Intercropping—Towards an Understanding of the Productivity and Profitability of Dryland Crop Mixtures in Southern Australia

Abstract

Intercropping using mixtures of dryland crop species for grain or seed production was investigated in southern Australia across a range of rainfall zones over three years. The objective was to understand the productivity and profitability of intercropping in extensive, high-input grain cropping systems. Previous research has shown large productivity benefits of mixtures; however, few farmers practice intercropping in Australia, and an analysis of profitability is needed to support future potential adoption. Experimental results showed strong mixture responses (in terms of yield, value and land equivalence), but not all were profitable compared to an equivalent share of monoculture crops (as measured by gross margins). The most promising mixtures were those containing high-value crops (canola) and legumes (field pea or faba bean) at the wetter sites where the additional gross margin over equivalent monoculture crops ranged from $12/ha to $576/ha. Mixtures containing highly competitive crops (wheat or barley) were generally unprofitable. Mixtures involving cereals were doubly disadvantaged by the aggressiveness of these lower-value crops in the mixtures we examined and the high grain separation costs post-harvest. Cost reduction in mixture systems involving high-value crops that are synergistic (grain legumes) should provide enduring opportunities for intercropping in southern Australia.

1. Introduction

Growing different crop species or varieties in mixtures and various combinations over space (mixtures) or time (relay cropping), generally referred to as intercropping, has been shown to offer a myriad of productivity benefits to the farm business. These are postulated to include enhancing nutrient, radiation and water use efficiencies that increase crop yields and profits within the growing season [1,2,3,4,5,6]. Other within-season benefits include reductions in pest and disease infestation (and hence chemical usage) [7,8], improved quality (hence grain price for human consumption or nutritional value of stock-feed) [9], and increased nutrient availability in-season and as carryover (e.g., nitrogen fixed by legumes included in the mix) [10,11]. In the long-term, there are the expected environmental benefits from building and turnover of soil organic matter [12] and general social benefits [7]. In the longer term, there is also the potential for reduced season-to-season yield variability, as one crop component may persist better under adverse weather conditions, though this benefit has been noted with reference to small landholders in developing countries [13].

Potential constraints to intercropping include increased system complexity and logistical challenges such as sowing, weed control, harvesting, and post-harvest grain separation [14,15,16], all of which increase costs. These constraints have the potential to resolve over time due to the uptake of advances like precision agriculture, modification of farm machinery to suit intercropping, robotics, and herbicide-tolerant crops. Some of these practical considerations are avoided if the intercrop is grown for forage rather than grain.

This paper has a strong focus on the economic and productivity benefits of intercropping. This focus is because intercropping systems have not been widely adopted by growers within modern, mechanised grain cropping systems that rely on high inputs in countries such as Australia [17], yet relative economic advantage is known to be a key predictor of how growers respond to a new practice or technology [18].

We hypothesised that there are annual crop mixtures that growers can experimentally test and profit from in the short term. Consequently, the objectives of this paper were to (1) report on yield data from replicated field trials conducted in the medium and high rainfall cropping regions of southern Australia and (2) use these data in suitable evaluation metrics to identify these mixtures and provide guidance to what intercropping systems need further research and validation by farmers.

Assessing the physical intercropping response and the relative economic advantage of intercropping is more complex than assessing that of monocultures, as comparisons are not directly observable and require some indirect algebraic metrics. Several mixture metrics are available, the Land Equivalent Ratio (LER) being the most used [19]. Other metrics that we were specifically interested in were the aggressivity index (AI) [20], the yield ratio (YR), the value ratio (VR) and the net gross margin (NetGM) [21]. The underlying basis is always a comparison of the outputs of the intercrop relative to the monocultures of the constituent crops.

A comprehensive series of winter cropping field experiments were undertaken across three years of wet and dry seasons and three representative rainfall zones of Victoria in southern Australia to quantify the relative yield responses of intercrops compared to monocultures of the constituent crops. Dryland farming in southern Australia is characterised by winter rainfall without irrigation that restricts a mixture of species to those adapted to cool winters where phenological development is generally synchronised. Crops of interest included cereal, legume and oilseed species with compatible herbicide regimes and sowing and harvest dates. Previously, very large apparent in-season yield gains have been reported, particularly with legume and oilseed mixtures in southern Australia [22,23,24]. With cereals dominating cropping in the region (wheat alone occupies over 50% of total cropland), the research represents increased complexity without intensification and increased legume and oilseed frequency in rotations.

Our research focused on the productivity and profitability benefits of intercropping under dryland cropping systems in southern Australia in the year of production. It was not possible to test all the possible system options. Production risk, as measured by long-run yield variance [25], was not considered. Also, the yield benefit of break crops like canola to subsequent cereal crops and other rotational benefits (any surplus N carryover from legumes) were not factored into the analysis. Societal benefits such as climate regulation (carbon sequestration) and increased biodiversity (crop richness, habitat for above and below-ground microorganisms) were not considered in this study.

2. Materials and Methods

2.1. Metrics of Intercropping Productivity and Profitability

We tested the likely extent of intercropping benefits by advancing the LER analysis through AI, YR, and VR to NetGM (

Table 1. Physical and economic evaluation metrics used in the evaluation of productivity and profitability. Metrics are for mixtures of two-component crops.

Metric 1	Data Required	Decision Criteria
Land Equivalent Ratio (LER) $\mathrm{L}\mathrm{E}\mathrm{R}=\frac{{\mathrm{Y}1}_{\mathrm{c}}}{{\mathrm{Y}1}_{\mathrm{m}}}+\frac{{\mathrm{Y}2}_{\mathrm{c}}}{{\mathrm{Y}2}_{\mathrm{m}}}$	Yields for crop components and monocultures.	LER > 1 indicates a positive intercropping response.
Aggressivity Index (AI) $\mathrm{A}\mathrm{I}=\frac{{\mathrm{Y}1}_{\mathrm{c}}}{{{\mathrm{Z}1}_{\mathrm{c}}\ast \mathrm{Y}1}_{\mathrm{m}}}-\frac{{\mathrm{Y}2}_{\mathrm{c}}}{{\mathrm{Z}2}_{\mathrm{c}}\ast {\mathrm{Y}2}_{\mathrm{m}}}$	Yields for crop components and monocultures and the proportional sown area for each crop component.	AI > 0 indicates crop 1 is dominant. AI < 0 indicates crop 2 is dominant. −1 < AI < 1 indicates compatible species. AI > 1 indicates crop 1 is aggressive. AI < −1 indicates crop 2 is aggressive in the mix.
Yield Ratio (YR) $\mathrm{Y}\mathrm{R}=\frac{({\mathrm{Y}1}_{\mathrm{c}}+{\mathrm{Y}2}_{\mathrm{c}})}{({\mathrm{Z}1}_{\mathrm{c}}\ast {\mathrm{Y}1}_{\mathrm{m}}+{\mathrm{Z}2}_{\mathrm{c}}\ast {\mathrm{Y}2}_{\mathrm{m}})}$	Yields for crop components and monocultures and the proportional sown area for each crop component.	YR > 1 indicates an intercropping advantage.
Value Ratio (VR) $\mathrm{V}\mathrm{R}=\frac{({\mathrm{Y}1}_{\mathrm{c}}\ast \mathrm{P}1+{\mathrm{Y}2}_{\mathrm{c}}\ast \mathrm{P}2)}{({{\mathrm{Z}1}_{\mathrm{c}}\ast \mathrm{Y}1}_{\mathrm{m}}\ast \mathrm{P}1+{\mathrm{Z}2}_{\mathrm{c}}{\ast \mathrm{Y}2}_{\mathrm{m}}\ast \mathrm{P}2)}$	Yields for crop components and monocultures, unit prices and the proportional sown area for each crop component.	VR > 1 indicates an intercropping advantage.
Net Gross Margin (NetGM, $/ha) $\mathrm{N}\mathrm{e}\mathrm{t}\mathrm{G}\mathrm{M}={\mathrm{G}\mathrm{M}}_{\mathrm{c}}-{\mathrm{G}\mathrm{M}}_{\mathrm{m}}$ where, $\begin{array}{cc}\hfill {\mathrm{G}\mathrm{M}}_{\mathrm{c}}& =\left({\mathrm{Y}1}_{\mathrm{c}}\ast \mathrm{P}1+{\mathrm{Y}2}_{\mathrm{c}}\ast \mathrm{P}2\right)-{\mathrm{C}}_3\hfill \\ \hfill {\mathrm{G}\mathrm{M}}_{\mathrm{m}}& =\{{\mathrm{Z}1}_{\mathrm{C}}\ast {(\mathrm{Y}1}_{\mathrm{m}}\ast \mathrm{P}1-\mathrm{C}1)\}+\hfill \\ & \{{\mathrm{Z}2}_{\mathrm{C}}\ast {(\mathrm{Y}2}_{\mathrm{m}}\ast \mathrm{P}2-\mathrm{C}2)\}\end{array}$	Yields for crop components and monocultures, the unit prices and the proportional sown area for each crop component, and activity variable costs.	NetGM > 0 indicates an intercropping advantage.

¹ Where Y1_c or Y2_c = Yield of component crops 1 or 2 in a mixture, Y1_m or Y2_m = Yield of crop 1 or 2 as a monoculture, GM_c = Gross Margin from intercropping, GM_m = Gross Margin from monoculture with same enterprise mix as in the mixture, P1 and P2 are the expected market prices of crop 1 and 2, C1, C2 and C3 are the variable costs of production for crop 1 and crop 2 and mixtures respectively, Z1_c and Z2_c = proportional sown area of crop 1 and 2 in the mixture.

).

The LER is commonly cited in the literature and was included for consistency with previously published work. The LER is a relative measure of the yields for the mixtures compared to those of the sole crops. It describes the additional land needed to grow the same quantity of both crops if they were grown as monocultures rather than as intercrops. Implicitly, it assumes a 50:50 mix ratio. When LER < 1, the intercropping system has a disadvantage in land use compared to the monocultures; when the LER > 1, there is a land-use advantage in intercropping. For example, an LER of 1.15 requires 15% more hectares when grown as monocultures to produce the equivalent yield as the mixture.

While the LER is a simple metric to calculate and interpret, it has limitations that the AI, YR, VR and NetGM overcome. This is because, unlike the LER, the other four metrics consider absolute yields and the proportion of each crop in the mix.

The Aggressivity Index (AI) is a measure of how much relative yield increase in crop 1 is greater than crop 2 in a mixture adjusted for enterprise mix ratio. It is a variation of the LER that highlights which crop is dominating the intercropping response and is useful for investigators interested in the yield of a particular crop in the mixture or wanting to understand the reason for any intercropping response (competition and/or facilitation [26].

For large-scale, commercialised agriculture, the YR, VR and NetGM are particularly important metrics that account for total yield. Additionally, the VR and NetGM consider relative crop prices, and the NetGM considers the change in variable costs for growing an intercrop compared to an equivalent mix of monocultures. The latter is the most comprehensive measure of the change in economic efficiency. When NetGM < 0, the intercropping system has a profit disadvantage compared to the monocultures. In contrast, when NetGM > 0, intercropping has a profit advantage.

2.2. Field Experiments

Monoculture and intercrop yields were obtained from replicated field experiments established in 2019 at three core sites (Hamilton, Horsham, and Rutherglen), where mostly the same treatments were applied. These core sites reflect the medium (400 to 500 mm annual rainfall) and high (600 to 1000 mm annual rainfall) rainfall zones of southern Australia. In 2020 and 2021, nearby sites were established, providing three full winter cropping seasons. In 2020 and 2021, we also established a smaller set of replicated experiments at satellite sites at Netherby, Curyo, Streatham, Wallup, Watchupga, Willaura, Inverleigh, Burramine, Dookie and Caniambo aimed to widen the mixture choices of interest to local farmers. These satellite sites represented a wider set of rainfall environments extending well into the low rainfall zone (300 to 400 mm annual rainfall). Figure 1 shows the location of all the experimental sites with details in

Table A1. Nineteen experimental sites were sown over three years, showing location, elevation, row spacing and time of sowing (TOS).

Trial	Latitude (d m s)	Longitude (d m s)	Elevation (m)	Row Spacing (m)	TOS
Hamilton core	−37 49 27	142 04 08	205	0.15	27 June 2019
Hamilton core	−37 49 27	142 04 08	205	0.20	19 May 2020
Hamilton core	−37 49 27	142 04 08	205	0.20	24–26 May 2021
Horsham core	−36 33 27	142 03 35	130	0.30	5 June 2019
Horsham core	−36 44 40	142 06 50	130	0.20	13 May 2020
Horsham core	−36 44 07	142 06 57	135	0.20	1–2 June 2021
Rutherglen core	−36 06 00	146 31 04	150	0.15	4 June 2019
Rutherglen core	−36 06 00	146 31 04	150	0.25	7 May 2020
Rutherglen core	−36 06 00	146 31 04	150	0.25	14 May 2021
Willaura satellite	−37 31 55	142 49 08	245	0.20	1 May 2020
Inverleigh satellite	−38 05 02	143 56 17	105	0.20	27 May 2020
Caniambo satellite	−36 26 40	145 35 48	130	0.23	13 May 2020
Burramine satellite	−36 03 11	145 52 27	120	0.23	29 May 2020
Curyo satellite	−35 50 57	142 42 02	90	0.30	28 May 2020
Netherby satellite	−36 01 13	141 44 33	100	0.30	25 May 2020
Dookie satellite	−36 23 36	145 42 50	145	0.23	14 May 2021
Burramine satellite	−36 03 11	145 52 27	120	0.23	18 May 2021
Wallup satellite	−36 21 00	142 12 00	125	0.30	1 June 2021
Watchupga satellite	−35 45 35	142 42 28	85	0.30	1 June 2021

.

Five crop species, wheat, barley, canola, faba bean, field pea and lentil, were sown both as a monoculture (100%) and in a mixture with another crop at the core sites with other mixtures of local interest.

Crop combinations at the core sites included:

• Barley (cv. Spartacus CL) + Canola (cv. Hyola 580 CT)
• Faba bean (cv. PBA Bendoc) + Canola (cv. Hyola 580 CT)
• Field pea (cv. BPA Butler) + Canola (cv. Hyola 580 CT)
• Faba bean + Wheat (cv. Sheriff CL) in the high rainfall zone only
• Lentil (cv. Hallmark XT) + Wheat (cv. Sheriff CL) in the medium rainfall zone only.

Combinations at the satellite sites included:

• Barley (cv. Spartacus CL) + Canola (cv. Hyola 580 CT)
• Faba bean (cv. PBA Bendoc) + Canola (cv. Hyola 580 CT)
• Field pea (cv. PBA Butler) + Canola (cv. Hyola 580 CT)
• Faba bean + Wheat (cv. Sheriff CL) in the HRZ only
• Lentil (cv. Hallmark XT) + Wheat (cv. Sheriff CL) in the MRZ only.

Cultivars were obtained from commercial sources, and the seeds were treated to protect them from pests and disease. Commercial peat inoculant Group F, strain WSM 1455 (Rhizobium leguminosarum bv. Viciae) was applied to faba bean, field pea, and lentil seed as a slurry (peat water mix) at double the commercial rates: 5 g/kg peat inoculant for faba bean and field pea and 10 g/kg seed for lentils.

Barley, faba bean, canola lentil and wheat varieties grown were tolerant of the broadleaf weed herbicide imidazolinone, while the canola variety chosen was also tolerant to the triazine group of herbicides, and the field pea variety chosen was tolerant of the broadleaf weed herbicide; imazamox. Each treatment received herbicide combinations appropriate for the crop species (turbuthylazine, carfentrazone, imazamox, imazapyr, trifluralin, glyphosate, clethodim and haloxyfop). Fungicides applied during the growing season were tebuconazole, azoxystrobin, bixafen and prothiocorazole. Starting fertiliser (100 kg/ha monoammonium phosphate) was applied across all treatments with no additional in-crop fertilisation.

The field plot layout at each site was a completely randomised block design with monoculture and mixtures as main plots (range 10–20 (length) × 4–6 (width) m) comprising four replications with monoculture target densities listed in

Table 2. Target mixture densities for monocultures at core sites (plants/m²).

Crop	Wheat	Barley	Canola	Faba Bean	Field Pea	Lentil
Cultivar	Sheriff CL	Sparticus CL	Hyola580 CT	PBA Bendoc	PBA Butler	Hallmark XT
Hamilton a		180	60	40	70
Horsham b	160	160	50	30	55	120
Rutherglen c		170	60	35	60

^a. Same planting densities at Hamilton satellite sites: Willaura, Inverleigh and Streatham. ^b. Same planting densities at Horsham satellite sites: Wallup, Netherby, Curyo, and Watchupga. ^c. Same planting densities at Rutherglen satellite sites: Burramine, Dookie, Caniambio.

. The mixtures studied were full-season (synchronised) intercropping, where the two crop components were planted together as a mix and harvested at the same time. Planting density was based on a ‘proportional’ design to achieve 100% absolute monoculture density for each mixture in the proportions 25:75, 50:50 and 75:25 so that the relative density of each component was the same as when grown in a pure stand. Crops were sown as a ‘mixed stand’ for all mixture proportions, where seeds of both species were planted in the same drill row that were fed from separate seed boxes into each drill row with row spacings ranging from 0.15 to 0.3 m (Table A1) in the nominated proportions across sites. For the 50:50 mixture proportion, an additional ‘skip-row’ treatment was included, where individual species were assigned to their own drill row. In this case, paired rows of each species were established across the width of the plot without changing the row spacing or overall target plant density.

2.3. Data

When applying the intercropping metrics to establish advantages or disadvantages, it is implicit that the monoculture yields are representative of the surrounding area. In small plot experiments, artefacts can affect monoculture yields disproportionately to those crops in the mixture (e.g., mis-sown seed, site variability). An unusually low monoculture yield would produce an unusually high intercropping metric. To reduce the risk of lower monoculture yields, we used the site mean yield of all monocultures. Yields used for the intercrops were averages of four replicates.

Representative crop prices and base guides for variable input costs (fertilisers, herbicides, pesticides, seed and farm operations) were obtained from an authoritative farm gross margin guide for cropping systems in the southern cropping regions of Australia [27]. Post-harvest cleaning costs were obtained from seed suppliers in northern Victoria [28].

Grain prices vary over time and can be volatile from year to year, so grain prices were averages for the ten years to February 2022. Nominal crop prices were adjusted to 2022 equivalent dollars using ABARES’ producer price index [29] and were net of charges based on the farm-gate value of the grain (endpoint royalties, insurance, and grower research levies). Real commodity prices for the 10-year period ending 2022 are shown in Figure 2. Cereal prices (wheat and barley) are more stable and relatively low, with wheat averaging $330/t and feed barley at $278/t. Canola prices are also relatively stable but higher at $626/t. Legumes prices are the most volatile and relatively high, averaging $470/t, $535/t and $782/t for faba bean, field pea and lentil, respectively. The greatest differences between the prices of crop components were for lentil-wheat ($452/t) and barley-canola mixtures ($348/t). The dominance of the higher-value crop in a mixture could enhance the profitability of the whole mixture compared to single stands of the monocultures.

Over the three experimental years, we adapted the base-line grower variable costs with input from local farm management consultants and farmers. It should be noted that in constructing these gross margins, fixed (overhead) costs were ignored, as it is considered that they will be incurred regardless of the level of the enterprise undertaken. It is acknowledged that the gross margins applied in this analysis are indicative only, as there is considerable heterogeneity between different farm businesses in capital resources already on the property and inputs used. Costs were reported in either $/ha or $/t; those denominated on a weight basis were converted to $/ha using the relevant experimental yield. Variable costs for the monocrops are shown in

Table 3. Variable costs for monocrops: Hamilton.

Monoculture	Area-Based Items ($/ha)									Yield-Based Items ($/t)
	Seed	Sowing	Spraying	Harvest	Fungicides	Insecticides	Herbicides	MAP 1	SOA 1	Grain cartage	Cleaning 2
Barley	62	50	65	80	27	0	49	48		25	35
Canola	63	50	50	90	45	15	53	48	43	25	35
Faba bean	95	50	60	90	39	63	42	48		25	35
Field pea	68	50	60	90	2	4	44	48		25	35
Wheat	55	50	65	80	35	0	57	48		25	35

¹ MAP is monoammonium phosphate and SOA is sulfate of ammonia. Costs include delivery and application. ² Cleaning is to “grower-dressed” standard.

,

Table 4. Variable costs for monocrops: Horsham.

Monoculture	Area-Based Items ($/ha)									Yield-Based Items ($/t)
	Seed	Sowing	Spraying	Harvest	Fungicides	Insecticides	Herbicides	MAP 1	SOA 1	Grain cartage	Cleaning 2
Barley	58	50	50	65	21	0	49	36		25	35
Canola	80	50	50	80	45	15	53	36	21	25	35
Faba bean	80	50	60	80	39	2	42	36		25	35
Field pea	69	50	60	80	2	3	50	36		25	35
Lentil	41	50	60	80	30	3	41	36		25	35
Wheat	44	50	60	65	35	0	57	36		25	35

¹ MAP is monoammonium phosphate and SOA is sulfate of ammonia. Costs include delivery and application. ² Cleaning is to “grower-dressed” standard.

and

Table 5. Variable costs for monocrops: Rutherglen.

Monoculture	Area-Based Items ($/ha)									Yield-Based Items ($/t)
	Seed	Sowing	Spraying	Harvest	Fungicides	Insecticides	Herbicides	MAP 1	SOA 1	Grain cartage	Cleaning 2
Barley	62	50	60	65	27	0	49	48		25	35
Canola	63	50	50	80	45	15	53	48	40	25	35
Faba bean	95	50	60	80	39	2	42	48		25	35
Field pea	68	50	60	80	2	4	44	48		25	35
Wheat	55	50	60	65	35	0	57	48		25	35

¹ MAP is monoammonium phosphate and SOA is sulfate of ammonia. Costs include delivery and application. ² Cleaning is to “grower-dressed” standard.

.

Changes in the variable costs for the mixtures compared to the monocultures and the rationale for the changes were as follows:

• Seed costs were the average of the monocultures weighted by the proportion of each crop in the mixture.
• Additional urea was not used in the field experiments or included in the costs. The cost for other fertilisers (monoammonium phosphate and sulfate of ammonia) is the average of fertiliser costs for the monocultures (including delivery and spreading) weighted by the proportion of each crop in the mixture.
• Imidazolinone herbicides were used exclusively on intercrops, but costs were assumed to be similar to those for monocultures.
• Fungicides were the maximum rate for the crop components in the mixture. However, one fewer spray application was assumed for the legume mixtures due to lower disease incidence.
• Insecticides were the maximum rate for the crop components in the mixture.
• Crop components were harvested together without any assumed logistic difficulty. However, contract harvest rates were higher for intercrops, reflecting slower work rates to avoid (for example) shattering of field peas.
• Sorting for intercrops ranges from easy and thus inexpensive to extremely difficult and thus prohibitively expensive. For example, field pea and canola are easily separated based on size using a set of screens; at the other extreme, lentil and wheat have the same bulk density and size and require colour sorting. Sorting costs for field pea or faba bean + canola was set at $85/t; faba bean + wheat: $120/t; barley + canola: $200/t; and lentil + wheat: $250/t.

For mixtures marketed for human consumption, post-harvest seed separation costs are potentially a major additional cost impost compared to monocultures. These costs are imposed on a $/t basis, so the more the intercrop out-performs compared to the monocultures of the component crops, the greater the additional cost penalty.

3. Results

3.1. Monoculture Yields and Gross Margins

The grain yield of monoculture plots varied significantly across both season and environment. The yield of cereal crops (wheat and barley) ranged from 2.1 to 8.7 t/ha (Figure 3a) and was positively correlated with annual rainfall. However, this relationship was not entirely consistent (R² 0.35–0.42), suggesting inefficiency of resource conversion driven by alternative limiting factors (examples include low wheat yield despite high rainfall at Inverleigh in 2020 and Dookie in 2021; Appendix B, Appendix C, Appendix D, Appendix E and Appendix F).

The relationship between annual rainfall and grain yield was stronger for broadleaf crops. Across all sites, canola yield ranged from 0.8 to 4.6 t/ha, while lentil yield (grown exclusively in low/medium rainfall environments) ranged from 1.1 to 4.8 t/ha (Figure 3b). Faba bean and field pea were the highest-yielding broadleaf crops, ranging from 1.0 to 7.6 t/ha (Figure 3c). The weak relationship between grain yield and annual rainfall indicates inefficiencies within the cropping system that could be addressed by planting multiple compatible crops to better capture and utilise available resources. Based on the range of grain yields measured, this study provides a comprehensive dataset testing the potential benefits of intercropping under a range of productivity scenarios and seasons.

Indicative gross margins at the core sites for each monoculture, together with crop yields and activity variable costs used in the calculation, are shown in

Table 6. Total revenue, activity variable costs and gross margins for monocultures at the core sites based on treatment yields for 2019 to 2021 (indicative ¹).

Site	Monoculture	Yield (t/ha)	Crop Price ($/t)	Gross Income ($/ha)	Activity Variable Costs ($/ha)	Gross Margin ($/ha)
Hamilton	Barley	8.0	278	2215	860	1355
	Canola	3.8	626	2407	687	1813
	Faba bean	6.4	534	3405	869	2535
	Field pea	6.1	470	2883	734	2149
	Wheat	7.7	330	2526	850	1676
Horsham	Barley	4.4	278	1233	596	637
	Canola	2.8	626	1779	601	1178
	Faba bean	5.3	534	2853	710	2143
	Field pea	3.2	470	1490	540	950
	Lentil	3.6	782	2846	558	2288
	Wheat	6.1	330	2006	712	1293
Rutherglen	Barley	7.3	278	2018	798	1220
	Canola	2.3	626	1466	584	882
	Faba bean	4.6	534	2445	690	1755
	Field pea	4.4	470	2082	622	1460
	Wheat	6.7	330	2211	773	1439

¹ Gross income may not equal yield multiplied by price due to rounding. Similarly, the gross margin may not equal gross income minus the variable costs due to rounding.

. Averages over the three years are shown for simplicity. Standard errors for yields and all the intercropping metrics for each year and all sites and treatments (including the monocultures) are shown in Appendix B, Appendix C, Appendix D, Appendix E and Appendix F.

Reflecting long-term average price levels, gross margins for the monocultures were generally lower for cereals (barley and wheat) compared to crops with higher (canola) and more volatile (legumes) average returns over the three years (2019 to 2021) of the experiment (Table 6). Barley, for example, had gross margin returns of $1355/ha, $637/ha, and $1220/ha at Hamilton, Horsham, and Rutherglen, respectively. By contrast, faba bean had gross margin returns of $2535/ha, $2143/ha, and $1755/ha at Hamilton, Horsham, and Rutherglen, respectively.

There were also substantial differences across the core sites in the gross margins due to differences in rainfall and inherent crop productivity. For example, canola performed relatively poorly at Rutherglen, where a 2.3 t/ha 3-year average yield translated to lower gross income of $1466/ha and a gross margin return of $882/ha. By contrast, the gross margin for the average 3.8 t/ha canola monoculture at Hamilton was $1813/ha.

3.2. Benefits of Intercropping from Biological and Economic Perspectives

The benefits of intercropping in terms of land productivity, as measured by the LER, were highly dependent on the crops being grown together and seasonal conditions. An LER greater than 1 occurred in 71% of the observed dataset for combinations of faba bean and canola, 70% for field pea and canola, and 62% for faba bean and wheat (Figure 4). By contrast, beneficial LER values for combinations of lentil and wheat in a low/medium rainfall environment only occurred in 50% of cases, while combinations of barley and canola only resulted in LER > 1 for 38% of the time (across all environments). Nonetheless, highly negative responses to intercropping were rare, with most observations resulting in LER > 0.8, indicating limited downside production risk. Rainfall was a key determinant of the likelihood of an LER being above 1. Where rainfall was in the lowest tercile of observations, LER > 1 occurred in 39% of observations. Conversely, LER > 1 occurred in 50% and 75% of observations at sites where rainfall was in the middle and upper seasonal rainfall terciles.

Assessment of mixture aggressiveness (AI) indicated that combinations of broadleaf crops were more likely to result in complementary responses between the two crops (defined here as AI values between −1 and 1) (Figure 4). For combinations of faba bean and canola, complimentary responses occurred in 78% of cases, while for combinations of field pea and canola, this occurred in 74% of cases. Conversely, these values were 64, 62 and 17% for combinations of barley and canola, faba bean and wheat, and lentil and wheat, respectively. Where broadleaf crops were grown in combination with cereals, the lower-value cereal crop tended to be more aggressive.

A linear relationship is not expected because the AI effectively decouples the LER into its components (Figure 4). Except for the legume-oilseed mixtures, the LER was also poorly associated with the NetGM (Figure 5), as highlighted by the coefficient of determination (

Table 7. Strength of the association (linear R²) between NetGM and LER, YR, and VR, and break-even LER and the number of treatments exceeding the threshold for each of these metrics; selected mixtures for all sites and all years.

Intercrop	Total Number of Treatments	LER		YR		VR		NetGM	Break-Even LER
		R2	n	R2	n	R2	n	n
Barley and Canola	45	0.06	17	0.22	32	0.26	18	0	1.03
Faba bean and Canola	41	0.68	29	0.88	23	0.89	25	18	1.08
Field pea and Canola	47	0.80	33	0.85	31	0.90	33	20	1.09
Faba bean and Wheat	47	0.52	29	0.60	32	0.76	25	4	1.16
Lentil and Wheat	12	0.18	6	0.13	12	0.54	3	0	1.21

R² is the coefficient of determination. n is the number of treatments exceeding the threshold for the land equivalent ratio (LER), yield ratio (YR), value ratio (VR), and net gross margin (NetGM). The break-even LER is the intercept estimated for a linear equation relating LER (dependent variable) to NetGM (independent variable).

). Absolute changes in yield (as measured by the YR) and absolute changes in gross value (as measured by the VR) were progressively better predictors of changes in profitability (as measured by the NetGM) than the LER (Table 7). This was particularly the case for the barley-canola and lentil-wheat mixtures in part due to the relative value of the more aggressive crop component accentuated by the very high post-harvest grain separation costs.

• The number of barley-canola treatments exceeding the YR threshold was 32, but the number exceeding the VR threshold fell to 18, reflecting the price penalty for the more aggressive barley (at $278/t) compared to canola (at $348/t).
• The number of wheat-lentil treatments exceeding the YR threshold was 12, but the number exceeding the VR threshold was only 3, reflecting the price penalty for the more aggressive wheat (at $330/t) compared to lentil (at $454/t).

Intercropping consistently resulted in increased variable costs that dragged down the NetGM, largely reflecting post-harvest seed separation costs (Figure 6). These costs were prohibitively high when there was cereal in the mix. Compared to variable activity costs averaged across all monocultures at the core sites, costs were, on average, 30% higher for faba bean and canola mixtures and 160% higher for lentil-wheat mixtures.

Summaries of the experimental means by site are tabulated with standard errors for all sites and mixtures in Appendix B, Appendix C, Appendix D, Appendix E and Appendix F. At Hamilton, the most profitable mixtures were field pea and canola and faba bean and canola. Profitable field pea and canola mixtures had profits ranging from $12/ha to $576/ha relative to their monocultures (

Table A13. Mixtures of field pea and canola grown at Hamilton in southwest Victoria (2019, 2020 and 2021).

	Monoculture 1	Monoculture 2	Mix 25:75		Mix 50:50		Mix SR 50:50		Mix 75:25
Year	Field Pea	Canola	Mix 1	Mix 2	Mix 1	Mix 2	Mix 1	Mix 2	Mix 1	Mix 2
2019
Yield (t/ha)	7.2 (0.50)	4.4 (0.23)	1.0 (0.45)	3.5 (0.56)					3.5 (0.79)	1.6 (0.41)
GM ($/ha)	2588 (204)	2045 (133)
LER			0.93 (0.11)						0.85 (0.18)
YR			0.88 (0.10)						0.78 (0.16)
VR			0.91 (0.10)						0.82 (0.17)
Net GM ($/ha)			−431 (247)						−766 (430)
2020
Yield (t/ha)	5.5 (0.91)	3.1 (0.56)	4.3 (1.13)	1.8 (0.20)	4.8 (1.15)	1.2 (0.36)	3.8 (0.12)	2.3 (0.23)	5.4 (0.77)	0.8 (0.32)
GM ($/ha)	1900 (373)	1321 (316)
LER			1.35 (0.20)		1.25 (0.15)		1.40 (0.08)		1.21 (0.21)
YR			1.62 (0.29)		1.39 (0.21)		1.39 (0.07)		1.24 (0.20)
VR			1.47 (0.24)		1.32 (0.18)		1.39 (0.08)		1.23 (0.20)
Net GM ($/ha)			576 (391)		326 (308)		491 (139)		158 (389)
2021
Yield (t/ha)	5.7 (0.56)	4.0 (0.46)	4.8 (0.97)	2.1 (0.34)	4.7 (1.00)	1.2 (0.53)	4.7 (0.48)	1.4 (0.46)	5.1 (0.39)	0.6 (0.53)
GM ($/ha)	1959 (229)	2092 (532)
LER			1.36 (0.13)		1.13 (0.14)		1.18 (0.15)		1.05 (0.11)
YR			1.55 (0.17)		1.22 (0.16)		1.26 (0.15)		1.09 (0.09)
VR			1.39 (0.13)		1.14 (0.14)		1.19 (0.15)		1.06 (0.11)
Net GM ($/ha)			522 (256)		12 (291)		112 (317)		−176 (234)

). Profitable faba bean and canola mixtures had profits ranging from $85/ha to $453/ha (

Table A7. Mixtures of faba bean and canola grown at Hamilton in southwest Victoria (2019, 2020 and 2021).

	Monoculture 1	Monoculture 2	Mix 25:75		Mix 50:50		Mix SR 50:50		Mix 75:25
Year	Faba Bean	Canola	Mix 1	Mix 2	Mix 1	Mix 2	Mix 1	Mix 2	Mix 1	Mix 2
2019
Yield (t/ha)	7.3 (0.52)	4.4 (0.23)	0.7 (0.12)	4.5 (0.18)					2.8 (0.50)	2.5 (0.19)
GM ($/ha)	2981 (245)	2045 (133)
LER			1.10 (0.05)						0.94 (0.03)
YR			0.99 (0.06)						0.79 (0.05)
VR			1.03 (0.06)						0.84 (0.04)
Net GM ($/ha)			−180 (138)						−769 (126)
2020
Yield (t/ha)	4.8 (1.17)	3.1 (0.56)	2.6 (0.25)	2.4 (0.32)	1.9 (1.22)	2.5 (0.39)	2.9 (1.03)	2.3 (0.22)	3.9 (1.27)	1.5 (0.49)
GM ($/ha)	1765 (553)	1321 (316)
LER			1.32 (0.10)		1.20 (0.25)		1.34 (0.24)		1.29 (0.13)
YR			1.43 (0.10)		1.12 (0.30)		1.32 (0.28)		1.23 (0.19)
VR			1.38 (0.10)		1.15 (0.28)		1.33 (0.27)		1.25 (0.17)
Net GM ($/ha)			453 (169)		85 (503)		405 (479)		278 (311)
2021
Yield (t/ha)	7.1 (0.89)	4.0 (0.37)	3.5 (2.25)	2.5 (1.01)	5.6 (2.08)	1.6 (0.86)	4.8 (1.17)	2.2 (0.13)	5.2 (2.58)	1.3 (0.72)
GM ($/ha)	2859 (419)	3274 (204)
LER			1.11 (0.07)		1.18 (0.10)		1.23 (0.15)		1.07 (0.26)
YR			1.25 (0.26)		1.29 (0.23)		1.27 (0.20)		1.04 (0.33)
VR			1.21 (0.20)		1.26 (0.19)		1.26 (0.19)		1.05 (0.31)
Net GM ($/ha)			204 (435)		358 (464)		366 (463)		−171 (843)

). At Horsham, the most profitable mixtures comprised field pea and canola. The most profitable mixtures had gross margins of $252/ha to $285/ha higher than the equivalent monocultures but required very strong relative yields and LERs exceeding 1.2 (

Table A15. Mixtures of field pea and canola grown at Horsham in northwest Victoria (2020 and 2021).

	Monoculture 1	Monoculture 2	Mix 25:75		Mix 50:50		Mix SR 50:50		Mix 75:25
Year	Field Pea	Canola	Mix 1	Mix 2	Mix 1	Mix 2	Mix 1	Mix 2	Mix 1	Mix 2
2020
Yield (t/ha)	3.7 (0.84)	4.6 (0.87)	1.0 (0.26)	3.8 (0.55)	1.5 (0.46)	2.8 (0.72)	3.1 (0.50)	2.3 (0.93)	2.6 (1.34)	2.2 (1.40)
GM ($/ha)	1168 (344)	2179 (494)
LER			1.10 (0.16)		1.00 (0.13)		1.35 (0.10)		1.19 (0.12)
YR			1.10 (0.15)		1.03 (0.14)		1.31 (0.13)		1.23 (0.11)
VR			1.10 (0.15)		1.05 (0.16)		1.26 (0.17)		1.29 (0.17)
Net GM ($/ha)			−2 (325)		−97 (300)		252 (326)		285 (303)
2021
Yield (t/ha)	2.6 (0.73)	1.2 (0.41)	1.0 (0.01)	0.6 (0.06)	1.1 (0.23)	0.6 (0.07)	1.6 (0.28)	0.5 (0.23)	1.6 (0.19)	0.3 (0.27)
GM ($/ha)	732 (298)	247 (234)
LER			0.92 (0.05)		0.90 (0.15)		1.05 (0.10)		0.88 (0.28)
YR			1.07 (0.04)		0.88 (0.16)		1.12 (0.05)		0.85 (0.18)
VR			1.01 (0.04)		0.89 (0.16)		1.09 (0.05)		0.86 (0.21)
Net GM ($/ha)			−68 (30)		−183 (122)		−26 (43)		−245 (191)

). At Rutherglen, the intercropping response was higher in the more favourable seasons (2020 and 2021) when a majority of the 16 mixtures evaluated demonstrated LERs greater than unity. The most profitable mixtures were faba bean and canola and field pea and canola. Profitable faba bean and canola mixtures were $64/ha to $594/ha higher (

Table A11. Mixtures of faba bean and canola grown at Rutherglen in northeast Victoria (2019, 2020 and 2021).

	Monoculture 1	Monoculture 2	Mix 25:75		Mix 50:50		Mix SR 50:50		Mix 75:25
Year	Faba Bean	Canola	Mix 1	Mix 2	Mix 1	Mix 2	Mix 1	Mix 2	Mix 1	Mix 2
2019
Yield (t/ha)	1.6 (0.20)	0.8 (0.18)	0.2 (0.06)	0.7 (0.26)					0.6 (0.10)	0.7 (0.15)
GM ($/ha)	356 (92)	15 (104)
LER			0.97 (0.33)						1.18 (0.23)
YR			0.88 (0.27)						0.88 (0.16)
VR			0.90 (0.29)						0.94 (0.18)
Net GM ($/ha)			−77 (142)						−91 (112)
2020
Yield (t/ha)	5.7 (0.42)	3.8 (0.37)	3.4 (0.75)	2.4 (0.24)	3.8 (0.64)	1.5 (0.17)	3.9 (0.60)	1.6 (0.28)	4.6 (0.69)	1.1 (0.26)
GM ($/ha)	2282 (197)	1698 (208)
LER			1.21 (0.08)		1.07 (0.13)		1.10 (0.14)		1.09 (0.08)
YR			1.34 (0.13)		1.12 (0.15)		1.16 (0.15)		1.08 (0.10)
VR			1.29 (0.11)		1.10 (0.14)		1.14 (0.15)		1.09 (0.09)
Net GM ($/ha)			384 (213)		−18 (307)		64 (326)		−55 (209)
2021
Yield (t/ha)	6.4 (0.68)	2.4 (0.54)	3.8 (0.48)	1.4 (0.59)	5.2 (0.64)	0.9 (0.39)	4.5 (1.07)	0.9 (0.34)	5.9 (0.39)	0.6 (0.35)
GM ($/ha)	2627 (322)	934 (307)
LER			1.19 (0.18)		1.19 (0.08)		1.08 (0.12)		1.15 (0.17)
YR			1.54 (0.07)		1.38 (0.07)		1.22 (0.19)		1.19 (0.11)
VR			1.47 (0.08)		1.36 (0.06)		1.20 (0.18)		1.18 (0.11)
Net GM ($/ha)			594 (143)		487 (120)		185 (346)		170 (269)

), while profitable field pea and canola mixtures had profits ranging from $32/ha to $336/ha relative to their monocultures (

Table A17. Mixtures of field pea and canola grown at Rutherglen in northeast Victoria (2019, 2020 and 2021).

	Monoculture 1	Monoculture 2	Mix 25:75		Mix 50:50		Mix SR 50:50		Mix 75:25
Year	Field Pea	Canola	Mix 1	Mix 2	Mix 1	Mix 2	Mix 1	Mix 2	Mix 1	Mix 2
2019
Yield (t/ha)	1.0 (0.19)	0.8 (0.18)	0.2 (0.05)	0.8 (0.05)					0.6 (0.11)	0.3 (0.07)
GM ($/ha)	46 (80)	15 (104)
LER			1.23 (0.10)						1.03 (0.12)
YR			1.22 (0.11)						1.00 (0.11)
VR			1.24 (0.10)						1.04 (0.12)
Net GM ($/ha)			69 (40)						−41 (44)
2020
Yield (t/ha)	5.8 (1.02)	3.8 (0.37)	1.7 (0.54)	3.1 (0.61)	2.9 (0.49)	2.8 (0.28)	1.4 (0.49)	3.1 (0.20)	3.4 (0.63)	2.4 (0.55)
GM ($/ha)	2023 (418)	1698 (208)
LER			1.12 (0.08)		1.25 (0.02)		1.06 (0.12)		1.23 (0.23)
YR			1.13 (0.04)		1.20 (0.05)		0.94 (0.13)		1.10 (0.20)
VR			1.12 (0.06)		1.23 (0.03)		1.02 (0.13)		1.18 (0.22)
Net GM ($/ha)			32 (137)		247 (43)		−162 (255)		140 (464)
2021
Yield (t/ha)	6.5 (1.59)	2.4 (0.54)	3.3 (1.34)	1.5 (0.53)	5.5 (1.69)	0.7 (0.40)	3.7 (1.25)	1.3 (0.16)	4.7 (0.79)	0.5 (0.34)
GM ($/ha)	2311 (651)	934 (307)
LER			1.12 (0.15)		1.15 (0.16)		1.10 (0.24)		0.93 (0.26)
YR			1.39 (0.29)		1.40 (0.31)		1.12 (0.31)		0.94 (0.20)
VR			1.30 (0.23)		1.33 (0.27)		1.12 (0.29)		0.94 (0.21)
Net GM ($/ha)			267 (329)		336 (458)		−19 (512)		−416 (451)

).

4. Discussion

This extensive experimental field program assessing dryland intercropping systems has confirmed that intercropping with mixtures can be productive and profitable in southern Australia. However, this observation was inconsistent across all crop mixtures when compared with their monoculture equivalent. We have identified those mixtures that were productive and profitable. The challenge for farmers is to identify low-risk options that are potentially profitable in their locations.

We focused on metrics that compare a mixture against its component monocultures at equivalent enterprise ratios to identify suitable crop mixtures. For this reason, we recommend the YR, VR and NetGM metrics over the more commonly reported LER to test the overall productivity and economic advantage of mixtures of interest.

Absolute changes in yield, as measured by the YR, and absolute changes in gross value, as measured by the VR, were progressively better predictors of changes in profitability, as measured by the NetGM (Table 7). We included the AI to understand whether a given productivity response was driven by the domination of one species over another, as the dominance of a higher yielding and/or higher unit value crop could enhance the profitability of the whole mixture compared to a similar ratio of monocultures. As well as absolute yields and unit crop prices, the NetGM metric introduces costs that farmers need to consider, importantly, grain sorting after harvest. For mixtures marketed for human consumption, these costs are potentially a major additional factor as they are imposed on a $/t basis, so the more the intercrop out-performs compared to a similar ratio of monocultures, the greater the additional cost penalty.

Biologically successful mixtures, assessed using the LER and YR, tended to be complementary and comprised faba bean and canola, field pea and canola, and faba bean-wheat (Figure 4, Appendix B, Appendix C, Appendix D, Appendix E and Appendix F) that were sown and harvested together. Highly negative responses were rare for these mixtures, indicating limited downside production risk. The largely positive responses are attributed to better resource use in the higher rainfall locations (Figure 4). Other studies support our findings, having reported very high LER for field pea and canola mixtures in Australia [22,23,24]. Fletcher et al. [24], for example, reported that ‘peaola’ (canola-field pea intercrops) performed the best out of the three main intercrop groups compared with the single varieties in large rain-fed grain cropping systems. Based on the LER metric, 70% of canola-field pea intercrops examined in their review paper had a 50% productivity increase over the monocultures, compared to 64% of cereal-grain legume intercrops. By contrast, mixtures of cereal varieties showed no evidence of a productivity increase.

Where we observed negative mixture responses, more aggressive and lower unit value cereals (wheat and barley) outcompeted the accompanying higher unit value oilseed or legume (Figure 4). Furthermore, the mixtures with cereals (wheat or barley) were more difficult to separate, and grain cleaning, according to Grain Trade Australia standards for human consumption, was prohibitively expensive. For these reasons, our results using the VR and NetGM show that many of these mixtures were often not more valuable or profitable even in higher rainfall environments (Figure 5, Appendix B, Appendix C, Appendix D, Appendix E and Appendix F).

A poor mixture productivity and profitability was generally seen in the drier growing environments (Figure 4 and Figure 5). This will help growers in the drier regions wanting to test intercropping principles to be more cautious and focus on cost reduction and possibly longer-term considerations like building soil organic matter and more stable nutrient supplies. Our results show that if seed separation costs could be reduced, then NetGMs above zero are possible, and agronomic methods or a change in enterprise mix could be pursued to reduce such costs. An example might be to avoid seed separation costs by harvesting the mixture as a livestock feed valued for its energy and protein components for intensive animal production or grazing by animals in a mixed crop-livestock system. Grain and feed quality are important factors that some growers need to consider as part of intercropping, particularly in mixed enterprises. Typically, farmers focus on their main enterprise, such as grain or grazing. Our initial study focused on grain production to examine the major effects of mixtures because there were few examples of intercropping in the grain regions of southern Australia.

Beyond productivity and profitability, intercropping offers many other advantages over monocultures. We focused on the benefits of mixtures in the year of production. We also avoided the confounding effect of growing mixtures at densities greater than parity (monoculture equivalent) so we could measure the true mixture response. Growing mixtures at over-parity densities requires a plot-by-plot comparison without reference to an enterprise mix ratio of the monocultures and represents a completely different farming system and is, therefore, not strictly a mixture comparison.

Given that our work was conducted over three years, it does not represent a long-term view of intercropping in the environments tested. Over the experimental period, a range of seasonal conditions was encountered, including one year where growth was poor due to limited rainfall (2019), followed by two average to above-average rainfall seasons (2020, 2021). Similarly, the experimental timeline was insufficient for a risk analysis using statistical dominance or efficiency techniques [21,30,31]. Such an analysis would require yields from sequenced or rotational field trials or a properly formulated and validated biophysical intercrop model.

We have identified that legume and oilseed mixtures generally had clear advantages over monoculture in wetter areas and seasons. New investigations should, therefore, explore the mechanisms of response, which were not able to be examined in this study. Probable mechanisms include improved N nutrition and resource sharing (notably water in rain-limited environments) or altered disease, pest or weed interactions [24]. Another probable mechanism, noted in a sister modelling review [32], is the morphological changes (plasticity effects) that are often seen, as we noted in field pea-oilseed mixtures where the field pea tended to trellis on the canola, thus gaining more height and using more water and nutrient resources, possibly more efficiently than would occur in a monoculture. Another mechanism we observed was a lower incidence of chocolate spot disease in a wet year (2020) within intercropped faba bean than in a monoculture in one of our experiments [33]. We also did not address disease and weed management across multiple years, as each experiment was sown into a new area. These issues need to be investigated to ensure the long-term productivity of companion cropping within an agricultural system.

5. Conclusions

This research confirmed our hypothesis that there are annual crop mixtures that can be grown profitably in modern, mechanised grain cropping systems such as those practised in southern Australia. Results show a strong bias towards legume and oilseed mixtures, with mixtures involving cereals being doubly disadvantaged by the aggressiveness of these lower-value crops in the mix and high grain separation costs post-harvest.

We also demonstrated the greater utility of the YR, VR and NetGM metrics postulated by Khanal et al. [22] compared to the more commonly used LER. The NetGM was the most important metric needed to evaluate the economic advantage of crop mixtures compared to the equivalent proportion of monocultures. This was empirically demonstrated by applying these metrics to yields obtained from a comprehensive series of winter cropping field experiments undertaken across three years of wet and dry seasons and three representative rainfall zones of Victoria in southern Australia.

We examined the shorter in-season, direct, private benefits of intercropping. Other system benefits need longer-term experiments to thoroughly assess the sustainability of these mixed crop systems in terms of pest and disease management, nutrient cycling, as well as public benefits such as improved soil and water quality, carbon sequestration, and biodiversity conservation. The assessment of the profit versus risk trade-off is also best assessed using a long-term time series of yields from sequenced or rotational field trials and crop models.