In 2009 the rainfall pattern was similar to long term trends, with a dry warm summer and autumn. However, the main measures of this experiment took place during an exceptionally wet winter period (166mm rain in December compared with a 10-year average of 63mm) followed by a very dry month in June (4.4mm compared with ten year average of 51mm). The soil moisture pattern follows that of the rainfall, with a slight lag time. Despite the differing regimes the spring/summer climate change and variable regimes never significantly differed from one another (Fig. 4.1). The effect of the spring phase of the spring/summer
Figure 4.1: Soil moisture compared with applied rainfall over the course of the experiment. Asterisks represent
differences in soil moisture between rainfall treatments (*p<0.05, **p<0.01, ***p<0.001). Error bars on the soil moisture lines are omitted for clarity.
climate change regime had little discernible effect on soil moisture, possibly because the differences in absolute rainfall volume were small (in 2010).
4.4.2 VEGETATION SURVEYS
Species richness per m2 was quite low across all climate regimes including ambient (2-6 spp) at all survey times. The spring/summer climate change regime had very constant species numbers throughout the experiment compared to the variable and the ambient treatments, which were more adversely affected by the drought in early spring 2010. The most dramatic of these was the variable treatment, which was most diverse in May 2010 (ten species), then lost an average of six species by September including Arrhenatherum elatius, Holcus mollis,
Rumex obtusifolius and Vicia species. The most abundant species was H. mollis in all
treatments and at all time points, except May 2010 when H. lanatus briefly superseded it in the spring/summer climate plots. The other common species throughout the experiment were
Cirsium arvense, Agrostis capillaris and Lotus corniculatus, although these were more
subject to fluctuation. RMANOVAs showed no effect of climate treatment on species richness per m2, but there was a very highly significant change over time (F4,36=5.93,
p<0.001, Fig. 4.2a). Percentage of bare earth was closely linked to soil moisture, being highest during the drier seasons, although the sustained drought in June 2010 did not lead to more bare earth than in June 2009 (Fig. 4.2b). There was a significant month by climate interaction, where the variable treatment plots had the least bare earth through the drought period, but had the most dramatic dieback by September despite the increase in soil moisture (RMANOVA: F8,36=11,72, p<0.001).
Grass percentage cover changed over time in all treatments, though most markedly in the variable treatment (RMANOVA: F4,36=11.72, p<0.01, Fig. 4.3a). This mirrors the loss of
species seen in Fig. 4.2a. The cover of forbs did not change significantly across treatment or time (RMANOVA: F4,36=1.63, p>0.05 Fig. 4.3b). Grass cover (Fig. 4.2a) increased up to
May 2010 then declined due to a combination of natural and manipulated drought. The variable treatment continued to decrease in grass cover to September after three months of prolonged droughts and heavy rainfall pulses, while the other two treatments showed signs of recovery. Forb cover was apparently unrelated to the amount of bare earth in the plots and appeared to be resilient to climate change. The spring/summer climate change treatment plots had less bare ground than the ambient (Fig. 4.3b).
Figure 4.2: a) Average species number in the climate treatments (RMANOVA Month: F4,36=5.93, p<0.001). b) Average percentage of bare earth in the plots in different climate treatments (RMANOVA Month*climate interaction: F8,36=11,72, p<0.001). Error bars depict ±1 SEM in both cases, and are staggered to illustrate differences between treatments.
Figure 4.3: Grass and forb percentage cover over time. a) Grass cover changes over time, but not
treatment (RMANOVA Month: F4,36=11.72, p<0.01) b) Forb cover does not significantly change over time or treatment (RMANOVA Month x Climate: F4,36=1.63, p>0.05). Error bars are staggered in order to illustrate differences between treatments and represent ±1 SEM.
Figure 4.4: a) Net ecosystem exchange showed no difference between climate change treatments
(RMANOVA: F2,45=1.37, p>0.05) or over time (RMANOVA: F10,45=0.32, p>0.05). Note that the x axis becomes more negative; this is to show greater reductions of CO2 in the chamber. b) Ecosystem respiration changed over time (RMANOVA: F5,45=5.11, p<0.001), but there was no effect of treatment (RMANOVA: F2,45=0.77, p>0.05) or an interaction between them (RMANOVA: F10,45=1.40, p>0.05). Error bars represent ±1 SEM, error bars are staggered to illustrate differences between treatments.
Net ecosystem exchange (NEE) is the balance between photosynthesis and ecosystem respiration (Reco); it was closer to zero (equilibrium) in winter and more negative in summer
which corresponds to with peak plant growth, becoming closer to equilibrium in June 10 when natural drought and mass plant dieback occurred (Fig. 4.4a). This change of NEE towards zero suggests photosynthesis was reduced, as the change was not reflected in Reco
rates (Fig. 4.4b). In February and May the key abiotic driver was photosynthetically active radiation, which had a strongly positive relationship with NEE (February: F1,3=21.71, p<0.05,
May: F1,9=12.64, p<0.01). In July and August 2010 soil temperature interacted with NEE, as
it became warmer the Reco component of NEE increased and brought it closer to equilibrium
(July: F1,9=9.13, p<0.05, August: F1,9=8.51, p<0.05).
Reco was not significantly affected by climate change manipulations (Fig. 4.4b); the
main driver appeared to be soil temperature, which was negatively correlated with Reco in
May (F1,9=10.94, p<0.05) and June (F1,9=14.96, p<0.01). It was also negatively correlated
with soil moisture in May, so warmer (and drier) soils had lower rates of Reco (F1,9=6.13,
p<0.05).
Figure 4.5:Evapotranspiration change over time with treatment effects (RMANOVA: F5,45=10.74, p<0.001). There are significant effects of climate change in July (one-way ANOVA F2,7=7.20, p<0.05) and August (one- way ANOVA F2,7=11.87, p<0.05). Error bars are staggered to elucidate treatment effects. Significance stars denote *p<0.05, **p<0.01, ***p<0.001.
Figure 4.6: Mineralisation rate over the growing season of 2010. Climate change was significant
(F2,8=4.11, p<0.05). Asterisks signify *p<0.05, **p<0.01, ***p<0.001.
The pattern of evapotranspiration (ET) was most likely to reflect the pattern of growth and natural water stress upon the system throughout the summer (Fig. 4.5). ET was strongly reduced by increasing soil temperature (closely linked to plant species cover Fig. 4.2b) in May and August 2010, (F1,9=85.31, p<0.001 and F1,9=8.64, p<0.05 respectively). In July, the
spring/summer treatment was significantly lower than the ambient (F2,8=7.2, p<0.05). In
August ambient had lower ET rates than spring/summer (F2,8=11.87, p<0.05).