Principal Investigator:

Kevin F. Bronson
Texas Agricultural Experiment Station
RR 3, Box 219
Lubbock, TX 79403
Email: k-bronson@tamu.edu

Cooperators:

College Station - S. Searcy
Lubbock - A. Brashears, R.J. Lascano, C. Trostle
T. Chua, R. Boman, P. Dotray, and E. Segarra

Agencies:

College Station and Lubbock, Texas
Texas Agricultural Extension Service
Texas A&M University
USDA-ARS, and Texas Tech University

Primary Research Location: Ropesville and Lamesa, Texas

Project Title: On-farm testing of precision agriculture technologies for Texas High Plains Cotton

Reporting Period: 1 September, 1999 - 31 August, 2000

Objectives:

1. To evaluate at two sites precision agriculture (PA) practices in cotton of grid soil sampling/analysis, and variable-rate technologies (VRT) compared to conventional, blanket applications of fertilizer, nematicide, and post-emergence herbicide.

2. Conduct economic analysis of PA/VRT vs. conventional, blanket farming practices.

3. Demonstrate PA field practices through extension outreach in conjunction with TAEX and Agripartners

A. Summary of Progress:

Objective 1. To evaluate at two sites precision agriculture (PA) practices in cotton of grid soil sampling/analysis, and variable-rate technologies (VRT) compared to conventional, blanket applications of fertilizer, herbicide and nematicide.

Precision agriculture (PA) technologies such as variable-rate fertilization and light-activated Weed Seeker shielded sprayers are now commercially available. However, little systematic/scientific testing of these technologies have been done with cotton in the Texas High Plains. Producers are interested in PA, but are not sure what the agronomic or economic benefits of shifting from conventional, whole-field, blanket-rate farming to VRT would be.

In the 2000 growing season we successfully implemented variable-rate P fertilization and nematicide applications at both the Ropesville and Lamesa sites. Variable-rate Roundup® herbicide post-direct applications were only done at Ropesville due to low weed pressure at Lamesa. Phosphorus fertilization was based Mehlich-3-extractable P from 0-6 in. soil samples. Sixty soil samples were taken on a 0.5-ac grid, which were global positioning system (GPS) referenced, at each of the 27-ac sites. Plots were 16, 40-in. rows wide and ranged from 2.1 to 3.9 ac in size. There were three replicates, which were in three center pivot spans. Irrigation was LEPA at 75 % estimated ET replacement at Lamesa and 80% ET replacement at Ropesville. Cotton variety Paymaster Roundup® Ready 2326 was planted at 15 lb seed/ac at both sites in May, 2000. Blanket doses of N fertilizer of 90 lb N/ac were applied to all plots at both sites.

The blanket treatment rate of P fertilization was based on an average of 20 soil samples taken from the blanket-rate plot areas. Phosphorus was applied 3 in deep and 3 in. from the top of rows with a spoke-applicator 1 wk before planting. Variable-rate P fertilizations were made with a liquid fertilizer rig fitted with variable-rate controllers and software (Agchem/SOILTEQ, Minnetonka, MN), and was GPS referenced and ground speed radar controlled. The blanket-rate of P fertilizer applied was 45 lb P2O5/ac at Ropeville and due to slightly greater average soil test P levels, the blanket P rate was 30 lb P2O5/ac at Lamesa (Fig. 1 and 2). The average variable-rate of P was 41.9 and 38.4 lb P2O5/ac at Ropesville and Lamesa, respectively.

Variable-rate or on/off post-direct Roundup® applications were made with an infrared Weed Seeker shielded sprayer (Patchen Selective Spray System, Los Gatos, CA, Fig. 3). Blanket and variable-rate Temik® applications were based on 1-ac grid soil samples taken the previous autumn and subjected to nematode counts. Cotton lint was hand-harvested from 4, 6.5 ft lengths of row at each of the 60, GPS-referenced, 0.5 ac grid points in October, 2000.

Variable rate Temik® applications were made by manually adjusting nematicide hoppers on the planter. The producer's rate of Temik® was 3 lbs/acre, while in the variable rate treatment, Temik® 15G was applied at 5 lbs/acre in areas with moderate root-knot nematode population densities, and 3 lbs/acre in areas with 0 or low root-knot nematode densities. Root-knot nematode density was obtained from soil samples taken in the fall of 1999.

The two test sites were photographed aerially using color infrared film during August of 1999. Soil samples were taken in the fall of 1999 at both sites, based on different reflection intensities with the near infrared (NIR), red, and green bands from the images, and band combinations. At the Ropesville site, there was a strong reflection pattern in the eastern part of the circle, from the rest of the circle (Fig. 4), which corresponds to a sandier soil texture and higher root-knot nematode densities.

Root-knot nematode densities > 1,000/500 cm3 soil was the threshold used to apply a rate of Temik® 15G higher (5 lbs/acre) than that of the producer (3 lbs/acre). At the Lamesa site, root-knot nematode densities were almost uniformly high, except in the dryland corners, which were outside of the test area (Fig. 5). A sidedress Temik® application was successful in increasing yields at this site in 1999, in areas where root-knot nematode densities were high.

This treatment, which is applied after 5 lbs of Temik®15G have been used at planting, is placed in the soil at pinhead size square, and is suppose to increase the time of protection to the plant against root-knot nematode. The decision rule for Temik® applications at this site was to apply 5 lbs of Temik® 15G at planting across the test area, and then apply 7.5 lbs of Temik® 15G sidedress at pinhead size square. Since the entire test area had a high density of root-knot nemtode, the middle area of the test, which contained vigorously growing cotton in 1999, was used for the variable rate area (Fig. 5).

These fields were photographed again during the 2000 growing season, and the test areas were intensively sampled for root-knot nematode. The Ropesville site again showed the highest concentration of root-knot nematode in the eastern part of the field. It appears that the variable rate treatment was adequately placed to protect against the highest density of root-knot nematode (Fig. 6).

The Lamesa site had lower root-knot nematode densities overall than during the 1999 growing season, but there was no indication that the Temik® 15G sidedress treatment lowered root-knot nematode density below that in the plots which did not receive sidedress treatments (Fig. 7).

Figure 1. Phosphorus fertilizer application map, Ropesville, TX, 2000 (B = blanket-rate, V = variable-rate, Z = zero P)

Figure 2. Phosphorus fertilizer application map, Lamesa, TX, 2000 (B = blanket-rate, V= variable-rate, Z = zero P)

Fig. 3. Infrared Weed Seeker shielded sprayer (Patchen Selective Spray System, Los Gatos, CA)

Fig. 4. False infrared image of Ropesville, TX, 11 August, 1999

Fig. 5. False-infrared image of Lamesa, TX, 11 August, 1999

Fig. 6. False infrared image of Ropesville, TX, 4 August, 2000

Fig. 7. False-infrared image of Lamesa, TX, 4 August, 2000

We are assuming that the two different herbicide application strategies did not differentially affect weed control or lint yields. Variable, i.e. higher, rates of Temik® nematicide were applied to about 2 ac of the 9 ac area of variable-rate/site-specific management plots at Lamesa and Ropesville, respectively. Table 1 shows in the area where variable-rate Temik® (and variable-rate P fertilizer) was applied, lint yields in the VRT plots were statistically similar to the blanket-rate plots and to the blanket-treated Temik® and zero P fertilizer areas. We can conclude, therefore that Temik® applications did not affect lint yields and we can therefore examine yields across the entire experimental area (10 ac of each treatment at each site) to determine P fertilizer affects. Lint yields with variable-rate P fertilizer were significantly greater than zero P plots at both sites (Table 2). The blanket rate of P fertilizer, however, did not affect lint yields, relative to the zero-P control.

Table 1. Cotton lint yields (lb/ac) for variable-rate, blanket-rate and zero-rate fertilizer application, Ropesville and Lamesa, 2000.

Table 2. Cotton lint yields (lb/ac) for variable-rate, blanket-rate and zero-rate P fertilizer application, Ropesville and Lamesa, 2000.

Objective 2. Conduct economic analysis of PA/VRT vs. conventional, blanket-rate farming practices.

Savings in costs of the Roundup® were realized at Ropesville, for VRT vs. blanket-rate ($2/ac and $9/ac, respectively, Table 3). However, greater costs were incurred for P fertilizer and Temik® with variable-rate. Considering income of $0.60/lb cotton and lint yields, $17/ac benefit was realized with VRT compared to blanket-rate applications, $11/ac of which were from variable-rate fertilization.

Phosphorus fertilizer costs at Lamesa were less with VRT than blanket-rate, and Temik® costs were again greater with VRT (Table 4). Variable-rate technologies resulted in a $20/ac benefit when cotton yields when net income was considered, all of which came from variable-rate fertilization.

Variable rate applications of nematicides did not improve yields at these test sites during 2000, and did result in higher pesticide costs than the blanket rate treatments. At six sites in 2000, Temik® 15G rate studies had none or small yield responses (< 10 %) over an untreated check. It is not surprising that a variable rate treatment had no response at the Ropesville and Lamesa test sites, given the poor yield response to Temik® 15G overall during 2000.

This preliminary economic analysis are simply based on input costs and cotton returns. We have not considered capital costs of variable-rate application equipment or the greater cost of 0.5-ac grid soil sampling and laboratory analysis for the VRT treatments. In the future we will do a complete economic analysis that considers these extra costs of VRT.

Table 3. Input application rates and cost and returns of input applications for variable-rate, blanket-rate and zero-rate applications, Ropesville, TX, 2000.

a Assumes no gain in yield b Does not considered capital costs of variable-rate application equipment or the greater cost of 0.5-ac grid soil sampling and laboratory analysis for the VRT treatments.

Table 4. Input application rates and cost and returns of input applications for variable-rate, blanket-rate and zero-rate applications, Lamesa, TX, 2000.

a Assumes no gain in yield b Does not considered capital costs of variable-rate application equipment or the greater cost of 0.5-ac grid soil sampling and laboratory analysis for the VRT treatments.

B. Education/technology transfer:

Objective 3. Demonstrate PA field practices through extension outreach in conjunction with TAEX and Agripartners.

On 1 August, 2000, an education/demonstration field day on Precision Cotton Farming was carried out in Ropesville by TAEX/TAES. Over 60 TAEX county agents and local producers attended. Participants were introduced to PA concepts such as grid soil sampling, remote sensing and yield monitoring. Additionally, training was provided in the use of back-pack GPS receivers.

Through-out the summer, one full-time demonstration technician was provided through the TAEX-Agripartners program. This individual provided valuable technical support in all aspects of the on-farm PA research: grid-soil sampling, data collection, GPS measurements, plant harvesting etc. The demonstration technician was thereby trained in practical PA methodology.

C. Milestones achieved:

Successfully tested variable-rate technologies in cotton on large, field-scale plots. Additionally, preliminary cost and return economic analysis were executed.

D. Publications:

Bronson, K.F. , J.D. Booker, R.J. Lascano, T.L. Provin, H.A. Torbert, and A.B. Onken. 2001. Irrigated cotton yields as affected by phosphorus fertilizer and landscape position. Commun. Soil Sci. Plant Anal. (In press).

Hong Li., R.J. Lascano, J. Booker, T. Wilson, and K.F. Bronson. 2000. State-space description of cotton lint yield variability in a heterogeneous soil on a landscape-scale. Soil Tillage Res (In press).

Bronson, K.F., J.D. Booker, J.W. Keeling, and J.L. Mabry. 2000. Remote sensing of in-season nitrogen requirements for irrigated cotton in the Southern High Plains. 2000 Proceedings Beltwide Cotton Conferences vol. 2. p. 1407. National Cotton Council of America, Memphis, TN.

Wheeler, T. A., L. V. Madden, R. C. Rowe, and R. M. Riedel. 2000. Effects of quadrat size and time of year for sampling of Verticillium dahliae and lesion nematodes in potato fields. Plant Disease 84:961-966.

Wheeler, T. A., B. Baugh, H. Kaufman, G. Schuster, and K. Siders. 2000. Variability in time and space of Meloidogyne incognita fall population density in cotton fields. Journal of Nematology: in press.

Wheeler, T. A., H. W. Kaufman, B. Baugh, P. Kidd, G. Schuster, and K. Siders. 1999. Comparison of variable and single-rate applications of aldicarb on cotton yield in fields infested with Meloidogyne incognita. Annals of Applied Nematology 31:700-708.

Wheeler, T. A. and H. W. Kaufman. 2000. Can remote sensing be used to create variable rate nematicide application maps? Phytopathology 90:S83.

Wheeler, T. A., J. R. Gannaway, and H. W. Kaufman. 2000. The relationship between incidence of Verticillium wilt and reflectance in a wilt nursery. Proceedings of the Beltwide Cotton Conferences, eds. P. Dugger and D. Richter. San Antonio, TX pp. 134-136.

E. Precision agriculture proposals:

Variable rate phosphorus fertilizer applications for irrigated cotton in the Southern High Plains, $3,000/year, 1999-2001, Phosphate and Potash Institute, Foundation for Agronomic Research, Norcross, GA (Principle investigator)

Improving nitrogen fertilizer use-efficiency in deficit-irrigation systems for cotton in the Southern High Plains, $318,500, 1999-2002, USDA/CSREES National Research Initiative (Principle investigator)

Updating Cotton Soil Test Calibrations and Fertilizer Recommendations for Nutrient Management Plans and Increased Profit in the High Plains, $20,000, 2001-2002, Cotton Incorporated, Texas State Support Committee (Principle investigator)

F. Precision Agriculture meetings attended/papers (posters) presented:

Bronson, K.F., R.J. Lascano, J.D. Booker, J. Booker, S. Machado, E. D. Bynum, Jr. T.L. Archer, H. Li, and J.W. Keeling. 2000. Grid soil sampling: Comparisons of grid size with landscape and soil texture-based sampling strategies. Precision Agriculture. Proceedings of the 5th International Conference, ASA, Minneapolis, MN. (In press).

G. Other developments: none