NUTRIENTS IN SOILS

Heterogeneity of Soil of Agricultural Land in Relation to Soil Sampling

MICAH W. M. LEO' Soils Department, Rutgers University, New Brunswick, N. J.

The heterogeneity of soil fertility in apparently uniform, agricultural lands was studied. One-hundred composite soil samples were taken from four New Jersey farms. Their available N, PI K, and Mg were determined, and statistical analysis was employed. The coefficient of variability of various available nutrients ranged from 10 to 537& The findings indicate that the heterogeneity of soil fertility of agricultural soils i s far beyond that commonly anticipated. Optimum sample sizes for various nutrients calculated from the data raised a question as to the validity of the present, generally accepted methods of soil sampling. New approaches should therefore be taken to meet the ever-changing soil fertility levels due to modern farming practices, especially intensive use of concen- trated chemicals. Further investigation in this direction appears to be urgently needed.

OIL has been considered as a hetero- S geneous body, a mixture of organic and inorganic materials. Naturally, the composition of soils varies from location to location and from site to site ( 7 , 2, 5, 7). However, to a certain extent, soil can also be considered as a homogenous body, in a physical sense, within a limited area ( 7 . 2). For these reasons. the knowledge of sampling has been recog- nized as one of the most important keys to soil tests. The validity of soil analysis depends on the fairness of samples as well as the technique employed in studying the samples.

Tisdale and Nelson (9 ) recommend that each composite sample should con- sist of borings from 15 to 20 locations over an average field of 10 acres. Graham (2) suggests that a final sample consisting of 5 to 10 of the composites of cores per 40 acres would give quite suf- ficient and accurate information in that field for most soil management practices.

The purpose of taking composite sam- ples is to attempt to minimize the in- fluence of nonuniformity of a given soil. Tisdale and Selson ( 9 ) : however, point out that the composite sample which is approximately equivalent to a pound from an average field of 10 acres (1 pound per 20,000.000 pounds of surface soil a t plow depth) is usually extremely small and would likely cause consider- able error. But for labor and time saving, a practical operation must be adapted.

The inherent soil fertility of virgin soils, as a rule, is less heterogeneous than the soil fertility of farmed lands where various agricultural activities such as liming, manuring, and fertilization have been conducted. This study reveals

Present address: Isotopes, Inc., 123 Woodland Ave., Westwood, N. J.

432 A G R I C U L T U R A L A N D

that the soil sampling methods widely used at present require further critical tests.

Materials and Methods

Four apparently uniform lands about 10 miles apart were selected from New Jersey farms ( d ) which had been cul- tivated since the early settlement of the area. The crops of the year before soil sampling were: oats-red clover mixture, corn. corn, and sod for the College farm, Forsgate farm, A. Pinter farm, and E. Pinter farm, respectively. An area of 60 X 250 feet was chosen from each of the farms and subdivided into 25 plots of 12 X 50 feet. Soil samples were taken from the plots a t a depth of about 7 inches with a Hoffer soil-sampling tube. Each composite sample from a plot con- sisted of 12 borings from different sites selected at random. Soil samples were air dried and mixed well to eliminate possible errors which may arise from subsampling. The soils were passed through a 2-mm. sieve and analyzed for their available P205, K20. and Mg content with the methods described by Hanna and Purvis (3) . The soils passed through a 1-mm. screen were tested for their potentially available nitrogen by the procedure of Purvis and Leo (4, 6 ) . Statistical analyses of data were em- ployed (8). The optimum number of randomly selected samples assumed with a precision level of the sample-mean within P = 5% were calculated from the coefficients of variation at the 0.10, 0.05, and 0.01 levels of significance.

Results and Discussion

Distributions of the available nutrients in the soils from the four New Jersey farms are shown semidiagrammatically in Tables I to IV.

F O O D C H E M I S T R Y

Nitrogen contents ranged from 62 pounds per acre at the Forsgate farm to 155 pounds per acre a t the A. Pinter farm; PeOj. from 1 pound per acre a t the E. Pinter farm to 244 pounds per acre a t the Forsgate farm; K20, from 35 pounds per acre a t the A. Pinter farm to 600 pounds per acre a t the Forsgate farm; Mg, from 15 pounds per acre a t the E. Pinter farm to 180 pounds per acre a t both the Forsgate and A. Pinter farms. There were definite differences in soil fertility among the various farms.

Table V shows the means, standard deviations, standard errors of means, calculated t-values. and coefficient of variation of the available nutrients from each of the four New Jersey soils. The standard deviations of all nutrients of the soils were beyond the confidence limits of significance a t the 1% level, and the variability of the nutrients in soils was remarkably great. As a whole, N tests showed the least variation, Mg the greatest, with K 2 0 and PZOS in between according to the order as listed. Mc- Kenzie (5) found that all nutrients in a soil tend to rise and fall together in virgin soils. Since there is no such tend- ency in the farmed soils studied here, it would be reasonable to assume that agricultural activities, such as intensive use of chemical amendments, have drastically changed the nutrient status of the soils.

In this experiment, as mentioned pre- viously, one composite sample of 12 borings was taken from each plot of 12 x 50 sq. ft. (or 1 j72 .6 acre). i.e., about 1 pound of soil sample from 27,550 pounds of surface soil. In comparison with the amounts of soils recommended by Tisdale and Nelson ( 9 ) and by Graham (21, i.e., 1 pound from 20 million pounds and from 80 million pounds. respectively, the size of the

Table 1. Distribution of Available Nutrients in Soil in the College Farm

(Values in Dounds Der acre)

Table II. Distribution of Available Nutrients in Soil of the Forsgate Farm

acre) (Values in pc

450 500 135 125

nds per 87 88

560 120

N . . , I 1 2 2 N . . . . . P20,. . . K 2 0 . . . M g . . , ,

N . . . . . P,Oa I . .

K20. . . M g . . , ,

N . . . . . P205. . . K20. . , M g . . , ,

N . . . . . PgOa.. . K 2 0 . . , M g . . . . N . . . . . . Piob . . . KZO.. , M g . , . .

100 5

75 45

111 5

90 40

117 7

95 50

99 8

65 40

101 12 90 45

104 23

140 100

99 7

75 30

106

30

PzOb.. . 6 K2O. . . 80 M E . . . . . 40

96 4

80 45

132 11

125 45

--

82 76

500 150

101 9

110 75

119 9

110 40

108 9

100 45

100 8

85 100

164 42 5 360 180 135

100 104

73 1 ! 128 I 2:; 1 71 120

N . . . . . 124

.Mg. . . .

N . . . .

K 2 0 . . . 120 M g . . . . 60

500 400 540 125 150 1 135 I

120 9

90 75

110 10 90 60

95 9

100 35

87 87 276 1 184 520 460

82 80 ’ 89 176 i 164 244 ~

44 0 520 ~ 520

+E- 138 5

100 30

105 6

100 30

92 5

130 30

116 116 420 I 3 2 0 160 165

156 480 540 400 135 125 140

w Table 111. Distribution of Available Nutrients in Soil

of the A. Pinter Farm (Values in pounds per acre)

2 5 5 45 80 55 ~ 180 180 180 180 110

Table IV. Distribution of Available Nutrients in Soil of the E. Pinter Farm

(Values in pounds Der acre) N . . . P205 KZO . M g . .

. . . . . . . . .

N . . . . . 1 9 : ‘ 8 : 1 7; 1 8; ~

P205.. . K 2 0 . . . 80 70 80 105 105 M g . . . . 25 30 30 30 30

N . . . . . Pzob. . . KZO.. , M g . . , .

N . . . . . P20,. . . K 2 0 . . . M g . . . .

180 180 1 180 180

125 130 121 127 118

~ l:i ~ 635 ~ 6; 1 3; ~ 5; ~ 180 180 180 135

N N . . . . . P20a. . . KZO.. . M g . . . .

PZOS.. . KZO. . . Mg . . . . I 40

1 125 I 75 I 90 I 5 5 I 120 150 125 180 150

N..... P 2 0 6 . . . K20. . . Mg . . . .

N . . . . . P 2 0 6 . . . KzO.. . Mg . . . .

VI 0

70 65 60 80 90 180 180 120 90 180 -

12 ft . - 12 ft .

sample used in this study should be con- sidered more reliable with less error introduced by sampling. Holyever, the statistical analysis indicates that there were still tremendous variations in vari- ous nutrients in the soils from different plots a t each location.

A stratified sampling method is sug- gested for a land with inherent hetero- geneity in soil fertility. HoLvever, it is neither practical nor possible to apply the method in most agricultural soils which exhibit apparent uniformity but actuall>- are heterogeneous.

Table VI shows the calculated opti- mum sizes of sample which are required to satisfy the probability at the 0.10, 0.05, and 0.01 significance levels; respectively.

The optimum size of sample for each nutrient varies significantly from location to location, suggesting that there is no definite pattern of optimum size of sample for various nutrients studied in any single location. For instance. should the 0.05 level of significance be con- sidered, 336 samples should be taken from the College farm for a reliable PaOs test: while only three samples would be enough from the same farm for a repre- sentative M g test. These results also reveal that there is no definite pattern of an optimum size of sample for any one nutrient in various locations. For ex- ample, should the 0.01 level of signifi- cance be demanded, 1331 observations should be required for a S test a t the

Forsgate farm. and only 2- should be needed for both the A. Pinter and E. Pinter farms. I t is. therefore. concluded that the optimum size of sample for vari- ous nutrients of the soils studied ranged from three to more than a thousand ob- servations.

The results strongly suggest that the sampling methods currently used are far from being ideal for rhose agricultural lands similar to the ones studied here. The sampling sizes of the methods cur- rently in use hardly approach the lowest extremes of the optimum sampling size calculated.

If a piece of land is fairly homogeneous in its available nutrients, as shokvn in rhe case of the E. Pinter farm where little

VOL. 1 1 , NO. 5, SEPT.-OCT. 1963 433

Table V. Statistical Analysis of Four Agricultural Soils in New Jersey

Available Sail Stofistical

Nutrients Measurea Col lege Farm

108.96 12 .38

43 .94 11 .36

8 . 1 6 3 . 7 4 0.756

10 .88 45.81

100.00 20 .41

24.50 20 ,41

50.00 2 .17 0.436

116.27 4 . 3 4

2.48h

4.08b

Forsgate Farm

79.00 17 .53

3.516 22 .51 22.19

142.88 49.79

9.966 14 .34 34 .85

474.00 73 .21 14.646 32.38 15.45

147.80 22 .96

32.20 15 .54

4.59b

A. Pinter Farm

124.32 12.36

2.47b 50.33

9 .94

5 .48 2 .89 9,585 9 .45

52.73

77 . O O 31.42

12 .26 40.81

162.40 28.36

5.76h 28.64 17.47

6.28‘

E. Pinfer Farm

90.16 8 , 9 4 1.79b

50.37 9 . 9 1

1 .80 0 . 4 1 0.08b

22.50 22.67

79.20 14 .91

26.58 18.82

25.20 6.99 1.406

18.00 27.75

2.986

?. mean; S. standard deviation; Sf , standard error of mean, .411 units are C. coefficient of \,ariation; t , calculated t - value.

in pounds per acre except C which is in percentage of the mean. h Significant at the 0.01 lekel.

Table VI. Optimum Size“ of Sample Calculated from Data of Four New Jersey Farms at the 0.1, 0.05, and

0.01 Levels Precision

Available Level as Sail Percent

Nutrienfs o f Mean

N 0 .10 0 .05 0 .01

PZOS 0 . 1 0 0 . 0 5 0 .01

K?O 0 .10 0 .05 0 .01 0 .10

Mg 0 . 0 5 0 .01

Col lege Farm

15 21 35

243 336 567

48 67

113 2 3 5

Forrgate Farm

569 788

1331 140 194 328

28 38 65 28 39 65

A. Pinter Farm

11 16 27

321 445 752 193 266 450

35 49 83

5 Formula used to obtain optimum sampling sizes is:

E. Pinfer Farm

11 16 27 6 8

14 41 57 96 89

123 208

where, n = slze of a sample; t = t-values at the 0.1, 0.05, and 0.01

levels, respectively: C = coefficient of variation (C = i) ; and p = a precision level, e. g., 5yc of the mean

fertilizer \vas used in the previous years. the eight to 16 composile samples may be adequate for PzOb and N tests. On the other hand, in a farmed land. such as the Forsgate farm. where fertilizers were heavily banded in rows for years, the variation of available nutrients Ivould be expected to be high, and therefore the size of samples should be appropriately increased.

In vieiv of progress made in chemical techniques for soil resting and in- creasingly extensive and intensive use of soil amendments on agricultural lands, further study on proper methods of soil sampling \i.ould appear to be urgently needed.

Acknowledgment

The author wishes to thank A. Owens and J. Trio, the Kew Jersey Agricultural Experimental Station, New Brunswick, N. J.? for most of the chemical analyses. Grateful acknowledgment is also ex- pressed to Martin H. Yeh. the University of Manitoba, LVinnipeg, Manitoba, Canada, for his suggestions on statistical analyses, and William J. Zwerman, LTni- versity of Alberta, Edmonton, ,41ta., Canada, for his final review of the paper.

Literature Cited (1) Cline, M. G., Soil Sci. 58,275 (1944). (2) Graham, E. R.: Llniu. Mo. Agr.

Eupt. Sta. Bull. 734 (1959).

(3) Hanna, W. J.. Purvis, E. R.. S. J . Agr. Expt. Sta. Bull. 780 (1955).

(4) Leo, M. LV. M., Ph. D. thesis, Rutgers Vniv.. New Brunswick, N. J. (1 960).

(5) McKenzie. R. M., Australian J . Aer. ~, - Res. 6, 699 (1955).

AGR. FOOD CHEM. 9, 15 (1961). (6) Purvis, E. R.. Leo, M. W. M., J.

(7) Reed, J. F.. Rigney, J. A., J . Am. SOC. Agron. 39, 26 (1947).

(8) Snedecor, G. W. “Statistical Meth- ods,” 4th ed., Iowa State College Press, Ames, Iowa. 1960.

(9) Tisdale, S. L., Nelson, W. L., “Soil Fertility and Fertilizers,” Mac- millan, New York, 1956.

Received f o r review February 20, 7 962. Accepted n’ouember 28, 1962.

434 A G R I C U L T U R A L A N D F O O D C H E M I S T R Y