So, I know. I’m doing this ass-backwards. First, I was supposed to have all this lead in, background on the project, telling you all about the details. But too bad – we’re already getting set up to get started so I’ll just have to add all that in later, rearrange the posts into a logical order, and go from there.
Also, a caveat. I’ve been tinkering around in permaculture and regenerative agriculture and trying to mix it up alchemically with my scientific background. I’ve been “designing” and “mapping” for 4 years post-PDC (plus a small fortune’s worth of every other workshop I could get myself into). This year, after taking Darren Doherty’s Regrarian/REX training, I realized very, very clearly that I could just start all over from scratch: download and install Google Earth Pro (dump Google Earth), learn QGIS (because it’s widely used, free, and has great online support) and Adobe Illustrator (which no one can afford but, well, hey, we all have friends), and hack it for real on a real adult-sized science project. So here we go.
The contour maps I’ve started with are through MapPass, a paid service that provides topographical maps for the US. So these are starting out with 20 ft contours, which is brutally crude, but… it’s a start. I’ve kept overlaying it with Google Earth Pro (from now on, GEP) to check the contours and how the layout is coming and so far – it’s okay. Except that I generated the original map image with a slant on the terrain so the border is a little off…. Whoops. I can redo it fairly quickly now though with all the other GEP pieces in place.
Caveat: I am learning this as I go. I have all kinds of reasons and excuses that I don’t know this better by now. Mostly I blame getting too much credit early on for being a faster learner, smarter, or more knowledgeable than I actually knew I was. I let it stall me, thinking I must just have a self-esteem issue or something. Now I’m starting over. So this is going to be kind of a real beginner’s hack into mapping a property with the purpose of designing both a functional agricultural system (on a very simple level; no fancy hyperdiverse food forestry swale mazes here) and a statistically testable, adaptively managed experimental system.
I’ll repeat that Google Earth Pro is now a free download, in case you don’t already have it (most reading this will). It will read .kmz and .kml files, which are basically folders that are packages of individual files that make up the map, in case you didn’t know that already. I “learned” that during my Masters 6 years ago. I finally understood it about a year ago. Now I’m actually using the information.
On to the maps.
The green blocks outline the flatter areas on the ridges from roughly 1140 ft (max elevation) down to about 1080 ft, minimizing slope where I have to cross contours. They are far from perfect but are a starting guideline to keep us off the steeper slopes, which we can figure out how to deal with as well.
I wanted to be able to specify exactly where my sampling points would be on the map, and for that, I thought it would be cool to generate a true random sampling from an independently generated grid. From years of ecological field work, working within others’ sampling strategies and generating my own, I know how easy it is to get led off-track in the field. Laying it all out beforehand is my way of nailing myself to a plan – which I can deviate from, if needed, but at least I’ll know I have a reasonable plan to fall back on that will satisfy the core questions I need answered.
But how to get those sampling points on a map? I Googled how to make a grid in GEP (GEPaths looks like a great tool… if you run Windows. I don’t). There are all kinds of hacks. Nothing I had the patience to grind through. And then I found the answer – an online software that generates points on a custom grid or random pattern. Better yet, it’s generously provided *FREE* by the Mangrove Lab, who provide clear PDF instructions for creating the points in the context of environmental science research. I very quickly and easily created a grid of 50 m x 50 m sampling points for three sampling blocks (Block1, Block2, Block3). (I went with metres because it’s my default, as a Canadian, and because that’s what most of the rest of the world outside the US uses. But it would take me about 5-10 minutes to recreate them in feet, if needed, which it may well be, because having contours in feet and a sampling grid in metres is ridiculous.)
Here’s the overview: For each block I generated a polygon in GEP; saved it as a .kml, which was uploaded to KMLTools; chose my grid specifications; saved the newly generated .kml file with the grid points; uploaded it to GEP; and then cleaned it up by de-selecting those points that lay outside the ridge blocks, where (for this discussion, at least) I want to focus my sampling efforts.
Each file might have taken me about 1-2 minutes to work with and I spent about 5 minutes reading the PDF and learning to use KMLTools. Really that easy.
So now for the most important part: the Research Objectives.
These are the main questions we’re looking to answer. Pretty straightforward. For the record, I think that Richard Teague and colleagues are right – most “intensive grazing management” and “holistic management” and “rotational grazing” studies are poorly designed, shoe-horned into a statistical framework that utterly ignores the realities and intricacies of adaptive management.
That said – I sympathize (to some degree) with the need for control-treatment type comparisons of grazed vs. ungrazed areas, in the same study area, under the same conditions, to really get a handle on how much (or if) we can impact soil restoration objectives using holistically managed/intensively managed/rotationally grazed livestock.
NB: Yes I understand the difference between intensively managed, holistically managed, and rotationally grazed livestock. Others have written about the nuances elsewhere (watch for a future blog post and links). So. The questions:
– How do grazing, compost application, and their combination (“treatments”) impact the soil?
- macro and micro nutrients
- hydrology (water retention and percolation depth)
- physical characteristics (texture, bulk density, compaction)
– How do treatments impact vegetation?
- native diversity returning in grazed vs. ungrazed areas?
- community shifts; functional traits (based on species IDs)
- nutrient quantity and quality; availability as forage
– Are there observable effects in tree health?
- measure with MultiSpeq in leaf nutrient, photosynthetic rate
- foliar nutrient analysis (lab-based)
Future areas for questions:
– Animal Health and Behaviour:
- behaviour (vegetation response to grazing; animal response to vegetation changes)
- parasites and health metrics
- supplement needs (minerals, protein, etc.) and linkage with vegetation metrics
- note changes in water cycle (greater soil retention, decreased run-off, changes in erosion patterns)
- water quality monitoring of ground water
- isotopic tracing of nutrients through soil -> vegetation -> animals
- nutrient quality of animal products
- additional crops, pollinator strips, diversified agroforestry, etc. plans to come…
And here’s what our data sheets will tell us:
- sample ID (use Amanda’s method or create our own, e.g. PlotID_ddmmyy)
- electronic file ref. – macro and micro nutrients (especially C and N breakdown; analysis of K to address lack of reference in soil literature); microbiology; bac:fungi
- bulk density, % coarse fragments (> 2mm), depth
- time of sampling
- percolation depth
- sampling day characteristics: ambient temp, soil temp, humidity
- sampling site characteristics: vegetation density, height, plant ID
VEGETATION (per soil sample)
- sampling day characteristics: ambient temp, soil temp, humidity
- plant ID (ref field guides used)
- Composite versus individual samples? depends on sample costs, time available, resources. Ideally I’d like to have samples from at least two depths (e.g. 10 cm and 40 cm) to understand how carbon and other nutrients are transported down through the soil profile over time, through vegetation, etc.
- Simple random, stratified random, or systematic sampling? We’ll be going with systematic sampling within bounded areas, although we may apply stratified random for grain areas and slopes where there are clear differences from our “ridge block” grazed areas. … This won’t make a lot of sense to people, or any sense to a lot of people, so I’ll dig into it in another post.
The brown areas on the right are the existing grain fields, which may or may not be open for grazing and sampling.
The small red area on the far east edge is the home site as it extends into the orchard property.
From these we can choose a representative number of samples that works with the time/budget/resources we have. From each block, we’ll keep a control that remains ungrazed and without soil amendments. There are several ways to do this: (1) randomly choose a control block of 20 x 20 m; (2) slice the end off of each block as control; or (3) some other method. Keeping the layout fairly straightforward and simple to visualize will be important for future grazing contractors/operations to be able to easily cordon off and/or move animals around.
The other option is just to shoot transects across each plot so there are, at minimum, sets of samples with each of Western slope – Ridgeline – Eastern slope accounted for.
From each soil sample a vegetation analysis will be done, e.g. within a 5 m radius, or 5 or 10 m2 plot around: height, density, identity, and potentially a clip for biomass estimation.
I’ve got a lot of papers still to comb through, focusing this time on Methods and Analyses – what size plots did they use? How many samples? What kind of analyses? What problems did they run into? Many of these I’ve read and reviewed before for conclusions, background info, and further references, but this will be another chance to dig deeper and really look at their experimental and statistical methods.
Let the good times roll!