Arrangement of experiments for simulating the effects of elevated temperatures and elevated C 0 2 levels onfield-sown crops in Finland

The experimental plants: spring wheat, winterwheat, spring barley, meadow fescue, potato, strawberry and black currant were sown or planted directly in the field, part of which was covered by an automatically controlled greenhouse to elevate the temperature by 3°C. The temperature of the other part of the field (open field) was not elevated, but the field was covered with the same plastic film as the greenhouse to achieve radiation and rainfall conditions comparable to those in the greenhouse. To elevate the C0 2 concentrations, four open top chambers (OTC) were built for the greenhouse, and four for the open field. Two of these, both in the greenhouse and in the open field, were supplied with pure C0 2 to elevate their C0 2 level to 700 ppm. The temperatures inside the greenhouse followed accurately the desired level. The relative humidity was somewhat higher in the greenhouse and in the OTC:s than in the open field, especially after the modifications in the ventilation of the greenhouse and in the OTC:s in 1994. Because the OTC:s were large (3 m in diameter), the temperatures inside them differed very little from the surrounding air temperature. The short-term variation in the C0 2 concentrations in the OTC:s with elevated C0 2 was, however, quite high. The control of the C0 2 concentrations improved each year from 1992 to 1994, as the C0 2 supplying system was modified. The effects of the experimental conditions on plant growth and phenology are discussed.

climate.However, the rate of development of cereals like wheat and barley is strongly related to temperature.In conditions of high tempera- tures and long days, the development of cereal crops may proceed excessively fast, resulting in a smaller volume of assimilating biomass, a shorter grain filling timeand subsequently, lower yields than would be possible at lower temperatures (Mela et al. 1994).Warming of climate can also be assumed to increase damage caused by pests and pathogens at northern latitudes where their development and reproduction rates are currently limited by low temperatures (Mela et  al. 1994).
The present investigation was undertaken in order to evaluate, through direct experimentation, the possible impact of climatic changes on the growth and yield of the cereal crops spring barley, winter wheat and spring wheat; potato; a grass crop, meadow fescue and two horticultural plants, strawberry and black currant.Although many experiments have been conducted to study the effects of temperature and C0 2 on individual cropplants, few investigations have been made of crop stands, especially those under long photoperiod conditions experienced during the growing season at high latitudes.The experimental conditions were maintained as close to natural field conditions as possible: the experimental plants were sown directly in the field, at normal sward density.Optimal fertilization and irrigation were applied so that soil moisture and nutrients would not limit growth.Pests and diseases were controlled in all the treat- ments.
To elevate the temperatures and the C0 2 levels, but without the resources to conduct "free air" CO, enrichment (FACE) experiments (e.g., see Drake et al., 1985), an artificial system (greenhouse and open top chambers, OTC;s) was constructed inside of which the experimental plants were grown.This paper describes the ar- rangement of the experiment and evaluates the level of environmental control achieved.Fig. 1.The greenhouse built on the experimental field in order to elevate the temperatures continuously by 3°C, and the adjacent open field, covered with the same material as the greenhouse.

Elevated temperatures
In order to maintain the experimental temperatures at a constant 3°C higher than ambient tem- peratures, a greenhouse (20 m x 30 m) was built over an experimental field at Jokioinen, Finland (60°49'N, 23°30'E).The greenhouse was built with white, pressure-impregnated, laminated wood arches in an east-west direction.The shape of the greenhouse was the pointed arch type (Fig. 1) with continuous ridge ventilation.A standard ethyl vinyl acetate (EVA) antifog film (light transmission 60 % at PAR (400-700 nm)  wavelengths, 0 % at wavelengths under 360 nm) was used as the covering material.Air tempera- tures were regulated by an automatic intelligent controller system, ITU computer HS -outstation (Itumic OY, Jyväskylä, Finland), which was installed inside the greenhouse.This operated by opening and closing the overhead hatches on the roof of the greenhouse, or by heating the greenhouse with heater fans.In 1992-1993, in order  to achieve better air circulation inside the greenhouse in warm and sunny periods during the growing season, the greenhouse was also venti- lated through openings running along its longitudinal sides at the height of 1-1.5 m.However, this type of ventilation was found to decrease daytime temperatures at the edges of the greenhouse by about 0-1 °C, depending on the dis- tance from the openings, and in 1994-1995 an extra fan (type RP, 8.1 kW, OY Wikström AB, Loviisa, Finland) was installed in the centre of the greenhouse to mix the air, while the open- ings at the sides of the greenhouse were kept closed.
The experimental field outside the green- house, at ambient temperature, was covered at the height of 3-4 m with the same plastic film that was used in the construction of the green- house, so as to achieve radiation and rainfall conditions comparable to those in the green- house.Temperature and humidity were measured in the greenhouse and in the open field with Itu- mic ventilated psychrometers with Pt 100 sensors (Itumic OY, Jyväskylä, Finland) during all of the experimental years (1992)(1993)(1994)(1995) and inside the OTC:s in 1993-1995.The measurements in the open field were made under the plastic film covering the field.The temperature meas- urements conducted in 1992 inside the OTC:s (Hakala et al. 1993) were later found to be incorrect, because the radiation shields of unventilated thermometers used in the OTC:s in 1992 could not entirely prevent the thermometers from being warmed by radiation.Temperature and humidity recordings were taken every 4 minutes and then transferred to a master unit, from which data files could be obtained, and information processed on a personal computer.

Elevated C0 2 levels
For the experiments on elevated CO, concentra- tions, OTC:s were constructed according to the basic design described by Ashenden et al. (1992).Their design was modified greatly, however, to meet our demands.The OTC:s were cylindrical in shape with a diameter of 3 m and a height of 2 m.They were constructed from colourless corrugat- ed acrylate sheets (Vetricell, ER, photosynthetically active radiation (PAR) transmission 90 %) bolted together on an aluminium frame.The thickness of the sheets was 1.5 mm, corrugation depth 18 mm and corrugation width 76 mm.One side of one sheet was left unbolted, which served as a door for entering the chamber.An overhead frustrum covered the upper end of the chamber; it was 35 cm wide and constructed from 1.0 mm thick polymethyl meta-acrylate sheets.Eight chambers were constructed, providing two independent replicates for each C0 2 x temperature treatment.Both wheat and meadow fescue were grown in each chamber, so that the meadow fes- cue occupied the southern side and the wheat, the northern side of the chamber.In this way, the meadow fescue canopy, which was cut from time to time, was not shaded by the wheat can- opy.
The C0 2 gas was supplied to the chambers from a set of gas cylinders by vapour withdraw- al, with each cylinder containing 28-30 kg of pure C0 2 gas (Woikoski OY in 1992 and AGA  OY in 1993-1995, Finland).There were 12 cyl- inders together in a set, from which the gas was released through a heating device and a two stage pressure regulator, into rubber tubing of 12 mm in diameter, which led to a delivery stand inside the greenhouse.This main flow was divided into 8 minor flows, stopped with separate main valves.From this point, the gas was passed through solenoid valves regulated by an auto- matic measuring-dosing feedback-system (Itumic OY, Jyväskylä, Finland).
The feedback system functioned by sampling and analysing the air from two points in each chamber with elevated C0 2 , both in the green- house and in the outside field.Depending on the analysis, the solenoid valves between the influx and outflux tubing were either opened or closed in order to adjust the C0 2 levels in the chambers with elevated C0 2 , to the target C0 2 concentration, 700 ppm.Two separate sets of C0 2 dosings were applied for each chamber with elevat- ed C0 2 , functioning independently according to the two analysing systems in each chamber.As the system was originally planned to meet the needs of large greenhouse departments, it was unsatisfactory for chambers as small as our OTC:s.Therefore, in 1993 only one analysingdosing-set of the C0 2 -enriched chambers was set to operate automatically, while the other chan- nel was open all day, and shut off only for the night.In 1994-1995, the automatic function of the C0 2 dosing system was shut off totally, and the C0 2 levels were operated manually, by ad- justing the C0 2 flow rates according to the lev- els in the chambers.The air of the chambers with ambient C0 2 and over open air plots was ana- lysed, but no C0 2 adjustment was applied.
From the solenoid valves, the C0 2 gas was delivered to the chambers individually, through separate flow meters and rubber tubing of 7 mm in diameter.Inside the chambers, the rubber tub- ing was perforated and criss-crossed over the plant canopy, in order to make the gas flow even.
Overhead fans were attached to each chamber to mix the C0 2 gas with the chamber air, and also in an attempt to prevent the temperature in the chamber from rising above that outside the chamber.In 1994-1995, the fans were connect- ed to a perforated plastic tunnel system, in which C0 2 from one of the separate dosing sets was introduced together with air from the overhead fans, while the other C0 2 dosing set was still connected to the perforated rubber tubing.In the ambient C0 2 chambers, a similar ventilation sys- tem was arranged, but with no C0 2 added.The tunnel was a standard greenhouse ventilation tunnel comprised of a thin plastic tube 24 cm in diameter.This configuration was built in order to achieve more uniform C0 2 levels and airflow inside the chambers.
The C0 2 source (the perforated rubber tub- ing or the plastic tunnel) was placed above, rather than within, the plant canopy in order to avoid the problem of C0 2 depletion due to plant uptake in the canopy during photosynthesis.The standard measurements of ambient atmospheric C0 2 concentrations were made in free air, be- tween the C0 2 source and the plants.It is known that concentrations within a plant canopy can be significantly lower than these.
In 1993-1994, the C0 2 concentrations were sampled every 5-8 minutes by one of the two measuring channels in an OTC while in 1992, the sampling interval in each chamber was about 12 minutes.C0 2 measurements were transferred to a master unit for storage and analysis.

Experimental plots
Before the onset of the experiments, the heavy clay soil of the experimental area was mixed with peat containing 35 % sand.The peat-sand mixture was limed, but not fertilized and was added in the proportion of 0.1 m 3 of the mixture /m 2 of soil.In the autumn of 1993,this soil was replaced by sandy loam soil brought to the experimental field from another field in Jokioinen.The nutri- ent levels of the soil were tested before the sowings, and nutrients were added according to the test results, so that the nitrogen level of the soil would be approximately 120 kg/ha and the oth- er nutrients would not limit growth.
At elevated temperatures, inside the greenhouse, the experimental plots for the spring-sown cereals and the grass crop were sown at normal sowing density shortly after the thermal grow- ing season had started (i.e. when the temperatures were expected to stay constantly above 5 °C).According to this principle, in 1992-1995 the growing season started inside the greenhouse 25 April, 22 April, 19April and 15 April, respec- tively.In the open field (ambient temperatures) the thermal growing season started 27 April, 22 April, 22 April and 20 April.In the open field there was a delay in the sowing because, while the temperatures rose rapidly at the onset of the growing season, the cold weather before the be- ginning of the growing season did not allow the soil to dry enough for sowing.The sowing dates were thus 14 May, 10 May, 9 May and 10 May from 1992 to 1995.Potato was planted in the outside field at the beginning of June to avoid late spring frosts and inside the greenhouse about 3 weeks earlier.

Crops and treatments
The spring wheat (Triticum aestivum L. var."Polkka") and the grass crop, meadow fescue (Festuca pratensis Hudson var."Kalevi") were grown in the following four treatments: a) am- bient air temperature and ambient C0 2 concentration; b) increased temperature (3°C above ambient) and ambient C0 2 ; c) ambient tempera- ture and elevated C0 2 (700 ppm), and d) increased temperature and elevated C0 2 .In an effort to estimate the chamber effect, the phenology, growth and yield of the experimental plants inside the chambers were compared to those of the plants growing on the adjacent plots with no chamber (open air plots).

Results and discussion
Temperatures Daily mean temperatures in the greenhouse and in the open field (both in the OTC:s and in free air) during the period 22 July -10 September in 1993 and 1994 are presented in Figure 2. The reason for choosing this period, is that the meas- urements of the temperatures were available in all the treatments, for both years, which makes comparisons between the different ventilation systems in the OTC:s possible.The results for 1995 are not shown here because the ventilation system of the greenhouse and in the OTC:s re- mained the same in 1995 as in 1994.
In the diagrams in Fig. 2, it can be seen that the greenhouse temperatures were constantly higher than the ambient temperatures.The tem- perature difference between the greenhouse and the open field was on average 3.2 °C (+ 0.1 °C, range 2-4.4 °C) in 1993 and 3.0 °C (± 0.0 °C, range 2.2-3.4 °C) in 1994 during the period 22 July -10 September.Temperatures within the OTC:s closely traded those of the surrounding air in both 1993 and 1994.In 1993, the daily mean temperatures in the OTC:s during the pe- riod 22 July -10 September were on average -0.4 °C (+ 0.1 °C, range +1.2-2.6)higher than the surrounding air inside the greenhouse and in 1994 the OTC temperatures were on average 0.5 °C (+O.l, range +l.3-1.3 °C) higher than the greenhouse temperatures.The difference in the OTC and the greenhouse temperature was approximately the same (within + 0.1 °C) both dur- ing day (from 9-19) and night (from 19-9).In the open field, the OTC temperatures were ap-proximately the same as the surrounding air in 1993 (within + 0.05 °C, range +0.5--0.9°C) and 0.2 °C (± 0.1 °C, range +O.6-1.6)lower than the surrounding air in 1994.The difference in temperature between the OTC:s and the open field was approximately the same during night and day in 1993; while in 1994, the temperatures were 0.4 °C lower in the OTC:s during the day, and the same as the open field temperatures dur- ing the night.In Fig. 3 it can be seen that the ventilation system installed inside the OTC:s in 1994 was quite efficient in keeping the OTC tem- peratures at the level of the surrounding air in the greenhouse for both sunny and cloudy days (Figs. 3 B and 3 D).In 1993, the daytime cham- ber temperatures inside the greenhouse were 1-3 °C higher than the surrounding air tempera- tures on sunny days and about 0.5 °C higher on cloudy days (Figs.3A and 3 C).In 1993, the chamber temperatures in the open field were more or less the same as the ambient surrounding temperatures both on sunny and cloudy days (Figs. 3 A and 3 C); while in 1994 the new ventilation system resulted in the daytime temperatures inside the OTC:s in the open field being about 1 °C lower than the ambient temperatures (Figs. 3 B and 3 D).The reason for the chamber temperatures being lower than ambient temperatures could be that, in the open field, the fans inside the OTC:s provided higher rates of air- mixing than in the surrounding area, as the air in the open field was not mixed with fans, al- though it was covered with plastic film.
1-2 °C increases in daytime temperatures and 0.2-1 °C increases in nighttime temperatures inside the OTC:s have been reported earlier (Hea-  gle 1989;Ashenden et al. 1992).The daily mean temperatures in the OTC:s in our experiments differed very little from the surrounding temper- atures, possibly because of the large size and the truly open structure of the OTC:s.However, in 1992-1993 the phenological development of wheat inside the OTC:s was 1-3 days ahead of the open air plots in the greenhouse and in the open field.In 1994, both in the greenhouse and in the open field, and in 1995 in the greenhouse, the phenological development of wheat proceeded at the same speed in the open air plots and in the OTC:s.In 1995, in the open field, the corre- sponding phenological stages were recorded 2-5 days earlier in the OTC:s than on the open air plots.The higher rate of development in the OTC:s may have been caused by the slightly higher OTC temperatures in 1992 and 1993.The As air temperatures increased steadily dur- ing April-July and the crops were sown approx- imately three weeks earlier in the greenhouse than in the open field, the temperatures experienced by the crops at comparable development stages were not necessarily higher in the elevat- ed temperature treatment than in the ambient temperature treatment.
The beginning of the growing season for spring-sown crops was mostly cooler in the greenhouse than in the open field (Fig. 4).The effective temperature sum (ETS), base tempera- ture 5 °C, accumulated faster in the open field during the first 12-15 days after sowing (DAS) in 1992 and in 1993, after which the ac- cumulation rate of ETS became higher in the greenhouse (Fig. 4A and 4B).The ETS in the greenhouse exceeded that of the open field 39 DAS in 1992 and 29 DAS in 1993.In 1994 the accumulation rates ofETS in the greenhouse and in the open field were more or less the same dur- ing the growing season (Fig. 4C).In 1995 ETS in both the greenhouse and in the open field ac- cumulated similarly during 20 DAS, then the accumulation of ETS was significantly slower in the greenhouse until 40 DAS and again faster towards the end of season (Fig. 4D).Because of the slow accumulation of ETS in the greenhouse, it did not exceed that in the open field until 65 DAS.The phenological stages of spring wheat grown in the open air plots are marked on the ETS diagrams in Fig. 4. The ETS for anthesis and yellow ripening was more or less the same, both at ambient, and at elevated temperatures in 1992-1994.In 1995 the ETS for anthesis was 140 degree-days and for yellow ripening 148 degree-days higher in ambient than in elevated temperatures (Fig. 4 D).The reason for this may be that the response of development rate to tem- Fig. 3D.Temperatures in the greenhouse and in the open field and inside the open top chambers in the greenhouse and in the open field during a cloudy day in 1994 (10.33 MJ/m 2 /day).perature may not have been linear, as the temperature peaked in the greenhouse 40-47 DAS and remained high (daily mean temperature about 20 °C) until heading (56 DAS) and anthe- sis (58 DAS).In the open field again, the high temperature period experienced by the crops in the greenhouse after 40 DAS, occurred at 18 DAS, and it was not as pronounced, while the temperatures were then 3 °C lower in the open field than in the greenhouse.

Relative humidity
The relative humidity (RH) of air was almost the same in the greenhouse as in the open field and the RH in the OTC:s was approximately the same as the RH of the surrounding air both in the Fig. 4A.Effective temperature sum (ETS) vs. days after sowing (DAS) of spring cereals at elevated temperatures (greenhouse) and at ambient temperatures (open field) in 1992.The phenological stages of spring wheat "Polkka" are marked on the ETS curves: A = anthesis, YR = yellow ripening.The sowing dates inside the greenhouse and in the open field were, respectively, 29th April and 14th May in 1992.
greenhouse and in the open field in 1993.However, on a sunny day the RH of the OTC air inside the greenhouse was on average 5 % higher than the surrounding air (Figs.5A and SC).Slight increases (5 % or less) in the RH in the OTC:s have been reported earlier (Heagle 1989, Ash- enden et al. 1992).In 1994, the RH in the green- house was 17-19 % higher than in the ambient air, and 7-10 % higher in the OTC:s than in the surrounding air both on sunny and cloudy days (Figs.5B and SD).The higher RH inside the greenhouse compared to the open field in 1994 was probably caused mainly by the permanent closure of the side openings of the greenhouse in 1994, when an extra fan was installed to mix the greenhouse air and to replace the ventilation through the side openings.The larger volume of perennial plants, which had been growing from Fig. 48.Effective temperature sum (ETS) vs. days after sowing (DAS) of spring cereals at elevated temperatures (green- house) and at ambient temperatures (open field) in 1993.The phenological stages of spring wheat "Polkka" are marked on the ETS curves: A = anthesis, YR = yellow ripening.The sowing dates inside the greenhouse and in the open field were, respectively, 16th April and 10th May in 1993.the year 1992, may have also contributed to the higher humidities inside the greenhouse.The temperature differences between the OTC:s and the surrounding air in 1994 and 1993 may partly explain why the difference between the RH inside the OTC:s and in the surrounding air was greater in 1994 than in 1993.In 1994 the temperatures inside the OTC:s were lower than am- bient temperatures in the open field, and inside the greenhouse the temperatures in the OTC:s were lower in 1994 than in 1993 during the daytime (Fig. 3).
The higher RH in the OTC:s and in the green- house could have given the plants a competitive advantage compared with those in the open field.At a higher RH, plants are able to maintain a higher stomatal conductance, resulting in better penetration ofC0 2 into the leaves and thus, high-Fig.4C.Effective temperature sum (ETS) vs. days after sowing (DAS) of spring cereals at elevated temperatures (greenhouse) and at ambient temperatures (open field) in 1994.The phenological stages of spring wheat "Polkka" are marked on the ETS curves: A = anthesis, YR = yellow ripening.The sowing dates inside the greenhouse and in the open field were, respectively, 15th April and 9th May in 1994.er assimilation rates (Collatz et al. 1991).The comparison of plant growth between the greenhouse and the open field is thus complicated, especially after the increase in the RH of the greenhouse after 1994.However, both inside the greenhouse and in the open field, the comparison of plant growth between the C0 2 enriched OTC:s and the OTC:s with ambient C0 2 can be made reliably, as the ventilation systems in the OTC:s were identical, and thus the RH levels in the OTC:s should have been identical.
Increases in temperature and the RH, decreases in light intensity and changes in wind velocity and boundary layer resistance of the plants caused by an OTC have sometimes been found to affect plant growth and biomass yield inside the chambers (Heagle 1989).The effects have, however, been quite small and they vary accord- Fig. 4D.Effective temperature sum (ETS) vs. days after sowing (DAS) of spring cereals at elevated temperatures (green- house) and at ambient temperatures (open field) in 1995.The phenological stages of spring wheat "Polkka" are marked on the ETS curves: A = anthesis, YR = yellow ripening..56 MJ/mVday) and during a cloudy day in C) 1993 (10.16 MJ/m 2 /day) and D) 1994 (10.33 MJ/m 2 /day).
ing to the plant species, plant cultivar, growing season and cultivation site (Heagle 1989).In our experiments, the biomass ofmeadow fescue was on average (over the years 1992-1995) 2 % high- er (varying from 13 % lower to 15 % higher) in the OTC than in open air plots in the greenhouse and 20 % higher (9-37 % higher) in the OTC in the open field.The yield of wheat was on aver- age 15 % higher and the biomass 7 % higher in- side the OTC than in open air plots in the open field, while in the greenhouse both the yield and biomass of wheat were 13 % lower in the OTC than in open air plots.
Even though the levels of radiation were about 40 % lower under the plastic film cover- ing both the greenhouse and the open field, the yield of meadow fescue "Kalevi" on the open air plots was 30 % higher in the open field and 70-170 % higher in the greenhouse as compared to the 8 year average yield of the same variety in the official variety tests over a comparable area in Finland.The yield of the spring wheat "Polkka" in the open air plots was 40-50 % higher in the open field and 20-60 % higher in the greenhouse than the 8 year average yield of the same variety in the official variety tests.

C0 2 levels
The average C0 2 concentrations in Table 1 were computed over the two C0 2 -enriched chambers, referred to with numbers 1 and 2 inside the greenhouse, and 3 and 4 in the open field.First, hourly mean values were calculated from the concentration samples drawn every 5-12 minutes.Then, the average concentrations were computed from the hourly values (6 a.m. to 6 p.m.) over periods varying from almost 5 months (1993, elevated temperatures) to 2.5 months (in 1994) (the exact periods are listed in the legend of Table 1), which means that the figures in Ta- ble 1 are computed from 147-76 hourly obser- vations.The variation of the C0 2 concentrations in the individual OTC:s is expressed as the stand- ard error of the mean.The concentrations in each chamber are the means of two measuring chan- nels.The data for the year 1995 is not present- ed, because the C0 2 dosing system was the same as in 1994.
It can be seen in Table 1 that the control of C0 2 levels inside the chambers improved from year to year as the system was modified.The best results were achieved with the system introduced in 1994, in which C0 2 -enriched air was delivered through the plastic tunnel.As C0 2 levels were substantially higher during the night, only the hours with levels of light intensity high enough to support photosynthesis throughout the growing season are listed here.The nighttime accumulation ofC0 2 in the chambers was great- er inside the greenhouse than in the outside field.This is partly due to crop and soil respiration.The C0 2 released could not escape from the greenhouse, as the hatches were usually closed at night to maintain the desired temperature lev- els.
The daytime variations of the C0 2 levels in the C0 2 -enriched chambers, both inside the greenhouse and in the open field in 1993 and 1994, are presented in Fig. 6.The variation in the C0 2 levels in the OTC:s in 1992 has been presented earlier (Hakala et al. 1993).The mean daytime C0 2 levels recorded from one measur- ing point in one chamber ranged from 600-800 ppm and 650-1000 ppm inside the greenhouse and from 550-650 ppm and 600-700 ppm in the open field in 1993 and 1994, respectively.The variation in the C0 2 levels was larger inside the greenhouse than in the open field in both 1993 and 1994.In 1993 the variation was smaller than in 1994, when the C0 2 levels were higher (Fig. 6).The larger variation of the C0 2 levels inside the greenhouse in 1992 compared to 1993 is explained by the different method of measurement used in 1992 which resulted in a longer time lag (12 minutes) of measuring the C0 2 levels in the individual chambers.This emphasises the problems of feedback systems with long time delays.In addition to the long time delay, both dosing channels were operated automatically in 1992, which led to unsatisfactorily low C0 2 levels.
The CO, levels in the chambers not fed with C0 2 , both inside the greenhouse and in the open Table 1.Mean C0 2 levels in the C0 2 -enriched open top chambers during the day inside the greenhouse (OTC 1 and 2) and in the open field (OTC 3 and 4), and the difference of the mean C0 2 levels (SEM) between the replicate OTC:s in 1992,   1993  field, also fluctuated diurnally, being highest during the night and lowest during the day.The mean daytime C0 2 concentrations were approximately the same both in the OTC:s with ambi- ent C0 2 and over open air plots, i.e. 380-400 ppm in the greenhouse and about 350 ppm in the open field.
In previous investigations with large OTC:s, the daily mean C0 2 values inside the chambers were within 9 % of the desired value (Drake et  al. 1985).The variation of the mean C0 2 values was smaller (less than 6 %), when the chambers were smaller (diameter 1.28 m and 1 m high) and partially covered (average of three cham- bers over a 24 h period, Ashenden et al. 1992)  or as big as the chambers in our experiment, but partially covered (average daily C0 2 levels over a period from May to August, Fig. 4 in Weigel  et al. 1992).In the present study, in an effort to keep the conditions as close to natural as possible, the OTC:s were large, and there were no extra covers other than the frustrum on top of the OTC:s to control the CO, levels.

Conclusion
The elevation of the temperature succeeded well with respect to the temperature difference between the greenhouse and the open field, and could be controlled quite accurately.The con- trol of the temperatures required the construc- tion of an artificial system, the greenhouse, inside which conditions were not fully compara- ble with the natural (i.e.field) environment.The major differences compared to natural conditions were 1) the light intensity, which was cut by 40 % at the PAR wavelenght (400-700 nm) region and totally at the UV-B region 2) the RH which was on average 17-19 % higher inside the greenhouse than in the open field during the daytime, particularly after the modification of the greenhouse ventilation system in 1994 (ventilation through the side openings was stopped and the air was mixed with a fan) and 3) wind velocity.The differences in wind velocity between the greenhouse and the open field were not studied, Fig. 6.Mean C0 2 levels during the day and the 0.05, 0.25, 0.75 and 0.95 percentiles of the observations in the greenhouse in A) 1993 and B) 1994 and in the open field in C) 1993 and D) 1994.The means and the percentiles were calculated from the hourly means of the observations during the periods 21st April -14th September in the greenhouse and Ist June -14th September in the open field in 1993 and Ist July -14th September in 1994.
but wind velocity is expected to be higher in the open field than inside the greenhouse.The tem- perature control plots that were sown on the ad- jacentfield and covered with the same plastic as the greenhouse were protected from rainfall and received approximately the same amount of incident light as those inside the greenhouse.How- ever, the RH and wind conditions in the open field remained different from the conditions in the greenhouse.
The OTC:s built for the CO z experiments had little effect on the temperatures, but the RH in the chambers was 7-10 % higher than in the sur- rounding air after the new ventilation system was installed in 1994.This system passed both C0 2 and air into the chamber through a perforated ventilation tunnel.In 1993, the RH in the OTC:s was more or less the same as in the surrounding air except in the greenhouse on sunny days, when the RH was about 5 % higher in the OTC:s.In 1992, the first year of experimentation, the C0 2 levels in the C0 2 -enriched OTC:s were quite satisfactory inside the greenhouse, but in the open field the C0 2 concentrations could not be kept at a high enough level.The control over the C0 2 levels in the C0 2 -enriched OTC:s improved in 1993 and still further in 1994, when the new ventilation system was installed in the OTC:s.
The variation in the CO, levels could not be re- moved even with this new system.In spite of the variation, the C0 2 levels in the C0 2 -enriched OTC:s were near the target of 700 ppm most of the time during the years 1993-1994 both in the open field and inside the greenhouse.

Fig
Fig. 2A.The ambient daily mean temperatures and the daily mean temperatures inside the open top chambers in the green- house and in the open field from 22nd July to 10th September in 1993.

Fig. 28 .
Fig. 28.The ambient daily mean temperatures and the daily mean temperatures inside the open top chambers in the green- house and in the open field from 22nd July to 10th September in 1994.

Fig. 3A .
Fig. 3A.Temperatures in the greenhouse and in the open field and inside the open top chambers in the greenhouse and in the open field during a sunny day in 1993 (21.49MJ/m 2 /day).
delay in the development in the open air plots in the open field in 1995 compared to the development in the OTC:s cannot be explained by higher OTC temperatures because the OTC temper- atures should have been lower than the ambient in the open field in 1995.

Fig. 38 .
Fig. 38.Temperatures in the greenhouse and in the open field and inside the open top chambers in the greenhouse and in the open field during a sunny day in 1994 (21,56 MJ/m 2 /day).

Fig. 3C .
Fig. 3C.Temperatures in the greenhouse and in the open field and inside the open top chambers in the greenhouse and in the open field during a cloudy day in 1993 (10.16 MJ/mVday).

Fig. 5 .
Fig. 5. Relative humidity in the greenhouse and in the open field and inside the open top chambers in the greenhouse and in the open field during a sunny day in A) 1993 (21.49MJ/mVday) and B) 1994 (21.56 MJ/mVday) and during a cloudy day in C) 1993(10.16MJ/m 2 /day) and D) 1994 (10.33 MJ/m 2 /day).
and 1994.The hourly C0 2 level values are means of the values during 14thMay -22nd September in 1992,215 t April -14th September in the greenhouse and Ist June -14th September in the open field in 1993 and Ist July -14th September in 1994.