Experimental design and measurements
Root runners of S. arvensis were collected in the autumn of 2005 from organically grown short-term ley fields with cereals in the rotation at Sala (59°40′N, 16°40′E) and planted in buckets filled with soil. This material constitutes a plant bank, which was stored over winters in buckets with a soil volume of 10 L from October until June in a dark cold store at +2 to +4°C, and grown outdoors in the same buckets during the summers of 2006–2008 from June to October. During each year, buckets were taken from the cold store in June, roots were replanted, and an early fertilisation of about 70 kg N ha−1 was provided, while the soil was kept moist during growing seasons.
On 29 June 2009, buckets were taken from the cold store, and roots were distributed into three thickness categories; fine (<2 mm), medium (2–4 mm) and thick (>4 mm) in diameter. Roots of each category were cut in 5- and 10-cm pieces, respectively, with at least two adventitious buds, but no sprouted buds, on each piece. These prepared root fragments were kept in the cold store prior to performing two experiments.
To determine the ratio between dry weight and fresh weight of the roots, 10 roots were randomly selected from each of the three thickness categories and of the 5 and 10 cm length categories, in total N = 60. After measuring fresh weight, the roots were dried at 50°C to constant weight (c. 3 days), and individual root dry weight was determined.
In the first experiment, 15 buckets with 5 cm root parts and 15 buckets with 10 cm root parts were placed in each of three dark chambers, with temperatures of 4, 8 and 18°C, respectively, on 30 June 2009. The soil in each of the 90 buckets (5 L per bucket) consisted of 85% moderately decomposed peat, 15% sand and about 1 g N per bucket and NPK proportion of 2:1:2 (Hasselfors Garden AB). For each temperature and root length, 81 roots were planted, so that three roots (one root for each thickness category) were present in each layer at a depth of 3, 10 or 17 cm below soil surface (all together 27 roots × 3 units = 81 roots), as described in Table 1. As emergence time is known to be negatively correlated with planting depth (Håkansson & Wallgren, 1972b), this design, which employs a wider depth range, was expected to give a larger time period over which soil grown (non-etiolated) shoots could be harvested.
Table 1. Experiment I and II: number of harvests, experimental units (buckets/pots), planting depths within units and number of roots per unit, for each temperature (4, 8, 18°C) and root length (5, 10 cm)
|Experiment||Harvest occasion||Experimental units||Planting depth (cm)||Number of roots/unit|
|I||1||3||3, 10, 17||9|
For the second experiment, referred to as the sand layer experiment, 60 pots with a diameter of 12 cm and a height of 11 cm (1 L) were prepared. A 2 cm washed, wet fine sand layer (Rådasand, Lidköping, Sweden) was placed at the bottom of each of the pots. In each of the pots, two randomly chosen roots of each of three thickness categories were placed on the sand layer, that is, six roots per pot, using 5 cm root parts in 30 of the pots and 10 cm root parts in the other 30 pots. The roots were not covered by any substrate. Ten pots with 5 cm root parts and 10 pots with 10 cm root parts were placed in each of the chambers, with temperatures of 4, 8 and 18°C, respectively, on 1 July 2009.
Prior to planting both experiments, fresh weight of the roots was measured for each experimental unit. All buckets and pots were sealed with plastic and kept moist throughout the experiment. Harvest of plants in buckets was performed well before or around the time the first shoot per bucket broke the soil surface and harvest took place during a time span ranging from about 120 days for roots at 4°C to about 20 days for roots at 18°C. Harvest of plants in pots (one pot per root length per harvest occasion) took place during a time span ranging from about 100 days for roots at 4°C to about 50 days for roots at 18°C. Dry weight of the remaining roots, fine roots and shoots per bucket or pot was determined after drying at 50°C to constant weight. In the bucket experiment, fine roots could not be retrieved from the soil and were discarded. Assuming that the allocation proportion to fine roots for a given temperature was equal in both experiments, and the fine root proportion resulting from the pot experiment was used to calculate fine root production in the buckets.
Estimation of dry weight loss rates
In analogy with weight loss due to respiration and microbial decay, we consider weight loss as a first order process, implying that dry weight loss overtime (dW/dt) is proportional to the amount of dry weight (W) present at that time:
where k is the daily relative weight loss rate. By integration over time from t = t0 when W = W(t0) to t = t1 when W = W(t1), we obtain:
For the relative decrease in W over time (WNorm(t1) = 100 × W(t1)/W(t0), that is, WNorm(t0) = 100), the value of k equals the slope of the linear relation between ln(WNorm(t1)) and time with a fixed intercept equal to 4.605 (=ln(WNorm(t0)) = ln(100)). We then obtain:
and estimated k by means of a linear regression procedure. Knowing the value of k, the half-lifetime (t½) of the biomass can be calculated as ln(2)/k.
When considering how the relative weight loss rate (k) is changing as a function of temperature, we may employ the temperature coefficient (Q10), which represents the factor by which k increases for every 10°C temperature (T) increment (Ryan, 1991). Q10 can be calculated as:
where k1 and k2 are weight loss rates at temperatures T1 and T2, respectively.
Application to current and elevated soil temperatures
To assess the effects of current climate variability on weight loss in below-ground overwintering structures of S. arvensis, the experimental results were extrapolated using records of daily average soil temperature, measured at 5 cm below soil surface (T5), retrieved from the Ultuna Meteorological Station (Karlsson & Fagerberg, 1995) close to Uppsala (59.48°N, 17.39°E), for the period 2004–2010, which included six entire overwintering periods. The start of the overwintering period with root activity was defined as days with soil temperature at 5 cm depth (T5) becoming <5.0°C, assuming a release of reduced sprouting at lower temperatures in S. arvensis roots (Brandsæter et al., 2010). The end of the overwintering season was set at T5 ≥ 4.0°C, as the emergence time of S. arvensis in spring was conceived as a function of soil warming (Lemna & Messersmith, 1990). Weight loss was estimated on a daily basis using Eqn (1) and a k-value as function of temperature according to Eqn (4), giving:
where kBase, Q10 and TBase were taken from the results of the controlled experiment in pots, where all biomass could be retrieved. The accumulated loss over the season was estimated by solving W(t + 1) = W(t)−k × W(t) daily from the day of start to the day of end of the overwintering season and evaluated for the relative decrease over the whole period (W(t = tEnd)/W(t = tStart)). Equation (5) was applied with two different parameterisations, to test for sensitivity to Q10 and to TBase. First, the Q10 value of 2.274, obtained over the entire experimental range of 4–18°C, a kBase-value of 0.0147 and a TBase of 18°C were used. Secondly, the Q10 value of 2.102, obtained over the range of 4–8°C, a kBase-value of 0.0047 and TBase of 4°C were applied.
To evaluate possible effects of alterations in soil temperature due to climate change on average temperature and length of the active below-ground period, the daily mean T5-values were raised constantly by 1, 2, 3 and 4°C, respectively. This gave in addition to the six actual seasons, 24 temperature-elevated seasons, for which cumulative weight loss was determined as described above.
Linear regression analysis procedures were used to establish relations between weight values and time. In case of known intercepts, the general model Y = a + b × x was reduced into a fixed intercept model. A general factorial anova (SAS Institute, 2002–2004) was used to assess the effects of root size categories on initial dry matter content, after having ensured an approximate normal distribution of the latter (SAS Institute, 2002–2004). A mixed model (Proc Mixed) was used to assess the effects of temperature (α), substrate (β) and root length (γ) on weight loss overtime (b). After testing for significance of interactions, the final model used was as follows:
In Eqn (6), rijk is normally distributed (N(0, )), where sijk is the standard error of bijk, and eijk is N(0, σ2).