Testing Variable-Rate Phosphorus Fertilization for Cotton in the Texas High Plains

Summary

            Phosphorus (P) fertilizer response in cotton is often not consistent.  Fertilizer applications in a variable-rate approach can theoretically match soil test P and P fertilizer rate on a site-specific basis.  The objective of this study was to compare variable-rate-P, blanket-rate-P, and zero-P applications on two 27-ac. irrigated sites in the Southern High Plains. Soil samples (0-6 in.) were taken in the spring of 2000 and 2001 in Ropesville and Lamesa, TX, on a 0.5 ac grid, and analyzed for Mehlich-3-P. Averaged across landscape position, lint yields in Lamesa responded to variable-rate-P in 2000.  In 2001, Lamesa lint yields responded to blanket-rate-P in the south-facing sideslope only.  At Ropesville in 2000, cotton biomass and P uptake at early squaring were less in a calcareous soil than a noncalcareous soil. Cotton lint response to variable-rate-P was observed on the non-calcareous soil only.  In summary, both variable-rate and blanket-rate-P responses were inconsistent in this study.  One of the most significant results of this study was that in two of three site-years, less P was applied with the variable-rate approach than with the recommended blanket-rate.  However, landscape position/slope at Lamesa, and soil type/soil calcium at Ropesville had greater impact on cotton than P management.

Introduction

Phosphorus fertilizer response in cotton is unpredictable, even with soil tests.  Nitrogen and water are the main limitations in the Southern High Plains, with P probably being the third limiting factor. There have not been many studies on cotton P fertilizer response at the landscape scale.  Landscape position has been widely reported to affect productivity and N response of crops like wheat, due to redistribution of water to lowerlying landscape positions. 

Our objectives were to: (i) compare variable-rate-P, blanket-rate-P, and zero-P applications on P accumulation and lint yields at two 27-ac irrigated cotton sites in the Southern High Plains, the secondary objective was to (ii) determine the effect of landscape position and soil properties on P fertilizer response, and lint yields.

Experimental design

Studies were conducted at two sites in a randomized complete block design with three replicates, each replicate within a center-pivot irrigation span.  Phosphorus treatments were zero-P, blanket-rate-P and variable-rate-P.  Plots were sixteen, 40-in. rows wide, and ranged up to 1000 yards in length.   Phosphorus fertilizer was applied as phosphoric acid (0-34-0) near the time of planting with a placement of 4 in. from the seed row and 4 in. deep.  The blanket rate of P was 45 and 30 lb P2O5/ac in both years at Lamesa, and Ropeville, respectively (Table 1) This was based on Mehlich-3-P in the 0-6-in. soil.  Variable-rate-P was applied with the same ground rig, which was fitted with an Agchem/Soil Tec Inc. Fertilizer Applicator Local Controls Operating Network (FALCONTM).  This consisted of variable-rate servo valves, field-duty computer, controlling software, ground-speed radar, hydraulic motor driven centrifugal pumps, flow meters, and shut-off valves.  A submeter accurate Satloc® differential global positioning system (GPS) receiver was used with the FALCONTM system.  The FALCONTM software uses the inverse-distance method of interpolation in calculating its variable-rate application maps (Lamesa site shown in Fig. 1).

Lamesa. 

The soil here is Amarillo sandy/sandy clay loam.  In March 2000, soil samples were taken at 63 GPS-referenced points in the 27-ac. experimental area.  Ten subsamples were taken, by hand soil probe, of the 0-6-in. depth and analyzed for Mehlich-3-P and nitrate.  Two subsamples were taken of the 6 to 24-in. depth for nitrate analysis.  A blanket nitrogen fertilizer application was based on an nitrogen requirement of 120 lb N/ ac for a 2 bale/ac yield goal and considering the 0-24-in. soil nitrate. On 10 May 2000, ‘Paymaster ® Round-up Ready 2326’ (‘PM 2326 RR’) cotton was planted into 40-in. rows.

Hand harvesting of lint was done on 22 ft. of row at each GPS referenced point.  Seedcotton was ginned and lint weighed for each sample.  In spring, 2001, the same soil sampling practices and soil analyses were done as in 2000. ‘PM 2326 RR’ cotton was planted on 28 May, at which time P fertilizer was applied, similar to 2000.

Ropesville

The two soil series here are Amarillo sandy/sandy clay loam, and Portales sandy clay/sandy clay loam. In March 2000, soil samples were taken at 60 GPS-referenced points in the 27-ac study area, in the same manner as in Lamesa and analyzed for nitrate and Mehlich-3-P.  On 6 May 2000, ‘PM 2326 RR’ cotton was planted in 40-in. rows.  A blanket nitrogen application was based on 0-24 in. soil nitrate.  Hand harvesting of 22 row ft. per GPS point was on 29 September.  On 2 May, 2001, ‘PM 2326 RR’ was planted, but a hail storm with 6 in. rain on 25 May destroyed the crop.

Results

Lamesa

Relative elevation and slope with the row were negatively correlated with biomass, and P accumulation at early squaring, and with lint yield (data not shown).  Soil properties did not have much affect on cotton in 2000 or 2001. 

Greater yields on the lowerlying bottomslope at this site were reported previously.  In early studies, we measured greater soil water content in the bottomslopes, presumably due to runoff of rain and irrigation.  Therefore, we delineated bottomslope from sideslope using slope with the row data (derived from elevation data). The bottomslope had the lowest slope with the row (< 0.4%).  Landscape position and landscape x P treatment was incorporated into the statistical analysis. 

Cotton biomass and P accumulation at early squaring were not affected by P fertilizer, but were greater in the bottomslope than in the sideslopes (data not shown).  Lint yield in the bottomslope was greatest in both years of the study (Table 1).  In 2000, lint yield in the north-facing sideslope was less than in the south-facing sideslope.   As mentioned above, water is apparently redistributed to the bottomslope in this location, resulting in greater plant growth.  Response to P fertilizer was less consistent than landscape position.  Lint yields responded to variable-rate-P in 2000, averaged across landscape position. In any particular landscape however, P did not affect lint yields.   In 2001, Lamesa lint yields responded to blanket-rate-P in the south-facing sideslope only (Table 1). 

In 2000, more P was applied to the variable-rate plots (average of 38 lb P2O5/ac) than to the blanket plots (Table 2).  Less P was applied in 2001 to the variable-rate plots (average of 16 lb P2O5/ac), despite similar soil P levels.

Ropesville

Strong negative correlations were observed between P accumulation, biomass at early squaring and soil calcium (data not shown).  High soil calcium levels were associated with the calcareous Portales soil, on the west half of the Ropesville study site.  Unlike the Lamesa site, there was no correlation between elevation or slope with the row and plant measurements. We therefore delineated the two soil types in our data sets, and incorporated soil type, and soil type x P management into the statistical analysis.

Phosphorus accumulation (Table 3) and biomass (data not shown) at early squaring were both lower in the Portales soil then in the Amarillo soil.  In both blanket- and variable-rate-P treatments, responses in P accumulation and biomass were observed, relative to the zero-P plots, but only in the Amarillo soil.  Apparently soil and added P was less available on the calcareous Portales soil.  A lint yield response to both blanket-rate and variable-rate-P was observed in the Amarillo soil only (Table 3).  Averaged across P management, lint yield was similar between soil types.  Less P was applied with variable-rate (on average) than with blanket-rate-P, despite similar soil test P levels (Table 2). 

Fig. 1.  Phosphorus fertilizer application map for cotton, Lamesa, spring, 2000.  V, B, and Z are variable-rate-P, blanket-rate-P, and zero-P, respectively.

Table 1. Lint yield as affected by phosphorus fertilization and landscape position, Lamesa, TX, 2000 and 2001.

Phosphorus treatment

North-facing sideslope

Bottomslope

South-facing sideslope

Mean
 

---------------------------------------- lb/ac -------------------------------------
 

2000

Variable-rate-P

687 a†

762 a

575 a

675 a†

Blanket-rate-P

602 a

690 a

564 a

619 ab

Zero-P

604 a

671 a

521 a

598 b

Average

631 b‡

707 a

553 c

  
 

2001

Variable-rate-P

 892 a†

1047 a

866 ab

935 a†

Blanket-rate-P

963 a

1013 a

885 a

953 a

Zero-P

896 a

1021 a

769 b

896 a

Mean

916 b‡

1026 a

840 b

  

† Means in a column followed by the same letter are not significantly different by pairwise t test (P=0.05)

‡  Means in a row followed by the same letter are not significantly different by pairwise t test (P=0.05)

Table 2. Cost and returns of phosphorus fertilizer applications, Ropesville, 2000 and Lamesa, TX, 2000-2001

 

Lint yield (lb/ac)

Lint income† ($/ac)

Lint income above zero-P

Ave rate of P2O5  (lb/ac)

Cost of P2O5 ‡ ($/ac)

Net return to P fertilizer ($/ac)
 
Ropesville, 2000

Variable-rate-P

643

321.50

35.50

41.9

14.66

20.84

Blanket-rate-P

623

311.50

25.50

45.0

15.72

9.78

Zero-P

572

286.00

  

0

0

  
 
Lamesa, 2000

Variable-rate-P

675

337.50

38.50

38.4

13.44

25.06

Blanket-rate-P

618

309.00

10.0

30.0

10.50

-0.50

Zero-P

598

299.00

  

0

0

  
 
Lamesa, 2001

Variable-rate-P

935

467.50

19.50

15.9

5.57

13.93

Blanket-rate-P

953

476.50

28.50

30.0

10.50

18.00

Zero-P

896

448.00

  

0

0

  

† $ 0.50/lb lint

‡ $ 0.35/lb P2O5

Cost and returns to variable-rate phosphorus fertilization

            Cost and returns analysis for both blanket-rate and variable-rate-P fertilization is shown in Table 3.  This analysis does not consider the extra costs incurred with variable-rate compared to blanket-rate-P fertilization.   These include cost of grid soil sampling, additional laboratory analyses costs of the larger number of soil samples, and the cost of retro-fitting a liquid fertilizer rig with variable-rate equipment, GPS and ground speed radar.

            Compared to zero-P, blanket-rate-P returned positive cash in Ropesville, 2000 and Lamesa, 2001, but gave a negative return at Lamesa in 2000 (Table 3).  Variable-rate-P applications gave positive cash returns in all three site-years.  However, in Lamesa, 2001, the cash return to P fertilization was greater for blanket-rate-P ($18.00/ac) than for variable-rate-P ($13.93/ac).

Kevin Bronson

J.W. Keeling

R.K. Boman

R.J. Lascano

Texas Agric. Exp. Stn. & Texas Coop. Extn.

 Lubbock

Ph. 806-746-6101

k-bronson@tamu.edu

Table 3.  Cotton phosphorus accumulation at early squaring, and lint yield as affected by phosphorus fertilization and soil type, Ropesville, TX, 2000.

Phosphorus treatment

Amarillo sandy loam

Portales clay loam

Mean
 

Phosphorus accumulation
 

--------------------------- lb P2O5/ ac ------------------------------

Variable-rate-P

5.2 a†

3.4 a

4.3 a†

Blanket-rate-P

5.0 a

2.7 a

3.9 ab

Zero-P

3.4 b

3.2 a

3.2 b

Mean

4.5 a‡

3.2 b

  
 

Lint yield
 

--------------------------------- lb/ac ------------------------------

Variable-rate-P

663 a†

622 a

643 a†

Blanket-rate-P

646 a

600 a

623 a

Zero-P

538 b

606 a

572 a

Mean

616 a‡

609 a

  

† Means in a column followed by the same letter are not significantly different by pairwise t test (P=0.05)

‡  Means in a row followed by the same letter are not significantly different by pairwise t test (P=0.05).