Appendix A. Species, growing conditions, temperature measurements, and sampling regime for the plant-growth experiment.
The species (nomenclature follows Stace (1997)) and their mean seed masses are Saxifraga tridactylites (0.006 mg), Erophila verna (0.025 mg), Cerastium diffusum (0.045 mg), Arenaria serpyllifolia (0.088 mg), Veronica arvensis (0.112 mg), Myosotis discolour (0.213 mg), Valerianella locusta (0.851 mg), Geranium molle (1.094 mg), and Erodium cicutarium (2.92 mg). Seeds were allowed to germinate at their natural germination time (September 2003) and plants were grown until they set seed in MayJune 2004. To mimic the typical small gaps in which these species are usually found, plants were grown singly in small cells measuring 13 × 13 mm and filled with sand plus a small amount of sieved peat and lime (in the proportions 6:1:1) to a depth of 26 mm. Cells were sown with more than one seed and thinned as soon as seedlings emerged to leave one seedling per cell (usually the first seedling to emerge). For the first five weeks following sowing, all cells were outside within glass-topped slug-proof enclosures in the University of Zurich experimental garden (47° 23’ N, 8° 33’ E, and 546 m a.s.l.). Fourteen days after sowing, 88% of the final plants had germinated and 15 individuals of each species were harvested, separated into above- and belowground parts, oven-dried and weighed. Following this initial harvest, each cell received one of five nutrient regimes (N ) by applying either water (N = 0) or a complete nutrient solution in one of four dilutions (N = 0.25, 0.5, 0.75, 1) every two weeks. Plants were otherwise watered daily.
At the end of five weeks, temperatures began to fall close to zero and following a second destructive harvest, half of the remaining plants were brought inside to a cool greenhouse while the rest remained outside to experience winter temperatures. Plants inside the greenhouse received no additional lighting. The greenhouse had no fine temperature control but we aimed to maintain temperatures between +5 and +20 degrees, so that plants would not experience extremes of either hot or cold. Approximately every 30 days we harvested, where possible, three individuals of each species from each nutrient regime both inside and outside (a total of seven destructive harvests) with the final one taking place in April (189 days after sowing). Three individuals from each species and nutrient regime both inside and outside were then left to finish setting seed. This scheme required a total of 1890 plants, of which we had 1724 (91%). The “missing” plants were more or less equally distributed across species, treatments, and harvests. From the beginning of December (harvest number 4) we also measured height, number of leaves and/or rosette diameter of all harvested and non-harvested plants. Following snowfall at the end of December, the glass covers were removed from the plants outside and they experienced ambient conditions, including snow cover, until the end of the experiment.
Daily minimum and maximum temperatures were obtained from a weather station within 1 km of the experimental garden (courtesy of MeteoSchweiz). In addition, weekly minimum and maximum temperatures were recorded inside the greenhouse. These fluctuated between a nighttime minimum of +3°C and a daytime maximum of +26 °C during the period November 2003 to April 2004 (compared to a nighttime minimum of –8.9 °C and a day-time maximum of +20.5 °C during the same period outside). For the initial period there were no thermometers under the glass covers. We therefore measured daily minimum and maximum temperatures under the same glass covers and the corresponding temperatures outside during a period when outside temperatures were similar to those during the growth experiment. By establishing regression relationships between the absolute temperature difference (outside vs. under glass) and the temperatures outside we were able to estimate temperatures under the glass. Hours of daylight on each day of the experiment were calculated using the latitude of Zurich and the formula presented in Forsythe et al. (1985), which is accurate to between 17 minutes per day (Forsythe et al. 1985).
One of the drawbacks of multiple-harvest experiments is that very large numbers of plants need to be grown, although each plant is only harvested once. However, we can potentially increase the sample size at each harvest if the biomass of unharvested plants can be estimated from nondestructive measures (height, diameter, and/or number of leaves; McGraw and Garbutt 1990). Clearly this is only valid if the non-destructive measures can usefully predict plant-to-plant variation in biomass within experimental treatments. To assess this, we fitted a full model to the biomass data from each species at all harvest dates containing the terms harvest date, nutrient treatment and temperature regime (inside or outside) and all possible interactions. Once this full model was fit, we fit the additional terms, height, number of leaves, and/or rosette diameter. For all species’ aboveground biomass, one or all of these nondestructive measures was highly significant, indicating that the non-destructive measures are an informative predictor of biomass variation within treatments. This model was then used to predict the biomass of unharvested plants; although these data were treated differently from those collected directly from destructive sampling (Appendix B).
Forsythe, W.C., E.J. Rykiel, R.S. Stahl, H.I. Wu, and R.M. Schoolfield. 1995. A model comparison for daylength as a function of latitude and day of year. Ecological Modelling 80:8795.
McGraw, J.B., and K. Garbutt. 1990. Demographic growth analysis. Ecology 71:11991204.
Stace, C. 1997. New flora of the British Isles. Cambridge University Press, Cambridge, UK.