African Crop Science Journal, Vol. 8. No. 3, pp. 273-282
EFFECT OF NITROGEN FERTILISER RATES AND PLANT DENSITY ON GRAIN YIELD OF MAIZE
(Received 13 December, 1999; accepted 14 August, 2000)
Code Number: CS00029
Maize (Zea mays L.) is a major cereal crop in the Southern Region of Ethiopia covering 28 % of the total area under cereal cultivation (CSA, 1997). Typical maize yields range from 1100 to 1600 kg ha-1, which is below the national mean yield of 1800 kg ha-1. The low yield is attributed to factors such as low soil fertility, inappropriate cultural practices and variety, erratic rainfall, weeds and other pests. About 57 and 67 % of farmers in the mid-altitude and lowland areas of Sidama district, respectively, consider low soil fertility as their major production problems. Soils are deficient mainly in nitrogen and only 19% of the total maize area is fertilised (CSA, 1998). To maintain and increase soil fertility, farmers mainly apply Diammonium phosphate (DAP), in about 59% of the total maize area (CSA, 1997). It is applied either at very low rate or not at all due to the high cost.
Various studies indicate that optimum fertiliser and plant populations provide better crop growth and yield. Response to fertiliser is, however, variable depending on amount and distribution of moisture, soil fertility and variety. Application of N fertiliser increases leaf area development which increases photosynthetic activity of the leaf (Sallah et al., 1998) and rooting depth which may provide available soil water for the plant when moisture stress is moderate i.e., crop water stress index <0.38 (Nielsen and Halvorson, 1991). Increased leaf area development and rooting depth will result in increased water-use efficiency. If there is severe moisture stress increasing N rate will increase the transpiration of the plant which may not be compensated by the increasing root volume. According to Onken and Wendt (1989), application of N under rainfed conditions increased water use efficiency (WUE) and grain yield of sorghum, but if there is limited supply of moisture N application does not increase WUE.
Hybrid maize favours high doses of fertiliser (Gardener et al., 1990; Killorn and Zourarakis, 1992 ). A study conducted by Killorn and Zourarakis (1992) at Iowa State, USA recorded least yield of hybrid maize from 0- N fertiliser and increased yield from the higher rate of N (196 kg ha-1). Other researchers (Nielsen and Halvorson, 1991; Sallah et al., 1998) reported similar results. On the other hand, response to plant population is dependent on location, variety and season. The open pollinated variety, A511, gave highest yield with a population of 53,000 and 74,000 plants ha-1 at Awassa and Upper Bir areas located in the Southern and Northern parts of Ethiopia, respectively. Nafziger (1994) in his study at Monmouth and Dekalb also in Ethiopia, found that maize hybrids were especially responsive to higher plant density under adequate rainfall but less responsive when there was insufficient precipitation. Currently the Bureau of Agriculture of the Southern Regional State of Ethiopia is introducing hybrid maize varieties to farmers to boost production. However, the optimum fertiliser requirement and plant population has not been determined. Hence, the objective of this study was to determine the effect of different levels of N fertiliser and plant population on grain yields of maize hybrid (BH-140) and recommend appropriate rates of both N and maize population.
MATERIALS AND METHODS
The experiment was conducted at Awasa Research Centre located in the Southern region of Ethiopia. The soil is silt loam with pH of 6.63, total N= 0.22%, P2O5 =35.84 ppm, CEC= 25.33 meq/100 g and OM= 4% and is classified as Eutric fluvisols. Maize hybrid, BH140, was used for this study which was conducted over four seasons (1995-1998). The experimental treatments were laid out in a randomised complete block design arranged as a 4 by 4 complete factorial with three replicates. The treatments were nitrogen (Urea as source of N) at 0, 46, 92 and 138 kg N ha-1; and four plant populations of 44x103, 53x103, 67x103 and 89x103 plants ha-1. The different plant populations were achieved by varying within plant spacing which ranged from 0.15 to 0.30 m., the difference from one spacing to the other being 0.05 m. Urea was applied as a single dose side dressed at the 8-10 leaf growth stage of maize. Two seeds were planted per hill and later thinned to one to produce the required population. The plot size was five rows, with 0.75 m between rows of 5.1 m. The planting dates were May 2, 6 and 12 in 1995, 1996 and 1997, respectively, and April 28 of 1998. The crop was hand weeded twice each season.
Data were taken on plant height, number of ears (cobs), number of seeds per cob, seed weight, total above ground biomass (1995,1996 and 1998), and grain yields. Grain yield was determined from the centre three rows of each plot and adjusted to 12.5 % moisture content. Data were analysed using analysis of variance with treatments (population and nitrogen combinations) as fixed and seasons as random effects. Agronomic components (harvest index, biological yield and productivity indexes , i.e., N fertiliser use efficiency)) were calculated only for 1995,1996 and 1998 since data on total above ground biomass was not taken in 1997 season. Combined analysis over years (seasons), and correlation and multiple regression analysis were carried out as outlined by Gomez and Gomez (1976). Fishers protected least significant (LSD) test was used to compare treatment means. Economic analysis was conducted using the approach of CIMMYT (1988). The regression model used was: Y=a+b1N+b2P, where: a = constant, N = nitrogen level (kg ha-1), P = population density (No. m-2), b1 = regression coefficient for nitrogen, b2 = regression coefficient for plant population.
Amount of rainfall in the growing seasons did not differ much from one year to the other (Fig. 1). However, seasonal distribution among seasons was variable. In 1995 and 1997 the rainfall pattern was inadequate particularly during the critical crop growth stages (tasseling, silking, and grain filling) while in 1996 and 1998 the distribution was favourable. Lower yields in 1995 and 1997 were attributable to seasonal fluctuation of rainfall. Occurrence of dry spells at critical crop growth stages was observed 61 and 58 % of the time in the growing seasons of 1995 and 1997, respectively. Distribution of precipitation in the whole of July and in the second halves of May, June and September and parts of August of 1995 were not favourable. Rainfall was also poorly distributed in the third and second decades of May and July, of the 1996 growing season. In the whole of June and the first and third decades of July as well as the first decade of September 1997 rainfall was un evenly distributed. It was also erratic in the third and first decades of June and September 1998, respectively.
The result of combined analysis indicated a significant yield variation among seasons (P<0.01), nitrogen (P<0.01) and interactions of season with nitrogen (P<0.01) and season with plant density (P<0.05)(Table 1). Seasonal distribution of precipitation significantly affected plant height, number of cobs, seeds/cob, kernel weight, total aboveground biomass and grain yields across N fertiliser rates and plant density. The significant interaction effect of season with both N and plant density meant that maize yields responded more to either N or plant density when distribution of precipitation was favourable (Fig 2 and Fig 3). Averaged over N and plant density, seasons with better distribution of rainfall (1996 and 1998) had higher mean yields of 7428 and 7813 kg ha-1, respectively, when compared with those (mean 5913 kg ha-1) obtained from seasons of erratic rainfall (1995 and 1997).
Agronomic traits (plant heights, number of cobs, biomass and grain yields) were significantly affected by variation in seasons and N levels (Table 1). When averaged over plant density and seasons of erratic precipitation, application of N gave 7,22,14,and 27% increase in plant height, number of cobs, seeds/cob, and biomass yield, respectively, as compared with zero N. N fertiliser resulted in higher yields for each increment of N added (Fig. 2). Mean grain yield at zero N averaged over seasons and plant density was 18,26 and 27% lower than grain yield at 46,92, and 138 kg N ha-1, respectively. Grain yield at 138 kg N ha-1 was 13 and 2% higher than the yield at 46 and 92 kg N ha-1, respectively. The significant interaction effect of nitrogen with season (P<0.01) indicated that response of plant height, number of cobs, seeds/cob, biomass and grain yields to N fertiliser was variable depending on the distribution of precipitation. During 1995 and 1997 crop growing seasons, the mean yield increments obtained over the check ( 0 N) due to addition of N were 1413 and 2537 kg ha-1, respectively. When averaged across plant density and seasons of erratic and even distribution of rainfall, the respective additional mean yields obtained were 1974 and 1185 kg ha-1 higher than their checks. The contribution of N to grain yield when distribution of rainfall was favourable was 53% while under erratic rainfall it was 81%.
Productivity index (yield per unit application of N) was more for the 46 kg N ha-1 (Table 2). The effect of N on total aboveground biomass was significant. Harvest index did not differ siginificantly among the three levels of N (46 to 138 kg N ha-1); however, there was an increase as the rate of N increased up to 92 kg N ha-1 (Table 3). Both the linear and quadratic responses to N were significant (p<0.01). When averaged over seasons and plant density, the highest N rate (138 kg ha-1) produced the highest grain yield of 7279 kg ha-1 although the increment over the 46 kg N ha-1 was minimal (only 13%). The largest increase in yield occurred between 0 and 138 kg ha-1. Generally 82.6% of the total variations in yield was mainly attributed to the effect of N fertiliser. Grain yield had significantly positive correlation with plant height (r=0.958), number of cobs (r=0.998), and total aboveground biomass yield (r=0.998).
There was a significant interaction effect of plant density and season on number of cobs (P<0.01) and grain yield of maize (P<0.05) (Table 1). It was observed that response to plant density was variable depending on seasonal distribution of moisture (Fig. 3). Percent response to plant density, over the optimum (53,000 plants ha-1), was low when distribution of rainfall was erratic while under even distribution the response was more. Under adequately distributed moisture, the highest population (89,000 plants ha-1) across seasons and N fertiliser produced the highest mean yield (8062 kg ha-1) whereas it produced the lowest under erratic rainfall. In contrast, under uneven distribution of precipitation, the lowest plant density (44,000 plants ha-1) gave the highest mean yield (5734 kg ha-1) while the same population gave 7085 kg ha-1 when the distribution was favourable. This meant that in seasons of better distribution of rainfall, the response was greater under higher plant density and accounted for 18 % of the variation in yield. In general, none of the density levels, averaged over seasons and N fertiliser, were superior in grain yield over the other (Table 2).
Economic analysis, with the minimum acceptable rate of return (MRR) being 100 %, shows that a combination of 46 kg N ha-1 and 53,000 plants ha-1 gave a marginal rate of return (MRR) of 286 % (Table 4). Similar results were obtained in the sensitivity and risk analyses (Tables 5 and 6). This combination seemed an optimum for resource-poor, risk averse farmers. On the other hand, a population of 53,000 plants ha-1 without N fertiliser was economical with MRR of 315 % which meant that for every birr (Ethiopian currency, 1US$ = 8 birr) a farmer invests he/she earns 3.15 birr; so also with sensitivity analysis.
Moisture is one of the major limiting factors for growth and development of a crop. Availability of nutrients and uptake by the crop is dependent on amount and distribution of precipitation in the growing seasons (Okalebo et al., 1999). Moisture affects leaf area development which serves as a means for light interception and photosynthetic activity. In this study, differences in seasonal distribution of rainfall were observed. In non-N fertilised plots, agronomic traits (plant height, number of cobs, seeds per cob, biomass and grain yields) were reduced due to uneven precipitation whereas the same characters showed more response to favourable distribution. Relative to erratic rainfall, mean grain yield obtained from favourable distribution of rainfall, averaged over N fertiliser and plant density, was 2183 kg ha-1 higher showing the significant effect of moisture distribution. The significant interaction of season with both nitrogen and population meant that the effect of either nitrogen or plant population was dependent on moisture distribution of the growing season.
This study showed the significant effect of N fertiliser in increasing agronomic traits and crop yield. Similar results were also reported by Sallah et al. (1998). The response to N fertiliser was more in seasons of erratic rainfall (1974 kg ha-1) when compared with 0-N. The reason for a better yield under uneven distribution could be that application of N fertiliser increased leaf area expansion and thus photosynthetic activity, rooting volume, and water-use efficiency which all contributed to better crop growth and development (Nielsen and Halvorson, 1991; Sallah et al., 1998). The results of Onken and Wendt (1989) showed that under rainfed condition application of N fertiliser increased water use efficiency and thus grain yield of sorghum; however, they stated that under severe moisture stress application of N did not increase water use efficiency and yield. Regardless of moisture distribution, crop yields without N fertiliser were reduced. Most likely, though not measured, the low yield in non-N fertilised plots could have been due to reduced leaf area development resulting in lesser radiation interception and, consequently, low efficiency in the conversion of solar radiation (Spedding et al., 1981). In addition, reduced rooting depth, which can extract stored soil moisture, could also be one factor for the low crop yield in zero N fertiliser. Lucas (1986) and Sallah et al. (1998) also reported reduced crop yield where N fertiliser was not applied. Increased application of N fertiliser, on the other hand, was accompanied by increased grain yield as previously reported by other researchers (Gardner et al., 1990; Oberle and Keeney, 1990; Nielsen and Halvorson, 1991; Killorn and Zourarakis, 1992; Sallah et al., 1998; Yaniag et al., 1998; Njui and Musandu, 1999; Mwato et al., 1999). Application of N might have improved rooting depth, leaf area expansion and, thus, the efficiency in the use of solar radiation, which indirectly increased yield (Spedding et al., 1981; Nielsen and Halvorson, 1991). If moisture is erratic during the growing season, growth and development of a crop will be retarded and this affects nutrient uptake (Killorn and Zourarakis, 1992; Okalebo et al., 1999). Analysis of variance over seasons and plant density showed a highly significant linear and quadratic effect of N. Overall, 82.6 % of the total variation in yield was accounted for by the addition of N fertiliser. Gardner et al. (1990) working in the US Corn Belt, reported that N fertiliser contributed most to grain yield of maize. In contrast Oberle and Keeney (1990) found relatively little variation (10 to 38 %) in maize yield due to N fertiliser on silt loam and silty clay loam soils, but this was probably related to relatively high initial amount of profile nitrate (NO3). Economic analysis indicated that 46 kg ha1 N combined with a population of 53,000 plants ha-1 was economical for the risk averse farmers because for every birr the farmer invests he/she gets a return of birr 2.86.
It was found that when moisture was adequate, yields were higher with higher plant population, which is in accordance with the finding of Nafziger (1994) whose work indicated that under drier condition response of maize to increased plant population was little while with adequate moisture the response was large. In this study, however, lower plant density gave higher grain yield when rainfall distribution was poor. It was observed that grain yield, averaged over seasons and N fertiliser, was not significantly affected by the difference in plant density. Various researchers (Truksa, 1993; Kaswende et al., 1997) also reported the non-significant effect of plant density on grain yield while Modarres et al. (1998) and Jagtap et al. (1998) indicated better yield response to increased plant density. The non-significance could probably be due to intra-and inter-competition at higher plant density. It might also be assumed that at lower plant density there was lower plant competition and thus a compensatory effect, i.e., there were more seeds per cob, kernel weight, and better plant growth at lower plant density. The result of multiple regression analysis indicated that plant density accounted for 18 and 2% of the total variation in yield under favourable and erratic weather conditions, respectively. Generally the contribution of plant population to grain yield was minimal (5.6%). In contrast to the results of this study, Nafziger (1994) reported that the contribution of plant density was large accounting for one-third of the total variation in yield. Increasing the plant density from 44x103 to 53x103 plants ha-1 was economical (marginal rate of return being 315.3 %). It means that one can use a plant density of 53x103 plants ha-1, which earns 3.153 birr for every birr invested, without fertiliser provided that the land is relatively fertile.
This study has shown that differences in seasonal distribution of precipitation significantly affected grain yield and other measured agronomic traits. N fertiliser had variable effect on grain yield of maize. Where there was uneven distribution of rainfall response to N was more and 81% of the variation in yield was attributed to application of N fertiliser. N fertiliser significantly affected agronomic traits of maize plant. Grain yield was increased over the yield from zero N following the application of 46,92,and 138 kg N ha-1. The linear and quadratic effects of N were significant and productivity index was more for the 46 kg N ha-1.
Grain yield had positively significant correlation with plant height, number of cobs, and total aboveground biomass yield. In seasons of favourable distribution of rainfall the highest plant density produced higher mean grain yield while under erratic rainfall the lowest plant density gave the highest yield. Eighteen percent of the total variation in yield under favourable distribution of precipitation was due to plant density. Generally, when averaged over seasons and N fertiliser, plant density did not significantly affect grain yield and contributed only 5.6% to the total variation in yield. Overall, application of 46 kg N ha-1 together with 53,000 plants ha-1 is economical; even planting of 53,000 plants ha-1 without fertiliser was economical.
Thanks are due to Dr.C. S.Wortmann for pre-reviewing the manuscript and the anonymous reviewers for their comments.
The following images related to this document are available:
Photo images[cs00029a.jpg] [cs00029c.jpg] [cs00029b.jpg]