On-Farm Intercropping Trials

Crop Picker's guild finder shows sets of crops that the literature says help each other. It cannot yet say how much. Does a corn–squash strip produce more per acre than corn alone and squash alone on the same land? Does it earn more than the better of the two? Those answers come from side-by-side trials, and very few exist for the crops and climate of the Lake Erie grape belt.

This page is a protocol for running one on your own farm. A trial run this way gives you an answer for your farm. Pooled with other growers' trials, it gives every.farm a measured number for the guild instead of a guess.

📄 Data sheet: intercrop-trial-template.csv. Open it in any spreadsheet. It has one row per event, with example rows to overwrite. 🧪 Not sure what to test? The Trials to run tab in the guild finder lists crop pairs whose documentation would complete the largest guilds.

#What the trial measures

Land equivalent ratio (LER) asks how much land you would need to grow the same harvest as separate sole crops.

For each crop:

partial LER = yield in the mix ÷ yield grown alone

Add the partial LERs together to get the LER.

• LER = 1.0: no advantage. The mix is the same as splitting the field.
• LER = 1.25: one acre of the mix yields what 1.25 acres of sole crops would. That's 25% more from the same land.
• LER below 1.0: the crops compete more than they help.

LER counts pounds and bushels, not dollars. So the trial also records what you sold, at what price, and the hours it took. A mix can have a high LER and still earn less than the best sole crop, as the worked example below shows.

#Design

#Plots

You need three kinds of plot, side by side:

1. the mix: your crops grown together, the way you would really grow them
2. a sole plot of each crop in the mix, grown the way you normally grow it alone

A two-crop trial therefore has three plots and a three-crop trial has four.

#Replicates

One mix plot beside one sole plot can't separate the mix effect from a wet corner or a richer patch of soil. Lay out the set of plots at least 3 times (each set is a replicate or block). Four is better. Within each replicate, shuffle the order of the plots: flip a coin or draw names from a hat, so the mix isn't always on the same side.

           ┌──────────┬──────────┬──────────┐
 Rep 1     │  Mix M1  │ Corn C1  │Squash S1 │
           ├──────────┼──────────┼──────────┤
 Rep 2     │Squash S2 │  Mix M2  │ Corn C2  │
           ├──────────┼──────────┼──────────┤
 Rep 3     │ Corn C3  │Squash S3 │  Mix M3  │
           └──────────┴──────────┴──────────┘
   Keep each replicate on similar ground: run reps ACROSS a slope or
   drainage change, so each one sits on a single soil type.

#Plot size

Plots should be large enough to plant, manage and harvest with your normal tools and to give a harvest you can weigh. Make every plot the same size if you can. For row crops, a plot of four or more rows is a good start, and you can harvest only the centre rows if the edge rows are shaded or crowded by the next plot. Write down the area you actually harvested.

#Management

Treat the mix and the sole plots the same: same field, same planting window, same fertility, same irrigation, same pest control. The trial compares growing systems. If the mix gets extra compost, the result measures the compost.

When the mix requires something different (e.g. a different row spacing), that's part of the system. Do it, and note it in the arrangement row.

#Replacement or additive?

• Replacement: the mix replaces part of each sole stand (e.g. two rows of corn, then two rows of squash). This is the usual design for strip and row intercrops.
• Additive: a companion is added to a full stand of the main crop (e.g. clover undersown in a full stand of oats, or flowering strips beside a full stand of squash).

Record which one you used. Both produce a valid LER.

#What to record

Every record goes on the data sheet, one row per event. The record column says what kind of event it is.

#The one rule that matters

Record each crop's harvest from the whole mix plot, divided by the whole mix plot's area. Don't divide by the area that crop occupied.

If squash took up half of a 1,000 sq ft mix plot and yielded 180 lb, record 180 lb on a 1,000 sq ft plot. Don't record 180 lb on 500 sq ft. The data sheet asks for each plot's area once, on the plot row, and every harvest refers back to it, so if you fill it in as laid out you're doing it right.

Also:

• Use the same unit for a crop in the mix and in its sole plot (ears and ears, lb and lb).
• Weigh or count everything harvested, then the marketable part separately.
• Record every picking for multi-harvest crops (beans, squash, tomatoes). The totals are added up for you.

#Worked example

From replicate 1 of the example rows in the data sheet (1,000 sq ft plots):

LER = 0.80 + 0.40 = 1.20. The mix produced on 1,000 sq ft what sole corn and sole squash would need 1,200 sq ft to produce. That's a real land advantage.

Now add the marketable harvest and prices from the same rows (corn $0.75 an ear, squash $1.50 a lb):

On this farm, at these prices, the mix beats sole corn but earns less than sole squash on the same ground, and takes more harvest time. That's a useful answer. It says the advantage would come from the mix's other benefits (fewer cucumber beetles in this example: 1.2 per plant against 2.8), or from a different arrangement giving squash more room. It doesn't come from yield alone. One replicate is one data point. The three replicates together, and other growers' trials, decide.

#Trees, vines and perennial alleys

Alley cropping (crops between tree or vine rows) and understory plantings take years to show their result. The design is the same, with three additions:

• Sole plots for the perennial too: trees or vines without the alley crop, managed the same way.
• Several seasons: record every year from planting. The alley crop usually yields well early and less as the canopy closes. That decline is part of the result.
• Yield by distance: in at least one alley, harvest the alley crop in bands (for example 0–5 ft, 5–10 ft and 10–15 ft from the tree row). Record each band as an observation with metric yield_at_distance and the distance in the notes. It shows how far the trees' shade and roots reach.

Each year, also record survival and one growth measure for the trees or vines (trunk diameter, height or cane count) in both the alley plots and the sole plots.

#Common mistakes

• No sole plots. Without them there's no LER, only a yield. Plant the sole plots even if they're small.
• Different management. Extra water or fertility on the mix turns the trial into a fertility trial.
• Mix always in the best spot. Shuffle plot order within every replicate.
• One replicate. It's still worth recording, but it can't count toward a guild's measured LER.
• Per-crop area. See the one rule that matters.
• Mixing units. Bushels in the mix and pounds in the sole plot can't be compared.
• Dropping the bad year. A season where the mix failed is as important as one where it shone.

#Submitting your trial

Send the completed data sheet to every.farm. Every.farm keeps your farm's records. Crop Picker receives only:

• the plot-level numbers on the sheet
• your county and state, and a location rounded to roughly 10 km
• an anonymous reference to the farm, never your name or address

Each submission is checked for sole plots, matching units, replicates and complete dates. If it passes, it's marked verified. Verified trials appear in Crop Picker beside the guild they test, and they count toward that guild's measured LER and per-acre value. Trials that don't pass are kept with your farm on every.farm.

Your trial won't be the only number shown. Each guild shows every trial behind it, how many there are, their range, and how closely each trial's soil and climate match the farm asking. Until a guild has trials, Crop Picker assumes no land advantage (LER 1.0) in any economic estimate.

#For developers

The data sheet maps onto the Crop Picker trial tables:

The view intercrop_trial_ler gives per-trial LER, area × time equivalent ratio (for relay and long-season mixes), mix value and best sole value. Published trials enter the same tables with every value cited to the paper, and ler_basis records whether a total was computed from yields or reported by the authors. The schema is a draft pending review; endpoints will be documented in the Crops API once it is applied.