Research and Development
Department for Environment, Food and Rural Affairs CSG 15
Research and Development
Final Project Report
(Not to be used for LINK projects)
|Two hard copies of this form should be returned to: |
|Research Policy and International Division, Final Reports Unit |
|DEFRA, Area 301 |
|Cromwell House, Dean Stanley Street, London, SW1P 3JH. |
|An electronic version should be e-mailed to resreports@defra..uk |
| |
|Project title |Sensing the physical and nutritional status of the root growth environment |
| | |
|DEFRA project code |AR0913 | |
|Contractor organisation and location |Silsoe Research Institute |
| |West Park |
| |Silsoe, Bedford, MK45 4HS |
|Total DEFRA project costs |£ 298,878.00 | |
|Project start date |1/10/2001 | |Project end date |30/09/04 |
|Executive summary (maximum 2 sides A4) |
|To tab in this section press the tab key and the Control key together |
|Press the DOWN arrow once to move to the next question. |
The root environment has a major effect on crop growth, both directly through the supply of water and nutrients to the shoot, and indirectly through root-to-shoot signalling. Better management of crop root systems through agronomic and genetic means has the potential to improve the efficiency of water and nutrient uptake, and limit root restrictions to crop growth. However, progress in this area is currently limited by the lack of sensors for in situ estimates of soil water status, strength and nutritional status in the field. We describe the development of low-maintenance sensors to measure matric potential of soil that can be used by crop physiologists, agronomists and farmers.
We use published models of Mullins and co-workers to predict soil strength from soil water status. These models predict soil strength from effective stress (i.e. product of the degree of saturation and matric potential). They have the advantage that they use a relationship that does not depend on soil type. We confirmed this and develop the use of effective stress to predict the penetrometer resistance of soil, which is an indication of the resistance to root penetration.
Whilst the negative impact of soil strength on crop yields has long been recognised, the physiological mechanisms by which plants respond to sub-optimal soil conditions has only recently been elucidated. This phenomenon has been investigated in the field, but never in conjunction with the direct sensing of soil conditions as is described in this project report. Our laboratory and field data suggest that in a drying soil the yield of wheat may be reduced because of a combination of a hydraulic signal controlling stomatal conductance and a non-hydraulic signal that decreases growth rates in strong soil. An index of soil strength determined from measurements of matric potential and degree of saturation was highly correlated with the final yield. Thus the use of the matric potential sensors developed in this project is likely to improve our understanding of the crop response to root stress and provide a useful tool for farmers and consultants.
We attempted to use the electrical conductivity of the soil water to predict N availability. This was not successful, however, and the approach being developed at Rothamsted by Dr A.J. Miller was found to be superior.
The objectives of this project have been fulfilled through collaboration between a group of scientists at Silsoe Research Institute, National Soil Resources Institute and the Open University that have a unique mix of skills. We have also worked closely with Dr Miller at Rothamsted Research who holds a related Defra grant (AR0910) to develop nitrate-selective electrodes.
The objectives agreed for the project were:
1. To develop a sensor for matric potential of soil water which can operate between 0 and –200 kPa without the need for maintenance.
2. To test the hypothesis that the product of soil water matric potential and the degree of saturation can be used as an index of soil strength in subsoils commonly found in the UK arable sector.
3. To determine the envelope of soil conditions in which the electrical conductivity of the pore water gives reliable estimates of the rate of N mineralisation. Here we will determine the range of soil types with reference to clay content and soil conditions with respect to water content that electrical conductivity of the pore water can be used as an index of N mineralisation.
4. To relate the index of soil strength developed in objective 2 to root and shoot development in a drying soil in both laboratory and field conditions.
5. To review the possibilities for monitoring and recording data from a number of sensors in the field environment. The output of the project will be summarised in a document that is understandable by farmers and gives a critical evaluation of the achievements and problems so that a future strategy can be developed.
Objective 1: We developed a novel porous matrix sensor to measure matric potentials as low as -300 kPa. This is the first sensor to give reliable measurements of matric potential in this range. A full description of the sensor design and use is described in a paper which has been submitted to Plant and Soil. Calibration of this device demonstrated the importance of using a hysteretic model with closed scanning loops to avoid noise resulting in a substantial drift that gave the impression of matric potential increasing in a drying soil. Data comparing the outputs of the new porous-matrix sensor and a conventional water-filled tensiometer indicated a good agreement over the range −50 to −80 kPa. Further, these comparisons also demonstrated that in a drying soil water-filled tensiometers can give stable but false high readings of matric potential even though the soil is considerably drier. We are the first to report this problem to the plant and soil science community. As such this project has made a significant contribution to the current state of knowledge. These results show that the best strategy, for any scientific study of soil drying, is to complement direct measurements of matric potential with indirect measurements. The porous-matrix sensor therefore has considerable potential as a tool for understanding plant responses to drying soil.
Objective 2: We confirmed the hypothesis that effective stress, estimated from the product of the degree of saturation and matric potential, can be used to predict the tensile strength of soil with a relationship that does not depend on soil type. In repacked soils with a low density, effective stress could also be used to predict the penetrometer resistance of soil with a calibration that was insensitive to soil type. As soil density was increased, the calibration between effective stress and penetrometer resistance differed between soil types, although penetrometer resistance still increased with effective stress in all soils. Soil moisture status, estimated with the sensors developed in objective 1, provides an indicator of the increase in soil strength due to soil drying and in loose soils penetrometer resistance can estimated with a calibration that does not depend on soil type. We measured the resistance to penetration of a rotating penetrometer. This gives a good estimate of the resistance to penetration by roots. These measurements suggested that soil can be impenetrable by roots at matric potentials as high as -100 kPa for clay and -250 kPa for loamy sand.
Objective 3: Data collected in field experiments did not support the use of electrical conductivity measurements as a method for estimating N mineralization. However, comparison between this technology and ion-selective technology being developed at Rothamsted Research showed that the latter method was extremely promising.
Objective 4: We designed and conducted laboratory and field experiments to determine how in a drying soil, water stress and soil strength interact to limit crop growth. In the field, we conducted experiments in 2003 and in 2004. Soil strength was manipulated by 4 levels of compaction and by 3 different irrigation treatments in a fully factorial experiment randomized in 3 blocks. In the non-irrigated plots, soil drying was monitored with the new sensors developed in objective 1. Changes in soil water status due to irrigation or soil drying had greater effects on soil strength than compaction. The index of soil strength developed in objective 2 explains a high percentage (approx 70%) of the variance in final yield in both field experiments.
In laboratory experiments using controlled environments, the separate effects of soil strength and matric potential on the growth of wheat were investigated. We suggest that two signalling mechanisms control the response of wheat to drying soil: hydraulic signalling due to the lower hydraulic conductivity of the soil that affects stomatal conductance and non-hydraulic signalling of mechanical impedance that directly affects shoots growth, independently of stomatal conductance.
Objective 5: We have had discussions with manufacturers of sensors and low power telemetry systems and there is no technical barrier to the deployment of remotely-logged sensors in the field. The deployment of such system will depend on an understanding of the spatial variability of soil strength which is being developed by R.M. Lark at Rothamsted Research.
|Scientific report (maximum 20 sides A4) |
|To tab in this section press the tab key and the Control key together |
|Press the DOWN arrow once to move to the next question. |
General introduction
The root environment has a major effect on crop growth, both directly through the supply of water and nutrients to the shoot, and indirectly through root-to-shoot signalling (Hussain et al., 1999; 2000). Abscisic acid (ABA) plays a central role in communicating the stress from root to shoot, though other chemical messengers may also be involved. This phenomenon has been investigated in the field (Tardieu et al., 1992), but never in conjunction with the direct sensing of the matric potential of soil water. Whilst recognising the importance of soil moisture in determining soil strength, recent physiological investigations have tended to use soil treatments that vary only in their bulk density, the soil moisture content being kept constant by adding water several times a day (Hussain et al., 1999; 2000). To understand the effects of root-to-shoot signalling in the field, one needs to study the simultaneous responses of both soil strength and plant growth as soil moisture varies.
Better management of crop root systems through agronomic and genetic means has the potential to improve the efficiency of water and nutrient uptake, and limit root restrictions to crop growth. However, progress in this area is currently limited by the lack of sensors for in situ estimates of soil water status, strength and nutritional status in the field. These data are needed to understand and predict the physiological response of a crop to its growth environment. The main aim of this research was therefore to develop practical, low-maintenance sensors that can be used to monitor soil matric potential, soil strength and N mineralisation in the field environment. The project was designed to lead to sensors that will allow a better understanding of how the root environment affects crop growth in the field. Specifically, it was anticipated that it will be possible to obtain better predictions of crop growth and yield if the soil strength, water availability and the available N can be determined in situ with the sensors that we propose to develop and test.
Recent advances in methods for monitoring soil matric potential (Whalley et al., 2001) and for determining the electrical conductivity of pore water (Hilhorst, 2000) were exploited in this project. We were also able to use published models of Mullins and co-workers (e.g. Mullins et al., 1992a, b) to predict soil strength from soil moisture status as we were able to measure the matric potential of soil water. The development of these sensors is timely because they will provide a route for the results of fundamental plant science research on root-to-shoot signalling to be transferred to the field environment. In particular, we will test the hypothesis that root-sourced signals can determine the rate of shoot growth in the field.
Project objectives:
1. To develop a sensor for matric potential of soil water which can operate between 0 and –200 kPa without the need for maintenance.
2. To test the hypothesis that the product of soil water matric potential and the degree of saturation can be used as an index of soil strength in subsoils commonly found in the UK arable sector.
3. To determine the envelope of soil conditions in which the electrical conductivity of the pore water gives reliable estimates of the rate of N mineralisation. Here we will determine the range of soil types (ie clay content) and soil conditions (ie water content) that electrical conductivity of the pore water can be used as an index of N mineralisation.
4. To relate the index of soil strength developed in objective 2 to root and shoot growth in a drying soil in both laboratory and field conditions.
5. To review the possibilities for monitoring and recording data from a number of sensors in the field environment. The output of the project will be summarised in a document that is understandable by farmers and gives a critical evaluation of the achievements and problems so that a future strategy can be developed.
1. The construction and testing of a matric potential sensor
In previous EU-funded work at Silsoe Research Institute, we developed a porous-matrix sensor to measure the matric potential of water in relatively wet soil (Whalley et al., 2001). In this Defra-funded project, we planned to construct and test a similar sensor that will work in much drier conditions (a lower limit of –200 kPa or drier, instead of −60 kPa). The sensor works by measuring the water content of a porous material in equilibrium with the soil water and then using the known moisture characteristics of the porous material to convert that water content into matric potential. We identified appropriate ceramics with moisture characteristics that will allow them to drain and rewet between saturation and –200 kPa (Whalley et al., 2001), while the moisture content of the ceramic was to be measured with dielectric moisture probes supplied by Delta-T. The advantage of using the Delta-T dielectric probe is that it provides an analogue output, which is easy to record automatically.
Materials and methods
The porous-matrix sensor
The design of the porous matrix sensor used in this work is shown in the schematic diagram in Figure 1.1, while the key components are shown in Figure 1.2. The porous matrix is an extruded ceramic having a pore size distribution suitable to measure matric potentials in the range −50 kPa to −300 kPa (Fairey Industrial Ceramics, Filly Brooks, Stone, ST15 0PU, UK). The water retention characteristic ensures a large (i.e. easily measurable) change in degree of saturation over the range of matric potentials of interest. The ceramic had a cylindrical hole in its axis 4 mm in diameter and a further six holes 4 mm in diameter equally spaced on a circle 12 mm in diameter. The length of the ceramic was 100 mm because at lengths much shorter than this the method used to measure the water content of the ceramic was sensitive to the electrical conductivity of water. The ceramic was located inside a stainless steel cylinder that had 4 slots milled along its length to allow ceramic-soil contact. The 4 mm hole in the axis of the ceramic held a 4 mm stainless steel rod which was glued in place with electrically-conducting epoxy resin. The six holes spaced in a circle functioned to increase the speed of equilibration of the sensor with its environment.
The water content of the ceramic was measured with a modified dielectric probe, which used the circuit board of the Theta probe (Delta-T Devices, 128 Low Road, Burwell, Cambridge, CB5 0EJ, UK). This works by detecting changes in the impedance at 100 MHz at the end of a coaxial transmission line. The steel rod in the central axis of the ceramic was connected to the centre wire of the transmission line. The slotted stainless steel cylinder surrounding the ceramic was connected to earth. The geometry of the ceramic and its electrodes (the central rod and the 4 earthed slots) was such that we obtained close to the maximum difference in output of the dielectric probe (nearly 1 V) between the wet and dry states of the ceramic. Although the output of the dielectric probe can be adjusted to some extent in terms of an offset and gain, this was not done because each sensor was calibrated individually against matric potential.
The ceramic’s electrodes and the dielectric probe were separated by a 0.8 m length of coaxial cable (¼ of a wavelength long), which ran inside an aluminium tube. This length was determined experimentally as the length that gave the maximum difference between the outputs of the dry and saturated ceramic. Electrically, the coaxial cable was invisible to the dielectric probe. The 0.8 m cable length was convenient because it allowed the dielectric probe to be housed above ground even when the porous matrix was installed at depths of up to 0.8 m. A plastic pipe with 1 mm bore was used to connect one of the empty 4 mm holes in the ceramic with the atmosphere. This allows the ceramic to drain easily as the matric potential of the soil becomes more negative.
Calibration of the sensor against matric potential
The hysteresis model of Kool and Parker (1987) was used as a basis to calculate matric potential from the sensor output. It uses the water retention curve of van Genuchten (1980), which for this purpose is written as
[pic]
where v is the sensor output (proportional to saturation) at matric potential ψ and vr, vm, m, n and α are fitted parameters. Here, vm and vr, which represent the voltage output from wet (−25 kPa in a pressure plate apparatus) and air-dry sensors, were determined experimentally. The shape of the main drying curve for the ceramic material was determined using the pressure plate apparatus and the van Genuchten equation was fitted to the data to obtain values of m, n and α.
The main wetting curve cannot easily be measured with the pressure plate, but the hysteresis model of Kool and Parker (1987) allows the main wetting curve to be scaled from the main drying curve by adjusting the α parameter. This was done using measured data on hysteresis over a range of matric potentials that could be reached without requiring pressure plate apparatus (see below). We also coded the hysteresis model of Dirksen et al. (1993) to calculate the matric potential from the sensor output. This model differs from that of Kool and Parker (1987) in that the scanning loops close (Figure 1.3).
The hysteresis between the sensor output and the matric potential was measured between saturation and −75 kPa. A sintered funnel was used with kaolinite sedimented on the surface of the sinter to a depth of approximately 4 mm in order to decrease the air-entry potential of the sinter. The sensor was then immersed in saturated silica paste on top of the kaolinite layer. The water-filled sintered funnel was connected to a reservoir of water in a glass jar. The matric potential was adjusted between 0 and −75 kPa with a regulated vacuum pump that controlled the air pressure above the water in the glass jar.
Results and discussion
Calibration data for porous matrix sensors
To obtain reasonable agreement with the measured scanning curves, the α parameter of the main wetting curve needed to be four times that of the main drying curve, rather than the value of two suggested by Kool and Parker (1987). Our measured scanning curves returned to a sensor output very close to the initial sensor output at saturation. This contrasts with the approach of Feng et al. (2002), who initially vacuum-saturated their sensors, which resulted in an initial drying curve that was never repeated following subsequent wetting and drying. Here, initial saturation was always under atmospheric pressure and under a small positive water potential, by immersing the ceramic in water. This approach has the advantages that the initial state of saturation can be repeated in the field and the additional primary drying curve is avoided.
Field use of porous-matrix sensors
Comparison of hysteresis models
The choice of hysteresis model greatly affected the calculated matric potential. The model of Kool and Parker (1987) indicated that ψ increased after an initial period of drying (1.4A). When the ratio between the α parameter of the main wetting and drying curves was changed from 2 to 4, the increase in ψ predicted from the sensor output increased further. However, the voltage output from the porous-matrix sensor showed that the soil was drying, as did the output from the Profile Probe (see below).
In contrast, the model of Dirksen et al. (1993) indicated that ψ decreased as the soil continued to dry (1.4B). When the ratio between the α parameter of the main wetting wetting and drying curves increased from 2 to 4, the noise in the predicted matric potential also increased, although the lowest estimates continued to follow the main drying curve. However, even estimates of matric potential calculated with just the main drying curve will be subject to some uncertainly because of the shallow gradient.
At low matric potentials, the closed scanning loops amplify the noise in the voltage output of the sensor and suggest large fluctuations in matric potential. This effect could be viewed as an advantage because it gives an indication that the sensor is approaching the lower limit of its range. The Kool and Parker (1987) model, however, drifts towards the higher matric potentials even when the soil is drying, because noise in the sensor output leads to modelled scanning curves that do not close (Figure 1.3), so it never returns to the main drying curve.
Comparison of porous-matrix sensors with water-filled tensiometers
We compared the outputs from the porous matrix sensors and water-filled tensiometers at 0.20 m depth (Figure 1.5). Water content was also measured with Profile Probes (Delta-T Devices Ltd), which indicated that the soil was drying throughout this period. The output from the tensiometers indicated that the matric potential became more negative during May, starting from close to 0 kPa. As the matric potential reached −50 kPa, the porous-matrix sensors started to respond to the decreased matric potential and indicated that the matric potential reached −300 kPa by the end of the second week of June. In plots 1, 2 and 4, the output from the tensiometers stabilised around −90 kPa and then drifted in a less negative direction before returning to atmospheric pressure. In plot 3, however, the tensiometer output followed that of the porous-matrix sensor down to −140 kPa, at which point its output rapidly deviated from the potential measured by the porous-matrix sensor and subsequently followed that of the other tensiometers.
Fig. 1.4. The effect of hysteresis models of Kool and Parker (1987: A) and Dirksen et al. (1993: B) on the matric potential calculated from the output of the porous-matrix sensor. Ratios of 2 and 4 between the α parameters of main wetting and drying curves of were tested in each model. Measurements were made on plot 1 (non-compacted) at 0.2 m depth.
[pic]
Interpretation of field measurements of matric potential
The discrepancy between the matric potentials measured by individual tensiometers and by porous-matrix sensors was investigated in laboratory experiments. These are not shown here but were consistent with the reported behaviour of tensiometers in the geotechnical literature. In brief, the presence of tiny air pockets in the ceramic of the tensiometer means that they are vulnerable to cavitation and so do not report matric potentials more negative than about −95 kPa. Occasionally, an individual tensiometer may be successfully saturated so that it is able to report more negative matric potentials, until it too cavitates, and then reports about −90 kPa.
Our interpretation of the field observations of matric potential as measured by porous-matrix sensors and water-filled tensiometers is therefore as follows. As the degree of saturation of the soil started to decrease, the tensiometers correctly indicated matric potential becoming more negative. As the soil dried to −50 kPa, the upper range of the porous-matrix sensor, so these sensors began to indicate decreasing matric potential. As the soil dried further, the tensiometers installed in plots 1, 2, and 4 experienced an apparent equilibrium matric potential of approximately −90 kPa (see Fig. 1.5), indicative of a pre-existing bubble in the water reservoir. The more highly saturated device in plot 3 (i.e. no free air bubble in the water reservoir) continued to measure correctly the decrease in matric potential until cavitation was triggered at −140 kPa (Fig. 1.5 plot 3). Thus, as the soil continued to dry, the water-filled tensiometers falsely indicated a stable matric potential around −90 kPa. Further drying was correctly captured by the porous-matrix sensors, whilst the water-filled tensiometers continued to desaturate until a continuous air void vented the vacuum within the device to the external pore air at atmospheric pressure.
The interpretation that porous-matrix sensors were correctly reporting changes in matric potential is supported by measurements made at the shallower depth of 0.1 m (Whalley et al., 2005). Both soil water content data from the Profile Probe and matric potential from the porous-matrix sensor indicated a drying trend, interspersed with wetting events due to rainfall. The internal consistency between these results gives confidence that the porous-matrix sensors yield an accurate guide to the matric potential of soil water in the range −50 to −300 kPa.
Conclusions
The new porous-matrix sensor reliably measured the matric potential of soil water in the range −50 to −300 kPa. Calibration of this device demonstrated the importance of using a hysteretic model with closed scanning loops to avoid noise resulting in a substantial drift that gave the impression of matric potential increasing in a drying soil. Comparisons demonstrated that water-filled tensiometers can give stable but false readings of matric potential yet be subject to much lower matric potentials. These results show that the best strategy is to complement direct measurements of matric potential with indirect measurements. The porous-matrix sensor therefore has considerable potential as a tool for understanding plant responses to drying soil.
2. An index of soil strength calculated from the soil moisture status
Although dry soils are stronger than wet soils, the relationship between soil strength and soil moisture status varies greatly with soil type. Mullins and co-workers have a model to predict soil tensile strength (Y), which is based on the product of the degree of saturation (S, volumetric water content / total porosity) and the absolute value of the soil water matric potential (ψ) as follows,
[pic] (2.1)
where c is the cohesion. Sψ is a simple estimate of effective stress (σe), which according to Mullins (2000) works well when S>0.5. The porous-matrix sensor allows us to measure ψ over a sufficiently wide range to provide a useful application for Equation 2.1. The degree of saturation, S, can be estimated if the water content and dry bulk density are known. Equation 2.1 can be expected to work reasonably well in structure-less subsoils. This theory was tested in laboratory experiments in a range of arable subsoils.
Materials and methods
Soils and treatments
Six soils were chosen to cover a range of textures and organic matter content and are listed in Table 2.1. Three separate experiments were conducted. For clarity the treatment combinations are given in Table 2.2. The dry soils were ground, sieved through a 2 mm sieve and then packed into cores 53 mm in diameter and 19 mm deep to a nominal high or low density.
For the low density treatments, we aimed to achieve a density of 1.2 g cm-3. The soils were weighed and packed into the cylinders and were gently tapped. To achieve the low density packing in the case of the clayey Evesham and Boot Field soils some light compaction using a rubber bung was required. For the high density packing we aimed to achieve a density of around 1.7 g cm-3. However this was impossible to do with air dry soil and the Wrest Park, Kirton and North Wyke soils were wetted to around 8.8% w/w water content and tamped into the cylinders. In the case of the Evesham and Boot Field soils the soils were wetted to 20% water content and considerable force was used to pack the cylinders. These two soils achieved 1.7 and 1.56 g cm-3 on average using this method. The exact value of dry bulk density at the time of measuring penetration pressure was recorded.
[pic]
The cores were brought to saturation and then equilibrated on tension tables for matric potentials greater than −30 kPa and on a pressure membrane apparatus for matric potentials less than −100 kPa. After equilibration the soil cores were weighed, so that the water content could be calculated, and then the strength of the soil was measured by either a tensile test or by a penetrometer as described below. The effective stress, σe, was calculated as Sψ.
Measurement of the tensile strength of soil
The tensile strength of discs of soil was measured with the “Brazilian test” (Dexter and Watts, 2001). Discs of soil 17.5 mm in diameter and 10 mm thick were equilibrated to a desired matric potential as described above. The equilibrated discs were crushed using either an Instron loading frame at a constant rate of strain of 2 mm s-1 or a device for applying a constant loading rate to the disc of soil. For both methods of crushing discs of soil, the tensile strength, Y can be calculated from the force, F required to fail the disc.
Table 2.2 Summary of the factor combinations and measurements made in the three experiments.
|Experiment number |Experimental factors in each experiment |Strength Measurements |
| |Soil |Matric potential |Bulk Density | |
| | |(−kPa) | | |
|Experiment 1 |Kirton |10 |High |Tensile strength |
| |Kingsmead |100 | | |
| |Wrest Park |200 | | |
| | |500 | | |
| | |1000 | | |
| | |1500 | | |
|Experiment 2 |Kirton |1 |Low |The resistance of fixed and |
| |Wrest Park |5 |High |rotating penetrometers |
| | |10 | | |
| | |15 | | |
| | |30 | | |
| | |100 | | |
|Experiment 3 |Boot Field |100 |Low |The resistance of fixed and |
| |Evesham | |High |rotating penetrometers |
| |Kirton | | | |
| |North Wyke | | | |
| |Wrest Park | | | |
Measurement of the resistance to penetration of fixed and rotating penetrometers
We measured the penetrometer strength of cores of soil 53 mm in diameter and 19 mm deep that were equilibrated to a desired matric potential as described in earlier. The penetrometer was conical with a 30° semi-angle and a diameter of 2 mm at the base of the cone. The penetrometer was pushed into the soil at a rate of 2 mm s-1 either fixed (non-rotating) or rotating at 26 rev min-1. Penetrometer pressure (Qr for rotating penetrometer, Qnon-rotating for non-rotating penetrometer) was calculated by dividing the maximum force required to push the penetrometer into the core by the cross-sectional area of the base of the cone. In each core we made three fixed and three rotating penetrations.
Results
Soil tensile strength
The tensile strength (Y) of the soil disc determined by constant rate of loading and constant rate of strain agree well with each other and analysis of variance showed that there was no interaction between the method for measuring Y and the soil treatment (combination of soil type and matric potential) (see Whalley et al., 2004). The data were log-transformed to remove the trend in residuals. From here on only the data collected using the Instron and constant rate of strain are reported.
[pic]
Fig. 2.1. Tensile strength of plotted effective stress for Kirton ((), Wrest Park (□) and Kingsmead (○) soils. A logistic function is fitted to the data.
Tensile strength decreased with increasing water content for the three soils tested, although there were three separate curves. However, when the same data for Y were plotted against σe, there was a common relationship for the three soils (Fig. 2.1). A logistic curve was fitted as there is not a single theory for Y that extends from very wet to dry soil (Mullins, 2000). These data confirm that σe can be used to define a common relationship between soil moisture status and moisture status as defined by effective stress.
Penetrometer resistance
The data collected with a rotating penetrometer in experiments 2 and 3 for the nominal low density treatments, as a function of soil water status is shown in Fig. 2.2. As for the tensile strength data, Qr, σe, and ψ required log-transformation to remove a trend in the residual. The fitted curves have the following equations:
[pic] (2.2) (3)
and
[pic] (2.3) (4)
which explain 81 and 74% of the variance, respectively. The slope of the relationship between Log Qr and Log σe was not significantly different to 1, whereas the slope between Log Qr and Log ψ was significantly different to 1 (P < 0.001).
There was little increase of penetrometer pressure with bulk density for the silty clay loam (Kirton soil) but the increase was much greater for the loamy sand (Wrest Park soil). Increasing soil density significantly (P < 0.001) decreased the slope in the relationship between Log Qr and Log σe for the Wrest Park soil, but not for the Kirton soil (P = 0.052). For both soils, the value of the intercept of the fitted curve increased significantly (P < 0.001) with density, although for the Kirton soil that increase was small (Whalley et al. 2004).
The effect of not rotating the penetrometer was to increase penetrometer pressure (Figure 2.3). The data includes those published by Farrell and Greacen (1966) as well as those from experiments 2 and 3 of this work. A curve fitted to the data for a non-rotating penetrometer had the equation
(2.4)
Both the slope and the intercept of this curve differ from that of a rotating penetrometer. The greatest difference is in the intercept, which was only log108.4 for a rotating penetrometer (Equation 2.3).
The data from experiment 3 is summarized in Tables 2.3 and 2.4. There appears to be a complex interaction between soil type and σe that determines penetrometer pressure. For Evesham, North Wyke and Wrest Park soils, which contained 33%, 56% and 80% sand, respectively, increasing density resulted in much higher penetrometer pressures (Table 2.4), despite relatively small effects on σe (Table 2.3). Although there was an interaction (P < 0.001) between penetrometer rotation and soil provenance, rotation decreased penetrometer pressure by a factor of between 2 to 3 (Table 2.4).
Field testing
The field experimentation set up to pursue objective 4 (see section below) gave the opportunity to test whether penetrometer resistance can be predicted from effective stress. This showed that predictions from effective stress did give a reasonable prediction of soil strength measured with a cone penetrometer (Fig. 2.4, note natural scale of axes).
Table 2.3 The effective stress of soil equilibrated at −100 kPa. The SED of the log-transformed data is 0.011 (170 df).
| |Log10 Effective stress |
| |(means of untransformed data are shown in brackets, kPa) |
|Density |Boot Field |Evesham |Kirton |North Wyke |Wrest Park |
|Low |1.92 |1.65 |1.49 |1.56 |1.32 |
| |(82) |(44) |(31) |(36) |(21) |
|High |1.93 |1.72 |1.60 |1.79 |1.45 |
| |(85) |(53) |(39) |(62) |(28) |
Table 2.4 Penetrometer pressures for rotating or fixed penetrometers during penetration in five soils at two densities when equilibrated at −100 kPa. The SED for the log-transformed data is 0.018 (160 df).
| |Log10 penetrometer pressure |
| |(means of untransformed data are shown in brackets, kPa) |
|Density |Penetrometer |Boot Field |Evesham |Kirton |North Wyke |Wrest Park |
|Low |Rotating |2.73 |2.37 |2.94 |2.41 |2.33 |
| | |(536) |(239) |(956) |(272) |(230) |
| |Fixed |2.96 |2.72 |3.26 |2.93 |2.84 |
| | |(904) |(528) |(1841) |(849) |(703) |
|High |Rotating |2.70 |2.90 |2.87 |3.16 |2.89 |
| | |(498} |(792) |(742) |(1457) |(797) |
| |Fixed |2.93 |3.15 |3.24 |3.57 |3.31 |
| | |(848) |(1431) |(1751) |(3715) |(2053) |
Discussion
Tensile strength
Estimates of Y obtained from constant rate of loading and from constant rate of strain were in good agreement. As far as we are aware this is the first comparison of the two different methods described by Dexter and Watts (2001) for measuring Y. An important feature of our data is that they need to be log-transformed so that the residual does not increase with Y. Log-transformation was also found to be necessary by Mullins et al. (1992).
The results presented here confirm that the effective stress concept can be used to give a common relationship between this measure of soil water status and Y for different soils (Mullins et al., 1992). For σe in the range −50 and −200 kPa, the slope of the logistic curve (figure 2.1) that we fitted to the data is approximately parallel to the 1:1 line. In wet soil, when σe is small, c is proportionally greater and estimates of Y were higher than the 1:1 line. This could have occurred if c was dependent on soil moisture and this effect was dependent upon soil type. Alternatively, it is possible that very wet discs of soil may be flattened before they fail and that the tensile strength at failure calculated with the “Brazilian method” could be an over-estimate (Dexter and Watts, 2001).
Penetrometer resistance and soil water status
In the low density soils (Fig. 2.2) the use of log ψ accounted for more of the variance in log Qr than did log σe. However, the relationship between log σe and log Qr has a slope not significantly different to 1. If we constrain the slope to 1, then the regression may simplified as Qr=a(Sψ) (as σe = Sψ and a is a fitted parameter). It should be remembered that this relationship could not be derived from the non-transformed data as log transformation was necessary to remove a trend in the residuals.
Although three of the soils (Evesham, North Wyke and Boot Field) were only equilibrated at −100 kPa, the saturations of these soils at this matric potential were different. This was particularly notable for the Boot Field soil (clay soil) which had a saturation of approximately 0.8 and had a much higher penetrometer pressure than the other soils at this saturation (Fig. 2.2c). The relationship between log Qr and log σe was parallel to the 1:1 line for the entire range of σe (Fig. 2.2b), in contrast to the relationship between log Y and log σe (Fig. 2.1). If penetrometer pressure is an accurate reflection of soil failure, this would suggest that the shape of the relationship between log Y and log σe is more likely to reflect over-estimates of Y by the “Brazilian method” in very wet soil.
The resistance to a non-rotating penetrometer could also be explained by a common relationship with log σe for a wide range of soils (Fig. 2.3). Although the slope of this relationship on a log scale was significantly different to that for a rotating penetrometer, the lines were approximately parallel for the range of σe considered. The difference between rotating and non-rotating penetrometer pressures therefore appears to be independent of soil water status.
Field testing of this theory showed that it was possible to gain reasonable predictions of soil strength from effective stress (Fig. 2.4). In effect, this means that it is possible to monitor soil strength in real time by logging measurements of degree of saturation and matric potential. This was exploited further in the work to achieve objective 4.
[pic]
[pic]
Fig. 2.4. Measured penetrometer resistance in the field in 2004 plotted against predictions from effective stress.
Dealing with high bulk density
The use of effective stress only provided a statistically useful prediction of penetrometer pressure when the data collected from the high density soils was excluded. For some (but not all) soils, an increase in density gave an increase in penetrometer pressure that was too great to be explained by an increase in σe. Vepraskas (1984) also found the relationship between penetrometer pressure and effective stress was not necessarily common for different soil types and there was some evidence that the relationship depended on soil density. There much evidence that demonstrates that the relationship between penetrometer resistance and soil density is nonlinear (Mulqueen et al., 1977; Hernanz et al., 2000). As soil density increases, initially penetrometer resistance is insensitive to changes in soil density, but at higher values of soil density penetrometer resistance increases rapidly with soil density. The strength of low density soil does increase rapidly with soil drying. It seems that for low density soils, the soil strength is explained almost entirely by the capillary pressure of the soil water, as described above. At high soil bulk densities, when the penetrometer deforms soil there will be insufficient void space to accommodate the displaced particles without significant soil particle rearrangement. Here it is likely that penetrometer pressure will depend on both soil bulk density and effective stress.
The data presented here provides a sound basis for deciding when the simple effective stress model can be used to predict penetrometer pressure without the need for soil-specific calibration. As far as we are aware, this is the first description of the soil conditions that define the validity of the use of effective stress to predict penetrometer pressure over a wide range of soil types. Future work should be directed at developing a simple model that includes the effect of soil density on penetrometer pressure without introducing parameters that depend on soil type.
Implications for root growth
The resistance to a rotating penetrometer gives a useful estimate of the resistance experienced by plant roots as they elongate through soil (Bengough et al., 1997). This allows us to estimate the extent to which soil drying will affect root growth due to an increase in soil strength by comparing Qr with the pressures that roots exert. The mean maximum growth pressure (σmax) of seedling roots of different species is about 400 kPa (Clark and Barraclough, 1999). However, water stress decreases σmax, by 70 kPa for each 100 kPa of osmotic stress in PEG solution for pea seedlings (Whalley et al., 1998). If we assume that the matric potential of soil water exposes the root to the same water stress as an equivalent osmotic solution of PEG, then the matric potential at which soil is impenetrable can be estimated (i.e. when σmax = Qr).
For the clay (Boot Field) and loamy sand (Wrest Park) soils, the matric potentials at which Qr is equal to σmax of pea roots are approximately −100 and −250 kPa, respectively. These estimates reflect differences in the water release characteristics of the two soils, which cause differences in σe. In practice, these estimates are likely to be approximate due to the different effects of osmotic solution and matric potential on the permeability of the root to water as well as the assumption that Qr does accurately predict the resistance to root elongation. It is interesting to note that water potentials of −250 kPa in the rooting medium have relatively little effect on root elongation when mechanical impedance is low (Sharp et al., 1988; Whalley et al., 1998). This result emphasizes the importance of good soil structure to provide channels for root exploration of the soil profile and how seriously soil can limit root exploration in the soil.
Models to predict penetrometer resistance
Farrell and Greacen (1966) developed a model that accurately predicts the penetrometer pressure of soils. Their model takes into account the complex processes of plastic and elastic failure of soil around the penetrometer as it deforms the soil. However, their model requires detailed measurements of soil mechanical properties that are not generally available.
There are a number of empirical models that relate penetrometer pressure to both soil density and soil water content in some nonlinear way. A recent example of such a model for a non-rotating penetrometer is presented by Hernanz et al. (2000). This and similar models all have parameters that depend on soil type, four in the case of Hernanz et al. (2000). In this report we have shown that in low density soil, if the soil water status is defined as effective stress, we can derive a model to predict penetrometer pressure of six soils using only two parameters that are independent of soil type (Equation 2.4, Fig. 2.3). More work is needed to determine the extent to which effective stress can be used to develop models with fewer parameters to predict the penetrometer pressure and hence resistance to root elongation of high density soils.
In practice, a model is only useful if the input variables are known or can be measured. This is the case for the degree of saturation, S, and the matric potential of the soil water, ψ. Recently, sensors have been described which are able to measure the matric potential of soil water over a wide range (Whalley et al., 2001; Whalley et al. 2005; also see objective 1). Thus in low density soils we are already in a position where we can predict the resistance to root elongation by measuring the soil water status (Whalley et al. 2004). It seems clear that better sensors to measure soil density are needed and research effort should be directed at this problem. The requirement to predict impedance to root elongation from sensors is likely to increase as the relationship between crop yield and root impedance becomes better understood and viewed as a possible management tool.
Conclusions
In compressible re-packed soils, there was a common relationship between effective stress and both tensile strength and penetrometer pressure, independent of soil type. However, when soils with a wide range of bulk densities were considered, there was no common relationship between penetrometer pressure and effective stress. The effect of rotation was to decrease the penetrometer pressure by a factor of between approximately 2 and 3. Assuming that the pressure of a rotating penetrometer is a good guide to the pressure roots need to exert in order to deform soil, we estimate that roots will typically be unable to penetrate unstructured soils with a matric potential lower than −250 kPa.
Objective 3. Estimating N mineralisation from the electrical conductivity of the pore water
Recent publications (De Neve et al., 2000; Nissen et al., 1998) describe how electrical conductivity of the soil water can be used to estimate N mineralisation under vegetable cropping on light soils. This is an important claim, which needs to be tested for the soils that are used in the UK arable sector. If this claim is true then it will be an important tool for assessing the capacity of a soil to supply N to a plant. In this part of the project, we used commercially-available Sigma probes, which have recently been introduced to the market by Delta-T Devices. They are designed to measure the electrical conductivity of the pore water (Hilhorst, 2000). This part of the project was carried out in close collaboration with Tony Miller and Darren Wells at Rothamsted, who were working on the development of a nitrate-sensitive electrode (Defra project AR 0910).
Materials and methods
We conducted investigations in both the laboratory and the field. Electrical conductivity and soil water content were measured with either a time domain reflectometer (TDR) or a sigma probe. Preliminary calibrations between electrical conductivity were made using a pure sand grade RH 65. This was followed by making field measurements of electrical conductivity in soils at Broadbalk at Rothamsted and also at Silsoe. At Rothamsted low and high N soils were used and at Silsoe Research Institute the electrical conductivity of the soil was monitored before and after the application of liquid N fertilizers to the plots either non-irrigated or irrigated to keep them well-watered (see objective 4).
Results and discussion
Initial results in sand were very promising. When bulk electrical conductivity as adjusted so that it reflected pore water electrical conductivity (Hilhorst, 2000) it was linearly related N concentration of the pore water solution (Fig. 3.1). However, field data were less encouraging. Neither measurements of bulk soil electrical conductivity nor estimated pore water electrical conductivity were affected by the application of N at 60 kg ha-1 in loamy sand plots that were well watered and un-irrigated. Measurements of bulk electrical conductively in the un-irrigated plots were less than those in the well watered plots, however, the calculated pore water electrical conductivity in both plots were similar. This suggests that even in a loamy sand soil the background electrical conductivity is high enough to dominated any change in electrical conductivity that may result from N fertilization. Comparison between measured electrical conductivity and N estimated with the Rothamsted nitrate-selective electrode at Rothamsted in low and high N plots, confirmed that the measurement of electrical conductivity to estimated N mineralization is an insensitive approach. Although we were able to detect differences at Rothamsted xx in the low N plot and xx in the high N plots, these differences were small in comparison with those detected by the N selective electrode technique. It should be noted that the approach of using the electrical conductivity of soil to estimate N content was developed on horticultural soils that have much higher rates of N application.
[pic]
Fig. 3.1. Relationship between electrical conductivity of the pore water solution in sand and the concentration N added as potassium nitrate
Conclusions
The results that we have obtained do not support the use of either bulk soil or pore water electrical conductivity as a useful estimate of the N content of arable soils. Instead, the nitrate-selective electrode developed by Rothamsted is a much better approach.
Objective 4. Relating the index of soil strength developed in objective 2 to root and shoot development
Information relating to the threshold value of soil strength to which plants respond is very limited. Reports tend to refer to either bulk density or soil moisture content, but not the more relevant variable, soil strength, which integrates the two. To obtain such information, initial studies were in controlled environments. The matric potential of soil was controlled by a negative-pressure circulation system (Lipiec et al., 1988). In another type of experiment, the matric potential was kept constant but the mechanical impedance of sand was varied independently of bulk density and aeration.
The main part of the work was field trials in which crop performance was be related to soil conditions. Continuous monitoring of soil conditions will allow crop growth to be related to the rooting environment in the field. Monitoring of xylem-sap ABA will indicate the role of root-shoot communication in co-ordinating this response. This experimental design will be used for two sowing occasions.
Materials and methods
Plant material
Wheat (Triticum aestivum L.) cv. Clare was used all of the work reported.
A controlled environment experiment on the effect of high soil strength
A sand culture system described by Clark et al. (2002) was used. This allows mechanical impedance to be varied independently of aeration and water status of the growing medium and uses non-flooded conditions. When a weight is placed on the surface of the sand core, the mechanical impedance of the medium is increased as the resistance of sand grains to displacement is increased, but there is negligible compaction of the sand. Two-day old wheat seedlings were planted into the cores, one seedling per core. The effect of two levels of impedance was tested: ‘impeded’ and ‘control’. The experiment was carried out in a controlled environment using a 16 h daylength with day/night temperatures of 22 and 16 °C, respectively, a relative humidity of 70%, and a photosynthetic photon flux density of 300-350 μmol m-2 s-1 by fluorescent tubes, supplemented by tungsten bulbs. The experiment was randomized in 6 blocks and each block contained 3 impeded plants and 3 control plants. The plants were harvested 54 d after planting, with the aid of a root harvesting apparatus (Clark et al., 2000). The lengths of the longest root axis of each plant were measured at harvest. The dry weights of the root and shoots were measured.
A controlled environment experiment on the effect of small difference in matric potential
An experimental facility was constructed to maintain soil moisture status at a constant matric potential using a negative-pressure circulation system (Lipiec, 1988) within a controlled environment. The 20 pots were filled with either sand, or soil (field soil (collected to a depth of 10 cm, then partially dried and sieved to 5 mm from the experimental site at Silsoe described below) depending on the experimental conditions. The plant containers were constructed of PVC pipe of 1.42 litre volume with a series of interconnected ceramic tubes placed through the soil matrix. The negative pressure was achieved by a series of water tanks set a predetermined distance from the height of the soil columns and connected by flexible PVC tubing to the ceramic pipes in a siphon system.
Wheat was sown in the pots after the soil had initially been set to field capacity then maintained for 3 days at −3 kPa. The growth room was set to a 14/10 hour day/night cycle with day/night temperatures of 20/16°C and day/night relative humidity of 80/70%. After 4 weeks the treatment siphon system was connected allowing the 10 treatment pots to be maintained at soil matric potentials of −19 kPa) with the control pots maintained at −3 kPa. The soil matric potential were continuously monitored using mini-tensiometers (SWT5, Delta-T Devices, 128 Low Road, Burwell, Cambridge, CB5 0EJ, UK), inserted through holes drilled at intervals from the base of the pot. The tensiometers were connected to a datalogger (Dl2e, Delta-T Devices), which allowed the monitoring of water potentials every 5 min and confirmed that the disturbance caused by weekly watering with 5 ml standard nutrient solution (Phostrogen, Monsanto, UK) persisted for less than 2 h.
At 7-10 weeks the above ground biomass was harvested, then roots recovered by soil washing and subsequently dried at 50 °C until constant weight was reached.
Field experiment
The field experiments were set up on the farm at Silsoe Research Institute on a field of loamy sand. The treatments were 4 levels of compaction and 3 irrigation regimes. The compaction treatments were applied with 0, 1, 4, or 8 passes of an 11 tonne tractor. The plots, 4 m wide and 12 long, were uniformly wheeled with the tractor. Irrigation treatments were applied using drip tape spaced at 37.5 cm. The plots were either not irrigated or they were irrigated to maintain a matric potential between −50 and −80 kPa or to a matric potential close to −5 kPa, as determined from measurements with a water-filled tensiometer. This gave 12 treatment combinations (4 compaction levels x 3 irrigation regimes), which were replicated in 3 randomized blocks. The plots were sown with winter wheat in 2002 and 2003. These will be referred to as experiments 1 and 2 respectively. The same field was used for both experiments but the second experiment had a new site to avoid any possible effects of residual compaction. A photograph of the 1st experiment at different stages of crop growth is shown in figure 1
The matric potential was measured in all of the plots of block 2 with a water-filled tensiometer (SWT6, Delta-T Devices) at a depth of 0.20 m. The output from these water-filled tensiometers was used to control the irrigation. Profile probes were installed in all the plots of block 2 to measure the soil water content at depths of 0.10, 0.20, 0.30, 0.40, 0.60, and 1.00 m (Whalley et al., 2004). In 2004, water potential was also monitored at these depths in the non-irrigated plots with novel porous-matrix sensors of matric potential (Whalley et al., 2001; 2005). The strength of the soil was measured with a Bush recording penetrometer (Findlay, Irvine Ltd., Bog Road, Penicuik, Midlothian, UK). In the 2003 experiment the penetrometer strength of one of the blocks was measured periodically during the spring and in 2004 experiment the penetrometer strength of all 3 blocks was recorded periodically. In the 2004 experiment the leaf area index was estimated (Sunscan, Delta-T devices) by measuring the portion of sunlight that was intercepted.
Adaxial and abaxial stomatal conductance to water vapour diffusion were measured, on 8 and 9 June 2004, sequentially, with a steady state porometer (PMR-3, PPsystems, Hitchin, Hertfordshire, UK). The recently fully expanded flag leaf was used.
Pre-dawn leaf water potential was measured using a Scholander-type pressure chamber (Skye Instruments, SKPM 1400) on 9 June 2004. Plants to be sampled on each occasion were covered in black plastic before dawn and measured immediately to avoid leaf responses to light and VPD. The pressure chamber was maintained at high relative humidity to avoid evaporation from the leaf by lining with wet tissue. The fully expanded flag leaf was cut and immediately transferred to the pressure chamber in a sealed container to maintain high relative humidity within.
Leaf ABA content was determined from a 5.5 cm2 (3/6/04) section, or 3 x 5mm diameter leaf discs (20/5/04) from the middle of a recently fully expanded leaf was taken from each plant and immediately frozen in liquid nitrogen, then stored at −20°C for later ABA analysis by radioimmunoassay (Quarrie et al., 1988). Leaf discs for ABA determination were sampled in both field experiments.
Xylem sap was collected using a similar protocol to that for leaf water potential using a Scholander-type pressure chamber (Skye Instruments, SKPM 1400). Plants to be sampled on each occasion were covered in black plastic before dawn and measured immediately to avoid leaf responses to light and VPD. The pressure chamber was maintained at high relative humidity to avoid evaporation from the leaf by lining with wet tissue. A plant was selected which had developed a true stem, removed from the soil and transferred to the pressure chamber in a sealed container to maintain a high relative humidity. The shoot was cut from the plant, inserted into the pressure chamber and pressurized to force xylem sap out of the cut stem (up to 4 bars over the balancing pressure). The sap was collected by pipette and immediately immersed in liquid nitrogen, then stored at −20°C for later ABA analysis by radioimmunoassay.
The pH of the xylem sap collected as above was measured in the 2004 experiment using a pH microelectrode (HI 1083, Hanna Instruments, UK) and meter (pH 211, Hanna Instruments, UK)
Each plot was harvested by hand. The amount of dry matter in two 0.5 m2 sub-plots was measured and the grains were separated by hand threshing. The C and N content of the grain was measured.
[pic]
Figure 4.1 The 2003 field experiment at different stages of crop growth.
Results
Effects of mechanical impedance and matric potential in controlled environments
The dry weight of root and shoots are both approximately halved by the high impedance treatment. The root to shoot ratio was increased by impedance. The length of the longest root was less in the impeded treatment, but it as approximately 90 percent of the length of the control plants. The number of tillers was halved by the impedance, but for both impeded and control plants the number of tillers was high in comparisons with the field growth of wheat.
Lowering the matric potential from −3 to −19 kPa had no significant effects on the dry matter of the wheat plants.
Effect of treatments on soil physical properties in the field
Soil strength: The compaction treatments increased the penetrometer pressure, although the effect was greater in 2003 than in 2004 (Fig. 4.2A). In both experiments, irrigation decreased penetrometer pressure (Fig. 4.2B) and the effect of irrigation was greater than the effect of compaction, although both treatment main effects were significant (P < 0.001). In 2004, there was a weak interaction (P = 0.03) between the effects of irrigation and compaction on penetrometer pressure. As only the penetrometer pressure of one block was measured in 2003, any interaction could not be tested for in that data.
Soil water status: Data from the dielectric soil moisture meters, expressed as degree of saturation (S), showed that the two seasons had very different patterns of soil drying. In 2003, the non-irrigated plots started drying from early April (Fig. 4.3A) but in 2004 the soil did not start to dry until the middle of May (Fig. 4.4A). In 2003, the measurement of matric potential (ψ) in the non-irrigated plots relied on water-filled tensiometers (Fig. 4.3B). Data from the water-filled tensiometers in the non-irrigated plots after 21 April are not shown as they indicated that ψ was becoming less negative as the soil dried. This behaviour is common in water-filled tensiometers that have been exposed to soil with ψ < −100 kPa and is indicative of partial failure (See objective 1). In 2004, this limitation was overcome by the use of the novel porous-matrix sensor, which demonstrated that ψ decreased to −400 kPa by mid-June (Fig. 4.4B).
Matric potential was plotted against S for the non-irrigated plots to give water release curves which were consistent with compaction decreasing pore sizes.
Effect of treatments on the growth of wheat
Irrigation significantly (P < 0.001) increased shoot dry weight and grain dry weight (Tables 4.1 and 4.2). This effect was seen in both years, although the effect was greater in 2004 than in 2003, reflecting significant (P < 0.001) interactions. There were no significant main effects of compaction treatments on shoot or grain dry weights, nor any significant interactions.
[pic]
Fig. 4.2 . The effect of compaction (A) and irrigation (B) treatments on the penetrometer pressure of the top 0.5 m of soil.
Table 4.1. Effect of irrigation on shoot dry weight in the field
Means are shown across compaction treatments.
The SED for comparison of means is 0.048 (10.7 df).
| Year |Shoot dry weight (kg m−2) |
| | |Irrigation | |
| |non-irrigated |−50 to −80 kPa |well-watered |
|2003 |1.40 |1.55 |1.63 |
|2004 |1.27 |1.45 |1.77 |
Table 4.2. Effect of irrigation on grain dry weight in the field
Means are shown across compaction treatments.
The SED for comparisons of means is 0.023 (13.8 22 df).
|Year |Grain dry weight (kg m−2) |
| | |Irrigation | |
| |non-irrigated |−50 to −80 kPa |well-watered |
|2003 |0.75 |0.88 |0.90 |
|2004 |0.69 |0.83 |1.07 |
[pic]
Fig. 4.3. Effect of irrigation treatments on degree of saturation (A) and soil matric potential (B) in 2003. Matric potential was measured with a water-filled tensiometer.
Fig. 4.4. Effect of irrigation treatments on degree of saturation (A) and soil matric potential (B) in 2004. Matric potential was measured with water-filled tensiometers for the irrigated plots. For the non-irrigated plots, estimates of matric potential greater than −50 kPa were obtained from water-filled tensiometers and those < −50 kPa from the porous-matrix sensor.
Relationships between wheat growth and soil physical conditions
Soil water status: The data of ψ, S and Sψ were accumulated with time (in a way analogous to thermal time) and then plotted against shoot dry weight. Shoot dry weight decreased with increased accumulated ψ and Sψ (Fig. 4.5A, C). Linear regression showed that the relationships between shoot dry weight and accumulated matric potential were not significantly different from parallel, so that for every MPa day of matric potential that was accumulated there was a reduction of 87 g m−2 (s.e. = 12). Similarly, for every MPa day of effective stress that was accumulated there was a reduction of 160 g m−2 (s.e. 23.2) in shoot dry weight. Accumulated matric potential accounted for 67% and 75% of the variance in dry matter yield in 2003 and 2004 respectively, whereas accumulated effective stress accounted for 80% and 67% of the variance, respectively. In contrast, the relationship between S and shoot dry weight was quite different between the two years (Fig. 4.5 B).
Here we note that porous-matrix sensors to measure matrix potential were not available in 2003, thus the matrix potential estimates form the water filled tensiometers in the un-irrigated plots may be underestimates of soil drying (Whalley et al., 2005). However, this relates to data from 2 non-irrigated plots in 2003 and this data did not have much leverage on the curve fitting.
Penetrometer pressure: To compare the penetrometer pressure with the dry matter yield, we chose a date each year when the soil in the non-irrigated plots had been drying for approximately 2 weeks according to tensiometers at a depth of 20 cm. Shoot dry weight decreased with penetrometer pressure in both years. Linear regression showed that the slopes of the relationships were not significantly different between the two years. For an increase in penetrometer pressure of 1 MPa, the shoot dry weight decreased by 223 g m−2.
[pic]
Fig. 4.5. Relationships between dry matter yield and accumulated matric potential (A), accumulated degree of saturation (B) and accumulated effective stress (C) in 2003 (○) and 2004 (●). In both years, these data were accumulated between 1 April and 14 June to coincide with the period of most rapid shoot growth.
ABA and pH measurements of the xylem sap
The treatments had little effect on shoot water potentials. The mean pressures were −0.10 MPa pre-dawn and −1.23 MPa at midday. However, irrigation significantly increased stomatal conductance from 247 to 468 mmol m-2 s-1.
We did not find any significant relationships between the imposed treatments and the ABA concentration in the xylem sap, ABA concentration in the leaf tissue nor the pH of the xylem sap.
Observations on root growth in the field
Soil pits dug in the plots in 2003 and 2004 showed that there were no gross differences in rooting. In both years the non-irrigated plots had roots at depths of more than 1 meter. It should also be noted that as it was winter wheat sown in the autumn soil drying in the surface is unlikely to have large affect on rooting depth. This is particularly the case for 2004 where soil drying in the surface did not start until the middle of May and deeper layer remained wet until late June.
Discussion
Effect of treatments on soil physical properties
The greater effect of the compaction treatments on penetrometer pressure in 2003 than in 2004 is consistent with the soil water contents at the time that the treatments were imposed. In the autumn of 2002, the soil moisture was 15% by weight, close to the optimum for compaction for this sandy loam, whereas in the autumn of 2003, the soil moisture was only 5%. Changes in soil bulk density due to compaction treatments were very small (about 0.1 g cm−3), especially in comparison with the range of bulk densities (1.1 to 1.6 g cm−3) that are commonly used to increase the strength of soil in pot experiments (e.g. Mulholland et al., 1996).
As irrigation treatments had a greater effect on penetrometer pressure than compaction treatments, differences in penetrometer pressure between plots mainly reflected soil moisture. Even the plots irrigated to keep them between −50 and −80 kPa had a penetrometer pressure more than twice that of the well-watered plots. This is consistent with laboratory penetrometer measurements on the same soil which showed four-fold increase in penetrometer pressure between saturation and a ψ of −30 kPa (Whalley et al., 2004; objective 2).
Effect of soil physical properties on growth: matric potential or soil strength?
In 2003, there was a common relationship between dry matter yield and penetrometer resistance, whether that resulted from compaction or irrigation. This is similar to the relationship reported in pot experiments by Masle and Passioura (1987) and Masle (1998). This suggests (but does not prove) that the wheat crop here responded to the increased soil strength in drying soils, rather than to the lack of water per se. In 2004, the relationship was less clear, presumably due to the smaller effect of compaction on soil strength that year and the later start of soil drying.
It is difficult to disentangle the direct effects of soil strength and ψ on yield because soil strength can be considered to be a proxy for ψ, and ψ a proxy for soil strength. Dry matter yield decreased with accumulated Sψ and accumulated ψ in both years. However, several comparisons lend weight to the view that the growth of wheat was limited directly by soil strength due to drying rather than ψ. In the field, even the limited drying (to ψ between −50 and −80 kPa) of the intermediate irrigation decreased dry matter and grain yield compared with the well-watered plots. This was especially evident in 2004 when there was better control of ψ in the well-watered treatments. In contrast, when wheat was grown in controlled environments, shoot dry matter production was insensitive to changes in ψ (in the absence of changes in strength) between −3 and −19 kPa. However, the growth of wheat was sensitive to mechanical impedance in the absence of water stress in controlled environments.
In contrast to measurements of penetrometer pressure, soil moisture status was monitored continually. Such data has the advantage that it can be accumulated with time in a manner analogous to thermal time (Fig. 4.4). If soil water status is accumulated in this way, it will give an estimate of the total exposure of the root system to soil drying. These estimates showed that the slopes of the relationships between dry matter yield and accumulated ψ and accumulated ψS were similar in both years. While both ψ and ψS gave similar relationships here, ψS has the advantage that it can predict penetration pressure that is largely independent of soil type (Whalley et al., 2004; objective 2). Although ψ is also strongly related to penetrometer pressure, the relationship will differ with soil type. If the yield reduction in wheat, because of soil drying, is related to the effects of increased soil strength, then the data in Fig. 4.4C would be expected to give a guide to the effects of soil drying on wheat yield over a range of soil types. This remains a hypothesis to test in the future.
The estimates of accumulated soil water status were based on data from a depth of 0.2 m. Accumulation of data at other depths would presumably rescale these relationships: shallower sensors would be affected by transient rainfall events while deeper sensors would give be less sensitive to soil drying. The growth response of shoots observed when roots intercept strong soil is frequently used to support the hypothesis that root-sourced signals determine the growth rate of shoots (Masle 1998; Montague et al., 2001). The current view is that a small number of roots attempting to penetrate strong soil will generate a signalling response that will give these roots a disproportionate effect on the shoot development. In our field experiment where we were looking at the response of winter wheat with a well developed root system to surface drying. This was particularly the case in the 2004 experiment where soil drying started in mid May. Between the 12 May and 25 May 2004 the percentage of variance in leaf area index accounted for by the penetrometer pressure in the top 0.5 m of soil increased from 0 (non significant) to 16% (P = 0.009). By the 9 June 2004, 67% of the variance in leaf area index was explained by the penetrometer pressure (P < 0.001). From Fig. 4b it should be noted that throughout this period the root system was well-watered at depth greater then 0.4 m and for much of the time greater then 0.3 m. This is evidence to support a hypothesis that local drying in the root system can determine shoot growth.
Evidence for signalling
The lack of treatment differences between the pre-dawn leaf water potentials shows that even the non-irrigated plants were able to re-hydrate overnight, consistent with the observed high soil moisture at depth. In the day, irrigation had only a small effect on leaf water potential, although it almost doubled gs. There was no discernable effect of irrigation on xylem sap ABA concentration, so ABA was not confirmed as the signal for effects of the treatments. We suggest that hydraulic signalling may be involved, according to the relationship below:
gs = (ψs − ψl) hc / D
where ψs is the soil water potential, ψl the leaf water potential, hc is the hydraulic conductance of liquid water through the soil and plant, and D is the leaf to air vapour pressure gradient (Comstock, 2002). This means that gs would have decreased with hydraulic conductance and ψs as the soil dried. Changes in ψs − ψl were likely to have been much smaller than changes in soil hydraulic conductivity, which would decrease approximately 100-fold as ψ decreased from −10 kPa to −100 kPa in this soil.
When, in controlled environment experiments, mechanical impedance was varied independently of water status in a controlled environment, there was a large effect on root and shoot growth, but no effect on gs, and only a small effect on leaf water potential. This is consistent with the effects of mechanical impedance being mediated by root to shoot signalling, although the lack of effect of impedance on gs is different to previous studies (Masle and Passioura, 1987; Masle, 1988). This also suggests that the response of gs to irrigation in the field was not a response to the mechanical impedance of the drying soil and supports the hydraulic signalling explanation above. There may be two signalling mechanisms working in wheat in the response to drying soil: hydraulic signalling due to the lower hydraulic conductivity of the soil that primarily affects gs and non-hydraulic signalling of mechanical impedance that directly affects shoots growth, independently of gs.
Conclusions
There was a common relationship between the yield of wheat in the field and estimates of soil strength, regardless of whether this was changed by soil drying or compaction. High mechanical impedance, in the absence of other changes of soil physical properties, also decreased the growth of wheat in controlled environments. Accumulated effective stress was a good predictor of yield, which emphasises the importance of the root environment. The large response of growth to irrigation was striking. While the response of stomatal conductance to soil drying in the field implies that there is hydraulic signalling in drying soil, we suggest that the role of high soil strength as a limitation to growth in moderately dry soil requires further research.
5. Scope for recording and processing data in the field and Technology Transfer
The value of this work to Defra depends on its dissemination to the arable farming community. The project was presented in poster from at a HGCA farmer/grower meeting, a BCPC meeting and at the Royal show. We have also described the project to LEAF during a visit to Silsoe Research Institute. We have written 5 papers for leading journals in soil and plant science.
Points 5.1 to 5.4 were identified as key questions in the proposal. We will comment on our progress and the scope for future development:-
5.1 Using the knowledge gained to build a picture of the soils and soil conditions that are likely to be prone to depressed shoot growth because of root to shoot signalling. Identification of the soils and/or soil conditions would allow the farmer to take preventive action prior to sowing.
The physiological mechanisms that lead to reduced yield because of soil drying remain to be fully explained (objective 4). We have shown that there is no doubt that high soil strength can reduce crop growth (from lab experiments), but in the field experiments stomatal conductance (gs) is also affected by soil drying. It seems reasonable to conclude that in wheat, two signalling mechanisms working in wheat in the response to drying soil: hydraulic signalling due to the lower hydraulic conductivity of the soil that affects gs and non-hydraulic signalling of mechanical impedance that directly affects shoots growth, independently of gs.
Taking the results from objectives 2 and 4 into account it would seem likely that the greatest benefits could be obtained by identifying soil management systems in which soil strength increases more slowly with soil drying. This remains a matter for future research. However, we have identified benefits for the crop if this were to be achieved.
5.2 Using the knowledge gained to irrigate to keep soil strengths below a threshold value. This action would not be appropriate for all crops.
We have shown that in wheat, any stress imposed on the roots results in a loss in yield. The work described in objectives 1, 2 and 4 provides an excellent basis to predict the effect of soil physical conditions on yield. In objective 4, we demonstrated that using the sensors developed in objective 1, matric potential and effective stress can be accumulated in a manner analogous to thermal time. The accumulated matric potential and effective stress at a depth of 20 cm were highly correlated with yield in field experiments conducted in 2 consecutive years. This knowledge is not only useful for irrigation; it also provides the basis for the early prediction of relative yield in a field. The sensors developed in this project could be distributed in a field to monitor, in this case, soil drying and its spatial variability (paper 5 below). We have held discussions with representative of industry (e.g. Plextek.co.uk) and there is no technical barrier to the development of low power telemetry. The application to agriculture will depend on the demonstration of the commercial value to developing such a system.
We have also shown that as winter wheat dries the soil in spring, the ensuing increase in soil strength after only 2 weeks of soil drying can explain a high proportion of the variance (more than 70%) in crop yield. In 2003 this prediction would have been possible as early as 24 April and in time to adjust N application should an appropriate protocol be available. This is an illustration of a possible empirical application of the output of this project, but further development is needed.
Crops other than wheat may respond differently to soil drying, but the sensing approaches we have developed are equally applicable. In this project we have identified serious limitations of water-filled tensiometers. This is the first work to identify that water filled tensiometers can give very misleading measurements to indicate that a soil that is actually drying, has a stable matric potential (Objective 2 and paper 3 below).
3. Using the knowledge gained to refine our estimates of the effect of soil strength on yield.
The evidence presented in this report suggests that any soil drying is likely to result in yield loss (Objective 4). We have also confirmed the role of high soil strength in reducing crop growth. The field evidence showed the soil drying had a greater effect on yield than compaction. However, we used a light soil in our field experiment and in heavier soils increased strength due to compaction would also depress yield. It is likely that by studying the interaction between soil drying and compaction on yield over a range of soils, we will be able to determine the relative importance of the two signalling pathways (see 5.1 objective 4 and paper 4 below). It is only the findings of this project that has placed us in a position to develop testable hypotheses.
4. Investigate the possibilities for using any data on N mineralization that can be deduced from electrical conductivity measurements of pore water in the planning of fertiliser application.
We did not find that this was a useful approach in the arable context. The approach to N measurement taken by Dr. Miller is superior.
By using the sensors developed in this project with those of Dr. Miller, there is an opportunity to understand how physical stress mediated N use by crops. This is a significant new opportunity that is the direct result of these 2 projects (AR0913 and A0910).
Summary of project output
Scientific papers
1. Whalley, W.R., Cope, R.E., Nicholl, C.J. and Whitmore, A.P. (2004) In-field calibration of a dielectric soil moisture meter designed for use in an access tube. Soil Use and Management. 20: 203-206.
2. Whalley, W.R., Leeds-Harrison, P.B., Clark, L.J., and Gowing, D.J.G. (2004) The use of effective stress to predict the penetrometer resistance of unsaturated soils. Soil and Tillage Research (In Press).
3. Whalley, W.R., Clark, L.J., Take, W.A, Bird, N.R.A., Leech, P.K., Cope, R.E. and Watts C.W. (2005) In-field measurements of the matric potential of soil water between saturation and −300 kPa. European Journal of Soil Science. (Submitted)
4. Whalley, W.R., Clark, L.J., Gowing, D.J.G., Cope, R.E., Lodge R.J., and Leeds-Harrison P.B. (2005) Reduction in the yield of wheat because of soil drying and high soil strength. Journal of Experimental Botany (To be submitted following internal review).
5. Clark, L.J., Gowing, D.J.G., Lark, R. M., Leeds-Harrison, P.B., Miller, A.J., Whalley W.R. and Whitmore, A.P. (2005) Measurement and interpretation of the soil conditions that limit the productivity of cereals: A review. Journal of Agricultural Science (To be submitted following internal review)
Conference papers
1. Miller, A, J., Wells, D.M., Braven, J., Ebdon, L., Le Goff, T., Clark, L.J. Whalley, W.R., Gowing, D.J.G. and Leeds-Harrison, P.B. (2003) Novel Sensors for Measuring Soil Nitrogen, Water Availability and Strength. British Crop Protection Council Meeting, Glasgow 1107-1114.
2. Whalley W.R., (2003) What can soil physics tell us about root growth? April meeting of the British Society of Soil Science, Nottingham University, UK.
3. Whalley W.R. (2004) Soil-plant interactions: A soil physics perspective. COST 859, Working Group 4 Meeting Integration and application of phytotechnologies, Leipzig.
4. Whalley W.R. (2005) Soil strength as a constraint to the growth of wheat. Roots and the Soil Environment.AAB meeting University of Nottingham, Sutton Bonington. 4-6 April.
Presentation to the public and at grower meetings
1. Royal Show, Warwickshire July 2003.
2. BCPC meeting Glasgow 2003.
3. HGCA workshop on managing soil and roots for profitable production, Heythrop Park, Oxford, March 2004 (Poster presentation)
4. Presentation to LEAF at Silsoe Research Institute May 2004.
Popular press
Sensing soil strength, July 15 2003 Arable Farming
References
1. Ayers, P.D., Bowen H.D., (1987) Trans. ASAE 30, 1331–1336.
2. Bengough, A.G., McKenzie, B.M., (1997) Journal of Experimental Botany 48, 885–893.
3. Bengough, A.G., Mullins, C.E., (1990) Journal of Soil Science 41, 341(358.
4. Bengough, A.G., Mullins, C.E., Wilson, G., (1997) European Journal of Soil Science 48, 603–612.
5. Clark, L.J., Barraclough, P.B., (1999) Journal of Experimental Botany. 50, 1263–1266.
6. Comstock JP. 2002.. Journal of Experimental Botany 53, 195-200.
7. De Neve, S., Van De Steene, J., Hartmann, R. & Hofmann, G. (2000) European Journal of Soil Science 51: 295-304.
8. Dexter, A.R., Watts, C.W., 2001. In: Smith, K.A., Mullins, C.E., (Eds.), Soil Environmental Analysis. Marcel Dekker, pp. 405–433.
9. Dirksen C, Kool J B, Koorevaar P and van Genuchten M T (1993) In Memoriam Eshel Bresler (1930-1991). Eds D Russo and G Dagan. pp. 99–122. Springer, Berlin.
10. Farrell, D.A., Greacen, E.L., (1966) Australian Journal of Soil Research. 4, 1–17.
11. Feng M, Fredlund D G and Fangsheng S (2002) Journal of Geotechnical Testing 25, 303–314.
12. Genstat, 2002. Genstat for Windows. Release 6.2. Sixth Edition. VSN International Ltd., Oxford.
13. Hernanz, J.L., Peixoto, H., Cerisola, C., Sanchez-Geron, V., (2000) Journal of Terramechanics. 37, 167–184.
14. Hilhorst, M.A. (2000) Soil Science Society of America. 64: 1922-1925.
15. Hussain, A., Black, C.R., Taylor, I.B., Mulholland, B.J. & Roberts, J.A. (1999) Plant, Cell and Environment 22: 1377-1388.
16. Hussain, A., Black, C.R., Taylor, I.B. & Roberts, J.A. (2000). Plant, Cell and Environment 23: 1217-1226.
17. Knapp R T, Daily J W and Hammitt F G (1970) Cavitation. McGraw-Hill.
18. Kool J B and Parker J C (1987) Water Resource Research. 23, 105–114.
19. Lipiec, J., Kubota, T., Iwama, H. & Hirose, J. (1988) Soil Science and Plant Nutrition 34: 417-428.
20. MAFF Nutrition and Management of Arable Crops Roame A.
21. Masle J, Passioura JB. (1987) Australian Journal of Plant Physiology 14, 643-656.
22. Montagu KD, Conroy JP, Atwell BJ. (2001) Journal of Experimental Botany 52, 2127-2133.
23. Mulholland B.J., Black C.R., Taylor, I.B., Roberts J.A., & Lnetaon J.R. (1996) Journal of Experimental Botany 47, 539-539.
24. Mullins, C.E., Blackwell, P.S. & Tisdall, J. M. (1992) Soil and Tillage Research 25: 113-128.
25. Mullins, C.E., Cass, A., MacLeod, D.A., Hall, D.J.M. & Blackwell, P.S. (1992) Soil and Tillage Reseach 25: 129-147.
26. Mulqueen, J., Stafford, J.V., Tanner, D.W., 1977. Journal of Terramechanics. 14, 137–151.
27. Nissen, H.H., Moldrop, P. & Henriksen, K. (1998) Soil Science Society of America Journal 62: 99-109.
28. Sharp, R.E., Silk, W.K., Hsiao, T.C., 1988. Plant Physiology. 87, 50–57.
29. Take W A and Bolton M D 2003 Tensiometer saturation and the reliable measurement of soil suction. Géotechnique 53, 159–172.
30. Tardieu, F., Zhang, J., Katerji, N., Bethenod, O., Palmer, S. & Davies, W.J. (1992) Plant, Cell and Environment 15: 193-197.
31. van Genuchten M T (1980) Soil Science Society of America Journal 44, 892–898.
32. Vepraskas, M.J., (1984) Soil Science Society of America Journal 48, 1220–1225.
33. Whalley, W.R., Bengough, A.G., Dexter A.R., (1998) Journal of Experimental Botany 49, 1689–1694.
34. Whalley W.R., Watts C.W., Hilhorst M.A. Bird N.R.A., Balendonck J. and Longstaff D. J. (2001) European journal of Soil Science 53: 511-519.
35. Whalley W R, Cope R E, Nicholl C J and Whitmore A P (2004a). Soil Use Man. 20, 203–206.
36. Whalley W R, Leeds-Harrison P B, Clark L J and Gowing D J G (2004b) Soil and Tillage Research (in press).
37. Young F R (1989) Cavitation. McGraw-Hill: London.
| Please press enter |
-----------------------
• To move from one fill-in location (field) to another, press TAB, RETURN, UP or DOWN arrow keys unless instructed to do otherwise. Locations may also be selected by clicking on them.
Fig. 1.1. Schematic diagram of the porous-matrix sensor. The arrangement of the ceramic is shown in exploded view in Fig. 2.
Fig. 1.2. Components of the porous-matrix sensor, showing outer cylinder, ceramic, central electrode, adapter and dielectric probe. In use, the dielectric probe is connected to the ceramic assembly with a coaxial cable.
Fig. 1.3 Hysteresis models of Kool and Parker (1987) and Dirksen et al. (1993).
Fig 1.5. Comparison of the outputs of tensiometers (w-FT) and porous-matrix sensors (PMS) of matric potential of soil water with degree of saturation (S) measured by Profile Probes. Measurements were made at a depth of 0.20 m in four plots in the field. Plots 1 to 4 correspond to 0, 1, 4 or 8 passes of the tractor, respectively. Porous-matrix sensors were calibrated using the main drying curve.
Table 2.1 Particle size distributions, textural classification and soil organic carbon content of the six soils.
Soil |Percentage of |Textural Classification |Soil organic carbon % | | |Sand |Silt |Clay | | | |Boot Field |9 |25 |65 |Clay |7.2 | |Evesham |33 |22 |45 |Clay |5.4 | |Kirton |8 |73 |20 |Silty clay loam |2.1 | |North Wyke |56 |23 |22 |Sandy clay loam |3.1 | |Kingsmead |80 |11 |9 |Loamy sand |1.4 | |Wrest Park |80 |15 |5 |Loamy sand |1.4 | |
Fig. 2.3. The effect of using a fixed or rotating penetrometer on penetrometer pressure. The data points are for a fixed penetrometer, with linear regression (solid line), including data from Farrell and Greacen (1966) for a range of compressible soils. The curve fitted to the rotating penetrometer data in (dashed line) and the 1:1 line (dotted) are also plotted.
Fig. 2.2. Rotating penetrometer pressure for the nominal low density treatments in experiments 2 and 3 plotted against matric potential (A), effective stress (B) and degree of saturation (C). Solid lines are linear regressions, dotted lines are 1:1 lines.
................
................
In order to avoid copyright disputes, this page is only a partial summary.
To fulfill the demand for quickly locating and searching documents.
It is intelligent file search solution for home and business.
Related searches
- performance and development plan
- learning and development definition
- education and development pdf
- research and development job description
- research and development manager
- education research and development institute
- research and development manager jd
- pharmaceutical research and development jobs
- research and development job
- defence research and development organisation
- pharmaceutical research and development process
- research and development director role