Field Trial Sample Control – Protecting Integrity from Plot to Lab

This topic is part of the SG Systems Global regulatory & operations glossary.

Updated December 2025 • Sampling Plans, Label Verification, LIMS, Data Integrity, Record Retention • Field Ops, QA, R&D, Regulatory, Labs

Field trial sample control is the end‑to‑end discipline of uniquely identifying, collecting, handling, transporting, splitting, storing, testing and archiving samples generated during agricultural field trials—so results can be trusted, reproduced and defended. In crop‑protection development and registration, the weak point is rarely the instrument in the lab. It’s the messy middle: mislabeled bottles, mixed-up plots, uncontrolled hold times on a hot truck, “helpful” relabeling in the field, missing metadata, and sample splits that no longer match the original material. When sample control is weak, trial results become arguable: you can’t prove what was tested, when it was collected, or whether it was representative. That is not a science problem. It’s a control problem—and it will waste money, time and credibility very quickly.

“If you can’t prove the sample’s identity and history, you’re not running a trial—you’re running a story.”

1) What Field Trial Sample Control Actually Is

This discipline answers three questions: (1) Is the sample unequivocally tied to the correct trial, plot, treatment, timepoint and collector? (2) Was the sample handled in a way that preserves what it is supposed to represent (no degradation, contamination, or selective sub‑sampling)? (3) Is there an auditable record of every handoff, storage condition and change to the sample’s status from collection through testing and archiving? A controlled program makes these questions easy to answer using time‑stamped records. An uncontrolled program relies on memory and good intentions—which fail the moment a trial scales, staff changes, or conditions get hectic.

2) Why This Is High-Stakes in Agrochemical Development

Field trials are expensive, seasonal, and often non-repeatable: you can’t easily recreate last month’s weather, pest pressure, crop stage or soil conditions. That gives sample integrity disproportionate leverage. If sample identity is uncertain, you lose the ability to interpret results confidently—whether for efficacy, residue, phytotoxicity, drift behavior, or formulation comparisons. Regulators and internal technical teams don’t require perfection; they require control. If your sample history is incomplete, you can’t defend decisions, and you will end up re-running work that should have been decisive the first time.

3) Sample Identity – Unique IDs Beat Handwritten Names

Field environments punish ambiguity. A robust approach assigns each sample a unique ID that encodes or links to trial metadata (trial ID, site, plot, treatment, timepoint, matrix, replicates) and is printed as a durable label with barcode support. Handwriting is a backup, not the system. Use barcode validation and label verification rules so “looks right” doesn’t substitute for “is right.” If an operator can relabel a sample without a record, you’ve created a fraud-friendly process—even if nobody intends fraud. Control systems are built for the days when things go wrong, not only for the days when everyone behaves perfectly.

4) Sampling Plans – Representative Collection Is Not Improvisation

Sampling must be representative and repeatable. Define how many subsamples, from which locations, at which times, and using what tools—then train and enforce it. A formal sampling plan prevents common field distortions: selectively grabbing “good-looking” leaves, sampling only from convenient edges, or changing the method mid-trial because the schedule is tight. Sampling plans also define what to do when reality conflicts with the plan (rain, missed timepoint, insufficient biomass): those deviations should be documented, not silently “fixed” by making up a new method on the fly.

5) Metadata – The Missing Data That Invalidates Good Samples

A physically perfect sample can still be analytically useless if the metadata is missing: crop stage, application details, nozzle type, spray volume, adjuvant used, water source or hardness, time since application, weather, soil conditions, and any unusual events. The control model is straightforward: decide what metadata is mandatory, capture it at the point of work, and prevent sample acceptance without it. In digital programs, this is where electronic work instructions and mobile capture matter more than “we’ll write it up later.” If you rely on later reconstruction, you will lose information—and the missing pieces will always be the ones you most need to interpret unexpected results.

6) Containers, Preservation, and Cross-Contamination Controls

Sample containers are part of the method. Use defined container types, closures, liners and preservatives appropriate to the matrix (plant tissue, soil, water, formulation, spray mix). Control tools and consumables to prevent cross‑contamination: clean tools, single-use implements where justified, and clear rules for glove changes and handling. In a busy field station, the easiest contamination route is not “lab contamination”—it’s the quick wipe and reuse of a cutter, scoop or bucket. If contamination is plausible, build controls so you can detect it (blanks, duplicates) and respond with data rather than argument.

7) Hold Time and Temperature – Samples Don’t Pause Their Chemistry

Field samples age and degrade. Control requires defined maximum hold times and storage conditions, supported by hold time studies and stability logic where needed. Temperature excursions in a vehicle, on a dock, or in a site fridge can change residues, moisture, and volatile components. Use temperature logging where risk demands it and apply the same discipline you would in manufacturing: detect, document, assess, disposition. If a sample spends eight hours in the sun, the correct response is not to pretend it didn’t happen; it’s to record it and decide what the data can still support.

8) Shipping and Handoffs – Logistics Is Part of the Method

Shipping is a controlled handoff, not a clerical step. Document what shipped, when, by whom, under what conditions, and to whom. Use shipping documents appropriate to your program such as a shipping manifest and, where useful, a bill of lading. Cold-chain or controlled conditions should be specified and verified, not assumed. If samples arrive partially thawed, broken, leaking, or unlabeled, the receiving process should force decisions: accept with justification, quarantine for assessment, or reject. Quietly absorbing bad shipments is how you create data you can’t defend later.

9) Receiving, Accessioning, and LIMS Linkage

Receiving is where field work becomes laboratory work. A controlled process includes: scan-based receipt, condition checks, reconciliation against the manifest, and immediate accessioning in LIMS so the sample’s identity and status are locked. LIMS should tie the sample to the study plan, methods, and required tests, and it should capture any pre-analytical exceptions (late arrival, temperature excursion, damaged container). If receipt happens on paper and LIMS entry happens later, you’ll get mismatches—because “later” is when humans forget and transpose data.

10) Sub-Sampling, Splits, and “Which Portion Was Tested?”

Sub-sampling is a high-risk step because it can create untraceable divergence. Define how homogenization happens (if applicable), how splits are created, and how each split is labeled and tracked. If a sample is split for multiple labs, each child sample should inherit traceability back to the parent, including the exact split method and date. Where possible, use controlled identifiers and analytical lot linkage concepts so analytical results can be connected back not only to the right parent sample but also to the correct processing history. If your system can’t answer “which portion was tested,” you can’t rule out selection bias or handling differences.

11) Data Integrity – Audit Trails, Roles, and Electronic Sign-Off

Field trial data is only as credible as its integrity controls. That means applying data integrity principles: unique user identities, controlled roles and permissions, tamper-evident audit trails, and (where appropriate) electronic sign-off for key events like sample collection confirmation, shipment release, receipt acceptance, and data approval. If staff can change IDs, dates, or results without a trace, you don’t have a controlled study system—you have spreadsheet culture with better branding.

12) Exceptions – Deviations, OOS/OOT, and Retesting Rules

Samples and results will not always behave. Define what constitutes a deviation (missed timepoint, broken chain, temperature excursion), how it is documented (see deviation/NC), and what actions are allowed. Establish clear retest and resample rules so teams don’t “shop” for cleaner outcomes when a result is inconvenient. For analytical results, use disciplined handling of OOS and OOT signals, including investigation logic that considers sample history as a first-class root cause candidate. If you don’t control exceptions, you create narratives instead of evidence.

13) Retains, Archiving, and Record Retention

Retains protect you when questions arrive later—often months or years after a trial. Define what retains are needed (sample, extract, reference standard aliquots, documentation), how they’re labeled, where they’re stored, and for how long. Align retention with record retention and archival governance so you can actually retrieve what you claim exists. A “retain” that no one can find is not a retain; it’s a comforting story. The best retention programs tie storage locations to digital records, apply environmental controls where needed, and periodically verify that retention is real, not theoretical.

14) Traceability Back to Manufacturing Lots and Formulation States

Field trials often compare formulations, co‑formulants, and “small changes” that can alter performance. To interpret results, you must tie each trial sample to the exact material state used: formulation batch, co‑formulant lots, and any site-specific adjustments. This is where manufacturing-style genealogy helps: end‑to‑end traceability, batch genealogy, and component lot traceability. Without that linkage, “the field result changed” becomes impossible to diagnose: was it the active, the surfactant system, the water, the application setup, or the samples? Traceability doesn’t solve every uncertainty, but it prevents avoidable uncertainty from dominating every interpretation.

15) Implementation Roadmap – From Field Chaos to Controlled Evidence

Most organisations mature in stages. Stage 1: samples are labeled by hand, stored inconsistently, and logged later; results are usable until they’re questioned. Stage 2: basic barcodes and manifests exist, but chain-of-custody and exception handling are inconsistent. Stage 3: defined sampling plans, scan-based receipt, LIMS accessioning, controlled holds, and disciplined deviation handling become routine. Stage 4: integrated digital capture across field and lab, with audit trails, retention governance, and trend review of sample exceptions to drive continuous improvement. The practical best first move is to lock down identity and metadata capture: unique IDs, durable labels, mandatory metadata fields, and scan-based receipt. If you can’t trust identity, nothing else you do will matter.

16) FAQ

Q1. What’s the most common failure mode in field trial sample integrity?
Identity and history ambiguity: mislabeled containers, relabeling without records, missing metadata, and untracked hold times. These failures don’t always ruin results—but they always make results easier to challenge.

Q2. Do we really need barcodes, or is handwriting “good enough”?
Handwriting works until it doesn’t. Barcodes reduce transcription, enforce uniqueness, and make receiving and reconciliation far more reliable. If a trial matters, barcode-based identity control is a cheap insurance policy.

Q3. How do we handle temperature excursions during shipping?
Document them, quarantine affected samples, assess impact based on sensitivity and time/temperature exposure, and decide whether data remains usable. Pretending excursions didn’t happen is how you create indefensible outcomes.

Q4. Can we resample or retest when results look wrong?
Only under defined rules. Retesting and resampling must be controlled to avoid bias and “result shopping.” If you allow ad-hoc redo decisions, your study integrity becomes negotiable.

Q5. What is the first practical step for a legacy field program?
Standardise IDs and metadata, implement a manifest-driven handoff, and enforce scan-based receipt into LIMS. Most programs see immediate improvement just by removing ambiguity and missing information.

Related Reading
• Identity & Labelling: Label Verification | Barcode Validation | User Access Management | Audit Trail
• Sampling & Methods: Sampling Plans | Hold Time Study | Stability Studies | Lab Tests & Review
• Systems & Traceability: LIMS | Analytical Lot Link | Lot Traceability | Batch Genealogy | Component Lot Traceability
• Shipping & Exceptions: Shipping Manifest | Bill of Lading | Deviation/NC | OOS | OOT | CAPA
• Governance: Data Integrity | Record Retention | SOP