| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Dep. of Natural Resource Sci. and Landscape Architecture, Univ. of Maryland, College Park, MD 20742 USA
b Crop Production Dep., Bunda College of Agriculture, Univ. of Malawi, P.O. Box 219, Lilongwe, Malawi
rw17{at}umail.umd.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMaterials and methodsResults and discussionConclusionsREFERENCES |
|---|
. Maize yields showed a N x S interaction such that there were no responses to S unless N was also supplied. With 80 kg/ha N, a significant response to S was shown by maize in all but one of 20 experiments. A response curve for S showed that 5 to 10 kg/ha of S was optimal, with mean yield responses ranging from 90 to 142 kg grain/kg S.
Abbreviations: DRIS, Diagnosis and Recommendation Integrated System • TSP, triple superphosphate
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and methodsResults and discussionConclusionsREFERENCES |
|---|
Some early reports recognized the problem. Jones (1977) reported symptoms of S deficiency on maize and groundnut (Arachis hypogaea L.) in parts of Malawi. A few experiments in the early literature reported significant responses by maize to S, generally in the range of 12 to 20% increases in yield (e.g., Allen [1976] in western Kenya, Grant and Rowell [1976] in Zimbabwe, and Kang and Osiname [1976] in Nigeria). However, comparatively little research was conducted to investigate the S problem. For example, a review of S fertility in Tanzania found only two experiments designed to test specifically for S response, and only one report of actual crop responses to S (Shenkalwa, 1986).
There are several probable reasons why S has not received adequate attention, especially in Africa. First, low-yield shifting–subsistence agriculture has been based on exploiting the natural reserves of S, mainly mineral S released as soil organic matter decomposes. Second, from the 1950s through the 1980s, adequate S was commonly supplied to cropland from animal manures, or, where available, from then-popular low-analysis fertilizers such as ammonium sulfate and single superphosphate. These fertilizers were applied for their N and P contents, but actually supplied more S than either N or P. In addition, certain technical issues have mitigated against research on S fertility. One was the lack of simple, inexpensive, reliable methods for analysis of plant and soil S and SO4 (du Toit and du Preez, 1995). Another is the fact that responses to S are easily overlooked where a basal dressing of P is applied as triple superphosphate (TSP), a high-analysis fertilizer commonly assumed by researchers to be free of S (Shenkalwa, 1986). However, as reported here, TSP may contain agronomically significant quantities of S.
The situation has changed in recent years. First, as organic matter has become depleted by continuous cropping, and as higher-yielding maize hybrids are finding wider use, the supply of S from organic matter mineralization has become inadequate in many instances. Second, during the past decade there has been a rapid shift in African agriculture from the use of low-analysis fertilizers such as ammonium sulfate and single superphosphate to high-analysis fertilizers such as urea or diammonium phosphate, which contain little if any S. Where these high-analysis fertilizers are used with high-yielding maize varieties, failure to supplement with S can be expected to rapidly deplete available S supplies in the soil.
Therefore, it is not surprising that deficiency of S is becoming increasingly common in Africa on land continuously cropped for food production. For example, MacColl (1984) observed severe S deficiency symptoms on maize at the Bunda College of Agriculture Crop Production Research Farm near Lilongwe, Malawi, on land that the college had acquired from smallholder farmers. On these soils, maize gave no response to N fertilizers unless S was also added. Weil and Mughogho (1993a) reported that recycling of deep soil S was associated with higher yields of maize grown under acacia trees [Faidherbia albida (Delile) A. Chev] in Malawi.
The potential for increasingly widespread S limitations to crop production in Africa in general, and Malawi in particular, calls for more research to better define the extent and degree of S limitations where continuous cultivation of crops is being practiced. Our aims were to survey the S status of smallholder maize crops in Malawi and to determine the response of maize to applications of S at sites in the Balaka, Lilongwe, Mzuzu, and Salima areas.
|
Materials and methods |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and methodsResults and discussionConclusionsREFERENCES |
|---|
|
For each field, the farmer was interviewed to determine management practices used. All sampled fields grew local, open-pollinated maize or one of two maize hybrids. All fields were under hand-hoe ridge cultivation. The vast majority of fields had not been fertilized in the previous 5 yr. Only one field had received regular cattle manure dressings, and only 15 fields had received any fertilizer in the sampling year. Of the latter, two had received 23–21–0–4 N–P–K–S, while the other 13 had received urea, diammonium phosphate, and/or calcium ammonium nitrate. In each field to be sampled, a homogeneous-appearing area of 100 m2 was marked for repeated sampling.
From each field, leaves from 15 to 20 maize plants were sampled on two dates: the most recently matured leaf on the early date and the earleaf (leaf blade opposite and below the ear) on the later date. No whole plants, but only leaves, were sampled, because we found that farmers objected to the removal of whole plants, and the ensuing negotiations would have greatly slowed the progress of the survey.
All sampled leaves were rinsed in distilled water and placed in cloth bags for transport to the lab, where they were dried in a forced-air oven at 70°C and ground in a stainless steel mill to <1 mm. A subsample of ground tissue was dry-ashed in a muffle furnace at 450°C for analysis of S content by turbidimetry (Chapman and Pratt, 1961). Total N in maize tissue was determined with a LECO high-temperature combustion CHN Determinator (Leco, St. Joseph, MI). The remaining nutrient elements (including P, B, and all the metallic cations) were determined by emission spectrophotometry after digestion and dissolution in acid at the Plant and Soil Analysis Lab of the University of Georgia (Plank, 1992).
1990 Experiments
In the 1989 to 1990 growing season, we established experiments (referred to as the 1990 experiments) using a randomized complete block design with two replications on each of five farmer maize fields in each of the four study areas. Eight treatments consisted of factorial combinations of two levels of N, S, and P (with or without), with rates of application of 80, 20, and 20 kg/ha for N, S, and P (supplied as urea, gypsum, and TSP). Depending on deficiencies suspected from the results of the preliminary survey, two additional nutrients (among Mn, Cu, Zn, and Mg at 10, 6, 8, and 20 kg/ha, respectively) were applied factorially in combination with the four N x S treatments to give an additional eight combinations, for a total of 16 treatments. All fertilizers were applied as a point-placed sidedressing 5 cm deep and 10 cm away from each hill of three maize plants. Each treatment plot was 9.7 m2. Hills were spaced 90 by 90 cm apart on ridges to give 37000 plants/ha. Since no significant effects or interactions of Mn, Cu, Mg, or Zn were observed (with the exception of one slightly negative response to Zn at a single site), those plots effectively served as additional replication of their underlying N x S treatments in the analysis of variance.
The experiments were laid out in November 1989, and treatments were applied in December 1989 to January 1990. Plot layout, fertilizer application, soil and plant sampling, and harvest were conducted by the authors. All other aspects of crop management, including sowing and weeding, were the responsibility of the farmer cooperator. Harvest of the experiments and measurements of grain yield were completed by the end of June 1990.
Maize tissue was sampled from the experimental plots and analyzed as described for the survey, except that for the experiments, whole plants were harvested at the V6 to V12 stage. Yield of stover and grain (expressed as grain with 130 g/kg moisture) was determined by weighing the entire production of each plot in the field and bringing a five-plant subsample from each plot back to the lab to determine field moisture content.
1991 Experiments
In 1990 to 1991, we established new experiments (referred to as the 1991 experiments) in collaboration with same 20 farmers as in the previous year, but with a somewhat altered treatment structure and with three replications on each farm. The new experiments included eight treatments that were factorial combinations of two levels of N, S, and P (with or without), but this time the applied levels were 80, 20, and 50 kg/ha for N, S, and P, respectively. In addition, plots receiving N and P plus either 5, 10, or 15 kg S were included so that a S response curve could be generated with levels ranging from 0 to 20 kg/ha S.
Planting was done by collaborating farmers in December 1990, and fertilizers were applied by the authors after stand establishment was assured in January 1991. Dry matter weights and tissue samples were obtained for whole maize plants at the V12 stage about one month after fertilizer application and earleaf samples were taken at tasseling. Grain and stover yields were determined and tissue nutrient content analyzed as described for the 1990 experiments.
Calculation of Nutrient Indices
The Diagnosis and Recommendation Integrated System is a method of interpreting plant tissue analyses that was designed to be especially useful when several nutrient deficiencies are suspected (Sumner, 1979). The DRIS ranks the importance of the various nutrients in limiting plant yield and estimates the degree to which each of the limiting nutrients is deficient. The DRIS uses a system of nutrient balance indices rather than critical values or ranges for each nutrient. We calculated DRIS indices for both young maize plants and for earleaf samples using the appropriate DRIS norms and computer programs of Walworth and Sumner (1987). The DRIS norms in these programs were determined from the means and standard deviations of the ratios of various nutrients (e.g., N:K, N:P, P:K, etc.) in a large worldwide population of high-yielding maize (Walworth et al., 1986). DRIS norms are used to calculate the DRIS index for a given sample. We calculated DRIS indices using ratios for all 11 nutrients analyzed (N, P, K, S, Ca, Mg, Fe, Mn, Zn, Cu, and B).
The general expression used to calculate DRIS indices was given by Walworth and Sumner (1987) as
 | (1) |
a/b,
 | (2) |
 | (3) |
The SYSTAT program (Wilkinson, 1990) was used to run analyses of variance on each experiment individually, and on the pooled experiments from each sampling area (with treatment, replications within farms, and sampling area as factors). SYSTAT was also used to run linear regressions and correlations between variables, and to estimate linear-plateau response equation parameters by nonlinear regression. An 
level <0.05 was considered to indicate statistical significance unless otherwise stated.
|
Results and discussion |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and methodsResults and discussionConclusionsREFERENCES |
|---|
|
There was only a weak overall correlation between S concentration of young plant leaf tissue and that of plants in the tassel stage (Table 2). The correlation ranged from
in the Salima area to a nonsignificant
in the Lilongwe area. The correlations were highest in the two areas with sandy, less weathered soil profiles and lowest in the Lilongwe area, which has well-developed soil profiles that include red subsoil accumulations of weathered, low-activity clays. We speculate that in the latter area, plants deficient in S early in the season may have accessed sulfates adsorbed on anion exchange sites deep in the profile later in the season, while in the two lake plain areas (Balaka and Salima), the plants were more dependent on mineralization of organic S for their S supply throughout the season.
The DRIS index data was characterized by much greater variation and more extreme values (ranging from 138 to -40 for earleaf indices) than were evident in the tissue concentration data. One reason for this was that, although we observed no plant symptoms of either B or Cu deficiency, about 10% of the samples had very low values for tissue B and Cu (<0.5 mg/kg B and <2 mg/kg Cu, data not shown). These extremely low B and Cu values resulted in very high DRIS indices for most other elements, including the S DRIS index, which exceeded 100 in these cases. This apparent distortion is related to a criticism leveled at the DRIS by Baldock and Schulte (1996), in which they pointed out that because the DRIS indices are not independent, the level of one nutrient can have marked effects on the indices of other nutrients.
The S DRIS indices for young plant leaves were generally only slightly positive (mean values of 4.6 to 8.4), except in Salima, where the mean S DRIS index was -5.7, significantly lower than in the other three areas. The S DRIS indices for earleaves at tasseling were mainly positive in the lake plain areas (Balaka and Salima), where the mean S DRIS indices for earleaves (20.2 and 27.8, respectively) were significantly higher than in the Lilongwe and Mzuzu areas (-3.4 and 5.7, respectively), and much higher than the corresponding young plant S DRIS indices. The S DRIS indices in the Lilongwe and Mzuzu areas did not appear to change much between the two stages of growth. There was little relationship between the young plant and earleaf DRIS indices (Table 2), except in Salima, where there was a weak but significant correlation
) between the two.
Since a major physiological function of S is the formation of proteins that contain both N and S, the ratio of N:S in plant tissue is commonly used as a measure of S nutritional status of plants, with an N:S ratio >12 often cited as indicative of S deficiency in maize. The N:S ratio in the surveyed maize plants varied from 5 to 26 (data not shown). The vast majority of young maize plants in all four areas had N:S ratios greater than the critical value, but plants in Mzuzu had a mean N:S ratio significantly higher than the other areas (Table 2). The overall mean N:S ratio was higher in the young plant leaves (14.7) than in the earleaves (11.5). The lake plain areas had significantly lower N:S ratios in the earleaves than did Lilongwe and Mzuzu, where in about half the cases the N:S ratio exceeded 12 in the earleaves. In none of the areas was the N:S ratio in young plant leaves significantly correlated with that in the earleaves (Table 2). In general, the S status of young maize plants in a field (whether indicated by tissue S, S DRIS index, or N:S ratio) was a very poor predictor of the S status of maize later in the season (tassel stage) in the same field.
1990 Experiments
Of the 20 experiments (40 total replications) established in 1990, yield data were successfully obtained from only 14.5 (29 replications), because of losses prior to grain harvest due to termites, drought, wayward cattle, thieves, or premature harvest by the farmer cooperator. The data within each of the four sampling areas (five to nine replications) was pooled for analysis of variance. The data shown in Fig. 1
are averages across treatments with and without P application (there were no interactions with P, and P had relatively little effect on yields, giving a significant increase in only one of the experiments). Yields of unfertilized maize (mainly open-pollinated land races) were highly variable, ranging from a low of 134 kg/ha in one plot in Balaka to 5809 kg/ha in the highest plot in Mzuzu (data not shown). Mean yields of unfertilized maize for the four areas (Fig. 1) ranged from 691 kg/ha in Salima to 2158 in Balaka, very close to the range reported by other surveys of on-farm yields in Malawi (Malawi Government, 1977, p. 19–46; Weil, 1985). The response to N, with or without S, was significant in all four areas (and on all but one individual farm), the average response to 80 kg N/ha ranging from 1300 kg/ha of maize grain in Salima to 2150 kg/ha in Mzuzu (Fig. 1). The N responses were thus in the range of 16 to 27 kg grain/kg N applied, very similar to the 11 to 23 kg grain/kg N reported by others in Malawi (Saka et al., 1992) for the first 60 kg N applied. The similarity of the unfertilized maize yields and N responses in our study to those in other reports suggests that our set of on-farm experiments succeeded in sampling typical Malawi smallholder conditions.
|
1991 Experiments
A second series of 20 on-farm field experiments was conducted in 1991. Yields with no N applied averaged 1121 kg grain/ha, similar to the results of the 1990 experiments. On all farms but one, maize responded significantly to the application of 80 kg N/ha, whether or not other nutrients were also applied. However, there was a significant N x S interaction in all four areas such that the effectiveness of N fertilizer was enhanced when accompanied by application of S fertilizer (Fig. 2)
. Responses to 80 kg N/ha ranged from 1530 kg/ha maize in Mzuzu to 2900 kg/ha maize in Salima with no S applied, but from 2600 to 3960 kg/ha maize when 20 kg/ha S was applied.
|
Figure 3 shows the mean effects of S and P application on maize yields for each area when 80 kg N/ha were applied. Just as S deficiency symptoms were observed more often than P deficiency symptoms in the 1990 survey, the maize response due to S in 1991 was more frequently significant and generally larger than that due to P. In the two lake plain areas (Balaka and Salima), P did not give any increase in maize yield, either where S and N were adequate (Fig. 3) or where P was applied alone (data not shown). In the upland areas of highly weathered soils (Lilongwe and Mzuzu), application of 50 kg/ha P along with 80 kg/ha N and 20 kg/ha S did give significant responses, and of somewhat greater magnitude than those due to S. Results for P application without S are not shown in Fig. 3 because of the potentially confounding S content of the P source used (discussed below).
|
|
The true response to S in our study may have differed somewhat from than the 450 to 850 kg maize/ha reported above, since all treatments making up the S response curve received 50 kg P/ha as triple superphosphate. After the experiment was completed, we analyzed the TSP used (which was of Norwegian origin), as well as three other TSP samples from the USA and Kenya, and found that both materials contained 25 to 31 g/kg total S. Therefore, the basal dressings of TSP probably alleviated some S deficiency in even the zero-S plots, since the TSP used to supply 50 kg P would have also supplied approximately 6 kg S. Thus, the yield plateau may be more accurately characterized as occurring at an S application level of 10 to 12 kg/ha. Since the true yield response to S would also be somewhat greater, we cannot calculate exactly what the response per kilogram S applied would have been had we used a P source that did not contain any S. The fact that many workers in soil fertility have not recognized the 20 to 30 g/kg S content of TSP suggests that crop responses ascribed to P application in many reports may actually have been due partly to S.
Based on the present results and those of other studies, it would seem that 5 to 10 kg S/ha is generally sufficient for maximum response by maize at yield levels <6 mg/ha. The size of the responses observed in our study would suggest that supplementing N fertilizers with S, perhaps by developing local sources of gypsum or elemental S, has the potential to be a very economical way of boosting maize production among those smallholder farmers who have access to moderate levels of N fertilizers. Given the low S-supplying capacity of the soils in this study, and the limited reports cited from the literature, we also speculate that S application should be experimented with for leguminous crops. However, in the absence of adequate levels of N (from either legumes, manure, or fertilizer), investments in S application would probably not be profitable.
In the low-fertilizer-input peasant agriculture typical of much of Africa, crop surveys using tissue analysis may be useful in delineating areas of particular nutrient deficiencies (Weil et al., 1991). One of the reasons for conducting the 1990 survey of maize in farmer fields was to test the hypothesis that the results of a crop tissue survey could be useful in predicting whether and to what degree maize would be likely to respond to S in a given area. Figure 5
shows the relationship between several measures of maize tissue S status on N-fertilized plots and the subsequent yield responses measured in the on-farm experiments. The earleaf S content by itself was not significantly related to yield responses to S. However, the S DRIS index did show a significant relationship
with yield response. As would be expected, the relationship between earleaf S DRIS index and yield response was negative, a negative DRIS index being indicative of a deficiency, which would likely result in a positive yield response. Interestingly, the regression line passes almost exactly through the origin, as would be predicted by the theory that a plant with an S index of 0 would be in balance and not show a response to added S.
|
than the S DRIS index. Yield responded positively to S application where the N:S ratio in the earleaf exceeded 10 (only slightly lower than the critical value of
). In two experiments where earleaf N:S was very low (<8), application of S resulted in a yield depression. The N:S ratio in the young maize plants, however, was not significantly related to maize grain yield responses (Fig. 5). Data available on soil S in our study (not shown) was too limited to test for useful correlations, partly because of difficulties in getting repeatable laboratory results for soil SO4–S in iron-rich, highly weathered soils (du Toit and du Preez, 1995), and partly because soil sampling to a greater depth than is practical in a rapid survey would be required to account for SO4–S potentially available to maize. |
Conclusions |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and methodsResults and discussionConclusionsREFERENCES |
|---|
Deficiency of S in farmer maize crops appears to be widespread in four areas of Malawi with greatly differing types of soils. A significant response to S was shown by maize on all but one of 20 farmer fields in which replicated experiments were conducted. Although some unintended S applied in TSP used for P application precluded exact calculations, our results suggest that the response to applied S in Malawi is likely to be between 80 and 160 kg of grain per kilogram of S applied, a rate of return that should make the use of S economically attractive in most instances. Optimal response to S was obtained with the application 5 to 6 kg S/ha in our experiments. Also, significant S x P interactions found in several of the 1991 experiments indicated that S response was reduced where TSP was applied and that P response was reduced where S was applied. These results, and our analysis of 25 to 31 g/kg S in TSP samples, suggest that care should be taken in future S and P response experiments to avoid S-containing TSP as a P source. Some responses to TSP in the literature should probably be reinterpreted, in that they may actually be partly due to the S rather than the P in this fertilizer material.
|
ACKNOWLEDGMENTS |
|---|
Received for publication July 20, 1999.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and methodsResults and discussionConclusionsREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  L. Chen, D. Kost, and W. A. Dick Flue Gas Desulfurization Products as Sulfur Sources for Corn Soil Sci. Soc. Am. J., August 20, 2008; 72(5): 1464 - 1470. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 V. D. Zheljazkov, T. Astatkie, C. D. Caldwell, J. MacLeod, and M. Grimmett Compost, Manure, and Gypsum Application to Timothy/Red Clover Forage J. Environ. Qual., October 27, 2006; 35(6): 2410 - 2418. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Crop Science | Vadose Zone Journal | |||
| Journal of Natural Resources and Life Sciences Education |
Soil Science Society of America Journal | ||||
| Journal of Plant Registrations | Journal of Environmental Quality |
The Plant Genome | |||
, where SUFF is the value of x at which y reaches the plateau, and a, b, and c are regression-estimated parameters. Corrected R2 values were 0.90, 0.99, 0.92, and 0.81 for Salima (diamonds), Lilongwe (squares), Balaka (circles), and Mzuzu (triangles)
.58**. *, ** Significant at the 0.05 and 0.01 probability levels, respectively