Irrigation scheduling and irrigation systems

4 downloads 15481 Views 743KB Size Report
good when grown on a sub-irrigated sand bed; this system ... increasingly on automated time clock systems (Beeson ...... overhead, drip or capillary irrigation.
Irrig Sci (2009) 27:139–153 DOI 10.1007/s00271-008-0128-x

ORIGINAL PAPER

Irrigation scheduling and irrigation systems: optimising irrigation efficiency for container ornamental shrubs Olga M. Grant Æ Michael J. Davies Æ Helen Longbottom Æ Christopher J. Atkinson

Received: 18 March 2008 / Accepted: 22 July 2008 / Published online: 12 August 2008 Ó Springer-Verlag 2008

Abstract Water use and plant growth and quality were compared across different nursery stock beds, different methods of applying irrigation, and different methods of scheduling irrigation. With overhead irrigation, scheduling of irrigation according to plant demand, along with an irrigation system designed to maximise irrigation uniformity, resulted in substantial water savings, without reducing plant quality. This was the case in both wet and dry years. In the dry year, plant quality was particularly good when grown on a sub-irrigated sand bed; this system also used less water than any of the overhead irrigation systems. Two different systems were effective in scheduling overhead irrigation, one based on the volumetric moisture in the growing substrate, and the other based on plant evapotranspiration. The latter was determined with a small sensor with wet and dry artificial ‘‘leaves’’, the output of which correlated with that obtained following the Penman–Monteith method based on a full set of meteorological data.

Introduction Water scarcity and the increasing competition for water resources between agriculture and other sectors are forcing growers to consider more seriously the adoption of water

Communicated by E. Fereres. O. M. Grant (&)  M. J. Davies  H. Longbottom  C. J. Atkinson East Malling Research, New Road, East Malling, Kent ME19 6BJ, UK e-mail: [email protected]

saving strategies, especially in areas of intensive horticultural production and limited water resources (Costa et al. 2007). Water requirements of food crops are relatively well quantified in agronomic literature, but there has been little quantification of the irrigation requirement in order to maintain healthy growth and acceptable quality of ornamental plants (Henson et al. 2006). In several regions with container production, water is either limited, restricted, or both (Beeson 2006). Nursery production of woody landscape ornamentals has shifted in the past 40 years from in ground to container production for small to medium sized shrubs and trees. Whilst nurseries occupy relatively small land areas compared to agronomic crops, their consumption of water is quite high, ranging between 1.8 and 2.9 m depth over the crop annually in the southeastern US, for example (Beeson 2004). In areas such as Florida restrictions have been imposed on the quantity of irrigation that may be used (2.3 m depth in 1992, reduced to 1.8 m depth where there is a strong competition with urban centres for drinking water), meaning that nurseries need to increase their irrigation efficiency in order to remain in their current profitable locations (Beeson and Brooks 2007). In the UK, awareness that water supplies may be more limited in future has been heightened by recent water shortages during the summer, particularly in the south-east of England. Nonetheless, 90% of smaller nurseries are dependent on mains water, few recycle water, and many use inefficient irrigation systems (Briercliffe et al. 2000). In addition to the problems associated with depletion of water resources, losses of nutrients from hardy nursery stock systems can represent a considerable eutrophication potential (Harris et al. 1997). To maximise growth, commercial nurseries generally strive to maintain plant available water to near 100% of container capacity (Beeson 2006). In large containers, this

123

140

is readily accomplished with micro-irrigation systems with minimum wastage of water. However, most smaller containers are irrigated with overhead sprinkler systems, and this is likely to continue (Briercliffe et al. 2000). Usually only 20–40% of the water applied through these systems is retained in the containers (Beeson and Knox 1991). Thus 60–80% of overhead irrigated water is of almost no value for plant growth. Overhead application efficiency is influenced by uniformity of irrigation application, application rate, total irrigation volume, sprinkler and canopy characteristics, and container size and spacing configurations (Beeson and Yeager 2003). The frequency and quantity of irrigation applied on most commercial nurseries is based on personal experience, or increasingly on automated time clock systems (Beeson 2004). Whilst plant water use can vary greatly from day to day, through the course of a season, and with growth, seldom does the grower respond to such changes by modifying irrigation. Hence under or over irrigation occurs more frequently than correct supply to the plants’ needs. The scientific tools needed to schedule irrigation are well developed (Stirzaker 2005). One method of scheduling irrigation is based on evaporative demand. This depends on the weather conditions where the plants are grown and in addition on plant spacing etc. For nursery stock production, the actual evapotranspiration of the crop (ETA) can be calculated from evapotranspiration from a short grass crop reference (ETo) and a crop coefficient (Kc) that defines the relationship between the two: ETA = ETo 9 Kc. However, Kc values are not generally known for ornamental plants, on account of the hundreds of species and thousands of cultivars in production (Beeson 2005). An alternative is to base scheduling on the moisture content of the growing substrate. Tensiometers have now been used for decades for monitoring soil water status (Stirzaker 2005) and can be applied to container plants (Lloyd et al. 2006). More recently a range of ‘‘user-friendly’’ capacitance devices have come on the market, based on the measurement of the dielectric property of soil (Charlesworth 2005). Several methods of irrigation scheduling based on plant responses to drying substrates also exist (Jones 2004) and can be applied to plants in containers (e.g. Grant et al. 2006). Unfortunately, the tools for irrigation scheduling have not been taken on board or are not systematically used by the majority of growers, perhaps because they are not perceived to work reliably in real conditions, or because growers cannot justify the time and expense associated with collecting, interpreting, and implementing the information that they provide (Stirzaker 2003). Over the last two decades the application of deficit irrigation regimes, where plants are provided with less than 100% of the water lost in evapotranspiration, have proved effective in saving water without severe consequences for

123

Irrig Sci (2009) 27:139–153

yield (dos Santos et al. 2003; Grant et al. 2004). Indeed, in some crops such as grapevine, whilst yield may be reduced by deficit irrigation, quality can be improved. Whilst this research has focussed on food crops, the technique can also be useful for controlling growth in ornamentals (Cameron et al. 2006). However, before any deficit irrigation regime is applied on nurseries, it is essential that irrigation uniformity is optimised, that the water use of container crops is well understood, and that the effective methods exist for determining irrigation requirements. The aims of the work described here were to clarify variation in water use across different nursery stock beds, different irrigation systems, and using different methods of scheduling irrigation. It was determined whether variation in water use, and uniformity of water application and substrate moisture content, influenced plant growth and quality. Such evaluation is necessary in order to provide information which can assist growers in making sound investments when upgrading or expanding water management systems in their production areas, and to bring nursery stock production inline with other crop types for which substantial progress is currently being made in modelling water use requirements and efficiently scheduling irrigation.

Materials and methods Irrigation A control ‘‘typical’’ nursery stock irrigation system, with poor uniformity of irrigation application and no adjustment of the quantity of irrigation applied in response to plant demand for water, was compared with five ‘‘improved’’ irrigation systems (Table 1). The experimental set-up consisted of six 10 m 9 5 m hardy nursery stock beds: four gravel beds (irrigation systems 2–5), a Mypex bed (irrigation system 1), and a sand bed (irrigation system 6). Each of the beds was lined with polythene. The sand bed was sub-irrigated. Three overhead sprinkler irrigation systems were compared. Two of these were conventional overhead systems: a traditional system with two 360° rotating sprinklers (Rotoframes) spaced 5 m apart on the bed and at 1 m height (irrigation system 5), and an improved uniformity overhead sprinkler system with rotator nozzles (MP2000, Walla Walla Sprinkler Company, Walla Walla, WA, USA) spraying inward onto the bed from a height of 1.2 m and spaced at 2.5 m between sprinklers so as to maximise the uniformity of irrigation on the bed, with the arc of the sprinkler projector set at 210° except on the corners, where the arc was set to 100° (irrigation systems 1–3). The third was a gantry overhead irrigation system (in the second year of the experiment,

Irrig Sci (2009) 27:139–153 Table 1 Nursery bed type, method of applying irrigation, and method of scheduling used in six irrigation systems

141

Irrigation system

Bed

Application of irrigation

Scheduling

1

Mypex

Improved uniformity overhead

Evaposensor

2

Gravel

Improved uniformity overhead

Evaposensor

3

Gravel

Improved uniformity overhead

Soil moisture sensor

4

Gravel

Gantry overhead

Evaposensor

5

Gravel

Poor uniformity overhead

None (timer only)

6

Sand

Sub-irrigated

Not required

irrigation system 4). To determine the application rate and uniformity of distribution under the different sprinklers, catch pots were spaced on each of the beds at 1 m interval to collect water during 5 min of irrigation. The captured water in each pot was measured to calculate the mean application rate of water onto the bed (MAR, mm irrigation water going onto the bed per hour of irrigation) and coefficients of uniformity [Christiansen’s coefficient of uniformity (CU): the absolute average difference between the catch in each saucer and the average catch, as a percentage of the average catch] and scheduling (scheduling coefficient = MAR/lowest catch rate). This was repeated on a number of different dates, including still and windy days to determine the impact of wind speed on uniformity. In addition, run-through was determined by placing the pots in larger containers with plastic liners, and measuring water delivery to the plant pot (weight gain) and the quantity of water running through the container below. This was carried out for each overhead system at a range of run lengths (or speeds in the case of the gantry) corresponding to typical run lengths/speeds used during the summer. Plant material Cistus creticus, Potentilla fruticosa ‘‘Tangerine’’, and Spiraea nipponica ‘‘Snowmound’’ were purchased as liners from New Place Nurseries (Pulborough, UK) in June 2006 and potted up into 3 L pots in a 100% peat substrate and placed on the experimental beds. Four blocks of each species were laid out on each bed, each block consisting of 45 plants, arranged in five columns of nine plants, with columns staggered and 3 cm spaces between pots. Blocks were arranged in three columns of four; the location of each species was randomly selected (Fig. 1). Cistus 9 pulverulentus ‘‘Sunset’’, Potentilla fruticosa ‘‘Tangerine’’, and Spiraea japonica ‘‘Shirobana’’ were placed on the beds in April 2007, using the same design as in 2006. Within each block, ten plants were randomly selected for measurements. In 2006, heights and two widths in perpendicular directions were measured when the plants were placed on the beds and again in September and in November. Mean projected canopy area was calculated as

Treatment 2 P1 C1 C2 S1

C3 S2 P2 S3

Treatment 1

C4 P3 P4 S4

C1 C2 S1 S2

P3 P4 S4 C4

P1 C1 C2 S1

Treatment 3 S1 P1 C1 S2

C2 C3 P2 S3

C3 S3 P1 P2

S4 P3 P4 C4

Treatment 4 C3 S2 P2 P3

S3 C4 P4 S4

Treatment 6 S1 P1 C1 P2

S2 C2 P3 S3

C3 P4 C4 S4

Treatment 5 C1 C2 S1 P1

P2 P3 S2 C3

C4 P4 S3 S4

N

Fig. 1 The layout of the experimental treatments and blocks of each species within those treatments. C, P, and S represent blocks of Cistus, Potentilla, and Spiraea plants, respectively. Numbers 1–3 indicate the number of the block of that species within a treatment. Each block consisted of 45 plants

the multiple of the two measured plant widths, following Beeson (2004). Percentage canopy closure was calculated as % canopy closure ¼ 100  mean projected canopy area= plant space allocation ð1Þ where plant space allocation is defined as the square of the distance between plant centres (Beeson 2004). Shoot and root quality were assessed in November. In March 2007 (corresponding to the usual marketing date for these shrubs) a rapid assessment (one measurement per block) of shoot quality was undertaken. In 2007, plant height, width, and quality of the shoots and roots were assessed from 20th to 24th August. For quality, plants were classified into three groups, class 1–3. For shoot quality, class 1 plants were those with good coverage of the pot, even heights of stems, not too many gaps when looked at from the side, and healthy foliage. Root systems were considered to have class 1 quality where roots could be seen over the whole of the substrate sides, were evenly distributed, and the roots reached the base of the pot. In both cases a class 2 system was one in which one of the requirements was not met, and a class 3 system was one in which two requirements were not met. Substrate moisture was measured at about 6 cm depth in the pots of each of the selected plants monthly

123

142

from July to September in both 2006 and 2007. At the end of the season in 2007 water extracts of approximately 1 L of substrate sampled from each of two pots per species per bed were analysed for pH and electrical conductivity (EC) with a pH and EC metre, ammonia and nitrate concentration with by colorimetry, and phosphorus, potassium, and magnesium concentration by ICP spectroscopy (Dove Associates, Diss, Norfolk, UK). Scheduling In the control irrigation system, irrigation was not adjusted to daily fluctuations in the demand of the plants for water. Rather, a timer (Galcon 6000 DC1, Galcon Irrigation Control Equipment, Upper Galilee, Israel) was set to run irrigation four times daily and the length of irrigation runs was adjusted only very occasionally to allow for plant growth, or periods of excessive rainfall. On the sub-irrigated sand bed, irrigation adjusts automatically to plant demand—therefore no intervention was necessary. With irrigation systems 1, 2, and 4, application of water was matched to evaporative demand. To achieve this, an ‘‘Evaposensor’’ (Skye Instruments Ltd., Powys, UK) was used, which has two temperature sensors, one dry and one kept wet with a wick in a small reservoir of deionised water. This sensor is attached to a meter (the ‘‘Evapometer’’) which records the difference in temperature between the two sensors and records ‘‘degree hours’’, where 1 degree hour is a difference of 1°C between the wet and the dry sensors over a period of 1 h. A prototype system has previously been described by Harrison-Murray (1991a, b). The accumulated number of degree hours over a day was monitored. Two plants per block, i.e. eight plants per species on either bed were weighed after they had been well irrigated and excess water had drained through, and a day later, to calculate how much water they used per degree hour. The pots were placed in saucers for these measurements, and any water draining into the saucer was subtracted from the calculated water loss, to ensure that only changes in mass due to evapotranspiration were used in the calibration. This calibration was repeated at intervals as the plants grew. To determine how much water the plants received in a given length of irrigation, in 2006 initially catch values were used, but later plants were weighed before and after irrigation to determine how much water plants took up during an irrigation event. For the gantry overhead irrigation, this was determined at a number of different speeds to construct a relationship between gantry speed setting and water delivery. From this, the speeds necessary to replace different numbers of degree hours of evapotranspiration were determined. Each day, the correct speed was selected to replace the amount of water lost from the plants the previous day. For all three

123

Irrig Sci (2009) 27:139–153

Evaposensor-scheduled beds, the irrigation was split into four events per day. In 2007, rainfall was monitored daily and based on the size of the pots it was estimated that 1 mm rainfall resulted in approximately 107 mL of water being delivered to an individual plant pot. Using the plant calibrations this was converted to number of degree hours of evapotranspiration replaced by 1 mm rainfall, and this figure was subtracted daily from the total degree hours before calculating the quantity of irrigation required to replace the previous day’s evapotranspiration. Where it was estimated that rainfall had met or exceeded evapotranspiration, no irrigation was applied. On another bed (irrigation system 3) a substrate moisture sensor (SM200, Delta-T Devices Ltd, Cambridge, UK) in the substrate in one of the pots measured the volumetric moisture content. This sensor was connected to a small logger (GP1, Delta-T Devices Ltd) which both monitored the substrate moisture over time and controlled whether or not the irrigation turned on. If the substrate was sufficiently wet, the irrigation stayed off. If the substrate moisture fell below a determined value, then the irrigation turned on. In 2006, 35% moisture (by volume) and 50% moisture were chosen as the lower and upper set-points. The GP1 was used with a battery-operated timer which was set to turn on four times a day for 15 min each time. When the substrate moisture fell below 35%, the irrigation turned on for 15 min. This occurred the next time the irrigation was set to turn on, if the substrate moisture had not reached the upper set value, but once the upper set value was reached then the GP1 overrode the signal from the solenoid timer and the irrigation stayed off, until the lower level was reached. In 2007, 30 and 45% were used as the upper and lower limits, respectively. In the case of both the Evaposensor and soil moisture sensor, irrigation was scheduled to suit the average plant on the bed, apart from late in the season of 2007 when Potentilla used more water than the other species, and due to its sensitivity to water deficits, scheduling was undertaken to meet the demands of Potentilla rather than the average species. A second sensor was used to monitor substrate moisture content in a pot of the same species but in a different location on the bed. Other soil sensors (Thetaprobes, Delta-T Devices Ltd) were used in 2006 to monitor the moisture in one pot of each species on this bed. The quantity of water applied to each bed was monitored with a water metre connected to a data-logger (Delta-T Devices Ltd). Total depth of water received over a canopy (irrigation + rainfall) was divided by the area of a container to estimate the volume of water received by a container. Total evapotranspiration from a container was estimated from measured evapotranspiration (from calibrations of the Evaposensor) and the total number of degree hours measured, whilst each calibration factor was in use. The percentage of water received by a container that

Irrig Sci (2009) 27:139–153

143

was evapotranspired was calculated from the above two estimates. Meteorological conditions were recorded at the East Malling Meteorological Station, a standard Meteorological Office station, located approximately 200 m from the experimental nursery beds. Evapotranspiration for a grass reference (ETo) was calculated from these data following the ASCE Penman–Monteith equation (Allen et al. 1989) using REF-ET software (Allen 2000). Water application to each bed was monitored with logged water meters (WMR, Arad Ltd, Israel). As the beds were sealed, all runoff was forced through single exit pipes. For systems 1 and 2, this was measured using logged and calibrated tipping buckets. Statistical analysis Differences between irrigation systems were tested using analysis of variance. Treatment and species were orthogonal Table 2 The best recorded irrigation performance of three overhead irrigation systems Performance

Irrigation system 1–3

4

5

MAR (mm/h)

19.9

NA

14.4

CU (%)

83

95

64

SC (%)

1.6

1.2

3.0

SC5% (%)

1.5

1.1

2.2

MAR is mean application rate, CU is Christiansen’s Coefficient of Uniformity, SC is the scheduling coefficient, and SC5% is the scheduling coefficient when the MAR is divided by the 5th percentile value rather than the lowest catch

factors and where there was more than one measurement per block, block was nested within species. The relationship between ETo and accumulated degree hours was tested by correlation analysis. Statistical analyses were performed using Genstat software (9.1, Rothamsted Experimental Station, UK).

Results Irrigation The ‘‘improved’’ sprinkler overhead irrigation systems (systems 1–3) resulted in greater uniformity of irrigation (CU) than the control system (system 5) (Table 2). Correspondingly, scheduling coefficients (SC) were higher for system 5. CU was highest, and scheduling coefficients lowest, for the gantry overhead irrigation system (system 4). The distribution of irrigation over a bed varied greatly according to how still or windy the weather (e.g. reducing the CU for the ‘‘improved’’ sprinkler system from 83 to only 52% on one occasion), as the experiment was performed on an exposed site. Pots on Mypex increased in mass during a given length of irrigation run by about twice as much as plants on gravel under a comparable system after irrigation. The experimental season in 2007 was on average cooler and slightly more humid than that in 2006 (Fig. 2a, b). The experiment ran for only 24 days longer in 2007, but the total rainfall during the 2007 season was considerably higher (366 mm) than during the 2006 season (241 mm). Little rainfall occurred early on in the experiment in 2006,

Fig. 2 Mean daily air temperature (Tair) and relative humidity (RH) (a, b) and total daily precipitation (c, d) measured at the experimental site in 2006 (a, c) and 2007 (b, d)

123

144

Irrig Sci (2009) 27:139–153

whereas in 2007 the driest period was quite late in the experiment, from late August to mid-September (Fig. 2c, d). Over the experimental period of 2006, rainfall occurred on 74 days, but quantities were generally low, with more than 5 mm rainfall only occurring on 15 of these dates. In 2007, rainfall occurred on 82 days during the experiment, with more than 5 mm rainfall on 21 of these dates. Air temperature and relative humidity varied over the season in both years, and, as a result, daily accumulated total degree hours varied considerably between dates. Accumulated degree hours read from the Evaposensor correlated well with ETo calculated from meteorological data using the method described by Allen et al. (1989) (Fig. 3). The control ‘‘typical’’ nursery irrigation system (system 5) used 2.3–3.2 times as much irrigation as the ‘‘improved’’ systems over the course of the experiment in 2006 and 2.6– 4.2 times as much over the longer experiment in 2007 (Fig. 4). Both years, irrigation system 2 was the next greatest user of irrigation water. In 2007, this system used 157 mm more water than system 4. In 2006, 1.4 times as much irrigation was applied with system 2 as with system 3 and in 2007, 1.2 times as much irrigation was applied with

system 2 as with system 3. Total irrigation application was reduced for system 1 in 2007 compared to 2006. This is because surface uptake from the Mypex bed was taken into account in irrigation scheduling throughout the experiment in 2007. The reduced application of irrigation water to the Mypex bed compared to system 2 (same scheduling method and sprinklers but a different bed surface) represents 39% savings of irrigation water. Uptake of water from the Mypex surface also meant that runoff was reduced with this system: runoff from system 1 (0.8 L/pot/week on average) over the season of 2007 was only 42% of that from system 2 (1.9 L/pot/week on average). In 2007, system 1 used only 78 mm more irrigation water than system 6, the system which used the least water both years. For the overhead irrigated plants, irrigation regime 3 involved far fewer irrigation events than the other regimes (Table 3). Hence the average application per irrigation event was far higher for this irrigation system than for the other scheduled overhead systems, and slightly higher than for the control, unscheduled, overhead system: irrigation runs for 15 min were used with irrigation system 3, compared to on average 4 min with irrigation system 1 during 2007. On average, 3% of irrigation ran through pots during

(a) (a)

(b) (b)

Fig. 3 Accumulated degree hours as measured with the Evaposensor and ETo estimated from meteorological data using the Penman– Monteith equation in 2006 (a) and 2007 (b). Spearman correlation coefficients and probabilities are displayed on either graph; for a df = 97, t = 10.59, and for b df = 148 and t = 17.18

123

Fig. 4 Depth of total irrigation and total irrigation plus rainfall falling above a plant canopy for each irrigation system during the experiment in 2006 (a) and 2007 (b). Each bar represents a single bed

Irrig Sci (2009) 27:139–153

145

Table 3 Number of irrigation events and average depth of irrigation per event (mm) over the experimental periods of 2006 and 2007 for each overhead irrigation regime Year Irrigation regime

2006 No. of events

2006 Depth/event (mm)

2007 No. of events

2007 Depth/event (mm)

1

456

1.9

456

1.6

2

458

2.3

445

2.4

3 4

117 NA

6.5 NA

136 452

6.5 2.0

5

503

4.8

532

5.2

Maximum possible number of irrigation events would have been 504 in 2006 and 600 in 2007

15 min irrigation runs, compared to only 0.7% during 8 min irrigation runs, and no run-through occurred during 3 min irrigation runs. Although relatively little irrigation water was used per event with the gantry (irrigation system 4), this will have represented quite a high rate of application above a given plant: approximately 5 mm per minute, compared to 0.4 mm per minute for systems 1–3. Hourly data indicated that irrigation system 6 used water in small quantities but almost continuously. For the overhead irrigated beds, the percentage of the estimated quantity of water received by a container (irrigation + rainfall) that was evapotranspired was highest for irrigation regimes 1 and lowest for regime 5 (Table 4). The percentage evapotranspired was also quite low for regime 2. These calculations are based on estimates of the depth of water received over the area of a pot. However, our measurements of plant water uptake during irrigation events showed that uptake was far lower than this on gravel beds. This is to be expected, as the canopy of the plants funnels water away from the pot and onto gravel, which runs through and is lost to the plant. Only about 58 and 62% of the water falling over the area of the pot was actually taken up by the plant with regimes 2 and 3, respectively, based on the average plant—this however varied between species, with about 90% of the water falling over Potentilla on gravel beds being taken up by the plant. In 2007, total evapotranspiration was greatest for Potentilla and least for Spiraea. As a result of taking up only about 60% of the water falling on it when placed on a gravel bed, and evapotranspiring less than the other species on the same beds, the percentage of water falling on a plant that was evapotranspired was least for Spiraea (72% with irrigation regime 1 and 48% with regime 2). Volumetric substrate moisture content varied within and between treatments and over time (Table 5 and examples in Fig. 5). In 2007, a very high percentage of plants irrigated with system 6 had quite dry substrates, often with lower than 20% volumetric substrate moisture contents (Fig. 5e, f). Substrate moisture content in pots under system 3 was generally between 35 and 55% in July (Fig 5c), but had fallen considerably for a number of pots on this bed

Table 4 Estimated percentage of irrigation falling over the area of a pot that was used in evapotranspiraton in 2006 and in 2007 Year Irrigation regime

2006 %

2007 %

1

72

83

2

46

58

3 4

59 NA

65 65

5

22

26

by August (Fig. 5d). Substrate moisture under system 1 was relatively high, compared to systems 3 and 6, in both months (Fig. 5a, b). Substrate sampled at the end of the summer of 2007 was found to differ significantly in electrical conductivity according to the irrigation system (Table 5), with higher electrical conductivity in substrate from pots irrigated with system 6 (345 ± 39 lS cm-1, n = 6) than with any other system (242 ± 32 lS cm-1, n = 30). There was also a significant difference between irrigation systems in the concentration of magnesium in the substrate (Table 5) with higher concentrations in substrate from pots irrigated with system 6 (78.8 ± 17.1 mg L-1) than with the other systems (40.8 ± 8.8 mg L-1). Concentrations of ammonium, nitrate, phosphorus, and potassium did not significantly differ between irrigation systems. pH differed significantly according to the system (Table 5), being less alkaline in substrate from pots irrigated with systems 1 and 6 (pH \ 5.5) than with the other systems (pH [ 5.7). Both years, plants grew substantially over the season on all beds. In 2006, by September the Cistus were tallest under irrigation systems 1 and 6 (Table 5 and Fig. 6a). Potentilla were shortest under irrigation system 2 (Fig. 6b). By November, plants of all three species were tallest with irrigation system 6 (Fig. 6). In 2007, significant differences in plant sizes (both height and width) were again found between beds (Table 5), when measured at the end of August (Fig. 7). Cistus were tallest and widest with irrigation systems 1 and 6 and shortest with irrigation system 5

123

146

Irrig Sci (2009) 27:139–153

Table 5 Results of analysis of variance for various plant and substrate properties Property

Crop

Month

Treatment

Species

Treatment 9 species

P

LSD

P

LSD

July

\0.001

0.02

\0.001

0.01

August

\0.001

0.02

NS

September

\0.001

0.02

\0.001

July

\0.001

0.02

0.003

August

\0.001

0.03

\0.001

P

LSD

Substrate properties Moisture content

2006

2007

\0.001

0.028

\0.001

0.036

0.02

\0.001

0.041

0.01

\0.001

0.036

0.02

\0.001

0.045 130.7 36.3

EC [Mg]

2007 2007

August August

0.039 0.004

75.5 20.9

NS 0.018

14.8

0.044 0.007

pH

2007

August

\0.001

0.32

0.03

0.23

NS

2006

September

\0.001

1.26

\0.001

0.98

\0.001

2.19

November

\0.001

1.74

NS

1.35

0.016

3.02

2007

August

\0.001

1.55

\0.001

1.10

\0.001

2.69

Plant width

2007

August

\0.001

1.55

\0.001

1.09

\0.001

2.68

Shoot quality

2006

November

\0.001

0.14

0.036

0.11

\0.001

0.24

March

\0.001

0.35

NS

0.035

0.60

2007

August

\0.001

0.15

\0.001

\0.001

0.26

2006

November

\0.001

0.13

NS

\0.001

0.23

2007

August

\0.001

0.13

0.005

0.09

0.003

0.23

Plant properties Plant height

Root quality

0.11

Shoot mass

2006

March

\0.001

2.89

\0.001

2.24

\0.001

5.00

Root mass

2006

March

\0.001

2.24

\0.001

1.73

\0.001

3.87

Root:shoot mass

2006

\0.001

0.04

0.036

0.08

NS

Crop refers to the year the crop was placed on the experimental beds. Degrees of freedom for the treatment effect are 4 and 5, and for the treatment 9 species interaction are 8 and 10, for the 2006 and 2007 crops, respectively. Degrees of freedom for the species effect are 2 throughout. LSD was tested where analysis of variance showed significant effects at P \ 0.05. February and March data refer to the 2006 crop, measured the following spring NS not significant, EC electrical conductivity

(Fig. 7a, b). Potentilla were tallest and widest with irrigation system 6 (Fig. 7c, d). Differences between beds were less marked for Spiraea, but they tended to be taller and wider with irrigation systems 1 and 6 (Fig. 7e, f). Overall, then, there was a tendency for greater growth with irrigation systems 6 and in some cases with system 1. In 2007, growth of Cistus and Potentilla appeared limited with system 5 (Fig. 7a, c, d). Differences in plant widths led to substantial variation between species and irrigation systems in percentage closure, with 888 ± 26% closure of Potentilla under system 6 by late August 2007, compared to 542 ± 37% closure with system 5, and only 300 ± 20% closure for Spiraea under system 3. In November 2006, there were significant differences between irrigation systems in shoot and root quality, but also significant interactions between irrigation systems and species (Table 5). Better shoot quality was found with irrigation systems 3 and 6 than with the other systems. For Cistus and Spiraea no plants on these beds fell into the poor quality category (class 3) (Fig. 8a, g) and for Cistus

123

in particular, a high proportion (88%) of plants irrigated with these systems fell into the best quality category (class 1). A relatively high proportion of Potentilla irrigated with system 5 (Fig. 8d) fell into class 3. A rapid survey of overall shoot quality in each block of plants late in March, when Spiraea was starting to produce new leaves, indicated little difference in quality between beds, with the exception of some poorer quality plants with irrigation system 5. In November 2006, the best root quality was found with irrigation system 6 (data not shown). In 2007, there were again significant differences between beds in both shoot and root quality (Table 5), this time measured late August. However, patterns of quality across beds were very dependent on species. For Cistus, the best shoot quality was found with irrigation system 5, and the worst with irrigation systems 1 and 4 (Fig. 8b). For Potentilla, the best shoot quality was found with irrigation system 6, with 75% of plants on that bed falling into the best quality category (Fig. 8e) and the worst with irrigation system 1. For

Irrig Sci (2009) 27:139–153

147

(a)

(a)

(b)

(b)

(c)

(d)

(e)

(f)

(c)

n

Fig. 5 Percentage of pots under each irrigation system with volumetric substrate moisture content falling between 5% categories from 5 to 90%, measured on 24th July (a, c, e) and 30th August (b, d, f) 2007, for irrigation systems 1 (a, b), 3 (c, d) and 6 (e, f). Approximately 120 pots were measured per bed (40 plants per species)

Spiraea, shoot quality was best for irrigation systems 3 and 6, and worst with system 2 (Fig. 8h). The best root quality for Cistus was found for systems 2 and 5 (Fig. 8c). The best root quality for Potentilla, however, was with irrigation system 4 (Fig. 8f). However, ‘‘best’’ and ‘‘worst’’ are perhaps misleading terms here, as in fact very few plants of any species on any bed fell into the poor quality category (Fig. 8c, f, i). Significant differences between irrigation systems occurred in both root and shoot mass of plants under the different systems during the summer of 2006 (harvested in spring 2007). Root and shoot dry mass both differed significantly between species, and a significant interaction of irrigation system and species was found (Table 5; Fig. 9),

Fig. 6 Heights measured on 11th September and 14th November 2006 of Cistus (a), Potentilla (b), and Spiraea (c). Bars represent mean ± SE, n = 4 blocks. Significant differences between irrigation systems (P \ 0.05) determined from LSD tests following ANOVA are indicated by different letters

with a trend for relatively large root and shoot dry mass with irrigation system 6 being particularly marked for Cistus (Fig. 9a). The ratio of root to shoot dry mass was affected both by species and its interaction with treatment (Table 5).

Discussion The lower uniformity of irrigation and higher scheduling coefficients for the control irrigation system (system 5) were to be expected given that under system 5 some areas of the bed will receive twice as much irrigation as others, on account of overlap on the center of the bed between the two sprinklers. High uniformity is apparently an advantage

123

148 Fig. 7 Plant heights (a, c, e) and widths (b, d, f) measured 22nd August 2007, of Cistus (a, b), Potentilla (c, d), and Spiraea (e, f). Bars represent mean ± SE, n = 4 blocks. Significant differences between irrigation systems (P \ 0.05) determined from LSD tests following ANOVA are indicated by different letters

Irrig Sci (2009) 27:139–153

(a)

(b)

(c)

(d)

(e)

(f)

of using the gantry overhead system, and uniformity with this system was less influenced by wind than under the conventional overhead irrigation systems. This improvement in uniformity may not, however, always justify the significant capital expenditure associated with purchase of a gantry system. The close agreement between accumulated degree hours read from the Evaposensor and ETo determined from meteorological data using the Penman–Monteith method is encouraging for the use of Evaposensors or similar small,

123

portable, and easily read sensors for scheduling of commercial nursery stock irrigation. The accuracy and application of the Evaposensor has not previously been evaluated. Daily adjustment of irrigation is much more practical where a measure of ETo can easily be assessed on a daily basis, such as with an Evaposensor, than where a meteorological station needs to be downloaded and data analysed prior to determining the correct adjustment. The Evaposensor therefore has clear advantages for daily scheduling. Additionally, automation should be possible

Irrig Sci (2009) 27:139–153 Fig. 8 Percentage of plants under each irrigation system falling into each of three quality categories with respect to shoots in November 2006 (a, d, g) and August 2007 (b, e, h), and with respect to roots in August 2007 (c, f, i) for Cistus (a–c), Potentilla (d–f), and Spiraea (g–i)

149

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

(i)

(Henson et al. 2006). A problem associated with the Evaposensor or with any measurement of ETo is that calibration is required to convert from potential evapotranspiration to actual crop evapotranspiration. In this study pots were weighed after irrigation and a day later and water loss was compared with the accumulation of degree hours during that time. This had to be repeated at intervals as the plants grew, and it is this need for repeated calibrations as well as potentially the need for separate calibrations for each species/variety grown on a nursery that is off-putting for the growers. The difficulty of obtaining calibration coefficients for ornamentals has been described by Irmak (2005), but for individual varieties models based on weeks after transplanting, growth index, cumulative reference evapotranspiration, or fraction of thermal units (a measure of cumulative temperature during the growth of the crop) have been described. Beeson (2004) suggested that projected canopy area or percentage canopy closure could be used to calculate calibration coefficients for use with ETo. In 2006, with irrigation systems 1 and 2 Cistus were found to use the least water over the season, and also had the smallest projected canopy areas and relatively low % canopy closure. In 2007, Spiraea was found to use the least water and showed the lowest canopy closure with those systems, though this was not much lower than for Cistus. Much further work, however, would be necessary to validate the use of such potential calibration factors with these and a wide range of other species.

Scheduling irrigation for different species with different water requirements but standing on the same beds is problematic as they have different rates of water use. If calibration coefficients can be established for different crop types, these could be used to determine which crops are placed together, minimising the risk of over or underwatering some crops, whether either evapotranspiration or substrate moisture content are used as the method of scheduling irrigation. However, since differences between species also occur in uptake of water, due to funnelling of water or the canopy throwing water away from the pot, grouping crops according to water use alone may not be sufficient to ensure efficient irrigation of all crops on a bed. Differing patterns of plant growth across the different irrigation systems according to species reflects that a given species may have received optimal irrigation with one system but a different species may have received optimal irrigation with another system. However, ‘‘optimal’’ irrigation is perhaps better deduced from plant quality rather than plant size. Uptake of water from Mypex led to a reduction in the amount of water that needed to be applied with system 1, but under rainy conditions leaving pots standing on wet Mypex can be deleterious to the roots. In general, the Efford sand-bed stood out in terms of plant quality in 2006, producing high-quality shoots and roots. Dry mass of both shoots and roots was also highest on this bed, indicating good growth, but whilst substantial root systems can aid garden establishment, long roots growing

123

150

Irrig Sci (2009) 27:139–153

(a)

(b)

(c)

Fig. 9 Dry mass of roots and shoots of Cistus (a), Potentilla (b), and Spiraea (c) measured in March 2007, after plants had been exposed to different irrigation systems through the summer of 2006. The plants remained on the same beds through the winter of 2006, but received only rainfall during this time. Bars represent mean ± SE, n = 4 blocks. Significant differences between irrigation systems (P \ 0.05) determined from LSD tests following ANOVA are indicated by different letters

out of the pots and into the sand may be lost taking the plants off the bed or in transit, so are not necessarily advantageous. Plants grew more on the Efford sand bed in 2007, but did not show higher quality. Quality has a larger influence on woody ornamental sales than plant size, especially in the landscape trade (Beeson and Brooks 2007). Two very different seasons, meteorologically, provided a good opportunity to determine how the different irrigation systems perform in both wet and dry years. It is clear that even in a wet year the control ‘‘typical’’ nursery irrigation system used far more water than any of the other systems. This partly relates to the lower efficiency of

123

irrigation application, and partly to the fact that scheduling of irrigation was not applied. Beeson and Brooks (2007) and Stanley and Harbaugh (1988) have also shown that water savings are possible when scheduling container plants based on crop evapotranspiration rather than a set irrigation quantity as applied commercially. In Beeson and Brooks’ (2007) study the reduction related partly to the crop being marketable earlier, partly due to reduced irrigation frequency, and partly due to accounting for effective rainfall. The gantry overhead system used less water than a comparable conventional overhead irrigation system (system 2, with the same bed surface and same scheduling method). The narrowing of the difference in water application between systems 2 and 3 in 2007 may be surprising given that lower thresholds were used with system 3 in 2007, causing irrigation to turn off at lower volumetric substrate moisture content than in the preceding year, but on the other hand taking rainfall into account with system 2 in 2007 resulted in more efficiency of irrigation application with that system than in 2006. As a result, only 23 mm more irrigation was applied with system 2 in 2007 compared to 2006, despite a longer experiment in 2007. Total irrigation application was actually reduced for system 1 in 2007 compared to 2006. This is because surface uptake from the Mypex bed was taken into account in irrigation scheduling throughout the experiment in 2007. The reduced application of irrigation water to the Mypex bed compared to system 2 (same scheduling method and sprinklers but a different bed surface) represents 39% savings of irrigation water. Runoff was much reduced from the Mypex compared to gravel bed, but even the runoff from the gravel bed is low compared to values of 3 L/pot/ week for overhead irrigation of plants in similarly sized pots and at similar spacing in Goodwin et al.’s (2003) study. This may indicate an advantage of the daily adjusted scheduling used with systems 1 and 2 in this study, compared to a standard daily quantity of irrigation in Goodwin et al.’s experiment, though it is likely that reduced runoff in our study relates more to the lower vapour pressure deficits in south-east England compared to New South Wales. The reduced runoff from irrigation system 1 compared to system 2 may mean a reduced impact on the environment in the form of nutrient pollution. System 6, the sub-irrigated sand bed, should lead to the least environmental contamination, due to recirculation of water (Harris et al. (1997) found that sub-surface irrigation per se did not reduce losses of pesticides or nutrients). The sub-irrigated sand bed system is clearly efficient in terms of water use, but as with the gantry, questions arise as to whether the reduction in water use with such a system justifies the initial expenditure: sub-irrigated sand beds need to be perfectly level, requiring more labour cost for installation than the other systems. The relatively long irrigation runs used with

Irrig Sci (2009) 27:139–153

irrigation system 3 may have resulted in some wastage of water via run-through. Wastage via run-through is also possible with the gantry, but the alternative, to run the gantry more frequently but at higher speeds, may be more wasteful in energy terms and less practical on large scale nurseries. Irrigation under several of the systems was divided up into four applications per day to reduce the problem of run-through. However, Beeson (2004) suggested that the reduced frequency of irrigation reduces evaporation from a substrate surface and promotes root growth deeper in the substrate, thus advocating longer, but less frequent irrigation runs. The relative loss of water via evaporation or via run-through is likely to depend on the environment in which the plants are being grown. For container plants growing in close proximity such that canopies overlap, a significant proportion of overhead sprinkler irrigation that falls above a plant canopy could be channelled to outside the substrate surface area and fall between the containers. Beeson (2006) suggests that the greater the amount of canopy overlap, the lower the percentage of sprinkler irrigation reaching the container surface. The degree of canopy overlap interacts with spacing in determination of the proportion of irrigation water applied that actually reaches the plants: as spacing increases linearly, percentage of overhead irrigation reaching a container surface declines exponentially, resulting in substantial irrigation waste at modest distances between containers (Beeson and Knox 1991). The variability shown in this study between irrigation systems in total water use, and in the percentage of applied water that is transpired by the plant, highlights the need to maximise the efficiency of uptake of water and to understand where losses occur, before deficit irrigation can be effectively applied. Using different deficit irrigation regimes, Beeson (2006) found a linear relationship between cumulative evapotranspiration and canopy dry mass, with Japanese ligustrum and Indian hawthorn both gaining about 2.9 g of shoot dry mass per litre of water lost via evapotranspiration. However, the rate of increase in canopy volume for Japanese ligustrum was nearly double that for sweet viburnum. The minimum cumulative evapotranspiration that must occur to reach a specified size is likely to vary widely between species. Irrigation regimes that limit daily evapotranspiration will extend the time required to reach a given size. Extending the length of production could effectively result in greater water use with deficit irrigation regimes than regimes that keep pots closer to container capacity (Beeson 2006). In 2006, volumetric substrate moisture in pots irrigated with system 1 was higher than on the corresponding gravel bed (system 2) in July and August, as a result of water being retained on the Mypex and distributed to the plants. The calibration factor was then reduced on this bed

151

compared to irrigation system 2, but nonetheless substrate moisture was still generally very high with system 1 in September. This may relate partly to rainfall, which would be expected to have more of an effect on the Mypex bed than on gravel beds. Pots irrigated with system 6 showed a fairly uniform distribution of substrate moisture in July with most pots having substrate moistures between 20 and 40%. Substrate moistures with this system were, however, less uniform in August, but by September most pots had substrate moistures of between 35 and 55%, with just a few pots with lower substrate moisture; these were pots in which either a good contact with the sand or a substantial root system was not established early on in the season. Substrate moisture in pots under irrigation system 3 was very similar to that with system 6 in July and August 2006, but was generally very low in September. This shows the danger of scheduling a bed based on the substrate moisture in one pot: the sensor was in the middle of the bed in one of the relatively few pots with substrate moisture above 35%, meaning that irrigation did not turn on even though most of the pots on the bed had much lower substrate moisture. Hence not only the choice of which species to use as the control, but also the location of the control pot on the bed, will influence water use and efficiency of irrigation with a substrate moisture based system. The system used here is based on control from one sensor alone; averaging over several pots could reduce the potentially deleterious impact of scheduling based on an anomalous location/plant, but would result in an increase in the cost of the sensing equipment. Increased variation over the season in substrate moisture content under some irrigation systems might relate to faster growth rates and hence greater water use in some of the plants compared to others; alternatively it may relate to the accumulated effect of variation in irrigation on a non-uniformly overhead irrigated bed. That some pots irrigated with systems 1, 4, and 5 were quite wet in both July and August 2007 may relate partly to rainfall, but cannot relate entirely to rainfall given that several pots measured at the same time on other beds were quite dry. Non-uniformity of volumetric substrate moisture implies that non-uniformity of irrigation application is a problem even under the ‘‘improved’’ overhead irrigation systems. Non-uniformity of substrate moisture leads to heterogeneity in plant size and quality, a problem for grading plants for sale. Stirzaker and Hutchinson (2005) have pointed out that variability remains the greatest enemy of automated scheduling by soil water status. However, in this study uniformity in plant quality and size was not obviously more apparent under those systems with a more uniform distribution of irrigation application. Excessive irrigation may result in loss of nutrients from the growing substrate, but in this study there was no evidence for a deleterious effect of greater irrigation use on nutrient concentration in the

123

152

substrate at the end of the summer. However, the pH of the substrate was more ideal (5.0 [ pH \ 5.5, Bunt 1988) in those systems which used less water (systems 1 and 6) at the end of the summer, than in the systems to which more water was applied. pH in the substrates under the latter systems increased over the summer, probably relating to bicarbonates in the irrigation water (Bunt 1988). The greater concentration of magnesium in substrate from pots that were sub-irrigated may relate to the build up of salts on this bed, which is sometimes a concern with sub-irrigation. However, electrical conductivity in those samples, whilst higher than that in substrate taken from the other systems, fell within the ‘‘moderate’’ range (301–500 lS cm-1, Bunt 1988). In conclusion, a reduction in water use under the improved scheduled systems was achieved partly due to reducing the need to irrigate excessively where irrigation is not uniform, in order to ensure that all plants on the bed receive sufficient water. In addition, scheduling allowed adjustment of irrigation to the daily fluctuations in the weather or volumetric substrate moisture content. Two different scheduling methods were shown to work effectively. Overall less irrigation water was used with the soil sensor than with the Evaposensor. This may relate to the greater difficulty associated with estimating the required reduction in irrigation after rainfall when scheduling with the Evaposensor. With the soil moisture sensor, irrigation did not turn on if rainfall had sufficiently wet up the substrate. Whether lower water application occurs with the soil moisture sensor for other species and other nursery beds and other types of irrigation application needs to be verified. With the soil moisture sensor, to maintain the control loop, at least one sensor is required per bed (or per crop, if a similar crop is grown on several beds). The Evaposensor lends itself more easily to scaling up from this study: one sensor could be used for several beds with crops with different water requirements, provided different calibrations are performed for the different beds. Ideally, automation would allow its use with a multiple-valve irrigation controller, with the user only needing to input calibration factors according to the crop and the rate of application from a given set of irrigation sprinklers. Acknowledgments We would like to acknowledge the advice of the steering group of HDC HNS project 122, the helpful comments provided by an anonymous reviewer, and the assistance of the EMR facilities team. This research was funded by the Horticultural Development Council.

References Allen RG (2000) REF-ET Reference Evapotranspiration Calculator. Version Windows 2.0. University of Idaho, Kimberley

123

Irrig Sci (2009) 27:139–153 Allen RG, Jensen ME, Wright JL, Burman RD (1989) Operational estimates of reference evapotranspiration. Agron J 81:650–662 Beeson RC Jr (2004) Modeling actual evapotranspiration of Ligustrum japonicum from rooted cuttings to commercially marketable plants in 12 liter black polyethylene containers. Acta Hortic (ISHS) 664:71–77 Beeson RC Jr (2005) Modeling irrigation requirements for landscape ornamentals. Horttechnology 15:18–22 Beeson RC Jr (2006) Relationship of plant growth and actual evapotranspiration to irrigation frequency based on management allowed deficits for container nursery stock. J Am Soc Hortsci 131:140–148 Beeson RC Jr, Knox GW (1991) Analysis of efficiency of overhead irrigation in container production. HortScience 26:848–850 Beeson RC Jr, Yeager TH (2003) Plant canopy affects sprinkler irrigation application efficiency of container-grown ornamentals. HortScience 38:1373–1377 Beeson RC Jr, Brooks J (2007) Evaluation of a model based on reference crop evapotranspiration (ETo) for precision irrigation using overhead sprinklers during nursery production of Ligustrum japonica. Acta Hortic (in press) Briercliffe T, Hewson A, Brough W (2000) Independent water audits for container grown nursery stock producers. Summary report for MAFF Water audits Bunt AC (1988) Media and mixes for container-grown plants, 2nd edn. Unwin Hyman, London Cameron RWF, Harrison-Murray RS, Atkinson CJ, Judd HL (2006) Regulated deficit irrigation–a means to control growth in woody ornamentals. J Hort Sci Biotech 81:435–443 Charlesworth PB (2005) Irrigation Insights No. 1—soil water monitoring. National Program for Irrigation Research and Development, 2nd edn. CSIRO Publishing, Melbourne Costa JM, Ortun˜a MF, Chaves MM (2007) Deficit irrigation as a strategy to save water: physiology and potential application to horticulture. J Integr Plant Biol 49:1421–1434 Goodwin PB, Murphy M, Melville P, Yiasoumi W (2003) Efficiency of water and nutrient use in containerised plants irrigated by overhead, drip or capillary irrigation. Aust J Exp Agr 43:189–194 Grant OM, Stoll M, Jones HG (2004) Partial rootzone drying does not affect fruit yields of raspberries. J Hort Sci Biotech 79:125– 130 Grant OM, Chaves MM, Jones HG (2006) Optimizing thermal imaging as a technique for detecting stomatal closure induced by drought stress under greenhouse conditions. Physiol Plantarum 127:507–518 Harris GL, Hodgkinson RA, Scott M, Mason DJ, Pepper TJ (1997) Impact of hardy ornamental nursery stock (HONS) systems on the environment: losses of nutrients and agrochemicals. Agric Water Manage 34:95–110 Harrison-Murray RS (1991a) A leaf-model evaporimeter for estimating potential transpiration in propagation environments. J Hort Sci 66:131–139 Harrison-Murray RS (1991b) An electrical sensor for potential transpiration: principle and prototype. J Hort Sci 66:141–149 Henson DY, Newman SE, Hartley DE (2006) Performance of selected herbaceous annual ornamentals grown at decreasing levels of irrigation. HortScience 41:1481–1486 Irmak S (2005) Crop evapotranspiration and crop coefficients of Viburnum odoratissimum (Ker.-Gawl). Appl Eng Agric 21:371– 381 Jones HG (2004) Irrigation scheduling: advantages and pitfalls of plant-based methods. J Exp Bot 55:2427–2436 Lloyd JE, Herms DA, Rose MA, Van Wagoner J (2006) Fertilization rate and irrigation scheduling in the nursery influence growth, insect performance, and stress tolerance of ‘‘Sutyzam’’ crabapple in the landscape. HortScience 41:442–445

Irrig Sci (2009) 27:139–153 dos Santos TP, Lopes CM, Rodriguez ML, de Souza CR, Maroco JP, Pereira JS, Silva JR, Chaves MM (2003) Partial rootzone drying: effects on growth and fruit quality of field-grown grapevines (Vitis vinifera). Funct Plant Biol 30:663–671 Stanley CD, Harbaugh BK (1988) Estimated poinsettia crop water requirements based on evaporative demand and plant growth characteristics. HortScience 23:66 (024 PS I-A)

153 Stirzaker RJ (2003) When to turn the water off: scheduling microirrigation with a wetting front detector. Irrig Sci 22:177–185 Stirzaker RJ (2005) Managing irrigation with a wetting front detector. UK Irrig 33:22–24 Stirzaker RJ, Hutchinson PA (2005) Irrigation controlled by a wetting front detector: field evaluation under sprinkler irrigation. Aust J Soil Res 43:935–943

123