. Sw; and time, 24 h vs. 96 h), which would decrease the power of the tests, contrasts were employed to evaluate specific differences between treatment levels (control, Dw 24, Dw 96, Sw 24 and Sw 96). In addition to the F statistic and its probability, we report effect sizes (η_p²) and power at P < 0.05 (power_0.05).