Water-use efficiency of irrigated biomass sorghum in a Mediterranean environment

A large interest is currently addressed to the no-food crops as an alternative source of energy. One of these crops is the biomass sorghum (Sorghum bicolor L. Moench) thanks to its high biomass productivity and high use efficiency of solar radiation and water. Aim of the research is assess the biomass sorghum response to the water in the Mediterranean environment. Biomass sorghum was subjected to four irrigation regimes, at 50, 75, 100 and 125% of ETc for three years (2008, 2009 and 2010). Water use efficiency (WUE), irrigation water use efficiency (IWUE) and water stress index (WSI) were calculated. Plant dry matter and green area index resulted different among the three years and the differences among irrigation treatments were more evident in 2009. The different soil water content at sowing among the three experimental years, affected the growth path during the growing crop cycle, explaining differences in term of biomass accumulation, leaf expansion and water consumption. WUE was higher in 2009 than in 2008 and 2010 with no differences among irrigation treatments for the first and third experimental year. WU ranged between 891 and 566 mm, the aboveground dry matter biomass between 4,097 and 1,825 g m and WUE between 8.49 and 4.00 kg m. IWUE, similarly to WUE, was higher in the second year than in the first and third year, but with differences among irrigation treatments in the 2008 and 2010. WUE calculated from WU normalized with VPD gave a more stable parameter in the three years. This research showed the suitability of biomass sorghum as energy crop in Mediterranean environment and its ability to use water efficiently. Additional key words: Sorghum bicolor; irrigation water use efficiency; green area index; biomass yield; water stress index; actual transpiration. * Corresponding author: michele.rinaldi@entecra.it Received: 01-03-13. Accepted: 11-09-13. This work has one Supplementary Table that does not appear in the printed article but that accompanies the paper online. Abbreviations used: ADM (Aboveground Dry Matter); CAW (Crop Available Water); ETc (Crop Evapotranspiration); ETm (Maximum Evapotranspiration); ET0 (Reference Evapotranspiration); GAI (Green Area Index); NPW (Not Productive Water); SWC (Soil Water Content); Tact (Actual Transpiration); Tp (Potential Transpiration); Tpi (Potential Transpiration at day i); VPD (Vapour Pressure Deficit); WSI (Water Stress Index); WU (Water Use); WUE (Water Use Efficiency); WUEvpd (Water Use Efficiency calculated with Water Use normalized with Vapour Pressure Deficit). Instituto Nacional de Investigación y Tecnología Agraria y Alimentaria (INIA) Spanish Journal of Agricultural Research 2013 11(4): 1153-1169 Available online at www.inia.es/sjar ISSN: 1695-971-X http://dx.doi.org/10.5424/sjar/2013114-4147 eISSN: 2171-9292

dition and water supply is the water use eff iciency (WUE; de Wit, 1958;Tanner & Sinclair, 1983) that is strictly related to biomass accumulation and water used. Therefore, WUE can be an indicator to assess the best water irrigation strategy of biomass sorghum in Mediterranean environments as an alternative energy crop.
A large number of researches have been carried out on grain or sweet sorghum in Mediterranean environments, but only few studies on biomass sorghum are reported (e.g. Habyarimana et al., 2004). This lack of information can be attributed to recent interest of biomass sorghum as a resource of bio-energy crop; thanks to cellulose, hemicellulose and lignin content in stems and leaves, biomass sorghum could represent an alternative renewable resource to fossil fuels (Cosentino et al., 2008).
The estimation of WUE for biomass sorghum is also important for obtaining a useful crop parameter, especially for the crop growth models that estimate biomass accumulation from water use efficiency, such as CropSyst (Stöckle et al., 2003), Parch (Hess et al., 1997) and the recent AquaCrop (Steduto et al., 2009) models. In fact, as reported by Hsiao (1993) and Hsiao & Bradford (1983), a correlation between above ground dry plant matter (ADM) and water use (WU) tends to remain linear (and so, simple to be applied) in both well-watered and water deficit conditions. Moreover, WUE seems to be influenced only by plant water status regardless of soil nutrient status (Stanhill, 1986).
Different authors report that WUE in sorghum is not a stable parameter since it changes among years, environments, phenological stages, soil water and nitrogen plant availability (4.1-6.0 kg m -3 , Mastrorilli et al., 1999;4.4-5.5 kg m -3 , Steduto & Albrizio, 2005;6.5-8.6 kg m -3 , Saeed & El-Nadi, 1998). This WUE variability underlines a limitation of applicability of a "fixed" value of WUE in sorghum in different climatic and environmental conditions, and so there is the need to find alternative approaches in order to make more flexible the use of WUE calculated from different years and locations. Two possible approaches in WUE estimation are the use of WU normalized (de Wit, 1958;Tanner & Sinclair, 1983) by evaporative demand of the atmosphere (ET 0 , mm; Steduto & Albrizio, 2005;Steduto et al., 2007) or by vapour pressure def icit (VPD, KPa, Stöckle et al., 2003).
One of the questions regarding WUE is that it does not provide constant indications on the effective use of water by the crop (transpiration) because it combines soil evaporation and crop transpiration in a single term. Moreover, WUE cannot be considered as an index of crop stress condition related to different water supplies. Water stress index (WSI; Idso et al., 1981), in fact, indicates the crop water availability level in relation to maximum evapotranspiration (ET m ). Furthermore, the gap between actual (T act ) and potential transpiration (T p ) gives the actual crop response to different water supply regimes, starting from reduction in canopy expansion to stomatal closure.
The aims of this work were to: i) determine the effects of four irrigation treatments on growth and yield of biomass sorghum, ii) assess several water use eff iciency indices at different scales, taking into account soil evaporation, potential and actual transpiration, in order to evaluate the effective water crop demand, use and efficiency, and iii) furnish parameterized values of these indices in a Mediterranean environment also useful for the most common crop simulation models.
The local climate is ''accentuated thermo-Mediterranean'' as classified by FAO-UNESCO (1963) Bioclimatic Maps, with daily temperatures below 0°C in the winter and above 40°C in the summer. Annual rainfall (average 550 mm) is mostly concentrated during the winter months, while only 101 mm of rainfall is recorded, on average, during sorghum crop cycle (1 st May-15 th August).
Biomass sorghum (cv. BIOMASS 133, Syngenta ® ) was sown on 9 th , 12 th and 4 th May in the three years, respectively, in rows 0.5 m apart and 0.08 m between seeds in each row (250,000 seeds ha -1 ). The crop was harvested before heading on 12 th , 20 th and 10 th August in 2008August in , 2009August in and 2010, respectively, at maximum dry matter accumulation, still rich in water and with simply glycosides composition (necessary for fermentative process in bio-ethanol production). The field experiments were carried out in a completely randomized block, setting four replications and elementary plots of 80 m 2 size, 16 rows per plot and 0.5 m apart. Water distribution was ensured by a drip irrigation system, with one line for each plant row and 4 L h -1 drippers and with one flow meter for each plot. As pre-sowing fertilization, 72 kg ha -1 of N and 87 kg ha -1 of P 2 O 5 as diammonium phosphate were supplied. Moldboard plow, disk arrow and rotary tiller were used to prepare the soil for the sowing, similarly to local farmer practices. Weeds were controlled by herbicides before sowing and by hand-hoeing during the first part of growing cycle. The health of the plants was ensured by fungicides and insecticides when required.

Irrigation and water use
Crop evapotranspiration (ET c , in mm) was measured in 2008 by means of two weighted lysimeters and crop coefficients (K c ) were estimated as ratio between ET c and the reference evapotranspiration (ET 0 , in mm), the latter was calculated using the FAO-Penman-Monteith model (Allen et al., 1998). K c derived from the first experimental year  were used in 2009 and 2010 to calculate ET c , as follows: Irrigation scheduling was set on the ET c basis, restoring the water used by the crop whenever the ET c reached 60 mm (subtracting rainfall), in order to compare four irrigation regimes: I_125 = 125% ET c , with each irrigation of 75 mm; I_100 = 100% ET c , with each irrigation of 60 mm; I_75 = 75% ET c , with each irrigation of 45 mm; and I_50 = 50% ET c , with each irrigation of 30 mm.

Growth analysis
Growth analysis was carried out at five sampling dates every two weeks from June to August: ADM was measured by taking a 0.5 linear meter sample from each plot and separated into stems, green and dead leaves. The plant material was dried at 80°C until the weight was constant. At harvest, the fresh biomass weight was determined on whole plot and the dry matter percentage on a 0.5 m linear meter sample.
To analyze the evolution of dry matter cumulated during crop growth cycle and compare the path of ADM among treatments and years, a sigmoid model (Vannella, 1998) was used: where ADM max is the maximal value of ADM, t the time expressed in days after sowing, t 0 represents the period between sowing and time to reach 50% of the final maximal value and b the f itting parameter of the model.
Green leaf area index (GAI) -with a destructive method-was determined using the Delta T Devices (Decagon Devices Inc., WA, USA) leaf area meter. Daily green area index (GAI i ) was obtained from the five values recoded at sampling dates according to Mailhol et al. (1997), as follows: where GAI max is the maximum GAI, t i is the time at day i, t e is the time at crop emergence, t m is the time at GAI max and α has a physical significance governing the GAI shape. GAI max , t m and a were the calculated values to fit the experimental data.
Daily potential transpiration (T pi ) was calculated starting from: where k (-0.7524) is the light extinction coefficient, calculated as the slope of regression line between the natural logarithm of diffuse non-intercepted sky radiation and GAI, both measured with a LI-COR 2000 portable area meter at sampling time, and K cmid is the K c measured at maximum canopy development. For each plot, the data derived from the average of six measurements carried out below the plant canopy during the middle of the day from 12:00 noon to 02:00 p.m., at each growing sample. GAI i is the green leaf area at day i and Cf is the clumping factor (Nilson, 1971;Lang, 1986Lang, , 1987, calculated with the following equation: Water-use efficiency of irrigated biomass sorghum in a Mediterranean environment where GAI i is the green leaf area index estimated with Eq. [3].

Irrigation and water use efficiencies
Gravimetric soil water measurements were carried out at 0.2, 0.4, 0.6, 0.8 and 1.2 m depths at sowing, harvest and growth analysis sampling dates, and soil moisture was expressed in volumetric content.
Seasonal water use (WU, in mm) was calculated according to the following simplified water balance equation: where ΔSWC is the variation, between seeding and harvest dates, of the volumetric soil water content in the 0-1.2 m depth layer, R is the rainfall and I the irrigations; all variable parameters are expressed in mm.
Usually, WUE and IWUE (kg m -3 ) are calculated applying the formula proposed by Tanner & Sinclair (1983), taking into account only the f inal value of ADM and the cumulated value of water used for irrigation or the water used by crops. In this work, WUE and IWUE were calculated as the slope of the linear regression between ADM (dependent variable) and WU (WUE) and between ADM and irrigation (IWUE). All the variables were measured at each sampling data (i).
The alternative approach to calculate WUE was with WU normalized by vapour pressure deficit (VPD, in kPa). For linear regression between ADM and irrigation, the intercept (b) was forced to zero, whereas in the regression between ADM and WU or WUvpd, the values of intercept on X axis (-b/a) provided an indication on water lost by soil evaporation (Passioura, 1977). [8] VPD (kPa; Murray, 1967) was calculated from daily maximum and minimum temperature and maximum and minimum relative humidity.

Water stress analysis
Plant efficiency to convert water in biomass was assessed with different indicators of water stress. One of these, the water stress index (WSI; Idso et al., 1981), was calculated as slope of linear regression, at intercept forced to zero, between cumulated maximum evapotranspiration (ET m ) and WU.
ET m in 2008 was measured by means of weighted lysimeters, whereas in 2009 and 2010 ET m was calculated by multiplying ET 0 by Kc max . This latter derived from K c estimated in 2008, but correcting K cmid with climatic conditions and plant height (Allen et al., 1998).
In particular: where K cb is K cmid corrected, u 2 is the wind speed (m 2 s -1 ) and h is the maximum plant height (m).
Since WSI considers also the water lost by soil evaporation, it does not involve the water effectively transpired by the crop. Therefore, a correct evaluation of water stress index could be done using the relationship between potential (T p ) and actual transpiration (T act ). T act was estimated with Eq. [6], but starting from GAI i greater than 3.0 m 2 m -2 , assuming that after this value, the soil is completely shaded by canopy and so evaporation is negligible (Ritchie, 1972).
At this point, the water stress index due to gap between cumulative T p and T act for each sampling date was calculated as slope of linear regression, with intercept forced to zero, as follows: From T p cumulated between two sampling times was derived daily T act (T acti ): [11] where D rel and f are parameters to fit cumulative T acti with T p . D rel can be considered as the fraction of total crop available water (CAW, mm) at which T p is reduced to T act through stomatal closure and f represents the effect of water depletion on stomatal closure; at higher values of f correspond low values of water stress. To assess accurately the D rel as a reference value for stomatal closure in biomass sorghum, it is necessary to relate D rel to CAW, the latter calculated as the sum of soil water content variation, rain and irrigation, taking as starting point the time when GAI is greater than 3.0 m 2 m -2 .
Analysis of variance of the data was carried out using a "randomized block" design model, and least signif icant difference (LSD) was used to compare mean values.

Climatic behaviours
In Suppl. Table 1 [pdf online] are reported the climatic data recorded during the years of experiment and the average values recorded at Foggia in a long term period . The maximum (T max ) and minimum (T min ) temperatures were different over the three years from the first part of growing cycle. May 2009 was characterized by T max and T min greater than those of 2008 and 2010, with T max characterized by values greater than 10°C compared with long term averages. However, 2010 was characterized by slightly lower T compared to 2008 and 2009 in the second part of growing cycle or from July to harvest time.
The same consideration can be made for daily global radiation (R g ), with greater differences found in May 2009 than in 2008 and 2010. In the first two weeks of June, R g was lower in 2008 than in the other two years, but globally, had no influence on crop growth (sowing dates: 9 th , 12 th and 4 th May; emergence dates: 20 th , 25 th and 13 th May, in 2008May, in , 2009May, in and 2010. The first and the third year were similar in terms of cumulated rainfall, 67 mm and 76 mm respectively, whereas in 2009, 92 mm were recorded. But a very large difference has been attributed to rainfall cumulated from 1 st January to the sowing date, equal to 168 mm, 418 mm and 255 mm for 2008, 2009 and 2010, respectively. Comparable averages were observed in the three years as regards daily reference evapotranspiration (ET 0 ), but these were slightly greater than long-term values. A detailed description of climatic behaviours is reported by .

Irrigation and water use
In Table 1, the number of irrigations, the amount of water applied, ΔSWC and the seasonal water use (WU) are reported. In the first and third year the greatest Water-use efficiency of irrigated biomass sorghum in a Mediterranean environment 1157 ) was irrigation (I), while in the second year was ΔSWC, the latter representing the water stored in the soil layers trough rainfall before sowing and subsequently used by crop during the growing cycle. This difference in water accumulated into soil during the winter and spring months could explain the increase in irrigation water supply ( The crop WU ranges between minimum and maximum of irrigation treatments and is greater in the first and third year than in the second one (Table 1).

Growth analysis
The main parameters (ADM max , t o and b) of sigmoid function (Eq. [2]) used to f it the function with the observed ADM data are reported in Table 2. The coeff icient of determination (R 2 ) was always high, especially in 2009 and 2010, but also in 2008 it was near to 0.90. The goodness of parameterization is shown by the curve of evolution of dry biomass (Fig. 2) where the fitted line is always close to mean of experimental data and its standard deviation. From emergence to maximum, ADM showed an exponential increase, even if some differences emerged among years. In fact, in 2008 the exponential phase is more pronounced, but within a shorter period and lower absolute values than 2009 and 2010. Moreover, ADM stopped earlier in 2008 than in the other two years. The final crop yield in terms of ADM was significantly different between I_125 and I_50. The exponential phase of crop growth in 2009 was more smoothed and delayed in time, reaching the maximum value of this phase at about 90 days after sowing; differences in ADM were observed already at 60 days and kept until harvest, with a clear separation between I_125 and I_100 compared to I_75 and I_50. The third year had an intermediate pattern between 2008 and 2009. Exponential crop growth phase stopped at about 80 days after sowing for I_125, I_100 and I_75, while I_50 showed a faster development but a lower dry matter accumulation up to the harvest. As shown in Values for GAI max , t m , a and R 2 are reported in Table 2. The lowest R 2 value was observed in 2009 but, globally, the sigmoid function curves were within the standard deviation in all the treatments in the three years (Fig. 2). As for ADM, in 2008 biomass sorghum had GAI max values lower than 2009 and 2010. The differences among irrigation treatments were noticeable after the second irrigation (Fig. 2) and kept until harvest, when I_125 and I_100 had similar GAI values and greater than I_75 and I_50. In 2009, the canopy expansion was very rapid especially compared to 2008 and 2010, but the canopy decline was fast as well. A high soil water content at sowing allowed I_125, I_100 and I_75 treatments to obtain similar GAI values regardless of different irrigation water supplies until 65 days after sowing; but after this point, I_75 showed a fast leaves senescence and at harvest was closer to I_50. The behaviour of GAI in 2010 was more diluted over time with the maximum GAI value reached later than in the previous two years. I_125 showed higher values than the other irrigation treatments during exponential canopy expansion (from 50 to 70 days after sowing).
More detailed results about GAI and ADM were reported by .

Irrigation and water use efficiencies
In the first year, reduction in water supply favoured an increment of IWUE, with the highest value in I_50 treatment (5.66 kg m -3 ), supplying 280 mm of water, followed by I_75, I_100 and I_125 treatments (Ta-ble 3). On the contrary, in the second year of experiment, no statistically significant differences between irrigation treatments were evident in IWUE, with an average value equal to 11.33 kg m -3 , more than double the average value recorded in 2008. In 2010, IWUE increased with decreasing irrigation water supply, making I_50 the treatment with the absolute highest value (12.42 kg m -3 ). Dercas & Liakatas (2007) reported that IWUE does not change with irrigation, and they found a value (4.45 kg m -3 ) closer to the first experimental year than the second and third one.
The slopes of regression lines between ADM and water used by the crop at each sampling (Table 4) correspond to the WUE (kg m -3 ). In the first and third year, an average value of 4.16 kg m -3 was statistically lower than WUE obtained in the second year (7.36 kg m -3 ). It was statistically similar among treatments in 2008 and 2009, while in 2010, I_125 and I_100 differed from I_75 and I_50. This large variability in WUE of sorghum as consequence of different water supplies is confirmed by different authors.
The not productive water, or the water lost by soil evaporation (Passioura, 1977), estimated as the ratio Water-use efficiency of irrigated biomass sorghum in a Mediterranean environment 1159 Moreover, some of not productive water might come from the saturation of vapour-pressure deficit (VPD), since the VPD decreases leaf conductance and photosynthesis, and also trough stomatal closure at high leaf water-potential (Bunce, 1985(Bunce, , 1988. As mentioned above, WUE in 2009 was about 75% greater than in 2008 and 2010. This gap was considerably reduced when the comparison was made with WUvpd (Table 5). In fact, despite 2009 showed the highest value (19.80 kg m -3 kPa -1 ) and 2008 and 2010 the lowest ones (14.36 and 12.99 kg m -3 kPa -1 , respectively), the gap was reduced to 27% in 2008 and 34% in 2010, compared with 2009.
We can observe that the not productive water was substantially similar among years (46, 76 and 64 mm), which disagrees with the observed results of WUE. This points out that normalization of WU with VPD takes into account the not productive water from canopy rather than from soil, since VPD between sub-stomatal cavity and outside air resulted in loss of water from leaf surface. Although differences between the three years diminished if we consider the WUEvpd rather than WUE, differences among treatments and years remained, underlining as other factors (for example, radiation interception and radiation use efficiency) linked to water use are involved in crop growth .

Water stress analysis
The linear regression (intercept forced to 0) between WU and ET m is reported in Fig. 3, where the slope coefficient (WSI) can be considered as an indicator of crop water status.
In 2008, WSI was equal to 1 in I_50 and I_75, with no evidence in water stress status, despite their water supply was reduced to 50% and 25%, respectively, compared with I_100. This is probably due to a great capacity of def icit irrigated sorghum to extract efficiently water from soil, especially in the deeper soil layers. Also in 2009, WSI was equal or higher than 1.0 in I_50 and I_75, and this shows as biomass sorghum is a crop with an elevated capacity to adapt itself to water stress conditions. On the contrary, observing the path in 2010, WSI was in agreement with the irrigation water supply.
An alternative method to estimate the water stress, which excluded the soil evaporative component, was assessed through the comparison between T p and T act . It can be considered the response of crop to reduction in water availability; in fact, the plant reduces leaves growth in order to adapt the transpiration process to soil water availability (T p ) and closes the stomata (T act ) in water stress condition. Daily T p for all years and all water (irrigation) treatments are reported in Fig. 4. Of course, the dynamic of T p is influenced by GAI, but differences in plant T p among treatments and years are highly reduced, especially at the maximum crop canopy expansion. As expected, T p in sorghum reached very high values (up to 6 mm), starting from 40 days after sowing to GAI max (up to 10 mm).
The regression lines between cumulative T act and T p are reported in Fig. 5, and the slopes represent WSI t . These values were similar in 2008 and 2010 for I_125 Water-use efficiency of irrigated biomass sorghum in a Mediterranean environment 1161 P. Garofalo and M. Rinaldi / Span J Agric Res (2013)    treatment and slightly higher in 2009. Similar WSI t was found in I_100 treatment in the 3 years, but 2010 showed more stressed plants for I_75 and I_50 water regimes compared to 2008 and 2009. From these results, it is evident that I_125 and I_100 treatments also suffered from water stress condition, probably due to the time elapsed between the irrigation events. WSI t is an indicator of the water stress magnitude: coupling it with the time when the stress occurs, further information can be obtained on plant drought resistance or when stomata begin to be closed.
In Table 6, the fitted values for Eq.
[11] are reported: they represent the fraction of CAW at which begins the gap between T p and T act (D rel ) and its inverse magnitude, f. From these values, T act was estimated as shown in Fig. 4. Unlike T p , a gap among treatments was observed for T act : I_125 and I_100 showed better performance than I_75 and I_50, especially in 2008 and 2009, while in 2010 a large superiority of I_125 was observed. D rel values increased with water supply in all the three years. The same behaviour for f value indicated a better adaptation to water stress in well watered regimes. Since CAW was different among years and irrigation treatments, to obtain a reference value of water availability threshold for a significant stomatal closure, CAW for each treatment was multiplied by D rel (Table 6). This threshold for plant water stress was similar for all treatments within each year, with a mean of 187, 234 and 253 mm for the f irst, second and third experimental year, respectively. These values indicate the minimum water supply (soil water content, rain and irrigation) that biomass sorghum needs not to reduce significantly actual transpiration (stomatal closure). Furthermore, this threshold represents a basal water requirement of sorghum and confirms as sorghum is a drought resistant crop also for a prolonged period of time.

Discussion
This research was conducted to assess the feasibility to introduce the biomass sorghum in Mediterranean environment as a renewable energy source, evaluating the productivity in terms of biomass produced and the capability to obtain the best water use eff iciency. Biomass produced and water used by crop cannot be evaluated separately and the parameter that relates these two factors should be stable and representative for the widest range of climatic management and soil conditions. Moreover, the knowledge of tolerance and/or the impact of water stress coupled to the soil water threshold at which sorghum suffers from water stress allow a more accurate irrigation management.

Irrigation and water use
Seasonal WU was different among years and these differences can be ascribed to the capability of sorghum to extract water from the deeper soil layers which were surely wetter in 2009 than in 2008 and 2010 because a lot of the rain fell before sowing date. WU in sorghum, as reported by other authors, varies with irrigation regime; Dercas & Liakatas (2007) in Greece indicate how water use in sweet sorghum passes from 662 mm to 397 mm with 512 and 175 mm of irrigation, respectively. Farrè & Faci (2006) observed a reduction equal to 314 mm in seasonal crop evapotranspiration, passing from 500 to 100 mm of water applied with irrigation in Spain.

Growth analysis
Sorghum ADM max attainable resulted influenced not only by irrigation treatment, but also by other factors. Indeed, although irrigation led to significant diffe-rences in term of ADM max with the highest values for the well irrigated regimes within years, among years the crop response did not result univocal, indicating a strong interaction between year and irrigation. Similar results were obtained by Farah et al. (1997) in grain sorghum, with values of ADM oscillating between 3,050 and 2,210 g m -2 , passing from 627 to 498 mm of water supplied in Sudan; lowest ADM was obtained by Farrè & Faci (2006) in Northern Spain, with values of 1,838 g m -2 for 588 mm of evapotranspiration and 522 g m -2 for 274 mm of water used.
In limited water supply conditions, typical of Mediterranean environment, biomass sorghum showed a similar or slightly better performance and stability for dry matter accumulation compared with sweet sorghum.

1164
P. Garofalo and M. Rinaldi / Span J Agric Res (2013) Curt et al. (1995), reported as in central Spain, in sweet sorghum cultivated in low watered regime, the ADM at harvest ranged between 1,200 and 2,300 g m -2 , whereas in our field experiments, the ADM max values (I_50) were between 1,800 and 2,299 g m -2 . In well irrigated conditions, the productive results were similar for biomass and sweet sorghum, about 3,200 g m -2 in Mastrorilli et al. (1995) vs 3,300 g m -2 in our experiment (I_125). A greater advantage in term of dry matter accumulation of biomass sorghum in drougth conditions, emerged strongly if compared with the data reported by Berenguer & Faci (2001) on grain sorghum. Indeed, in grain sorghum, at water consumption comparable with WU in I_50 treatment, the authors indicated values of aerial dry matter ranging between 1,026 and 1,249 g m -2 and so almost halved if compared to the results obtained for biomass sorghum in this research.
Reduction of GAI in sweet sorghum, as a consequence of reduction in water supply, is reported by Dercas & Liakatas (1999) who observed that by halving the water regime the peak of GAI was reduced by ~33%.

Irrigation and water use efficiencies
This interaction between year and irrigation could explain the differences in IWUE recorded for the three experimental years. Also in literature, the path of IWUE can change as consequence of irrigation management. Tolk & Howell (2003) showed IWUE decline at increasing irrigation, whereas Farrè & Faci (2006) gave an opposite indication on maize crop and reported a decrease in IWUE as consequence of increase of irrigation water supply going from 3.57 kg m -3 with 100 mm to 2.89 kg m -3 with 380 mm of irrigation water.
Other studies, however, reported contrasting results about sorghum. Amaducci et al. (2000) reported that fibre sorghum did not take advantage of irrigation in a well-watered environment of Northern Italy and this was also conf irmed by Monti et al. (2002), with positive but not significant relationships of irrigation on IWUE. From these authors and from this research, it appears clear as in sorghum the IWUE is a parameter that is influenced by some other conditions which make this parameter not very reliable in different agricultural conditions.
One of these conditions could be the soil water content at sowing time. In barley, this is a crucial point for roots lengths and density and consequently for ADM as reported by Sahnoune et al. (2004). In addition, early water status seems to influence the number of tillers in cereals (Baldy, 1986;Guedira et al., 1997;Volkmar, 1997). Considering the average value recorded during all crop cycles, the number of tillers per plant in 2009 was 1.54, greater than in the other two years (1.22, on average).
The difference in soil water content among the three years of experiment was due to rain fell before sowing date; IWUE does not take into account this "aspect" that we consider, instead, very important when we need to parameterize above ground biomass and the available water for crop. This limitation can be overcome using WUE. The first correlation between water used by crops and biomass was developed by de Wit (1958). Afterwards, several works are reported about the relationship between WU by the crops and the crop production parameters (Hanks, 1974;Stanhill, 1986;Monteith, 1993). As reported by Lindroth et al. (1994) and Beale et al. (1999), WUE, based on harvestable biomass and total annual evapotranspiration from the field, could be a useful tool to identify crops suitable for energy purpose. Mastrorilli et al. (1999), in a Mediterranean environment, reported values of WUE in sweet sorghum in well watered regimes, ranging between 5.6 and 4.1 kg m -3 despite small differences in water consumption (580 and 552 mm). In this research, at water stress condition for biomass sorghum (I_50), the water consumption was of 538 mm (three years average), comparable with the water use of sweet sorghum in optimal water supply (Mastrorilli et al., 1995) showing better water extraction from the soil and appreciable use efficiency to convert water in biomass (5.7 kg m -3 for I_50 vs 5.3 kg m -3 for sweet sorghum) in Mediterranean environment. Reduction in WUE as a consequence of a decline of water used is also reported in grain sorghum: Steduto & Albrizio (2005), in Southern Italy, found WUE equal to 5.7 kg m -3 with 510 mm of water supply, but this value decreased by 23% when the WU decreased by only 5%. Values of WUE observed in 2009 are close to those reported in forage sorghum by Saeed & El-Nadi (1998), in Sudan, with a variation from 8.6 to 6.9 kg m -3 , using a fixed level of water amount (700 mm), but varying the time between irrigation events and the relative amount of water. According to these assumptions, the values of WUE present in this research show sorghum as a valid energy crop in Mediterranean environment. In fact, considering an average value of WUE for the three years equal to 5.07 kg m -3 , it results comparable with other values reported for different energy crops such as Cynara (between 3.1 and 5.5 kg m -3 ) reported by Fernandez & Curt (1996), Spartina (between 5.1 and 8.2 kg m -3 ), and Miscanthus (between 7.8 and 9.5 kg m -3 ) observed by Beale & Long (1997).
WUE seems to indicate a conservative behaviour of this parameter on water productivity in sorghum with small fluctuations among treatments within year (Table 5). WUE is a more conservative indicator compared tothe IWUE within year, but it shows weakness if we use it among years.  that probably different climatic conditions do not allow considering this parameter as representative in crop water productivity. Many researches were carried out to identify which climatic variables could influence or drive plant transpiration and soil evaporation and to link the water productivity to climatic variables. In many of these studies, the vapour pressure gradient between leaf and air (Δe) is considered main engine in canopy transpiration (Norman, 1979). Tanner & Sinclair (1983) reported that "normalization" of transpiration by flux gradient (Δe, between leaf and air saturation vapour pressures) allows obtaining the best estimation of biomass as a function of water used. Since vapour pressures are temperature dependent, it is necessary to measure leaf temperature as much as above air temperature; if it is not possible, it is necessary to introduce simplif ication in gradient-flux calculation. This simplif ication can be obtained using evapotranspiration from water balance and VPD as reported in this paper. Paw & Gao (1988) and Asseng & Hsiao (2000) showed some weakness when VPD is used in substitution of leaf-to-air water vapour pressure difference (Δe), when canopy temperature is cooler or hotter than air temperature that causes a value of Δe substantially lower or higher, respectively. In crops such as sorghum, in which LAI overcomes quickly 3.0 m 2 m -2 , the shaded leaves are the most representative portion of the canopy, and thus, it is reasonable the assumption that all canopy leaves are at air temperature. WUEvpd was more eff icient than IWUE and WUE, and the response of dry matter accumulation as a result of the water availability among years indicates that WUEvpd is more suitable to different climatic conditions. An average value of WUEvpd equal to 15.9 kg m -3 kPa -1 let us to assess biomass sorghum as a high energy biomass crop, similar to or better than other crops such as Miscanthus for which Beale et al. (1999) reported values of WUEvpd equal to 10.7 kg m -3 kPa -1 .

Water stress analysis
WUE and WUEvpd provided a global judgement on sorghum water productivity. Reducing irrigation water supply, sorghum kept an appropriate canopy development among treatments without reducing the potential transpiration. But differences emerged when the actual versus potential transpiration was evaluated, with an average difference of only 23 mm in the most irrigated treatment and of 270 mm in the least irrigated one. The threshold of CAW to avoid physiological water stress was estimated about 225 mm; this threshold was similar for all treatments, although the sorghum in well watered regimes showed a better adaptation to water stress time, as highlighted by the higher value for f (adaptation to stomatal closure).

Conclusions
These results suggest that biomass sorghum has a high potential productivity (3-4,000 g m -2 ) of dry matter in Mediterranean environments if it is supplied with an adequate seasonal water amount, not less than 300 mm. However, sorghum showed a good adaptation to water stress; on average, it shows a reduction of potential transpiration only below 225 mm of CAW.
The suitability of biomass sorghum as bioenergy crop in Mediterranean environment is underlined by similar or higher values of WUE and ADM if compared with sweet and grain sorghum, when the water supply is reduced. Moreover, the WU comparable with other sorghum cultivars in water stress conditions, in a shorter growing cycle, indicates as the biomass sorghum has a better capacity to extract water from deeper soil layers, avoiding prolonged water stress condition.
For this reason, by exploiting crop characteristics, it is possible to schedule deficit irrigation, obtaining good biomass yield, but saving water resources.