Trial validation protocol - AHDB Cereals & Oilseeds : … · Web viewProtocol VALID 2017 –...
21
Click here to load reader
Transcript of Trial validation protocol - AHDB Cereals & Oilseeds : … · Web viewProtocol VALID 2017 –...
PROTOCOL VALID 2017 – OCTOBER 2013 (UPDATED OCTOBER 2015)
AHDB RECOMMENDED LIST
TRIAL INSPECTION AND VALIDATION
PROTOCOL
RL Project Consortium. Registered Office: AHDB Cereals & Oilseeds, Stoneleigh Park, Kenilworth, Warwickshire, CV8 2TL
1
AHDB Recommended List trials
Trial inspection and validation
Protocol
HGCA RECOMMENDED LIST......................................................................................................................1
1 General information...........................................................................................................................................3
1.1 RL Team contact details...........................................................................................................................3
2 Trial inspections.................................................................................................................................................4
2.1 Timing of inspections................................................................................................................................4
2.2 Access requirements.................................................................................................................................4
2.3 Reports......................................................................................................................................................4
2.4 Inspection guidelines................................................................................................................................5
3. Post-harvest statistical and technical validation of data............................................................................6
3.1. Preliminary validation of data - individual trials..............................................................................6
3.2. Non-yield data....................................................................................................................................7
3.3. Yield data...........................................................................................................................................7
3.4. Data ranges........................................................................................................................................8
3.5. Analysis of variance statistics:...........................................................................................................9
3.6. Statistical and technical validation following analysis – overtrial..................................................10
Appendix 1a Inspection report form (field inspection) Cereals & oilseeds....................................................12
Appendix 1bPost-harvest validation Cereals & oilseeds................................................................................13
2
3
Part 1 - General information
1.1 RL Team contact details
Headquarters postal address: RL Project Consortium, AHDB Cereals & Oilseeds, Stoneleigh Park, Kenilworth, Warwickshire, CV8 2TL
Huntingdon office address: Ashton House, Ambury Road South, Huntingdon, Cambridgeshire, PE29 3EH
Workbooks and all other data should be e-mailed to [email protected]
RL & Agronomy Trials Coordinator (Primary technical contact)Mark Bollebakker Tel: 01480 482989 (Huntingdon office)
Mobile: 07972 637899E mail: [email protected]
Field operations team leader (Technical contact)Sean Burns Tel: 01480 482584 (Huntingdon office)
Mobile: 07773639936E mail: [email protected]
RL & Agronomy Data Analyst (Primary contact: workbooks and data handling)Ms Beatrice Rey Tel: 01480 482987
E mail: [email protected]
RL & Agronomy Data Analyst (Primary contact: workbooks and data handling)Bastiaan Brak Tel: 02476 478868
E mail: [email protected]
RL & Agronomy Data Analyst (Primary contact: for purchase orders, invoices etc.)Mrs Tara Ross Tel: 024 7647 8744
Mobile: 07816171863E mail: [email protected]
RL Data Analysis and Reports ManagerMrs Jane Ryall Tel: 01480 482982
Mobile: 07805 820243E mail: [email protected]
Technical Manager Bill Handley Tel: 01954 232071
Mobile: 07743 195581E-mail: [email protected]
Senior Research & Knowledge Transfer ManagerDr Simon Oxley Tel: 02476 478866
Mobile: 07779 330853E mail: [email protected]
4
2 Trial inspections
2.1 Timing of inspectionsTrials and special field tests will normally be inspected at least once by AHDB inspectors. Additional inspections may be required if problems are encountered. Inspection visits will be arranged to coincide with the best time to observe any defects in the trial. This is likely to be between February and March for winter oilseed rape crops (to allow the assessment of plant population and uniformity of growth after the winter) and cereals will be inspected between May and July to assess the effectiveness of fungicide, plant growth regulator and fertiliser applications and the uniformity of growth and development. Inspections may be earlier if there are serious concerns relating to establishment or growth and development prior to the normal trial inspection period.
2.2 Access requirements
It is a contract condition that all trials will be inspected by an AHDB/RL Project Consortium approved inspector and, in some cases, it may be necessary to visit on more than one occasion.
The requirements for Trial Operators in respect of inspections are as follows:
To give reasonable access to trials and to be available to accompany the inspector if requested at a mutually convenient time.
Details must be supplied (normally via the trial workbook) of site location, layout of trials within the field and of plots/varieties within the trial.
To supply the inspector with information (for example sprays applied etc) within seven days of a request. To co-operate with the inspector in making any non-routine assessments required to establish the validity of the
trial (for example population counts). To carry out any action agreed in consultation with the inspector. In particular it is important that any requirement
to shorten plots is undertaken and that missing values are returned on any plots which have been rejected.
The trials inspection also provides an opportunity for feed-back to AHDB/RL Project Consortium about any problems with the trials protocol and Trial Operators are encouraged to make any points to the inspector so that these can be considered for future revisions to the protocol.
2.3 Reports
The trial inspector will record his/her findings on a standard form (see Appendix 1). The trial inspector will discuss the state of the trial with the trial manager during the inspection visit or, if unaccompanied, by email shortly afterwards. If any non-routine action is thought necessary, a plan of action will be agreed at and the details included in the report. Having agreed on a course of action it is the responsibility of the trial manager to ensure that this is carried out.
Individual reports will be posted on the admin area of the AHDB website (usually within the trial workbook) and it is the responsibility of the trial manager to read the relevant reports and to implement the agreed course of action. The trial manager will be given a user-name and password to allow access to the admin area.
5
2.4 Inspection guidelinesThe trial inspectors will assess each trial using the following guidelines:
Sowing datePrevious cropSoil type
Do these meet those given in the protocol (if defined) and/or appropriate to the trial crop?
Suitability of fieldPosition in field
Is the soil apparently uniform in terms of texture, depth, structure and drainage?
Is there water nearby that might lead to water-logging?
Is the field steeply sloping?
Are there features such as trees and hedgerows that might give rise to pest problems or effects such as shading or wind effects that might cause abnormal lodging?
Are there inoculated disease plots nearby that might give abnormal disease pressure?
Is the site free of problems from previous cropping e.g. volunteers, club root (in oilseed rape) or herbicide effects?
Are there genetically modified (GM) plants in the field or nearby?
Standard of drilling and field operations
Has a bordered drill been used? If so: is the inter-plot border width ≥ to the harvested plot-width?
Are there any interruptions in the plot drilling?
Are there consistent distances between neighbouring rows and inter-plot gaps?
Are tractor wheelings/tramlines at right-angles to the direction of plot drilling?
If the field is sloping, has the trial been laid out such that the plots are at right-angles to the contours?
Are there any staggered plot ends?
Drilled to plan? Were there any changes to the plan supplied and, if so, have these changes been relayed to the Trials Coordinator?
Plant population Does the plant population appear to be correct? If not the trial manager should be asked to conduct a plant count on controls varieties.
Are there buffers? Have buffers (i.e. between hybrids and open pollinating oilseed varieties) been drilled as required?
Weed control What is the size and population of weeds?
Are there any pernicious weeds such as Blackgrass, couch, wild oats, brome?
Have they been sprayed and, if so, has the herbicide been effective?
Are weeds competing or likely to compete with the crop?
Pest control Is there any damage by pests such as insects, birds, rodents, molluscs?
Is the crop significantly damaged and will it recover?
What measures are being/have been taken to minimise the problem?
Disease control If the trial should have been sprayed, does it appear to have been effective (is the level of any disease >10% in any plot?)
If there is an untreated trial in the field, compare the levels of disease between the treated and untreated plots.
6
Volunteers Indicate approximately how many volunteers are present by assessing in the interplot gaps.
How big are the volunteer plants? Are they likely to be suppressed by the crop or compete with it?
Is the volunteer population constant across the trial or do they appear in bands? If they appear in bands are they across the direction of plot drilling?
Conduct a count if there are greater than 10 volunteers per m2 and estimate the percentage ground cover.
Uniformity Indicate whether the trial is growing uniformly within the reps.
Indicate if there is any difference in growth between reps.
Indicate if there is a serious problem with specific plots (note which plots are affected).
Indicate if there is a problem with individual varieties (note which varieties are affected and inform the plant breeder/agent).
If there is any lodging indicate if it appears to be caused by differences in soil fertility or environmental effects rather than variety.
Any conflict of protocol?
Does the trial meet the protocol specification for soil type, rotation, sowing date or any other definition?
Are the harvestable plot dimensions acceptable?
Is the trial acceptable? The trial will be rated as follows:
Good: Evenly established well-grown trial that meets protocol requirements.
Satisfactory: Some problems, such as small areas of poor growth, missing plots or missing rows within plots. Some plots or parts of plot may need to be excluded but overall trials should provide satisfactory data.
Of concern: Larger areas of poor establishment or growth, affecting replicates. Disease levels >10% in fungicide treated trials. A second trial Inspection may be carried out to assess subsequent development. Requires careful validation at harvest.
Reject: Problems cannot be resolved.
3. Post-harvest statistical and technical validation of data
3.1. Preliminary validation of data - individual trials
Preliminary validation of data should be undertaken by the trial manager responsible for the trial. Checks should include that:
the correct character names and units have been used.
the trial ID is shown.
the minimum and maximum values recorded are sensible.
the date and growth stage been recorded.
These checks will also normally be conducted by the Data Analyst.
7
3.2. Non-yield dataDisease data:
Was the record taken at the correct time, as defined in the protocol?
Is the pattern and level of disease recorded unexpected?
If the disease recorded is in a fungicide treated trial were the levels >5% or >10% in any plot?
Agronomic data:
Is the pattern and level of the recorded characteristic unexpected?
If root and stem lodging has been recorded, do the totals add up to the figures given for the combined lodging% score?
Have ‘obligatory’ records been submitted?
3.3. Yield data
Yield is the most important validation character because it a composite indicator of the overall quality of data from a trial and therefore deserves the most careful scrutiny.
Are yield and dry matter data both available?
If a moisture analyser has been used for cereal trials are all plot dry matter values in the range 83-88%?
If an oilseed trial, have dry matters been conducted using the oven method? Moisture analysers may be used by
prior arrangement.
Have all plots or reps identified by the trial inspector as being of concern been adjusted for size or omitted?
Is the trial mean in the following expected ranges? If not, is it between 75% and 100% of the mean for the trial
series in that year?
Are there data available from at least two replications? If observations on a variety are lost for all but one
replicate, then the results for that variety are normally treated as missing for the trial, with the following
exceptions.
If a dry matter assessment is lost, can it reliably be estimated from the remaining observation(s)? If so, its value
may be replaced by an estimated value, e.g. the mean of the other assessments.
Where all or part of a plot area is lost, the standardised residuals should be checked carefully before deciding to
include it. If 50% or more of the plot area is lost, it should be excluded.
If more than one third of the plots in a replication are missing, the residuals should be scrutinised before
accepting data from the remaining plots.
If more than half of the plots in a sub-block of an incomplete block design are missing, the residuals should be
scrutinised before accepting data from the rest of the sub-block.
8
3.4. Data ranges
Most data received at harvest will be within the ranges shown below. If the data received are outside of these ranges they may be correct but should be queried.
Cereals
Crop Growth stage Dry matter Yield
Fung treated UntreatedWinter wheat 92-93 80-90% 6.5 - 13.0 4.1 - 11.1Late sown wheat 92-93 80-90% 6.0 - 11.5 -Spring wheat 92-93 80-90% 4.0 - 11.5 -Winter barley 92-93 80-90% 5.5 - 10.8 4.1 - 8.9
Spring barley 92-93 80-90% 4.5 - 10.5 3.5 - 8.5Spring oats 92-93 80-90% 3.0 - 9.0 2.5 - 9.0Winter rye 92-93 80-90% 5.5 - 8.5 -Winter triticale 92-93 80-90% - 3.5 - 9.0
Oilseeds
Growth stage
Dry matter Gross output
Seed yield Oil content
Fung treated Untreated
Fung treated Untreated
Fung treated Untreated
Winter oilseed rape 6.3-10.0 86-96 3.0 - 6.5 2.7 - 5.7 2.8 - 6.1 2.6 - 5.2
38.9 - 47.8 37.7 - 46.6
Spring oilseed rape 6.3-10.0 86-96 - 1.0-3.5 - 1.0-3.5 - 34.0-46.0
Spring linseed 90-99 86-96 - 1.0-2.5 - 38.5-42.0 -
3.5. Analysis of variance statistics:
Check for variance accounted for by the replicates and by the blocking for incomplete block designs. Investigate any large differences i.e. more than +/- 0.5 t/ha.
Coefficient of variation (CV)
The Coefficient of Variation (CV) (trial standard deviation*100/trial mean) may be low because of high yield and high because of trial variability or low yield, so the standard error (variety mean) should also be checked. Any trial should be further investigated if it has a CV% higher than given below. These CVs are for guidance only and are not strict thresholds for accepting or rejecting trials. Trials with higher CVs may be acceptable and trials with CVs lower than
9
the guidelines may be unacceptable. Any trial should be further investigated if it has a CV% >8% and/or a standard error of >0.4.
Winter wheat 7% Winter oilseed rape 10%Winter barley 7% Spring oilseed rape 10%Winter oats 9% Linseed 10%Spring wheat 7%Spring barley 7%Minor crops (cereals) 10%
Residuals
A residual is the plot yield with the variety and replicate effects removed. Check for any values which may indicate problems with individual plots or sets of plots. Using these residuals, plot yields are flagged for attention as follows:
? Moderately up or downx Worth looking at* Should be checked
Check the residuals for any large, flagged values, which may indicate problems. Do high residuals appear in patches or individual plots? The validator will decide, in consultation with the trial manager, if they are atypical and should be excluded from the summary report or reflect an aspect of the variety's performance that is valid for RL assessment.
Example reasons for high variation:
Soil effects.Pest damage.Lodging.Disease.
Shedding / ear loss.Sprouting.Combine losses.
Equipment faults.Human error.Chance.
Submitted data may offer a likely explanation of the ‘odd’ results or they may correspond to notes taken earlier either by the trial manager or inspector. If no obvious likely reason can be found, the trial manager should be consulted to see whether any factor has been missed which, if rectified, might increase the accuracy of the trial.
Variety F ratio.
The ratio of the variety mean square to the residual mean square in the analysis of variance is expected to be statistically significant at the 5% level or lower. However, trials with low numbers of varieties may not show statistically significant differences between variety means.
3.6. Statistical and technical validation following analysis – overtrial
Standardised residuals
Standardised residuals provide a means of identifying varieties or trials which are producing atypical results within a series and which should be analysed further. They will not be used alone to reject a trial and may help to identify situations where a variety is particularly adapted, or less well adapted, to conditions at a location. A pattern in the residuals for a variety can be linked with a feature of the site, e.g. water availability.
10
A residual indicates the extent to which a variety in a particular trial performs greater (+) or less (-) than expected as judged by its overall performance.
SD ratio column (1% probability)
Indicates variability of site within variety and depends on the number of varieties in the trial. The average is 1.0 and the following thresholds suggest sites that are more variable than ‘average’ and worth further investigation. The average is 1.0 and >1.3 is more variable than ‘average’ and worth further investigation.
SD ratio row(1% probability)
Indicates variability of variety within site. The average is 1.0 and >1.3 is more variable than ‘average’ and worth further investigation.
CV% A measure of the internal variability of individual trials. A high CV can result for legitimate reasons, such as low yields. >8% is worthy of further investigation.
Correlation coefficient (x100) A low correlation coefficient (<40) indicates that at that site, varieties are performing differently from the average.
Environment effect The amount by which site yield is different (+ or -) from the overall trial mean.
Summary
The balance of judgement is to accept trial data unless there is a clear reason to reject them.
Trials will not normally be rejected because they appear to be different to others in the series but the trial manager and validator will attempt to explain why such differences have occurred.
Any plant breeder/agent who is concerned over any trial data should advise the RL Trials Coordinator immediately.
Crop Evaluation will meet with representatives of BSPB if requested prior to final validation of a series. No BSPB member with a direct interest in the validity of any data will attend such a meeting without prior agreement but their views would normally be relayed via the appointed representative.
Ultimately the validation of a trial is a judgement, based on a number of criteria and the experience of the validator. A decision will be taken only after these criteria have been discussed with the trial manager.
Final judgement on the acceptability of all RL trials rests with RL Project Consortium Board.
11
Appendix 1a Inspection report form (field inspection) Cereals & oilseeds
National List VCU / AHDB Recommended List trial validation report
Trial inspection report - cereals & oilseedsCropYearOperator codeTrial CodeTreatmentTrial seriesTypeSite (Operator/County/Nearest settlement)OS Map Reference and/or postcodeTime of sowing requested / actualSoil type requested / actualRotation requested / actualType of drill used (border/direct, conv double-width) *If bordered, is the inter-plot border width ≥to the harvested plot-width? Date of InspectionName of trial InspectorName of trial contractor presentDoes contractor agree with report?
Management & husbandry checklist (assume all answers are 'YES' unless a comment is made):Does the trial meet protocol specifications (e.g. soil type)?Is the field suitable?Is the trial in a suitable position in the field?Is seedbed preparation & drilling of a good standard?Are the harvest plot dimensions OK?Have buffers been sown as required?Is the plant population OK?Is weed control acceptable?Is the trial free of pest damage?Are numbers of volunteers acceptable?If a fungicide required, has it been effective?
Current state of trialIs trial uniformity acceptable?Is rep uniformity acceptable?Are individual plots free of problems?Are all varieties free of problems?Is the trial acceptable on the day of the visit?
State of trial on day of trial inspection
Summary inspection report including reasons for concern or rejection
If trial is rejected and it is an NL or NL/RL trial, give the date on which BSPB and APHA notified
12
Appendix 1bPost-harvest validation Cereals & oilseeds
Post harvest validationCrop Year Operator code Trial Code Treatment Trial series Type Site (Operator/County/Nearest settlement) Is the plot length/width constant? Yes
Have actions suggested in the trial report been implemented? Yes
Was the harvesting method (swathing/direct) appropriate?* Yes
Have moistures been determined using the oven method? Yes
Is the yield / gross output within a sensible range? Yes
Is the trial free from seed loss, sprouting etc? Yes
If a +F trial, is it free of significant disease? Yes
Is it free of significant lodging? Yes
Statistical AnalysesFirst Run
Yield (t/ha) corrected for DM/MC% CV% =
Dry matter (DM)% or moisture content (MC)% Mean DM% for trial
CV% =
Seed yield* CV% =
Notes:
FINAL ANALYSIS
Yield (t/ha) corrected for DM/MC% CV% =
Dry matter (DM)% or moisture content (MC)% Mean DM% for trial
CV% =
Gross output (oilseed crops only) CV% =
Final trial validationIs the trial valid?
Reason if rejected
Is the non-yield data valid? If trial is rejected and it is an NL or NL/RL trial, give the date on which BSPB and APHA notified
Validator's name
Date of validation
13
14
