Features Agronomy Tillage
The predictive ability of the plant bioassay

Many growers, retailers and agronomists have relied on a plant bioassay to
detect herbicide residues in soil samples. Developed by the Alberta Research
Council (ARC), the bioassay is a qualitative predictive tool, designed to determine
potential re-crop injury risk resulting from soil residual herbicides, and is
currently one of the only formal diagnostic tools available to assist with cropping
decisions.

New research, however, calls into question the ability of this tool to accurately
predict potential crop injury. Two independent field studies, one by Agriculture
and Agri-Food Canada (AAFC) and the other by BASF Canada, were conducted in
Manitoba, Saskatchewan and Alberta during the 2003 to 2005 period to validate
the accuracy of the residual herbicide bioassay. In both studies, bioassay results
were compared with field rotational cropping studies for soil residual herbicides.

Eric Johnson, weed biologist with AAFC, led the first study in collaboration
with Alberta Agriculture, Food and Rural Development. "We undertook this
research to determine the accuracy of the bioassay because we know that many
growers use this tool to help guide their cropping decisions," explains
Johnson. "The condition of the soil is an important factor in the success
of a crop, and any steps that can be taken to avoid crop injury are key."

The AAFC research was conducted on soil sites across Saskatchewan, including
Scott, Melfort and Vanscoy, where herbicide stacking trials had been conducted.
Both Scott and Vanscoy are Dark Brown soil zones, while Melfort is Thin Black.

BASF used existing rotational cropping sites established in 2002 and 2003 across
five locations for its research: Morden, Manitoba; Vanscoy and Estlin in Saskatchewan;
and Lethbridge and Nisku in Alberta. These sites represent Black, Dark Brown
and Brown soil zones, as well as a variety of environmental conditions: from
severe drought to above-normal rainfall.

While there were slight differences in the AAFC and BASF methodologies, in
both cases soil samples were collected from the respective fields in the manner
required by ARC, then sent to ARC for bioassay analysis.

Field ratings were quantitative visual ratings taken at three assessment dates.
Field crops were considered injured if a visual injury rating of greater than
15 percent was recorded on any single evaluation date. The bioassay results
returned from ARC, which evaluates for symptoms consistent with herbicide damage,
included photographs of the bioassays and an accompanying letter with visual
observations of the plant responses. The results of both the plant bioassay
and the field ratings were then compared and categorized into three possible
outcomes:

• Corroboration – field and greenhouse bioassay results are the same;
• False positive – injury occurs in the bioassay but is not observed
  in the field;
• False negative – no injury is observed in the bioassay but is observed
  in the field.

Results compared
When the data were aggregated, AAFC's research found a greater proportion of
false positives than false negatives, whereas BASF's research revealed a greater
proportion of false negatives than false positives.

The chart (Figure 1) shows the percentage of agreement between the canola bioassay
and field results from samples where injury occurred in the bioassay and/or
the field. While corroboration in the Scott soils (which one highly responsive
to herbicide residue) was high at 71 percent, AAFC's two other research sites
produced false positive results in 35 percent of the samples and false negative
results in 18 percent. Conversely, only seven percent of BASF's samples resulted
in false positives and 64 percent produced false negatives.

According to the researchers, the discrepancies may be accounted for by soil
variations, slight differences in sample collection, or variation in the behaviours
of the chemistries used in the fields (the BASF study focussed exclusively on
the imidazolinone chemistry, while the AAFC study included other chemistries).

Implications for growers
While the results of the research showed that the bioassay provided a reasonable
level of accuracy in predicting injury on the Scott soils, and in predicting
sugar beet (a very sensitive crop) injury in the field, the frequency of false
positive and false negative results for all other soil types and crops was considerable.

"What the results of this research really mean for agronomists and growers
is that if they proceed with a plant bioassay, there is a significant chance
that they may be making cropping decisions based upon inaccurate information,"
says AAFC's Johnson. "While a cropping recommendation based on a false
positive reading poses no risk to a grower from a crop injury perspective, it
may represent the loss of cropping options and income potential. A cropping
recommendation based on a false negative, though, could represent serious risk
or liability, as considerable crop injury or loss may occur."

Paul Watson, a research scientist with ARC, says that a bioassay rarely produces
black and white results; there is a lot of gray area between no damage and significant
damage. Just because an ARC bioassay showed minimal damage in canola does not
necessarily mean that it is safe to plant canola. "You have to take other
risk factors into account," Watson says.

Growers and agronomists are advised to consider numerous factors when determining
which crops to seed, including previous herbicide use on the field, soil conditions
(such as pH, organic matter, soil texture and soil moisture), historical weather
conditions (drier conditions inhibit the breakdown of herbicide residues in
the soil) and herbicide label recommendations.

Bioassays are not perfect, Watson acknowledges, but he maintains that in the
hands of knowledgeable agronomists and agrologists, they do provide a useful
tool for growers. "We pass on the results we find and then the agronomist
or agrologist makes a recommendation to the grower on what he should or shouldn't
plant based on the local conditions," Watson says. For example, "In
the Peace River area, it's my understanding that if a bioassay shows minor damage
on sugar beets, many agrologists won't recommend canola either. The soil pH
in that area is low and growers have had enough problems in the past to know
that if sugar beets are showing damage there is a real risk for canola too."

While the bioassay is rarely used (nor should be) in isolation, the question
arises, however, as to what value it provides. Based on the results from two
studies, the bioassay was shown to give a correct result less than half the
time for most soils.

"An accuracy reading of less than 50 percent certainly doesn't instill
confidence. With all the information and experience we have regarding the behaviour
of residual herbicides, and taking into consideration associated soil, herbicide
and agronomic conditions, fairly accurate predictions can be made about the
potential for crop injury without the bioassay," concludes Johnson. "I
have received feedback from experienced agronomists who believe a bioassay to
be a helpful tool; however, those with limited experience and knowledge in herbicide
carryover need to understand the likelihood of it producing inaccurate results."
-30-