Crop Systems Unit · 6 active projects · 11 saved cohorts

Research workspace

Start visit authorised + public projections synthetic demo data

17 new compatible case files match saved cohort "Biological seed treatments · winter wheat"

cohort v4 · membership unchanged until you accept · COH-BST-WW-v4

Review candidates

3 method incompatibilities flagged in "N-use efficiency" cohort

2 unit mismatches (kg N/ha vs lb N/ac, approved conversion available) · 1 boundary mismatch (cannot pool)

Inspect matrix

Addendum on a cited record affects 2 snapshots and 1 export

PRF-UK-A-000188 corrected 18 Jul · affected outputs listed for review, originals untouched

Open impact map

Trial site · Brook Ground strip trial — mid-season assessment

09:30 · protocol v2 checklist · counts + canopy photos · offline capture supported

Start visit

Cohort site · Wychwood Estate — SOM sampling round 3

13:00 · LOI protocol · fixed sampling points from 2023 baseline · deviation form ready if timing slips

Start visit

Record explorer · faceted cohort builder

live denominator · k=5 enforced

Active filters

crop: winter wheat

practice: biological input

soil: medium

region: East of England

metric: yield t/ha

strictly compatible

Theme preset: arable biological inputs

presets expose the exact canonical filters beneath — "organic" reads as management context unless certification evidence is recorded

128
records match · live denominator
31
case files · holding the 54 strictly compatible records
37
related context (not poolable)
RecordTypeSeasonOutcomeState
PRF-UK-A-000231Comparison2025+0.2 t/hameasured
PRF-UK-D-002204Outcome2025no clear differencemeasured
PRF-UK-F-001873Outcome2025−0.3 t/hameasured
PRF-FR-K-000966Outcome2025pendingnot yet measured
Sub-group "organic management × chalk soils" has fewer than 5 records — values withheld. Broaden in the explorer

Compatibility matrix

why records can and can't be pooled
MethodUnitBasisPoolable
-000231weighbridget/hafieldyes
-002204weighbridget/hafieldyes
-001873yield monitort/hafieldcalibration differs
-000966plot combinet/haplotno — basis mismatch

The tool refuses to combine apples and pears silently — each "no" carries its plain-language reason.

Exclusion waterfall · cohort v4

total N → included N, every step named
128 matched 92 · outcome measured 54 records · strictly compatible (31 case files) 23 · reference arm present excluded: 36 no outcome · 38 method mismatch · 31 no ref arm

Forest plot · biological seed treatment vs untreated · yield difference

n=23 comparisons · t/ha · ordered by season, explicitly not ranked
0 · no difference PRF-UK-A-000231 · 2025 PRF-UK-D-002204 · 2025 PRF-UK-F-001873 · 2025 PRF-UK-H-000554 · 2024 PRF-UK-B-001990 · 2024 +0.2 −0.1 −0.3 +0.4 0.0

18 further comparisons below · every point opens its record and inclusion reason · no pooled estimate shown where boundaries differ · no conclusion drawn for you.

11
saved cohorts · versioned
4
reproducible snapshots frozen
27
records cited or reused · frequency ≠ strength
This workspace never shows: an evidence-quality score, a validity badge, or a ready-made conclusion. Incompatible measures are never silently combined; citation frequency is never evidence strength; every aggregate opens to sources; theme presets always reveal the canonical filters beneath; k=5 throughout. Every view exports its manifest — filters, exclusions, methods and limitations.