A combined rain shelter and free-air CO2 enrichment system to study climate change impacts on plants in the field


  • Martin Erbs,

    Corresponding author
      Correspondence author. E-mail: martin.erbs@vti.bund.de
    Search for more papers by this author
  • Remy Manderscheid,

    1. Institute of Biodiversity, Johann Heinrich von Thünen-Institute, Federal Research Institute for Rural Areas, Forestry and Fisheries, Bundesallee 50, 38116 Braunschweig, Germany
    Search for more papers by this author
  • Hans-Joachim Weigel

    1. Institute of Biodiversity, Johann Heinrich von Thünen-Institute, Federal Research Institute for Rural Areas, Forestry and Fisheries, Bundesallee 50, 38116 Braunschweig, Germany
    Search for more papers by this author

Correspondence author. E-mail: martin.erbs@vti.bund.de


1. There is a clear need for field experiments to estimate the effects of global climate change like decreasing precipitation and rising atmospheric CO2 concentration. Adequate methods for controlled manipulations of these environmental parameters under field conditions are scarce, particularly in regard to multi-factor experiments. Here, we describe a new flexible rain shelter system, which can be assembled manually and easily be combined with further experimental treatments in field studies.

2. Frames of tents with a ground area of 12 m × 20 m were assembled on a field site after the sowing of the maize and sorghum crop and after the equipment for free-air CO2 enrichment has been set up. The tents were equipped with transparent tarpaulins, which were installed on the frames only in cases of high amounts of precipitations forecast. In autumn, the entire experimental equipment was removed from the field site.

3. The rain shelter tents were operated in the growing seasons 2008 and 2010 for 9 and 20 days with 54 mm and 176 mm of precipitation excluded, respectively. In the months with rain shelters in operation, pronounced reductions in precipitation were achieved (2008: 39·5%, 2010: 58·6%). The tent frames did not affect temperature or CO2 concentration, but slightly decreased incident photosynthetic active radiation (PAR) by 6·6%. In times with tarpaulins installed, PAR decreased by 24·1%. Comparing times without and with tarpaulins installed, the fraction of time in which 1-min mean CO2 concentration was within ±20% limits of the setpoint was decreased from 99·7% to 97·8% in 2008 and from 99·0% to 96·7% in 2010, respectively.

4. The rain shelter tents provide a suitable and versatile tool for excluding precipitation from larger areas in the field without relevant disturbances to the soil and aerial environment except a slight decrease in incident radiation, which can be accounted for i.e. in the evaluation of plant growth data. Furthermore, they can be easily combined with further experimental treatments like free-air CO2 enrichment.


Current model ensemble assessments predict an increase in land area prone to drought within the next few decades and a global trend for extended drought periods (Burke, Brown & Christidis 2006; Meehl et al. 2007). Combined with the rising global population, an over-all decrease in water available for agriculture of 18% is projected for the year 2050 (Strzepek & Boehlert 2010). The resulting drought stress periods will have pronounced negative impact on crop growth (Gornall et al. 2010). In contrast, feeding a global population of about nine billion people sufficiently will necessitate an increase in food production of 50–70% (FAO 2009). The imperative is to develop coping strategies for extreme events such as drought stress, with experiments providing the essential insights in the underlying processes (Jentsch, Kreyling & Beierkuhnlein 2007). This clearly points out the need for research on the effects of reduced water availability and temporal drought stress on crops at the field scale.

In the last decades, experiments with altered and reduced plant water supply in the field were enabled by the benefit of large-scale rain shelter facilities. Rain shelters are commonly constructed on the basis of two different concepts. Fixed shelters are permanent structures and mostly intercept the entire precipitation on the area of examination, thus offering quantitative control of water supply (e.g. Svejcar, Angell & Miller 1999; Fay et al. 2000; Yahdjian & Sala 2002; De Boeck et al. 2011). Two relevant disadvantages of this approach are a continuous reduction in incident radiation with the extent depending on roof material and unavoidable impacts on microclimate, e.g. air temperature and relative humidity. The second rain shelter approach frequently chosen is to use mobile shelters, which are moved over the examination area only in cases of precipitation. By doing so, impacts on incident radiation and microclimate can largely be reduced. However, mobile rain shelters need more complex technical facilities and elaborate controlling mechanisms for automatic operation (e.g. Foale, Davis & Upchurch 1986; Beier et al. 2004). This does not only raise the costs for automatic shelters, but also is a potential source of experimental error because of technical failures. Both concepts have a further drawback. In most cases and particularly for large-scale facilities, the installation of rain shelters requires extensive construction works using heavy machinery and concrete foundations. Consequently, there is a great risk of causing permanent disturbances to soil properties like compaction and changes in soil-water relations.

Plant water relation has a direct physiological connection to atmospheric carbon dioxide concentration ([CO2]) by the stomatal gas exchange. The [CO2] has increased from about 280 μmol mol−1 in pre-industrial times to 390 μmol mol−1 today (Tans 2009). Current model ensemble scenarios predict that [CO2] of at least 550 μmol mol−1 will be reached in the middle of this century (Meehl et al. 2007). This will have major impacts on plant productivity and water balance (Cramer et al. 2001). Elevated [CO2] reduces plant transpiration, while still enough CO2 is available for photosynthesis (Morison 2001), resulting in an increase in water use efficiency (Kimball, Kobayashi & Bindi 2002). In return, plant responses to CO2 enrichment largely vary with water availability (Mooney et al. 1999). This effect might be pronounced in C4 species like the crops maize and sorghum in which CO2 effects are suspected to be solely driven by the impacts on water relations (Leakey et al. 2006; Ghannoum 2008). Because current crop yield projections still differ the most in regard to the combined effects of elevated [CO2] and drought stress, it is of major importance to examine these combined effects in the field (Parry et al. 2004; Long et al. 2006). Furthermore, severity, timing and frequency of drought during the growing season could off-set beneficial effects of elevated [CO2] on crop growth (Oliver, Finch & Taylor 2009). Based on these findings, it is crucial to expand current knowledge by the means of results from experiments with crops under both drought stress and elevated [CO2] conditions.

Concerning experiments with elevated [CO2] among different approaches, free-air CO2 enrichment (FACE) has proven to be the most adequate one, because it largely avoids impacts on microclimate and allows for investigations under field conditions (Lewin et al. 1994; Erbs & Fangmeier 2006; Hendrey & Miglietta 2006). Hence, secondary impacts on plant-water relations and responses to elevated [CO2] are avoided (Leakey et al. 2009). While CO2 impacts on plants have been well documented in FACE experiments, studies on interactions with expected changes in other environmental variables such as water availability are scarce and should be a major research objective (Ainsworth et al. 2008).

The first objective of this study was to create temporal drought stress conditions on tall-stature crops by the means of large-sized tents used as rain shelters. Here, we examine whether the rain shelters and the experimental protocol chosen were capable of creating temporal drought stress by the interception of precipitation from the areas under examination. Potential secondary impacts of the tents on air temperature, on relative humidity and on incident total and diffuse radiation, which is important for plant growth (Sinclair, Shiraiwa & Hammer 1992), are evaluated. A further objective of the study was to use the rain shelter tents in combination with FACE, to create a scenario with two of the most relevant variables of global change that affect plant growth. In this paper, we examine the impact of the tents on the performance of CO2 enrichment on the field plots. To our knowledge, the approach presented here is the first larger scale rain shelter system described that allows for precipitation control and that does not require permanent installations, thus offering an extended flexibility.

Materials and methods

Field Site

The experiments were carried out on a field site of 20 ha in size at the Johann Heinrich von Thünen-Institute, Federal Research Institute for Rural Areas, Forestry and Fisheries, in Braunschweig, Germany (52°18′N, 10°26′E, 79 m a.s.l.). Annual climate averages of the field site are as follows: mean temperature 8·8 °C, mean temperature July (warmest month) 17·0 °C, total precipitation 618 mm (half of it deposited between May and September) and 1514 h of sunshine. The seasonal climate data are presented in Table 1. The soil at the experimental area is a luvisol of a loamy sand texture in the plough horizon (69% sand, 24% silt, 7% clay). In a depth of about 0·8 m, it is followed by a sand and gravel layer of more than 3 m in size. A more detailed description of the experimental site is given elsewhere (Weigel et al. 2005). In 2008, the experiment was carried out on maize (Zea mays L., cv. ‘Romario’; row spacing 0·75 m; 15 plants m−2). In 2010, each half of a plot was divided into six small subplots with a ground area of ca. 20 m2 each, with one maize cultivar (Zea mays L., cv. ‘Simao’; row spacing 0·75 m; 10 plants m−2) and five different sorghum cultivars (4 cvs. Sorghum bicolour Moench, 1 cv. Sorghum bicolour × sudanenese; all cultivars: row spacing 0·75 m; 20 plants m−2) established.

Table 1.   Monthly climate data of the two experimental seasons and the respective 30-year averages for the field site. Data on temperature and global radiation (Global rad.) are given as means. Data on precipitation and precipitation excluded are totals, with percentage of precipitation excluded referring to the respective months.
 Period in timeJuneJulyAugustSeptemberMean/Total
Temperature [°C]1971–200015·717·717·613·915·6
Global rad. [MJ m−2 day−1]1971–200018·818·216·110·516·3
Precipitation [mm]1971–200069·955·856·848·1283·2
Precipitation excluded [mm/%]2008 25·7/44·9%28·1/35·7% 53·8/39·5%
2010  159·2/81·8%16·4/15·6%175·6/58·6%

Rain Shelters

In the years 2008 and 2010, six large-sized tents (Röder, Büdingen, Germany) with a ground area of 12 m × 20 m, a ridge height of 5·5 m and a height of the side walls of 3·3 m equipped with transparent polyvinyl chloride (PVC) tarpaulins were used as rain shelters (Fig. 1). The construction principle of the tents consisted of basic aluminium framework elements of 12 m in width and 5 m in length each, with four of these chosen for each tent to obtain the respective ground area. Of each of six circular field plots of 20 m in diameter, one half was designated as water stress subplot (WS), with the tents placed in the manner that they completely covered the respective WS areas. The other halves of the plots were designated as well-watered subplots (WW), with drip irrigation applied to keep volumetric soil-water content above 50%, thus ensuring sufficient water supply to the plants. Both treatments were separated by walkways of 1·5 m in width to obtain access to the subplots and to have a sufficient distance between the WW and WS areas. Because of the aforementioned coarse, sandy texture and the very good drainage characteristics of the soil at the experimental site, no soil barriers were used between the treatments. The assembly of the tent frames in the field was carried out manually in the beginning of the respective experimental seasons. At all of the 10 posts of each tent frame, base plates were installed, each of which fixed with four ground anchors (20 mm × 0·8 m) on the soil surface. Most of the time, the frames were left without tarpaulins, and only during periods with high amounts of precipitation forecast (>10 mm day−1), the transparent tarpaulins were installed manually. To drain off the intercepted amounts of water at least 20 m away from the plots, gutters (180 mm in width), downspouts and drainage pipes (125 mm in diameter) were installed at each corner of the frames. In times when the tarpaulins were not installed, they were stored rolled up on reeling devices, which were installed on the vertical posts of the frames. The installation of the reeling devices and tarpaulins was performed later in the growing seasons, when the plants were already larger in height than the mounting position of the reeling devices (1·2 m and 1·5 m above ground) to minimise potential shading of the plants.

Figure 1.

 Field plots with the frames of the rain shelter tents (no tarpaulins or reeling devices installed). At the plot in the front, the free-air CO2 enrichment (FACE) area is denoted by the white circle (small black tubes: vertical vent pipes for CO2 release). One half of the FACE plot is covered by the tent (water stress treatment) and the other half outside the tent is equipped with drip irrigation (well-watered treatment).

Free-air CO2 Enrichment

The tents were used in an experiment in combination with FACE, engineered by Brookhaven National Laboratory (NY, USA). Each FACE apparatus consisted of an octagonal plenum on the ground serving as CO2 supply line that surrounded the plots. Facing towards the plot’s centre, 32 vertical vent pipes were connected to the plenum. Atmospheric CO2 enrichment was achieved by releasing CO2 enriched air on the upwind side right above the canopy from nozzles in the vertical vent pipes. Depending on wind speed and air refresh rate on the plots, the amount of CO2 was adjusted combined with a feedback control in accordance to the [CO2] measured at the plots. Throughout the growing seasons, the height of the CO2 release nozzles at the vertical vent pipes was adapted to crop growth. For further details on the FACE apparatus, see Lewin et al. (1994) and Hendrey & Miglietta (2006). The treatments included three plots enriched with CO2 (i.e. FACE) and three control plots with ambient air [CO2] (i.e. AMB). The set point [CO2] in the FACE plots during daylight hours was 550 μmol mol−1 in 2008 and 600 μmol mol−1 in 2010. Atmospheric CO2 enrichment was started when the leaf area index of plants was about 1 m2 m−2 (9 June 2008; 14 July 2010) and stopped at final harvest. At a mean wind speed >5·5 m s−1 of longer than 5 min CO2 enrichment was interrupted until wind speed was below this limit again.

Microclimate Measurements

Radiation transmission of the PVC used for the tarpaulins was examined in the laboratory with a spectrophotometer (U-2000; Hitachi, Tokyo, Japan) on the basis of three replicates. Measurements were conducted with wavelength between 300 nm and 1100 nm that also comprises the wavelength of photosynthetic active radiation (PAR – 400 nm to 700 nm). The impact of the experimental set-up on total and diffuse incident PAR in the field was evaluated from measurements conducted in the 2008 growing season at a WW and a WS subplot. Sensors were placed right above the plant canopy (BF3; Delta-T, Cambridge, UK). In the evaluation, daily PAR means for the entire growing seasons, the times without tarpaulin and with tarpaulins are compared. At all WW and WS subplots of each AMB and FACE plot, data on air temperature and relative air humidity in the canopy were recorded during the growing seasons (dry and humid air temperature by Pt100 thermocouples, Deutscher Wetterdienst, Braunschweig, Germany). Sensors were placed and regularly adjusted to three-quarter of canopy height.

Measurements of [CO2]

Throughout the periods in time with CO2 enrichment, [CO2] at the AMB and FACE plots were recorded continuously with infrared gas analyzers (LI-6262 and LI-840A; LI-COR, Lincoln, NE, USA). An important criterion to evaluate the performance of CO2 enrichment on plants is the fraction of time in which the desired [CO2] was achieved. Hendrey et al. (1997) examined the duration and the amplitude of fluctuations in [CO2] sensed by photosynthesis and found that fluctuations of less than ±20% from the chosen setpoint [CO2] and of less than 1 min in time have no effect on photosynthesis. Hence, a commonly used criterion to evaluate whether the predefined [CO2] was reached satisfactorily is the fraction of time of CO2 enrichment in which 1-min [CO2] means are within ±20% of the setpoint. To describe the impact of the tents on the performance of CO2 enrichment, results for the periods in time with and without tarpaulins as well as for the entire growing seasons were examined.


The operation time of the tents with transparent tarpaulins installed was based on the experimental protocol (installation of tarpaulins in cases of precipitation >10 mm day−1 forecast), with the aim to create drought stress in the WS treatment areas. The two growing seasons of the experiment were considerably different from each other (Table 1). In the 2008 growing season, the tents were operated with tarpaulins for 9 days in July and August (6·3% of the growing season). In contrast, in the August of the 2010 growing season, the highest precipitation amounts ever were recorded for Braunschweig (194·6 mm). Hence, the tarpaulins had to be installed for 20 days in August and September of 2010 (24·1% of the growing season). The operation of the rain shelters resulted in a precipitation intercept of 39·5% in 2008 and of 58·6% in 2010 in the respective months. Results for 2010 could have been even better, but from the middle of September, it was decided not to continue the drought stress treatment because of the progressed maturity of the sorghum crop. Effects on plant-available soil water content will be published elsewhere (R. Manderscheid, M. Erbs & H.-J. Weigel, unpublished data).

The radiation measurements conducted under laboratory conditions revealed that the PVC of the tarpaulins had a transmission of 85·7% and 78·3% in the range of PAR and global radiation, respectively. For wavelengths below 400 nm, a sharp decrease in transmission was found, so UV radiation is largely blocked by the PVC (Fig. 2). In the 2008 growing season, the frames alone caused a 6·6% decrease in total PAR in the field (Fig. 3), while diffuse PAR was reduced by 2·2%. In times with tarpaulins installed, total PAR was decreased by 24·1% and diffuse PAR was decreased by 23·8%. For the entire growing season 2008, this resulted in a 7·3% decrease in total PAR in the WS subplots. Differences in air temperature and air relative humidity between the WW and the WS treatment areas for the respective 2 months with rain shelters in operation are shown in Fig. 4. Except for short periods early in both of the seasons, higher temperature and a lower relative humidity were found at the WS compared with the WW subplots. Without tarpaulins, temperature was increased in the WS subplots by 0·3 °C and 0·2 °C, and relative humidity was reduced by 1·7% and 1·6% in the respective times in 2008 and 2010. Having the tarpaulins installed (Fig. 4– shaded areas), temperature was increased by 0·3 °C in the two seasons, and relative humidity was reduced by 2·7% and 1·8%, respectively.

Figure 2.

 Radiation transmission of the PVC used for the tarpaulins determined by spectrophotometry, with (b) results for photosynthetic active radiation (PAR) given with standard deviations of measurements.

Figure 3.

 Comparisons of daily means (24 h) of total incident photosynthetic active radiation (PAR) measured in the open field and under a tent frame in the 2008 growing season. Open circles and upper linear equation: times without tarpaulins; closed squares and lower linear equation: times with tarpaulins installed.

Figure 4.

 Difference of daily means (24 h) of air temperature (Temp) and relative humidity (Rh) between the results for the well-watered (WW – no tents) and the water stress subplots (WS – rain shelter tents) given for ambient air CO2 concentration (AMB) in the growing seasons 2008 (upper graph) and 2010 (lower graph), respectively. Shaded areas represent periods in time, in which the tarpaulins were installed on the tent frames.

In both growing seasons, diurnal hourly means of elevated [CO2] did not vary significantly with the predefined setpoints, with lower standard deviations found in 2008. Compared with results for the entire growing seasons, slightly more precise [CO2] and lower standard deviations were found in the periods of time with no tarpaulins installed. In times with tarpaulins, [CO2] varied stronger and had increased standard deviations. The fractions of time in which 1-min [CO2] means were within limits of ±20% from the setpoint [CO2] are shown in Fig. 5. For the entire growing seasons, [CO2] was found to be within ±20% limits for 99·6% and 98·6% of the time in 2008 and 2010, respectively. When the frames were left without tarpaulins, the fraction of time within ±20% limits increased to 99·7% and 99·0% in 2008 and 2010, respectively. When tarpaulins were installed, the time in ±20% limits was decreased to 97·8% in 2008 and to 96·7% in 2010.

Figure 5.

 Fractions of time, in which 1-min mean CO2 concentrations were within limits of ±20% of the setpoint CO2 concentrations. Results for the 2008 growing season (a) refer to a setpoint CO2 concentration of 550 μmol mol−1 and results for 2010 (b) refer to a setpoint of 600 μmol mol−1, respectively. Tarps: tarpaulins.

For further information about impacts on total and diffuse radiation and on 1-h mean [CO2], see the Supporting information.


The aim of the study presented here was to grow crops under climate change scenarios with temporal drought stress caused by rain exclusion. This aim was reached in both growing seasons. Relevant amounts of precipitation were intercepted when the rain shelter tents were in operation. On the basis of the experimental protocol, tarpaulins were installed three times in 2008 and four times in 2010. This resulted in an interception of 39·5% and 58·6% of the precipitation in 2 months in 2008 and 2010, respectively. Remarkable results were achieved for the very rainy August of 2010, when 81·8% of the monthly precipitation was intercepted. A rain shelter system operated during night times only is reported by Beier et al. (2004), with efficiency in rain exclusion between 64% and 95%. These results are comparable to the ones of the current system and underline the effective operation. The tents demonstrated to be capable for manual assembly in the field during the growing season without the need for using heavy-weight construction machinery and concrete foundations. Hence, the installation of the tents caused no relevant negative secondary effects on soil conditions like compaction and impacts on soil-water relations. This is a clear advantage over other large-scale rain shelter systems described before (Foale, Davis & Upchurch 1986; Svejcar, Angell & Miller 1999; Fay et al. 2000), all of which need some kind of foundation or basement for their assembly.

As a matter of principle, the basic objective of a rain shelter is to exclude precipitation from the area below, but not to cause any secondary effects. In reality, the best agreement between intercepting rain and the risk of causing microclimatic alterations (e.g. impacts on temperature, relative air humidity and radiation) has to be considered. In times when the tarpaulins were installed, a 24·1% reduction in PAR was found in the current study. The transparent PVC used for the tarpaulins decreased PAR by 14·3%, while the frames alone caused a 6·6% decrease in PAR. Compared with reductions in total PAR of the tents with tarpaulins installed, the difference of 3·2% is suspected to attribute to dust formation on the tarpaulins. Radiation use efficiency for plant biomass production increases when the fraction of diffuse radiation increases (Sinclair, Shiraiwa & Hammer 1992). We expected an increase in the diffuse radiation owing to the radiation scattering from the frames and the tarpaulins, which could compensate for the decrease in total radiation. However, measurements in 2008 showed no substantial change in the ratio of the fractions of total and diffuse radiation. Although the tents with tarpaulins installed reduced incident PAR by 24·1%, this reduction is suspected to be of minor importance for plant growth, because in most of these times, cloudy weather with low light intensity predominated. Nevertheless, net primary production of a plant canopy is linearly related to incident PAR as shown for biomass data obtained over periods of weeks (Monteith 1977) and also for canopy net CO2 uptake over periods of minutes (Ruimy et al. 1995). Thus, the reduction of 7·3% found for a growing season of the present study is suspected to have slight impact on plants and has to be accounted for in the evaluation of data.

The results of the current study are at least parallel to those reported for other rain shelter systems. The scaffolding of the automatic rain shelters presented by Beier et al. (2004) covered about 5% of the area of the experimental plots, which results at least in a similar size of shaded area. Under permanent rain shelters covered with polyethylene foil, incident radiation was reduced by 22% (Fay et al. 2000). The permanent shelters presented by Svejcar, Angell & Miller (1999) had a roof covered with transparent fibreglass, which caused a decrease in global radiation of 29% and an even greater decrease in PAR of 50%. A permanent rain shelter system with a more transparent but also more costly material used for the roofs is described by Yahdjian & Sala (2002). For a treatment with 80% of the roof area covered with bands of acrylic glass, a mean reduction in PAR of 10% and a maximum decrease of 25% is reported. Relying on roofs of highly transparent but also expensive polycarbonate, De Boeck et al. (2011) found reductions of PAR between 5% and 15% under their rain shelters. Facing the comparably low price of the PVC used for transparent tarpaulins in the current study, an excellent cost-benefit ratio was achieved on a per area basis. However, as the transparent PVC largely blocks UV radiation, it is not well suited for permanent roof installations.

As a matter of fact, every transparent roof material causes not only a reduction in incident radiation but is also a barrier for radiation reflection and airflow, all of which lead to alterations in air temperature and relative humidity under a rain shelter. For their smaller-scaled (3·35 m × 3·35 m) permanent rain shelters, De Boeck et al. (2011) reported an increase in temperature of 0·2 °C and an increase in relative air humidity of 3%. The rain shelters presented by Yahdjian & Sala (2002) that had the roofs permanently covered for 80% with acrylic glass caused a decrease in air temperature of 5·6 °C under high ambient air temperatures. The authors suspect the reduction in PAR of 10% as the reason for this unexpected finding, while in other studies, temperatures have been found to be increased under rain shelters. In spite of the larger size of our rain shelters, results from the current study are in the order of magnitude as those presented by De Boeck et al. (2011). Drought stress conditions like in the WS subplots reduce the transpiration of plants, thus raising air temperature and reducing relative air humidity. This is also reflected in the differences between the WW and WS subplots, which were temporarily even more prominent in the times with no tarpaulins installed. Hence, the canopy feedback on the microclimatic conditions at the WS subplots is clearly demonstrated. In comparison with this, it is assumed that impacts on temperature and relative humidity by the rain shelters must have been negligible, what is also reflected in microclimate data.

The drought stress experiment presented here was carried out in combination with free-air CO2 enrichment as a second treatment, having setpoint [CO2] of 550 μmol mol−1 and 600 μmol mol−1 in 2008 and 2010, respectively. Seasonal [CO2] means exactly reached the respective setpoints, while hourly seasonal means revealed slightly negative impacts of the tarpaulins. Hourly results for 2010 vary slightly stronger from the setpoint, because of the fact that a higher setpoint [CO2] was chosen making CO2 enrichment more prone to changes in wind speed. Results on the performance of CO2 enrichment on the basis of the percentage of time within ±20% limits (Hendrey et al. 1997) have been published for several other studies and are summarised in the review of Hendrey & Miglietta (2006). In the present study, the ±20% limit was reached nearly all of the time, with a slight decrease for the times with tarpaulins installed. However, results are at least in the order of magnitude reported for normal operation of other FACE facilities (Hendrey & Miglietta 2006).

Based on the technical prerequisites and the experimental protocol, the set-up of the current study was capable to create drought stress by causing prolonged intervals between rainfall events. Prolonged drought stress periods are forecast to be very likely for the next decades under the advent of climate change (Burke, Brown & Christidis 2006; Meehl et al. 2007). In the study by Fay et al. (2000), different alterations in precipitation pattern were tested and prolonged intervals were shown to have the most pronounced impact on the plants, while reductions in precipitation quantity alone had much less effect. Hence, experimental studies with temporal rain exclusion like the current one will be crucial for the understanding of anticipated drought stress impacts on crops. Field experiments on crops necessitate facilities that are large enough in size to cover an adequate area under examination. The rain shelter tents presented here were even large enough to carry out an experiment on six different subplots with tall-stature crops per semi circle. Because the tents have a variable construction principle, it is possible to reduce the height of the ridge and the side walls to 4·9 m and 2·4 m, respectively and to elongate the frames by steps of 5 m (e.g. 25 m, 30 m etc.). This offers the opportunity to adapt the assembly to different experiments. A further advantage of the tents is that by relying on a commercial product, the stability requirements of the construction are supplied by the producer.

The assembly of a single rain shelter tent in the field can be performed within 1 day (frame, gutters and tarpaulins). In our experiment, it took eight workers about 1 week to do the complete installation of the six tents with the equipment necessary for operation. All of this work was carried out manually without relevant impacts on the vegetation or soil and without permanent installations on the experimental site. A major advantage of the tents is that they can easily be used in different places in consecutive experiments resulting in a much higher flexibility compared with other large-scale rain shelters. Their flexible construction further offers the opportunity to combine them with other experimental techniques in multi-factorial experiments, e.g. FACE. The rain shelter tents presented here provide a versatile and cost-effective solution for well-replicated large-scale field experiments on the impacts of drought stress, one of the most relevant climatic changes.


The Johann Heinrich von Thünen-Institute, Federal Research Institute for Rural Areas, Forestry and Fisheries, Braunschweig, Germany, is part of the German Federal Ministry of Food, Agriculture and Consumer Protection (BMELV). In the growing season 2008, the experiment was part of the project LandCaRe2020. In 2010, the experiment was part of the project Bioenergie2021. Both projects were additionally funded by the German Federal Ministry of Education and Research (BMBF, support codes 01LS05108 and 0315421D). We are grateful to Florian Hackelsperger of the Friedrich-Loeffler-Institute, and Dr. Frank Höppner of the Julius Kühn-Institute, both Braunschweig, Germany, for their kind support with the agricultural fieldworks. Thanks to Dr. Klaus-Peter Wittich of the Deutscher Wetterdienst, Braunschweig, Germany, for microclimate measurements and providing climate data. We are grateful to Keith Lewin and Dr. John Nagy of the Brookhaven National Laboratory, NY, USA, for their support with the FACE system. And last but not least, we appreciate the work of the people involved in the experiments: Peter Braunisch, Andrea Kremling, Roland Isaak, Lothar Jurczyk, Reiner Mohr, Anke Mundt, Enrico Nozinski, Evelin Schummer and Ralf-Dietrich Staudte of the Johann Heinrich von Thünen-Institute, Braunschweig, Germany.