- Email: [email protected]
Effect of subsurface drip irrigation on processing tomato yield, water table depth, soil salinity, and profitability
Agricultural Water Management 68 (2004) 1–17
Effect of subsurface drip irrigation on processing tomato yield, water table depth, soil salinity, and profitability B. Hanson∗ , D. May Department of Land, Air, and Water Resources, University of California, One Shields Avenue, Davis, CA 95616-8627, USA Accepted 25 March 2004
Abstract The potential of subsurface drip irrigation of processing tomatoes to reduce subsurface drainage, control soil salinity, and increase farm profits in areas affected by saline, shallow ground water was evaluated at three fields with fine-textured, salt-affected soil along the west side of the San Joaquin Valley of California. No subsurface drainage systems were installed in these fields. Yield and quality of tomato of the drip systems were compared with sprinkler irrigation. Yield increases of 12.90–22.62 Mg/ha were found for the drip systems compared to the sprinkler systems with similar amounts of applied water. Soluble solids of the drip-irrigated tomatoes were acceptable. Response of water table levels during drip irrigation showed that properly managed drip systems could reduce percolation below the root zone. Yields of the drip systems were similar over the range of soil salinity levels that occurred near the drip lines. Profits under drip irrigation were 867 to $ 1493 ha−1 more compared to sprinkler irrigation, depending on the amount of yield increase and the interest rate used in the economic analysis. © 2004 Elsevier B.V. All rights reserved. Keywords: Drip irrigation; Processing tomato; Salinity; Shallow groundwater
1. Introduction The traditional approach to dealing with shallow ground water problems is to install subsurface drainage systems for water table control and improved leaching. Proper operation of the drainage systems requires disposal of the subsurface drainage water. No economically, technically, and environmentally feasible drain water disposal method exists for the San ∗
Corresponding author. Tel.: +1-5307524639; fax: +1-5307525262. E-mail address: [email protected] (B. Hanson). 0378-3774/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.agwat.2004.03.003
2
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17
Joaquin Valley of California (USA), and thus, the drainage problem must be addressed through options such as better management of irrigation water to reduce drainage below the root zone, increasing crop water use of the shallow groundwater without any yield reductions, and drainage water reuse for irrigation (Hanson and Ayars, 2002). One option for improving irrigation water management is to convert from furrow or sprinkler irrigation to drip irrigation. Drip irrigation can apply water both precisely and uniformly compared with furrow and sprinkler irrigation resulting in the potential to reduce subsurface drainage, control soil salinity, and increase yield. The main disadvantage of drip irrigation is its cost, which based on grower experience can be as much as $ 2470 ha−1 . For drip irrigation to be at least as profitable as the other irrigation methods, more revenue from higher yields and reduced irrigation and cultural costs must occur. Yet, several large-scale comparisons of furrow and drip irrigation of cotton revealed uncertainty in the economic benefits of drip irrigation (discussed later). Thus, growers converting to drip irrigation face uncertainty about the economics risks involved. Subsurface drip irrigation was compared with improved furrow irrigation of cotton on clay loam (Fulton et al., 1991). The improved furrow irrigation consisted of both surge irrigation and reduced furrow lengths for the preplant irrigation. Results showed that 61 mm more water was applied with the furrow systems compared to the drip system. Cotton yield was 163 kg/ha more for the drip system than for the furrow systems. However, profit was US$ 990/ha for the furrow systems and US$ 504/ha for drip irrigation. Another study also compared subsurface drip irrigation, improved furrow irrigation, and historic furrow irrigation of cotton under saline, shallow ground water conditions over several years (Styles et al., 1997). The improved furrow system consisted of furrow lengths one-half of those of the historic system. The historic furrow system applied 98 mm more water compared to the drip system, while the improved furrow system applied 37 mm more water. Cotton yield of the drip system was 16% higher than that of the furrow systems. Profit was US$ 1623/ha for the drip system, US$ 1249/ha for the improved furrow system, and US$ 1457/ha for the historic furrow system. During the past 3 years, subsurface drip irrigation of processing tomatoes was evaluated to determine its effect on crop yield and quality, soil salinity, water table depth, and profitability in salt-affected, fine-textured soil underlain by saline, shallow groundwater. Because tomatoes are a high cash value crop, a better potential for increased profitability with drip irrigation exists compared to cotton. However, tomatoes are much more sensitive to soil salinity, which could result in reduced crop yields in salt-affected soil. Several studies have investigated the effect of soil salinity on drip-irrigated tomatoes. Hand-harvested tomato yields ranged from 129.1 to 140.5 Mg/ha in 1991 and from 110.7 to 145 Mg/ha in 1993 under saline, shallow ground water conditions (Ayars et al., 2001). Machine-harvested yields of 1993 ranged from 71.7 to 112.0 Mg/ha. Depth to the shallow ground water was less than 2 m and its salinity was about 5 dS/m. Soil salinity ranged from about 4 to 10 dS/m for depths less than 1 m. About 10% of the water requirement of tomatoes was supplied by upward flow of the shallow ground water. Surface drip irrigation was used to irrigate tomatoes with irrigation water electrical conductivities of 1.2, 4.5, and 7.5 dS/m (Pasternak et al., 1986). The saline water was used for stand establishment. Results showed a yield reduction of about 10–12% for the 4.5 dS/m water compared to the 1.2 dS/m irrigation water, while yields of the 7.5 dS/m water were reduced by about
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17
3
60%. However, little difference was found between relative yield verses soil salinity for the Pasternak data and that of the salinity tolerance data for tomatoes (Maas and Grattan, 1999), which were developed from experiments using furrow irrigation (Grattan, S.R. personal communication).
2. Materials and methods 2.1. Field-wide comparison Subsurface drip irrigation systems were installed in three fields of tomatoes, each about 64.8 ha, located in the Westlands Water District, about 70 km southwest of Fresno CA in the San Joaquin Valley of California, USA. Sites DI (32.4 ha of drip irrigation) and BR (16.2 ha) were installed in 1999, while site DE (16.2 ha) was installed in 2000. Sprinkler irrigation was used for the rest of each field, the current irrigation method of tomatoes in these soils. Westlands Water District irrigation water was used at DI and BR, while well water was used at DE. Sprinkler irrigation was used for stand establishment at DE (direct-seeded) and BR (direct-seeded), while the drip system was used at DI (transplants). Measurements made at all sites were field-wide red fruit yield (machine harvested), yield quality, depth to the water table, electrical conductivity of irrigation water and groundwater, and applied water. Irrigation scheduling at each site was determined using appropriate crop coefficients and reference crop evapotranspiration from the California Irrigation Management Information System (CIMIS). No subsurface drainage systems existed at the drip-irrigated sites. Low-flow drip tape (149 l/h–100 m), 22 mm diameter, was buried about 0.20 m deep with one drip line per bed although two drip lines per bed were used for BR2001 (site per year). Emitter spacing ranged from 0.305 to 0.46 m depending on the type of tape. Drip line lengths were about 396 m at all sites. Irrigations were twice per week during the period of maximum canopy size. Soil type was clay loam at the three sites. 2.2. Small-plot differential irrigation experiment In addition, an experiment that consisted of applying different amounts of irrigation water to small plots was conducted in the drip-irrigated area of each field to determine the minimum amount of water that can be applied under saline, shallow ground water conditions without reducing crop yield. The DI1999 experiment used the varieties H9557, H9665, and H8892, while the DE2000 experiment used the varieties Halley 3155, H9665, and H8892. Otherwise, the growers’ varieties were used (Table 1). The experimental design used in 1999 and 2000 for the differential irrigation treatments consisted of dividing the field length into five blocks, each block 15 beds wide and 80 m long. Bed width ranged from 5 to 5.5 feet depending on the particular grower. Width of each plot was three bed spacings. This design allowed three varieties per plot to be grown with a minimum of disruption of field area to the growers’ operation. Each treatment occurred only once in a particular bed to minimize any bed effects. Applied water of each treatment was measured at the inlet of the first block. Differential irrigation amounts were obtained by using smaller irrigation set times compared to the growers’ irrigation time. Irrigation
4
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17
Table 1 Summary of field-wide applied water and yield characteristics for all sites and yearsa Irrigation system
Variety
Applied water (mm)
Yield (Mg/ha)
Soluble solids (%)
Color
BR Sprinkler (1999)b Drip (1999)b Drip (2000) Drip (2001) Drip (2002)
H8892 H8892 Halley 3551 H9665 Peto303
427 406 427 521
81.86 103.71 78.40 71.45 109.54
5.3 6.0 5.4 4.6 4.8
24.2 21.1 23.4 25.3 24.1
DI Sprinkler (1999)b Drip (1999)b Drip (2000) Drip (2001)
H9557 H9557 H9492 H9492
564 737 582
78.85 90.94 103.94 115.80
5.2 5.0 4.8 4.9
24.8 22.8 21.0 24.1
DE Sprinkler (2000)b Drip (2000)b Drip (2001)
H9557 H9557 H8892
579 711 561
63.84 86.46 102.59
5.5 5.6 5.2
23.9 23.7 23.6
c
c
a
BR, DI, and DE identify the various sites. Comparison year. c Data not available. b
frequency was the same used by the grower. The data were analyzed by regressing total yield along the field length against cumulative applied water for each treatment. A randomized block design was used in 2001 consisting of four irrigation treatments replicated five times with each block extending along the field length. Each plot consisted of one continuous bed along the field length. The same irrigation set time and irrigation frequency were used for each treatment, but differential irrigation amounts were obtained by using different manifold pressures compared to the field system pressure. Data collected in the irrigation plots were applied water for each treatment, total red fruit yield (machine harvested), soluble solids, color; percent red, green, and non-marketable fruit; soil salinity; and weekly measurements of both canopy coverage and soil water content. Sampling locations for both soil water content and soil salinity were 0.25 m from the drip line at 0.15 m depth intervals down to 0.76–0.91 m deep (depending on site conditions) at the head, the middle, and the end of the field. The electrical conductivity of the saturated extract (ECe) was determined for each soil sample. A digital infrared camera and appropriate software were used to measure canopy coverage; a neutron moisture meter was used for soil water measurements. In addition to the soil water and soil salinity measurements made 0.25 m from the drip line, patterns of soil water content and soil salinity around the drip line were determined by a one-time sampling with depth at various distances from the drip line. A ThetaProbe was used for these cross-sectional measurements, which measures volumetric soil water content. Seasonal crop evapotranspiration (ET) was estimated using a computer ET model (Hsiao and Henderson, 1985) and reference crop (grass) ET. The model was calibrated for tomatoes with ET data from an unrelated project (Hanson, 2003). Differences between measured seasonal ET and that estimated with the model were 5% or less.
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17
5
3. Results 3.1. Field-wide comparison At each site, only 1 year of comparing drip versus sprinkler irrigation was possible (1999 for DI and BR; 2000 for DE). After the first year at each site, the rest of the field was converted to drip irrigation at BR and DE, while at DI, a different crop was planted in the rest of the field. After the comparison year, yields of the drip-irrigated fields were monitored for several additional years. The comparisons showed that the field-wide yields under drip irrigation were 12.10–22.62 Mg/ha more than those under sprinkler irrigation (Table 1). Average yields were 93.63 and 74.82 Mg/ha for drip- and sprinkler irrigation, respectively. Differences between the sprinkler and drip irrigation field-wide yields were statistically significant using the t-test with a level of significance of 5%. Drip yields were considered to be high for these fine-textured, salt-affected soils. After the first year, yields at DI and DE continued to be high (Table 1). Yields at BR for 2000 and 2001 were relatively low due to late plantings although these yields were higher than normally experienced for late plantings. A high yield was found for BR2002, which had an earlier planting date compared to 2000 and 2001. Soluble solids of the drip-irrigated fields, determined by a commercial grader, were acceptable for all years with average soluble solids of 5.3 and 5.5% for sprinkler and drip irrigation, respectively. Soluble solids increased with increasing soil salinity with average soluble solids of 4.9, 5.3, and 5.4 for DI (the lowest salinity level), BR, and DE (the highest salinity level), respectively. This behavior was similar to that found by Mitchell et al. (1991). The average color, also determined by a commercial grader, was 24.3 and 22.5 for sprinkler and drip irrigation, respectively. Differences in soluble solids and color between drip and sprinkler irrigation were not statistically significant. Applied water at BR1999 was similar for drip and sprinkler irrigation (Table 1). About 0.15 m more water was applied to the drip field compared with the sprinkler field for DE2000, partly because the drip field was irrigated for about 2 weeks longer. Applied water data for the sprinkler field at DI1999 were not available. 3.2. Differential irrigation treatments yield characteristics Results of the differential irrigation experiments showed that plot yield declined with decreasing irrigation water applications for all sites and all years although differences in behavior occurred among the sites and years (Fig. 1 for 2001 data). At DI, the overall average yield of all tomato varieties (yield differences between varieties were statistically insignificant) decreased from 105.28 to 94.08 Mg/ha as applied water decreased from 587 to 378 mm in 1999; and in 2001, it decreased from 113.12 to 103.94 Mg/ha as applied water decreased from 509 to 343 mm. At BR, 1999 yields decreased from 84.45 (730 mm of water) to 61.38 Mg/ha (508 mm), while in 2001, 64.06 Mg/ha occurred for 569 mm of water and 58.91 Mg/ha for 246 mm. DE2001 yields decreased from 110.43 to 94.75 Mg/ha as applied water decreased from 477 to 264 mm. Similar trends occurred in 2000, but the range of applied water was relatively small (data not shown). Soluble solids increased with decreasing applied water for all sites and all years (Fig. 1 for 2001 data). In 1999, overall average soluble solids at DI increased from 4.6 to 5.0% as
6
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17 120 2001
Yield (Mg/ha)
100 80 60
DI DI Regression DE DE Regression BR BR Regression
40 20
Site DI: Yield = 84.05 x AW +0.06, r2 = 0.90 Site DE: Yield = 78.51 x AW + 0.08, r2 = 0.72 Site BR: Yield = 54.06 x AW + 0.02, r2 = 0.91
0
(A)
0
100
200
300
400
500
600
6 2001 Soluble Solids (%)
5 4 3
DI DI Regression DE DE Regression BR BR Regression
2 1
Site DI: Brix = -0.00056 x AW + 5.18, r2 = 0.05 Site DE: Brix = -0.0014 x AW + 5.74, r2 = 0.57 Site BR: Brix = -0.00050 x AW + 4.92, r2 = 0.92
0
(B)
0
100
200
300
400
500
600
25 2001
Color
20 15 DI DI Regression DE DE Regression BR BR Regression
10 5
Site DI: Brix = -0.0034 x AW + 24.2, r2 = 0.36 Site DE: Brix = 0.0086 x AW + 18.7, r2 = 0.24 Site BR: Brix = 0.0016 x AW + 22.3, r2 = 0.18
0 0
(C)
100
200
300
400
500
600
Applied Water (mm)
Fig. 1. Results of the 2001 differential drip-irrigated experiment for (A) crop yield, (B) soluble solids, and (C) color. BR, DI, and DE identify the various sites.
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17
7
applied water decreased from 587 to 378 mm, while at BR, soluble solids increased from 5.9 to 6.8% as applied water decreased from 730 to 508 mm. The 2001 data showed soluble solids to increase from 4.8 to 5.2% as applied water decreased from 509 to 343 mm at DI; 5.2–5.4% as applied water decreased from 477 to 264 mm at DE; and 4.7–4.8% as applied water decreased from 568 to 246 mm at BR. Applied water had little effect on color (Fig. 1) and percent red fruit (data not shown). Linear regression equations relating yield characteristics with applied water were tested for their statistical significance and statistical similarity among the sites. Results of the statistical tests were mixed, preventing any conclusions from being developed about differences between sites and between years. There were no statistical differences in yield, solids, and color between varieties at DI in 1999 and at DE in 2000. 3.3. Water quality The electrical conductivity (EC) of the Westlands Water District irrigation water at BR and DI normally was about 0.34 dS/m. At DE, the electrical conductivity of the well water was about 1.06–1.2 dS/m. The EC of the shallow ground water at BR ranged from 4.7 to 7.4 dS/m, while at DI, it ranged from 7.9 to 11.1 dS/m for 1999 and 2000 and was 4.0–4.7 dS/m in 2001. Reasons for the small 2001 values are unknown even though sampling locations were within 10 m of each other. Ground water EC values at DE were 13.6–16.4 dS/m in 2000 and 9.0–9.5 dS/m in 2001. These differences may reflect different sampling locations in the field from year-to-year. 3.4. Water table response The water table depth at DI1999 decreased with time until about July 20 and then increased to about 1.8 m deep, while the water table remained below 1.8 m deep in 2000 and 2002 (Fig. 2). No response of water table depth to drip irrigation was evident. At BR1999, the water table depth increased from about 0.6 to about 1.3 m, but in 2000, drip irrigations caused the water table to rise to nearly 0.5 m deep before July 15, the result of applying about 10% more water than the estimated crop evapotranspiration. After mid-July, the water table depth increased to 1.52–1.83 m deep due to reduced water applications. Water table levels were not measured at DE in 2000 because of problems in installing observation wells. In 2001, water table depth at DE fluctuated between about 0.61 and 1.22 m with a definite response to drip irrigation. The gaps in the 2001 data for BR and DI were caused by the water level in the observation wells dropping below the pressure transducers, corrected by deepening the wells. 3.5. Soil salinity Soil salinity as measured by the electrical conductivity of the saturated extract (ECe) differed considerably among the three sites (Fig. 3 for 2001 data). ECe values at DI were generally less than 2.5 dS/m (threshold value for tomato). (Note: the threshold ECe value is the maximum average root zone ECe at which no yield reduction should occur (Maas and
8
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17
Depth to Water Table (m)
0.0
(A)
1999
BR
0.5 1.0 1.5 2.0 2.5
DI Jun 1
Jun 15
Jul 1
Jul 15
Aug 1
Aug 15
Sep 1 Sep 15
3.0 140 150 160 170 180 190 200 210 220 230 240 250 260
Depth to Water Table (m)
0.0
(B)
2000
BR
0.5 1.0 DI
1.5 2.0 2.5
Jun 1
Jun 15
Jul 1
Jul 15
Aug 1
Aug 15
Sep 1 Sep 15
3.0 140 150 160 170 180 190 200 210 220 230 240 250 260
Depth to Water Table (m)
0.0 2001 0.5
DE
1.0 BR
1.5 DI
2.0 2.5
Jun 1
Jun 15
Jul 1
Jul 15
Aug 1
Aug 15
Sep 1 Sep 15
3.0 140 150 160 170 180 190 200 210 220 230 240 250 260
(C)
Day of Year
Fig. 2. Depth to water table for (A) 1999, (B) 2000 and (C) 2001. BR, DI, and DE identify the various sites.
Grattan, 1999). The actual root zone salinity under drip irrigation at these sites is unknown because of spatially varying patterns of soil salinity, soil water, and probably root density around drip lines. The threshold value is provided as a reference only to indicate a potential for yield reduction.) At BR2001, ECe increased considerably with depth and exceeded the threshold value except for depths less than about 0.54 m. ECe values at DE2001 increased
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17 0.00
0.00 DI2001 Wet
Depth (m)
0.25
DI2001 Dry 0.25
1 May 15 Aug.
0.50
0.50
0.75
0.75
1.00
1 May 14 Aug
1.00 0 1 2 3 4 5 6 7 8 9 10 11 12
0.00
0 1 2 3 4 5 6 7 8 9 10 11 12 0.00
BR2001 Wet
BR2001 Dry
Depth (m)
0.25
0.25 2 May 19 Sept.
2 May 19 Sept.
0.50
0.50
0.75
0.75
1.00
1.00 0 1 2 3 4 5 6 7 8 9 10 11 12
0.00
0 1 2 3 4 5 6 7 8 9 10 11 12 0.00
DE2001 Wet
Depth (m)
9
DE2001 Dry
0.25
0.25
0.50
0.50
0.75
1 May 14 Aug
0.75 1 May 14 Aug.
1.00
1.00 0 1 2 3 4 5 6 7 8 9 10 11 12
0 1 2 3 4 5 6 7 8 9 10 11 12
ECe (dS/m)
Fig. 3. Electrical conductivity of saturated extracts (ECe) with depth for wet (receiving the most irrigation water) and dry (receiving the least irrigation water) differential drip-irrigated treatment in 2001. Dashed line is the reference salinity threshold value for tomato. Units are dS/m.
with depth with all values exceeding the threshold value. At all sites, differences between wet and dry irrigation treatments were slight. (In the differential drip-irrigated experiment, the “wet” treatment received the most irrigation water and the “dry” treatment received the least irrigation water.)
10
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17
Depth (mm)
DI2000
-200 Drip Line
ECe (dS/m)
-400
-600
a -600
(A)
-400
-200
0
200
400
600
Distance From Drip Line (mm)
Depth (mm)
BR2000
-200 Drip Line -400
-600
b -600
(B)
12 11 10 9 8 7 6 5 4 3 2 1 0
-400
-200
0
200
400
600
Distance From Drip Line (mm)
Fig. 4. Patterns of electrical conductivity of saturated extracts (ECe) around the drip line for (A) DI2000 and (B) BR2000. Units are dS/m.
The pattern of ECe around the drip line showed values less than the threshold value throughout the soil profile at DI (Fig. 4A). At BR2000, salinity was the least near the drip line with values less than about 1 dS/m, but salinity increased with horizontal distance from the drip line and with depth to values of about 7 dS/m (Fig. 4B). Salinity also increased above the drip line as depth decreased indicating salt accumulation above the drip line. The zone of ECe values less than the threshold value extended about 0.41 m horizontally from the drip line and about 0.21 m deep below the drip line. At DE2000, ECe was the highest near the drip line with values of 3–4 dS/m and decreased with horizontal distance to values less than 2.5 dS/m at distances beyond about 0.21–0.42 m (Fig. 5A). Salinity above the drip line increased as depth decreased. The high salinity near the drip line reflected the well water EC. The low levels of salinity near the edge of the pattern probably reflected leaching of salts due to ponding from a severe late spring rain. In 2001, ECe levels ranged between 5 and 7 dS/m throughout most of the soil profile except near the drip line where ECe values were between 3 and 4 dS/m (Fig. 5B). The main source of salt in these fields is from the upward flow of saline, shallow ground water into the root zone. However, poor correlation was found between soil salinity near the
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17
11
DE2000
Depth (mm)
Drip Line
ECe (dS/m)
-200
-400 a -600
(A) -600
-400
-200
0
200
400
600
Distance From Drip Line (mm)
Depth (mm)
DE2001 Drip Line
-200
-400
-600
b -600
(B)
12 11 10 9 8 7 6 5 4 3 2 1 0
-400
-200
0
200
400
600
Distance From Drip Line (mm)
Fig. 5. Patterns of electrical conductivity of saturated extracts (ECe) around the drip line for (A) DE2000 and (B) DE2001. Units are dS/m.
bottom of the sampled soil profile and ground water salinity. At DI, soil salinity at the deeper depths of the sampled soil was generally was less than 2 dS/m, but the ground water salinity was much higher. At BR, soil salinity levels at the deeper sampled depths were similar to the ground water salinity; while at DE, soil salinity was less than the ground water salinity. Reasons for the behavior at DI and DE are not clear, but the deeper water table depth at DI (generally 1.82 m or deeper) probably contributed to the smaller soil salinity values, whereas at the other sites, much smaller water table depths occurred. The deeper depth at DI may have greatly reduced upward flow of shallow ground water into the root zone. 3.6. Soil water Soil water content decreased with time throughout the irrigation season for all irrigation treatments (data not shown). Water contents at the start of the measurement period were generally between 37 and 43% and decreased to between 27 and 38% prior to harvest. Average water contents of the wet irrigation treatments were slightly higher than those of the dry irrigation treatments. Seasonal changes in soil water content, estimated from the neutron moisture meter data, ranged from 33 to 86 mm. Wetting patterns around the drip line showed water moving laterally to about 0.41 m from the drip line at DI (Fig. 6) and BR (data not shown). At about that distance, soil
12
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17
DI1999 (July 27)
0 -200
Drip Line
-400
(%)
-600
50
Depth (mm)
45
-800
Wet
(A) 0
200
40
400
600
800 1000 1200 1400 1600
35 30
0
25 20
-200 Drip Line
15 10
-400
5 0
-600 Dry
0
200
400
(B)
600 800 1000 1200 1400 1600 Distance Across Bed (mm)
Fig. 6. Patterns of soil water content (%) around drip lines for (A) wet (receiving the most irrigation water) and (B) dry (receiving the least irrigation water) differential drip-irrigation treatments in DI1999. Units are volumetric soil water content in %.
water content was the least for a given depth. Soil water content increased with depth, but changes with depth were small below the drip line. At about 0.51 m from the drip line (in the furrow), slightly higher soil water contents occurred compared to 0.41 m suggesting less water extraction at 0.51 m compared to 0.41 m. Soil water contents above about 0.38–0.51 m deep were less for the dry treatment compared to the wet treatment. Similar behavior occurred at BR. Wetting patterns at DE were not measured, but based on the salinity pattern in Fig. 5A, lateral flow was between 0.20 and 0.41 m from the drip line. 3.7. Crop evapotranspiration A variety of canopy growth curves were found with maximum coverage ranging from 74 to 99% (data not shown). Canopy growth curves showed a rapid increase in canopy development starting between 20 and 50 DAP (days after planting). Maximum canopy coverage generally occurred between 60 and 80 DAP. In some case, canopy coverage decreased later
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17
13
100 2001
Canopy Coverage (%)
90 80 70 60 50 40 30
DE DI BR
20 10 0 0
20
40
60
(A)
80
100
120
140
Days After Planting
Crop Evapotranspiration (mm/day)
10 2001 8
6 BR DE DI ETo
4
2 Apr 1
May 1
Jun 1
Jul 1
Aug 1
Sep 1
0 80
(B)
100
120
140
160
180
200
220
240
260
280
Day of Year (DOY)
Fig. 7. (A) canopy coverage, and (B) crop evapotranspiration for 2001. BR, DI, and DE identify the various sites.
in the season due to pruning of vines by growers and canopy aging. No differences in canopy coverage were found between the irrigation treatments. The canopy growth curves of 2001 illustrate the effect of different cultural practices on canopy growth and crop evapotranspiration (Fig. 7A). The planting date of DE (April 10) was much earlier than those of BR (May 5) and DI (May 2). At both DE and BR, plants were direct-seeded, while transplants were used at DI. Sprinkler irrigation was used for stand establishment at BR and DE, while drip irrigation was used at DI. Canopy growth at DI was about 10–20 days ahead of that of the direct-seeded plants at BR (Fig. 7A). High evapotranspiration occurred just after planting at BR and DE due to evaporation from soil wetted by sprinkler irrigation (Fig. 7B). Very small ET rates occurred at DI just after planting because of the dry soil surface at that time due to using drip irrigation for stand establishment. Maximum ET rates occurred at about DOY170 (day of year) at BR, DOY180 at DI,
14
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17
Table 2 Total amount of applied water, cumulative crop ET, and irrigation efficiency (IE) Year
Applied water (mm)
Seasonal change in soil water (mm)
Cumulative crop ET (mm)
IE (%)
BR 1999 2000 2001
406 427 521
86 69 69
516 544 582
105 110 99
DI 1999 2000 2001
564 737 582
38 86 76
638 640 676
106 78 103
DE 2000 2001
711 561
33 81
615 587
83 91
The irrigation efficiency is the ratio of the cumulative crop ET to the sum of applied water and seasonal soil water change. BR, DI, and DE identify the various sites.
and DOY190 at BR. ET rates were similar during the mid-season growth stage, and then decreased during the late-season stage at all sites, primarily the results of canopy pruning. Seasonal cumulative ET for all years calculated using the computer model and canopy growth curves showed ET values to range from 516 to 676 mm (Table 2). Seasonal irrigation efficiency, defined as the ratio of cumulative ET to the sum of the cumulative applied water and seasonal change in soil water, ranged from 78 to 110%. Values near or exceeding 100% indicate deficit irrigation and plausible use of the shallow ground water. 3.8. Economics The economics of converting to subsurface drip irrigation from sprinkler irrigation was examined using cost data provided by one of the grower participants. Assumptions used in this analysis were: • The existing sprinkler irrigation system was used elsewhere on the farm. • The economic life of the drip system was 20 years. • Replacement of the drip tape occurred every 5 years. This replacement schedule is based on the growers’ practice in this area. • Filters and pumps were replaced every 10 years. • Drip irrigation yield increases ranged from 12.90 to 22.62 Mg/ha compared to sprinkler irrigation (Table 1). • Equivalent annual capital cost of the drip irrigation system was determined for interest rates of 5 and 10%. • The same amount of irrigation water was applied by both irrigation methods. • Area irrigated was 32.4 ha. The benefits of converting to drip irrigation were increased revenue from higher yields and annual savings of cultural costs and energy costs of sprinkler irrigation. The costs of
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17
15
Table 3 Economic analysis of benefits and cost ($ ha−1 ) of converting from an existing sprinkler irrigation system to a subsurface drip system for interest rates of 5 and 10% Interest rate (%) 5
10
Benefits Revenue increase Savings of sprinkler energy costs Savings of sprinkler cultural costs Subtotal
709a , 1244b 141 1568 2418a , 2958b
709a , 1244b 141 1568 2418a , 2958b
Costs ($ ha−1 ) Equivalent annual capital cost of drip system Drip energy costs Drip cultural costs Subtotal
296 69 1099 1465
383 69 1099 1551
($ ha−1 )
Net returns ($ ha−1 )
953a , 1493b
867a , 1407b
Yield increases of 12.90 and 22.62 Mg/ha were used for this analysis. Crop price was $ 55/Mg. a Yield increase of 12.90 Mg/ha. b Yield increase of 22.62 Mg/ha.
the conversion were the equivalent annualized capital cost of the drip system and its annual cultural costs and energy costs. Annual net returns ranged from 953 to $ 1493 ha−1 for a 5% interest rate and from 867 to $ 1407 ha−1 for a 10% rate (Table 3). Returns to land, farm management costs, taxes, and insurance costs (data not available) were not included. Capital cost of the drip system was $ 1998 ha−1 . The equivalent annual capital cost of the drip system was 296 and $ 383 ha−1 for the 5 and 10% interest rates, respectively. Reasons for the higher profits under drip irrigation were mainly due to higher yields, and thus higher revenue, and lower cultural costs compared to sprinkler irrigation. Reduced weeding costs and eliminating the costs of setting up and moving hand-move sprinklers contributed to the smaller cultural costs for drip irrigation. This economic analysis assumed a drip line replacement every 5 years. However, drip line replacements of 10 years or more have been reported. Assuming a 10-year replacement schedule resulted in an equivalent annual capital cost of $ 316 ha−1 for the drip system at a 10% interest rate compared to $ 383 ha−1 for the 5-year replacement.
4. Discussion and conclusions Subsurface drip irrigation in these fine-textured salt-affected soils can increase yield and profit of tomatoes compared to sprinkler irrigation with acceptable levels of soluble solids (mainly due to the soil salinity at these locations). Drip irrigation also can control subsurface drainage to the shallow ground water. Little correlation was found between soil salinity and crop yield over the range of ECe’s found at these sites even though ECe values higher than the threshold ECe were found around the drip line at one site, suggesting soil salinity under drip irrigation may affect crop yield less compared to other irrigation methods, as
16
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17
suggested by Shalhevet (1994). Subsurface drip irrigation also provided better late-season water management, a time when careful water management is needed to prevent excessive deficit irrigation or phytophthora from excessive wet soil. Little, if any, water savings on a per hectare basis are likely to occur by converting to drip irrigation from sprinkler irrigation. The higher yields of the drip irrigation suggest increased evapotranspiration compared to sprinkler irrigation. This ET increase may offset any water savings due to reduced percolation and evaporation losses found with drip irrigation compared to those losses under sprinkler irrigation. However, because higher yields occurred under subsurface drip irrigation, the same total yield could be grown on fewer hectares compared to sprinkler irrigation, which would save water. The results of the differential irrigation treatments showed that subsurface drip irrigation must be carefully managed to prevent yield reduction and excessive drainage to the ground water. Recommended irrigation amounts are near 100% of the potential crop ET as a compromise between reducing drainage and controlling soil salinity in the root zone. Irrigation should occur two to three times per week (Hanson et al., 2003). In summary, the long-term sustainability of tomato yield under subsurface drip irrigation in these salt-affected soils will require the following: • Sufficient leaching must occur to maintain acceptable levels of soil salinity near the drip lines where the root density is probably the greatest. • Periodic leaching of salt accumulated above the buried drip lines with sprinklers will be necessary for stand establishment if winter and spring rainfall is insufficient to leach the salts. • Careful management of irrigation water will be required to apply sufficient water for crop evapotranspiration and leaching yet prevent excessive subsurface drainage. • Periodic system maintenance must be performed to prevent clogging of drip lines. Clogging due to root intrusion was found to be a severe problem at one site where little or no chlorination occurred. Clogging will not only reduce the applied water needed for crop ET, but also reduce the leaching. Subsurface drip irrigation in these marginal soils is very profitable, which has encouraged the conversion to subsurface drip irrigation from sprinkler irrigation in this area. However, where high tomato yields are obtained under furrow and sprinkler irrigation, converting to drip irrigation may not be profitable because the potential for large yield increases under drip irrigation may not exist. Any increase in revenue under drip irrigation may be insufficient to offset capital, energy, maintenance, and management costs of subsurface drip irrigation. Also, using drip irrigation on lower-valued crops also may be unprofitable even if yield increases occur because the relatively small revenue from these crops may be insufficient to offset the costs of drip irrigation.
Acknowledgements We wish to acknowledge the contributions of the US Bureau of Reclamation, Westlands Water District, Britz Farming Corp., Farming D. Inc., Agri-Valley Irrigation Company,
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17
17
T-Systems International, Netafim Irrigation Company, Roberts Irrigation Products, Rain Bird, Toro Ag, and the University of California Salinity/Drainage Program.
References Ayars, J.E., Schoneman, R.A., Dale, F., Meson, B., Shouse, P., 2001. Managing subsurface drip irrigation in the presence of shallow ground water. Agric. Water Manage. 47, 242–264. Fulton, A.E., Oster, J.D., Hanson, B.R., Phene, C.J., Goldhamer, D.A., 1991. Reducing drains water: furrow vs. subsurface drip irrigation. Calif. Agric. 45 (2), 4–8. Hanson, B.R., 2003. Evapotranspiration of tomato under furrow/sprinkler and subsurface irrigation. HortScience 38 (5), 759. Hanson, B.R., Ayars, J.E., 2002. Strategies for reducing subsurface drainage in irrigated agriculture through improved irrigation. Irrigation Drainage Syst. 16, 261–277. Hanson, B.R., May, D.M., Schwankl, L.J., 2003. Effect of irrigation frequency on subsurface drip irrigated vegetables. HortTechnology 13 (1), 115–120. Hsiao, T.C., Henderson, D.W., 1985. Improvement of crop coefficients for evapotranspiration. California Irrigation Management Information System Final Report. Maas, E.V., Grattan, S.R., 1999. Crop yields as affected by salinity. In: Skaggs, R.W., van Schilfgaarde, J. (Eds.), Agricultural Drainage, Agronomy. Monograph 38. ASA, CSSA, SSSA, Madison, WI. Mitchell, J.P., Shennan, C., Grattan, S.R., May, D.M., 1991. Tomato fruit yields and quality under water deficit and salinity. J. Am. Soc. Hort. Sci. 116 (2), 215–221. Pasternak, D., De Malach, Y., Borovic, I., 1986. Irrigation with brackish water under desert conditions. VII. Effect of time of application of brackish water on production of tomatoes (Lycopersion esculentum Mill.). Agric. Water Manage. 12, 149–158. Shalhevet, J., 1994. Using water of marginal quality for crop production: major issues. Agric. Water Manage. 25, 233–269. Styles, S., Oster, J.D., Bernaxconi, P., Fulton, A., Phene, C., 1997. Demonstration of emerging technologies. In: Guitjens, J., Dudley, L. (Eds.), Agroecosystems: Sources, Control and Remediation. Pacific Division, Am. Assoc. Adv. Sci., San Francisco, CA, pp. 183–206.
Effect of subsurface drip irrigation on processing tomato yield, water table depth, soil salinity, and profitability B. Hanson∗ , D. May Department of Land, Air, and Water Resources, University of California, One Shields Avenue, Davis, CA 95616-8627, USA Accepted 25 March 2004
Abstract The potential of subsurface drip irrigation of processing tomatoes to reduce subsurface drainage, control soil salinity, and increase farm profits in areas affected by saline, shallow ground water was evaluated at three fields with fine-textured, salt-affected soil along the west side of the San Joaquin Valley of California. No subsurface drainage systems were installed in these fields. Yield and quality of tomato of the drip systems were compared with sprinkler irrigation. Yield increases of 12.90–22.62 Mg/ha were found for the drip systems compared to the sprinkler systems with similar amounts of applied water. Soluble solids of the drip-irrigated tomatoes were acceptable. Response of water table levels during drip irrigation showed that properly managed drip systems could reduce percolation below the root zone. Yields of the drip systems were similar over the range of soil salinity levels that occurred near the drip lines. Profits under drip irrigation were 867 to $ 1493 ha−1 more compared to sprinkler irrigation, depending on the amount of yield increase and the interest rate used in the economic analysis. © 2004 Elsevier B.V. All rights reserved. Keywords: Drip irrigation; Processing tomato; Salinity; Shallow groundwater
1. Introduction The traditional approach to dealing with shallow ground water problems is to install subsurface drainage systems for water table control and improved leaching. Proper operation of the drainage systems requires disposal of the subsurface drainage water. No economically, technically, and environmentally feasible drain water disposal method exists for the San ∗
Corresponding author. Tel.: +1-5307524639; fax: +1-5307525262. E-mail address: [email protected] (B. Hanson). 0378-3774/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.agwat.2004.03.003
2
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17
Joaquin Valley of California (USA), and thus, the drainage problem must be addressed through options such as better management of irrigation water to reduce drainage below the root zone, increasing crop water use of the shallow groundwater without any yield reductions, and drainage water reuse for irrigation (Hanson and Ayars, 2002). One option for improving irrigation water management is to convert from furrow or sprinkler irrigation to drip irrigation. Drip irrigation can apply water both precisely and uniformly compared with furrow and sprinkler irrigation resulting in the potential to reduce subsurface drainage, control soil salinity, and increase yield. The main disadvantage of drip irrigation is its cost, which based on grower experience can be as much as $ 2470 ha−1 . For drip irrigation to be at least as profitable as the other irrigation methods, more revenue from higher yields and reduced irrigation and cultural costs must occur. Yet, several large-scale comparisons of furrow and drip irrigation of cotton revealed uncertainty in the economic benefits of drip irrigation (discussed later). Thus, growers converting to drip irrigation face uncertainty about the economics risks involved. Subsurface drip irrigation was compared with improved furrow irrigation of cotton on clay loam (Fulton et al., 1991). The improved furrow irrigation consisted of both surge irrigation and reduced furrow lengths for the preplant irrigation. Results showed that 61 mm more water was applied with the furrow systems compared to the drip system. Cotton yield was 163 kg/ha more for the drip system than for the furrow systems. However, profit was US$ 990/ha for the furrow systems and US$ 504/ha for drip irrigation. Another study also compared subsurface drip irrigation, improved furrow irrigation, and historic furrow irrigation of cotton under saline, shallow ground water conditions over several years (Styles et al., 1997). The improved furrow system consisted of furrow lengths one-half of those of the historic system. The historic furrow system applied 98 mm more water compared to the drip system, while the improved furrow system applied 37 mm more water. Cotton yield of the drip system was 16% higher than that of the furrow systems. Profit was US$ 1623/ha for the drip system, US$ 1249/ha for the improved furrow system, and US$ 1457/ha for the historic furrow system. During the past 3 years, subsurface drip irrigation of processing tomatoes was evaluated to determine its effect on crop yield and quality, soil salinity, water table depth, and profitability in salt-affected, fine-textured soil underlain by saline, shallow groundwater. Because tomatoes are a high cash value crop, a better potential for increased profitability with drip irrigation exists compared to cotton. However, tomatoes are much more sensitive to soil salinity, which could result in reduced crop yields in salt-affected soil. Several studies have investigated the effect of soil salinity on drip-irrigated tomatoes. Hand-harvested tomato yields ranged from 129.1 to 140.5 Mg/ha in 1991 and from 110.7 to 145 Mg/ha in 1993 under saline, shallow ground water conditions (Ayars et al., 2001). Machine-harvested yields of 1993 ranged from 71.7 to 112.0 Mg/ha. Depth to the shallow ground water was less than 2 m and its salinity was about 5 dS/m. Soil salinity ranged from about 4 to 10 dS/m for depths less than 1 m. About 10% of the water requirement of tomatoes was supplied by upward flow of the shallow ground water. Surface drip irrigation was used to irrigate tomatoes with irrigation water electrical conductivities of 1.2, 4.5, and 7.5 dS/m (Pasternak et al., 1986). The saline water was used for stand establishment. Results showed a yield reduction of about 10–12% for the 4.5 dS/m water compared to the 1.2 dS/m irrigation water, while yields of the 7.5 dS/m water were reduced by about
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17
3
60%. However, little difference was found between relative yield verses soil salinity for the Pasternak data and that of the salinity tolerance data for tomatoes (Maas and Grattan, 1999), which were developed from experiments using furrow irrigation (Grattan, S.R. personal communication).
2. Materials and methods 2.1. Field-wide comparison Subsurface drip irrigation systems were installed in three fields of tomatoes, each about 64.8 ha, located in the Westlands Water District, about 70 km southwest of Fresno CA in the San Joaquin Valley of California, USA. Sites DI (32.4 ha of drip irrigation) and BR (16.2 ha) were installed in 1999, while site DE (16.2 ha) was installed in 2000. Sprinkler irrigation was used for the rest of each field, the current irrigation method of tomatoes in these soils. Westlands Water District irrigation water was used at DI and BR, while well water was used at DE. Sprinkler irrigation was used for stand establishment at DE (direct-seeded) and BR (direct-seeded), while the drip system was used at DI (transplants). Measurements made at all sites were field-wide red fruit yield (machine harvested), yield quality, depth to the water table, electrical conductivity of irrigation water and groundwater, and applied water. Irrigation scheduling at each site was determined using appropriate crop coefficients and reference crop evapotranspiration from the California Irrigation Management Information System (CIMIS). No subsurface drainage systems existed at the drip-irrigated sites. Low-flow drip tape (149 l/h–100 m), 22 mm diameter, was buried about 0.20 m deep with one drip line per bed although two drip lines per bed were used for BR2001 (site per year). Emitter spacing ranged from 0.305 to 0.46 m depending on the type of tape. Drip line lengths were about 396 m at all sites. Irrigations were twice per week during the period of maximum canopy size. Soil type was clay loam at the three sites. 2.2. Small-plot differential irrigation experiment In addition, an experiment that consisted of applying different amounts of irrigation water to small plots was conducted in the drip-irrigated area of each field to determine the minimum amount of water that can be applied under saline, shallow ground water conditions without reducing crop yield. The DI1999 experiment used the varieties H9557, H9665, and H8892, while the DE2000 experiment used the varieties Halley 3155, H9665, and H8892. Otherwise, the growers’ varieties were used (Table 1). The experimental design used in 1999 and 2000 for the differential irrigation treatments consisted of dividing the field length into five blocks, each block 15 beds wide and 80 m long. Bed width ranged from 5 to 5.5 feet depending on the particular grower. Width of each plot was three bed spacings. This design allowed three varieties per plot to be grown with a minimum of disruption of field area to the growers’ operation. Each treatment occurred only once in a particular bed to minimize any bed effects. Applied water of each treatment was measured at the inlet of the first block. Differential irrigation amounts were obtained by using smaller irrigation set times compared to the growers’ irrigation time. Irrigation
4
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17
Table 1 Summary of field-wide applied water and yield characteristics for all sites and yearsa Irrigation system
Variety
Applied water (mm)
Yield (Mg/ha)
Soluble solids (%)
Color
BR Sprinkler (1999)b Drip (1999)b Drip (2000) Drip (2001) Drip (2002)
H8892 H8892 Halley 3551 H9665 Peto303
427 406 427 521
81.86 103.71 78.40 71.45 109.54
5.3 6.0 5.4 4.6 4.8
24.2 21.1 23.4 25.3 24.1
DI Sprinkler (1999)b Drip (1999)b Drip (2000) Drip (2001)
H9557 H9557 H9492 H9492
564 737 582
78.85 90.94 103.94 115.80
5.2 5.0 4.8 4.9
24.8 22.8 21.0 24.1
DE Sprinkler (2000)b Drip (2000)b Drip (2001)
H9557 H9557 H8892
579 711 561
63.84 86.46 102.59
5.5 5.6 5.2
23.9 23.7 23.6
c
c
a
BR, DI, and DE identify the various sites. Comparison year. c Data not available. b
frequency was the same used by the grower. The data were analyzed by regressing total yield along the field length against cumulative applied water for each treatment. A randomized block design was used in 2001 consisting of four irrigation treatments replicated five times with each block extending along the field length. Each plot consisted of one continuous bed along the field length. The same irrigation set time and irrigation frequency were used for each treatment, but differential irrigation amounts were obtained by using different manifold pressures compared to the field system pressure. Data collected in the irrigation plots were applied water for each treatment, total red fruit yield (machine harvested), soluble solids, color; percent red, green, and non-marketable fruit; soil salinity; and weekly measurements of both canopy coverage and soil water content. Sampling locations for both soil water content and soil salinity were 0.25 m from the drip line at 0.15 m depth intervals down to 0.76–0.91 m deep (depending on site conditions) at the head, the middle, and the end of the field. The electrical conductivity of the saturated extract (ECe) was determined for each soil sample. A digital infrared camera and appropriate software were used to measure canopy coverage; a neutron moisture meter was used for soil water measurements. In addition to the soil water and soil salinity measurements made 0.25 m from the drip line, patterns of soil water content and soil salinity around the drip line were determined by a one-time sampling with depth at various distances from the drip line. A ThetaProbe was used for these cross-sectional measurements, which measures volumetric soil water content. Seasonal crop evapotranspiration (ET) was estimated using a computer ET model (Hsiao and Henderson, 1985) and reference crop (grass) ET. The model was calibrated for tomatoes with ET data from an unrelated project (Hanson, 2003). Differences between measured seasonal ET and that estimated with the model were 5% or less.
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17
5
3. Results 3.1. Field-wide comparison At each site, only 1 year of comparing drip versus sprinkler irrigation was possible (1999 for DI and BR; 2000 for DE). After the first year at each site, the rest of the field was converted to drip irrigation at BR and DE, while at DI, a different crop was planted in the rest of the field. After the comparison year, yields of the drip-irrigated fields were monitored for several additional years. The comparisons showed that the field-wide yields under drip irrigation were 12.10–22.62 Mg/ha more than those under sprinkler irrigation (Table 1). Average yields were 93.63 and 74.82 Mg/ha for drip- and sprinkler irrigation, respectively. Differences between the sprinkler and drip irrigation field-wide yields were statistically significant using the t-test with a level of significance of 5%. Drip yields were considered to be high for these fine-textured, salt-affected soils. After the first year, yields at DI and DE continued to be high (Table 1). Yields at BR for 2000 and 2001 were relatively low due to late plantings although these yields were higher than normally experienced for late plantings. A high yield was found for BR2002, which had an earlier planting date compared to 2000 and 2001. Soluble solids of the drip-irrigated fields, determined by a commercial grader, were acceptable for all years with average soluble solids of 5.3 and 5.5% for sprinkler and drip irrigation, respectively. Soluble solids increased with increasing soil salinity with average soluble solids of 4.9, 5.3, and 5.4 for DI (the lowest salinity level), BR, and DE (the highest salinity level), respectively. This behavior was similar to that found by Mitchell et al. (1991). The average color, also determined by a commercial grader, was 24.3 and 22.5 for sprinkler and drip irrigation, respectively. Differences in soluble solids and color between drip and sprinkler irrigation were not statistically significant. Applied water at BR1999 was similar for drip and sprinkler irrigation (Table 1). About 0.15 m more water was applied to the drip field compared with the sprinkler field for DE2000, partly because the drip field was irrigated for about 2 weeks longer. Applied water data for the sprinkler field at DI1999 were not available. 3.2. Differential irrigation treatments yield characteristics Results of the differential irrigation experiments showed that plot yield declined with decreasing irrigation water applications for all sites and all years although differences in behavior occurred among the sites and years (Fig. 1 for 2001 data). At DI, the overall average yield of all tomato varieties (yield differences between varieties were statistically insignificant) decreased from 105.28 to 94.08 Mg/ha as applied water decreased from 587 to 378 mm in 1999; and in 2001, it decreased from 113.12 to 103.94 Mg/ha as applied water decreased from 509 to 343 mm. At BR, 1999 yields decreased from 84.45 (730 mm of water) to 61.38 Mg/ha (508 mm), while in 2001, 64.06 Mg/ha occurred for 569 mm of water and 58.91 Mg/ha for 246 mm. DE2001 yields decreased from 110.43 to 94.75 Mg/ha as applied water decreased from 477 to 264 mm. Similar trends occurred in 2000, but the range of applied water was relatively small (data not shown). Soluble solids increased with decreasing applied water for all sites and all years (Fig. 1 for 2001 data). In 1999, overall average soluble solids at DI increased from 4.6 to 5.0% as
6
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17 120 2001
Yield (Mg/ha)
100 80 60
DI DI Regression DE DE Regression BR BR Regression
40 20
Site DI: Yield = 84.05 x AW +0.06, r2 = 0.90 Site DE: Yield = 78.51 x AW + 0.08, r2 = 0.72 Site BR: Yield = 54.06 x AW + 0.02, r2 = 0.91
0
(A)
0
100
200
300
400
500
600
6 2001 Soluble Solids (%)
5 4 3
DI DI Regression DE DE Regression BR BR Regression
2 1
Site DI: Brix = -0.00056 x AW + 5.18, r2 = 0.05 Site DE: Brix = -0.0014 x AW + 5.74, r2 = 0.57 Site BR: Brix = -0.00050 x AW + 4.92, r2 = 0.92
0
(B)
0
100
200
300
400
500
600
25 2001
Color
20 15 DI DI Regression DE DE Regression BR BR Regression
10 5
Site DI: Brix = -0.0034 x AW + 24.2, r2 = 0.36 Site DE: Brix = 0.0086 x AW + 18.7, r2 = 0.24 Site BR: Brix = 0.0016 x AW + 22.3, r2 = 0.18
0 0
(C)
100
200
300
400
500
600
Applied Water (mm)
Fig. 1. Results of the 2001 differential drip-irrigated experiment for (A) crop yield, (B) soluble solids, and (C) color. BR, DI, and DE identify the various sites.
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17
7
applied water decreased from 587 to 378 mm, while at BR, soluble solids increased from 5.9 to 6.8% as applied water decreased from 730 to 508 mm. The 2001 data showed soluble solids to increase from 4.8 to 5.2% as applied water decreased from 509 to 343 mm at DI; 5.2–5.4% as applied water decreased from 477 to 264 mm at DE; and 4.7–4.8% as applied water decreased from 568 to 246 mm at BR. Applied water had little effect on color (Fig. 1) and percent red fruit (data not shown). Linear regression equations relating yield characteristics with applied water were tested for their statistical significance and statistical similarity among the sites. Results of the statistical tests were mixed, preventing any conclusions from being developed about differences between sites and between years. There were no statistical differences in yield, solids, and color between varieties at DI in 1999 and at DE in 2000. 3.3. Water quality The electrical conductivity (EC) of the Westlands Water District irrigation water at BR and DI normally was about 0.34 dS/m. At DE, the electrical conductivity of the well water was about 1.06–1.2 dS/m. The EC of the shallow ground water at BR ranged from 4.7 to 7.4 dS/m, while at DI, it ranged from 7.9 to 11.1 dS/m for 1999 and 2000 and was 4.0–4.7 dS/m in 2001. Reasons for the small 2001 values are unknown even though sampling locations were within 10 m of each other. Ground water EC values at DE were 13.6–16.4 dS/m in 2000 and 9.0–9.5 dS/m in 2001. These differences may reflect different sampling locations in the field from year-to-year. 3.4. Water table response The water table depth at DI1999 decreased with time until about July 20 and then increased to about 1.8 m deep, while the water table remained below 1.8 m deep in 2000 and 2002 (Fig. 2). No response of water table depth to drip irrigation was evident. At BR1999, the water table depth increased from about 0.6 to about 1.3 m, but in 2000, drip irrigations caused the water table to rise to nearly 0.5 m deep before July 15, the result of applying about 10% more water than the estimated crop evapotranspiration. After mid-July, the water table depth increased to 1.52–1.83 m deep due to reduced water applications. Water table levels were not measured at DE in 2000 because of problems in installing observation wells. In 2001, water table depth at DE fluctuated between about 0.61 and 1.22 m with a definite response to drip irrigation. The gaps in the 2001 data for BR and DI were caused by the water level in the observation wells dropping below the pressure transducers, corrected by deepening the wells. 3.5. Soil salinity Soil salinity as measured by the electrical conductivity of the saturated extract (ECe) differed considerably among the three sites (Fig. 3 for 2001 data). ECe values at DI were generally less than 2.5 dS/m (threshold value for tomato). (Note: the threshold ECe value is the maximum average root zone ECe at which no yield reduction should occur (Maas and
8
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17
Depth to Water Table (m)
0.0
(A)
1999
BR
0.5 1.0 1.5 2.0 2.5
DI Jun 1
Jun 15
Jul 1
Jul 15
Aug 1
Aug 15
Sep 1 Sep 15
3.0 140 150 160 170 180 190 200 210 220 230 240 250 260
Depth to Water Table (m)
0.0
(B)
2000
BR
0.5 1.0 DI
1.5 2.0 2.5
Jun 1
Jun 15
Jul 1
Jul 15
Aug 1
Aug 15
Sep 1 Sep 15
3.0 140 150 160 170 180 190 200 210 220 230 240 250 260
Depth to Water Table (m)
0.0 2001 0.5
DE
1.0 BR
1.5 DI
2.0 2.5
Jun 1
Jun 15
Jul 1
Jul 15
Aug 1
Aug 15
Sep 1 Sep 15
3.0 140 150 160 170 180 190 200 210 220 230 240 250 260
(C)
Day of Year
Fig. 2. Depth to water table for (A) 1999, (B) 2000 and (C) 2001. BR, DI, and DE identify the various sites.
Grattan, 1999). The actual root zone salinity under drip irrigation at these sites is unknown because of spatially varying patterns of soil salinity, soil water, and probably root density around drip lines. The threshold value is provided as a reference only to indicate a potential for yield reduction.) At BR2001, ECe increased considerably with depth and exceeded the threshold value except for depths less than about 0.54 m. ECe values at DE2001 increased
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17 0.00
0.00 DI2001 Wet
Depth (m)
0.25
DI2001 Dry 0.25
1 May 15 Aug.
0.50
0.50
0.75
0.75
1.00
1 May 14 Aug
1.00 0 1 2 3 4 5 6 7 8 9 10 11 12
0.00
0 1 2 3 4 5 6 7 8 9 10 11 12 0.00
BR2001 Wet
BR2001 Dry
Depth (m)
0.25
0.25 2 May 19 Sept.
2 May 19 Sept.
0.50
0.50
0.75
0.75
1.00
1.00 0 1 2 3 4 5 6 7 8 9 10 11 12
0.00
0 1 2 3 4 5 6 7 8 9 10 11 12 0.00
DE2001 Wet
Depth (m)
9
DE2001 Dry
0.25
0.25
0.50
0.50
0.75
1 May 14 Aug
0.75 1 May 14 Aug.
1.00
1.00 0 1 2 3 4 5 6 7 8 9 10 11 12
0 1 2 3 4 5 6 7 8 9 10 11 12
ECe (dS/m)
Fig. 3. Electrical conductivity of saturated extracts (ECe) with depth for wet (receiving the most irrigation water) and dry (receiving the least irrigation water) differential drip-irrigated treatment in 2001. Dashed line is the reference salinity threshold value for tomato. Units are dS/m.
with depth with all values exceeding the threshold value. At all sites, differences between wet and dry irrigation treatments were slight. (In the differential drip-irrigated experiment, the “wet” treatment received the most irrigation water and the “dry” treatment received the least irrigation water.)
10
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17
Depth (mm)
DI2000
-200 Drip Line
ECe (dS/m)
-400
-600
a -600
(A)
-400
-200
0
200
400
600
Distance From Drip Line (mm)
Depth (mm)
BR2000
-200 Drip Line -400
-600
b -600
(B)
12 11 10 9 8 7 6 5 4 3 2 1 0
-400
-200
0
200
400
600
Distance From Drip Line (mm)
Fig. 4. Patterns of electrical conductivity of saturated extracts (ECe) around the drip line for (A) DI2000 and (B) BR2000. Units are dS/m.
The pattern of ECe around the drip line showed values less than the threshold value throughout the soil profile at DI (Fig. 4A). At BR2000, salinity was the least near the drip line with values less than about 1 dS/m, but salinity increased with horizontal distance from the drip line and with depth to values of about 7 dS/m (Fig. 4B). Salinity also increased above the drip line as depth decreased indicating salt accumulation above the drip line. The zone of ECe values less than the threshold value extended about 0.41 m horizontally from the drip line and about 0.21 m deep below the drip line. At DE2000, ECe was the highest near the drip line with values of 3–4 dS/m and decreased with horizontal distance to values less than 2.5 dS/m at distances beyond about 0.21–0.42 m (Fig. 5A). Salinity above the drip line increased as depth decreased. The high salinity near the drip line reflected the well water EC. The low levels of salinity near the edge of the pattern probably reflected leaching of salts due to ponding from a severe late spring rain. In 2001, ECe levels ranged between 5 and 7 dS/m throughout most of the soil profile except near the drip line where ECe values were between 3 and 4 dS/m (Fig. 5B). The main source of salt in these fields is from the upward flow of saline, shallow ground water into the root zone. However, poor correlation was found between soil salinity near the
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17
11
DE2000
Depth (mm)
Drip Line
ECe (dS/m)
-200
-400 a -600
(A) -600
-400
-200
0
200
400
600
Distance From Drip Line (mm)
Depth (mm)
DE2001 Drip Line
-200
-400
-600
b -600
(B)
12 11 10 9 8 7 6 5 4 3 2 1 0
-400
-200
0
200
400
600
Distance From Drip Line (mm)
Fig. 5. Patterns of electrical conductivity of saturated extracts (ECe) around the drip line for (A) DE2000 and (B) DE2001. Units are dS/m.
bottom of the sampled soil profile and ground water salinity. At DI, soil salinity at the deeper depths of the sampled soil was generally was less than 2 dS/m, but the ground water salinity was much higher. At BR, soil salinity levels at the deeper sampled depths were similar to the ground water salinity; while at DE, soil salinity was less than the ground water salinity. Reasons for the behavior at DI and DE are not clear, but the deeper water table depth at DI (generally 1.82 m or deeper) probably contributed to the smaller soil salinity values, whereas at the other sites, much smaller water table depths occurred. The deeper depth at DI may have greatly reduced upward flow of shallow ground water into the root zone. 3.6. Soil water Soil water content decreased with time throughout the irrigation season for all irrigation treatments (data not shown). Water contents at the start of the measurement period were generally between 37 and 43% and decreased to between 27 and 38% prior to harvest. Average water contents of the wet irrigation treatments were slightly higher than those of the dry irrigation treatments. Seasonal changes in soil water content, estimated from the neutron moisture meter data, ranged from 33 to 86 mm. Wetting patterns around the drip line showed water moving laterally to about 0.41 m from the drip line at DI (Fig. 6) and BR (data not shown). At about that distance, soil
12
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17
DI1999 (July 27)
0 -200
Drip Line
-400
(%)
-600
50
Depth (mm)
45
-800
Wet
(A) 0
200
40
400
600
800 1000 1200 1400 1600
35 30
0
25 20
-200 Drip Line
15 10
-400
5 0
-600 Dry
0
200
400
(B)
600 800 1000 1200 1400 1600 Distance Across Bed (mm)
Fig. 6. Patterns of soil water content (%) around drip lines for (A) wet (receiving the most irrigation water) and (B) dry (receiving the least irrigation water) differential drip-irrigation treatments in DI1999. Units are volumetric soil water content in %.
water content was the least for a given depth. Soil water content increased with depth, but changes with depth were small below the drip line. At about 0.51 m from the drip line (in the furrow), slightly higher soil water contents occurred compared to 0.41 m suggesting less water extraction at 0.51 m compared to 0.41 m. Soil water contents above about 0.38–0.51 m deep were less for the dry treatment compared to the wet treatment. Similar behavior occurred at BR. Wetting patterns at DE were not measured, but based on the salinity pattern in Fig. 5A, lateral flow was between 0.20 and 0.41 m from the drip line. 3.7. Crop evapotranspiration A variety of canopy growth curves were found with maximum coverage ranging from 74 to 99% (data not shown). Canopy growth curves showed a rapid increase in canopy development starting between 20 and 50 DAP (days after planting). Maximum canopy coverage generally occurred between 60 and 80 DAP. In some case, canopy coverage decreased later
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17
13
100 2001
Canopy Coverage (%)
90 80 70 60 50 40 30
DE DI BR
20 10 0 0
20
40
60
(A)
80
100
120
140
Days After Planting
Crop Evapotranspiration (mm/day)
10 2001 8
6 BR DE DI ETo
4
2 Apr 1
May 1
Jun 1
Jul 1
Aug 1
Sep 1
0 80
(B)
100
120
140
160
180
200
220
240
260
280
Day of Year (DOY)
Fig. 7. (A) canopy coverage, and (B) crop evapotranspiration for 2001. BR, DI, and DE identify the various sites.
in the season due to pruning of vines by growers and canopy aging. No differences in canopy coverage were found between the irrigation treatments. The canopy growth curves of 2001 illustrate the effect of different cultural practices on canopy growth and crop evapotranspiration (Fig. 7A). The planting date of DE (April 10) was much earlier than those of BR (May 5) and DI (May 2). At both DE and BR, plants were direct-seeded, while transplants were used at DI. Sprinkler irrigation was used for stand establishment at BR and DE, while drip irrigation was used at DI. Canopy growth at DI was about 10–20 days ahead of that of the direct-seeded plants at BR (Fig. 7A). High evapotranspiration occurred just after planting at BR and DE due to evaporation from soil wetted by sprinkler irrigation (Fig. 7B). Very small ET rates occurred at DI just after planting because of the dry soil surface at that time due to using drip irrigation for stand establishment. Maximum ET rates occurred at about DOY170 (day of year) at BR, DOY180 at DI,
14
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17
Table 2 Total amount of applied water, cumulative crop ET, and irrigation efficiency (IE) Year
Applied water (mm)
Seasonal change in soil water (mm)
Cumulative crop ET (mm)
IE (%)
BR 1999 2000 2001
406 427 521
86 69 69
516 544 582
105 110 99
DI 1999 2000 2001
564 737 582
38 86 76
638 640 676
106 78 103
DE 2000 2001
711 561
33 81
615 587
83 91
The irrigation efficiency is the ratio of the cumulative crop ET to the sum of applied water and seasonal soil water change. BR, DI, and DE identify the various sites.
and DOY190 at BR. ET rates were similar during the mid-season growth stage, and then decreased during the late-season stage at all sites, primarily the results of canopy pruning. Seasonal cumulative ET for all years calculated using the computer model and canopy growth curves showed ET values to range from 516 to 676 mm (Table 2). Seasonal irrigation efficiency, defined as the ratio of cumulative ET to the sum of the cumulative applied water and seasonal change in soil water, ranged from 78 to 110%. Values near or exceeding 100% indicate deficit irrigation and plausible use of the shallow ground water. 3.8. Economics The economics of converting to subsurface drip irrigation from sprinkler irrigation was examined using cost data provided by one of the grower participants. Assumptions used in this analysis were: • The existing sprinkler irrigation system was used elsewhere on the farm. • The economic life of the drip system was 20 years. • Replacement of the drip tape occurred every 5 years. This replacement schedule is based on the growers’ practice in this area. • Filters and pumps were replaced every 10 years. • Drip irrigation yield increases ranged from 12.90 to 22.62 Mg/ha compared to sprinkler irrigation (Table 1). • Equivalent annual capital cost of the drip irrigation system was determined for interest rates of 5 and 10%. • The same amount of irrigation water was applied by both irrigation methods. • Area irrigated was 32.4 ha. The benefits of converting to drip irrigation were increased revenue from higher yields and annual savings of cultural costs and energy costs of sprinkler irrigation. The costs of
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17
15
Table 3 Economic analysis of benefits and cost ($ ha−1 ) of converting from an existing sprinkler irrigation system to a subsurface drip system for interest rates of 5 and 10% Interest rate (%) 5
10
Benefits Revenue increase Savings of sprinkler energy costs Savings of sprinkler cultural costs Subtotal
709a , 1244b 141 1568 2418a , 2958b
709a , 1244b 141 1568 2418a , 2958b
Costs ($ ha−1 ) Equivalent annual capital cost of drip system Drip energy costs Drip cultural costs Subtotal
296 69 1099 1465
383 69 1099 1551
($ ha−1 )
Net returns ($ ha−1 )
953a , 1493b
867a , 1407b
Yield increases of 12.90 and 22.62 Mg/ha were used for this analysis. Crop price was $ 55/Mg. a Yield increase of 12.90 Mg/ha. b Yield increase of 22.62 Mg/ha.
the conversion were the equivalent annualized capital cost of the drip system and its annual cultural costs and energy costs. Annual net returns ranged from 953 to $ 1493 ha−1 for a 5% interest rate and from 867 to $ 1407 ha−1 for a 10% rate (Table 3). Returns to land, farm management costs, taxes, and insurance costs (data not available) were not included. Capital cost of the drip system was $ 1998 ha−1 . The equivalent annual capital cost of the drip system was 296 and $ 383 ha−1 for the 5 and 10% interest rates, respectively. Reasons for the higher profits under drip irrigation were mainly due to higher yields, and thus higher revenue, and lower cultural costs compared to sprinkler irrigation. Reduced weeding costs and eliminating the costs of setting up and moving hand-move sprinklers contributed to the smaller cultural costs for drip irrigation. This economic analysis assumed a drip line replacement every 5 years. However, drip line replacements of 10 years or more have been reported. Assuming a 10-year replacement schedule resulted in an equivalent annual capital cost of $ 316 ha−1 for the drip system at a 10% interest rate compared to $ 383 ha−1 for the 5-year replacement.
4. Discussion and conclusions Subsurface drip irrigation in these fine-textured salt-affected soils can increase yield and profit of tomatoes compared to sprinkler irrigation with acceptable levels of soluble solids (mainly due to the soil salinity at these locations). Drip irrigation also can control subsurface drainage to the shallow ground water. Little correlation was found between soil salinity and crop yield over the range of ECe’s found at these sites even though ECe values higher than the threshold ECe were found around the drip line at one site, suggesting soil salinity under drip irrigation may affect crop yield less compared to other irrigation methods, as
16
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17
suggested by Shalhevet (1994). Subsurface drip irrigation also provided better late-season water management, a time when careful water management is needed to prevent excessive deficit irrigation or phytophthora from excessive wet soil. Little, if any, water savings on a per hectare basis are likely to occur by converting to drip irrigation from sprinkler irrigation. The higher yields of the drip irrigation suggest increased evapotranspiration compared to sprinkler irrigation. This ET increase may offset any water savings due to reduced percolation and evaporation losses found with drip irrigation compared to those losses under sprinkler irrigation. However, because higher yields occurred under subsurface drip irrigation, the same total yield could be grown on fewer hectares compared to sprinkler irrigation, which would save water. The results of the differential irrigation treatments showed that subsurface drip irrigation must be carefully managed to prevent yield reduction and excessive drainage to the ground water. Recommended irrigation amounts are near 100% of the potential crop ET as a compromise between reducing drainage and controlling soil salinity in the root zone. Irrigation should occur two to three times per week (Hanson et al., 2003). In summary, the long-term sustainability of tomato yield under subsurface drip irrigation in these salt-affected soils will require the following: • Sufficient leaching must occur to maintain acceptable levels of soil salinity near the drip lines where the root density is probably the greatest. • Periodic leaching of salt accumulated above the buried drip lines with sprinklers will be necessary for stand establishment if winter and spring rainfall is insufficient to leach the salts. • Careful management of irrigation water will be required to apply sufficient water for crop evapotranspiration and leaching yet prevent excessive subsurface drainage. • Periodic system maintenance must be performed to prevent clogging of drip lines. Clogging due to root intrusion was found to be a severe problem at one site where little or no chlorination occurred. Clogging will not only reduce the applied water needed for crop ET, but also reduce the leaching. Subsurface drip irrigation in these marginal soils is very profitable, which has encouraged the conversion to subsurface drip irrigation from sprinkler irrigation in this area. However, where high tomato yields are obtained under furrow and sprinkler irrigation, converting to drip irrigation may not be profitable because the potential for large yield increases under drip irrigation may not exist. Any increase in revenue under drip irrigation may be insufficient to offset capital, energy, maintenance, and management costs of subsurface drip irrigation. Also, using drip irrigation on lower-valued crops also may be unprofitable even if yield increases occur because the relatively small revenue from these crops may be insufficient to offset the costs of drip irrigation.
Acknowledgements We wish to acknowledge the contributions of the US Bureau of Reclamation, Westlands Water District, Britz Farming Corp., Farming D. Inc., Agri-Valley Irrigation Company,
B. Hanson, D. May / Agricultural Water Management 68 (2004) 1–17
17
T-Systems International, Netafim Irrigation Company, Roberts Irrigation Products, Rain Bird, Toro Ag, and the University of California Salinity/Drainage Program.
References Ayars, J.E., Schoneman, R.A., Dale, F., Meson, B., Shouse, P., 2001. Managing subsurface drip irrigation in the presence of shallow ground water. Agric. Water Manage. 47, 242–264. Fulton, A.E., Oster, J.D., Hanson, B.R., Phene, C.J., Goldhamer, D.A., 1991. Reducing drains water: furrow vs. subsurface drip irrigation. Calif. Agric. 45 (2), 4–8. Hanson, B.R., 2003. Evapotranspiration of tomato under furrow/sprinkler and subsurface irrigation. HortScience 38 (5), 759. Hanson, B.R., Ayars, J.E., 2002. Strategies for reducing subsurface drainage in irrigated agriculture through improved irrigation. Irrigation Drainage Syst. 16, 261–277. Hanson, B.R., May, D.M., Schwankl, L.J., 2003. Effect of irrigation frequency on subsurface drip irrigated vegetables. HortTechnology 13 (1), 115–120. Hsiao, T.C., Henderson, D.W., 1985. Improvement of crop coefficients for evapotranspiration. California Irrigation Management Information System Final Report. Maas, E.V., Grattan, S.R., 1999. Crop yields as affected by salinity. In: Skaggs, R.W., van Schilfgaarde, J. (Eds.), Agricultural Drainage, Agronomy. Monograph 38. ASA, CSSA, SSSA, Madison, WI. Mitchell, J.P., Shennan, C., Grattan, S.R., May, D.M., 1991. Tomato fruit yields and quality under water deficit and salinity. J. Am. Soc. Hort. Sci. 116 (2), 215–221. Pasternak, D., De Malach, Y., Borovic, I., 1986. Irrigation with brackish water under desert conditions. VII. Effect of time of application of brackish water on production of tomatoes (Lycopersion esculentum Mill.). Agric. Water Manage. 12, 149–158. Shalhevet, J., 1994. Using water of marginal quality for crop production: major issues. Agric. Water Manage. 25, 233–269. Styles, S., Oster, J.D., Bernaxconi, P., Fulton, A., Phene, C., 1997. Demonstration of emerging technologies. In: Guitjens, J., Dudley, L. (Eds.), Agroecosystems: Sources, Control and Remediation. Pacific Division, Am. Assoc. Adv. Sci., San Francisco, CA, pp. 183–206.