Published online 3 January 2006
Published in Agron J 98:94-99 (2006)
DOI: 10.2134/agronj2005.0111
© 2006 American Society of Agronomy
677 S. Segoe Rd., Madison, WI 53711 USA
Modeling
Effect of Genotype, Nitrogen, Plant Density, and Row Spacing on the Area-per-Leaf Profile in Maize
Oscar R. Valentinuza and
Matthijs Tollenaarb,*
a Instituto Nacional de Tecnología Agropecuaria, EEA Paraná Ruta 11 km 12.5 (3101) Oro Verde, Entre Ríos, Argentina, and Universidad Nacional de Entre Ríos, Facultad de Ciencias Agropecuarias, CC 24 (3100) Paraná, Entre Ríos, Argentina
b Dep. of Plant Agriculture, Univ. of Guelph, Guelph, ON, Canada N1G 2W1
* Corresponding author (mtollena{at}uoguelph.ca)
Received for publication April 15, 2005.
 |
ABSTRACT
|
|---|
Accurate estimates of total leaf area and the vertical leaf area profile are important in process-based crop growth models. The bell-shaped function that quantifies the area-per-leaf profile of a maize (Zea mays L.) plant can be used to estimate the area-per-leaf profile. The objectives of this study were to quantify the effects of maize hybrid, soil N, plant density, and row spacing on the coefficients of the bell-shaped function. The coefficients of the bell-shaped function that quantify (i) the breadth of the area-per-leaf profile, (ii) the skewness of the area-per-leaf profile, and (iii) the position of the largest leaf were estimated using nonlinear regression in four datasets. Datasets consisted of the fully expanded leaf areas of all leaves on maize plants grown in studies performed in Ontario, Canada, between 1997 and 2001 that included combinations of maize hybrids, plant densities, N levels, and row spacing. Observations fitted well to the bell-shaped function (r2 > 0.95). The breadth of the area-per-leaf profile decreased under high soil N level and high plant density, and was lower for a newer than an older hybrid, whereas the opposite occurred with the position of the largest leaf. In contrast, the degree of skewness was not significantly altered by any of the factors examined in this study. Because of the relatively small impact of the examined agronomic factors on the coefficients of the bell-shaped function, a general model using mean coefficient values was validated with independent datasets. Results showed that this general bell-shaped function is a robust predictor of the area-per-leaf profile in maize.
Abbreviations: CHU, crop heat units • LAI, leaf area index
|
INTRODUCTION
|
|---|
LEAF AREA and the vertical leaf area profile influence the interception and utilization of solar radiation of maize crop canopies and, consequently, maize dry matter accumulation and grain yield. Rate of leaf expansion, maximum leaf area, and rate of leaf senescence are important factors in the estimation of canopy photosynthesis in crop growth simulation models that compute dry matter accumulation from temporal integration of canopy photosynthesis. In addition to total leaf area, the area-per-leaf profile or the vertical distribution of leaf area is also required when the calculation of canopy photosynthesis is based on sunlit and shaded leaf area across various layers in the crop canopy (e.g., Boote et al., 1996). In the model MAIS (Drost, 2001), a process-based crop growth model for maize grown under nonlimiting soil conditions, the growth of each individual leaf on the stem of the maize plant is estimated. The size of each leaf is computed from the time of its appearance, based on the relationship between rate of leaf tip appearance and temperature (Tollenaar et al., 1979), and the duration and rate of its expansion (Stewart and Dwyer, 1994a, 1994b). The distribution of leaf area by position conforms to a slightly skewed bell-shaped curve (Dwyer and Stewart, 1986). This function for any leaf number n is described by the equation:
where LAn is the area of the leaf on the nth position (leaves numbered from the bottom to the top), the amplitude (LAo) represents the area of the largest leaf, (xo) is the leaf position of the largest leaf, and b and c control the degree of breadth and skewness of the area-per-leaf profile, respectively. The four coefficients defining the bell-shaped function (i.e., LAo, b, xo, and c) can be biologically interpreted (Keating and Wafula, 1992) and, consequently, are useful in analyzing changes in the area per leaf profile inherent to growing seasons and agronomic practices.
Coefficients defining the bell-shaped function have been studied in both temperate and tropical maize. In temperate maize, Dwyer and Stewart (1986) compared the area-per-leaf profile of six datasets representing three growing seasons by normalizing the bell-shaped function. They found little year-to-year variation in the b, c, and xo coefficients, and their results showed that the prediction of the total LAI and area-per-leaf profile could be estimated from single estimations of the largest leaf, similar to that suggested by Francis et al. (1969). Keating and Wafula (1992) studied the area-per-leaf profile under a range of total number of leaves per plant in a tropical maize open-pollinated population grown without water and N limitations until silking. They reported that coefficients of the bell-shaped function could be estimated from the total number of initiated leaves per plant. The wide range of total number of initiated leaves per plant (i.e., 12–17) in the tropical maize population used by Keating and Wafula (1992) is not common in temperate maize hybrids and a single relationship using total number of initiated leaves per plant as the independent variable does not seem adequate for estimating coefficients of this function in temperate single-cross hybrids. The impact of important agronomic factors such as maize hybrid, soil N, row spacing, and plant density on the coefficients of the bell-shaped function needs to be documented before this function can be effectively used in the estimation of the vertical profile of leaf area in maize simulation models.
The objectives of this study were to examine the effect of a number of agronomic variables on the coefficients of the bell-shaped function that quantifies the area-per-leaf profile. Values of the three coefficients in the bell-shaped function that determine the leaf area distribution across vertical plane (i.e., b, xo, and c) were estimated using nonlinear regression in datasets of studies performed between 1997 and 2001, which included combinations of maize hybrids and large ranges in plant densities, N levels, and row spacing. Because of the relatively small impact of a wide range of conditions studied on the coefficients quantifying the area-per-leaf profile, we were able to use a general function to estimate areas of individual leaves and compare them with measured values from an independent dataset.
|
MATERIALS AND METHODS
|
|---|
Datasets used in this study were generated from field studies that were performed at the Woodstock and Elora Research Stations (Ontario, Canada) between 1997 and 2001 (Table 1). In short, studies included the maize hybrids Pioneer 3893 and Pioneer 39P06 grown at 7 and 9 plants m– 2 in three row spacings (51, 76, and 76 cm in twin rows) at a range of N fertilizer amendments, from 1997 to 2000 at the Woodstock Research Station (unpublished data, 2001) and the maize hybrids Pride 5, Pioneer 3902, and Pioneer 3893 grown at three plant densities (1, 3.5, and 12 plant m– 2) during 2001 at the Elora Research Station (Valentinuz and Tollenaar, 2004). Twin rows have a row configuration where two rows with an 18-cm row spacing (twins) are positioned on 76-cm centers. Pioneer 3893 and Pioneer 39P06 are single-cross hybrids that were introduced during the 1990s, Pioneer 3902 is a single-cross hybrid that was introduced during the late 1980s, and Pride 5 is a double-cross hybrid that was introduced during the late 1950s. Relative maturity of the hybrids ranged from approximately 2650 crop head units (CHU) (Brown and Bootsma, 1993) for Pride 5 and Pioneer 3902 to approximately 2750 CHU for Pioneer 3893 and Pioneer 39P06. Experimental design of all studies was a complete randomized block with four replications, arranged in a split, split plot design. Weeds and pests were chemically controlled. When soil N was not a factor to be tested, plots were fertilized with N according to soil test recommendations. At Woodstock, experiments were machine planted and at Elora all plots were hand planted and thinned to the target plant density. Data for the first three datasets in Table 1 overlapped and were selected to focus on year effects (Dataset 1), plant density effects (Dataset 2), and hybrid and soil N effects (Dataset 3).
View this table:
[in this window]
[in a new window]
|
Table 1. Agronomic inputs for the datasets where the area of each individual leaf was measured. Data for the first three datasets overlapped and were selected to focus on year effects (Dataset 1), plant density effects (Dataset 2), and hybrid and soil N effects (Dataset 3).
|
|
Area of all initiated leaves (i.e., from the first seedling leaf to the topmost leaf that emerges before tasseling) was measured soon after leaves were fully expanded. Leaves were numbered from the bottom to the top and length and maximum width of all leaves on five consecutive plants in a row per plot were measured and the leaf area of each individual leaf calculated by multiplying length by maximum leaf width by 0.75 (Montgomery, 1911). Leaf length was measured as the distance between the collar and the tip. Measurements were taken three times between plant emergence and silking to obtain a measurement of all fully expanded leaves before leaf senescence or leaf breakage occurred. The area of each individual leaf was normalized with respect to the largest leaf. Each five-plant plot was bordered by at least 3 m on each side within the row and two rows on each side between the rows. For each plot, the b, c, and xo coefficients of the bell-shaped function were estimated using nonlinear regression performed in the Proc N-LIN of SAS (SAS Institute, 1997). Each dataset was fitted to the bell-shaped function (i.e., r2 > 0.95) and, subsequently, coefficients were compared using an analysis of variance performed in the PROC-GLM of SAS. Mean comparisons were made with the LSD test.
The predictive value of the general bell-shaped function, using mean values of the coefficients across the four datasets in Table 1 (i.e., b = 0.04019, c = 0.00015, xo = 12) and the measured area of the largest leaf (i.e., LAo), was evaluated in datasets that had not been used to quantify the b, c, and xo coefficients. The general function was evaluated in three independent datasets that included area-per-leaf (mean of five plants) of the hybrid Pioneer 3893 grown in 1997 at 7 plants m– 2 in three row spacings, with an N amendment of 60 kg N ha– 1, and the hybrid Pioneer 3893 grown in 1998 at 7 and 9 plants m– 2, with an N amendment of 60 kg ha– 1 (cf., Table 1), and the hybrids Pride 5, Pioneer 3902, and Pioneer 3893 grown in 2001 at 1 plant m– 2 (Valentinuz and Tollenaar, 2004).
|
RESULTS
|
|---|
There was good agreement between normalized observations and the function describing the bell-shaped function (r2 > 0.95) for the variables year, hybrid, soil N, plant density, and row spacing. Analyses of variance for the b, c, and xo coefficients (Table 2) show that the c coefficient, which describes the degree of skewness in the area-per-leaf profile, was not affected by any factor included in our datasets. Therefore, our analyses were focused on the other two coefficients of the bell-shaped function (b and xo) that were significantly influenced by the treatments.
View this table:
[in this window]
[in a new window]
|
Table 2. ANOVA of the four datasets studied. The parameters b, c, and xo are the coefficients of the bell-shaped function normalized in respect to the largest leaf.
|
|
Year, plant density, and hybrid had a large impact on the b coefficient, whereas the effects of soil N and row spacing on the b coefficient were relatively small. Analysis of variance showed significant effects of year on the b coefficient (Table 2). The variation between maximum and minimum values of b due to years was 18% in Dataset 1, 6% in Dataset 2, and 13% in Dataset 3 (Table 3). Analysis of variance showed that plant density affected values of b only in the dataset with a wide range of plant density. The value of b increased 16% as plant density increased from 3.5 to 12 plants m– 2 in Dataset 4 (Table 3), whereas differences were nonsignificant in Dataset 2 when plant density was close to the commercial range (7 and 9 plants m– 2). Analysis of variance showed that hybrids differed significantly in respect to b values. In Dataset 3, the value of b was 13% greater for Pioneer 3893 than for Pioneer 39P06 and in Dataset 4, b was 19% greater for Pride 5 than for Pioneer 3893 (Table 3). Row spacing did not significantly affect the value of b in two out of three datasets (Datasets 1 and 2), however, in Dataset 3 the coefficient b was greater in plants established in twin rows (p < 0.07) than in plants established in single rows at a row width of 76 cm (Table 3). Analysis of variance showed that soil N significantly altered the values of b in the one dataset in which soil N was varied. Although soil N levels (0 vs. 180 kg ha– 1) resulted in significant differences in leaf area index and grain yield (data not shown), the difference in the b coefficient was only 7% (Table 3).
View this table:
[in this window]
[in a new window]
|
Table 3. Values of b and xo coefficients in four datasets. Only those factors that had a significant effect on the coefficients are shown.
|
|
The response of the xo coefficient to year, hybrid, plant density, soil N, and row spacing generally mirrored that of the b coefficient (Table 2). The coefficient xo varied with year in two of the three datasets. These two datasets included 1998, a year characterized by high temperature at the beginning of the growing season that resulted in smaller total number of initiated leaves (18 vs. 20) and, consequently, in a higher position of the largest leaf in 1998 than in the other years. The value of xo varied significantly when a wide range of plant density was tested (Table 3), but no difference was found when 6.9 and 8.9 plants m– 2 were compared. Hybrids varied significantly in relation to the position of the largest leaf, probably as a result of differences in total number of leaves. The value of xo was 12.5 in Pioneer 39P06 and 11.7 in Pioneer 3893 in Dataset 3. In Dataset 4, the largest leaf was positioned lower in Pride 5 than in either Pioneer 3902 or Pioneer 3893 (11.4 vs. 12.2). Similarly to that observed for coefficient b, row spacing altered the value of xo only in Dataset 3, where small but significant differences were observed between 76 cm and 76 cm twin rows (12.0 vs. 12.2). The largest leaf was positioned in a higher position when N application increased from 0 to 180 kg ha– 1 (11.9 vs. 12.3).
In spite of significant changes observed in b and xo in response to the treatments, the impact per se of b and xo on leaf area per plant (expressed as the sum of all normalized values of leaf area) was small. For instance, coefficient b in Pride 5 was 12% greater than the mean value across the three hybrids, but the increase in leaf area per plant attributable to a flatter area profile was not greater than 7% (Table 4). Similarly, maximum changes in the position of the largest leaf due to hybrids and years (4%) had little effect on total leaf area per plant (Table 4).
View this table:
[in this window]
[in a new window]
|
Table 4. Largest difference between mean values for b and xo used in the general bell-shaped function (i.e., b = 0.04019 and xo = 12) and values encountered in Datasets 1 to 4 and the impact of the difference on total leaf area per plant. When changes in b were evaluated, xo was the mean of all four datasets and vice versa
|
|
The ability of the bell-shaped function to predict the area of individual maize leaves and total plant leaf area was evaluated using independent datasets (i.e., datasets that had not been used in the quantification of the coefficients of the bell-shaped function). Because of the small impact of variations in b and xo on total leaf area in addition to low sensitivity of the c coefficient to the environmental variables, a general bell-shaped function with mean values of b, c, and xo (i.e., b = 0.04019, c = 0.00015, xo = 12) was used to compare predicted and measured leaf areas. The general bell-shaped function was evaluated by comparing the areas of 544 individual leaves in three independent datasets with the leaf areas estimated by the general function and the area of the largest leaf (i.e., the absolute value of LAo). Although the linear relationship was significant (P < 0.05, r2 = 0.93), the slope was lower than 1 (Fig. 1
), indicating that the general function underestimated the observed values. Estimated values continued to be slightly lower than the observed values when the independent set of data was divided into groups differing in row spacing, plant density, and hybrid (Table 5).
View larger version (33K):
[in this window]
[in a new window]
|
Fig. 1. Observed values of individual leaf area compared with those estimated from the bell-shaped function. For each individual leaf, its area (LAn) was estimated as LAn = LAo x exp[0.04019 x (xn – 12)2 + 0.00015 x (xn – 12)]3, where LAo was the area of the largest leaf and xn was the leaf position of the nth leaf.
|
|
View this table:
[in this window]
[in a new window]
|
Table 5. Slope and coefficient of determination of each linear relationship between observed and estimated values in three independent datasets.
|
|
|
DISCUSSION
|
|---|
The objective of this study was to evaluate changes in the area-per-leaf profile in response to a range of conditions, including year effects, management practices, and genotypes. Coefficients describing the normalized bell-shaped function were used to quantify changes in the area-per-leaf profile. Our analysis revealed that year, hybrid, plant density, N, and row spacing influenced only coefficients that describe the breadth of the bell-shaped function and the position of the largest leaf. Although effects on the b and xo coefficients were significant, the overall impacts were fairly small (Fig. 2
).
View larger version (18K):
[in this window]
[in a new window]
|
Fig. 2. Normalized leaf area profile for (A) two maize hybrids, and (B) for maize plants grown at either two plant densities or (C) two levels of N application.
|
|
The year effect on values of b and xo may be related to incident solar radiation and temperature during early phases of development. Values of b were inversely associated with the mean daily incident solar radiation during a period of approximately 30 d after planting (Fig. 3A
), indicating that area-per-leaf profile becomes flatter as incident solar radiation increases during a period when the first 12 leaf tips appear (of a total of 18–20 leaves). These results are consistent with the response of b to plant density, which shows that the area-per-leaf profile was flatter in low than in high plant density (i.e., a higher b value at a lower incident solar radiation per plant) (Table 3). Total initiated leaf number and rate of leaf appearance in maize increase when incident shortwave radiation increases from approximately 10 to 20 MJ m– 2 d– 1 during the period from planting to the 15-leaftip stage (Tollenaar, 1999), but it is not clear whether there exists a direct relationship between total leaf number and rate of leaf appearance and the breadth of the area-per-leaf profile. Values of xo were associated with mean temperature during a period of approximately 30 d after planting in a quadratic fashion with an optimum temperature (i.e., the temperature in which the position of the largest leaf is the greatest) around 16°C (Fig. 3B). Curiously, it has been reported that the relationship between the total number of initiated leaves in maize and temperature is also quadratic, but the response is inverted with the lowest number of leaves at 15°C (Warrington and Kanemasu, 1983). The response of xo to temperature is similar to the response of dry matter accumulation and LAI at the 12-leaf stage to temperature for maize grown under controlled-environment conditions, which showed a quadratic response with an optimum temperature of about 15°C for dry matter and 19°C for LAI (Tollenaar, 1989). The lower position of the largest leaf (xo) at higher plant density (Table 3) may be a result of earlier competition for assimilates at high plant densities. Small changes in b and xo with a reduction in row spacing suggest that a more uniform plant stand is not effective in alleviating early competition among plants. Flénet et al. (1996) reported a reduction in the extinction coefficient (i.e., a more uniform light distribution across the canopy) as row spacing decreased from 1 to 0.35 m, but our results indicate that this reduction may not be related to the area-per-leaf profile. Breadth of the area-per-leaf profile seems to be a genotype specific trait and the largest differences among hybrids in coefficient b ranged from 13% (Dataset 3) to 19% (Dataset 4). Interestingly, the b values reported by Dwyer and Stewart (1986) and values estimated from the relationship reported by Keating and Wafula (1992) for a tropical maize plant with 19 leaves were within the range of b values we observed in Pioneer 3893 in our datasets.
View larger version (17K):
[in this window]
[in a new window]
|
Fig. 3. Relationship between (A) the b coefficient and mean daily incident solar radiation and (B) between the xo coefficient and mean temperature during a period of 30 d after planting. Data included 4 yr in Woodstock (squares) and 1 yr in Elora (diamonds).
|
|
Variations in b and xo had small effects on total leaf area as estimated from the normalized bell-shaped function. Reduction in the value of the b coefficient resulted in an increase in leaf area per plant, but the increase in leaf area resulting from a flatter area-per-leaf profile due to increased incident solar radiation, low plant density, and newer hybrids was never greater than 10% (Table 4). This percentage represented the per se impact of changes in the breadth of the profile whenever the area of the largest leaf was comparable. In addition, the increase on total leaf area attributable to a higher placement of the largest leaf was negligible (Table 4).
Based on the relatively small impact of a wide range of conditions studied on coefficients quantifying the area per leaf profile, we were able to use a general function to estimate areas of individual leaves from the area of the largest leaf and compare them with measured values from an independent dataset. Although estimated values of individual leaf area were lower than observed values, the underestimation was consistent across different management practices and hybrids. Therefore, one can speculate that by quantifying the changes in the area of largest leaf associated with experimental year, diverse management practices, or genotype, their impact on full leaf area could be easily determined in yield trials, breeding nurseries, and extensive field test. Thus, our results constitute an expansion of that reported by Francis et al. (1969), who also reported a relationship between the area of the largest leaf and the leaf area per plant before variations inherent to years, genotypes, and plant densities.
In summary, this study supports the contention that the bell-shaped function is a robust predictor of the area-per-leaf profile in maize. A single general function was proposed to estimate the area-per-leaf profile in a maize canopy. Even though two out of the three coefficients of the function varied with year and agronomic practices, their impact per se on the total leaf area estimated from the area of the largest leaf was never greater than 10%. Since the calculation of the profile depends on knowing the area of the largest leaf, we postulate that by measuring the impact of varying environmental or management conditions on the largest leaf, their respective effects on total leaf area could be determined.
|
REFERENCES
|
|---|
- Boote, K.B., J.W. Jones, and N.B. Pickering. 1996. Potential uses and limitations of crop models. Agron. J. 88:704–716.[Abstract/Free Full Text]
- Brown, D.M., and A. Bootsma. 1993. Crop heat units for corn and other warm season crops in Ontario. Factsheet Agdex 111/31. Ontario Ministry of Agric. and Food, Queen's Park, ON, Canada.
- Drost, R.G. 2001. MAIS: A mechanistic model of maize growth and development. M.S. thesis. Univ. of Guelph, Guelph, ON, Canada.
- Dwyer, L.M., and D.W. Stewart. 1986. Leaf area development in field-grown maize. Agron. J. 78:334–343.[Abstract/Free Full Text]
- Flénet, F., J.R. Kiniry, J.E. Board, M.E. Wesgate, and D.C. Reicosky. 1996. Row spacing effects on light extintion coefficients of corn, sorghum, soybean, and sunflower. Agron. J. 88:185–190.[Abstract/Free Full Text]
- Francis, C.A., J.N. Rutger, and F.E. Palmer. 1969. A rapid method for plant leaf area estimation in maize (Zea mays L.). Crop Sci. 9: 537–539.[Abstract/Free Full Text]
- Keating, B.A., and B.M. Wafula. 1992. Modelling the fully expanded area of maize leaves. Field Crops Res. 29:163–176.
- Montgomery, E.G. 1911. Correlation studies in corn. Nebraska Agric. Exp. Stn. Annu. Rep. 24:108–159.
- SAS Institute. 1997. SAS/STAT user's guide. 6th ed. Vol. 2. SAS Inst., Cary, NC.
- Stewart, D.W., and L.M. Dwyer. 1994a. Appearance time, expansion rate and expansion duration for leaves of field-grown maize (Zea mays L.). Can. J. Plant Sci. 74:31–36.
- Stewart, D.W., and L.M. Dwyer. 1994b. A model of expansion and senescence of individual leaves of field-grown maize (Zea mays L.). Can. J. Plant Sci. 74:37–42.
- Tollenaar, M. 1989. Response of dry matter accumulation of maize to temperature: I. Dry matter partitioning. Crop Sci. 29:1239–1246.[Abstract/Free Full Text]
- Tollenaar, M. 1999. Duration of the grain filling period in maize is not affected by the photoperiod and incident PPFD during the vegetative phase. Field Crops Res. 62:15–21.[CrossRef]
- Tollenaar, M., T.B. Daynard, and R.B. Hunter. 1979. Effect of temperature on rate of leaf appearance and flowering date in maize. Crop Sci. 19:363–366.[Abstract/Free Full Text]
- Valentinuz, O., and M. Tollenaar. 2004. Vertical profile of leaf area and leaf senescence during the grain-filling period in maize. Crop Sci. 44:827–834.[Abstract/Free Full Text]
- Warrington, I.J., and E.T. Kanemasu. 1983. Corn growth response to temperature and photoperiod: II. Leaf initiation and leaf appearance rates. Agron. J. 75:755–761.[Abstract/Free Full Text]
This article has been cited by other articles:
|
|
|
|
 
V. Ciganda, A. Gitelson, and J. Schepers
Vertical Profile and Temporal Variation of Chlorophyll in Maize Canopy: Quantitative "Crop Vigor" Indicator by Means of Reflectance-Based Techniques
Agron. J.,
September 8, 2008;
100(5):
1409 - 1417.
[Abstract]
[Full Text]
[PDF]
|
|
|
TOP
ABSTRACT
INTRODUCTION
