Two experiments were conducted in 2018 and 2019 using sugar beet varieties from different breeders with contrasting canopies ( Figure 1 ), Cayman with a prostrate canopy (Prostrate) and Sabatina KWS with an upright canopy (Upright). To simulate a realistic canopy environment sugar beet were grown in plastic pallet boxes with a volume of 610 L, depth of 60 cm and surface area of 1.1 m². Boxes had drainage holes drilled in the bottom with membrane overlaid and filled with a sandy clay loam, (Landscape20, Topsoil, Cambridgeshire, UK). The boxes were filled in four stages, with a minimum of 6 days between each stage, and hand watered to settle the soil. Volumetric water sensors (ECH₂0 EC-5, Meter group Inc, USA) were buried 15 cm from the bottom of the box in 2018 whilst in 2019 larger sensors (ECH₂0 10HS, Meter group Inc, USA) were buried at 30 cm to get a reading over a larger soil volume. Sensors were calibrated specifically to the soil used, as directed by the manufacturer’s protocol. Volumetric water content (VWC) was logged every hour (Em5b, Meter group Inc, USA). The boxes were placed in an open ended polytunnel without environmental controls orientated East to West and covered in a diffuse polythene (SunMaster Diffused, XL Horticulture LTD, Devon, UK) in 2018 and a clear polythene (SunMaster Clear, XL Horticulture LTD, Devon, UK) in 2019. In both years, the boxes were arranged in a split plot design, with irrigation regime on the main plot and variety on the sub plot. Measurements were taken on 32 boxes divided into four blocks of eight with discard boxes at the end of each row to minimise edge effects. A temperature and humidity sensor was suspended at canopy height and logged measurements every hour (TinyTag Ultra 2, Chichester, UK). In 2019, an additional sensor was suspended at the end of blocks 2 and 4 to identify if a temperature gradient was present but no differences were identified. Thermal time was higher in 2019 than 2018 with differences most evident in August and September ( Supplementary Figure S1 ).

To ensure an optimal seedbed the boxes were raked to produce a fine tilth. A plywood board with holes drilled for correct seed spacing ( Supplementary Figure S2 ) was placed over the box and three seeds sown in each hole. Boxes were hand watered at regular intervals, and timed to ensure equal watering, to prevent soil drying and ensure good establishment. Plants were thinned once two true leaves were evident to give a total of 12 plants per box. Boxes were fertilised with ammonium nitrate in 2018 and ammonium sulphate in 2019, using a split application equating to 40 kg N ha^-1 followed by 80 kg N ha^-1. After each application of fertiliser, the boxes were watered equally. In both years the seeds were sown on 9 April, in 2018 fertiliser was applied at 15 and 29 DAS with thinning at 32 DAS, whilst in 2019 fertiliser was applied at 16 and 29 DAS with thinning at 29 DAS.

An irrigation system consisting of drip irrigation pipe was installed after emergence with three lengths of pipe running between the sugar beet rows ( Supplementary Figure S2 ). Four irrigation treatments were applied; a fully irrigated control (Full), a single drought (SD), a double drought (DD) which had the SD treatment plus a second period of drought and a water limited treatment (Ltd) which was kept at approximately 50% field capacity. The second water withdrawal for the DD started when the maximum rate of assimilation (A_max) returned to a level similar to the fully irrigated plants. Boxes were irrigated back to field capacity immediately at the end of a water withdrawal period. Timings were comparable between the two years, except for the DD which was later in 2018 than 2019. The first water withdrawal was at 65-96 DAS in 2018 and 73-92 DAS in 2019 and the second at 151-200 DAS in 2018 and 129-148 DAS in 2019. In 2018, the total amount of water applied per box was; Full – 110.9 L, SD- 102.1 L, DD-75.9 L and Ltd- 41.3 L and in 2019; Full – 102 L, SD- 90.1 L, DD-83.6 L and Ltd- 28.9 L. Soil moisture was monitored using the VWC sensors in 2018 and 2019 ( Supplementary Figures S3A, B ) and irrigation adjusted accordingly.

Spot measurements of gas exchange measurements of maximum assimilation (A_max) and stomatal conductance (g_s) were taken using an infra-red gas analyser (LiCor Li6800, LiCor, USA) and used to calculate WUE_i (Condon et al., 2002).

The two-sugar beet located centrally in the box (6 and 7, Supplementary Figure S2 ) were measured to avoid any potential edge effects. Settings were: flow 500 µmol s⁻¹, heat exchanger temperature 20°C (which gave a leaf temperature between 20°C and 28°C dependent on ambient conditions), RH 50%, light 1200 µmol m⁻² s⁻¹, and matched every 10 minutes. Chlorophyll fluorescence parameters of F_v’/F_m’ (maximum PSII efficiency in the light), ΦPSII (quantum efficiency of PSII electron transport in the light) and q_p (photochemical quenching) were measured at the same time using a multiphase flash fluorometer (LiCor Li6800, LiCor, USA) and a dark pulse (Murchie and Lawson, 2013). Data was logged once A, g_s, CO₂ sample and H₂O sample were stable, which took between 5-10 minutes per leaf. Two leaves from each plant, referred to as ‘measurement leaf 1’ and ‘measurement leaf 2’ were focused on each year. Measurements started once the leaves were fully expanded with measurement leaf 1 used from 77-133 DAS in 2018 (includes the first drought from 65-96 DAS) and in 2019 this leaf was used from 69-110 DAS (includes the drought at 73-92 DAS). Measurement leaf 2 was used from 105-210 DAS in 2018, (second drought 151-200 DAS) and from 118-182 DAS in 2019 (second drought 129-148 DAS).

In 2019 measurements were also taken of ‘measurement leaf 3’ at 162-182 DAS to correlate with Δ¹³C, which will be outlined later. Measurements were taken between 8:00 hr and 14:00 hr over 2 consecutive days with blocks one and two measured on the first day and blocks three and four on the second. The DAS of the first day of measurements is used to denote the timing of the measurement. When gas exchange measurements were completed, canopy temperature was assessed using thermal images of each box taken at a distance of 1 metre perpendicular to the edge of the box with a handheld camera (FLIR C2 thermal imaging camera, FLIR, USA) and analysed using thermal analysis and reporting software (FLIR Tools, FLIR, USA).

In 2019, chlorophyll extraction was undertaken on leaf discs collected at 203 DAS from plants one, two and three ( Supplementary Figure S2 ). Each leaf disc was wrapped in aluminium foil and placed in liquid nitrogen before storage in a -80°C freezer. At the same time three additional leaf discs were cut from each leaf and weighed together to determine FW and oven dried to record DW.

The frozen leaf discs were added to 2 ml microcentrifuge tubes with 1.5 ml 80% acetone and a ceramic bead before milling using a fast prep for two 20 sec cycles. The extracted chlorophyll was then transferred to 15 ml centrifuge tubes and topped up to 4 ml using 3 ml 80% acetone before centrifuging for 5 min at 3000 rpm. Chlorophyll a and b concentrations were then measured using a spectrophotometer (Cary 50, Agilent, Santa Clara, US) using the method of Porra (2002).

Relative water content (RWC) was measured at regular intervals from 66-213 DAS in 2018 and 74-177 DAS in 2019. Using a cork borer three 1 cm diameter leaf discs were cut from a leaf on plants 2 and 3 representative of the new measurement leaf on plants 6 and 7 ( Supplementary Figure S2 ). The 6 leaf discs were processed using the method of Weatherley (1950) and RWC calculated using the following equation:

Stomatal impressions of the abaxial and adaxial leaf surface of leaves of a similar age were taken at 219 DAS and 203 DAS in 2018 and 2019 respectively. Clear nail varnish was applied and left to dry for 20 minutes until no longer tacky, lifted using clear tape and mounted on a microscope sample slide. Three images were taken from each sample slide using a microscope (Leica 5000B, Wetzlar, Germany) with a light source (Leica CTR5000 Wetzlar, Germany) at 100x magnification and cropped to 1 mm² using the microscope scale for reference in Fiji (Schindelin et al., 2012). The stomata in the cropped images were manually counted in Fiji using the Cell Counter plug in (Author: Kurt De Vos), with an average stomatal density (SD) for each sample calculated from the three 1 mm² areas.

Boxes were harvested at 226 DAS and 211 DAS in 2018 and 2019 respectively. The sugar beet were hand lifted with plants 6 and 7 taken for further analysis. The 10 remaining beet were topped and the leaves and roots weighed separately to determine FW. The canopy was then discarded, and the roots taken to the BBRO tare house at Wissington Sugar Beet factory, Norfolk, UK to determine sugar % using polarimetry. K and Na impurities were also measured using flame photometry and AN content using colorimetry as per standard methods (ICUMSA, 2022). The leaves and roots of plants 6 and 7 were combined and weighed to determine FW before oven drying at 70°C and weighed to determine leaf and root dry matter (DM). The %DM of leaves and roots from plants 6 and 7 in each box was used to calculate the total DM from the total FW. The white sugar yield (WS) was calculated by multiplying the total FW by the sugar percentage. The total DM and WS for each box and the total amount of irrigation applied was then used to calculate the box level total dry matter water use efficiency (WUE_DM) and WS water use efficiency (WUE_WS).

Specific leaf weight (SLW) was calculated from measurement leaf 3, before processing to determine Δ¹³C, with the leaf passed through a leaf area meter (Li-3100, LiCor, USA) to determine leaf area with DW calculated from the FW multiplied by the %DW derived from beet 6 and 7 at harvest.

In 2019, measurement leaf 2 and 3 were removed at 209 DAS and freeze dried to determine the ration of ¹²C to ¹³C (δ13C). Samples were milled (ZM200, RETSCH, Haan, Germany) to a fine homogenised powder and analysed at the British Geological Survey in Keyworth, Nottinghamshire. Leaf δ¹³C analyses were performed by combustion in a Costech Elemental Analyser (EA) on-line to a VG TripleTrap and Optima dual-inlet mass spectrometer, with δ13C values calculated to the VPDB scale using a within-run laboratory standards calibrated against NBS 18, NBS 19 and NBS 22. Replicate analysis of well-mixed samples indicated a precision of ±<0.1‰ (1 SD). δ¹³C was used to calculate carbon isotope discrimination (Δ¹³C), which is inversely proportional to WUE.

Where δ_p is the δ13C calculated from the leaf tissue and δ_a is the atmospheric ratio of ¹²C to ¹³C taken to be -8‰ Δ¹³C (Farquhar et al., 1989).

Δ¹³C was plotted against average WUE_i on a per leaf basis, (calculated from the gas exchange values taken from measurement leaf 2 and 3), and WUE_DM, WUE_WS and SLW using averages calculated on a per box basis.

Repeated measures ANOVA for a split plot design was conducted on the A_max, g_s, C_i/C_a, WUE_i, F_v/’F_m’, ΦPSII, q_p, canopy temperature and RWC data. There were four blocks and whole plots dictated by the irrigation regime with variety and irrigation as factors. For the total leaf chlorophyll content, leaf water content, stomatal density, WS yield, Total DM, WUE_WS, WUE_DM, SLW and Δ¹³C data a general split plot ANOVA with the same factors as the repeated measures ANOVA was used. All analysis was undertaken in GenStat 19^th edition (VSN International Ltd., Hemel Hempstead, United Kingdom).

No leaf gas exchange interactions were observed between water availability x variety in the spot measurements, so the data was averaged across the two sugar beet varieties to examine the effect of water availability on WUE. Differences in some of the gas exchange parameters were observed in 2019 between the varieties and in Δ¹³C, which gives a longer-term measure of WUE_i, and are presented later. A_max and g_s were significantly reduced in all treatments over time as leaves aged ( Figure 2 ). Drought reduced A_max and g_s compared to the fully irrigated plants 92 DAS, 23 days after water withdrawal in 2018 (P<0.001; Figures 2A, B ), in 2019 A_max was reduced 81 DAS, only 9 days after water withdrawal (P<0.001; Figure 2C ) and g_s was reduced, but not significantly, compared to the fully irrigated ( Figure 2D ). In 2018 the slight reduction in A_max and g_s in the DD treatment compared to the fully irrigated at 196 DAS ( Figures 2A, B ) may be attributed to the drought being later in the year when the lower temperature reduced transpiration, as shown by the lower thermal time during the second drought in 2018 compared to 2019 ( Supplementary Figure S1 ). During the second drought in 2019, A_max and g_s were reduced in the DD treatment (P<0.001; Figures 2C, D ) at 140 DAS, 11 days after water withdrawal. At 140 and 146 DAS the SD treatment had a higher A_max and g_s than the fully irrigated (P<0.001; Figures 2C, D ). During this period air temperatures exceeded 40°C from 136 – 140 DAS and the fully irrigated treatment saw a decline in VWC to levels similar to the water limited treatment before additional irrigation was applied at 141 and 142 DAS with this difference no longer evident at 153 DAS.

Recovery from the first drought, when A_max was no longer significantly different to the control, in 2018 was at 133 DAS, 37 days after rewatering ( Figure 2A ), and in 2019 was at 109 DAS, 18 days after rewatering ( Figure 2C ). Due to the lack of significant decline in A and g_s under the second drought in 2018, recovery was only measured in 2019 and this was evidenced at 174 DAS, 46 days after rewatering. The close relationship between A_max and g_s means that g_s showed the same trends in recovery as A_max. The decline and recovery of A_max is further supported by the changes in chlorophyll fluorescence parameters ( Supplementary Figures S4A–D ) of maximum photochemical efficiency in the light, F_v/’F_m’ ( Figures S4A, D ), PSII operating efficiency, ΦPSII ( Figures S4B, E ), and photochemical quenching, q_p ( Figures S4C, F ), which are largely consistent with those expected from a change in A_max as a result of stomatal closure.

The continually water limited treatment had a reduced A_max and g_s comparable to the droughted treatments during water withdrawal, with the slower reduction in VWC in 2018 ( Figures 2A, B ) meaning the decline in A_max and g_s was not as rapid as it was in 2019 ( Figures 2C, D ). Once VWC was maintained at approximately 50% field capacity A_max and g_s were lower in the water limited treatment compared to the fully irrigated throughout the measurement periods in both years (P<0.001; Figure 2 ), except from 183 DAS onwards in 2018 ( Figures 2A, B ), and g_s at 133 DAS as the fully irrigated leaf g_s had also declined ( Figure 2B ).

WUE_i was greater in the drought treatments of measurement leaf 1 in 2018 with all treatments having a higher WUE_i than the fully irrigated until 133 DAS (P=0.004; Figure 3A ), which could be attributed to a lower relative decline of A_max and g_s in comparison as leaves age. The C_i/C_a values show a significant reduction compared the fully irrigated until 133 DAS (P=0.006; Figure 3B ) as stomatal aperture is reduced to conserve water which is also driving the increased WUE_i. In 2019, the difference in WUE_i between treatments was not consistent and from 97-110 DAS the Ltd treatment had a lower WUE_i than the fully irrigated, the only example of a water deficit treatment having a lower WUE_i than the fully irrigated (P=0.003; Figure 3A ). This was also reflected in the C_i/C_a values ( Figure 3B ). In 2018, the Ltd treatment had a higher WUE_i at 105 and 113 DAS in measurement leaf 2 where it covers the first drought (P=0.004; Figure 3A ). The C_i/C_a also reflects this with the Ltd treatment being lower than the SD and DD treatment which were themselves lower than the fully irrigated (P=0.004; Figure 3B ). At 196 DAS a higher WUE_i was also evident in the Ltd treatment (P=0.004; Figure 3A ) alongside a reduced C_i/C_a (P=0.004; Figure 3B ) compared to the other treatments. The DD treatment showed no increase in WUE_i ( Figure 3A ) but A_max and g_s were not significantly reduced as previously outlined. In 2019, the second drought increased the WUE_i of the DD treatment at 153 and 163 DAS with this difference no longer significant at 169 DAS (P=0.022; Figure 3C ), just before recovery of A_max and g_s at 174 DAS. This also resulted in a decline in C_i/C_a compared to the fully irrigated and SD to levels similar to the Ltd treatment (P=0.027; Figure 3D ). The increase in WUE_i (P=0.022; Figure 3C ) and decrease in C_i/C_a (P=0.027; Figure 3D ) in the fully irrigated compared to the SD at 140 and 146 DAS can be attributed to the decline in VWC previously outlined.

Averaged over all treatments at all measurement points there were no consistent differences in A_max, g_s, WUE_i and C_i/C_a between varieties in 2018. In 2019 there was no significant difference in A_max between the two sugar beet varieties, however g_s was significantly higher in the upright variety of both measurement leaf 1: 0.284 µmol m⁻² s⁻¹ compared to 0.196 µmol m⁻² s⁻¹ (P=0.012) and 2: 0.279 µmol m⁻² s⁻¹ compared to 0.197 µmol m⁻² s⁻¹ (P=0.042). The non-significant difference in A_max between varieties coupled with a significantly lower g_s resulted in a trend (P=0.072) of greater WUE_i in the prostrate variety in measurement leaf 1 of 77.2 µmol CO₂ mol^-1 H₂O compared to 70.4 µmol CO₂ mol^-1 H₂O (P=0.072) and was significantly greater for measurement leaf 2 at 86.0 µmol CO₂ mol^-1 H₂O compared to 75.8 µmol CO₂ mol^-1 H₂O (P=0.011). This greater WUE_i was also associated with a lower C_i/C_a for measurement leaf 2 of the prostrate variety of 0.655 compared to 0.681 (P=0.012) and a similar trend for measurement leaf 1 at 0.615 compared to 0.657 (P=0.069).

Canopy temperature can increase relative to air temperature as a result of reduced stomatal aperture and lowered transpiration. Looking at the effect of water availability across the two varieties during the first drought in 2018 and 2019 and the second drought in 2019 (not the second drought in 2018 due to the cool temperatures later in the season), an increase in absolute canopy temperature was evident ( Supplementary Figure S7 ).

In 2019 there was a variety x treatment x time interaction of canopy temperature. In the fully irrigated sugar beet the prostrate variety had a warmer canopy compared to the upright at 104, 111, 140, 147, 163, 174 and 182 DAS (P=0.003; Figure 4A ). In the SD ( Figure 4B ) and DD ( Figure 4C ) treatments there were no significant differences between varieties during the first drought but during the second drought at 140 DAS the prostrate variety had a significantly warmer canopy than the upright in the DD treatment (P=0.003; Figure 4C ), although both were above air temperature suggesting they had closed stomata. In the Ltd treatment the upright variety had a warmer canopy than the prostrate (P=0.003; Figure 4D ), at 104, 111, 140 and 163 DAS, opposite to the observations in the fully irrigated treatment.

RWC declined in the sugar beet under drought as water availability reduced. This was evident during the first drought in 2018 at 93 - 101 DAS with recovery by 108 DAS except in the SD which was lower than the fully irrigated but similar to the DD and Ltd treatments (P<0.001; Figure 5A ). In 2019, the first drought reduced RWC from 84 – 106 DAS with recovery by 121 DAS (P<0.001; Figure 5B ). The second drought had the same effect in 2019 reducing RWC from 141-148 DAS with recovery by 154 DAS (P<0.001; Figure 5A ). In 2018, the late drought did not significantly reduce RWC ( Figure 5B ). The Ltd treatment had a lower RWC compared to the fully irrigated once VWC had declined from 101-108 DAS in 2018 (P<0.001; Figure 5A ) and 84-170 DAS 2019 (P<0.001; Figure 5B ). Despite the decline in VWC in the fully irrigated at 140-146 DAS in 2019 and the concurrent reduction in A_max and g_s, RWC did not significantly decline.

The prostrate variety had a greater RWC than the upright variety when averaged across treatments in 2018 at 81.1% compared to 73.4% (P<0.001) and in 2019 84.8% compared to 75.5% (P<0.001).

The prostrate variety had greater total chlorophyll content on a per unit area basis in the Ltd treatment than all the other treatments, while in the upright variety, the fully irrigated and the DD had greater chlorophyll content than the SD and the Ltd (P=0.008; Figure 6A ). The upright variety had greater chlorophyll content than the prostrate in the fully irrigated, SD and DD treatments whilst there was no significant difference in the Ltd treatment (P=0.008; Figure 6A ). The ratio of chlorophyll a to b was not significantly different between irrigation treatments or variety. Leaf water content was also assessed to identify if this varied with chlorophyll content but the only difference was a greater water content in the prostrate variety compared to the upright in the full and Ltd, with the Ltd water content being significantly greater than that observed in any other treatment combination (P=0.016; Figure 6B ).

The prostrate variety had a significantly lower stomatal density on both the adaxial and abaxial leaf surface in 2018 and 2019 (P<0.001; Figure 7 ). There was no consistent relationship between stomatal density and irrgation in either year.

There were no varietal differences in the total dry matter or white sugar yields and the WUE_DM and WUE_WS so these observations were averaged across the two varieties to focus on the effect of water availability. The fully irrigated treatment had a greater total DM and WS yield in 2018 (P<0.001; Figure 8A ) and 2019 (P<0.001; Figure 8B ) than the other three treatments. In 2019, the Ltd treatment resulted in lower total DM and WS than the SD and DD (P<0.001; Figure 8B ). In 2019 the sugar beet achieved a higher total DM and WS than 2018, except in the Ltd treatment, highlighting the differences in the crop’s growth between years.

The WUE_DM and WUE_WS was higher in the Ltd treatment than the other three treatments and nearly double that of the fully irrigated and SD treatments in 2018 (P<0.001; Figure 8C ). The DD had a higher WUE_DM and WUE_WS compared with the fully irrigated and SD in 2018 (P<0.001; Figure 8C ), despite having a similar total DM and WS. In 2019, the WUE_DM of the Ltd treatment was higher than the other three treatments ( Figure 8D ; P<0.001). The WUE_DM of the fully irrigated, SD and DD were similar but the WUE_WS in the SD and DD was lower than the fully irrigated ( Figure 8D ; P<0.001). Overall, the extreme water deficit of the Ltd treatment increased WUE but the SD and DD treatments had inconsistent effects.

There were no differences in impurities (K, Na and AN) between the irrigation treatments in 2018. In 2019 impurities significantly increased in the SD, DD and Ltd treatments compared to the irrigated control ( Supplementary Figure S8 ). The SD and DD had similar levels of K whilst NA was significantly (P<0.001) higher in the DD treatment than the SD and Ltd treatments. AN levels increased with declining water availability. There were also differences in impurities between the varieties ( Supplementary Figure S9 ) with the upright variety having significantly higher K (P<0.001) and Na (P<0.001) than the prostrate whilst AN levels were similar.

Δ¹³C was negatively related to the average WUE_i (P<0.001; Figure 9A ), WUE_DM (P=0.001) ( Figure 9B ) and WUE_WS (P=0.004) ( Figure 9C ).

Water deficit reduced Δ¹³C indicating increased WUE_i. The Δ¹³C of measurement leaf 2, which was fully expanded at 118 DAS; before the second drought, was similar in the fully irrigated and SD treatments, with the DD lower, but not significantly ( Figure 10A ). The Ltd treatment had a lower Δ¹³C than the fully irrigated and SD but was not different to the DD treatment ( Figure 10A ; P=0.016).

Measurement leaf 3 was fully expanded at 162 DAS; after the second drought, with Δ¹³C generally higher than in measurement leaf 2 ( Figure 10A ). This can be attributed to younger leaves being more active with higher transpiration, as shown by the gas exchange measurements, resulting in reduced WUE_i. The Δ¹³C of measurement leaf 3 was lower in the DD than the fully irrigated and the difference was significant, unlike in measurement leaf 2, showing that water deficit was reducing Δ¹³C and hence increasing WUE_i at a greater magnitude in younger than older leaves ( Figure 10A ; P=0.016). The SD remained similar to the fully irrigated and the Ltd treatment Δ¹³C was lower than the fully irrigated and SD but not the DD ( Figure 10A ; P=0.016).

Averaged across the watering regimes, the prostrate variety had a lower Δ¹³C than the upright in both measurement leaf 2 (P=0.001) and measurement leaf 3 (P=0.012) ( Figure 10B ).

Δ¹³C was negatively related to SLW (P=0.001; Figure 11A ). There was no significant difference in SLW between the varieties but high levels of water deficit increased SLW with the DD having a greater SLW than the fully irrigated and SD (P=0.016; Figure 11B ). The SLW of the Ltd treatment was almost double the next nearest SLW value in the DD.

Increased WUE_i under water deficit, as evident in our results, has been shown previously in sugar beet (Bloch et al., 2006; Rinaldi and Vonella, 2006; Fitters et al., 2018). As the water deficit increased the stomata began to close to conserve water and g_s declined with the reduction in transpiration causing an increased canopy temperature (Baker et al., 2007). The reduction in stomatal aperture also reduced C_i/C_a which is associated with a greater WUE_i. This in turn is related to a lower Δ¹³C as the CO₂ in the sub stomatal cavity is not replenished as readily leading to greater a proportion of ¹³C being fixed, thereby lowering the ratio of ¹³C to ¹²C (Seibt et al., 2008). There were two exceptions to this observation. Firstly in 2019 the fully irrigated, SD and DD had similar WUE_DM but the fully irrigated had a higher WUE_WS. This suggests that water deficit did not alter the relationship between DM accumulation and water use but reduced the proportion of the DM partitioned to sugar in the root. It has previously been identified that under drought phloem loading can be reduced, lowering the amount of sugar in the root but not in the plant overall (Mäck and Hoffmann, 2006). Secondly WUE_i was reduced compared to the fully irrigated in the Ltd treatment of the first measurement leaf in 2019 whilst it was generally higher in all other observations. Across the same time period as the reduced WUE_i the Ltd treatment also showed a greater C_i/C_a ratio, whilst A_max and g_s were close to zero. An increase in C_i has been observed in extreme cases of water deficit previously and is driven by an increase in non-stomatal limitations to photosynthesis (Flexas and Medrano, 2002).

The measured gas exchange parameters recovered after drought, but only partially, due to the underlying leaf age related decline. Leaves respond differently to heat stress depending on age, with younger leaves showing responses of a greater magnitude compared to older leaves (Marias et al., 2017). This occurs because shading of older leaves drives reallocation of Nitrogen from Rubisco to light harvesting proteins reducing the maximum photosynthetic rate (Sims and Pearcy, 1991). However, when looking at the overlap of measurement leaf 1 and 2 from 105-133 DAS in 2018 it is evident that measurement leaf 1 reflected the trend of the response in measurement leaf 2 but not the magnitude. This means that, despite the overall decline in leaf activity with age, they still provided a reliable measure of the onset of drought and the subsequent recovery.

Both sugar beet varieties responded similarly to the first and second drought, suggesting the crop did not acclimate to avoid or better cope with water deficit. This has also been observed in glasshouse studies where sugar beet was exposed to three temporary water deficits but no acclimation was evident in photosynthetic or biochemical responses (Leufen et al., 2016; Ebmeyer and Hoffmann, 2022). This recovery is mediated by a range of physiological and biochemical mechanisms. How these interact to enable plant survival and recovery under drought stress is not clear and requires further study (Chaves et al., 2009). The maritime ancestry of sugar beet (Ribeiro et al., 2016) may be a driver of this with the plants showing drought tolerance (Dunham, 1993) and the photosynthetic apparatus being able to withstand severe water deficit and recover rapidly (Monti et al., 2006). This resilience is partly driven by the accumulation of K, Na and AN, as observed in 2019, to maintain osmotic potential for longer to ensure cell function. This means there is no need to avoid a decline in leaf RWC, as the plant can continue to photosynthesize until the most severe levels of drought. Even then no long-term damage occurs to the PSII, as shown by the recovery of the F_v’/F_m’, ΦPSII and q_p values which returned to levels similar to the fully irrigated once re watered. However, the recovery of maximum PSII efficiency in the light, and A_max, was not immediate and may have contributed to the reduced DM accumulation in the droughted treatments as evidenced in sugar beet by Bloch et al. (2006). It has been suggested that measuring photosynthetic induction coupled with A_max provides better data on photosynthesis and would have been a valuable approach in this study. This is because induction can provide a performance in the field under fluctuating light regimes whereas A_max is not always as sensitive to drought stress, especially during recovery (Sakoda et al., 2022).

The lack of acclimation from the first drought was also reflected in the decline in WUE_i and increase in C_i/C_a as the sugar beet opened stomata and began to reach similar levels of g_s and A_max to the fully irrigated treatment. The rapid recovery of leaf RWC, which recovered faster than PSII and leaf gas exchange, helps drive this recovery by ensuring the leaf has optimal conditions for photosynthesis (Lawlor, 2002), with the rapid recovery of RWC over daily cycles previously observed in sugar beet (Geiger et al., 1991). Not only was this recovery evident in these short-term measurements but also in the Δ¹³C results where the SD leaves showed no difference compared to the fully irrigated treatment and therefore no long-term change in WUE_i.

Water deficit has been shown to alter the stomatal density of leaves which develop under drought (Xu and Zhou, 2008; Sun et al., 2014). There were no consistent changes in stomatal density highlighting further that sugar beet physiology changes little under water deficit. There is an exception to this which is the reduction in plant biomass which appears to have resulted in the crop having a reduced demand for water, likely due to a reduction in canopy size. Canopy size was not measured but water deficit has been shown to reduce leaf area in sugar beet (Fitters et al., 2017) which results in reduced radiation interception and DM accumulation (Brown et al., 1987). This was evidenced by the reduction in VWC in the fully irrigated treatment at 140 DAS which the SD did not encounter despite receiving the same amount of irrigation. This seems to have been beneficial under a slight water deficit but did not alter the response to the severe second drought in 2019. In the late second drought in 2018, the slight decline in A_max and g_s was not reflected in any other parameter and shows how late season drought can be hard to detect as the forces driving transpiration are reduced.

The difference in Δ¹³C and associated traits between the two varieties supports the findings of Ober et al. (2004); Ober et al. (2005) that there is variation in traits associated with drought tolerance and water use in sugar beet, despite suggestions that sugar beet varieties lack diverse traits due to being derived from a single population (Davis, 2006). The varieties also differed in leaf physiology, but it is not clear if this can be attributed to the differences in canopy architecture and requires further study with more varieties.

The traits identified differing between the varieties suggest that the prostrate variety is more conservative in its use of water. The greater RWC is linked with more drought resistant phenotypes as a greater RWC can enable plants to function for longer under water deficit (Xu et al., 2000). Ober et al. (2005) did not find RWC to be linked to DTI but leaf succulence was. This suggests there is a relationship between the water content of leaves and drought tolerance, which this study has further shown to be linked to a greater WUE_i. RWC did not reduce the rate of carbon uptake and assimilation as has been observed in wheat (Farquhar et al., 1980). This is supported by the observation that RWC is not the driver for reduced Rubisco activity and that reduced CO₂ concentrations in the chloroplast due to reduced stomatal conductance have a greater regulatory effect (Flexas et al., 2006). This shows that stomata have an important role to play in plant carbon dynamics and therefore WUE.

The lower stomatal density in the prostrate variety may be associated with WUE_i but complex interactions between stomatal density and size and the speed of stomatal response means that the relationship between stomatal density and water use is debated. In potatoes a higher stomatal density, which developed under drought, led to a greater Δ¹³C and WUE_i (Sun et al., 2014) and it has been suggested the smaller stomata are faster to close and reduce transpiration and increase WUE_i (Drake et al., 2013; Barratt et al., 2021). However, in Arabidopsis a lower stomatal density was associated with reduced susceptibility to water deficit (Doheny-Adams et al., 2012), which could explain why the prostrate variety had a cooler canopy than the upright under extreme water deficit (Ltd treatment).

The difference in stomatal traits could be driven by the differences in canopy architecture. This is because the leaves of the upright variety are less shaded and intercept more light on the abaxial surface. This can result in a greater number of stomata on the abaxial surface of the upright variety to optimize CO₂ uptake for photosynthesis (Harrison et al., 2020). This is supported by a study that used the same sugar beet varieties which showed that the upright sugar beet variety has a greater RUE than the prostrate (Tillier et al., 2023). Despite the difference in RUE, and as observed in this study, there was no significant difference in yield. The difference in stomatal number and potential link to photosynthetic performance was not supported by the spot gas exchange measurements as both varieties performed similarly. A difference could have been expected between the older measurement leaves as greater shading in the prostrate could result in a quicker decline in photosynthetic performance as N is reallocated from Rubisco as previously outlined. However, the observation that both varieties have similar photosynthetic performance has been made before (Tillier et al., 2023) and suggests that the prostrate variety performs as well as the upright despite intercepting less radiation to achieve an equal sugar yield.

Büssis et al. (2006) showed that increased stomatal aperture can compensate fewer stomata to enable similar levels of A and g_s but that under high light intensities g_s was reduced compared to the plants with a higher stomatal density. This may explain the higher canopy temperature in the prostrate variety in fully irrigated conditions as lower g_s rate results in reduced cooling by transpiration. This reduction in g_s was also observed in 2019, leading to a greater WUE_i which highlights how different stomatal densities may be suited to different environments with a lower density making a plant more prone to heat stress under higher temperatures and irradiance (Lu et al., 1998) although such conditions are only present for short periods in much of the temperate beet growing area. This also highlights the importance of understanding the genotype x environment interaction of different traits, as environmental conditions can significantly affect how traits drive sugar beet yield (Hoffmann et al., 2009). In addition to this other factors that drive photosynthetic performance and linked to stomatal traits must also be considered such as stomatal size and leaf mesophyll conductance (Gago et al., 2014; Bertolino et al., 2019).

The varieties had a similar SLW despite observing that a higher SLW was associated with a lower Δ¹³C and greater WUE_DM, and this relationship did not vary depending on water deficit as had been previously observed (Rajabi et al., 2008). In this study SLW was correlated with Δ¹³C but this was driven by difference in SLW due to water deficit, with plants exposed to a greater water deficit having a lower Δ¹³C and WUE_DM. This suggests that the strong relationship between SLW, Δ¹³C and WUE_DM observed by may not always be evident on a varietal basis and other factors may be more strongly correlated with WUE_DM. Additionally, the observation of lower levels of K and Na in the prostrate variety suggest water regulation in the root was different to the upright which could be linked to the drought tolerant traits observed.

Overall, the prostrate variety was more efficient in its use of water and had a reduced stomatal density and greater RWC which are traits associated with increased WUE and drought tolerance. It is not known whether these traits are present in all sugar beet genotypes with prostate canopies, as only one was examined in this study, and this is an area for further research.

Leaf chlorophyll content was greater in the upright variety except in the Ltd treatment, where the prostrate variety had a similar chlorophyll content and a greater leaf water content suggesting that under severe and prolonged water deficit the leaf morphology had changed, which was not evident in the upright variety.

The increase in chlorophyll content was opposite to that typically observed under drought for many species such as maize (Din et al., 2011) and rice (Pirdashti et al., 2009) and there was no change in the ratio of chlorophyll a and b which is also sometimes observed (Saeidi and Zabihi-e-Mahmoodabad, 2009). However, an increase in chlorophyll content has previously been reported in sugar beet exposed to drought (Hussein et al., 2008) and increased SPAD values (a proxy for chlorophyll content) were observed in sugar beet exposed to high level water deficits by Fitters et al. (2018), although these differences were no longer present after re-watering and were attributed to a dilution effect. No such dilution effect is evident in this study as leaf water content was measured at the same time as collection of leaf discs for chlorophyll extraction. Additionally, the prostrate variety had a more stable chlorophyll content with an increase only evident in the Ltd treatment, whilst the upright variety was more variable, but this did not seem to affect the photosynthetic performance of the two varieties. In barley, varieties with a more stable chlorophyll content across varying moisture deficits were found to have greater drought tolerance (Li et al., 2006) but in wheat greater chlorophyll content was associated with greater drought tolerance (Talebi, 2011). The relationship between leaf chlorophyll content and WUE is therefore unclear in sugar beet and is an area for future research.