02/06/2026
Part of last week's trip was all-paddock testing on a couple of dairy farms I work with 🐄. I'm not one to jump the gun when it comes to testing as it’s the quality of the data that's just as important than the number of samples. The biggest issue is trusting the data that has already been collected; typically as a 'box ticking' exercise rather than focus on statistical validity or applicability. With all-paddock testing a big investment, focusing on a selection of representative paddocks to assess for soil & corresponding plant nutrient availability is a good first step. When done well, despite being ‘selective’, the dataset is a valuable tool to form the basis of nutrient plans on, quickly & cost-effectively.
Selection protocol is based on influencing factors such as soil type, texture, topography, paddock history, irrigation management etc.
Once the major system limitations have been identified, all-paddock testing can be an asset to dive into the detail further & confirm or alter the previously stipulated treatment boundaries.
Whether all-paddock or selective testing, the data is only as good as the sampling. I’d argue the ‘text book’ instructions of only collecting 20-25 soil cores or 250-500gms straight away reduces statistical power in that sample. That is, the ability of that sample to correctly detect true effect or true average is not great.
If there’s 1T of topsoil per hectare & we’re only sampling 500gms, that’s 0.05% of what’s potentially in that one hectare. We then apply that result beyond one hectare, we may even apply it to 100Ha or more in the case of hill country sheep & beef testing & suddenly the probability of what’s in that 500gms being applicable & representative to the larger scale becomes very small (the sample volume < 0.00000053% of the area applied to).
You should sample as much as you physically can within the bag provided (or send bigger bags!). Each sample I collect will typically be >4Kg, still a very small volume relative to the bigger picture but nearly 10-fold more than what the ‘standard’ is. The more sample, the more area covered, the more robust your sample is to anomalies or ‘hot spots’ that we cannot visually see.
All paddock testing is not going to have the same value to all systems & logistically, no top dresser will want to do a different blend for every paddock so a bit of commonsense applies but regardless of testing protocol, the same importance for statistical & sampling awareness apply.
📊 All of the statistical principles apply to ALL testing, soil, herbage, feed, animal…the data is only as good as what’s going in & if you are basing your livelihood’s profits on that data, your sure as 💩 want that data to be as accurate as possible.
It is not uncommon for fertiliser reps or testing contractors to only walk 50m, sample 250-500gms soil, not pay attention to the volume in the auger & in some cases, even put their car in drive while they stick the auger out the door every now & then 😲 The industry as we know it does not promote good data.
The actual volume collected in each auger sample is another very important part of accurate testing. If the auger volume is not flush from top to bottom (this is the calibrated volume) or there are stones etc. once again the sample & it's meaningfulness will be a waste of time & money.
What needs to be kept in mind is the fact fertiliser reps are not paid for soil testing, they're paid to sell product so there's no doubt that this incentive (or lack of) impacts sampling quality. Just as worringly is the fact soil contractors are typically paid per paddock, not for time so taking their time to collect quality data is not an incentive for them (& the fact the fert reps don't change what they sell to you regardless of soil tests takes away the importance of the accuracy anyway!).
The photo is one of the farms I visited last week, way in the distance if you squint you can see my car🚶♀️➡️ Walking the length of paddocks in a 'W' to ensure the sample at the outset is as good as it can be for the farmer's, animal's & environment's benefit.
