| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a CREA Tandil, Bolívar 710, Tandil 7000, Argentina
b Universidad de Mar del Plata-INTA Balcarce, CC 276, Balcarce 7620, Argentina
c CSIRO, PMB 2, Glen Osmond, SA 5064, Australia
* Corresponding author (pcalvino{at}infovia.com.ar)
Received for publication October 16, 2001.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Abbreviations: ETmax, maximum evapotranspiration • PAW, plant available soil water • W, water availability • W1, water use by the crop from 20 d before to 20 d after flowering • W2, rainfall from 20 d before to 20 d after flowering plus available soil water at the beginning of this period • W3, rainfall from 20 d before to 20 d after flowering • W4, rainfall from 30 d before to 20 d after flowering • 
Y, difference between actual yield and calculated yield
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
One of the methods to analyze variation in yield consists of simple, agronomically meaningful models based on a few key environmental variables. Examples of this approach include research on the association of wheat (Triticum aestivum L.) yield with radiation and temperature in Argentina (Magrin et al., 1993) or with rainfall in Australia (French and Schultz, 1984a, 1984b) and of maize yield with rainfall, temperature, and stored soil moisture in the USA (Leeper et al., 1974). Other tools involve multivariate analysis and crop simulation models. However, standard statistics may be limited in biological meaning while crop simulation models have their own problems (Monteith, 1996; Passioura, 1996; Sinclair and Seligman, 1996; Sadras and Trápani, 1999).
In Argentina, the traditional Corn Belt is located between 32° S lat and 35° S lat. In the last two decades, improved management practices and better-adapted, shorter-season hybrids contributed to successful inclusion of maize in cropping systems down to 37° S lat, the region that is the focus of our study. In this region, actual yield averages 7 t ha-1; attainable yield, i.e., maximum yield in growers' fields, is around 9 t ha-1, and yield potential, measured in experimental plots, is 15 t ha-1 (Andrade, 1995; Cirilo and Andrade, 1994). Opportunities therefore exist to adjust management practices to increase yield, provided we are able to identify the factors underlying the gap between actual, attainable, and potential yield.
Maize is most sensitive to environmental stresses in the period bracketing flowering (Robins and Domingo, 1953; Shaw, 1988). Runge (1968) and Thompson (1975) evaluated long-term experiments under constant or average management using regression techniques and concluded that yield in the U.S. Corn Belt was highly correlated with W at flowering. Of the likely environmental stresses in our region, water deficit is also a prime candidate owing to the combination of shallow soils and erratic summer rainfall (Sadras and Calviño, 2001).
Management practices have changed substantially during the past two decades in our region. These changes include a shift from conventional tillage to no-tillage, increase in the use of N and P fertilizers, improved weed control, longer fallow duration, and better adjustment of plant density and sowing date (Calviño and Grosse, 1998).
Our objective was to investigate the influence of W (as related to rainfall and soil depth) and technological changes throughout the last 15 yr on the yield of dryland maize crops of the Argentine Pampas using data from large, grower-managed fields. Involvement of growers in one or more stages of the research process can partially compensate for the lack of formal channels of communication between the production system and research organizations that is common in less developed countries.
We followed the three-steps approach proposed by Calviño and Sadras (1999). First, we quantified the relationship between yield and W during the critical period for kernel set (Andrade and Sadras, 2000). Baselines, comparable to those developed by French and Schultz (1984a)(1984b) and Calviño and Sadras (1999)( 2002) were thus developed. Second, the baseline was tested using independent data sets. Third, using the baseline function as a benchmark, the effects on yield of water available to the plant as determined by soil depth and of crop management practices were evaluated.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Sites
Crops were grown in the region of Tandil (37° S, 59° W) and Marcos Juarez (32° S, 62° W). At Tandil, average annual rain is 940 mm, of which 69% falls between September and March. Soils include deep Typic Argiudols and shallow Petrocalcic Paleudolls (USDA taxonomy). Organic matter averages 6.2% and available water-holding capacity 1.5 to 1.6 mm cm-1 soil (Travasso and Suero, 1994). Available P just before sowing and fertilization was between 3 and 10 mg kg-1 (Bray and Kurtz, 1945). Shallow soils dominate areas with slopes from 2 to 5%, whereas deep soils are typical of flat areas. Experiment 3 compared two soil types: deep (depth 
1.2 m), belonging to class I or IIe (Klingebiel and Montgomery, 1961), and shallow (0.75 m depth 0.5 m) of class IIIes or IIIs. Only deep soils were included in the other experiments. Recommended sowing date for maize is mid-October to early November. Adapted cultivars (FAO 550–600) flower in early January (Andrade, 1995).
At Marcos Juarez, average annual rainfall is 900 mm, of which 80% falls between September and March. Soils are deep Typic Argiudols (USDA taxonomy) belonging to class I or IIc (Klingebiel and Montgomery, 1961). Organic matter averages 2.6% and available water-holding capacity 1.5 mm cm-1 soil (Travasso and Suero, 1994). Available P at sowing was between 12 and 20 mg kg-1 (Bray and Kurtz, 1945). Recommended sowing date is between late August to late October. The most common cultivars flower between 25 November and 25 December (FAO 550–600 and 700–750). Solar radiation and temperature data at these two locations were obtained from a nearby meteorological station. Photothermal quotient [Rad/(t - tb)] was calculated for the period flowering ± 20 d for each location (Rad = solar radiation, t = average temperature, and tb = base temperature = 10°C).
Experiment 1: Baseline Function
Yield and rainfall data were collected from 77 (1996), 67 (1997), and 72 (1998) crops at Tandil. Management practices for all crops included direct drill; weed-free fallow period of at least 90 d, which ensured a soil water content at sowing close to maximum; 0.7 m between rows; plant population density between 7.2 and 7.7 plants m-2; adequate control of weeds and insects; and sowing dates between 10 and 25 October. Hybrids used included ‘Dekalb 664’, ‘Dekalb 669’, ‘SPS 2601’, ‘Dekalb 639’, and ‘Nidera 788’. Crops were fertilized with 17 to 25 kg P ha-1 and 40 to 60 kg N ha-1 according to available diagnosis methods (García et al., 1997). On average, paddocks were cropped for 6 yr since the last pasture.
The combination of three growing seasons and the large number of farms generated a considerable range in the data for yield and rainfall. The analyzed period was short enough to meet the criterion of unchanged technology. To further reduce the influence of management as a source of variation, we used yield data from only the 15% highest-yielding fields out of the total fields grown each season and at least the highest-yielding field from each commercial farm (Calviño and Sadras, 1999, 2002).
The relationship between yield (Y) and water use by the crop or available to the crop at flowering (W) was described with the model:
 | [1] |
 | [2] |
Four estimates of W (mm) were compared:
Maximal plant available soil water in first 1.2-m depth for these soils is 180 mm (Travasso and Suero, 1994). Based on this value, the cropping system (long and weed-free fallow and a substantial amount of crop residues on soil surface), and the high probability of rainfall before sowing (Fig. 1) , the available soil water at sowing was conservatively assumed to be 80% of field capacity. Available water in the soil 20 d before flowering and water use during the period bracketing flowering (flowering ± 20 d) was calculated with a hydrological soil balance, as described in Sadras and Calviño (2001). Maximum evapotranspiration (ETmax) was calculated as the product between reference evapotranspiration (Penman, 1948) and locally tested, phenology-dependent crop coefficients (Della Maggiora et al., 2000). Published upper and lower limits were used for the calculation of relative plant available soil water (PAW) (fraction) (Travasso and Suero, 1994). Actual evapotranspiration was assumed to be equal to ETmax when PAW > 0.5 and to decline linearly with PAW between 0.5 and 0 (Sadras and Milroy, 1996). Daily rain was measured at each farm. Long-term data (30 yr) from a meteorological station at Tandil and the hydrological soil balance described above were used to calculate the cumulative probability of (i) available water in the soil at 20 and 30 d before flowering and (ii) W1 considering a maize crop sown in mid-October in deep and shallow soils. Rainfall data were also used to calculate the cumulative probability of rainfall from 1 July to 15 October and from 30 d before to 20 d after flowering.
|
Experiment 3: Soil Depth
Grain yield and rainfall were measured for 77 crops grown in shallow soils (soil depth 0.5–0.7 m) during three growing seasons from 1996 to 1998 at Tandil. Management practices were similar to those described for Exp. 1.
Experiments 4–6: Crop Management
Experiment 4: Yield and rainfall data were collected from 40 (1987), 42 (1988), and 41 (1989) crops at Tandil. Management practices were quite different from those described for Exp. 1, including: (i) conventional tillage; (ii) no N nor P fertilizer; (iii) short fallow period with poor weed control, which did not ensure a close-to-maximum soil water content at sowing (Zimdahl, 1980); (iv) older hybrids with less yield potential and stability; and (v) plant population density from 5.5 to 6 plants m-2, as recommended at this time (Darwich, 1987).
Experiment 5: Yield and rainfall data were collected from 45 (1994) and 52 (1995) crops at Tandil. Management practices were those described for Exp. 1, except for: (i) conventional tillage, (ii) no N fertilizer, and (iii) plant population density from 6 to 6.3 plants m-2. On average, paddocks had three cropping seasons after pasture.
Experiment 6: Yield and rainfall data were collected from 56 (1999) and 61 (2000) crops grown at Tandil in 1999 and 2000. Management practices were those described for Exp. 1, except for a greater rate of N fertilizer, between 70 and 100 kg ha-1, which is currently recommended in no-till systems (Sainz Rozas et al., 2000).
The objective of these trials was not to assess the effect specific management practices on yield—which would have required a factorial design—but rather to compare different management systems using the baseline function between yield and W as a benchmark.
Criteria for crop selection in Exp. 3–6 were the same as in Exp. 1: The 15% highest-yielding fields out of the total fields grown each season and at least the highest-yielding field from each commercial farm were considered. To assess the interaction between rainfall and soil or management factors, we followed the approach of Calviño and Sadras (1999)(2002). First, we calculated Y, the difference between actual yield and yield calculated using the baseline function (Eq. [1]) at a fixed level of W4; Y is therefore a measure of yield differences attributable to soil or management factors. Second, we tested the association between Y and W using regression analysis. A significant association between these variables implies a significant interaction between soil or management factors and rainfall, as discussed in Calviño and Sadras (1999)( 2002). In all experiments, curves were fitted with Jandel Scientific (1991).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
|
Y) varied between 1.6 and more than 5 t ha-1 and were not related to W4.
|
Y was unrelated to W4.
|
Y) varied between 0.3 and 1.6 t ha-1 and were unrelated to W4 (p > 0.05).
Experiment 6: Yield varied between 8 and 11.5 t ha-1, with W4 accounting for 55% of the variation (Fig. 5C). Grain yields at Exp. 6 were mostly greater (up to 2.4 t ha-1) than those predicted by the baseline function derived from Exp. 1. Again, Y was not related to W4.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Our rainfall-based model (W4 as independent variable) did not account for a number of factors, including (i) W during early vegetative growth stages, (ii) the contribution of stored soil water, (iii) runoff associated with both reduced storage capacity and the prevalent position of shallow soils in steep sections of the landscape and run-on to foot plain, (iv) within-month rainfall distribution, and (v) variation in actual evapotranspiration. Despite these limitations, a rainfall-based model worked reasonably well because it considered both the critical period for grain number determination (Shaw, 1988; Kiniry and Ritchie, 1985; Andrade et al., 1999) and the most relevant environmental factor for yield, i.e., W (Runge, 1968; Leeper et al., 1974; Thompson, 1975). Moreover, a model considering water use by the crop (W1) did not improve yield prediction.
Advantages and drawbacks of simulation models as tools for research of complex interactions have been widely discussed (Monteith, 1996; Passioura, 1996; Sinclair and Seligman, 1996; Sadras and Trápani, 1999; Calviño and Sadras, 1999). Following Passioura (1996), we developed a simple analytical method based on robust empirical relationships between the main variables. Crop simulation models like CERES-Maize are not fully used because of the amount and complexity of variables needed to run them. Moreover, in some cases, simple models explain grain yield variability better than complex simulation models (Otegui et al., 1996). However, being developed for a specific cropping system, our approach lacks the flexibility of more complex crop simulation models (Calviño and Sadras, 1999, 2002).
Yield Responses to Rainfall, Soil Depth, and Crop Management
The strong association between yield and water available during the period bracketing flowering found in commercial maize crops at two locations highlights the driving effect of rainfall in these farming systems (Magrin et al., 1998). Similar results were reported by Leeper et al. (1974) for the U.S. Corn Belt. Differences in the radiation–temperature quotient during the period bracketing flowering among locations and years were small (<6%), suggesting that these environmental variables did not affect grain number. Differences in attainable yield between locations were mainly due to technological inputs. Yield increased in response to increased W, but the response was not linear. The high yield response at low W agrees with that found by Muchow (1989) and NeSmith and Ritchie (1992). With milder water deficits, the response of grain yield to W was lower and comparable to that reported in experimental plots in the same area (Otegui et al., 1995; Andrade et al., 1996) and in other areas (Muchow, 1989).
In the cropping systems of the southeast Pampas, the probability of having 50% available water content in the soil (90 mm) at crop emergence (not shown) and at 30 d before flowering (Fig. 6) is higher than 90%. This is one of the reasons for (i) the good agreement between the information shown in this work and that presented in more controlled experiments in experimental plots in which water stress was imposed only at flowering and (ii) the high value of W4 as yield predictor (Fig. 2D). Most of the variation in maize grain yields was accounted for by W4. Based on the probability of rainfall during this critical period (Fig. 7)
and on the baseline function from Fig. 2D, it is concluded that the probability of obtaining maize yields 8000 kg ha-1 is 60%, considering well-managed crops in Tandil using the technology available during 1996–1998.
|
|
|
Grain yield in the area steadily increased during the past 15 yr. Predicted increases at W4 = 170 mm (50% cumulative probability; Fig. 7) were 2.3 t ha-1 from the late 1980s to the mid-1990s, 0.9 from the mid-1990s to 1996–1998, and 0.8 from 1996–1998 to 1999–2000 (Fig. 5). Within each step of technological improvement, increments were not related to W. Management techniques differed among these periods. In the late 1980s, crop management was mainly characterized by short cropping cycle after pasture, conventional tillage, no N nor P fertilizer, older hybrids that had less yield potential and stability, and low plant density (5.5–6 plants m-2). Management of 1980s would be the reason for the lower water use efficiencies, i.e., yield per unit amount of rainfall at the period bracketing flowering, typical of the late 1980s and early 1990s. Owing to the low P availability in these soils, lack of P fertilization would have imposed limitations to crop growth and yield (García et al., 1997). By the mid-1990s, improved crop management involved P fertilization, better and earlier weed control (Hall et al., 1992), and the use of hybrids with higher yield potential and stability. Improved crop nutrition and weed control, together with better cultivars, allowed higher yields at a fixed W4. Important management changes took place in 1996. In the 1996–1998 growing seasons, management changes included direct drill, N fertilization, and higher plant densities (7.5 plants m-2). Nitrogen fertilization was required because N mineralization rate is lower (at least during the first years after no-till implementation) and N losses are greater with direct drill than with conventional tillage (Meisinger et al., 1985; Rizzalli, 1998). Differences in grain yield between 1994–1995 and 1996–1998 are probably explained by a better crop water balance because direct planting (with residue cover) would reduce soil evaporation and improve water infiltration (Brandt, 1992; Blevins et al., 1971). In the 1999 growing season, the only change in relation to the previous years was an increase in N fertilization rate. This was due to an improvement in the N diagnosis methods for no-till practices (Calviño and Echeverría, 2000; Sainz Rozas et al., 2000). Thus, N fertilization rates were lower than optimum in 1996–1998. Improved water use efficiency during the last season was likely associated with better N nutrition.
In summary, a curvilinear relationship between yield and W at flowering was found. This relationship was quite different for deep vs. shallow soils. Finally, an impact of technology on maize yield was found using the above relationship to remove environmental effects.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  A. Pagani, H. E. Echeverria, F. H. Andrade, and H. R. Sainz Rozas Characterization of Corn Nitrogen Status with a Greenness Index under Different Availability of Sulfur Agron. J., March 4, 2009; 101(2): 315 - 322. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. C. Mason, D. Kathol, K. M. Eskridge, and T. D. Galusha Yield Increase Has Been More Rapid for Maize than for Grain Sorghum Crop Sci., July 1, 2008; 48(4): 1560 - 1568. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. R. Capristo, R. H. Rizzalli, and F. H. Andrade Ecophysiological Yield Components of Maize Hybrids with Contrasting Maturity Agron. J., June 26, 2007; 99(4): 1111 - 1118. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. L. Ma, K. D. Subedi, and T. Q. Zhang Pre-Sidedress Nitrate Test and Other Crop-Based Indicators for Fresh Market and Processing Sweet Corn Agron. J., January 1, 2007; 99(1): 174 - 183. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
X. Zhang, D. Pei, S. Chen, H. Sun, and Y. Yang Performance of Double-Cropped Winter Wheat-Summer Maize under Minimum Irrigation in the North China Plain Agron. J., October 31, 2006; 98(6): 1620 - 1626. [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 | |||
) expressed as a fraction of the maximum. Parameters of the fitted curves are (A) a = 11781 ± 439, W0 = 39.1 ± 74.5, and b = 43.4 ± 53.2 (P < 0.001) and (B) a = 0.91 ± 0.01, W0 = 46.6 ± 5.9, and b = 39.8 ± 5.2 (P < 0.001).
