CHEMISTRY AND GROWTH OF COTTON IN RELATION TO SOIL FERTILITY AND ROOT-ROT1

J. E. ADAMS, R. C. WILSON, L. E. HESSLER, AND D. R.

NUMEROUS reports have been made of investi-gations concerned with the factors influencing

the incidence of root-rot, a disease caused by the fun-gus Phymatotrichmn omnivorum (Shear) Duggar.Summaries of many of these studies have been madeby Streets (8)3 and by Rea (5).

Experiments of the Division of Soil Fertility In-vestigations, Bureau of Plant Industry (4), haveshown that, on Wilson very fine sandy loam, soil ofthe Blackland prairie section of Texas, highly signifi-cant reductions in mortality have been associated withfertilizers in which the effect of nitrogen dominated.Also, highly significant increases in amounts of deadcotton were found on plats receiving fertilizers inwhich the effect is that of phosphate. Other work ofthis Division on Wilson clay loam soil has shown thatthe composition of the root bark of cotton is mate-rially altered by fertilizer treatment (3). Watkins(9) has presented evidence that chemical action, bysecretions of the fungus, is a factor involved in theentry of the fungus into the plant root. Combinedfacts emphasize the study of the root bark as relatedto attack by the fungus. Studies of other plant parts,or combination of parts (i, 2), offer fundamentalinformation on the effect of fertilizer on the com-position of cotton plants at various stages of growthon different soil types.

The work presented here shows that differences inmortality of cotton secured on Wilson very fine sandyloam soil are associated with outstanding differencesin the total nitrogen and total P2O5 contents of theroot bark, and to a somewhat lesser extent with thecomposition of the leaves. The mortality data havebeen reported (4), but are repeated here in part topresent the relation of chemical composition to theincidence of root-rot.

EXPERIMENTAL PROCEDURE

Sampling technic and methods of analysis were the sameas outlined in previous publications (i, 6). Treatments werereplicated six times on plats4 arranged as a 6 x 6 Latin square.

The fertilizers used were 0-15-0," 3-9-3, 9-3-3, 9-3-0, and15-0-0, with unfertilized soil serving as a check. The appli-cation was 600 pounds per acre for all ratios other than the9-3-0; it was applied at the rate of 750 pounds to furnish thesame total amount of plant food as the other ratios. Theexperiment was located on an area of Wilson very fine sandyloam soil near Caddo Mills, Hunt County, Texas. This fieldhad been cropped continuously to cotton for at least five yearsprior to this experiment, which was begun in 1936.

Samples of root bark were taken op August 15, and Sep-tember 20, for each of the 36 plats. Analysis of individualsamples allowed the data to be treated by variance methods.Leaf samples were taken on July 20, and August 15; theanalytical data are not presented but are discussed in termsof their correlation with the mortality data.

EXPERIMENTAL DATA

Table I gives the average composition of the root-bark samples for all treatments. The sampling onAugust 15 was intermediate in respect to the dates ofrecord for percentages of dead cotton, so the root-rotdata are given for both August 2 and 30. Actualvalues for the unfertilized check are given, but thevalues for the five treatments are expressed as differ-ences in respect to the check.

On August 15 it was found that the 15-0-0, 9-3-0,and 9-3-3 samples contained amounts of nitrogengreater than that of the check; these differences werehighly significant. The 3-9-3 sample did not differfrom the check, but phosphate alone, 0-15-0, reducedthe nitrogen content by a highly significant amount.The statistical import of the root-rot records is thesame for both August 2 and August 30, i.e., highlysignificant reductions in percentages dead cotton areassociated with the 15-0-0, 9-3-0, and 9-3-3 treat-ments, and a highly significant increase with the0—15—0 fertilizer. The values of the correlation coeffi-cient, r (7), are given in Table 2. That for the rela-tion between nitrogen contents of root bark and per-centages dead cotton is —0.950 for August 2, and—0.833 f°r August 30. The relation between P2O5content and dead cotton is +0.228 and +0.242, re-spectively, for August 2 and 30. These data indicate

'Contribution from the Division of Soil Fertility Investigations, Bureau of Plant Industry, U. S. Dept, of Agriculture.2Soil technologist in Charge of the Soil Fertility Cotton Root-Rot Investigations at Austin, Texas; Junior Chemist and Assist-

ant Chemists, respectively. These investigations are under the general suprvision of Dr. J. J. Skinner, Senior Biochemist, in chargegf cotton soil and fertilizer investigations.

''Figures in parenthesis refer to "Literature Cited", p. 332.Acknowledgment is made to Mr. D. R. Hooton of the U. S. Cotton Field Station, Greenville, Texas, and to Messrs. H. V.

Jordan, H. A. Nelson, and P. M. Jenkins for the management of the plats and assistance in preparation of the samples.6N—P2O5-K2O. One-half of the nitrogen was derived from nitrate of soda, and one-half from sulfate of ammonia; the P2OS from

18% superphosphate, and the K2O from sulfate of potash.

329

33° SOIL SCIENCE'SOCIETY PROCEEDINGS 1939

TABLE i.—Composition of root bark of cotton plants and percentages of dead cotton on Wilson very fine sandy loam soil, Hunt County,Texas; 1939 season.

Fertilizertreatment

0-15-03-9-39-3-39-3-oi=;-o-o

Check. . . . .

P = o.osP=O.OI

Sample of Aug. 15

N in root bark%

-0.13**+0.05+o.3ift+0.32ft+o.34tt

1.92

0.09O.I2

P2OS in rootbark, %

+o.oi9f+o.o38ft+0.028|t+O.O2lf-O.OlS

0.159

O.OI90.026

Sample of Sept. 20

N in root bark,%

-0.06-0.06+0.05+ O.I2f+o.i7tt

i-93O.IO0.14

P2O5 in rootbark, %

+o.ositt+o.03off+0.024!+0.013-0.029*

0.196

0.022O.O3O

Percentage dead cotton

Aug. 2

+ H.6tt- 4.8-24.8**-23.8**-25.9**

26.5

9-i12.4

Aug. 30

+ i8.4tt- 5-6-34.2**-33-7"-35-7**

36.911.615-8

Sept. 19

+22.8ft0.0

-37.8"-37.2**-40.4**

41.814.72O.O

*Less than on check plats (P =0.05).**Less than on check plats (P=o.oi).•(Greater than on check plats (P =0.05).

ttGreater than on check plats (P =0.01).

a highly significant inverse association of nitrogencontent of root bark and the mortality of cotton. ThePsOs relation is not significant.

The data for September 20 show that differencesin total nitrogen content of the root bark were ob-tained only with the 15-0-0 and 9-3-0 fertilizers;these furnished 90 and 67.5 pounds of nitrogen peracre, as compared with 54 pounds for the 9-3-3 ferti-lizer. The increase due to the 15-0-0 ratio was highlysignificant, while that for the 9-3-0 differed from theunfertilized check by a significant margin. While thedifferences in nitrogen content of the bark have be-come less numerous and outstanding as the seasonadvanced those for P2Og have improved in orderli-ness. The highly significant increase for nitrogen dueto the 15-0-0 fertilizer is accompanied by a signifi-cant reduction in PgOg content. While the 9-3-3,3-9-3, and 0-15-0 treatments are without effect onthe nitrogen content, they have increased the P2C>5content. The differences in amounts of dead cottoncarry the same statistical import on September 19 ason August 2 and 30, i.e., highly significant reductionsare associated with the 15-0-0, 9-3-0, and 9-3-3fertilizers, with a highly significant increase in per-centages of dead cotton on the 0-15-0 plats.- The

value of r for the nitrogen contents of the root-barksamples of September 20 with percentages dead cot-ton on September 19 is —0.618, a highly significantinverse correlation. The r value for PzOs contentswith percentages of dead cotton is +0.596, a highlysignificant positive correlation. That for percentagesnitrogen with percentages PzO5 is —0.500, a highlysignificant inverse correlation. The relation of nitro-gen and PzOs contents on August 15 was not signifi-cant, the value of r being —0.131.

Samples of 20 leaves per plat were taken at random,from near the middle of the plants, and composited foranalysis. These samples were not as comparable asthose of root bark as it was difficult as the seasonadvanced to be certain that the leaves were of com-parable physiological age. The dates of sampling wereJuly 20, and August 15. These data are not presentedin tabular form. The value of r for nitrogen contentsof July 20 with percentages dead cotton for July 19was —0.721, a highly significant value. The value forthe samples of August 15 with the August 2 mortalitydata was —0.478, and with the August 30 data,—0.501. Both of these values indicate a highly signifi-cant inverse correlation of nitrogen in the leaves andmortality of cotton. The P2Os contents of the leaves

TABLE 2.— Values of correlation coefficient, r,for leaf and root-bark samples from cotton plants, and per cent dead cotton, Wilson veryfine sandy loam soil, Hunt County, 1(139.

Leaves:07 NT%P 2 0 S . . .

Root bark:% N . . . . . .%p a o s . . .

July 20

Leaves, %P20S

+0.184

——

Root-rot, %July 19

-0.721**-0.056

• ——

Aug. 15

Leaves, %P;0S

+o.378f

——

Root bark,% P,05

-0.131

Root-rot, %Aug. 2

-0.478**+0.317

-0.950**+0.228

Root-rot, %Aug. 30

-0.501**+0.328

-0.833**+0.242

Sept. 20

Root bark,% P,OS

———

-0.500**

Root-rot, %Sept. 19

-0.618"+o.596ft

tor * =Significant value =0.330.f-f-Or ** =Highly significant value =0.424.

ADAMS, ET AL. I COTTON ROOT-ROT 331

on August 15 with percentages dead cotton on August2 give a value of +0.317, and for August 30 the valueis +0.328. These values approach significance, 0.330being required. The correlation between nitrogen andPzQs contents of leaves is significantly positive onAugust 15, but no significance was found on July 20.

In an experiment of this Division (4) on Houstonblack clay soil 0-15-0, 3-9-3, 9-3-3, and 15-0-0fertilizers were used under a continuous cottonsequence. An additional series of 3-9-3 plats wereused to complete a 6 x 6 Latin square; this secondseries differed from the first 3-9-3, and other series,in that the stalks were turned as soon as possible afterthe cotton was picked, with bedding at a later date,as compared with the prevailing practice of beddingin December or January. This early plowing waspracticed in 1937 and 1938. Analyses of leaf sampleson July 14, 1938, showed significant increases in thenitrogen content for the 15-0-0 treatment and the3-9-3 plats which received the fall plowing, and ahighly significant decrease on the 0—15-0 plat. The0-15-0 samples also showed a highly significant in-crease in P2O5 content. The level of infestation wastoo low, however, to allow the mortality data to beanalyzed with assurance. Composite samples of leavesfrom the 15-0-0 and both 3-9-3 series of plats takenin 1939 failed to show any appreciable difference innitrogen content. No significant difference in mortal-ity was found in 1939 associated with fertilizer treat-ment, but the turning of the stalks in 1937 and 1938as soon as the cotton was picked resulted in highlysignificant and economically important reductions inpercentages of dead cotton at all dates of record. Theaverage reduction was from 29 to 2% dead cotton.The results on this soil from fertilizers used undercontinuous-cotton sequence show the same trends inrespect to mortality of cotton as shown for the Wilsonsoils. It would seem that the difference is in degreerather than in kind. Differences are associated withthe extremes in fertilizer composition, i.e., 15-0-0and 0—15-0 ratios, largely, and are not so pronouncedas on the Wilson soils.

DISCUSSIONIt has been shown that differences in mortality of

cotton, under root-rot conditions on the lighter soilsof the Blackland prairie section of Texas, are asso-ciated with fertilizer treatment. These differences be-come less pronounced as a shift is made from theWilson very fine sandy loam to Wilson clay loam,which are non-calcareous or slightly acid soils, to thecalcareous Houston black clay soil. Reductions in

mortality have been secured on the Houston soil byturning the stalks as soon as the cotton was picked,with bedding done later in the winter, as comparedwith the prevailing practice of allowing the stalks tostand until the land is bedded in late winter or earlyspring.

It appears reasonable in the experiment on Wilsonvery fine sandy loam soil, in Hunt County, to at-tribute both the increases and decreases in the mor-tality of cotton to induced chemical composition, in-asmuch as tillage was the same for all treatments.The chemical data, as a whole, indicate that the nitro-gen and P2Os concentrations are concerned in theextent to which cotton plants are killed by the root-rot fungus. Whether or not this is due to some definitenitrogen or phosphorus-bearing compound in the barkof the root is not known. The data presented herewould seem to indicate that the chemical factor is oneof equilibrium in which nitrogen and phosphorus, andperhaps other compounds, are involved. If it is a mat-ter of equilibrium then it would seem that the factorswhich affect the response of plants to nitrogen andphosphate fertilizers could be expected to influence,indirectly, the plant in its relation to root rot.

In the experiment where reduced mortality accom-panied early fall plowing it would seem that some fac-tor other than chemical composition has played themajor role, augmented, perhaps, by favorable com-position. The major effect may have been one ofsmaller numbers and lessened viability of sclerotia.The effect of fall plowing might oppose the influenceof chemical composition, e.g., that of an 0-15-0 ferti-lizer, if applied to the same area of cotton.

SUMMARY

The total nitrogen and ?2O5 contents of samples ofroot bark and leaves of cotton plants produced onWilson very fine sandy loam soil are discussed inrelation to the mortality of cotton caused by the fun-gus Phymatotrichum omnivorum (Shear) Duggar.Samples were taken from plats arranged as a 6x6Latin square to test the effect of 0-15-0, 3-9-3,9-3—3, 9-3-0, and 15-0-0 fertilizers on the incidenceof cotton root-rot. Data analyzed by variance methodsare quoted to show that highly significant reductionsin mortality were associated with the 15-0-0, 9—3-0,and 9—3-3 fertilizers, and highly significant increaseswith the 0-15-0 ratio.

The nitrogen contents of the root bark samples ofAugust 15 and September 20, and leaf samples ofJuly 20 and August 15 were affected sufficiently to

332 SOIL SCIENCE SOCIETY PROCEEDINGS 1939

give highly significant negative values for the cor-relation coefficient, r, between nitrogen contents andmortality data. The value of r between ^zQz contentsof root bark on September 20 and mortality at a com-parable date is positive and highly significant. Ahighly inverse correlation is found between PsOs andnitrogen contents of root-bark samples taken on Sep-tember 20. These data indicate that chemical com-position of root bark and leaves of cotton plants, asaffected by fertilizer treatment, is correlated with themortality of cotton on the plats from which sampleswere taken.

Differences in the incidence of the disease as af-fected by fertilizer treatment are not as numerous oras outstanding for the soils of the Houston series asfor those of the Wilson. The turning of the stalks assoon as the cotton was picked, with bedding later inthe season, as compared with the prevailing practiceof bedding in December or January, has resulted inoutstanding reductions in mortality. This appears tobe due to an effect on the fungus, abetted by chemicalcomposition.