Development and performance of wheat roots above shallow saline groundwater.
Extensive areas of irrigated land are affected by saline groundwater at shallow depth. Hoffman and Durnford (1999) reported how salinity and waterlogging have developed worldwide during recorded history, and the speed at which they are still advancing. Ghassemi et al. (1995) reviewed various estimates and concluded that, of some 230 Mha of irrigated land around the world, ~45Mha suffer from irrigation-induced salinisation. Conventional wisdom holds that the best solution to dealing with saline groundwater at shallow depth is to lower the water-table through provision of engineered drainage disposal systems, but the sustainability of such systems is open to question (Konukcu et al. 2006). Existing practice in irrigated areas of Australia results in excessive salt loads being discharged into receiving waters (Christen et al. 2001). Ayars et al. (2006a) reviewed drainage practice more widely in arid and semi-arid regions and identified a similar concern over poorly managed water and salt balances.
In view of increasing concern over water scarcity, there is another reason to re-evaluate the wisdom of the conventional drainage approach. This shallow saline groundwater should not be seen as a waste product, but rather as a part of the available water resource, accessible to the crop by capillary rise (i.e. sub-irrigation) and supplementing rainfall and/or conventional irrigation. However, proper management of irrigation requires understanding of direct water use by crops from the shallow groundwater. Ayars et al. (2006b) reviewed research on in situ water use by crops from shallow groundwater and concluded that the amount of water use for the conditions of the experiment was generally quantified but how to extend this information to other situations was rarely specified. Most of the accumulated experimental evidence was obtained with non-saline groundwater, and relatively few studies have considered the influence of salinity. The limited evidence (Ayars and Schoneman 2006; Ayars et al. 2006a) indicates that direct uptake from groundwater can sustain 30-40% of total crop water use with water-tables at up to 1.5 m depth and salinities up to 5 dS/m. This indicates that shallow saline groundwater is a resource with the potential to meet a significant proportion of crop demand provided that a sound deficit irrigation strategy is adopted.
Gowing et al. (2009) demonstrated that this was feasible by investigating the performance of a wheat crop in lysimeters under a combination of surface and sub-irrigation, each with 3 different levels of salinity. They noted, however, that the shallow saline groundwater prevented the roots from ramifying deeply because of the presence of a saline capillary fringe above the water-table. In this paper we focus on the development and performance of the root system, estimated in 2 ways (direct sampling and simple minirhizotrons), and how these are separately affected by the salinities of the groundwater and irrigation water.
We have not found any other major quantitative study of cereal roots above shallow saline groundwater over the whole growing season. Zuo et al. (2006) measured the distribution of root-length density (RLD) and water uptake of winter wheat under sub-irrigation with sweet water in a greenhouse, complementing the field study of Xue et al. (2003) under deficit irrigation. There are several detailed studies of root dynamics of rainfed temperate cereals in England (e.g. Welbank et al. 1974; Gregory et al. 1978; Barraclough and Leigh 1984; Ellis et al. 1984) and in Western Australia (Hamblin and Hamblin 1985). For techniques of measuring and describing roots, see the comprehensive monograph of Smit et al. (2000) or the authoritative review of Atkinson and Dawson (2001).
Design of the investigation
The investigation comprised 3 separate experiments with 9 lysimeters. In each experiment, 3 treatments were imposed by maintaining groundwater salinities of 2, 6, and 8 dS/m, each in triplicate. The 3 sequential experiments differed only in that the salinity of the irrigation water (applied at the soil surface) was successively 2, 4, and 1 dS/m. The lysimeters were in an unchanging controlled environment so that the experiment could be considered as a balanced 2-factor experimental design (3 groundwater salinities x 3 irrigation salinities x 3 replicates) with the irrigation salinity treatment repeated in time rather than replicated in space.
The experiments took place under isothermal conditions in a ventilated growth room at a temperature of 25 [+ or -] 2[degrees]C. The photoperiod was 12h with an average light intensity of 200 [micro]mol/[m.sup.2] x s. There was no humidity control, but the relative humidity in the room, measured by wet- and dry-bulb thermometers every 30 min, averaged 34%. These conditions led to high evaporative demand with average open-water evaporation of 5.1 [+ or -] 1.0 mm/day.
The lysimeters (Fig. 1) were 9 cylinders of polyvinyl chloride (PVC), 130 cm deep with walls 2 cm thick, the bases closed with PVC sheet of the same thickness. There were 3 cylinders per treatment, 2 with internal diameters of 55 cm and one of 35 cm. Each lysimeter contained, from the base upwards, 5 cm of gravel and 5 cm of sand (to allow unrestricted exchange with the supply of groundwater), then soil to 5 cm below the rim. Air-dry topsoil of the Rivington series (sandy loam: sand 76%, silt 11%, clay 13%) from Cockle Park, Northumberland, that had been steam-sterilised then passed through a 1-cm grid was packed as uniformly as possible in 15-cm layers to bulk density of 1.53 Mg/[m.sup.3]. The desorption soil-water characteristic for the Rivington soil is in Hassan et al. (1996).
Each column was instrumented with tensiometers connected to mercury manometers, to measure the matric potential of the soil solution and thus locate the plane of zero flux, at depths of 10, 20, 30, 45, 60, and 80 cm below the soil surface. The larger columns also contained insulated TDR probes (Trime-FM with P2G probes, 6 mm diameter and 16 cm long) at the same depths to monitor the salinity of the soil solution and to measure the soil-water content using calibrations previously developed for the Rivington soil over the range of salinities and water contents encountered in these investigations. The smaller columns contained a minirhizotron installed vertically in the centre. Each column was connected at 3 cm above the base to a Mariotte siphon, which maintained the water-table at a constant depth of 100 cm below the soil surface and also metered the solution that moved into the lysimeter to replace that evaporated from the surface or transpired by the crop.
For each experiment the lysimeters were initially saturated from below with saline water then covered and allowed to drain for 9 weeks until an equilibrium soil-water profile was attained; the overflow tap was then closed. The plane of zero flux (ZFP) (McGowan and Williams 1980) was then at the water-table. Each lysimeter was then sown with a salinity-tolerant variety of wheat (KRL 1-4), supplied by CAZS Natural Resources (University of Bangor, Wales), in rows 8cm apart at a seeding density of 250/[m.sup.2]. An initial surface irrigation of 10mm was applied after 2 days to aid germination and establishment. Each salinity combination received subsequent irrigations of 25, 50, and 25 mm at times which differed between treatments and experiments, applied as a controlled spray to the soil surface to avoid direct contact between the saline irrigation water and the leaves of the wheat plants. As time passed, the ZFP moved upwards from the water-table, and although water moved downwards from the surface after each irrigation, the ZFP never retreated to the water-table, so that the working of the Mariotte syphons was unaffected at all times.
[FIGURE 1 OMITTED]
Plant-based measurements were used to schedule irrigation applications; leaf water potential (LWP) was measured with a thermocouple psychrometer and stomatal conductance was measured with a steady-state porometer. A threshold LWP value of -1.5MPa was adopted and this was found to correspond to a stomatal conductance of 0.15cm/s. The plants received fertiliser based on soil analysis at the rate of 170kgN/ha as urea, 100kg [P.sub.2][O.sub.5]/ha as superphosphate, and 120 kg [K.sub.2]O/ha via the irrigation water.
The experiments were terminated after 112 days because the crop had reached the level of the lights in the growth room. Gowing et al. (2009) give details of the yields and above-ground plant attributes measured at this time. Typical crop duration, however, would be 120-130 days (Allen et al. 1998) and this may have restricted the period of grain filling and root development after anthesis. The total water use was given by the sum of the surface irrigation (110 mm), the groundwater use, measured as the outflow from the Mariotte siphons, and the profile-water use, estimated by the method of McGowan and Williams (1980) from sequential soil-water profiles measured daily by the TDR system in the larger lysimeters.
After each crop was harvested and the roots sampled, the lysimeters were repaired with fresh soil and then thoroughly leached with tap water to produce a non-saline profile before the preparation of the next experiment as described above.
The following strategy was adopted because of the small volumes of soil in our lysimeters. The minirhizotrons were in the smaller lysimeters. The progression of roots down the profile was estimated from exploratory auger cores extracted from soil between rows in one of the larger lysimeters (A) to avoid disturbing the growing crop; the soil was replaced afterwards. Root-length densities were measured on larger cores, extracted from the other larger lysimeter (B) after the above-ground vegetation had been cleared. Thus, direct comparison between estimates from minirhizotrons and adjacent soil cores was not possible; this was not ideal. There was, however, no effect of lysimeter diameter or the presence of the minirhizotrons on the amounts of water used by or the above-ground performance of the crop (Gowing et al. 2009). We therefore assumed, to a good first approximation, that the rooting behaviour of the crop was likewise unaffected.
Root depth was estimated on seven occasions by sampling the soil between rows in each larger lysimeter A with a 1-cm-diameter auger in 5-cm increments. After harvest, triplicate soil cores were extracted from the middle of the rows of each larger lysimeter B in successive depth increments with aluminium pipe, 7 cm in diameter and 5 cm deep. The soil was removed and placed in a plastic bag which was sealed immediately and stored at 4[degrees]C. The soil was later soaked overnight in a solution of tetra-sodium pyrophosphate (3 g/L) and then washed through a 0.5-mm mesh. The roots were collected and stored in sealed plastic bags at 4[degrees]C to keep fresh. Root length was measured by the grid-intercept method on a 1-cm grid, as described by Atkinson and Dawson (2001). Root length density (RLD) expressed on a volume basis, [L.sub.V] cm (root) per [cm.sup.3] (soil), was found from the total root length divided by the volume of soil from which the roots had been extracted. Total root length over the depth of the profile was expressed per unit of surface, [L.sub.A] km (root) per [m.sup.2].
A simple minirhizotron, closely based on the design of Gregory (1979), was installed vertically in the centre of each smaller lysimeter before they were filled with soil. Each minirhizotron comprised a perspex cylinder (6cm internal diameter and 100cm long) containing a steel rod (110 cm long) on which was mounted a light-source adjacent to a rotatable mirror (3 cm diameter) that could move vertically and reflect its image to a similar rotatable mirror 10 cm above the soil surface. This image was transmitted via a magnifying lens to a glass eyepiece with a 5-mm grid. Rooting depths were observed on seven occasions. Just before harvesting, the number of visible roots intersecting the eyepiece grid was recorded at 10 equally spaced orientations in each 5-cm depth interval. These were converted to RLD, [L.sub.V], by the analysis of Gregory (1979) and Atkinson and Dawson (2001) with an appropriate scaling factor to account for the difference in size between the field and the image. We did not, however, ascertain whether the roots were living or dead, active or inactive.
Results and discussion
Development of the root systems
The progression of the roots down the profile was well fitted by rectangular hyperbolae of the form of the Michaelis-Menten equation:
[z.sub.t] = [Z.sub.max]t/(t + k) (1)
in which [z.sub.t] (cm) is the observed root depth at time t (days), [Z.sub.max] (cm) is the hypothetical maximum root depth, and k (day) is a constant equal to the time at which [z.sub.t]=[z.sub.max]/2.
Roots penetrated deeper and faster the less saline the imposed treatment. Figure 2 shows measured values and fitted curves for the deepest and shallowest root systems, found in the least and most saline treatments, respectively. Table 1 contains fitted parameters from the Lineweaver-Burke linearisation. The Hofstee linearisation gave similar but not identical fits. This is an empirical observation because there is no theoretical reason why Eqn 1 should apply to the penetration of roots into the soil.
[FIGURE 2 OMITTED]
In all cases, [z.sub.max] was less than the depth of the water-table (100 cm) and decreased as salinity increased. The roots, however, never penetrated to [z.sub.max] because of (i) the physical barrier of the water-filled pores close to saturation in the capillary fringe above the water-table, and (ii) the decreasing water potential arising from increasing salinity as the profile dried. Table 2 lists the ultimate depths, [z.sub.f], reached; the ratio [z.sub.f]/[z.sub.max] increased as salinity increased. In contrast, parameter k decreased as salinity increased.
Rates of root growth are given by the slopes of the hyperbolae (Eqn 1, Fig. 2), which decrease continuously from maxima (=[z.sub.max]/k) at time zero. Thus, rates of downward extension fall continuously from initial values of 6-12 mm/day, which increase as salinity decreases. This contrasts with roots in soil or solution culture, which commonly grow at uniform rates for long periods if unconstrained (Hackett and Rose 1972). For example, Gregory et al. (1978) observed mean rates of extension of wheat roots of 6 mm/day in winter and 18 mm/day in spring, continuing for 7 months to an ultimate depth of 2 m.
At all times, root depths estimated by the minirhizotrons (y, cm) exceeded those estimated from the auger samples from lysimeters A (x, cm), although differences were small. The relation was y=2.14 + 1.039x, with r=0.986 and n=63. Likewise, fitted values of [z.sub.max] and k were generally larger for the minirhizotron than for the auger estimates, and the ratios [z.sub.f]/[z.sub.max] were generally smaller.
Table 2 summarises various attributes of the root systems as a series of 3 x 3 matrices, those from auger or core samples in plain type and those from the minirhizotrons in italic type within parentheses. All measured properties decreased as the salinities of both groundwater and irrigation water increased, mirroring observations on above-ground vegetation (Gowing et al. 2009). Extreme values of RLD for both methods of measurement and each depth differ by factors of 3.
All estimates of ultimate rooting depth from lysimeters A were shallower than those from the minirhizotrons. All measurements of [L.sub.V] in the top 10 cm of the profile in cores from lysimeters B exceeded those from the minirhizotrons, whereas the reverse was obtained at depths below 10 cm (Table 2). This observation is consistent with several reports in the literature (Ephrath et al. 1999; Atkinson and Dawson 2001). There were highly significant correlations between estimates of [L.sub.V] from minirhizotrons (y, cm/[cm.sup.3]) and those from cores (x) in each of the 4 depth intervals (Tables 2 and 3) but the relations weakened as depth increased. Overall these estimates were related by the regression equation y=0.116+0.752x with r=0.991 and n=36. This indicates that estimates from minirhizotrons exceed those from cores when [L.sub.V] is <0.47cm/[cm.sup.3], those from cores being larger above this value. This is probably, however, the result of proximity to the surface rather than a critical value of [L.sub.V] per se.
Total root length, [L.sub.A], expressed per unit area of surface, was greater for core samples when the groundwater salinity was 2dS/m but greater for the minirhizotron estimates at groundwater salinities of 6 and 8 dS/m (Table 2). The proportion of roots in the top 10 cm, a derived measure, was always greater for core (average 56.5 [+ or -] 4.8%) than for minirhizotron (45.1 [+ or -] 5.5%) estimates as a consequence of the systematic differences in [L.sub.V] obtained from the 2 methods. This proportion increased as the salinity of the irrigation water increased, but decreased as that of the groundwater increased, although differences are small for the core estimates (Table 2).
In the field, roots of winter wheat penetrate much deeper than the results in Table 2, e.g. to 2 m (Gregory et al. 1978), 1.8-2.0 m (Xue et al. 2003), 1.0-1.4 m (Hamblin and Hamblin 1985), >1 m (Barraclough and Leigh 1984), and 1 m (Welbank et al. 1974; Ellis et al. 1984). Likewise, RLDs in the surface 10 or 15 cm in the field greatly exceed those in Table 2, e.g. [L.sub.V] >8 cm/[cm.sup.3] (Welbank et al. 1974; Xue et al. 2003), >7cm/[cm.sup.3] in April falling to 2-3 cm/[cm.sup.3] in June (Ellis et al. 1984), and >6 cm/[cm.sup.3] from May to August (Gregory et al. 1978). Similar differences persist lower in the profile. The deeper penetration and larger values of [L.sub.V] combine to produce much longer overall root lengths in the field than those in Table 2 by up to an order of magnitude, e.g. values of [L.sub.A] 22-30 km/[m.sup.2] (Barraclough and Leigh 1984), 23.5 km/[m.sup.2] (Gregory et al. 1978), 19.88-23.1 km/ [m.sup.2] (Xue et al. 2003), 13.8 km/[m.sup.2] in drained and 8.5 km/[m.sup.2] in undrained clay soil (Ellis et al. 1984), 11.8 km/[m.sup.2] (Welbank et al. 1974), and 2.2-5.9 km/[m.sup.2] in sandy soils (Hamblin and Hamblin 1985).
Zuo et al. (2006) grew wheat at a high density of 400 seeds/ [m.sup.2] in PVC columns of 15 cm diameter in a greenhouse. By 56 days aider sowing, roots penetrated to 80-90 cm above sweet groundwater at 1.2m depth but only to 50 cm under surface irrigation. In compensation, however, [L.sub.V] exceeded 50 cm/[cm.sup.3] near the surface when irrigated but was only 5-10 cm/[cm.sup.3] above the groundwater.
Table 3 summarises average values of RLD for all 9 combinations of salinity. RLDs decrease consistently with depth, the decrease being more marked for the core than for the minirhizotron estimates. The standard deviations and coefficients of variability summarise the range of values in a given depth interval arising from the imposed salinities. The variability in the core estimates exceeds that of the minirhizotron estimates in each depth interval; coefficients of variability increase down the profile for the core estimates but remain roughly constant at 33% for the minirhizotron estimates. The ratio of RLDs between minirhizotron and core estimates increases from 0.84 to 1.45 down the profile, with an overall value of 1.03.
We fitted the distributions of root length with depth for the 9 individual salinity treatments and each method of measurement to several empirical models. That of Gerwitz and Page (1974) assumes an exponential decrease with depth:
P = 100[1 - exp(-fz)] (2)
where P is the percentage of the total root length contained at an average depth z, and f is a parameter such that half of the total root length is above (and below) a depth of 0.693/f. Fits were reasonably good (P<0.05) with average values of r=0.971 [+ or -] 0.015 (cores) and 0.967 [+ or -] 0.014 (minirhizotrons); the 50% depths were 11.2 [+ or -] 0.7cm (cores) and 14.9 [+ or -] 2.1 cm (minirhizotrons), compared to around 30 cm for winter wheat in the field (Gerwitz and Page 1974). There were, however, systematic deviations, with the model underestimating the proportion of length in the top 10 cm and overestimating that at 10-20cm; these 50% depths, therefore, exceeded those expected from the results in Table 2.
Other simple models take the form of a power law [L.sub.V] [varies] [z.sup.-n]. Monteith et al. (1989), working with sorghum and pearl millet, used n=0.5, and this fitted our data better than did the exponential model, with r = 0.990 [+ or -] 0.007 (cores) and 0.992 [+ or -] 0.004 (minirhizotrons). The best fit was found assuming n = 1, with r rising to 0.996 [+ or -] 0.005 (cores) and 0.994 [+ or -] 0.010 (minirhizotrons). A free fit, allowing n to take any value, gave variable results, with n=0.952 [+ or -] 0.080 and r=0.992 [+ or -] 0.007 (cores) and n = 0.659 [+ or -] 0.129 and r=0.990 [+ or -] 0.006 (minirhizotrons). For all these power-law representations P<0.01, and so we conclude that these are better alternatives to the exponential model for our dataset.
A preliminary analysis by GLM (generalised linear model) showed that there were statistically significant effects of the salinities of both groundwater and irrigation water, as well as of depth, on RLD. There was, however, no effect of any interaction between the salinities of the groundwater and irrigation water. Therefore, we separated the individual effects of the salinities of the groundwater and irrigation water by multiple linear regression of the data in Table 2. We set:
y = a + [bx.sub.1] + [cx.sub.2] (3)
where y is an attribute of the root system, [x.sub.i] is the imposed salinity of the groundwater (dS/m), [x.sub.2] is the salinity of the irrigation water (dS/m), and a, b, and c are regression coefficients. The results are in Table 4. There are 6 degrees of freedom, and the fit to Eqn 3 is better for each minirhizotron attribute than for the corresponding core attribute. The coefficient a is the value of a given attribute extrapolated to zero salinity and may be regarded as a hypothetical maximum value in non-saline soil from which losses due to salinity may be calculated. This modest extrapolation is considered acceptable, given the highly significant correlation coefficients (Table 4), in the absence of a confirmatory zero-salinity treatment in our investigation. The ratio c/b is the relative effect on an attribute of the salinity of the irrigation water to that of groundwater. Equation 3 is the simplest, and thus the most parsimonious, representation of the individual effects of the salinities of the groundwater and irrigation water.
For all measured attributes, except RLD in the top 10 cm, the decreases due to the salinity of the irrigation water exceed those due to groundwater of equal concentration, particularly in the minirhizotron data. The reverse is true for roots near the surface. The rate of root development is clearly dependent on the imposed salinities of the groundwater and soil-water, but salt accumulates near the surface as time passes because of convection accompanying capillary rise. This salt is displaced downwards when irrigation is applied, decreasing concentrations near the surface but increasing them lower down the profile. Thus, irrigation with water, even of quality equal to or better than that of the groundwater, as in 8 of our 9 experimental combinations, appears only to benefit the development of roots close to the surface but not sufficiently to enhance total root length. As a result, the proportion of total root length in the top 10 cm increases as the salinity of the irrigation water increases.
Table 5 summarises the water use by the crop as a series of 3 x 3 matrices. The use of both groundwater (GWU) and soil water (SWU) decreased as the salinities of groundwater and irrigation water increased. Extreme values in both cases differ by a factor of 3, as with those of RLD (Table 2). Total water use (TWU, Table 5), which includes 110 mm irrigation, likewise decreases. We correlated these components of water use to measured root attributes to ascertain which sources of water were preferred by different parts of the root system (Table 6). As with Table 4, all correlations were statistically significant (to better than P < 0.015) and were stronger with the minirhizotron than with core measures. For both methods of measurement, correlation coefficients decreased in the order TWU > GWU > SWU for the surface RLD, SWU > TWU > GWU for RLDs between 10 and 45 cm, and TWU>SWU>GWU for total root length and ultimate root depth; differences were, however, small. Overall, the best correlations were with ultimate root depth.
Total water use, which includes irrigation, is most strongly correlated with surface RLD and total root length. We could not, however, be certain that all 110 mm of the surface irrigation was used by the crop because we had no unambiguous way of distinguishing the use of the irrigation water once it has entered the soil from the use of the water originally stored in the profile. In fact, the soil in the top 10 cm of the profile dried to water contents of around 0.05 [m.sup.3]/[m.sup.3], corresponding to matric suctions exceeding 1.5 MPa (Hassan et al. 1996), and became increasingly saline during the intervals between successive irrigations. These then not only replenished the soil water in the upper zone but also displaced accumulated surface salt towards the water table, so providing more favourable conditions for continued root growth near the surface. In contrast, RLDs below 10 cm correlated best with soil-water use, presumably being less affected by the irrigation.
There are no unambiguous estimates of inflow to the roots, defined as the volume of water transpired per unit root length per unit time ([cm.sup.2]/day), because we could not separate transpiration from evaporation in our experiments. Table 5, however, contains the total water use per unit root length over the duration of the experiments ([cm.sup.2]) for the core results; those for the minirhizotron results follow a similar but less extreme pattern. This measure increases as groundwater salinity increases and, for groundwater salinity of 2 dS/m, as irrigation-water salinity increases. For more saline groundwater, it decreases when the salinity of the irrigation water exceeds 2 dS/m.
Relations between root length and yield
Barraclough and Leigh (1984) demonstrated a linear relation between grain yield and total root length at anthesis for winter wheat in England with high yields between 4.4 and 11.1 t/ha. In contrast, our grain yields, recorded in table 2 of Gowing et al. (2009), were small, ranging between 0.4 and 1.2t/ha. Nevertheless, we also found statistically significant linear relations between yields, Y (g/[m.sup.2]), and total root length at harvest, [L.sub.A] (km/[m.sup.2]):
Y = a + [bL.sub.A] (4)
Figure 3 illustrates these relations and Table 7 contains the regression parameters in Eqn 4 for both biomass and grain yields. Reworking the original data of Barraclough and Leigh gives a=0.78t/ha, b=30.0 g/km, r=0.75, n=21, P<0.001.
For each method of measurement, correlation coefficients decreased in the order dry biomass > grain > fresh biomass, and were larger for the minirhizotron than for the core estimates. Parameter a was smaller and b larger for the minirhizotron than the corresponding core estimate, a consequence of the more extreme range of [L.sub.A] values (Table 2) when estimated by the latter method. Our results were close to the extrapolated line of Barraclough and Leigh (1984); both datasets if extrapolated, however, imply a finite yield in the absence of roots (i.e. a > 0) which is absurd.
[FIGURE 3 OMITTED]
The slope of the regression line, parameter b, is the marginal yield per unit length of root. Our values for grain of 30 g/km (core) and 35 g/km (minirhizotron) correspond to that of Barraclough and Leigh (1984), though the agreement is probably fortuitous. These slopes, however, underestimate the true value because a > 0. Mean values of overall yield per unit root length for our 9 salinity combinations are 498 [+ or -] 42, 152 [+ or -] 9, and 48 [+ or -] 3 g/km for fresh biomass, dry biomass, and grain, respectively, for core estimates and somewhat smaller and less variable for the corresponding minirhizotron estimates (471 [+ or -] 32, 144 [+ or -] 6, and 45 [+ or -] 3 g/km). The high yields of Barraclough and Leigh (1984) give a mean value of 34 [+ or -] 2g/km. The data of Hamblin and Hamblin (1985) from Western Australia give smaller values between 13 and 36 g/km (grain) and between 31 and 96 g/km (dry biomass) depending on rainfall and fertiliser input, for low yields of 0.3-1.05 t/ha. The results of Xue et al. (2003) yield positive but non-significant associations with root length, with a > 0 and mean values of b of 20 g/km (grain) and 57 g/km (dry biomass); the increase in both yields with root mass is, however, significant (P< 0.05).
These results suggest that root length sets a limit to the quantity of resources that a crop can capture from the soil, which ultimately controls the yields that can be achieved. If nutrients are adequate, the limiting resource is water, as in this study, given the strong dependence of yield on total water use demonstrated in figure 2 and table 3 of Gowing et al. (2009).
Our finding that estimates of RLD from cores exceed those from minirhizotrons near the surface, but are smaller deeper in the soil, accords with the literature (Smit et al. 2000; Atkinson and Dawson 2001), despite the exigencies of our sampling strategy given that Welbank et al. (1974) measured larger values of both [L.sub.A] and dry mass beneath crop rows than beneath the spaces between rows. Our soil, however, was repacked sandy loam from which cores were freely extracted and roots easily separated without the damage that can occur in compacted clay soils in the field. Results from each method at all depths were closely related despite the minirhizotron tubes being vertical because of the limited size of our smaller lysimeters, although there is some evidence of better agreement from tubes angled at 45[degrees] (Atkinson and Dawson 2001). Ephrath et al. (1999), however, found no significant difference for wheat and acacia between the 2 insertion angles, though the scatter in their data was large. Our measurements were immediately before (minirhizotrons) and after (cores) harvest, but had time permitted, we could have monitored RLD throughout the season via the minirhizotrons. For winter wheat, RLD and total root length frequently rise to a maximum at anthesis and fall thereafter, despite roots continuing to grow downwards until harvest (Gregory et al. 1978; Ellis et al. 1984). Although our roots grew continuously (Fig. 2), we unfortunately have no information on the time course of [L.sub.V] and [L.sub.A] above shallow saline groundwater.
Our findings (Tables 2 and 4), that measured root attributes decrease continuously as the salinities of both sources of water increase, differ from expectations on the basis of the Maas-Hoffman salinity response function (Maas and Grattan 1999). In this, for a wide range of crops, yield remains constant up to a threshold salinity (6 dS/m for wheat), before falling linearly at higher salinities. The Maas-Hoffman function is based on the salinity of the saturation extract from the soil in the root-zone as an integrated value over the growing season. In contrast, Eqn 3 is based on the imposed salinities of groundwater and irrigation water and not on the subsequent variable salinity within the root-zone. In addition, our experiments with shallow saline groundwater in lysimeters in a controlled environment differ from field conditions under conventional irrigation, particularly in that the lysimeters were brought to equilibrium at the start of each experiment with soil water at the same salinity as groundwater. This represents a severe condition, and in practice salinity in the root-zone at the start of the season would be reduced as a result of leaching by rain or irrigation. The Maas-Hoffman relation strictly applies only to harvested yield, but given the close correspondence between yield and total root length, there is no reason why it should not apply to cereal roots. We did not, however, measure the dry mass of our root systems.
Formal mathematical treatment of the complex dynamics of water and salt in the root-zones of our lysimeters is beyond the scope of this paper. Consequently, much of our detailed analysis relies on statistical relations between root attributes and the imposed salinities of the groundwater and irrigation water or some consequence of those salinities. Thus, inferences from Eqn 3 and subsequent analyses are broadly similar because of the inter-correlations of salinity with water use and yield (Gowing et al. 2009). Overall, correlation coefficients are larger for regression equations involving minirhizotron estimates than with core estimates of root attributes, but this does not imply that the former technique is superior, although it is more convenient to use. It is clear that the primary effect of the salinity is to restrict the physiological performance of the wheat plants despite the apparent non-limiting supply of water accessible from the shallow water-table. However, the mechanisms causing this restriction, and the stages of growth at which they might occur, cannot be deduced from these experiments.
A saline water-table at 1 m restricted both the depth of rooting and the total root length of our crops compared to those reported for winter wheat where roots could ramify throughout the wetted profile to depths of more than 1 m. Ellis et al. (1984), however, noted that a temporary perched water-table 20 cm below the surface of a clay soil severely restricted both the distribution and total length of roots throughout winter, compared to adjacent drained soil, but the differences did not persist once the soil had dried in spring. Our growing period was short (112 days) and our grain yields small (Gowing et al. 2009) compared to winter wheat. These yields were, however, comparable to those from dryland and short-season irrigated crops (c. 1 t/ha) reported by Musick and Porter (1990) for which there is no information on root systems, although Hamblin and Hamblin (1985) noted low values of both [L.sub.A] and yield on sandy soils in areas of modest rainfall. There is evidence of a linear relation between yield and [L.sub.A], possibly because the latter determines the amount of water a plant can transpire. And, since total root length and other root attributes decrease as the salinities of both groundwater and irrigation water increase, the uptake of water and production of biomass likewise decrease.
We have analysed our results in some detail and compared them with data from field and lysimeter studies on non-saline soils. We have been unable to find comparable information from saline soils, especially above shallow groundwater. We acknowledge that there are lacunae in our study (e.g. changes in RLD and total root length over time) and in our discussions, particularly the mechanisms by which the combination of sources of saline water operate to adversely affect crop performance. As demand on global water resources inevitably rises, saline groundwater will increasingly be needed as a component in crop production in semi-arid parts of the world. Our results, which stand alone, may help in devising optimum strategies of water use. It is of particular interest to note that shallow roots indicate a need for frequent irrigation, but that the adoption of such a strategy would be likely to restrict uptake from the water-table and at the same time increase non-beneficial evaporation. We cannot propose an optimal strategy, but rather present our results for debate and in the hope that they will stimulate others to extend the investigation.
Shallow saline groundwater is a resource that can supply part of a crop's water requirement by sub-irrigation when supplemented by rainfall or by surface irrigation with water of better quality. Above such groundwater at 1 m depth, however, root systems of wheat were both shallow and sparse, whether measured on soil cores or estimated via minirhizotrons. Root attributes decreased as the salinities of both groundwater and irrigation water increased. Nevertheless, yields of biomass and grain per unit length of root were comparable to those of both dryland and high-yielding crops.
Manuscript received 15 September 2009, accepted 18 June 2010
Allen RG, Pereira LS, Raes D, Smith M (1998) 'Crop evapotranspiration: guidelines for computing crop water requirements.' FAO Irrigation and Drainage Paper 56. (FAO: Rome)
Atkinson D, Dawson [L.sub.A] (2001) Root growth: methods of measurement. In 'Soil and environmental analysis'. (Eds KA Smith, CE Mullins) pp. 435-497. (Marcel Dekker: New York)
Ayars JE, Christen EW, Hombuckle JW (2006a) Controlled drainage for improved water management in arid regions irrigated agriculture. Agricultural Water Management 86, 128-139. doi:10.1016/j.agwat. 2006.07.004
Ayars JE, Christen EW, Soppe RW, Meyer WS (2006b) The resource potential of in-situ shallow ground water use in irrigated agriculture: a review. Irrigation Science 24, 147-160. doi:10.1007/s00271-005-0003-y Ayars JE, Schoneman RA (2006) Irrigating field crops in the presence of saline groundwater. Irrigation and Drainage 55, 265-279. doi: 10.1002/ ird.258
Barraclough PB, Leigh RA (1984) The growth and activity of winter wheat roots in the field: the effect of sowing date and soil type on root growth of high-yielding crops. Journal of Agricultural Science, Cambridge 103, 59-74. doi:10.1017/S002185960004332X
Christen EW, Ayars JE, Hombuckle JW (2001) Sub-surface drainage design and management in irrigated areas of Australia. Irrigation Science 21, 35-43. doi: 10.1007/s002710100048
Ellis FB, Christian DG, Bragg PL, Henderson FKG, Prew RD, Cannell RQ (1984) A study of mole drainage with simplified cultivation for autumnsown crops on a clay soil. Journal of Agricultural Science, Cambridge 102, 583-594. doi:10.1017/S002185960004212X
Ephrath JE, Silberbush M, Berliner PR (1999) Calibration of minirhizotron readings against root length density data obtained from soil cores. Plant and Soil 209, 201-208. doi:10.1023/A:1004556100253
Gerwitz A, Page ER (1974) An empirical mathematical model to describe plant root systems. Journal of Applied Ecology 11, 773-781. doi: 10.2307/2402227
Ghassemi F, Jakeman A J, Nix HA (1995) 'Salinisation of land and water resources.' (University of New South Wales Press Ltd: Sydney)
Gowing JW, Rose DA, Ghamarnia H (2009) The effect of salinity on water productivity of wheat under deficit irrigation above shallow groundwater. Agricultural Water Management 96, 517-524. doi: 10.1016/j.agwat.2008.09.024
Gregory PJ (1979) A periscope method for observing root growth and distribution in field soil. Journal of Experimental Botany 30, 205-214. doi: 10.1093/jxb/30.1.205
Gregory PJ, McGowan M, Biscoe PV, Hunter B (1978) Water relations of winter wheat. 1. Growth of the root system. Journal of Agricultural Science, Cambridge 91, 91-102. doi:10.1017/S0021859600056653
Hackett C, Rose DA (1972) A model of the extension and branching of a seminal root of barley, and its use in studying relations between root dimensions. 1. The model. Australian Journal of Biological Sciences 25, 669-679.
Hamblin AP, Hamblin J (1985) Root characteristics of some temperate legume species and varieties on deep, free-draining Entisols. Australian Journal of Agricultural Research 36, 63-72. doi:10.1071/AR9850063
Hassan AA, Rose DA, O'Callaghan JR (1996) Model of the soil moisture characteristics. The Journal of Agricultural Science 29, 349-356.
Hoffman G J, Dumford DS (1999) Drainage design for salinity control. In 'Agricultural drainage'. Agronomy Monograph No. 38. (Eds RW Skaggs, J van Schilfgaarde) pp. 579-614. (American Society of Agronomy: Madison, WI)
Konukcu F, Gowing JW, Rose DA (2006) Dry drainage: a sustainable solution to waterlogging and salinity problems in irrigation areas? Agricultural Water Management 83, 1-12. doi:10.1016/j.agwat.2005. 09.003
Maas EV, Grattan SR (1999) Crop yields as affected by salinity. In 'Agricultural drainage'. Agronomy Monograph No. 38. (Eds RW Skaggs, J van Schilfgaarde) pp. 55-108. (American Society of Agronomy: Madison, WI)
McGowan M, Williams JB (1980) The water balance of an agricultural catchment. 1. Estimation of evaporation from soil-water records. Journal of Soil Science 31, 217-230. doi:10.111l/j.1365-2389.1980.tb02077.x
Monteith JL, Huda AKS, Midya D (1989) RESCAP: a resource capture model for sorghum and pearl millet. In 'Modeling the growth and development of sorghum and pearl millet'. ICRISAT Research Bulletin No. 12. (Eds SM Virmani, HLS Tandon, G Alagarswamy) pp. 30-34.
Musick JT, Porter KB (1990) Wheat. In 'Irrigation of agricultural crops'. Agronomy Monograph No. 30. (Eds BA Stewart, DR Nielsen) pp. 597-638. (American Society of Agronomy: Madison, WI)
Smit AL, Bengough AG, Engels C, van Noordwijk M, Pellerin S, van de Geijn SC (Eds) (2000) 'Root methods: a handbook.' (Springer: Berlin)
Welbank PJ, Gibb MJ, Taylor PJ, Williams ED (1974) Root growth of cereal crops. Rothamsted Experimental Station Report 1973(Part 2), 26-66.
Xue Q, Zhu Z, Musick JT, Stewart BA, Dusek DA (2003) Root growth and water uptake in winter wheat under deficit irrigation. Plant and Soil 257, 151-161. doi:10.1023/A:1026230527597
Zuo Q, Shi J, Li Y, Zhang R (2006) Root length density and water uptake distributions of winter wheat under sub-irrigation. Plant and Soil 285, 45-55. doi:10.1007/s11104-005-4827-2
D. A. Rose (A), H. M. Ghamarnia (A,B), and J. W. Gowing (A,C)
(A) School of Agriculture, Food and Rural Development, Newcastle University, Newcastle upon Tyne NE1 7RU, UK.
(B) Permanent address: Department of Irrigation and Water Resource Engineering, College of Agriculture, Razi University, PO Box 1158, Kermanshah, Iran.
(C) Corresponding author. Email: firstname.lastname@example.org
Table 1. Fitted parameters for root penetration (Lineweaver-Burke linearisation) Numbers under Treatment are the salinities of groundwater and irrigation water, respectively, in dS/m [z.sub.max] [Z.sub.p]/ Method Treatment (cm) k (day) [R.sup.2] [Z.sub.max] Minirhizotron 2.1 88.4 78.1 0.999 0.57 8.4 29.1 39.9 0.974 0.83 Cote 2.1 69.8 59.5 0.971 0.67 8.4 26.3 44.9 0.956 0.84 Table 2. Effect of salinity on root attributes RLD, Root-length density. Numbers in plain type refer to measurements on cores, those in italic type within parentheses to measurements with minirhizotrons Irrigation water salinity (dS/m): Groundwater salinity (dS/m) 1 2 4 Ultimate root depth (cm) 2 47 (50) 41 (45) 33 (35) 6 42 (45) 35 (40) 30 (31) 8 32 (35) 30 (33) 22 (24) RLD, [L.sub.v], 0-10cm (cm/[cm.sub.3]) 2 1.76(l.47) 1.61 (1.32) 1.30 (1.10) 6 0.98 (0.88) 0.86 (0.78) 0.75 (0.68) 8 0.88 (0.68) 0.70 (0.52) 0.62 (0.50) RLD, [L.sub.v], 10-20cm (cm/[cm.sub.3]) 2 0.60 (0.65) 0.47 (0.52) 0.27 (0.35) 6 0.49 (0.52) 0.34 (0.35) 0.25 (0.28) 8 0.35 (0.42) 0.21 (0.28) 0.18 (0.25) RLD, [L.sub.v], 20-30cm (cm/[cm.sub.3]) 2 0.40 (0.46) 0.30 (0.35) 0.18 (0.23) 6 0.22 (0.32) 0.17 (0.28) 0.14 (0.21) 8 0.19 (0.25) 0.15 (0.25) 0.13 (0.16) Total root length, [L.sub.A] (km/[m.sup.2]) 2 3.24 (3.17) 2.73 (2.67) 1.97(l.95) 6 1.99 (2.13) 1.57 (1.80) 1.29(l.41) 8 1.63 (1.65) 1.21 (1.37) 1.08 (1.12) % Root length in top l0 cm 2 51.8 (43.7) 59.1 (49.4) 65.8 (56.4) 6 49.2 (41.4) 55.0 (43.3) 58.1 (48.2) 8 54.0 (41.2) 57.9 (38.1) 57.4 (44.6) Table 3. Effect of depth on root-length density, Lv Values in parentheses are coefficients of variability Depth interval [L.sub.v] (cm/[cm.sup.3]) Ratio (cm) Core Minirhizotron 0-10 1.051 [+ or -] 0.410 0.881 [+ or -] 0.345 0.838 (39.0%) (32.9%) 10-20 0.35110. 142 0.402 [+ or -] 0.136 1.145 (40.5%) (33.3%) 20-30 0.209 [+ or -] 0.088 0.279 4 0.088 1.335 (42.2%) (31.7%) 30-45 0.163 10. 074 0.237 t 0.081 1.459 (45.5%) (34.3%) Total 1.856 [+ or -] 0.723 1.917 [+ or -] 0.657 1.033 (38.9%) (34.3%) Table 4. Multiple regression parameters relating salinities of water sources to root attributes AU individual regression parameters were statistically significant, except for one marked ([dagger]) Method Attribute a b Core Lv (0-10 cm), cm/[cm.sup.3] 2.049 0.1424 Lv (10-20cm), cm/[cm.sup.3] 0.702 -0.0317 Lv (20-30 cm), cm/[cm.sup.3] 0.425 -0.0237 Lv (30-45 cm), cm/[cm.sup.3] 0.347 0.0204 LA, km/[m.sup.2] 3.697 -0.2283 Ultimate root depth, zf, cm 54.1 -1.93 Roots in top 10 cm 52.9 -0.52 ([dagger]) Minirhizotron Lv (0-10 cm), cm/[cm.sup.3] 1.721 -0.1227 Lv (10-20cm), cm/[cm.sup.3] 0.743 -0.0315 Lv (20-30 cm), cm/[cm.sup.3] 0.501 -0.0208 Lv (30-45 cm), cm/[cm.sup.3] 0.437 -0.0186 LA, km/[m.sup.2] 3.622 -0.2030 Ultimate root depth, 58.5 -1.98 [z.sub.f], cm % Roots in top 10 cm 46.6 1.42 Method Attribute c [R.sup.2] Core Lv (0-10 cm), cm/[cm.sup.3] -0.102 0.951 Lv (10-20cm), cm/[cm.sup.3] -0.0781 0.876 Lv (20-30 cm), cm/[cm.sup.3] -0.0383 0.838 Lv (30-45 cm), cm/[cm.sup.3] -0.0321 0.852 LA, km/[m.sup.2] -0.2670 0.937 Ultimate root depth, zf, cm -3.93 0.923 Roots in top 10 cm 2.73 0.654 Minirhizotron Lv (0-10 cm), cm/[cm.sup.3] -0.0795 0.977 Lv (10-20cm), cm/[cm.sup.3] -0.0741 0.896 Lv (20-30 cm), cm/[cm.sup.3] -0.0476 0.897 Lv (30-45 cm), cm/[cm.sup.3] -0.0436 0.872 LA, km/[m.sup.2] -0.2665 0.957 Ultimate root depth, -4.48 0.938 [z.sub.f], cm % Roots in top 10 cm 2.62 0.872 Method Attribute P c/b Core Lv (0-10 cm), cm/[cm.sup.3] <0.001 0.72 Lv (10-20cm), cm/[cm.sup.3] <0.01 2.47 Lv (20-30 cm), cm/[cm.sup.3] <0.01 1.62 Lv (30-45 cm), cm/[cm.sup.3] <0.01 1.58 LA, km/[m.sup.2] <0.001 1.17 Ultimate root depth, zf, cm <0.001 2.03 Roots in top 10 cm <0.05 -5.20 Minirhizotron Lv (0-10 cm), cm/[cm.sup.3] <0.001 0.65 Lv (10-20cm), cm/[cm.sup.3] <0.01 2.35 Lv (20-30 cm), cm/[cm.sup.3] <0.01 2.29 Lv (30-45 cm), cm/[cm.sup.3] <0.01 2.35 LA, km/[m.sup.2] <0.001 1.31 Ultimate root depth, <0.001 2.54 [z.sub.f], cm % Roots in top 10 cm <0.01 -1.85 Table 5. Effect of salinity on water use Groundwater Irrigation water salinity salinity (dS/m) (ds/M) 1 2 4 Groundwater use (mm) 2 147 126 114 6 121 112 78 8 107 104 53 Soil-water use (mm) 2 81 71 38 6 63 56 26 8 45 45 24 Total water use, TWU (mm) (A) 2 338 308 262 6 294 278 214 8 263 259 187 TWU per unit mot length ([cm.sup.2]) 2 1.04 1.13 1.33 6 1.48 1.78 1.66 8 1.61 2.14 1.73 (A) Including 110 mm irrigation. Table 6. Correlation coefficients between root attributes and water use Method Attribute GWU SWU TWU Core [L.sub.V] (0-10 cm) 0.800 0.770 0.807 [L.sub.V] (10-20 cm) 0.837 0.924 0.895 [L.sub.V] (20-30 cm) 0.806 0.883 0.859 [L.sub.V] (30-45 cm) 0.811 0.884 0.862 LA 0.842 0.863 0.872 Ultimate depth, [z.sub.f] 0.924 0.935 0.952 Minirhizotron [L.sub.V] (0-10 cm) 0.798 0.769 0.805 [L.sub.V] (10-20 cm) 0.847 0.913 0.896 [L.sub.V] (20-30 cm) 0.894 0.959 0.944 [L.sub.V] (30-45 cm) 0.867 0.980 0.937 [L.sub.A] 0.876 0.904 0.909 Ultimate depth, [z.sub.f] 0.935 0.963 0.970 Table 7. Regression parameters relating yield to total root length at harvest a (g/ Method Attribute [m.sup.2]) b (g/km) r P Core Fresh biomass 326.4 304.6 0.791 <0.02 Dry biomass 110.7 86.2 0.887 <0.01 Grain 30.3 29.8 0.802 <0.01 Minirhizotron Fresh biomass 206.0 357.7 0.845 <0.01 Dry biomass 80.8 99.1 0.927 <0.001 Grain 18.7 34.8 0.853 <0.01
|Printer friendly Cite/link Email Feedback|
|Author:||Rose, D.A.; Ghamarnia, H.M.; Gowing, J.W.|
|Publication:||Australian Journal of Soil Research|
|Date:||Dec 1, 2010|
|Previous Article:||Retention capacity of biochar-amended New Zealand dairy farm soil for an estrogenic steroid hormone and its primary metabolite.|
|Next Article:||Accelerated testing of nutrient release rates from fertiliser granules.|