T-4-108: Efficacy data requirements for fertilizers and supplements regulated under the Fertilizers Act

August 2008

TABLE OF CONTENTS

• A) INTRODUCTION
  • 1. PURPOSE
  • 2. REGULATORY AUTHORITY
  • 3. SCOPE
• B) SOURCES OF EFFICACY EVIDENCE
  • 1. RESEARCH TRIALS
    • 1.1 Canadian Trials
    • 1.2 International Trials
  • 2. EXTRAPOLATION
  • 3. PUBLIC DOMAIN EVIDENCE
  • 4. PRIVATELY OWNED DATA
• C) TESTING AND PERFORMANCE STANDARDS
  • 1. GOOD EXPERIMENTAL PRACTICES
    • 1.1 Responsible Researchers
  • 2. LOCATION OF TESTING AND SITE SELECTION
    • 2.1 Physical Soil Characteristics
    • 2.2 Trial Site Selection
    • 2.3 Production Regions
  • 3. DATA REQUIREMENTS FOR PRODUCT REGISTRATION/APPROVAL
    • 3.1 Minimum Number of Trials
    • 3.2 Annual Trial Distribution
    • 3.3 Representative Crops
    • 3.4 Foreign Data
    • 3.5 Accelerated Trial Requirements
    • 3.6 Petitions To Remove Trials From Consideration
    • 3.7 Bridging Data
    • 3.8 Controlled Environment Testing
  • 4. TRIAL DESIGN
    • 4.1 Plot Size
      • 4.1.1 On-Farm Field Scale Trials
    • 4.2 Treatment Structures
    • 4.2.1 Reduced Pest Pressure
    • 4.3 Response Variables
• D) DATA ANALYSIS AND INTERPRETATION
  • 1. INTRODUCTION
  • 2. TRIAL PARAMETERS
    • 2.1 Level of Significance
    • 2.2 Degrees of Freedom
    • 2.3 Descriptive Statistics
  • 3. STATISTICAL TESTS
    • 3.1 Analysis of Variance
    • 3.2 Post hoc tests
    • 3.3 Non-parametric Tests
    • 3.4 Multiple Site/Trial Series Analysis
• E) FORMAT OF EFFICACY DOSSIERS
• F) MECHANISM OF APPEAL
• G) CONCLUSIONS
• H) CONTACT INFORMATION
• I) REFERENCES
• Appendix A: Efficacy Data Triggers
• Appendix B: Experimental Trial Designs

A) INTRODUCTION

The objective of the Fertilizers Act and Regulations is to ensure that all fertilizer and supplement products imported into, and sold in, the Canadian market are; safe for humans, plants, animals and the environment, efficacious for their intended purpose(s), and properly labelled. Therefore, efficacy assessments are a key element of the registration/approval process for fertilizers and supplements.

Efficacy is defined as the ability of a fertilizer or supplement to fulfil any label claims and to produce a desired or intended result based on the labelled guarantees and directions for use. This definition includes the ability to clearly demonstrate a benefit to the end user from the application of the product. The purpose of an efficacy assessment is to evaluate product performance in order to establish appropriate label claim(s), active ingredient guarantee(s), and usage pattern(s).

In addition to specific claims and guarantee(s), each usage pattern or direction for application on the product label will be treated as a claim, and must be supported by scientifically valid efficacy data. Anecdotal or testimonial evidence is not a scientifically valid form of efficacy data, and therefore will not be accepted to support product registrations or approvals.

1. PURPOSE

These guidelines are intended to outline the general requirements for conducting efficacy trials, and to describe the criteria and procedures for efficacy data evaluation. Furthermore, these guidelines also provide direction as to when bridging data, extrapolation and/or a scientific rationale are relevant or acceptable in order to reduce the level of efficacy data required to support product approval or registration.

2. REGULATORY AUTHORITY

The regulatory authority for the creation and implementation of efficacy requirements for the registration or approval of fertilizers and supplements is provided under Section 11.(2) of the Fertilizers Regulations.

3. SCOPE

Efficacy assessments are performed on all fertilizers and supplements evaluated by the CFIA, Fertilizer Section. However, not all products require the submission of efficacy data to support their registration or approval. Most supplement products require registration and the submission of efficacy data to support their acceptance. However, most fertilizers do not require registration, and also do not require the submission of efficacy data to support their acceptance. Please see Appendix A of this document for a complete listing of product categories/types that require efficacy data to support their registration or approval.

Upon reviewing the nature of guarantees or claims displayed on a product label, the CFIA, Fertilizer Section may determine that the submission of efficacy data is required for a specific product even though it does not appear in Appendix A of this document.

This document is intended to provide guidance to the applicant on the general requirements for the submission of efficacy data to the CFIA, Fertilizer Section. Please be aware that efficacy data requirements may differ based on product type or category. To ensure that the correct information and appropriate scope of efficacy data is submitted to the CFIA, please review all CFIA Trade Memoranda applicable to your product type or category.

B) SOURCES OF EFFICACY EVIDENCE

There are four main sources of efficacy evidence; research trials, extrapolation, public domain data and privately owned data.

The data and/or information submitted to support the efficacy of a product can be comprised of one or more of the following sources:

1. RESEARCH TRIALS

Statistical analysis generated from scientifically valid experimental trials conducted in accordance with the product's directions for use and consistent with the proposed label claims are considered direct evidence.

1.1 Canadian Trials

Research trials conducted in Canada, as described in Section C of this document can be submitted. Please note that a Research Authorization must be obtained prior to the environmental release of a novel supplement in Canada. Further information on Research Authorizations can be found in Trade Memorandum: T-4-103 - Guidelines for Research Authorizations for Testing of Novel Supplements.

1.2 International Trials

Research trials conducted outside Canada may be acceptable to support product registration/approval. The trial information required for international trials is the same as for Canadian trials (see Section C). International trials must be accompanied by an acceptable rationale clearly explaining how each foreign-derived trial supports the efficacy of the product for the proposed use in Canada. The rationale should detail the similarity between the foreign sites and Canadian agricultural production areas with respect to climate, soils, agronomic conditions (cultural practices, application methods, cultivars, planting and harvest dates) and other parameters that may affect the efficacy of the product. These parameters are dictated by the mode of action, usage pattern of the product, and intended crop species for application of the product.

2. EXTRAPOLATION

Extrapolation from indirect evidence may support an amendment to a registered product, or a new product that is similar in nature to an existing product from the same company that is supported by a full efficacy package. Examples of sources of indirect data and information include: previously generated evidence to support a registered product, data from efficacy trials that were not specifically designed to verify the proposed claims, and data and/or information from other sources such as scientific articles.

Extrapolation of evidence is based on the product's mode of action and use pattern in combination with the experimental design and measured parameters. Generally, extrapolating information from one product to another will support the same or reduced performance claims. Expanded performance claims will usually require further evidence.

The acceptability of extrapolated evidence to support an assumption of efficacy for a new product registration, or product amendment, is dependent on the quantity, quality and relevance of submitted information and the accompanying rationale. Factors that may need to be addressed for extrapolation include: crop type and physiology, agricultural practices, agronomic and environmental conditions, and the directions for use of the product(s). All extrapolation submissions are assessed on a case-by-case basis and their acceptance is determined on the merits and applicability of evidence submitted.

3. PUBLIC DOMAIN EVIDENCE

Public domain evidence includes peer reviewed published papers, conference proceedings, academic technical bulletins, and scientific texts. The onus is on the product proponent to adequately describe how the information supports the proposed claims on the product label. Summary tables and a scientific rationale citing the submitted references should be submitted to clearly link the information to the proposed claims. Each peer reviewed article submitted as evidence to support product efficacy must be accompanied by a summary that explains the comparability of the product as used in the research and the product proposed for registration/approval in terms of; formulation, active ingredient(s), crop type, application rate, method, frequency and timing. Additionally, it is the responsibility of the company to provide copies of all articles/text submitted as public domain evidence to support product efficacy.

4. PRIVATELY OWNED DATA

A product proponent can submit privately owned data from another party as evidence of product efficacy if the owner of the data provides written consent; this consent must be in the form of an original signed letter on company letterhead submitted with the application package. Alternatively, the third party may submit the data directly to the CFIA, Fertilizer Section, with the letter of consent. If the data was not generated with an identical product at the same use pattern to support the same claims as proposed for registration/approval, the product proponent must provide a rationale to support the extrapolation of supporting evidence. See the above sections on extrapolation and public domain evidence.

C) TESTING AND PERFORMANCE STANDARDS

1. GOOD EXPERIMENTAL PRACTICES

Although efficacy trials do not currently need to be conducted in accordance with Good Laboratory Practices, it is strongly recommended that all trials be conducted using Good Experimental Practices (GEP) as described below.

The primary goal of GEP is to ensure that all efficacy trials conducted are of a high quality and the results derived are reliable and can be used by different registration authorities for comparison in verifying product efficacy. GEP requires that quality trial management practices are implemented to ensure proper trial planning, execution, assessment, record-keeping, and interpretation of data. Other factors that are considered in GEP are; the qualifications of staff conducting and/or analysing the trials, the use of proper equipment, facilities, protocols, methods of application and data recording. Additionally, another important element is the establishment of internal procedures that provide verification of the proper use of GEP.

A summary of key GEP requirements:

• a thorough description of trial site where the experiment will be conducted
• a plan of the trial sites and surrounding areas
• trial conditions of the plot and crop should be included
• validity of choosing trial sites explained
• all trials should be conducted using the specific formulation and methods of application which are to appear on the product label
• alternative formulations/methods used at the trial site
• complete address and geographical coordinates of trial locations
• appropriate number of trials at proposed label rates in order to adequately demonstrate product efficacy

1.1 Responsible Researchers

As per the requirements of GEP described above, the researchers conducting the trials must be qualified scientists and/or technicians with the appropriate knowledge, qualifications, training, and experience to conduct agronomic experiments and data analysis in a scientific manner. These qualifications may be derived from formal education in agricultural science or a relevant field, appropriate training, and/or professional experience.

Research can be conducted by qualified personnel employed by the company seeking product registration or approval, or by an independent researcher (e.g. university, college, independent agriculture research contractor, etc.). For each efficacy trial submitted, a researcher must be identified who assumes responsibility for the validity and integrity of the trial, the data, and/or the analyses submitted. The efficacy data dossier sent to the CFIA, Fertilizer Section must include the name, contact information, and signature of the responsible researcher for all submitted trials.

2. LOCATION OF TESTING AND SITE SELECTION

2.1 Physical Soil Characteristics

Trial sites must be representative of the climatic and soil conditions intended for the end use of the product. A number of variables relating to the soil, such as texture, moisture content, fertility, organic matter content and pH, may measurably influence the efficacy of the product, especially in the case of a soil treatment. Thus, these factors must be considered in selecting the sites of the tests, and should be documented. Therefore, in many situations soil analysis for the trial locations must be provided, which may include the soil zone and type, percent organic matter, soil pH, nutrient content, and cation exchange capacity (CEC). Additional parameters of soil analysis may be required depending on the type or category of product being tested.

2.2 Trial Site Selection

Efficacy trials should be conducted in a variety of locations which span the environmental conditions, geographic regions, seasonal variation, soil zones, and agricultural production regions in which the product/crop combination are found and for which registration or approval is being sought. It is the responsibility of the product proponent to ensure that trials are conducted in the appropriate location(s). An adequate rationale referencing crop production statistics and/or climatic conditions must be provided in order for the trial locations to be considered relevant and acceptable.

2.3 Production Regions

In cases where >85% of the national crop production area is in one region (primary region), only two trials in each year of testing are required in the smaller or secondary region in order expand registration into the secondary region. Census data or the equivalent on the production area for the crops in question must be provided to support the reduced trial allowance in secondary use regions (<15% of National production). Additionally, data from non-Canadian sources (e.g. United States) may be submitted for this secondary region, accompanied by a rationale demonstrating the equivalence/comparability with the Canadian production region.

Examples of trial breakdown in support of national registration (12 trials overall, from two years for use on a specific crop):

Soybean

Production Statistics
Ontario 70%
Quebec 19%
Manitoba 11%

Trial Breakdown
A) Where ON/QC has been demonstrated to be a single production region:
Year 1: 4 trials ON/QC, 2 trials MB
Year 2: 4 trials ON/QC, 2 trials MB

B)
Year 1: 2 trials in each of ON, QC, MB
Year 2: 2 trials in each of ON, QC, MB

C)
Year 1: 6 trials in ON, 2 trials in MB
Year 2: 2 trials in ON, 2 trials in MB

Examples of trial breakdown in support of primary region registration (should be where >85% of production is in one region) (6 trials overall, from two years):

Soybean

Production Statistics
Ontario 70%
Quebec 19%
Manitoba 11%

Trial Breakdown
Where case has been made to consider ON/QC a single production region (i.e. 89% of soybean production):
Year 1: 3 trials ON/QC
Year 2: 3 trials ON/QC

3. DATA REQUIREMENTS FOR PRODUCT REGISTRATION/APPROVAL

The number of trials required to support product efficacy will vary depending on the claims listed on the label, the product type or category, the intended usage pattern, and characteristics of the active ingredient(s). For example, a product with a single active ingredient, intended to be used on a single crop with the claim of increased yield would generally require only a minimum number of trials (12). Please note, product proponents must ensure that the appropriate parameters and response variables are used in testing to support the claims, crops, usage pattern, etc. displayed on the proposed label. If the appropriate response variables and parameters were not employed, the product proponent may need to conduct additional testing. In general, all label claims must be adequately supported by scientific data that has been statistically analyzed and demonstrates a statistically significant benefit. For more information regarding data analysis, including acceptable levels of significance and multi-location analyses, please refer to Section D: Data Analysis and Interpretation.

For new uses and/or products, in general, a full data package is required as evidence in supporting efficacy claims and establishment of appropriate application rates.

For products intended for use on a crop grown in a limited production area (e.g., peaches), it will be necessary to consult with the CFIA, Fertilizer Section to determine the appropriate number and location of the field trials required to support product efficacy.

3.1 Minimum Number of Trials

A minimum of twelve efficacy trials are required to obtain national registration/approval, and a minimum of six trials are required for regional registration/approval. Trials should be conducted over a minimum of a two-year period to account for seasonal variability in environmental conditions such as soil moisture, humidity, precipitation, and soil and air temperature. As indicated above, additional trials may be necessary if a product has multiple claims, usage patterns and/or active ingredients. Additionally, companies may choose to conduct more than the minimum number of trials, to ensure that the number of trials demonstrating a statistically significant benefit is adequate to support product efficacy.

3.2 Annual Trial Distribution

The distribution of trials on a per year basis does not have to be symmetrical, so long as a minimum of two trials per region are conducted in each year of testing, resulting in overall achievement of the minimum required number of trials (i.e. twelve trials for national registration/approval and six trials for regional registration/approval).

3.3 Representative Crops

All crops that are being claimed to benefit from use of the product must be tested. If the intended crops for use claimed on the label are considered to be similar, then a) trials from similar crop species can be grouped together or b) a representative crop species can be chosen for testing and can be used to extrapolate the data to similar crop species. In both instances, an adequate rationale to establish the similarities between crops in terms of plant physiology and production patterns must be provided in order for the data to be considered relevant and acceptable.

3.4 Foreign Data

A maximum of 50% of the trials can be derived from non-Canadian sources, provided that the climatic conditions have been demonstrated to be similar, and a rationale is submitted that adequately explains the similarity or comparability between the foreign location and Canadian agricultural production regions (please see Section B.(1.2)).

3.5 Accelerated Trial Requirements

If a company wishes to conduct all efficacy trials in a single year to support product efficacy, the product proponent must increase the number of trials to a minimum of sixteen trials in order to support a national registration/approval, and eight trials for regional registration/approval. Additionally, in order to ensure that the product is subjected to a range of environmental conditions in the single year of testing, the product proponent must demonstrate that the trials are adequately spatially distributed within the production region(s).

For registration/approval of a product in a secondary region using a single year of trials, the number of trials required increases from two trials to four trials.

3.6 Petitions To Remove Trials From Consideration

During the course of conducting field trials, circumstances may arise which negatively impact the results of a trial to such a point that the trial should be removed from consideration in the evaluation of the overall efficacy of the product. Petitions for the removal of efficacy trials must be made before data analysis is conducted. In these instances, the product proponent must submit, in writing, to the CFIA, Fertilizer Section a petition for the removal of such a trial from consideration. The removal of a trial must be substantiated by documented evidence of the circumstances which the product proponent feels have negated the results of the trial. Petitions for removal of efficacy trials are assessed only on the merits of the evidence submitted.

Examples where the exclusion of a trial may be appropriate include severe drought, hail, flood, wind lodging, killing frost, severe pest damage, or human error in the application of treatments.

3.7 Bridging Data

The use of a reduced number of trials or laboratory studies to support expanded, or amended performance claims, is called bridging data. Bridging data can be used, in some cases, to support an amendment to a registered/approved product, or to support a product registration/approval that is similar to one or more of a company's currently registered products. The reduced number of trials will be acceptable when there is an appropriate amount of initial supporting direct or indirect evidence. Therefore, bridging data is generally acceptable if a full data package was previously submitted to the CFIA, Fertilizer Section, and deemed adequate to support the product registration/approval. Trials must be carefully planned to ensure they can support the intended use in combination with other information/data. It is recommended that the applicant consult with the Fertilizer Section during the trial design phase.

Examples where bridging data may be appropriate:

• New formulation (same active ingredient)
• Co-formulation of two or more registered/approved products
• Change in application method
• Change in application rate
• Change in volume
• Change in application technique
• Change in guarantee
• Change in fertilizer/supplement constituent
• Change in carrier
• Addition of a crop species

In general, only one year of testing is necessary where bridging data has been deemed adequate to support product approval/registration. Therefore, a minimum of three trials in a single year would be required for regional registrations/approvals, and six trials in a single year for national registrations/approvals. For information regarding controlled environment testing, please see section 3.8 below.

The structure of agronomic trials for the purposes of bridging may require the inclusion of the following treatments for comparative purposes: a) the original product/formulation, b) the amended product, and c) an untreated control.

3.8 Controlled Environment Testing

There are instances where controlled environment testing is sufficient to support product registration/approval, depending on the nature of the product claims and its intended use. Laboratory, growth chamber, or greenhouse bench trials (controlled environment tests) may be acceptable for proposed label uses such as indoor use, greenhouses and/or nurseries. Greenhouse, laboratory, and growth chamber studies conducted outside of Canada are deemed to be equivalent to controlled environment testing conducted in Canada if the studies were carried out under appropriate conditions. The number of trials required to support a controlled environment use pattern as described above is two trials conducted in each of two independent growing cycles for each of the crop species claimed on the label. For example, a supplement intended for use on greenhouse tomatoes would need a total of four trials; two trials in one growing cycle, and an additional two trials in the next growing cycle.

Controlled environment tests are very useful in preliminary screening of candidate products for; demonstrating appropriate application rates, choosing crops for testing, determining the most effective formulation, and proving the mode of action for an active ingredient. Nevertheless, data generated under controlled conditions may not be a realistic indicator of a product's field performance. Thus, in general, trials conducted under controlled conditions will not wholly replace agronomic field trial data. However, a company may request that the CFIA, Fertilizer Section accept a controlled environment trial (greenhouse/growth chamber) in replacement of a field trial when that field trial failed to perform due to environmental conditions or human error. Please consult the above section 3.6: Petitions To Remove Trials From Consideration for details on what is required for discarding trials from evaluation.

4. TRIAL DESIGN

Efficacy tests should be designed in such a manner that the appropriate statistical analysis of data can be performed in order to ensure that any differences observed are due to product treatments. Typically, a randomized complete block design or split plot design are employed in which the treatment plots are randomly distributed within each block. Other designs such as the completely randomized design may be used for trials conducted in completely homogeneous environments, such as greenhouses or growth chambers. It is advisable that an Efficacy Evaluator from the CFIA's Fertilizer Section be consulted prior to employing any trial design. Please refer to Appendix B: Experimental Trial Designs for additional information.

4.1 Plot Size

Determination of the most suitable plot size depends upon many factors, including the characteristics of the particular crop and the product, the method of application of the product, and the equipment used for treating and harvesting the crop. The individual plots should also be large enough to provide meaningful data when sampled, such that each plot adequately represents the effects of the factors or variables being measured. Guard rows or border plots are often included in order to lessen variability or treatment overlap, and potentially reduce drift of product or treatment from one plot to another. Guard rows are especially important if the plot size is small.

4.1.1 On-Farm Field Scale Trials

The CFIA, Fertilizer Section recognizes that small plot tests may not always be indicative of normal field-use conditions. Large scale farm trials may serve the useful purpose of demonstrating the performance of the product under varying commercial conditions using typical application equipment and techniques. Data obtained from large scale trials may be considered as complementary to small plot tests. Therefore the CFIA, Fertilizer Section will consider requests for the employment of large scale field trials only after one year of small plot research has been successfully conducted and product performance demonstrated (≥ 60% of trials). All large scale field trials must be conducted in accordance with acceptable experimental designs and must be properly analyzed. For more information, please review Section D of this document. For large scale trials that are intended to be conducted on novel supplements, please ensure that a Research Authorization has been obtained from the CFIA, Fertilizer Safety Office, as described in Trade Memorandum: T-4-103 - Guidelines for Research Authorizations for Testing of Novel Supplements.

4.2 Treatment Structures

The objectives for the trial and hypothesis, as well as the criteria by which they are to be assessed should be clearly defined prior to conducting the trial(s). Treatments in the trial should be selected carefully in order to meet the specified objectives. The timing of treatment application should be reported in the way most appropriate to support the objective of the trials; calendar date, developmental stage of the target crop, days after planting of the treatment site, or days after application of the treatment.

In order for a test to be scientifically valid, all treatments must be compared to one or more check treatments: (a) an untreated control; and if necessary (b) an appropriate comparative treatment (positive control).

4.2.1 Reduced Pest Pressure

In cases where a product contains an active ingredient with known pest control properties, it is necessary for the product proponent to ensure that the trial(s) are conducted in such a way as to remove the potential pest-suppression effect from consideration in the evaluation of the efficacy of the product. Generally, this can be done by conducting the trial under reduced pest pressure. The onus is on the product proponent to adequately demonstrate that the potential pesticidal properties of the product have been controlled for in the submitted trial(s).

4.3 Response Variables

Response variables, or sampling parameters, should be chosen in such a manner as to support the objective of the experiment and therefore the product label claims. In general, quantitative variables should be used (counts and measures) to assess responses to a treatment and allow for a parametric method of statistical analysis. Some examples of quantitative variables include; yield, plant height, root weight, plant weight, plant stand, nutrient availability etc. In some instances the trial objective does not support the use of quantitative variables, but is conducive to the use of qualitative parameters such as ranking and scoring (e.g. plant vigour, leaf colour, etc). Qualitative variable data should be analyzed using non-parametric methods.

D) DATA ANALYSIS AND INTERPRETATION

1. INTRODUCTION

The following is intended to provide general guidelines for conducting the data analysis necessary in supporting product efficacy. This is not a comprehensive review of all available statistical methods, but an overview of some common statistical tests encountered in the evaluation of fertilizer and supplement efficacy, and a brief description of the general suitability of these approaches. Overall, a case may be made to employ additional statistical tests not covered here, or to deviate from these guidelines regarding the circumstances in which a given test is applicable. In these instances, it is expected that the product proponent will be able to provide sound scientific evidence that the analysis conducted follows accepted statistical practices. It is also advisable that the product proponent seeks the advice of the CFIA, Fertilizer Section.

Generally, the practices of good science dictate that the method of statistical analysis should be considered prior to conducting the experiment. The decision as to the appropriate statistical test will be strongly dependent on experimental parameters that are encompassed in the experimental design, such as the type of parameter being measured (e.g. qualitative or quantitative), the number of replications, the sample size etc. Therefore, whenever possible, the statistical analysis should be set out before beginning the analysis.

In general, all label claims must be adequately supported by scientific data that has been statistically analyzed and demonstrates a statistically significant benefit.

2. TRIAL PARAMETERS

2.1 Level of Significance

It is an accepted convention in the biological sciences to employ a 95% probability (p ≤ 0.05) for hypothesis testing. The CFIA, Fertilizer Section recognizes that there is no universal agreement on the correct standard of significance used in agricultural research; in published agronomic research a p ≤ 0.10 level of significance is not uncommon. However, from a regulatory perspective, rigorous precision and an appropriate level of certainty is required to determine whether a product is efficacious or not. As such, data in support of product efficacy should be analysed at the 95% probability (p ≤ 0.05). Recognizing that in agronomic research, experiments are often conducted at levels of significance other than p ≤ 0.05, the CFIA, Fertilizer Section will accept data analysed up to a 90% probability (p ≤ 0.10) if accompanied by a rationale indicating why this is an appropriate level for the testing conducted.

2.2 Degrees of Freedom

The degrees of freedom (d.f.) are the number of values in the final calculation of a statistic that are free to vary, and can be calculated from the sample size. The residual d.f. should be sufficiently large to conduct a useful statistical analysis (i.e. with adequate power). Generally, agricultural science research relies on a minimum of 12 degrees of freedom (e.g. four sites with two treatments and four replicates). Degrees of freedom can be increased by increasing the number of replications or the number of sites.

2.3 Descriptive Statistics

Two classes of analytical tests are parametric and non-parametric. Data can be considered non-parametric if the response variable utilized in the trial is qualitative in nature (e.g. a vigor index), or if the data resulting from the trial does not fit a parameterized distribution (e.g. normal distribution), and cannot be transformed. Generally, nonparametric tests have less power than the appropriate parametric tests, but are more robust when the assumptions underlying the parametric test are not satisfied.

Descriptive statistics should be employed prior to analysis, to ensure that the data fits the assumptions required by the test method (e.g. normality and equal sample variance).

3. STATISTICAL TESTS

3.1 Analysis of Variance

An analysis-of-variance (ANOVA) is, with few exceptions, the most appropriate statistical method for determining if there is a statistical difference between agronomic data sets. The assumptions that must be satisfied in order to perform an ANOVA are that; experimental units are applied independently, there is a normal distribution of errors, and, variances between samples are equal. The product proponent may be requested to demonstrate that these assumptions are met. If these assumptions are not met, various data transformations may be undertaken, an alternative model may be necessary, or the data set may be rejected.

In some instances, a case may be made to pool data across treatment groups. As described below, good statistical practices generally dictate that the statistical analyses are set out prior to beginning the data evaluation, which avoids the possibility of having the data influence the choice of statistics, and thus the significant outcome, if any. Therefore, data pooling may increase the likelihood of making a Type 1 error. If data pooling is conducted, the applicant must provide statistical evidence that there is no significant interaction between factors, and a rationale explaining why pooling was necessary.

For two-way ANOVAs, F tables should be presented in order to verify that there is no interaction between main effect factors.

A significant F value only reveals that the treatment means are not equal. Post hoc tests (i.e. multiple comparison procedures/means separation tests) must be used to determine which treatment means are significantly different from each other.

3.2 Post hoc Tests

There are two general categories of mean separation methods: a) tests which control for comparison-wise (pairwise) error rates (e.g. Fisher LSD) and b) tests which control for experiment-wise error rates (e.g. Scheffe, Tukey). The least stringent but most powerful test is Fisher Least Significant Difference test which is effective when applied to a small number of preplanned comparisons. However, with poor control of experiment wise error rates, this test is most likely to result in a Type 1 error when used to analyse experiments with several treatments. Generally, Fisher LSD is not an acceptable means separation test for agronomic data when comparisons between means will exceed four treatments. When employing means separation tests for experiments that exceed four treatments, a separation test should be chosen that controls for experiment-wise error rates.

3.3 Non-parametric Tests

The non-parametric equivalent of the ANOVA is the Kruskal-Wallis test, a one-way analysis of variance by ranks that makes no assumption of normality or equal sample variance. Generally, non-parametric tests rank the data points from low to high and then analyze the ranks. These tests generally lack statistical power with small sample sizes, however when the assumptions of an ANOVA are not met concerning normality of data, and the data cannot be transformed, it may be advisable to use a non-parametric test.

3.4 Multiple Site/Trial Series Analysis

The advantage of combining trials to form a trial series is a direct increase in the number of replications and thus, additional degrees of freedom; increasing replication results in greater precision. This is especially relevant where the differences in the benefit (e.g. yield) expected by individual treatments are relatively small (< 10%). The data from the trial series (consisting of multiple experiments at several sites, and possibly over multiple years if testing was conducted at the same locations) may be analyzed at the end of each crop season, or over all the seasons, to examine the treatment x site interaction effect and the average effects of the treatments over homogeneous sites (generally employing a randomized block design). In effect, by combining multiple sites into a single data set and analyzing as a multiple site/trial series evaluation, it may be possible to evaluate the data with less emphasis on random effects (years or locations) and obtain a better assessment of the fixed effect (treatment).

The analysis of multiple sites combined into a single data set may include: an additional variable within any growing season (i.e. location), analysis of a single site over time (i.e. growing seasons), or combinations of both (multiple sites over multiple years). Statistically, these factors will be analyzed as main effects, and should reveal site to site variation. This variation is to be expected if sites are separated geographically, environmentally and/or by time.

Multiple site analysis must be conducted and presented to the CFIA, Fertilizer Section in the following manner:

• An analysis of variance should be conducted for each individual experiment presented.
• A combined homogeneity test should be conducted across all experiments.
• A combined analysis of variance across all experiments should be conducted. This analysis should test the significance of location x treatment interaction, and report the level of significance for treatments.

E) FORMAT OF EFFICACY DOSSIERS

Submissions for product registration/approval that contain efficacy data must be submitted in an organised and complete format. To facilitate the timely review of efficacy evidence by the CFIA, Fertilizer Section, the following information should be provided, in the described format.

• Overall summary
• Hypothesis / Study objective
• Table of contents
• Mode of action of active ingredient
• Experimental methods/ Standard operating procedures for laboratory studies (when applicable)
• For each trial (when applicable):
  • Trial name
  • Trial identification number (submission control number if trial conducted in Canada under a Reasearch Authorization)
  • Trial objective
  • Trial location
  • Active ingredient(s) and formulation(s)
  • Responsible researcher and organization
  • Experimental design (including size, number of plots and layout)
  • Crop species tested
  • Soil factors (e.g. type, organic matter, pH) and analysis if required.
  • Application date(s)
  • Application rate(s) (application method, equipment)
  • Growth stage of crop at application
  • Information on other applications to the plot (e.g. fertilizer, pesticide, adjuvants, surfactants, maintenance activities)
  • Weather conditions during the trial(s) (e.g. temperature, rainfall)
  • Sampling information (dates, parameters)
  • Statistical analysis
    • data summary table
    • statistical method used
    • stepwise presentation of analyses conducted (for multi-site analysis)
    • statistical outputs
      • treatment means, standard deviation/coefficient of variance
      • normal distribution
      • homogeneity of variance
      • ANOVA tables
      • f value
      • means separation tests
    • supporting raw data
  • Trial summary and conclusions relating to the label claims

F) MECHANISM OF APPEAL

When a product proponent disagrees with the findings of an efficacy evaluation conducted by the CFIA, Fertilizer Section, they may apply in writing for a review of the initial evaluation. This review will be conducted by either the Chief of the Efficacy Data Unit, or the National Manager of the Fertilizer Section. If, at the outcome of the review process, it is decided that an additional efficacy evaluation is warranted, this evaluation will be completed by a separate evaluator, by the Chief, or by the National Manager.

G) CONCLUSIONS

The efficacy data requirements set out in this document are intended to be comprehensive and applicable to all fertilizer and supplement products for which efficacy data is required. The CFIA, Fertilizer Section will also be developing product-specific efficacy guidelines which will more narrowly define the efficacy requirements for certain product types whose mode of action is well-characterized. This risk-based approach to efficacy data requirements will result in a subset of efficacy requirements; as the product-specific efficacy guidelines are developed, applicable references will be added to this document, and Appendix A: Efficacy Data Triggers will be updated as required.

H) CONTACT INFORMATION

If you have any questions about this document, please do not hesitate to contact the Fertilizer Program.

Mail:
59 Camelot Dr.
Ottawa, ON
K1A 0Y9
E-mail: fertilizer@inspection.gc.ca
Phone: 613-221-7519
Fax: 613-228-4552

I) REFERENCES

Canadian Fertilizer Products Forum, Efficacy Task Force (2007) Efficacy Recommendations.

Cochran, W.G. and Cox, G.M. (1992) Experimental Designs. John Wiley and Sons, Inc. USA.

Crop Protection Association (2002) Climatic Justification Document: A justification for climate comparability to allow the use of efficacy trials data from defined areas of continental Europe in support of applications for authorisation in plant protection products in the UK.

European and Mediterranean Plant Protection Organization (2000) Efficacy evaluation of plant protection products. EPPO Standard PP 1/214(1): Principles of acceptable efficacy.

European and Mediterranean Plant Protection Organization (2003) Efficacy evaluation of plant protection products. EPPO Standard PP 1/152(3): Design and analysis of efficacy evaluation trials.

European and Mediterranean Plant Protection Organization (2003) Efficacy evaluation of plant protection products. EPPO Standard PP 1/181(3): Conduct and reporting of efficacy evaluation trials, including good experimental practice.

European and Mediterranean Plant Protection Organization (2003) Efficacy evaluation of plant protection products. EPPO Standard PP 1/223(1): Introduction to the efficacy evaluation of plant protection products.

European and Mediterranean Plant Protection Organization (2003) Efficacy evaluation of plant protection products. EPPO Standard PP 1/224(1): Principles of efficacy evaluation for minor uses.

European and Mediterranean Plant Protection Organization (2003) Efficacy evaluation of plant protection products. EPPO Standard PP 1/226(1): Number of efficacy trials.

European and Mediterranean Plant Protection Organization (2005) Efficacy evaluation of plant protection products. EPPO Standard PP 1/241(1): Guidance on comparable climates.

European and Mediterranean Plant Protection Organization (2006) Efficacy evaluation of plant protection products. EPPO Standard PP 1/152(3): Design and analysis of efficacy evaluation trials.

European and Mediterranean Plant Protection Organization (2007) Efficacy evaluation of plant protection products. EPPO Standard PP 1/257(1): Efficacy and crop safety extrapolations for minor uses.

Government of Canada, Canadian Food Inspection Agency (1997) Trade Memoranda T-4-108: Guidelines for Field Testing Supplements.

Government of Canada, Health Canada, Pest Management Regulatory Agency (1996) Regulatory Directive DIR96-01: Guidelines for Efficacy Assessment of Fungicides, Bactericides and Nematicides.

Government of Canada, Health Canada, Pest Management Regulatory Agency (2003) Regulatory Directive DIR2003-04: Efficacy Guidelines for Plant Protection Products.

Government of the United Kingdom, Department for Environment, Food & Rural Affairs, Pesticides Safety Directorate (2000) Efficacy Guideline 102: Efficacy Overviews: A Guide to Applicants.

Government of the United Kingdom, Department for Environment, Food & Rural Affairs, Pesticides Safety Directorate (2000) Efficacy Guideline 111: Guideline on Statistical Analysis of Efficacy Data.

Government of the United Kingdom, Department for Environment, Food & Rural Affairs, Pesticides Safety Directorate (2001) Efficacy Guideline 604: Efficacy Data Required for Tank Mixtures and Sequences of Pesticides.

Government of the United Kingdom, Department for Environment, Food & Rural Affairs, Pesticides Safety Directorate (2002) Efficacy Guideline 208: Seed Treatments – Efficacy and Physical/Mechanical Data Requirements.

Government of the United Kingdom, Department for Environment, Food & Rural Affairs, Pesticides Safety Directorate (2003) Data Requirements Handbook Chapter 8: Guidance on Efficacy Requirements.

Government of the United Kingdom, Department for Environment, Food & Rural Affairs, Pesticides Safety Directorate (2004) Efficacy Guideline 115: Guidance on Numbers of Trials in Target Crops for Demonstration of Efficacy and Crop Safety.

Government of the United Kingdom, Department for Environment, Food & Rural Affairs, Pesticides Safety Directorate (2005) Efficacy Guideline 103: Efficacy Requirements for the Re-registration of Products Containing Existing Active Substances After Inclusion in Annex I.

Government of the United Kingdom, Department for Environment, Food & Rural Affairs, Pesticides Safety Directorate (2005) Efficacy Guideline 104: Guidance of Efficacy Sifting of New Active Substance Submissions.

Government of the United Kingdom, Department for Environment, Food & Rural Affairs, Pesticides Safety Directorate (2005) Efficacy Guideline 211: Home Garden Products.

Government of the United Kingdom, Department for Environment, Food & Rural Affairs, Pesticides Safety Directorate (2005) Efficacy Guideline 406: Data Requirements for Turf Fungicides, Herbicides, Insecticides and Plant Growth Regulators.

Government of the United Kingdom, Department for Environment, Food & Rural Affairs, Pesticides Safety Directorate (2005) Report to the European Commission: Proposals for Extending and Harmonizing Efficacy and Crop Safety Extrapolations to Reduce the Need for Efficacy Trials on Minor Crops.

Keuhl, R.O. (1984) Statistical Principles of Research Design and Analysis. Duxbury Press, Belmont, California.

Peterson, R.G. (1994) Agricultural field experiments: design and analysis. Marcel Dekker, Inc. New York, New York.

Stoskopf, N.C. (1981) Understanding Crop Production. Reston Publishing Company, Inc. Reston, Virginia.

Snedecor, G.W. and Cochran, W.G. (1980) Statistical Methods. Iowa State University Press, Ames, Iowa.

Appendix A: Efficacy Data Triggers

Products that do NOT require Efficacy Data to support approval or registration

Fertilizers:

• Soil applied chemical N-P-K fertilizers (with or without micronutrients) and not comprised of new slow or controlled release technology
• Lesser plant nutrients; Calcium (Ca), Magnesium (Mg), and Sulphur (S)
• Products that meet section a) of the definition of a Specialty Fertilizer and not comprised of new slow release technology
• Fertilizer-Pesticide combinations as found in the Fertilizer Pesticide Compendium
• Organic and organic based N-P-K fertilizers (e.g. processed waste) without supplement claims
• EDTA chelated micronutrient fertilizers
• Micronutrient fertilizers in sulphate form
• Micronutrient fertilizers in oxy-sulphate form (>80% sulphate form)

Supplements:

• Products used only to correct soil acidity or alkalinity (e.g. liming materials)
• Soil amendments represented to improve the condition of the soil by increasing organic matter (OM)
• IBA within currently approved ranges

Products that REQUIRE Efficacy Data to support approval or registration

Fertilizers:

• Foliar applied N-P-K fertilizers
• Low Analysis Farm Fertilizers (Mixed grade chemical fertilizers less than 24% N-P-K)
• Non-EDTA chelated micronutrient fertilizers (e.g. HEEDTA, DPTA, CDTA, etc.) or micronutrient products that are claimed to be complexed
• New Slow/Controlled Release coatings/constituents for N-P-K Fertilizers (Polymers, Urease Inhibitors, etc.)
• CFIA and PMRA approved Pesticides for use in Fertilizer-Pesticide combinations, but currently not found in the Fertilizer-Pesticide Compendium
• Micronutrient fertilizers in oxide form
• Micronutrient fertilizers in oxy-sulphate form (<80% sulphate form)
• Seed applied nutrients
• Products represented as sources of plant nutrients and contain an element other than; N, P, K, Ca, Mg, S, B, Cl, Fe, Mn, Mo, Zn (e.g. Si, Va)

Supplements:

• Plant growth regulators (e.g. NAA, Kinetin), including IBA if not within the currently approved ranges
• Signalling compounds (e.g. LCO's, Hesperetin, Naringenin)
• Wetting Agents and Surfactants
• Soil Flocculants/Stabilizers and water holding polymers(e.g. anionic polyacrylamides)
• Any product containing micro-organisms (e.g. Rhizobia, Mycorrhizha, Penicillium sp.)
• Inoculant Extenders (products represented to improve inoculant on-seed survival)
• Any product represented as a “Bio-stimulant” (e.g. vitamins, enzymes)
• Products represented to improve the physical condition of the soil other than fibrous organic matter.
• Supplements sold only for the correction of soil acidity or alkalinity other than Ca, Mg, or S.

Appendix B: Experimental Trial Designs

The choice of experimental trial design is dependent upon the objectives of the trial, the practicality of implementation, environmental conditions, application methods, crops, and treatments. All these factors should be carefully considered before choosing an appropriate design that will fulfill the objectives of the experiment in demonstrating product efficacy. Below are descriptions of various trial designs used in agronomic research and information on their appropriate usage:

Completely Randomized Design (CRD)
In this type of design, all treatments are assigned to experimental units (plots) in a truly random manner without restriction. Any treatment is equally likely to be assigned to any plot; therefore this type of design can be statistically very powerful because it allows for maximum degrees of freedom. However, this type of design is only suitable for trials conducted in homogenous environments (e.g. greenhouse, growth chamber, and laboratory). This design should not be used when significant heterogeneity may exist in the trial and can result in a high level of experimental error.

Randomized Complete Block Design (RCBD)
The randomized complete block design is utilized most often in agricultural research. In the RCBD, plots are randomized and uniformly assigned within blocks. Treatments are assigned to plots so that each one will appear an equal number of times within a block, usually only once per block, and therefore each block is a replicate. The layout of blocks within the design is influenced by the characteristics of the testing site such as nutrient or water gradients. One of the advantages of this type of design is that it can remove one source of variation from experimental error and therefore increase the overall precision of the trial. Additionally, any number of treatments and blocks can be used so long as each treatment appears the same number of times within the block.

Latin Square Design
This type of design is very similar to the RCBD except that it controls for two sources of variation. The structure of this design is such that a treatment will appear only once in each row and column of the experiment. Although this design can be very effective in controlling two sources of variation, it has one significant disadvantage; since each treatment can only appear once per row and column, the number of plots is the square of the number of treatments (4x4, 5x5…10x10). Consequently, the practical use of this type of trial is limited to between four and ten treatments.

Split Plot Design
In general, multifactorial experiments are carried out using a randomized complete block design, with each treatment combination (AxB, AxC, BxC, etc.) assigned once per block. However, there are instances in which the practical considerations of treatment application make it convenient to have one factor applied to a large plot, and another factor applied to smaller sub-units within the large plot. In the split-plot design, the levels of one factor are randomly assigned to large plots, and the second factor is applied to smaller plots within the large plot. The larger units are referred to as main plots and the smaller units as split-plots (sub plots). The primary advantage of this design is that it can be a very efficient use of the trial site when the factors allow for the use of different plot sizes. Additionally, the split-plot design can improve estimation of factorial effects.

Strip Block Design (Strip-Plot Design)
This type of design is a variation on the Split Plot Design with the principle difference being that a larger plot can be used for application of the different factors. One of the advantages of this design is it facilitates the use conventional agricultural equipment in the application of treatments. For example, levels of one factor such as seeding rates, can be applied in strips using one direction of the field. Then application of a second factor (e.g. foliar application of various product including an untreated control) can be applied in strips perpendicular to the direction of the first factor.

Field Scale Trials
The premise of field scale trials is to engage the active participation of the agricultural producer in testing new technologies within the context of an “on-farm” production system. Typically variables that may affect the outcome of a trial are not as well controlled in comparison to traditional agronomic research methods. As such, field scale trials complement, but do not replace mainstream agronomic research methods for demonstrating product efficacy ( 4.1.1 On-Farm Field Scale Trials). Due to some of the constraints and limitations of Field Scale Trials, enough sites or locations should be chosen to detect differences between the treatments in the trial and ensure sufficient error degrees of freedom in the analysis of variance. As such, at minimum, 11 different locations should be chosen for On-Farm Field Scale Testing. It is advisable that the product proponent seeks the advice of the CFIA, Fertilizer Section prior to designing a testing protocol using Field Scale Trials.