- What prompted Seed Terminator to invest in testing of other mills?
The success of the mill industry making a difference to sustainability of farming relies on the mills actually being able to kill weed seeds. The challenge for the consumer is that it is not easy to determine if it's working or not, so data was needed. The testing protocol had already been ratified through SAGIT funding Trengrove Consulting and the University of Adelaide Weed Science Group to do a thorough evaluation of the Seed Terminator in a whole range of weed species. To benchmark other mills in the market using the ratified testing protocol was an obvious next step that our research partners were really keen to conduct. As it was beyond the scope of the project, it meant a new application for funds. It takes a huge amount of time to go through the grant application process, right through to getting the results reported and by that time, technology is likely to have changed. To get some cut through we just funded the work and got it done. Going forward, we believe that the mill industry needs to take a legislative brand approach, where manufacturers take responsibility for their own validation and push the industry forward for the betterment of sustainable agriculture; it shouldn't be up to farmer levy funded organisations to hold manufacturers accountable.
- How should we interpret the numbers? (new question or put in additional comments)
Kill is calculated as a percentage relative to an unprocessed batch of seed germinated in the same soil/chaff environment. It represents a level of control only relevant to the seed type and the actual batch of seed processed. The actual % kill can be a little misleading. 90% sounds quite good. The challenge is that we are not controlling this year's plants, we are controlling next year's plants. 1 ryegrass plant may have 200 seeds. So if a mill processes just one plant's worth of seeds with 90% kill rate - that one plant could turn into upto 20 plants next year. By comparison 99% kill rate could turn into upto 2 plants next year. This effect compounds over time, if you look at the effect of a mill alone after three years that one plant could go from 1 - 20 (y1) - 400 (y2)- 8000 (y3) plants @90% kill Compared to 1, 2 (y1), 4 (y2), 8 (y3) plants @99% kill. It is like having an interest rate of your seedbank of 2000% compared to 200%! This is agriculture so there are a whole heap of factors that make this number purely theoretical but it does illustrate the compounding effect of survivors.
There is the argument that you only capture a % of weeds with the front which I have bought into for a long time. But the “aha” moment was talking to Josh Lade in Canada about wild oats. Josh said, I know that wild oats shed a lot of seeds before harvest but then why do they always grow in header trails? Turns out the answer is that it is only the seeds that are captured that are spread, the ones that are shed stay put in existing patches. The seeds that are spreading are the ones causing more damage to more area of the crop in the years to come. We found some research by Steve Shirtliffe from the University of Saskatchewan looking at wild oat dispersal by the combine that showed how the combine would spread the seeds up and down the paddock a long way. Typically this is due to the tailings returns system on the combine that cycles weed seeds and can mean seeds are transported 100s of meters down a paddock.
- Are there any variables that may have influenced results and could those learnings be used for future test protocol?
First and foremost seed species is the most important key parameter. Seeds are of all different shapes, sizes and structure and therefore require totally different amounts of energy to control with a mill. The ST kill results that the University of Missouri finds in waterhemp or the Swedish University of Ag Science finds in Canola have almost no relevance to a farmer in Australia trying to kill annual ryegrass. In Australia, Annual ryegrass (Lolium Rigidum) has been a major focus for testing, firstly because of its prevalence as a weed species, secondly because it is bloody hard to kill with a mill. This is due to its high fibre to starch ratio (meaning it is very tough) and because it is very light (hitting it doesn’t provide a lot of energy for seed damage).
*picture of seed size comparison 
After seed species, seed weight is the next most critical because batch to batch there is a big variance in seed weight. The seeds used in this testing had a 1000 gram weight of 2.8 g/1000 seeds - this is on the heavy side for annual ryegrass seed. For comparison, my PhD had seeds of 2.17 g/1000 seeds. The energy available to damage a seed in a mill is proportional to its mass (E=1/2mV^2). Heavy seeds are easier to kill. If we had used a lighter batch of seeds we would see lower kill levels. You can’t take a test using one batch of seed and compare it to another test using another batch of seed. Since we haven’t figured out how to get standardised seed of the same size and strength, internally we use benchmark mills in our procedure to understand the relative difference between designs (so we know differences in kill are not due to the batch of seeds being a different weight).
Other factors include chaff type, chaff flow rate and moisture content. Interestingly, of the studies I am aware of, the impact of mill wear on kill has not been statistically significant for iHSD or ST (the difference is smaller than the variation in results so that we can’t tell if mill wear decreases kill, increases kill or stays the same).
The method of mounting a conveyor in the combine was very effective and means that lab level control of material flow rate can be achieved by commercial machines. The iHSD was difficult to feed as the auger would throw the material everywhere so a feed chute had to be manufactured. We are unsure if this would have resulted in additional seed damage through having tighter clearances or not.
- The report specifies a test protocol that deviates from previous bench testing – how well do the results align with bench testing data?
The protocol used was developed during the SAGIT project taking on the lessons from Mike Walsh’s extensive work as well as the extensive knowledge of the University of Adelaide Weed Science Group (and a lesson or two from my PhD). It has been shown to be highly repeatable across a range of weed species and we use it every year for validating new mill technologies benchmarking against existing (to take out seed weight issue).
The results here using the soil bin method align well with testing done by Mike Walsh comparing Ihsd (95%) and ST (99%) using the agar germination method using a different batch of ryegrass seeds. The main issue with the agar germination method is the huge amount of labour needed to find the individual dyed seeds. The time and cost of doing this method limits the amount of testing you can do. I have seen that Mike has also moved towards the soil bin method.
- Is Seed Terminator open to working with other manufacturers and Weedsmart to develop and ratify a standard test protocol?
Very interested in looking at ways to standardise testing as it would be a great asset to the mill industry. It is likely to be someone's whole PhD given the number of variables (variation in seed weight being one of the main ones)… the side by side comparison like this one and the one Mike Walsh did, is so powerful compared to a claimed kill % because it takes the variables between tests away and provides an accurate comparison with statistical significance for the particular seed batch used.
While we come up with a standardised test, I would encourage all manufacturers to take the method and results from this report. Improve their mill technology and then use their existing mill as the benchmark to compare to their improved technology. When there are new manufacturers or significant changes to existing manufacturers' mills we would be happy to be involved in benchmarking again.
- Is there any other comment ST would like to make regarding the data or future testing?
I would very much like to thank the many groups who have been involved in my PhD and then testing Seed Terminator and have ultimately moved the space forward over the years. Including but not limited to Trengrove Consulting, The University of Adelaide, University of Hohenheim, University of Missouri, University of Western Australia, University of Sydney, University of South Australia and the Swedish University of Agricultural Sciences.