Assessing Sustainability of Shrimp Aquaculture and Integration with a Field Crop

Final Report for SW01-062

Project Type: Research and Education
Funds awarded in 2001: $68,523.00
Projected End Date: 12/31/2004
Matching Non-Federal Funds: $20,550.00
Region: Western
State: Arizona
Principal Investigator:
Kevin Fitzsimmons
Univ of AZ Environmental Research Lab
Expand All

Project Information

Abstract:

The project focused on integrating aquaculture with irrigated agriculture. Our primary site was a shrimp farm in southwestern Arizona that used effluent from the farm to irrigate olive trees. At that site we demonstrated that the effluent could be used benficially to grow olive trees. There was no significant difference in growth rate of trees receiving effluent compared to trees receiving well water with fertilizer. However, the effluent treated trees did grow significantly greater than trees that received only well water, with no fertilizer added.

A second experiment, conducted in a greenhouse at the Environmental Research Lab in Tucson, utilized sludge collected from the bottom of harvested shrimp ponds as a soil amendment for tomatos. Tomato plants with the sludge soil amendment produced signifcantly greater amounts of tomato fruit compared to plants grown in unamended soil.

A third trial was conducted at the Maricopa Agriculture Center in central Arizona. In this trial, effluent from a fish pond, stocked with koi and tilapia, was used to irrigate a field planted with barley for two winters and cotton planted for two summers.

Project Objectives:

There are five specific objectives in undertaking this project:
1. Determine the benefits of irrigating olives with low-salinity aquacultural effluents by measuring growth of trees
2. Determine any detrimental effects on soil caused by the application of saline irrigation water through the monitoring of soil salinity and macro-nutrients
3. Reduce the reliance on chemical fertilizers through close monitoring of nutrients applied and through the application of nutrient rich aquacultural effluents
4. Efficient utilization of scarce water resources through the multiple use of water for shrimp production and irrigation
5. Initiate an integrated aquaculture/agriculture extension program in Arizona by hosting an integrated agriculture field day, distributing a newsletter, and developing a bulletin and website reporting the findings of the research and attitudes of the farmers involved with the trials.

Introduction:

In 2001, global aquaculture production exceeded 37 million tons (FAO 2002), and it has become a continuingly growing multi-billion dollar per year industry in the United States. In the southwestern U.S., the largest sectors of agriculture are uniquely situated in relation to aquaculture in that many necessary input resource requirements of field agriculture can be provided through utilizing the waste outputs of aquaculture. Therefore, the agriculture industry in southwestern states has a great potential to benefit from incorporating aquaculture into its current management practices.

In 1998, aquaculture produced 25% of the total shrimp supply (FAO, 1998). Demand for shrimp and high market value have led to rapid expansion in shrimp aquaculture (Deb, 1998). Use of coastal areas for aquaculture often conflicts with other users, such as recreationalists and homeowners, or impacts sensitive mangrove habitat (Alongi, 2002). The high concentration of farms in coastal areas also leads to a self-polluting industry (Corea et al., 1998), as nutrient rich waters or disease exit one farm near the intake of the next. As a result of these factors, inquiries into the feasibility of inland low-salinity aquaculture operations are becoming more common (Smith and Lawrence, 1990; Flaherty and Vandergeest, 1998; Flaherty et al., 2000). Inland aquaculture reduces coastal conflicts and the risk of disease (Menasveta, 2002), but requires a new approach to water management.

Several studies have focused on water quality parameters and acclimation of marine shrimp for inland growth in low salinity water (McGraw et al., 2002; McGraw and Scarpa, 2003; Saoud et al., 2003). Lab and field experience has shown the importance of maintaining proper mineral ratios in the water (Zhu et al., 2004), and has identified potential areas to establish aquaculture facilities based on water quality. Reduced water exchange in inland culture should not affect yields (Thakur and Lin, 2003), but total shrimp farm groundwater use has not been quantified.

In achieving a more effective use of aquaculture pond effluent, farmers could make an important contribution to agricultural industries in arid regions, particularly by maximizing farm production without increasing water consumption (McIntosh & Fitzsimmons 2003). In many areas, integrating aquaculture with agriculture has become a channel for increasing the use of limited water resources (Prinsloo & Schoonbee 1993; Ingram et al. 2000), decreasing dependence on chemical fertilizers (Fernando & Halwart 2000), and providing a greater economic return per unit of water. In Arizona alone, over a trillion liters of irrigation water per year are used in agriculture that could be used for aquaculture first. In return, fish waste and algae production have the potential to sustain crop growth and yield, while lowering the usage and cost of chemical fertilization. There is a great need for an environmentally safe best management practice that will alleviate the use of these fertilizers, but which will also continue to profit growers.

Effluent and sludge from aquaculture ponds can act as valuable inputs into arid climate agricultural systems where natural irrigation water may be insufficient and depleted of nutrients (Hosetti & Frost 1995). Sludges, which may be discharged with liquid effluent, are most often comprised of sand, mud, uneaten feed, particulate matter, organic matter, bacteria, dead algal biomass, or fish waste. They can become thick, hard to move, and difficult to transport or collect. Algae, which assimilate available, aqueous forms of N and P into biomass, may settle at the bottom of ponds after death, sequestering nutrients. In areas where clays line the bottom of ponds, ammonium ions are weakly attracted to and retained by negatively charged cation exchange sites of clays and organic colloids in pond bottom solids (Boyd & Tucker 1998). Phosphates are more strongly adsorbed by pond muds, which become the eventual recipients of most PO4 added through feed input (Hepher 1958). Adsorption of both N and P to pond muds or poor spatial distribution may allow the benthic region of an aquaculture pond to become a temporary nutrient sink, which is unfortunate because they are cost-effective by-products useful for agriculture and they represent valuable sustainable resources as raw materials (Hosetti & Frost 1995).

It is not yet clear as to what kinds of integrative aquaculture-agriculture (IAA) systems are able to produce sustainable, positive production increases of field crops through land application of effluents. Regardless of increases, there may still be economic gain using IAA through simply maintaining original field crop yields and then raising fish in excess water to sell for profit. Stocking fish crops into ponds or irrigation reservoirs can provide greater economic security for farmers through diversification, as well as provide further income that would help to offset heavy irrigation costs of farming in the desert. It has been proposed that even a simple IAA system, such as introducing high densities of fish into irrigation systems, may enhance land crops, alleviate the pressure of terrestrial and aquatic pests, and lower the populations of vectors of diseases of man and domestic animals (Fernando & Halwart 2000). In Arizona and other arid regions, where water costs continue to rise and where crop values are considered marginal, it is extremely sensible for fish to be cultured in pre-irrigation water before it is used to irrigate crops (Fitzsimmons 1988; Brooks 1989; Budhabhatti 1991; Fitzsimmons 1992).

Previous studies have shown that synergisms existing between inland aquaculture and agriculture contribute to positive ecological and environmental effects (Bacon et al. 1993; Edwards 1993; Gomiero et al. 1999; Fernando & Halwart 2000; Jamu & Piedrahita 2002; Prein 2002). Integrative methods used in some of these studies may allow for nutrient recycling of otherwise unused waste materials, nutrient and energy recovery, better sanitation, increased natural resource efficiency, low environmental loading, and little dependence on fossil energy inputs. Other studies have shown that utilizing an integrated system has the potential to positively affect crop production, net income, and/or sustainability (Bacon et al 1993; Hopkins & Bowman 1993; Lightfoot et al. 1993; Olsen et al. 1993; Shereif et al. 1995; Mosher 1996; Dalsgaard & Oficial 1997; Devendra 1997; Duong et al. 1998; Edwards 1998; Hosetti and Frost 1995; ICLARM 2000; Fernando & Halwart 2001; Prein 2002; McIntosh and Fitzsimmons 2003).

In 1993, a study by Olsen et al. in Maricopa, AZ, found that by integrating tilapia and channel catfish farming with cotton farming, there was an increase in total nitrogen (N) and phosphorous (P) content of irrigation water. However, effluent irrigations did not produce higher cotton yields or increase N and P in the soil. This was attributed to low fish densities and, therefore, low effluent concentrations. Our research has built upon this study and has taken into account the fact that concentrations of N and P in many integrated systems may be limited by two things working in conjunction: 1.) large quantities of diatoms, cyanobacteria, and green algae, which assimilate available nutrients in the water and 2.) the inability of most pond systems to effectively transport nutrients in the form of sludge (sediment, ammonia, urea, solid fish waste, and dead algal biomass) from a pond to a field. With respect to pond nutrient content, specific attention must be paid, not only to feed input, absorption, and excretion, but also to all algae and bacteria communities because of their enormous impact on water quality and nutrient flow within aquatic systems.

The aim of our studies were to use integrated systems that would allow us to determine the effects of fish and shrimp effluent irrigations on the growth and yield of field crops in an arid region. This was accomplished by making comparisons between different irrigation and fertilization treatments. We also measured the nutrient content of irrigation water in an IAA system as a function of increasing fish densities to compare with studies done by Olsen et al. (1993). Through this study, we can determine if healthy and profitable amounts of fish and field crops may be grown in this sort of pond-field system. Perhaps most importantly, we plan on discussing the significance of pond biosolids in the IAA system. Each study was designed so that biosolids not exported during effluent discharge could accumulate at the bottom of the pond. Therefore we have made a determination as to what effect the collection and application (or lack thereof) would have upon the system as a whole. Finally, we plan to discuss the environmental and economic benefits of this particular system and make assumptions about the sustainability of IAA in arid regions. Within IAA, sustainability has become a major objective, and is surrounded by many questions. Prein (2002) notes that on-farm performance of IAA systems have been successful, yet not sustainable in large scale systems, but Jamu and Piedrahita (2002) argue that IAA systems are increasingly being promoted as an environmentally sustainable method for producing aquatic and terrestrial crops. Therefore, these studies were designed in order to further test the prospects and procedures.

.

Cooperators

Click linked name(s) to expand
  • Craig Collins
  • Kevin Fitzsimmons
  • Chad King
  • Dennis McIntosh
  • Kalb Stevenson

Research

Materials and methods:

Shrimp Trials

Experimental Design
We hypothesized that the use of shrimp effluent as an irrigation source would increase olive tree growth and water use efficiency over the use of well water, with no detrimental effects on soil salinity and productivity. We compared differences in tree height as a response to irrigation treatments between effluent, well water, and irrigation with standard fertilization. The goal was to determine changes in growth due to the effluent irrigation and to compare those changes to growth expected with recommended fertilizer application. The economic savings from growing two crops with the same water was also examined.

An experimental plot covering 0.133 ha (0.329 acres) was laid out on a commercial shrimp farm growing Litopenaeus vannamei, the Pacific white shrimp, in Gila Bend, Arizona. Soils in this area have been classified as a torrifluvent association (Hendricks, 1985). The experimental plot was isolated from other olive groves, and the top layer of soil had been removed and used as a source of soil during pond building. Olive trees (one year old from cuttings) were planted in ten rows of twelve trees (120 trees total)(Figure 1). The design was a randomized complete block. It was unbalanced with respect to the effluent treatment in order to gain more knowledge about response to effluent. The treatment assigned to each row was randomized by lottery, and trees to be planted within the rows were selected randomly, with order assigned by a random number generator. Each row was an experimental unit, with data reported as mean height for trees in each row. There was not a significant difference in tree height between treatments at the beginning of the study (F2, 117 = 0.31, p = 0.73).

The experiment was designed to approximate farm conditions. For this reason, trees were placed in rows receiving furrow irrigation. We planted as many rows as would fit across the experimental plot, with the extra row assigned as an effluent replicate to gain more information on response to effluent. Trees were planted in the bottom of a single furrow 30 cm wide and 30 cm deep, and watered by flood irrigation. Tree height was measured monthly, from a mark painted on the trunk, five cm above the original soil level, to the end of the longest branch. Trunk diameter was also measured initially, but was found to vary considerably depending on placement of the calipers. Due to this variability, this measurement was abandoned.

Irrigation

Trees were irrigated every week in the summer and every third week in the winter, approximating farm procedures. On the rest of the farm, trees were irrigated weekly in the summer, and as trees showed signs of water deficiency the rest of the year. Irrigation rates were 2.5 cm for each application from March through May and October through December, and five cm for each application from May through October. During shrimp production (approximately June to October), effluent from pond water discharge was used to irrigate the effluent treatment rows. The well water + fertilization treatment groups received urea fertilizer applications with the scheduled fertilizer applications for the rest of the farm (March through April). The rest of the year, all trees received well water.

Fertilizer was applied in four applications the first year and five during the second, with a target of a total of 0.23 kg of N per tree per year, or 188 kg/ha. This is half of what is recommended in the literature for large olive trees (Freeman et al., 1994), to account for the small starting size. In year one, 1.64 kg of urea (45% N) was applied per row in four applications, and a total of 10 cm of irrigation water (a rate of 112 kg/ha). In year two, five fertilizer treatments totaling 5.56 kg urea/row were applied in 12.5 cm of irrigation water (371 kg/ha), the full recommendation for olive trees. Urea was mixed with well water in 7,571-L water tanks before application in irrigation water.

Duplicate water samples were taken for each treatment during every irrigation event to determine levels of nitrogen and salinity addition. A HACH DR-890 spectrophotometer (Hach Co., Loveland, CO) was used to analyze the samples (Table 1) for ammonia-nitrogen (NH3-N), nitrite-nitrogen (NO2-N), nitrate-nitrogen (NO3-N) and total nitrogen. To confirm the nitrate-nitrogen results, a standard curve was developed, and all nitrate-nitrogen samples were adjusted accordingly.

Statistical methods

We compared mean tree growth among treatments from beginning to end of the experiment and the mean water quality parameters, using a one-way analysis of variance (ANOVA). We performed all analyses with JMP IN 4 statistical software (SAS Institute Inc., Pacific Grove, CA)

Fish-Pond Trials

Construction of the IAA system began in 2001at the Maricopa Agricultural Center (M.A.C.) in Maricopa, AZ. An elevated, oval-shaped pond holding 1.8x106 L of water was used as an irrigation reservoir. A drain manifold was constructed using long PVC pipe, and uniform perforations were made along the top and sides of the entire pipe to allow for nutrient extraction from a wider distribution of the pond area. An aerator and floating stationary cages were added to the pond and fastened into place. Fish were then added to the pond, both in the cages and free-swimming, depending on the species. Water was siphoned from an adjacent reservoir in order to replace evaporated or discharged water as needed. Water discharge occurred only during irrigation events, and was carefully directed onto randomized plots of an agricultural field through slotted irrigation pipe. Some of the randomized plots received pond effluent applications, which were pumped from the pond near the southwestern corner of the field, while the others received well water applications, which were tapped from an alfalfa well at the eastern end of the field. The same slotted irrigation pipe was used for each treatment, with the water source location being the only variable factor.
The agricultural portion of the study was sectioned into a randomized complete block (RCB) with four treatments and four repetitions in a 4x2 factorial design (four treatments x two types of field crops). The four treatments differed according to their source of irrigation and chemical fertilizer application, and are defined as (1) well water irrigations only (W.W.), (2) well water irrigations + standard chemical fertilizer applications (W.W. + S.F.), (3) fish effluent irrigations only (F.E.), and (4) fish effluent irrigations + standard chemical fertilizer applications (F.E. + S.F.). Sixteen total plots were used, each containing a surface area of 0.024 ha with dimensions ([6]1.02m rows x 39.6m). Of the six rows in each plot, only plants in the center four were used in the study. During barley season, however, a 5.5m combine harvested the center of the 6.1m wide plots. Therefore, each harvestable plot area was approximately 0.021ha for cotton trials and 0.022ha for barley trials.

Prior to the study, the field was tilled, lasered, disked, and groomed with a bed-shaper. The soil type in the field was classified as sandy loam, and herbicide, insecticide, and defoliant applications were used as needed. For plots requiring chemical fertilizer treatment, ammonium sulfate (21-0-0) was applied as needed during all growing seasons, and monoammonium phosphate (11-52-0) was applied prior to each barley season. Approximately 106kgN/ha and 1157kgP/ha during every barley season, and 179 kgN/ha during cotton season was applied using chemical fertilizers. After each field crop harvest, a conservation tillage system was implemented, in which soils were tilled horizontally within their own plots in order to reduce contamination.

Irrigation
McCrometers were fastened onto the pipes nearest to the functioning water sources on each side of the field to measure the total volume and rate of water flow onto the agricultural field. The 6-row plots were diked along the edges and corners of rows one and six to allow for equal distribution of nutrients within each plot and to eliminate treatment contamination between plots. Barley and cotton were both watered up by allowing irrigation water to reach the end of the row and slowly rise to cover the tops of irrigation beds, thereby saturating the soil around the planted seeds completely. Irrigation scheduling was outlined using data from the Arizona Meteorological Network for the city of Maricopa, and AZSCHED, a software program designed to manage and schedule watering events. Irrigation most often occurred when the amount of water used was at 50% depletion of plant available water (PAW) in the rooting zone. Multiple water samples were also taken from each irrigation treatment to determine the concentration of nutrients in the water applied to the field. Initially, the applied volume of water (per irrigation and total) differed between treatments of well water and pond water because of flow-rate differences between the gas-powered pump and the alfalfa well, but during the second half of the study, BMPs were developed to increase flow rates for well water treatments. A total of 667.6cm of water were applied to the 0.386 ha research field over three cropping seasons. Because of slight differences in water pressure, 17.98% more water was applied to plots receiving well water. The amount of N and P applied through irrigation was adjusted proportionally by increasing nutrient concentrations of effluent by the same percentage.

At the start of each irrigation event the effluent was concentrated and black in color, but then became slightly less concentrated. Therefore, time trials were performed on 8/25/02 to determine the rate of change in nutrient concentration over time during effluent irrigations. It was concluded from these trials that the surge of concentrated water had a negligible effect upon the total amount of N and P added to the field during each irrigation event.

Fish Cropping
Different species of fish were stocked into and harvested from the pond throughout the year (see Table 1). Koi (Cyprinus carpio), tilapia (Oreochromis niloticus), and channel catfish (Ictalurus punctatus) were selected for this study because of their abundance in M.A.C ponds and because of their pertinence to U.S. arid lands aquaculture. Tilapia and catfish are popular in inland aquaculture because of their favorable food market values, while koi are raised primarily for ornamental purposes.

Koi fingerlings were purchased from an independent distributor (Pisces Aquaculture Inc.) and were stocked into the research pond, while juvenile tilapia and catfish were seasonally added to floating cages in the north end of the pond. The 765m3 floating cages were made of mesh wire, metal, and styrofoam. They were tied together and fastened to the edges of the pond using rope and iron stakes. Additional juvenile and adult koi were added at different times during the season to increase the total pond biomass.

All fish were fed 2-3% of their biomass once per day as recommended by Tucker and Robinson (1990), five days per week, from 12/27/01 through 4/17/03 with a floating aquaculture feed (see Table 2 for nutritional information). Caged fish were fed through long pipes that allowed the feed to drop through and remain in the cage. A total of 449.6kg of feed were given over a 441d period. Fish were also allowed to feed upon pond algae, which were prolific during the summer. The pond’s pH level was routinely monitored and was found to be between 7.66 and 8.01 on a consistent basis. The concentration of total dissolved salts also slightly increased over the course of the study.

Field Cropping
Short-season barley (Poco variety) was planted on 12/21/01 using a JD 8200 planter with a seeding rate of 122kg/ha. Seeds were planted 4cm deep on rows and in furrows with 18cm spacing. During the first barley season, chemical fertilizers for selected treatments were deposited onto the field using a standard applicator prior to irrigation. In Arizona, short-season barley requires one to three irrigations after emergence, whereas late season cotton can require between eight and ten. Although it is not common to find barley planted on rows and in furrows, it was necessary to select a crop that would fit into a rotational-crop system with late-season upland cotton, an important agricultural product of central Arizona. Yield and repetitive growth measurements were collected and calculated from barley plants at random (seeTable 3 for all field crop measurements and techniques). Poco barley was harvested on 5/13/02 using a 5.5m wide International 1440 axial flow combine and weigh wagon.

DP-458 BR late-season cotton was planted using a MF-4263 Monosoem planter on 5/15/02 at a seeding rate of 83kg/ha. During the cotton season, both mechanical side-dressing techniques and manual techniques were used in the application of chemical fertilizers. Cotton was harvested using a four-row spindle picker, sampled for fiber analysis, and weighed using an SK-CrustBuster Boll Buggy, and harvested stalks were extracted using a root puller. DP-458 BR cotton was selected for this study based on its importance to the Arizona agricultural market, and because of the ease with which it can permit a field to be double cropped, allowing for year-round production. Yield and repetitive growth measurements were collected and calculated from cotton plants at random (see Table 3).

A second term of short-season barley (Quick variety) was planted on 12/19/02 using methods identical to those used for Poco Barley. Quick barley was deemed most similar to the Poco variety, which was taken off the market in 2002. Because of heavy rains in the winter of 2003, Quick barley received one less irrigation than Poco barley during the 2001-2002 season. Quick barley was harvested on 4/29/03. Heavy and unexpected rains fell during the winter of 2003 and did not allow for the necessary application of 2,4-D herbicide. Therefore, weed production increased and the number of total barley irrigations decreased from four to three. The abundance of weeds caused some proportional clogging of the grain hopper weigh station, but final yields were able to be adjusted proportionally.

Fiber Quality Analysis
Ginned samples (400g) of harvested cotton were sent to USDA labs in Phoenix, AZ for fiber quality analysis and were tested for color grade (a measure of cotton whiteness), micronaire (a measure of fiber thickness or fineness), staple (a measure of how long cotton fibers reach in a season), strength (how difficult it is to break a single fiber), and uniformity (consistency and similarity between cotton fibers). These measurements are used by ginning and processing plants to determine what prices and discounts cotton farmers will receive. For instance, cotton fibers with low micronaire ratings will receive more money per bale of cotton than fibers with high micronaire ratings; favorable staple and ratings combined with a low color grade will produce average returns for farmers. These analyses were important in determining if fish-effluent irrigations would affect cotton fiber quality.

Water Analysis
Three to four water samples were taken at each irrigation event of both fish effluent and well water. Concentrated sulfuric acid (H2SO4) was added to every other sample in order to eliminate any contaminating effects of algae. All samples were frozen until they could be analyzed for levels of electrical conductivity (EC), total dissolved solids (TDS), total Kjeldhal nitrogen (TKN), ammonia (NH4-N), organic nitrogen (Org-N), nitrates (NO3-N), phosphates (PO4-P), and total N (Total N) according to standard methods in the lab.

Soil Analysis
Soil samples were taken to determine the impact of nutrient leaching and adsorption. Using a 2m auger, replicate samples were taken from the soils of each treatment following the 2002 cotton season on 11/19/02. Samples were taken from the soil column at 15cm, 30cm, 60cm, and 90cm and were analyzed by IAS Labs in Phoenix, AZ.

Pond Biosolid Analysis
Pond sludge was sampled from the bottom of the research pond after the final fish harvest and analyzed for nutritive value at IAS Labs in Phoenix. In preparation for this harvest, however, many biosolids were uncontrollably lost during draining. Therefore, a total sludge production mass could only be estimated. Samples were taken from several different depths and locations within the pond and compiled to analyze the total sludge nutrient content.

Data Analysis
Because several dependant variables were measured in order to accurately characterize plant growth, the analytical data were summarized in a fashion that provides the greatest clarity (See Figure 4 for a list of comparisons). One-way ANOVAs with Bonferroni Post Hoc tests were used in comparing treatments, and a level of significance was set at p<0.05. It was not difficult to predict that plots receiving chemical fertilizers would produce higher numerical height and yield values than chemically unfertilized plots. Therefore, there were three major statistically comparative foci with respect to the effects of fish effluent irrigations on field crops. First, comparisons were made between non-fertilized plots, or treatments 1 (W.W.) & 3 (F.E.). Next, comparisons were made between fertilized plots, or treatments 2 (W.W.+S.F.) & 4 (F.E.+S.F.). Finally, data from treatments 2 & 3 were compared in order to determine whether or not fish effluent irrigations alone (T3) can replace current chemical fertilization and irrigation practices (T2).

Research results and discussion:

Shrimp - Olive Results

The experiment was designed to approximate farm conditions. For this reason, trees were placed in rows receiving furrow irrigation. We planted as many rows as would fit across the experimental plot, with the extra row assigned as an effluent replicate to gain more information on response to effluent. Trees were planted in the bottom of a single furrow 30 cm wide and 30 cm deep, and watered by flood irrigation. Tree height was measured monthly, from a mark painted on the trunk, five cm above the original soil level, to the end of the longest branch. Trunk diameter was also measured initially, but was found to vary considerably depending on placement of the calipers. Due to this variability, this measurement was abandoned.

Irrigation

Trees were irrigated every week in the summer and every third week in the winter, approximating farm procedures. On the rest of the farm, trees were irrigated weekly in the summer, and as trees showed signs of water deficiency the rest of the year. Irrigation rates were 2.5 cm for each application from March through May and October through December, and five cm for each application from May through October. During shrimp production (approximately June to October), effluent from pond water discharge was used to irrigate the effluent treatment rows. The well water + fertilization treatment groups received urea fertilizer applications with the scheduled fertilizer applications for the rest of the farm (March through April). The rest of the year, all trees received well water.

Fertilizer was applied in four applications the first year and five during the second, with a target of a total of 0.23 kg of N per tree per year, or 188 kg/ha. This is half of what is recommended in the literature for large olive trees (Freeman et al., 1994), to account for the small starting size. In year one, 1.64 kg of urea (45% N) was applied per row in four applications, and a total of 10 cm of irrigation water (a rate of 112 kg/ha). In year two, five fertilizer treatments totaling 5.56 kg urea/row were applied in 12.5 cm of irrigation water (371 kg/ha), the full recommendation for olive trees. Urea was mixed with well water in 7,571-L water tanks before application in irrigation water.

Duplicate water samples were taken for each treatment during every irrigation event, to determine levels of nitrogen and salinity addition. A HACH DR-890 spectrophotometer (Hach Co., Loveland, CO) was used to analyze the samples (Table 1) for ammonia-nitrogen (NH3-N), nitrite-nitrogen (NO2-N), nitrate-nitrogen (NO3-N) and total nitrogen. To confirm the nitrate-nitrogen results, a standard curve was developed, and all nitrate-nitrogen samples were adjusted accordingly.

Statistical methods

We compared mean tree growth among treatments from beginning to end of the experiment and the mean water quality parameters, using a one-way analysis of variance (ANOVA). We performed all analyses with JMP IN 4 statistical software (SAS Institute Inc., Pacific Grove, CA)

Fish-Pond Trials

Construction of the IAA system began in 2001at the Maricopa Agricultural Center (M.A.C.) in Maricopa, AZ. An elevated, oval-shaped pond, holding 1.8x106 L of water was used as an irrigation reservoir. A drain manifold was constructed using long PVC pipe, and uniform perforations were made along the top and sides of the entire pipe to allow for nutrient extraction from a wider distribution of the pond area. An aerator and floating, stationary cages were added to the pond, and fastened into place. Fish were then added to the pond, both in the cages and free-swimming, depending on the species. Water was siphoned from an adjacent reservoir in order to replace evaporated or discharged water as needed. Water discharge occurred only during irrigation events, and was carefully directed onto randomized plots of an agricultural field through slotted irrigation pipe. Some of the randomized plots received pond effluent applications, which were pumped from the pond near the southwestern corner of the field, while the others received well water applications, which were tapped from an alfalfa well at the eastern end of the field. The same slotted irrigation pipe was used for each treatment, with the water source location being the only variable factor.
The agricultural portion of the study was sectioned into a randomized complete block (RCB) with four treatments and four repetitions in a 4x2 factorial design (four treatments x two types of field crops). The four treatments differed according to their source of irrigation and chemical fertilizer application, and are defined as (1) well water irrigations only (W.W.), (2) well water irrigations + standard chemical fertilizer applications (W.W. + S.F.), (3) fish effluent irrigations only (F.E.), and (4) fish effluent irrigations + standard chemical fertilizer applications (F.E. + S.F.). Sixteen total plots were used, each containing a surface area of 0.024 ha with dimensions ([6]1.02m rows x 39.6m). Of the six rows in each plot, only plants in the center four were used in the study. During barley season, however, a 5.5m combine harvested the center of the 6.1m wide plots. Therefore, each harvestable plot area was approximately 0.021ha for cotton trials and 0.022ha for barley trials.

Prior to the study, the field was tilled, lasered, disked, and groomed with a bed-shaper. The soil type in the field was classified as sandy loam, and herbicide, insecticide, and defoliant applications were used as needed. For plots requiring chemical fertilizer treatment, ammonium sulfate (21-0-0) was applied as needed during all growing seasons, and monoammonium phosphate (11-52-0) was applied prior to each barley season. Approximately 106kgN/ha and 1157kgP/ha during every barley season, and 179 kgN/ha during cotton season was applied using chemical fertilizers. After each field crop harvest, a conservation tillage system was implemented, in which soils were tilled horizontally within their own plots in order to reduce contamination.

Irrigation
McCrometers were fastened onto the pipes nearest to the functioning water sources on each side of the field to measure the total volume and rate of water flow onto the agricultural field. The 6-row plots were diked along the edges and corners of rows one and six to allow for equal distribution of nutrients within each plot, and to eliminate treatment contamination between plots. Barley and cotton were both watered up by allowing irrigation water to reach the end of the row and slowly rise to cover the tops of irrigation beds, thereby saturating the soil around the planted seeds completely. Irrigation scheduling was outlined using data from the Arizona Meteorological Network for the city of Maricopa, and AZSCHED, a software program designed to manage and schedule watering events. Irrigation most often occurred when the amount of water used was at 50% depletion of plant available water (PAW) in the rooting zone. Multiple water samples were also taken from each irrigation treatment to determine the concentration of nutrients in the water applied to the field. Initially, the applied volume of water (per irrigation and total) differed between treatments of well water and pond water because of flow-rate differences between the gas-powered pump and the alfalfa well, but during the second half of the study, BMPs were developed to increase flow rates for well water treatments. A total of 667.6cm of water were applied to the 0.386 ha research field over three cropping seasons. Because of slight differences in water pressure, 17.98% more water was applied to plots receiving well water. The amount of N and P applied through irrigation was adjusted proportionally by increasing nutrient concentrations of effluent by the same percentage.

At the start of each irrigation event the effluent was concentrated and black in color, but then became slightly less concentrated. Therefore, time trials were performed on 8/25/02 to determine the rate of change in nutrient concentration over time during effluent irrigations. It was concluded from these trials that the surge of concentrated water had a negligible effect upon the total amount of N and P added to the field during each irrigation event.

Fish Cropping
Different species of fish were stocked into and harvested from the pond throughout the year (see Table 1). Koi (Cyprinus carpio), tilapia (Oreochromis niloticus), and channel catfish (Ictalurus punctatus) were selected for this study because of their abundance in M.A.C ponds and because of their pertinence to U.S. arid lands aquaculture. Tilapia and catfish are popular in inland aquaculture because of their favorable food market values, while koi are raised primarily for ornamental purposes.

Koi fingerlings were purchased from an independent distributor (Pisces Aquaculture Inc.) and were stocked into the research pond, while juvenile tilapia and catfish were seasonally added to floating cages in the north end of the pond. The 765m3 floating cages were made of mesh wire, metal, and styrofoam. They were tied together and fastened to the edges of the pond using rope and iron stakes. Additional juvenile and adult koi were added at different times during the season to increase the total pond biomass.

All fish were fed 2-3% of their biomass once per day as recommended by Tucker and Robinson (1990), five days per week, from 12/27/01 through 4/17/03 with a floating aquaculture feed (see Table 2 for nutritional information). Caged fish were fed through long pipes that allowed the feed to drop through and remain in the cage. A total of 449.6kg of feed were given over a 441d period. Fish were also allowed to feed upon pond algae, which were prolific during the summer. The pond’s pH level was routinely monitored and was found to be between 7.66 and 8.01 on a consistent basis. The concentration of total dissolved salts also slightly increased over the course of the study.

Field Cropping
Short-season barley (Poco variety) was planted on 12/21/01 using a JD 8200 planter with a seeding rate of 122kg/ha. Seeds were planted 4cm deep on rows and in furrows with 18cm spacing. During the first barley season, chemical fertilizers for selected treatments were deposited onto the field using a standard applicator prior to irrigation. In Arizona, short-season barley requires one to three irrigations after emergence, whereas late season cotton can require between eight and ten. Although it is not common to find barley planted on rows and in furrows, it was necessary to select a crop that would fit into a rotational-crop system with late-season upland cotton, an important agricultural product of central Arizona. Yield and repetitive growth measurements were collected and calculated from barley plants at random (seeTable 3 for all field crop measurements and techniques). Poco barley was harvested on 5/13/02 using a 5.5m wide International 1440 axial flow combine and weigh wagon.

DP-458 BR late-season cotton was planted using a MF-4263 Monosoem planter on 5/15/02 at a seeding rate of 83kg/ha. During the cotton season, both mechanical side-dressing techniques and manual techniques were used in the application of chemical fertilizers. Cotton was harvested using a four-row spindle picker, sampled for fiber analysis, and weighed using an SK-CrustBuster Boll Buggy, and harvested stalks were extracted using a root puller. DP-458 BR cotton was selected for this study based on its importance to the Arizona agricultural market, and because of the ease with which it can permit a field to be double cropped, allowing for year-round production. Yield and repetitive growth measurements were collected and calculated from cotton plants at random (see Table 3).

A second term of short-season barley (Quick variety) was planted on 12/19/02 using methods identical to those used for Poco Barley. Quick barley was deemed most similar to the Poco variety, which was taken off the market in 2002. Because of heavy rains in the winter of 2003, Quick barley received one less irrigation than Poco barley during the 2001-2002 season. Quick barley was harvested on 4/29/03. Heavy and unexpected rains fell during the winter of 2003 and did not allow for the necessary application of 2,4-D herbicide. Therefore, weed production increased and the number of total barley irrigations decreased form four to three. The abundance of weeds caused some proportional clogging of the grain hopper weigh station, but final yields were able to be adjusted proportionally.

Fiber Quality Analysis
Ginned samples (400g) of harvested cotton were sent to USDA labs in Phoenix, AZ for fiber quality analysis and were tested for color grade (a measure of cotton whiteness), micronaire (a measure of fiber thickness or fineness), staple (a measure of how long cotton fibers reach in a season), strength (how difficult it is to break a single fiber), and uniformity (consistency and similarity between cotton fibers). These measurements are used by ginning and processing plants to determine what prices and discounts cotton farmers will receive. For instance, cotton fibers with low micronaire ratings will receive more money per bale of cotton than fibers with high micronaire ratings; favorable staple and ratings combined with a low color grade will produce average returns for farmers. These analyses were important in determining if fish-effluent irrigations would affect cotton fiber quality.

Water Analysis
Three to four water samples were taken at each irrigation event of both fish effluent and well water. Concentrated sulfuric acid (H2SO4) was added to every other sample in order to eliminate any contaminating effects of algae. All samples were frozen until they could be analyzed for levels of electrical conductivity (EC), total dissolved solids (TDS), total Kjeldhal nitrogen (TKN), ammonia (NH4-N), organic nitrogen (Org-N), nitrates (NO3-N), phosphates (PO4-P), and total N (Total N) according to standard methods in the lab.

Soil Analysis
Soil samples were taken to determine the impact of nutrient leaching and adsorption. Using a 2m auger, replicate samples were taken from the soils of each treatment following the 2002 cotton season on 11/19/02. Samples were taken from the soil column at 15cm, 30cm, 60cm, and 90cm and were analyzed by IAS Labs in Phoenix, AZ.

Pond Biosolid Analysis
Pond sludge was sampled from the bottom of the research pond after the final fish harvest and analyzed for nutritive value at IAS Labs in Phoenix. In preparation for this harvest, however, many biosolids were uncontrollably lost during draining. Therefore, a total sludge production mass could only be estimated. Samples were taken from several different depths and locations within the pond and compiled to analyze the total sludge nutrient content.

Data Analysis
Because several dependant variables were measured in order to accurately characterize plant growth, the analytical data were summarized in a fashion that provides the greatest clarity (See Figure 4 for a list of comparisons). One-way ANOVAs with Bonferroni Post Hoc tests were used in comparing treatments, and a level of significance was set at p<0.05. It was not difficult to predict that plots receiving chemical fertilizers would produce higher numerical height and yield values than chemically unfertilized plots. Therefore, there were three major statistically comparative foci with respect to the effects of fish effluent irrigations on field crops. First, comparisons were made between non-fertilized plots, or treatments 1 (W.W.) & 3 (F.E.). Next, comparisons were made between fertilized plots, or treatments 2 (W.W.+S.F.) & 4 (F.E.+S.F.). Finally, data from treatments 2 & 3 were compared in order to determine whether or not fish effluent irrigations alone (T3) can replace current chemical fertilization and irrigation practices (T2). Shrimp – Olive Results
Water Treatment Characterization
Mean nutrient content in irrigation waters varied considerably across treatments (Table 2). NO2-N values for the effluent treatment were 0.3 mg/L and 0.5 mg/L higher than the fertilizer and well water treatments respectively (F2, 73 = 28.5, p<0.0001, one way ANOVA). The fertilizer applications averaged 0.5 mg/L NH4-N higher than the effluent and 1.0 mg/L NH4-N higher than well water values (F2, 66 = 11.77, p<0.0001). NO3-N levels were comparable across treatments (F2,73 = 0.127, p = 0.88). The TN was much higher in the fertilizer treatment, 67.2 mg/L compared to 6.87 mg/L for the effluent treatment, and 8.53 for the well water (F2,66 = 60.5, p<0.0001). TN levels reveal that urea additions had not converted to other forms of nitrogen during application. While TN was lower than the sum of the nitrogen components for the effluent and well water treatments, given the standard error for each measurement, this difference is insignificant. Salinity varied little among the treatments throughout the study, ranging from 1.63 ppt to 1.86 ppt (F2, 67 = 1.46, p = 0.24).
Total water applied was 159 cm (2,460 m3 for the whole experimental plot). Total evapotranspiration (ET) over the two-year experiment was 405 cm, giving a crop coefficient (Kc) of 0.39. There was a total of 38.4 cm of rainfall. Effluent irrigation contributed 113 cm (71% of the total irrigation). In the first year, the fertilizer treatment received a total of 1.64 kg of urea per row (112 kg/ha) in 10 cm of well water over four applications. In the second year, 5.76 kg of urea was added per row (392 kg/ha) in 12.5 cm of well water over five applications. Water with fertilizer contributed 14.5% of the total irrigation for this treatment.
Nitrogen additions were extrapolated from water quality parameters and total water applied. Total nutrient additions - a summation of nutrients in the treatments and in the well water the rest of the year - were highest in the fertilizer treatment, with the well water treatment providing slightly more TN than the nitrate-rich well water treatment. NO3-N and TN additions were comparable across treatments, while NO2-N and NH4-N were highest in the effluent treatment (Table 3).

Tree Growth
Irrigation with shrimp effluent did not harm the olive trees. Growth of trees receiving effluent was not different than the fertilizer or well water treatments (Figure 2). The well water treatment grew substantially less than the fertilizer treatments when comparing changes in growth from the beginning to the end of the experiment (F2, 62 = 3.19, p = 0.048, one-way ANOVA). Overall, trees receiving effluent averaged 61.0 cm of growth over the experiment, compared to 70.4 cm for the fertilizer and 48.4 cm for the well water treatment.
Discussion
Water Treatment Characterization
Nitrogen parameters in the water were similar to levels from previous studies (Table 4). However, high nitrate-nitrogen levels in the groundwater distinguish well water at this site from other inland farms. Effluent supplied 159 kg/ha NO3-N, 6.3 kg/ha NO2-N, 5.04 kg/ha NH4-N and 116 kg/ha TN over the two-year experiment (Table 3). While total nitrogen additions were highest in the fertilizer treatment, the effluent treatment contained the highest nitrate, nitrite and ammonia additions (Table 3). TN additions were not different in the well water and effluent treatments, despite nitrogen additions in shrimp feed, possibly due to phytoplankton nitrogen assimilation. Since salinity levels in effluent were not significantly different from well water, salinity in effluent is not considered to be harmful.
The timing of these additions likely impacts tree growth more than the total nitrogen contributions. Fertilizer was applied in the spring, as temperatures warmed, stimulating an increase in the rate of tree growth (Figure 2). Effluent became available for irrigation in July, when tree growth rate was not at its highest levels, suggesting trees were assimilating fewer nutrients. However, the slow addition of plant available NO3-N and NH4-N over the course of the growing season may allow for constant nutrient uptake and more efficient assimilation.

Tree Growth
Effluent irrigation did not significantly increase olive tree growth over the two-year study. While we did not reject the null hypothesis, we also found no significant difference between trees receiving fertilizer and those receiving effluent. Despite the fact that the effluent treatment TN additions were not different than the well water treatment, tree growth was not different than the fertilizer treatment with its higher TN levels. Effluent from low salinity inland shrimp culture can be reused as a source of irrigation water, increasing water use efficiency in arid lands as water is pumped once and used twice, without detriment to either crop.
In retrospect, a number of changes could have strengthened this experiment. Testing a broader range of water and soil nutrient parameters to include micronutrients, organic matter, and ash could provide further understanding of contributions of effluent to field crops. This may help quantify contributions of algal biomass to soil nutrients and composition. Sampling soil and leaf nutrients from each experimental unit would have quantified nutrient uptake and deposition. Randomly selecting trees from each row for removal to determine surface leaf area and total biomass every six months would have also provided relative growth rate, a better measure of tree growth. Decreasing the number of trees in each row, while increasing the number of rows would increase the number of experimental units, improving statistical power without increasing the number of trees. More experimental units would have also allowed the addition of another treatment of a blend of effluent and fertilizer, to quantify fertilizer savings and potential fertilizer-effluent interactions. Small test plots of other field crops could have provided more complete information on water use and production with effluent irrigation.

Water Use Efficiency
Integrating shrimp aquaculture into farm production cycles reduces water costs and increases water use efficiency. On this farm, water is constantly pumped from May to October to provide water for aquaculture production. Early in the season much of this is lost to seepage, before algae and fine solids seal the ponds. Evaporation losses are also great, with evapotranspiration averaging 90 cm during the course of the growing season. Water exchange in the shrimp ponds is estimated at 1% per day. Previously, it was determined that this would contribute 2,725 m3 of water for irrigation daily (McIntosh and Fitzsimmons, 2003). Over the course of the 100-day shrimp growing season, this provides approximately 2,700 ha cm of irrigation water. As the ponds are drained at harvest, another 2,700 ha cm of irrigation water is made available. This is enough water to irrigate nearly 48 ha of mature olive trees (Kc(crop coefficient)=0.75) annually, without having to pump any additional water on the farm.
Reducing water consumption will lead to economic savings in electrical or water costs. Average pumping costs for this irrigation district was $2.10 per ha cm ($26 per acre foot). By using discharge water twice, the farmer can realize over $5,000 of pumping savings if the farm is designed to gravity drain from ponds to agriculture fields. If the farm has the capacity to hold the harvest water for use in irrigation, the farmer can attain another $5,000 in savings. Economic gains can be even greater if the aquaculture portion of the farm is financially successful, effectively subsidizing agriculture water costs.
In the last year of this study, approximately 55.5 metric tons of shrimp were produced, with a gross farm gate value of over $246,000. This valuable secondary crop more than pays for the pumping cost of the water lost to seepage and evaporation. Producing shrimp in water that is already being pumped for agriculture irrigation increases water use efficiency, with greater production per unit of water than agriculture alone. Any increase in field crop growth due to effluent irrigation also increases water use efficiency without any financial input by the farmer. So while nutrient addition from inland shrimp effluent is minimal, financial savings in water costs and increased water use efficiency make inland shrimp production a valuable option for integration with existing irrigated agriculture.
Benefits of integrated shrimp culture and irrigated field crops are based on pond water exchange and financially viable shrimp production. As shrimp producers in Arizona have looked for ways to maximize production with reduced inputs, they have decreased water exchange. Some are not exchanging any water, simply replacing seepage and evaporation losses. In these cases, no water is available for reuse during the summer months, with a large amount available at harvest when plant growth and ET is reduced. Given the decrease in shrimp prices over the last five years, production costs are often higher than wholesale prices, leading to a net financial loss in aquaculture production on most of the farms. Water management practices, shrimp prices and production costs will therefore dictate the viability of effluent reuse for field crop irrigation.

Fish-Cotton-Barley Cropping
The total koi biomass continually increased from December of 2001 until April of 2003. The total input biomass of all koi was 77.83kg. The total harvestable biomass was 286.39kg, a 270% increase from the initial input biomass, despite a moderate fingerling mortality during acclimation (32%).
After 145 days (Summer-Fall 2002), 107.63kg of tilapia, were harvested from cages, resulting in a 95% increase in total biomass. The average mass per live individual had more than doubled from 197g (+/-92.7) to 432g (+/-160.7). There was an 89% recovery rate with most mortalities taking place during acclimation. The highest rate of tilapia mass increase occurred from early August to late September of 2002.
After 121 days (Winter-Spring 2003), 35.38kg of catfish were harvested from cages resulting in a 37% increase of total biomass. The average individual fish had grown from 172.8g (+/-72.0) to 290g (+/-112.1), a 68% increase. There was an 82% recovery rate, with several mortalities taking place during acclimation. During the season, losses resulted from both disease and theft. Also, the parasite, Ichthyophthirius multifiliis, appeared to cause considerable damage. Carcasses were often found floating in cages before harvest, and many fish had visible signs of lesions and outer skin deterioration.

Field Cropping
Subgroup A: Comparing the Effects of Well Water Irrigations (T1) and Fish Effluent
Irrigations (T3) on Non-Fertilized Field Crops
Poco Barley; 2001-2002
Vegetative Growth: Stands were found to be uniform, and no significant differences in vegetative growth were observed in subgroup A.
Reproductive Growth: No significant differences in reproductive growth were observed in subgroup A.
DP-458 BR Cotton; 2002
Vegetative Growth: Stands were found to be uniform, and no significant differences in vegetative growth were observed in subgroup A.
Reproductive Growth: No significant differences in reproductive growth were observed in subgroup A.
Quick Barley; 2002-2003
Vegetative Growth: Stands were found to be uniform, and plant heights were recorded on three different occasions. These were labeled phA-phC: A (2/3), B (3/19), and C (4/16). Using an analysis of variance (ANOVA) it was determined that within data subgroup A, plants receiving fish effluent (T3) exhibited significantly more vegetative growth in the form of plant height than untreated plants (T1), (phB = 0.005; phC = 0.003).
Reproductive Growth: No significant differences in reproductive growth were observed in subgroup A.

Subgroup B: Comparing the Effects of Well Water Irrigations (T2) and Fish Effluent
Irrigations (T4) on Fertilized Field Crops
Poco Barley; 2001-2002
Vegetative Growth: No significant differences in vegetative growth were observed in subgroup B.
Reproductive Growth: No significant differences in reproductive growth were observed in subgroup B.
DP-458 BR Cotton; 2002
Vegetative Growth: Cotton plant heights were recorded at five different times. These were labeled phA-phE: A (6/12), B (7/8), C (7/16), D (8/19), and E (9/21). Within subgroup B, effluent irrigated crops (T4) produced significantly taller plants (at midseason, phC, p=0.003; and just before harvest, phE, p=0.05) than plants receiving irrigations from the well (T2) (see Figure 1).
Reproductive Growth: No significant differences in reproductive growth were observed in subgroup B.
Quick Barley; 2002-2003
Vegetative Growth: No significant differences in vegetative growth were observed in subgroup B.
Reproductive Growth: No significant differences in reproduction were observed in subgroup B.

Subgroup C: Comparing the Effects of Well Water Irrigations on Fertilized Field Crops (T2) vs. the effects of Fish Effluent Irrigations on Non- Fertilized Field Crops (T3)
Poco Barley; 2001-2002
Vegetative Growth: Barley plant heights were recorded at three different times. These were labeled phA-phC: A (2/22), B (3/22), C (5/13). Within subgroup C, T2 produced significantly taller barley than T3 (for phB, p<0.001; for phC, p<0.001)
Reproductive Growth: No significant differences in reproduction were observed in subgroup C.
DP-458 BR Cotton; 2002
Vegetative Growth: T2 produced taller cotton plants (for phD, p=0.04; for phE, p<0.001). and produced more nodes per plant than T3 (p<0.001). On average, plants receiving T2 produced significantly more nodes per plant than plants receiving T3 (p<0.001; mean difference = 4.4 (+/-0.83) nodes per plant). Significant differences in cotton plant height:node ratio were only found within subgroup C, where T2 always produced higher ratios than T3 (p=0.05). Petiole samples were taken on 8/14/02 to determine plant tissue concentrations of N and P. Across all treatments (T1, T2, T3, and T4), plants receiving fish effluent irrigations (T3 and T4) stored an average of 639ppm (+/-748) more nitrate and 635ppm (+/-673) more phosphate in petiole tissue than plants receiving well water irrigations (see Figure 2). Within comparative subgroup C, however, T3 (F.E.) plants stored significantly more phosphates than T2 plants (W.W.+S.F.) (p=0.027). If more petiole sample repetitions had occurred, the same may have been true for nitrate.
Reproductive Growth: No significant differences in reproductive growth were observed in subgroup C.
Quick Barley; 2002-2003
Vegetative Growth: T2 produced significantly more vegetative growth in barley than T3 at all times during the growing season (phA < 0.001; phB < 0.001, phC < 0.001).
Reproductive Growth: No significant differences in reproductive growth were observed in subgroup C.

Water Analysis
From 12/01 through 04/03, the total amounts of nitrogen added to the field through both fish effluent and well water irrigations were 30.39kgN (+/-2.14) and 42.48kgN (+/-2.25) respectively (see Figure 3 for progressive data). Adjusting for unequal volumes of irrigation water between treatments, total applied nitrogen for effluent treatments would have increased to approximately 35.66kgN. Also during this time, the total amounts of phosphorous added to the field through both fish effluent and well water irrigations were 0.54kgP (+/-0.02) and 0.0kgP respectively (see Figure 4 for progressive data). Adjusting for unequal volumes of irrigation water between treatments, applied phosphate would have increased to 0.65kgby the end of the study.

Soil and Pond Biosolid Analysis
Soils from the different treatment plots were collected and tested in December 2002, but no significant differences were found.
On 4/17/03, the research pond was drained for final fish harvest, in which a portion of biosolids were uncontrollably lost. The nutritive value of pond biosolids has been summarized in Table 5. As expected, the sludge depth was shallow near the center pipeline, but rapidly became deeper and thicker at further distances. Sludge depths were found to be between 7cm and 61cm in the pond, and the possible range of total biosolid retention over three cropping seasons was estimated to be between 50m3 and 400m3.

DISCUSSION

Fish cropping
The main goal, with regard to raising fish, was to produce a healthy and profitable crop that would generate large quantities of waste that could, in turn, be used as an organic fertilizer. This was accomplished, and even though water exchange occurred only at crop irrigation, at no time did fish show any of the common signs signaling ammonia toxicity (Wright 1995).
Each component of fish cropping was successful in its own way. Over a greater length of time, tilapia might have generated the highest net profit because of both their favorable market value and because they are omnivorous and will eat almost anything naturally occurring in the pond. Although the market is currently favorable for live, whole tilapia, one drawback is that they are a tropical fish and must be harvested or kept in a heated area to avoid the effects of the cool southwestern winters.
Koi were the most resilient after initial acclimation, because they are adaptive animals, able to survive both temperature extremes and prolonged periods of hypoxia. Koi are a secure crop because they can be kept cheaply with little need for strict management until a demand in the market occurs. By design, koi biomass increased the most becausae their inherent environmental stability allowed them to remain in the water during the entire study.
Like many desert aquaculturists, we faced several problems in raising channel catfish in our IAA system. Not only did several catfish die from handling and acclimation stress, but they were also most susceptible to the parasite, Ichthyophthirius multifiliis, or ‘ick.’ Every fish became infected, and several showed visible signs of sickness or death (although fish with ick may still be treated or harvested). Because they are such a desirable crop, catfish are also susceptible to thieves (as we found out!), but vandalism and larceny are common problems in unsecured ponds. Although market prices for catfish generally produce good returns, it can be a risky investment in the desert. They are not well acclimated to extremely hot desert temperatures and may not survive handling in the summers. If an aerator breaks down on an extremely warm day, an entire crop of these delicate fish may be lost.

Determining the nature of treatments through water and sludge analysis
Before discussing the effects of the different treatments on field crop growth and yield, it is necessary to first understand the nature of the integrated system itself, including its internal processes and resulting products that comprise each treatment (see Table 6). Concentrations of total Kjeldahl nitrogen (TKN) were higher, on average, in effluent while well water treatments contained a much higher overall concentration of NO3. At first glance, this seems opposite of what one might expect, but perhaps not if we consider the nitrogen cycle. Ammonia is the predominant nitrogenous waste product excreted by fish (Wright 1995) and is not likely to get “locked up” in benthic sludges like many solid wastes because it is very soluble in water. It is also unlikely to be affected by algae or pond bacteria because the ammonium ion (NH4+) must first be slowly converted to NO2- by the aqueous bacteria nitrosomonas, and then to NO3- by nitrobacter. It can be assumed that denitrification occurred along the bottom layers of the pond, including the sludge/water gradient, where, over time, anaerobic bacteria stripped the nitrogen atoms from the nitrates. Aqueous nitrate ions above the anaerobic layer were quickly assimilated by cyanobacteria and green algae floating in the pond, which eventually died and settled to the bottom, forming a thick sludge. These biosolids sequestered valuable nitrogen and phosphorous-based compounds, creating a nutrient sink and yielding an overall high TKN concentration and an overall extremely low concentration of NO3.
To confirm this, we measured the nutrient outflow from the pond at initial discharge and again after 60 minutes (see Figure 5). NO3 was absent from the initial discharge (0.0ppm) and very little was present 60 minutes later (1.87ppm). The pond effluent averaged overall nitrate levels of 1.61ppm (+/- 1.44), but this water was initially introduced into the pond from the same source of water used in the well water treatments, which had a seasonal average of 4.39ppm (+/-1.00) NO3. The concentrations of NH4 and organic nitrogen in the initial discharge were much greater than in the sample taken an hour later. The concentration of total N in the few seconds of the initial discharge was almost ten times greater than the concentration of total N in the sample taken at 60 minutes.
If this material was to be distributed onto the agricultural field with the irrigation water, we would likely see increases in plant growth and yield. It would be as similar as applying manure to promote growth. However, the sludge in the pond was so thick and adhesive that it could not easily be brushed manually toward the perforated discharge pipe running along the center of the pond. Some biosolids that had settled near the pipe by chance were immediately pumped onto the field during the start of irrigation, but after a short time, the thick, black sludge yielded to a cleaner, green colored pond water, signaling that the majority of the nutrients remained in the pond. Poor spatial distribution of sludges along the bottom of the pond was a key factor in producing a low total N content of the pond effluent treatment.
Biosolids were allowed to accumulate over time to show the significance of its collection and application in a sustainable IAA system. Because the groundwater at the M.A.C. is high in nitrates, and because much of the NO3 in the open pond reservoir was assimilated by bacteria or algae, the average total amount of nitrogen applied through irrigation alone was actually higher for well water treatments (+0.94ppm). Figure 3 shows the total amount of nitrogen applied through irrigation by well water and fish effluent from December of 2001 through April of 2003. As time progressed, the difference between the amount of nitrogen applied (by treatment source) became greater and greater. It is safe to say that, although we don’t know to what extent, there was an inverse relationship between the amount of nitrogen applied through effluent irrigations and the amount of nitrogen sequestered by microorganisms in the pond.
In nature, plants may readily take up nitrogen in the form of NO3 or NH4, and phosphorous in its inorganic form. Organic nitrogen or phosphorous must be broken down by soil bacteria before plant uptake and use. NO3 and NH4, however, leach through the soil column much more quickly than the nitrogen that is sequestered by algae. Because of its larger size, decomposing algae will travel more slowly through the soil as it is broken down into nutrients that plants can use. In our study, the total amount of nitrogen applied through irrigation for well water treatments was much higher than that of fish effluent treatments, but a large percentage of this was in the form of NO3 (see Table 6). Conversely, fish effluent contained greater percentages of ammonia and organic nitrogen, which may have leached more slowly. Also, there was no phosphorous in any of the well water, while an average of 0.18ppm (+/-0.09) PO4 was added to the field with every effluent irrigation.
Soil samples taken at the end of the 2002 cotton season revealed that applied chemical fertilizers and nutrients that were applied through irrigation earlier in the year had either been taken up by plants or had leached below 90cm in depth by the end of the cotton season. As previously mentioned, the adhesive properties of the biosolids transformed the bottom of the pond into a nutrient sink. Consistent fish effluent irrigations during the growing season did not significantly increase the amount of N or P in soils, confirming that land application of pond biosolids would be necessary to compete with any current practices involving chemical fertilizer applications.

Treatment effects upon field crops
When faced with nutritional challenges, plants will often sacrifice vegetative growth before limiting reproductive growth, so it is not surprising that we did not find any significant differences with respect to reproduction (i.e. barley and cotton yield, the number of fruit per cotton plant, or the first fruiting branch in cotton). However, within each of our subgroup comparisons, we found differences in vegetative growth which arose from the plants’ exposure to the different treatments. Plants receiving T3 and T4 received small amounts of organic nitrogen and ammonia in every irrigation, while those receiving T1 and T2 received slightly more total nitrogen distributed in the form of small amounts of nitrates through irrigation. Plants receiving T2 and T4 were exposed to concentrated chemical fertilization events, while no chemical fertilizers were applied to the plants receiving T1 and T3.

Subgroup A: Comparing the Effects of Well Water Irrigations (T1) and Fish Effluent
Irrigations (T3) on Non-Fertilized Field Crops
The total biomass within the research pond was relatively small during the initial effluent irrigations, so few significant differences in growth were expected during the first growing season (Poco barley; 2001-2002) for all comparative subgroups. During the winter of 2002-2003, however, when cumulative effects of effluent irrigations had built up most in soils, vegetative growth in Quick barely was measured. In both mid-season and late-season, plants receiving only fish effluent irrigations were significantly taller than those receiving only well water irrigations. Over time, as the fish biomass increased along with the aqueous nutrient load in the pond, phosphates applied through effluent irrigations may have generated this difference. Subgroup A had no external fertilizer applications, and it is also possible that, as organic matter broke down, there may have been sufficient nitrogen taken up by cotton and barley plants (see Figure 2 for petiole analysis) to produce a change in vegetative growth.

Subgroup B: Comparing the Effects of Well Water Irrigations (T2) and Fish Effluent
Irrigations (T4) on Fertilized Field Crops
When soils were fertilized, no differences were seen in either Poco barley (2001-2002) or Quick Barley (2002-2003). However, differences in vegetative growth were visible in cotton plants (2002). Applications of phosphorous occurred just prior to each barley-growing season, but not before the cotton-growing season. Like comparative subgroup A, plants in effluent irrigated plots grew significantly taller than plants receiving only well water for subgroup B. This may, again, be due to the fact that all phosphorous applied in the form of chemical fertilizers may have leached through the soils well before cotton plants could consume these nutrients. Small concentrations of phosphorous (0.18ppm (+/-0.09) PO4) were added to the soils through the application of water in each of the nine irrigation events during the 2002 cotton season. This would also explain why no effects were seen in barley vegetative growth, as chemical fertilizers were added just before watering up.

Subgroup C: Comparing the Effects of Well Water Irrigations on Fertilized Field Crops (T2) vs. the effects of Fish Effluent Irrigations on Non- Fertilized Field Crops (T3)
As expected, almost all of the data from this comparison confirms that T2 was more effective than T3. This is because T3 contained lower amounts of total N applied (both through irrigation and chemical fertilizer application) and lower amounts of total P applied (because of the supplementary chemical fertilizer applications in T2). In Poco barley (2001-2002), DP-458 BR cotton (2002), and Quick barley (2002-2003), plants receiving well water and chemical fertilizer grew significantly taller than plants receiving effluent alone. For all three growing seasons, a significant difference was seen by at least midseason, but in the final growing season (Quick barley), significant early-season differences were also observed. This confirms that, without biosolid application, this particular IAA system would not be sustainable over time. In our study, so many nutrients were sequestered by algae and remained in the immovable pond sludges that the crops suffered.
It is important to note the petiole data from the DP-458 BR cotton season (prior to which, no chemical phosphate was added). Tissues from plants receiving T3 contained significantly higher concentrations of P than the tissues of plants receiving T2. However, height was probably not affected because chemical nitrate (in the form of ammonium sulfate) was added to T2 during the season, thereby maintaining healthy plant growth.
It is also important to note that, even without biosolid application, no differences in yield were seen between any of the subgroups. Although this result would not likely prove to be sustainable, our study has shown that an IAA design in which chemical fertilizers are reduced and effluent is added would maintain profitable yields for several seasons until nutrients could be replenished with the collection and application of pond biosolids.

Research conclusions:

The long-term sustainabilty of integrative aquaculture and agriculture in arid regions is still debatable, but we have helped to provide clarity to some of the issues. The importance of effluent and biosolid application to agricultural fields cannot be understated. Further studies should be designed to determine the effectiveness of different effluent and biosolid management processes. Although the overall best solution may vary depending on individual farm conditions, it is certain that aquaculture waste water application is the key to effective, profitable, and sustainable integrative aquaculture and agriculture.
We have shown that healthy and profitable amounts of fish and shrimp may be raised in an aerated pond reservoir that is discharged only at each irrigation event. It is recommended, however, that sludges be collected and applied to soils whenever possible to increase the overall health of the pond and the field. We also determined the nutrient content of well water and effluent applied to the field in an isolated study. Through our examinations, it was determined that algae and pond bacteria play an enormous role in the form and quantity of N and P present in the discharged effluent. These small organisms ultimately determine the effectiveness of effluent treatments on long-term field crop growth.
With respect to vegetative growth, field crops were negatively affected by the withholding of the nutrient-rich pond biosolids. Effluent irrigations may have increased vegetative growth because of the addition of small amounts of phosphorous at each irrigation event, but were still lacking in N.
The best solution to arid lands integrative aquaculture and agriculture may, in fact, be found in a combination of effluent irrigation, biosolid application, or even reduced chemical fertilizer applications. If possible, individual growers should conduct their own studies to determine the best management practices for their farm that will maximize profits without adding health risks or nutritional challenges to fish or field crops. Natural resource conservation is necessary for any sustainable integrative aquaculture and agriculture system that has a goal to to increase profits, reduce water usage, and decrease the load of applied chemical fertilizers that often cause health and contamination problems.

Participation Summary

Educational & Outreach Activities

Participation Summary:

Education/outreach description:

McIntosh, D. and Fitzsimmons, K. 2003. Characterization of effluent from an inland, low-salinity shrimp farm: What contribution could this water make if used for irrigation. Aquacultural Engineering 27:147-156.

McIntosh, D., K. Fitzsimmons, J. Aguilar and C. Collins. 2002. Towards Integrating Olive Production with Inland Shrimp Farming. World Aquaculture 34(1):16-20.

Abstracts and Presentations

Fitzsimmons, K, King, C. and McIntosh, D. 2004. Overview of inland-saline aquaculture. Australasian Aquaculture 2004. Sydney, Australia. September, 2004. Abstracts p.130.

Fitzsimmons, K., Piedrahita, R. Chen, S, and Hardy, R. 2004. Aquaculture effluent research in the Western Regional Aquaculture Center. World Aquaculture Society – Honolulu, Hawaii. March 2004. Abstracts p. 196.

McIntosh, D., King, C. Ryder, E. and Fitzsimmons, K. 2003. Integrating shrimp and olive production in Arizona – Is it working? World Aquaculture Society – Bahia, Brazil. May 2003. Abstracts p. 478.

King, C. and Fitzsimmons, K. 2003. Land application of aquaculture shrimp sludge as a soil amendment for olive trees (Olea europaea). USAS Abstracts. Louisville, KY

M.N. Asghar, A.S. Qureshi and K. Fitzsimmons. - Socio-ecology of Saline Groundwater: Integration of Aquaculture within Drainage Systems as Marketable Produce in Developing Countries . 9th International ICID Drainage Workshop. September 2003 in Utrecht, the Netherlands

McIntosh, D., Fitzsimmons, K., Aguilar, J., and Collins, C. 2002. Integrating olive production with inland shrimp farming. p. 206. USAS Abstracts. San Diego. CA.

An Arid Lands Aquaculture newsletter was distributed in April of 2003.

Project Outcomes

Project outcomes:

We have not conducted any formal economic analyses as part of this study. However, associated work has demonstrated that the farmer can achieve considerable savings in water costs by splitting water pumping expenses between a fish or shrimp operation and the irrigated field crop.

Farmer Adoption

The farmer cooperator at Wood Brothers farm has changed farm practices to utilize all of the effluent to irrigate olive trees and is testing it on wheat. Another shrimp farmer, Arizona Mariculture, about 40 km west of Wood Brothers has also put in olive trees to mimic our experimental procedures.

Phelps-Dodge, a major copper mining corporation in Arizona has asked us to design a demonstration project of integrated aquaculture-agriculture at an inactive mine location to utilize currently unused assets of water, land, and staff.

The Maricopa Agriculture Center has continued the fish-irrigated crop system and now highlights the methodology to many visitors at the demo farm.

A large Native American farm operation in Central Arizona had instituted an integrated fish-irrigation project for two years. However, with major revenues now being generated from tribal casinos, interest in the aquaculture operation has waned and the fish farm has been closed. One of the Native American High Schools has continued with a demo project on school grounds.

Recommendations:

Areas needing additional study

We had hoped to demonstrate conclusively that fish and shrimp effluents would prove beneficial for irrigated crops and that they would exhibit significantly higher growth rates. We did not show this consistently, but we did show that plants grew better than controls grown with well water alone. We also demonstrated that the effluent did not have a detrimental impact on irrigated plants.

Additional study is needed to determine if prolonged use of effluents, especially elevated TDS effluents from shrimp operations, might have a detrimental effect, such as soil salinization. It would also be worthwhile to determine if more intensive aquaculture might increase effluent nutrients, especially nitrogen and phosphorus, to levels that did elicit significantly greater plant growth.

Any opinions, findings, conclusions, or recommendations expressed in this publication are those of the author(s) and do not necessarily reflect the view of the U.S. Department of Agriculture or SARE.