Agronomy Journal 94:734-742 (2002)
© 2002 American Society of Agronomy
MODELING
Intercropping System Optimization for Yield, Quality, and Weed Suppression Combining Mechanistic and Descriptive Models
Daniel T. Baumann*,a,
Lammert Bastiaansb and
Martin J. Kropffb
a Swiss Federal Res. Stn. for Fruit-Growing, Viticulture and Hortic., P.O. Box 185, CH-8820 Wädenswil, Switzerland
b Group of Crop and Weed Ecology, Dep. of Plant Sci., Wageningen Univ., P.O. Box 430, NL-6700 AK Wageningen, the Netherlands
* Corresponding author (daniel.baumann{at}faw.admin.ch)
Received for publication October 18, 2000.
 |
ABSTRACT
|
|---|
Intercropping leek (Allium porrum L.) with celery (Apium graveolens L.) is an option to reduce growth and reproductive potential of weeds while maintaining productivity. In this study, a combined modeling approach is used to optimize a leek—celery intercropping system with respect to crop yield, product quality, and weed suppression. An ecophysiological model for interplant competition was used to simulate yield and product quality of the crops as well as biomass and seed production of the weed Senecio vulgaris L. for a wide range of crop mixtures and weed infestations. The results of the simulations were summarized using a descriptive hyperbolic yield–density model, which then allowed evaluation of the intercropping system in terms of productivity, product quality, and ability to suppress weeds. In a weed-free mixture, the competitive ability of celery was six times higher than that of leek. With respect to late-emerging S. vulgaris, the relative competitive ability of leek was 5.4 times lower than that of celery. Replacing two leek plants of a leek monoculture by one celery plant resulted in almost 20% biomass reduction of late-emerging S. vulgaris. Crop mixtures with a leek density of about 20 plants m-2 and a leek/celery ratio of 2 proved to be the optimum intercropping system, given the current price ratios. Compared with leek monoculture, profitability was maintained, and late-season weed suppression was greatly increased, resulting in reduced weed seed production.
Abbreviations: NDI, niche differentiation index • RCA, relative competitive ability
|
INTRODUCTION
|
|---|
RECENT STUDIES have addressed intercropping as an option for an integrated weed management, particularly in farming systems with low external inputs (Caporali et al., 1998; Itulya and Aguyoh, 1998; Liebman and Davis, 2000; Rana and Pal, 1999; Schoofs and Entz, 2000). Effects of crop diversification on weeds have been reviewed by Liebman and Dyck (1993), Liebman and Ohno (1998), and Teasdale (1998). As an example of functional biodiversity, intercropping leek (Allium porrum L.) with celery (Apium graveolens L.) showed various beneficial effects, such as the reduction of weeds and pests and an improved resource capture, while cropping practices were not hampered (Baumann et al., 2000, 2001a). Celery improved weed suppression by the canopy by increasing its light interception. As a result, incoming radiation was captured more efficiently by the intercrop canopy, and less radiation was available for germination and growth of weeds. However, the strong relative competitive ability (RCA) of celery in the intercropping system resulted in a loss of leek quality because stem diameter was reduced to <20 mm (market criterion) (Baumann et al., 2001a). The authors, therefore, concluded that the intercropping system needed to be optimized with respect to crop quality and weed suppression for successful implementation and suggested applying ecophysiological simulation models to optimize the system. Earlier, Kropff and Van Laar (1993) advocated the use of modeling to develop and optimize weed management systems with respect to cost effectiveness and minimization of environmental effects.
Ecophysiological crop growth models can be very effective to evaluate and develop complex systems, such as multispecies plant communities (Kropff and Van Laar, 1993). Based on physiological, morphological, and phenological processes, such models provide insight into the competitive relationships of the system. These models facilitate the exploration of complex systems without extensive field experimentation to investigate all options in a wide range of conditions. Empirical models and regression techniques can help analyze the final outcome of competition trials and describe plant interference in cropping systems. Approaches to describe interplant relationships have been developed for a long time and have helped improve understanding of competitive effects between crops and between crops and weeds (De Wit, 1960; Kira et al., 1953; Shinozaki and Kira, 1956).
The current study attempts to combine a mechanistic and descriptive modeling approach to optimize the system. A well-evaluated ecophysiological model, such as INTERCOM (Kropff and Van Laar, 1993), provides the necessary insight into the processes and plant characteristics determining mutual competitive effects and allows generating a large number of data sets for a wide range of densities and environments. Subsequent application of a descriptive model to the generated data sets can help summarize the results, calculate the RCA of the system components, and describe yield and product quality of the component crops in relation to plant density and mixing ratios. The objective of this study was to evaluate the use of combined modeling approaches for analysis and design of a leek and celery intercropping system to optimize this system with respect to yield and quality while improving weed suppression.
|
MATERIALS AND METHODS
|
|---|
Model Description
The ecophysiological competition model INTERCOM (Kropff and Van Laar, 1993) was used to simulate interplant competition in a leek–celery intercropping system. The model was simplified with respect to physiological processes but included the original detailed simulation of competition for light (Baumann et al., 2002). Because water and nutrients were available in ample supply in the experimental system, competition for these resources was not simulated in this version of the model. The competition model was parameterized using experimental data from pure stands of the crops. Validation with independent data showed that the model accurately simulated growth in both monocultures and mixtures. For a detailed description of the model, the ecophysiological characteristics of the crops, and the underlying experiments, the authors refer to Baumann et al. (2002). To study the growth of Senecio vulgaris L. and its effect on intercrop performance, the model was extended to include this weed species. Parameter values were derived from field experiments and additionally from earlier studies carried out by Schnieders (1999); they are summarized in Table 1. The model was validated with independent data from monocultures and mixtures of the three species collected in two field experiments carried out in 1997 and 1998.
View this table:
[in this window]
[in a new window]
|
Table 1. Summary of the parameter estimates used for parameterization of the model INTERCOM for leek, celery, and Senecio vulgaris.
|
|
Field Experiments I and II (Replacement Series of Leek and Celery with Additive S. vulgaris)
Two field experiments, referred to as Exp. I (1997) and II (1998), were carried out on a sandy loam soil (Inceptisol; 17% clay, 24.5% silt, 54.2% sand, pH 7.8, and 4.3% organic matter) at the Sandhof experimental farm of the Swiss Federal Research Station for Fruit-Growing, Viticulture, and Horticulture at Wädenswil, Switzerland (47°13' N, 08°40' E). The experiments were set up to examine the effect of pure and mixed stands of leek and celery on the biomass and reproduction of S. vulgaris in relation to its relative time of emergence in the crops. Rows of S. vulgaris were sown between crop rows at a density of 50 plants m-2. For both experiments, a split-plot block design with three replicates was used. Crop system (monoculture leek, monoculture celery, and intercrop of the two species) was the main-plot factor. Plant densities were 18 and 9 plants m-2 for leek and celery, respectively, and the intercrop was arranged as a row-based replacement series of the two crops. The relative emergence time of S. vulgaris was the split-plot factor. In Exp. I, S. vulgaris was sown at seven times, each 10 d apart, with the first sowing at 10 d after crop establishment. In Exp. II, there was five sowings, each 10 d apart, starting at crop establishment. In both experiments, a weed-free plot was included. In Exp. II, a monoculture of S. vulgaris was included, which was used to generate data for model parameterization. A comprehensive description of the experimental details of the two experiments is given in Baumann et al. (2001b).
Simulations
After validation of the model (Baumann et al., 2002), the performance of pure and mixed crop stands with and without S. vulgaris was simulated for the environmental conditions of the Sandhof experimental farm. Plant density for leek was varied between 0 and 25 plants m-2, and plant density of celery was varied between 0 and 20 plants m-2. Plant density of S. vulgaris remained constant at 50 plants m-2 at a relative emergence time of 0, 10, 20, 30, and 40 d after crop establishment. Simulation runs were conducted with weather data of 1997 and 1998 from Wädenswil, Switzerland, for all combinations of crop densities with and without S. vulgaris. Biomass production and per-plant mass of the species, after a growing period of 88 d for 1997 and 92 d for 1998, were output of the model. For leek, the diameter of the pseudostem, which is used as a quality parameter, was calculated based on the per-plant mass because a high correlation (r2 = 0.92) between the dry mass of aboveground organs and pseudostem diameter had been found in earlier experiments (Baumann et al., 2002). For celery, the per-plant fresh mass was calculated based on an average dry matter content of 7.3%, which was found in Exp. I and II, and did not differ significantly between the various treatments. For S. vulgaris, seed production was estimated based on the established linear relationship between per-plant dry mass and number of seeds per plant (Baumann et al., 2001b; Schnieders, 1999).
Data Analysis
Biomass production of S. vulgaris in the intercrop was expressed relative to its biomass production in leek monoculture. The effect of progressively replacing leek by celery in the mixture on S. vulgaris biomass was then analyzed with a hyperbolic function using celery density as an explanatory variable (Cousens, 1985):
 | [1] |
where RY is the relative yield of S. vulgaris (as biomass fraction of its biomass in leek monoculture), NC is the plant density of celery (plants m-2), a is a parameter describing the effect of replacing the first leek with celery (m2 plant-1), and m is the maximum relative yield loss of S. vulgaris in intercropping.
To analyze the crop performance of the intercropping system, the relative yield total (RYT) was calculated for all replacement series of the simulated crop stands (De Wit, 1960):
 | [2] |
where Y is the crop yield (kg ha-1) and the suffixes L and C denote leek and celery, respectively. Thus, YLC is the yield of leek when grown in mixture, and YLL is the yield of leek when grown in monoculture. YCL and YCC are the corresponding yields for celery in mixture and monoculture, respectively.
Additionally, the RCA of the crops was analyzed using an approach proposed by Spitters (1983), Watkinson (1981), and Wright (1981). This approach is based on the notion that the biomass–plant density response can be described by a rectangular hyperbola (De Wit, 1960; Spitters, 1983). The model relates the biomass of each species to the density of both species in the mixture, and the yield, Y (g m-2), of a component crop can be calculated by:
 | [3] |
where N1 and N2 are the plant densities (plants m-2) of Crop 1 and 2, respectively; b1,0 is the intercept denoting the reciprocal of the virtual biomass of an isolated plant of Crop 1 (plant g-1); and b1,1 and b1,2 (m2 g-1) are parameters for intra- and interspecific competition, respectively. The ratio of these last two parameters denotes the RCA between both crops with respect to the production of the first crop. A similar ratio was calculated with respect to the production of the second crop.
Based on the coefficients for intra- and interspecific competition of leek and celery, the niche differentiation index (NDI) was calculated (Spitters, 1983):
 | [4] |
The same approach was used to analyze interplant competition in a system with three components—leek, celery, and S. vulgaris—by expanding Eq. [3] with an additional parameter to account for the third species:
 | [5] |
where YL,C,S is the yield of leek in presence of celery and S. vulgaris (g m-2) and NL, NC, and NS are the plant densities of leek, celery, and S. vulgaris, respectively. Dividing yield by plant density of corresponding species results in the per-plant mass, which was used to derive crop quality parameters and seed production for S. vulgaris:
 | [6] |
where WL,C,S is the per-plant dry mass (g plant-1) of leek in the presence of celery and S. vulgaris.
Optimization of the Intercropping System
To optimize the intercropping system, crop mixtures with either the same quality, the same yield, or a similar weed-suppressive ability were determined by calculating isolines for these parameters. For this purpose, Eq. [5] was rewritten to obtain an expression for plant density of celery (NC). Isolines with equal biomass production of each of the components of the mixture were then determined by taking a specific crop yield (Y) and calculating the corresponding NC for a range of leek densities (NL). To calculate isolines for mixtures with similar quality, quality parameters for leek and celery and seed production for S. vulgaris were first converted into per-plant dry mass. Consequently, a similar procedure as described for yield isolines was followed for per-plant dry mass isolines using Eq. [6]. Isolines for total biomass production of the intercrop were calculated in an identical way. Here Eq. [5] for biomass production of leek and celery were first added, after which the combined equation was rewritten to obtain an equation for NC. Accordingly, the total financial yield, YT, of the mixture was calculated as:
 | [7] |
where P is the price of the product received by the farmer
(€ kg-1),
the other parameters are defined as indicated for Eq. [3], and the suffixes L and C are for leek and celery, respectively. Rewriting Eq. [7] for NC, which results in a quadratic equation, allowed calculation of isolines for crop stands with equal financial yield. For the calculation, average prices achieved by farmers over a 5-yr period between 1993 and 1998 were used (Spigt and Janssen, 1997). The crop stand with the highest financial gross return was detected by determining the intersection of the YT isoline and the minimum quality isoline for leek. This was established by introducing the equation for the minimum quality isoline into the equation of the YT isoline. Calculating the celery density for which the first derivative, with respect to financial yield, of this combined equation equals zero made it possible to determine the crop densities of the mixture with the highest financial yield. The sensitivity of the yield and crop densities to a 5% change of the prices was tested.
|
RESULTS
|
|---|
Model Performance
The model was calibrated based on data from monocultures of the crops grown in field experiments carried out in 1996 and 1998 (Baumann et al., 2002). Calibration for S. vulgaris was based on experimental data from monocultures of the weed in Exp. II (1998) and from literature (Schnieders, 1999). For the model evaluation, independent data sets from mixed stand treatments of Exp. I and II were used. Dry matter production was simulated accurately for leek monoculture and mixture in 1997 and 1998 (Fig. 1A)
. For celery, simulations with 1998 weather data underestimated the observed biomass production in the mixed stand compared with observed data in Exp. II (Fig. 1B). For the mixture in Exp. I and celery pure stands of both years, simulation of celery production was acceptable. Standard errors for celery dry matter production in the experiments were high in both years. The model very accurately simulated S. vulgaris biomass in all crop stands (Fig. 2)
for 1998. The model could not be evaluated for the 1997 data because S. vulgaris had been infected by Puccinia lagenophorae Cook., which caused early senescence and a strong reduction of the biomass. Biomass of S. vulgaris was more reduced in all crop stands when the plants emerged later than the crop. Biomass was more reduced in the crop mixture and celery monoculture compared with leek monoculture, particularly for early dates of emergence.
View larger version (24K):
[in this window]
[in a new window]
|
Fig. 1. Measured (symbols) and simulated (lines) shoot dry matter of (A) leek and (B) celery at harvest in monoculture (filled symbols) and mixture (open symbols) as affected by the relative time of emergence of Senecio vulgaris. Results of 1997 (Exp. I; squares and dashed lines) and 1998 (Exp. II; circles and solid lines). Error bars are standard errors of means.
|
|
View larger version (17K):
[in this window]
[in a new window]
|
Fig. 2. Measured (symbols) and simulated (lines) shoot dry matter of Senecio vulgaris grown in (A) leek monoculture, (B) leek–celery intercrop, and (C) celery monoculture as affected by the relative time of emergence. Error bars are standard errors of means.
|
|
Relative Competitive Ability
Parameter estimates for nonlinear regression of density–yield relationships for simulated competition results using Eq. [5] are given in Table 2. Under weed-free conditions, the competitive ability of celery was about three times higher than the that of leek, both with respect to celery and leek production (Fig. 3A) . If S. vulgaris was introduced in the intercrop at the time of crop transplanting, both RCAs were slightly changed (about 15%) to the benefit of leek. For later planting dates of S. vulgaris, the RCA of both crops differed <3% from that of the weed-free mixture. For the weed-free mixture of leek and celery, an NDI of 1.45 was calculated (Eq. [4]), indicating a slight complementarity in capture and/or use of light for leek and celery. The value NDI for leek and S. vulgaris ranged between 2 and 5, whereas NDIs between 1.7 and 3.2 were found for celery and S. vulgaris, depending on the relative time of weed emergence.
View this table:
[in this window]
[in a new window]
|
Table 2. Estimates and standard errors of the parameters for density response functions of simulated biomass production (Eq. [5]) in a leek–celery intercropping system with Senecio vulgaris introduced at different times relative to crop establishment. Adjusted r2 values were  0.99 for all estimates.
|
|
View larger version (17K):
[in this window]
[in a new window]
|
Fig. 3. Effect of relative time of emergence of Senecio vulgaris (50 plants m-2) on relative competitive ability (RCA) of (A) leek vs. celery with respect to leek production ( ; bL,L/bL,C) and celery vs. leek with respect to celery production (•; bC,C/bC,L) (dotted lines indicate the relative competitive ability of the weed-free mixtures), (B) leek vs. S. vulgaris with respect to leek production (; bL,L/bL,S) and celery vs. S. vulgaris with respect to celery production (•; bC,C/bC,S), and (C) S. vulgaris vs. leek with respect to production of S. vulgaris (; bS,S/bS,L) and S. vulgaris vs. celery with respect to production of S. vulgaris (•; bS,S/bS,C).
|
|
When S. vulgaris was planted together with the crops, 12 plants of S. vulgaris were found equally competitive with either one leek or one celery plant. For later planting dates of S. vulgaris, the ratio between the RCA of celery vs. S. vulgaris and leek vs. S. vulgaris with respect to the productivity of the respective crops (RCACS/RCALS) increased linearly, reaching a value of 5.4 when S. vulgaris was introduced as late as 40 d after crop establishment (Fig. 3B). The large differences in competitive strength between leek and celery with respect to S. vulgaris were also reflected in RCASC and RCASL, which differed markedly for the early plantings of S. vulgaris (Fig. 3C).
The relative yield total of the intercrop (Eq. [2]) ranged from 1.0 to 1.03 over a wide range of densities of the two crops and was not affected by introduction of S. vulgaris (Fig. 4A)
. The relative biomass of S. vulgaris was reduced when the proportion of celery density was increased in the mixture (Fig. 4B). The response function could be well described with a rectangular hyperbola using Eq. [1], irrespective of the relative emergence time of S. vulgaris (r2 = 0.99). Parameter estimates are given in Table 3. Replacing two leek plants per square meter of the monoculture with one celery plant resulted in a 2.9% biomass reduction of S. vulgaris when planted together with the crops. This reduction percentage, represented by the initial slope of the hyperbolic curve, increased steadily for later dates of introduction, finally reaching 19.3% if S. vulgaris emerged 40 d after crop establishment (Table 3).
View this table:
[in this window]
[in a new window]
|
Table 3. Parameter estimates and standard errors for the hyperbolic function (Eq. [1]) describing the response of Senecio vulgaris biomass expressed relative to the biomass in leek monoculture to plant density of celery in a leek–celery intercropping system.
|
|
Isolines for Crop Quality, Yield, and Weed Biomass
Isolines for crop stands with equal quality were calculated using the hyperbolic competition model (Eq. [6]), which was fitted to the simulated data for leek (Fig. 5A)
and celery (Fig. 5B). For leek, the diameter of the pseudostem was used as a quality parameter, and isolines for diameters ranging between 15 and 30 mm were calculated. A minimum pseudostem diameter of 20 mm is required for marketable leek plants in many European countries (Brewster, 1994). For celery, isolines for the per-plant fresh mass are given. Market requirements range between 0.25 and 1 kg or more, with larger plants being used for industrial processing.
View larger version (28K):
[in this window]
[in a new window]
|
Fig. 5. Isolines for crop stands producing (A) leek with similar pseudostem diameters (mm), (B) celery with similar per-plant fresh mass (g), (C) similar leek yield (kg ha-1 fresh mass), and (D) similar celery yield (kg ha-1 fresh mass). For Compound Fig. 6C and 6D, the quality isoline for leek given by a minimum pseudostem diameter of 20 mm is included.
|
|
View larger version (26K):
[in this window]
[in a new window]
|
Fig. 6. Isolines for crop stands with (A) similar total biomass production (kg ha-1 dry matter), (B) similar total financial yield (€ ha-1), and (C) similar seed production with initial density of 50 Senecio vulgaris plants m-2 (seeds m-2). Compound Fig. 6D combines isolines for financial yield, minimum required leek pseudostem diameter, and seed production of S. vulgaris. The isoline for minimum required leek pseudostem diameter of 20 mm is also included in Compound Fig. 6A and 6B.
|
|
A second set of isolines indicates crop stands with equal yield levels for leek (Fig. 5C) and celery (Fig. 5D). For both crops, the slopes of the isolines differed sixfold if the yield level was tripled. In combining isolines for yield with the isoline for an acceptable leek quality, a solution space indicating crop stands with acceptable quality and high yields could be determined. Isolines for crop stands with equal total yield could be drawn by adding leek and celery yield (Fig. 6A)
. The highest biomass production was achieved with celery monocultures.
Financial rather than physical yield determines solutions with the highest economic value. Isolines for total financial yield were calculated using Eq. [7] and average prices
of
€ 0.35 kg-1
and
€ 0.19 kg-1
for leek and celery, respectively. By combining isolines for financial yield with the quality isoline for leek, the mixture with the highest financial yield could be determined (Fig. 6B). With a crop mixture of 9.4 and 19 plants m-2 for celery and leek, respectively, indicated by the point where the isoline for financial yield touches the leek quality isoline, a financial yield of
€ 27854
could be achieved. This yield was 7% higher than the maximum financial yield that could have been achieved with a leek monoculture and 9% higher than a maximum financial yield of a celery monoculture with a per-plant fresh mass of 730 g, which is equal to the per-plant fresh mass achieved in the optimum intercrop. Increasing the price for either leek or celery by 5% while keeping the price of the other crop constant resulted in a 2.5 and 3% increase of the financial yield for leek and celery, respectively. Decreasing the prices in the same way by 5% caused a financial yield reduction of 2 and 2.6% for leek and celery, respectively. The optimal leek and celery density was more sensitive to altering the leek price than to altering the celery price.
The effect of the cropping system on the reproductive potential of 50 S. vulgaris plants m-2, which were introduced 40 d after crop establishment, is shown by the isolines with equal production of S. vulgaris seeds per square meter in Fig 6C. The slope of the curves reflects the five to six times higher sensitivity of S. vulgaris to competition by celery compared with leek. To reduce the seed production of 50 initial S. vulgaris plants m-2 from 500 to 250 seeds m-2, a 2.7 times increase of plant densities was required in the crop stands. A similar effect was achieved when, for a given crop stand, the initial S. vulgaris density was reduced from 50 to 18.5 plants m-2. Seed production of S. vulgaris emerging 30 instead of 40 d after crop establishment was about 5.6 times higher. Hence, the extension of the weed-free period from 30 to 40 d reduced the seed production of S. vulgaris by 82%.
Combining isolines for financial yield, leek quality, and S. vulgaris seed production created a solution space including crop mixtures with high yield level, quality production, and high suppressive ability for S. vulgaris (Fig. 6D). The maximum financial yield did not coincide with highest suppressive ability. The latter could be further increased with increasing numbers of celery in the mixture, which, however, will cause a dramatic reduction of financial yield as quality criteria for leek will not be met anymore. Seed production of S. vulgaris could be reduced by 38% by growing a celery monoculture at density of 25 plants m-2, which would produce plants with a per-plant fresh mass of about 500 g and result in the same financial return as the highest-yielding crop mixture. The highest-yielding leek monoculture, on the other hand, resulted not only in an 7% lower financial return than the highest-yielding mixture, it also caused 35% higher seed production of S. vulgaris (Fig. 4D). Similar comparisons could be made for yield, quality, and levels of weed suppression between other crops stands.
|
DISCUSSION
|
|---|
Modeling Weed Growth in Monoculture and Intercrop Systems
Calibration of the model INTERCOM for two crops and one weed species demonstrated a large increase in complexity of interplant competitive relations when numbers of species considered increased from two to three. Morphological characteristics such as plant height and leaf area dynamics, which in the weed-free crop mixture proved to be determinant for competition (Baumann et al., 2002), were also critical for the simulation of weed suppression. Adaptations to the model had to be made with respect to early leaf area development, which is often temperature determined (Horie et al., 1979). The transition from sink- to source-limited simulation of leaf area development was erratic because transition from one state to the other is predetermined and abrupt. This might be improved by explicitly simulating sink and source size for each species independently, followed by determining the most limiting factor.
The model underestimated the biomass production of celery in the 1998 crop mixture (Exp. II; Fig. 1B), whereas the two other species were simulated accurately. This was possibly the result of a different response of the leaf morphology (e.g., higher specific leaf area) of celery if grown in mixture compared with monoculture. The model, parameterized for monoculture, was not able to account for these adaptations occurring in the mixture. Model performance was considered acceptable because the effect of S. vulgaris on the crops (Fig. 1) and, inversely, the effect of the crops on S. vulgaris (Fig. 2) was simulated correctly for the other crop stands in both years.
Competitive Relations among Leek, Celery, and S. vulgaris
Differences between the RCA of leek and celery found with model analysis confirmed results found in earlier experiments with a leek–celery intercropping system (Baumann et al., 2001a). Only small yield advantages were detected for crop mixtures if the relative yield total was calculated over a wide range of crop densities (Fig. 4A). In the simulated data sets, a near-balanced competitive relation was reached with a leek/celery ratio of about 2.0, which was also found in earlier experiments with a leek–celery intercropping system (Baumann et al., 2001a). Because the response of relative yield to mixing ratio depends on total density in a replacement design, it does not reflect the proper RCA of the crops if they are not grown at a density where the total yield reaches the asymptote on the density–yield response curve (Connolly, 1986). The NDI defined by Spitters (1983) reflects the true degree of niche differentiation. The NDI calculated for the weed-free crop mixture slightly exceeds unity, indicating complementarity in light capture between the crops. In earlier experiments, NDIs around 1.0 were found, and it was concluded that no complementarity in resource capture occurred between leek and celery (Baumann et al., 2001a).
Senecio vulgaris affected the competitive relation between leek and celery only when emerging at the time of crop establishment. Celery, which is more competitive than leek, was more affected by early emerging S. vulgaris (Fig. 3A). Probably due to transplanting shock and retarded early development (Rubatzky et al., 1999), celery was more susceptible to early weed competition than leek. The latter could profit from the reduced competitive ability of celery, which was reflected by the higher RCALC in the weedy situation compared with RCALC in the weed-free crop stand. This illustrates the complexity of mixtures with more than two species; competition relations between the first two species were mediated through the introduction of a third, and the degree of change was affected by the time of introduction of the third species. The initial size advantage of the transplanted crops resulted in a weak response of the crops to later-sown S. vulgaris, which had to germinate from seeds. Leek and celery showed a similar competitive advantage over S. vulgaris when the weed was sown at the same time the crop was planted. However, at later times of weed introduction, RCACS increased much faster than RCALS, reflecting the faster leaf area development of celery and its better ability to intercept light compared with leek. As a result, a negative response of S. vulgaris biomass to the proportion of celery in the mixture was found (Fig. 4B). Cousens (1985) demonstrated that a yield loss–weed density relationship could be well described using a rectangular hyperbolic function (Eq. [1]). In the current study, it was observed that this rectangular hyperbola could be equally well used to relate the reduction in S. vulgaris biomass to the proportion of celery in the crop mixture.
The degree of S. vulgaris biomass reduction was not only affected by celery density but, to an even larger extent, by the time of emergence of the weed relative to the crops (Fig. 4B). Weed density and the time between crop and weed emergence have been found earlier to be critical for predicting yield loss due to weed competition (Cousens, 1987; Kropff and Spitters, 1992; Kropff et al., 1984). Moreover, the relative time of emergence between crops and weeds is crucial for period thresholds, which predict when, rather than if, weeds need to be controlled to prevent yield and quality losses (Dawson, 1986). Period thresholds, however, are generally based on expected yield reduction of the current crop and do not account for seed production of late-emerging weeds, which may cause considerable problems in subsequent crops (Cousens and Mortimer, 1995). The experiments and simulation studies showed that the replacement of a few leek plants by celery in the crop stand contributed considerably to improving suppressive ability of the cropping system, particularly with respect to late-emerging weeds (Table 1). At the same time, leek yield and quality could be maintained (Fig. 5). Because of the improved competitive ability of the intercrop canopy, the critical period for weed control of the intercrop will be reduced compared with leek monoculture (Baumann et al., 2000). Though weeds emerging early in the season still require direct control measures, the current study indicates that it is likely that, compared with a leek monoculture, the number of required weed control treatments to obtain a successful weed control strategy will be lower in a leek–celery intercropping system.
Simulation runs with crop densities as used in practice showed that late-germinating S. vulgaris might still produce up to 1000 seeds m-2 in a leek monoculture (Fig. 6C). In crop mixtures with six celery plants, seed production could be reduced by 50% due to increased light competition. Although only S. vulgaris was considered in this study, similar effects of increased light competition have been found for other species, such as black nightshade (Solanum nigrum L.), lambsquarters (Chenopodium album L.), and barnyardgrass (Echinochloa crus-galli L.) (Lotz et al., 1993; Paolini et al., 1999; Schnieders, 1999).
Optimization of the Intercropping System
Insight into the competitive relations between crops and weed enabled the optimization of the system with respect to financial yield and weed suppression. Crop quality plays a predominant role as it is critical for the profitability of the system. For leek and celery, there is a strong response of quality parameters to intra- and interspecific competition (Baumann et al., 2001a). For celery, quality requirements depend on whether the produce is used for industrial processing, convenience food, or the fresh market. Leek pseudostem diameter proved to be the limiting factor for crop quality in the intercropping system. Therefore, crop mixtures represented by the isoline for leek plants with a pseudostem diameter of 20 mm delimit the solution space for profitable mixed stands (Fig. 5 and 6). Although high biomass yields can be achieved with high proportions of celery in the mixture (Fig. 6A), producing leek is more profitable because its price is higher than that of celery. A large yield gap was found between the calculated maximum financial yield and the yield level obtained with plant densities as used in practice where lower densities are usually planted to ensure high plant quality and enable efficient and labor-saving cultivation and harvesting. In particular, leek is generally grown at row distances between 0.5 and 0.75 m. For high plant densities (e.g., >30 plants m-2), in-row spacing would need to be between 4 and 6 cm, which would increase the plant-to-plant variability and result in a higher proportion of undersized plants (Brewster, 1994). Therefore, limitations for the spatial arrangement of the crop directed by the cultivation practices, as well as the use of below-optimal densities that meet the risk perception of the farmer, have to be taken into account.
Depending on whether the intercropping system is compared with a monoculture production of leek or celery, a double advantage or a trade-off between financial yield and weed suppression arises. For leek production, the yield advantage of an intercropping system is combined with a reduction of S. vulgaris seed production. If celery production is considered, a monoculture with the same yield as a mixture suppresses S. vulgaris better (Fig. 6D). In this study, leek was the crop of interest due to its economic potential in many European countries and weak competitive ability against weeds. It was shown that high quality leek can be produced at a high yield level in an intercropping system with celery, which at the same time, has distinct advantages with respect to weed suppression.
|
CONCLUSIONS
|
|---|
A combined approach using mechanistic and descriptive models for analyses and optimization of an intercropping system of leek and celery proved to be very effective. With a new version of the model INTERCOM, biomass production, product quality, and weed seed production for monocultures and mixtures could accurately be simulated. Application of a descriptive regression model for summarizing the simulation results was very effective and facilitated optimization of the intercropping system. It is concluded that this combined modeling approach enlarges the potential of mechanistic crop growth and competition modeling to be used in the optimization and design of cropping systems.
|
ACKNOWLEDGMENTS
|
|---|
We thank MeteoSwiss for providing the weather data and Dr. W. van der Werf for his mathematical advise. The research was founded by the Swiss Federal Office for Agriculture.
|
NOTES
|
|---|
Research financed by The Swiss Federal Office for Agriculture.
|
REFERENCES
|
|---|
- Baumann, D.T., L. Bastiaans, J. Goudriaan, H.H. Van Laar, and M.J. Kropff. 2002. Analysing crop yield and plant quality in an intercropping system using an eco-physiological model for interplant competition. Agric. Syst. (in press).
- Baumann, D.T., L. Bastiaans, and M.J. Kropff. 2001a. Competition and crop performance in a leek–celery intercropping system. Crop Sci. 41:764–774.[Abstract/Free Full Text]
- Baumann, D.T., L. Bastiaans, and M.J. Kropff. 2001b. Effects of intercropping on growth and reproductive capacity of late emerging Senecio vulgaris L., with special reference to competition for light. Ann. Bot. 87:209–217.[Abstract/Free Full Text]
- Baumann, D.T., M.J. Kropff, and L. Bastiaans. 2000. Intercropping leeks to suppress weeds. Weed Res. 40:359–374.
- Brewster, J.L. 1994. Onions and other vegetable alliums. CAB Int., Wallingford, Oxon, UK.
- Caporali, F., E. Campiglia, R. Paolini, and R. Mancinelli. 1998. The effect of crop species, nitrogen fertilization and weeds on winter cereal/pea intercropping. Ital. J. Agron. 2:1–9.
- Connolly, J. 1986. On difficulties with replacement-series methodology in mixture experiments. J. Appl. Ecol. 23:125–137.
- Cousens, R. 1985. An empirical model relating crop yield to weed and crop density and a statistical comparison with other models. J. Agric. Sci. (Cambridge) 105:513–521.
- Cousens, R. 1987. Theory and reality of weed control thresholds. Plant Prot. Q. 2:13–20.
- Cousens, R., and M. Mortimer. 1995. Dynamics of weed populations. Cambridge Univ. Press, Cambridge, UK.
- Dawson, J.H. 1986. The concept of period thresholds. p. 327–331. In Proc. Eur. Weed Res. Soc. Symp. on Econ. Weed Control, Stuttgart, Germany. 12–14 Mar. 1986. Eur. Weed Res. Soc., Stuttgart-Hohenheim, Germany.
- De Wit, C.T. 1960. On competition. Versl. Landbouwkd. Onderz. 66:1–82.
- Horie, T., C. de Wit, J. Goudriaan, and J. Bensink. 1979. A formal template for the development of cucumber and its vegetative stage. Proc. K. Ned. Akad. Wet., Ser. C 82:433–479.
- Itulya, F.M., and J.N. Aguyoh. 1998. The effects of intercropping kale with beans on yield and suppression of redroot pigweed under high altitude conditions in Kenya. Exp. Agric. 34:171–176.
- Kira, T., H. Ogawa, and N. Sakazaki. 1953. Intraspecific competition among higher plants: I. Competition–yield–density interrelationships in regularly dispersed populations. J. Inst. Polytech., Osaka City Univ., Ser. D 4:1–16.
- Kropff, M.J., and C.J.T. Spitters. 1992. An eco-physiological model for interspecific competition, applied to the influence of Chenopodium album L. on sugar beet: I. Model description and parameterization. Weed Res. 32:437–450.
- Kropff, M.J., and H.H. Van Laar. 1993. MODELING crop–weed interactions CAB Int., Wallingford, Oxon, UK.
- Kropff, M.J., F.J.H. Vossen, C.J.T. Spitters, and W. de Groot. 1984. Competition between a maize crop and a natural population of Echinochloa crus-galli (L.). Neth. J. Agric. Sci. 32:324–327.
- Liebman, M., and A.S. Davis. 2000. Integration of soil, crop and weed management in low-external-input farming systems. Weed Res. 40:27–47.
- Liebman, M., and E. Dyck. 1993. Crop rotation and intercropping strategies for weed management. Ecol. Applic. 3:92–122.
- Liebman, M., and T. Ohno. 1998. Crop rotation and legume residue effects on weed emergence and growth: Implications for weed management. p. 181–221. In J.L. Hatfield et al. (ed.) Integrated weed and soil management. Ann Arbor Press, Chelsea, MI.
- Lotz, L.A.P., R.W.M. Groeneveld, and B.J. Schnieders. 1993. Evaluation of the population dynamics of annual weeds to test integrated weed management at a farming system level. Landscape Urban Plann. 27:185–189.
- Paolini, R., M. Principi, R.J. Froud-Williams, S. Del Puglia, O. Barcellona, and E. Biancardi. 1999. Competition between sugarbeet and Sinapis arvensis and Chenopodium album, as affected by timing of nitrogen fertilization. Weed Res. 39:425–440.
- Rana, K.S., and M. Pal. 1999. Effect of intercropping systems and weed control on crop–weed competition and grain yield of pigeonpea. Crop Res. 17:179–182.
- Rubatzky, V.E., C.F. Quiros, and P.W. Simon. 1999. Carrots and related vegetable Umbelliferae. CABI Int., Wallingford, Oxon, UK.
- Schnieders, B.J. 1999. A quantitative analysis of inter-specific competition in crops with a row structure. Ph.D. thesis. Wageningen Univ., Wageningen, the Netherlands.
- Schoofs, A., and M.H. Entz. 2000. Influence of annual forages on weed dynamics in a cropping system. Can. J. Plant Sci. 80:187–198.
- Shinozaki, K., and T. Kira. 1956. Intraspecific competition among higher plants: VII. Logistic theory of the C-D effect. J. Inst. Polytech., Osaka City Univ., Ser. D 7:35–72.
- Spigt, R.M., and T.L.J. Janssen. 1997. Kwantitatieve informatie akkerbouw en vollegrondsgroenteteelt. Praktijkonderzoek voor de akkerbouw en de vollegrondsgroenteteelt, Lelystad, the Netherlands.
- Spitters, C.J.T. 1983. An alternative approach to the analysis of mixed cropping experiments: 1. Estimation of competition effects. Neth. J. Agric. Sci. 31:1–11.
- Teasdale, J.R. 1998. Cover plants, smother plants, and weed management. p. 247–270. In J.L. Hatfield et al. (ed.) Integrated weed and soil management. Ann Arbor Press, Chelsea, MI.
- Watkinson, A.R. 1981. Interference on pure and mixed populations of Agrostemma githago. J. Appl. Ecol. 18:967–976.
- Wright, A.J. 1981. The analysis of yield–density relationships in binary mixtures using inverse polynomials. J. Agric. Sci. (Cambridge) 96:561–567.
This article has been cited by other articles:
|
|
|
|
 
C. Shennan
Biotic interactions, ecological knowledge and agriculture
Phil Trans R Soc B,
February 27, 2008;
363(1492):
717 - 739.
[Abstract]
[Full Text]
[PDF]
|
|
|
TOP
ABSTRACT
INTRODUCTION
), leek–celery intercrop (
), and celery monoculture (
) for (A) three crop densities [leek/celery = 40:20 (—), 20:10 (---), and 10:5 (···) plants m-2; and (B) one crop density (leek/celery = 20:10 plants m-2); and the yield of Senecio vulgaris grown in crop mixture expressed relative to its yield in leek monoculture for various relative times of emergence (0, 10, 20, 30, and 40 d after crop establishment).