Effect of repeated spring drought and summer heavy rain on managed grassland biomass production and CO₂ efflux

Abstract

Aims

Ecosystem respiration represents an important component of the carbon cycle. The response of respiration to climate change can have a significant effect on carbon sequestration in terrestrial ecosystems in the future when, according to climate scenarios, spring drought and consequent summer heavy rains are expected. Aims of our study were to determine the effect of repeated spring drought on biomass production and CO₂ efflux from a mountain grassland in Beskydy Mountains in the Czech Republic and to evaluate factors driving the differences among the study years.

Methods

CO₂ efflux was measured on plots with ambient precipitation conditions (AMB) and on plots where drought conditions (DRY) in the first half of the growing seasons and consequent heavy rain were simulated in 2011–14.

Important Findings

The spring drought significantly decreased the amount of above-ground biomass sampled just after the simulated drought in all years except for 2014. On the contrary, the spring drought stimulated root production. The drought also resulted in a rapid decrease in CO₂ efflux. It was lower by up to 46% for the DRY treatment compared to AMB treatment. After the simulated drought period, differences in CO₂ efflux between the treatments gradually decreased. Simulated heavy rains in DRY resulted in fast but temporary increase in CO₂ efflux. We can assume that the future spring drought will have a significant effect on carbon balance of grassland ecosystems. The impact will depend on the length of the dry period and the time between the beginning of the growing season and the dry period.

INTRODUCTION

Current climate models predict future changes in precipitation volume and pattern. Many scenarios presume an increase in the frequency of spring droughts and more intensive rain in summer (Frederick and Major 1997; Frei et al. 2006; IPCC 2013). As precipitation is one of the crucial factors controlling ecosystem carbon fluxes, such as ecosystem primary production and respiration, the changes in precipitation are expected to have a great impact on terrestrial ecosystems.

Soil CO₂ efflux represents a major part of total ecosystem CO₂ efflux. It constitutes, after photosynthetic carbon assimilation, the second largest flux of carbon between terrestrial ecosystems and the atmosphere (Raich and Schlesinger 1992) and is one of the key determinants of net ecosystem carbon exchange. Grassland ecosystems cover around 40% of the ice-free land on the Earth’s surface (White et al. 2000), they also represent an important component in the global carbon cycle as they store about 33% of total terrestrial soil organic carbon (Lal 2004). Therefore, they will play an important role in the carbon cycle under changing climatic conditions.

Soil moisture can affect soil carbon fluxes in several ways. The severe lack of water results in slower diffusion of substrates to the microbes (Stark and Firestone 1995) and reduces the activity of enzymes (Steinweg et al. 2012). On the other hand, high soil moisture or even flooding can create anoxic conditions in soil and inhibit the oxidative respiration (Freeman et al. 2001). Therefore, the predicted lower precipitation amount would have different impacts on different ecosystems. Decreases in soil moisture in permanently wet or flooded soils will result in soil aeration and enhanced soil respiration (Couwenberg et al. 2010), causing further release of CO₂ into the atmosphere especially from northern ecosystems (Tarnocai 2009). On the contrary, lower soil water availability in mesic or dry soils can decrease soil respiration (Borken et al. 2006) as well as ecosystem productivity (Zhang et al. 2012). A decrease in the amount of the above-ground green biomass caused by drought decreases the assimilate supply to roots and soil (Barthel et al. 2011; Fuchslueger et al. 2014).

Whether ecosystems represent a sink or source for CO₂ depends on the balance between total photosynthesis and respiration, particularly during the growing season. As soil CO₂ efflux is the second largest CO₂ flux in terrestrial ecosystems, as mentioned above, changes in soil CO₂ efflux have a crucial impact on the carbon balance of the whole ecosystem.

The effect of drought on soil CO₂ efflux has been investigated in different ecosystems, e.g. tropical forests (Sotta et al. 2007), temperate forests (Borken et al. 2006), grasslands (Fay et al. 2000) or cropland (Larionova et al. 2010). The studies made on grassland ecosystems have focused mainly on the effect of drought and changed precipitation patterns on biomass production (Bates et al. 2006; Yahdjian and Sala 2006), species composition (Baez et al. 2012; Grant et al. 2014) and carbon flux (e.g. Harper et al. 2005; Joos et al. 2010). Water stress conditions affect both soil and plants as mentioned above. Moreover, reduction of the above-ground biomass can result in other impacts on soil CO₂ efflux through decrease in transport of newly assimilated carbon to roots and soil and, therefore, the reduction of easily decomposable substrate for respiration of rhizosphere microorganisms (Kuzyakov and Cheng 2001).

The impact of both drought and rain events on the ecosystem carbon balance depends on their frequency, extremity and timing. For example, Fay et al. (2000) and Laporte et al. (2002) found that the pattern of lower precipitation frequency with higher intensity resulted in a lower amount of released CO₂ than the pattern of higher precipitation frequency with lower intensity. Therefore, we can expect projected dry periods to cause lower CO₂ efflux even though the total annual precipitation would not change. The greatest impacts of drought can be expected to occur during the period of plant growth (Kwon et al. 2008). The rain causes the most marked changes when the soil is dry (Chou et al. 2008). Based on these findings, we would hypothesize that the projected spring droughts followed by heavy summer rains will result in a substantial reduction in plant growth and CO₂ efflux from grassland ecosystems during the drought conditions, followed by a rapid release of large quantities of CO₂ into the atmosphere in response to heavy rainfall.

There have been few studies that have analyzed the effect of drought on the soil or ecosystem carbon balance using long-term observations. One-year studies record only the immediate response of the ecosystems to drought; however, repeated droughts in successive years may result in a long-term decrease in plant productivity (Yahdjian and Sala 2006), changes in species composition of plants (Baez et al., 2012) and soil microorganisms (Fuchslueger et al. 2014), which can affect CO₂ exchange. Backhaus et al. (2014), on contrary, have found drought resistance of plant communities to improve after several years of drought stress. Therefore, studies based on repeated year-on-year droughts are needed (Beier et al. 2012).

Thus, the present study focused on the response of the mountain grassland ecosystem to repeatedly simulated spring droughts and an intensive summer watering during four consecutive years.

The aims of our study were (i) to determine the effect of 4-year pattern of the repeated spring droughts on soil temperature, the above- and below-ground biomass production and CO₂ efflux (soil and bellow- and above-ground biomass together) in the grassland ecosystem; (ii) to evaluate the effect of factors, such as the timing of the beginning of growing season on biomass and CO₂ efflux; and (iii) to evaluate the effect of the end of the dry period and extreme summer rainfall events on CO₂ efflux.

MATERIALS AND METHODS

Site description

The experiment was carried out in the mountain grassland that is a part of the Ecological Experimental Study Site Bily Kriz (49°30′N, 18°32′E) situated in the Moravian–Silesian Beskydy Mountains, the Czech Republic. The site is characterized by a mean annual temperature of 6.8±1.1°C and precipitation of 1318±215mm.

The grassland is located at an altitude of 855 m a.s.l. and on a slope of 8.5° with south-east exposure. The soil type is Gleyic Luvisol (FAO classification). The studied grassland is dominated by moor matgrass (Nardus stricta L.), red fescue (Festuca rubra L.), creeping soft grass (Holcus mollis L.), common sorrel (Rumex acetosa L.), imperforate St John’s-wort (Hypericum maculatum Crantz.) and common yarrow (Achillea millefolium L.). The grassland is mowed once a year in the course of the growing season in July.

Experiment design

Six experimental plots (2.0×3.0 m) were established at the grassland site during 2011–14. During the experimental simulated drought period (SDP) roofs were built over the plots (Fig. 1). The roofs consisted of a wooden construction with 12 acrylate strip plates that had a 20° inclination. There were two arrangements of the plates. On three roofs, the plates were arranged so that the bottom edges of the upper plates overlapped the top edges of the lower plates. The roofs were equipped with a gutter that intercepted the water on their lowest side and drain the water out of the experimental plot. This arrangement resulted in the capture of precipitation and the simulation of drought conditions (DRY treatment). On the other three roofs, the bottom edges of the plates were placed under the top edges of the bellow-neighboring plates. This arrangement provided full precipitation release through the roofs, resulting in ambient rainfall conditions (AMB treatment) being enabled. A 0.2-m wide trench was dug and sheathed with a plastic foil to separate the soil of the roofed areas from the neighboring soil. A 0.25-m wide zone beneath the edge of the roofs was excluded from all measurements and samples. The distance between the roof and soil surface was 1.3 m to allow natural air recirculation and prevent greenhouse effect.

In the summer, watering of DRY was performed (Table 1). It simulated about 30mm of extreme summer rainfall.

CO₂ efflux, air and soil temperature and soil moisture measurements

CO₂ efflux from the grassland ecosystem was measured from four permanently installed cylindrical collars (diameter 20cm, height 15cm, insertion depth 3cm) per plot (total collar n = 24). The measurements were performed using a portable closed gasometrical system Li-8100 (Li-Cor Inc., Lincoln, USA) with a 20-cm survey chamber. After the chamber closed, the period (dead band) of 15s was set to allow steady mixing of the air in the chamber. During the following 60s CO₂ concentration was measured with 1-s time sequence, then a linear fit was used to calculate soil CO₂ efflux. CO₂ efflux measurements were carried out 6–10 times per growing season (shown in Fig. 2). Each measurement was made between 10 am and 12 pm.

During each measurement of CO₂ efflux, soil temperature at a depth of 1.5cm was measured 5cm outside the collar using a penetration thermometer (TPD32, Omega, USA). On each plot two sensors (ThetaProbe ML2x, Delta-T Devices, UK) recorded soil moisture at a depth of 10cm every 10min. Precipitation was measured by a 386C rain gauge (MetOne Instruments, Inc., USA).

Moreover, soil (5cm) and air (2 m) temperatures were measured continuously at the site using sensor PT100 (Sensit, CZ) and EMS 33 (EMS, CZ), respectively. These measurements were used to determine the beginning of the growing season, which was defined as the date when minimum daily air temperature did not fall below 5°C and soil temperature did not fall below 0°C for three consecutive days.

Biomass analyses

The above-ground biomass in the collars (n = 24) for CO₂ efflux was sampled once a year. When the whole grassland was mowed, the grass in the collars was cut to the same height as the surrounding grass (Table 1). The samples were dried to the constant weight for 48h at a temperature of 80°C and weighed.

Total root biomass was sampled in all treatments at the end of SDP in 2012–14. Four soil cores per treatment (9.4cm in diameter, 20cm depth) were collected. The collected samples were divided according to soil depth on 0–5cm, 5–10cm and 10–20cm, then washed in nylon bags and on sieves of 0.5mm mesh size, and dried to the constant weight and weighed.

Statistical analyses

One-way analysis of variance (ANOVA) was run separately for each treatment and sampling date to test the homogeneity of CO₂ efflux and soil temperature among three plots. As each plot consisted of four pseudo replicates, the mean values of CO₂ efflux and soil temperature were calculated for each plot. Differences in CO₂ efflux and soil temperature between AMB and DRY were tested using t-tests for each measurement campaign. T-test was also used for comparison of daily means of soil moisture between AMB and DRY and for comparison of the amount of above- and below-ground biomass between AMB and DRY in individual years. To analyze differences in the amount of above-ground biomass among individual years, one-way repeated measures ANOVA was implied separately for AMB and DRY. The analyses were performed using SigmaPlot 11.0 analytical software (Systat Software, San Jose, CA, USA).

RESULTS

The growing seasons at the study site began in March or April. The earliest onset of the growing season occurred in 2014, while the latest was in 2013 and occurred >1 month later than in 2014 (Table 2).

Micrometeorological measurements

Seasonal sums of precipitation (1 May–31 October), SDP sums of precipitation and number of rainy days during SDP are presented in Table 2. The growing season for 2011 showed the highest total precipitation out of the experimental years. The roofs captured 470.6mm of precipitation in DRY during the SDP, which was about two or three times more than in other years. This was caused by the longest SDP occurring in 2011 (Table 1).

Roof installation resulted in a significant reduction in soil moisture at the 10cm depth in DRY, the minima ranged between 15% and 20%. After the roof removal at the end of SDP, differences in soil moisture between treatments declined as soil moisture in DRY gradually increased due to increases in precipitation (Fig. 2). The statistical differences in soil moisture between AMB and DRY were found for the whole period when moisture was measured in 2011 (8 June–8 August) and then on 13 May–25 May and 4 June–13 July 2012, 28 May–31 July 2013, 26 June–1 August 2014.

No significant differences in soil temperature in 1.5cm were found among three plots of each treatment. However, soil temperature was significantly affected by the simulated drought (Fig. 3). The soil temperature in DRY was significantly higher (P < 0.05) than in AMB during and shortly after the SDP as indicated by asterisks in the graph. In 2013, the higher soil temperature in DRY compared to AMB occurred about 3 weeks after roof removal (30 July), probably as a result of the very low amount of precipitation during this period and persisting difference in soil moisture.

Biomass analyses

The mean dry weight of the above-ground biomass ranged from 267 to 434.6g m⁻² in AMB and from 267.3 to 319.6g m⁻² in DRY over four experimental years. In AMB, the amount of above-ground biomass in 2011 was significantly higher than in 2012 and 2013 (Fig. 4) and the amount of above-ground biomass in AMB positively correlated with the percentage of rainy days in the period from 1 May to the day of the biomass harvesting (Pearson correlation, P < 0.05). The high amount of above-ground biomass in 2011 could also be related to the fact that harvesting of the above-ground biomass was provided at a later date than during other years (Table 1).

In DRY, the amount of above-ground biomass had an increasing tendency from 2011 to 2014 and it was significantly higher in 2014 than in 2011 and 2012. The amount of above-ground biomass in DRY was significantly reduced in 2011, 2012 and 2013 compared to AMB. In 2014, no significant difference between treatments was found (Fig. 4).

The amount of root biomass was found to be highest in the top layer (0–5cm) and decreased with depth in both treatments. The amount of roots tended to be higher in DRY compared to AMB except for 5–10cm and 10–20cm in 2013. However, because of high variability, the significant differences between AMB and DRY (P < 0.05) were found only for depths 5–10cm and 10–20cm in 2014 (Fig. 5).

CO₂ efflux

CO₂ efflux in AMB ranged from 2.5 to 14 µmol m⁻² s⁻¹ and peaked between the end of May and beginning of June as a result of both high temperature and the high amount of biomass growing above ground. In the second half of the growing season, after the harvesting of the above-ground biomass, CO₂ efflux did not reach values as high as before cutting, despite higher soil temperatures (Fig. 3). No significant differences in CO₂ efflux were found among three plots of each treatment. This indicated homogeneity of CO₂ efflux at the grassland site and that three plots per treatment were sufficient. In the beginning of the growing seasons, before SDP, CO₂ efflux did not differ between the treatments except for 2012, when CO₂ efflux in DRY was lower by 22.3% than in AMB. Pretreatment data in 2011 are not available.

CO₂ efflux in DRY was significantly reduced (P < 0.05) by up to 46% during the simulated drought as indicated by asterisks in Fig. 3. The highest differences were in 2011 and 2012. CO₂ efflux in DRY, measured just before harvesting of the grass, significantly correlated with the amount of above-ground biomass during the first 2 years of the experiment (P < 0.05, Pearson correlation). However, no correlation was found for DRY in 2013 and 2014 or for AMB in any year (P > 0.05; Fig. 6).

Shortly after roof removal, CO₂ efflux in DRY remained mostly lower than that in AMB and difference in CO₂ efflux between AMB and DRY gradually decreased. CO₂ efflux in autumn 2013 did not differ between treatments, and CO₂ efflux in DRY even slightly exceeded CO₂ efflux in AMB in 2014. On the contrary, CO₂ efflux in September 2012 was significantly lower in DRY than in AMB (P < 0.05). These results may correspond to the low precipitation from August to September in 2012, which was approximately half of the values for August to September in 2013 and 2014.

In 2012, CO₂ efflux in DRY reached values of AMB 3 weeks (30 July) after roof removal. After this measurement, watering of the DRY was performed and then, 2h later, CO₂ efflux in DRY was significantly higher than in AMB (P < 0.05).

DISCUSSION

Expected global climate change includes a rise in global surface temperatures and changes in the intensity and frequency of precipitation in most regions of the Earth (IPCC 2013). In the context of carbon losses from soil, the changes in frequency of rainfall can have a more substantial impact than changes in precipitation intensity (e.g. Fay et al. 2000; Harper et al. 2005; Jentsch et al. 2007). Many climate change projections suggest that periodic droughts, especially in the spring, will become more common in Central Europe (van Haren et al. 2015). As the spring is the period of intensive plant development and growth, it would have a severe effect on carbon fluxes in terrestrial ecosystems (Wolf et al. 2013).

In our study, two arrangements of shelters over a grassland ecosystem resulted in the full release (ambient precipitation treatment—AMB) or capture (spring drought treatment—DRY) of precipitation. This arrangement, roofs over both treatments, was supposed to avoid differences in energy fluxes between AMB and DRY as Fay et al. (2000) observed higher temperature in sheltered plots compared to unsheltered ones. As temperature is one of the factors driving CO₂ efflux (Davidson et al. 2006), these differences are undesirable.

During and shortly after the SDP higher daytime soil temperatures were observed in DRY compared to AMB, which is in agreement with the study of Al-Kayssi et al. (1990). They observed higher temperature maxima during daytime after the decrease in soil moisture. This can be assumed to be a result of higher evaporation from wet soils, which leads to cooling of the soil surface, keeping the soil from warming as fast during the day. At night the soil surface cools more slowly because the high water content results in a greater soil heat capacity (Al-Kayssi et al. 1990). The above-ground biomass works also as an insulation layer, which prevents the sun’s radiation from warming the soil surface. Therefore, a low amount of biomass results in faster warming of the soil surface (Zhou et al. 2007). Although soil temperature is usually the driving factor for soil respiration (Davidson et al. 2006), we do not assume a great impact of its increase in DRY during the dry period on CO₂ efflux. Under dry conditions, soil water becomes a limiting factor and the effect of soil temperature is minimized (Correia et al. 2012; Xu and Qi 2001).

Rain exclusion resulted in significant suppression of the amount of above-ground biomass except for 2014 (Fig. 4). The highest difference between AMB and DRY was observed during the first year of the experiment. This was caused by the high amount of above-ground biomass, which exceeded the amounts in other years. The main reasons were that in 2011 the sampling of the above-ground biomass was provided over 20 days later than in other years, so the vegetation had a longer time to grow, and the high amount of precipitation and high soil moisture were observed in this year. As the water supply is one of the most important factors influencing the grassland productivity (Parton et al. 2012), the rainfall regime can have a great impact on its inter-annual variability. On the other hand, no difference in the above-ground biomass between treatments was found in 2014, when the growing season began the earliest out of all the experimental years (Table 1) and the period between the beginning of the growing season and beginning of the SDP was the longest. Therefore, the biomass had more time to fully develop and become more resistant to the subsequent drought period.

Other studies have also described biomass reduction under dry conditions, which has a strong effect on ecosystem productivity, especially when the drought occurs in the first half of growing season. As spring is the most active period of plant growth, it has the highest potential to influence plant production (Bates et al. 2006). This assumption suggests that spring drought can be more critical to carbon dynamics than a lack of summer precipitation, e.g., as suggested by Kwon et al. (2008). However, our results suggest that the effect of the spring drought can be weakened by the early onset of the growing season.

The simulated drought significantly reduced CO₂ efflux from the experimental grassland ecosystem by as much as 46%. Other studies have observed reductions in soil CO₂ efflux during drought conditions as high as 60% (Burri et al. 2014; Hagedorn and Joos 2014; Joos et al. 2010). All these results show that a reduction of precipitation has a significant effect on the grassland carbon dynamics. The lower reduction in our study can be attributed to respiration of the above-ground biomass as we measured CO₂ efflux from the whole ecosystem.

In contrast to the reduction of the above-ground biomass observed during simulated spring drought, we observed that root biomass tended to be higher for simulated spring drought conditions. This corresponds with common knowledge that root growth is less affected by drought stress compared to leaf growth (Saab et al. 1990). Rapid changes in allocation strategy were observed in response to drought in the previous studies, when larger portion of assimilates were transported to roots by drought-stressed plants compared to water-sufficient plants (Burri et al. 2014). As a result of the assimilate supply, higher root growth and root exudation of water-stressed plants have been observed (Kahmen et al. 2005; Sanaullah et al. 2012). In conclusion, drying of the shallow horizons can stimulate the plants for preferential transport of fresh assimilates below ground, which are then used for root growth to sustain water uptake (Burri et al. 2014). The plants can then become more resistant to drought stress.

The low water availability has an immediate effect on carbon flux during the dry period as described in the text above. However, the drought can result in changes in these processes after the dry period is over, as the ecosystem has ‘memory’ about these conditions through changed abiotic drivers and microbial community structure. The soils previously exposed to drought have, e.g., slower respiratory response and bacterial growth during rain compared to not-stressed soils (Evans and Wallenstein 2012). In addition, CO₂ efflux remains lower in previously stressed soils than in not-stressed soils (Goeransson et al. 2013). Water stress conditions also results in the soil microbial composition being dominated by fungi as they are more resistant to water stress than bacteria (Manzoni et al. 2012). The change of the microbial composition may also influence soil CO₂ efflux (Wang et al. 2014). Reduced leaf production caused by drought (Bates et al. 2006), which can reflect also in the year following the drought (Reichmann et al. 2013), would lead to lower assimilation, and therefore a lower transport of assimilates to roots, resulting in lower rhizosphere respiration and soil CO₂ efflux (Kuzyakov and Cheng 2001). How long these effects persist depends on the extremity of the drought (Evans and Wallenstein 2012). These findings can contribute to the explanation for the lower CO₂ efflux for DRY in the periods of late summer and autumn or before drought simulation in the beginning of the growing season in 2012. The long-term effects of severe droughts can result in difficulties in predicting the behavior of ecosystems exposed to drought (Shi et al. 2014), especially if the dry periods are repeated annually.

First rains after roof removal or heavy summer rains, simulated in our study by watering of DRY, suddenly changed the conditions of the ecosystem suffering from water stress. The ecosystem carbon balance rapidly responded to the new conditions by increasing the CO₂ efflux. CO₂ efflux in DRY reached rates of AMB fast after watering or roof removal especially in 2011 and 2012 or with some delay as in 2013 and 2014. Several mechanisms can contribute to this CO₂ efflux increase.

According to Wang et al. (2015), the spring drought, which suppressed both above- and below-ground productivity of a grassland ecosystem, resulted in accumulation of soil nutrients. These accumulated nutrients could contribute to increases in soil respiration in DRY after the end of the dry period and the restoring of ambient precipitation conditions (and increase on soil moisture).

The fastest and dominant response to precipitation after the dry period is usually attributed to the heterotrophic component of soil respiration, which is also supported by the stable carbon isotope study of Casals et al. (2011). In contrast, an increase in respiration of newly assimilated carbon follows the wetting with a time lag, as assimilation has a delayed response to wetting (Shim et al. 2009) and it also requires a higher amount of precipitation than heterotrophic respiration (Parton et al. 2012).

After watering of the DRY treatment in 2012, we observed a fast increase in CO₂ efflux, which exceeded the rates in AMB. This response can be described as a result of so called ‘Birch effect’ (Birch 1958; Jarvis et al. 2007). It represents a pulse in CO₂ efflux, after the wetting of dry soils, caused by enhanced microbial activity, mineralization of previously unavailable and easily decomposable substrates and lysis of microbial cells (Borken and Matzner 2009). Borken et al. (2003) found this effect to be fast but short term, and dependent on the intensity of watering and the length of the dry period. In a grassland ecosystem, these pulses of CO₂ did not compensate for the reduced carbon losses during the dry period (Hagedorn and Joos 2014).

CONCLUSIONS

Simulated spring drought significantly reduced CO₂ efflux from the grassland ecosystem. Moreover, it resulted in suppression of the production of above-ground biomass and stimulation of root growth.

Our findings indicate that spring droughts will strongly affect the carbon fluxes of grassland ecosystems. The magnitude of drought responses will differ, depending on the drought length and possible shifts of the beginning of the growing season to earlier dates, which can result in the sufficient production of plants in their early stages so they are more resistant to the following spring drought stress.

Simulated intensive summer rain resulted in a fast but short-term increase in CO₂ efflux. These results could be useful for model development and validation at landscape and regional scales to improve predictions of carbon dynamics in grasslands in response to climate change.

FUNDING

Ministry of Education, Youth and Sports of Czech Republic within the National Sustainability Program I (NPU I), grant number LO1415.

ACKNOWLEDGEMENTS

Authors wish to thank Dr Ryan Patrick McGloin for revising the English language of this manuscript.

Conflict of interest statement. None declared.

REFERENCES

Author notes