The camera pans over a vast, treeless plain, mountains hazy in the distance, suitably stirring music setting the tone. A small child appears, doing childlike stuff. A beat, for the viewer to drink in the awesome scenery, before the words appear: exhorting the viewer to plant trees, with the logo of a well-known charity.

Charity websites proudly display counters with the number of trees planted. They number in the millions. Photos of singing and dancing villagers are shown with their hundred thousandth or their millionth tree. "Deforestation" proclaim the sites "ruins livelihoods and contributes to global warming. We can fix it".

Schoolchildren celebrate arbour day by planting trees in their schoolyard. They are taught from an early age how important trees are to the environment, and how, if we are to save the biosphere, we must plant trees, more trees, millions of trees. These kids are the next generation; when they grow up, they will understand the importance of forests and trees and they will make a difference. They will influence policymakers and set up charities of their own.

All this has some ecologists very, very worried.

Forests around the world are threatened with habitat destruction, and reforestation efforts are a major tool in conservationists' toolbox to restore degraded forest ecosystems. Reforestation efforts are generally intended to provide employment and business opportunities for local communities, restore forest ecosystems, and sequester carbon for carbon credits.

All laudable goals. But what about when forest restoration efforts harm other ecosystems?

That's the question posed by Professor William Bond in a paper in the journal Science.  Bond, an ecologist at the University of Cape Town, has studied grasslands, savannas, fire, grazing and biodiversity for decades. And he was deeply concerned by the implications of a new global reforestation initiative.

Bond looked at the recently-announced Atlas of Forest and Landscape Restoration Opportunities, an ambitious and much-needed effort by the World Resources Institute to map degraded forests globally. They have identified 1.5 million square kilometres of land for restoration by 2020, much of which, Bond argued, may, in fact, be natural grassland.

To the frustration of grassland ecologists, grasslands have frequently been overlooked in the international consciousness as places worthy of, well, anything. But grasslands store more than a third of the world's terrestrial carbon stocks (almost all of it underground), and hold a huge proportion of global terrestrial biodiversity, second only to tropical forests. Yet to many people, and particularly policymakers and the public, they are simply tidy places to plant crops or build suburbs. Many natural grasslands are viewed by influential thinkers as "degraded forest", partly because it can often be difficult to tell the difference between ancient grassland and grassland that was once forest.

The question that Prof. Bond asks in his paper is: how much of that 1.5 million square kilometres is, in fact, degraded forest, and how much is ancient natural grasslands?

There is still a lot of work to be done to understand grasslands globally. But ancient grasslands can be identified by a range of clues, including the number of endemic species (species that occur in a particular location and nowhere else), and the evolutionary adaptions of those species to current ecological conditions. For instance, fire has been a major driver in grassland ecosystems for millions of years, and many of the plants and animals that occur there are well adapted to regular fires. In fact, many plants will disappear in the absence of fire. They need the sunlight shining on the soil surface, and often the smoke and heat of the fires themselves to germinate. Their entire structure is adapted to being grazed, to surviving drought and frost and floods and flames.

So Bond's concerns are real. If the Atlas project identifies areas of natural grassland as "degraded forest", huge amounts of resources could be misdirected towards transforming an intact and healthy grassland ecosystem into an artificial forest ecosystem.

Encouragingly, the authors of the Atlas project, in their response to Bond's concerns, invited grassland ecologists to collaborate with them to refine the model. But at some point, models reach their limits and knowledge and experience take over. Once projects get to the point of actually preparing to plant trees, they will need experienced local biologists to guide project managers in making the final assessments about a particular patch of land.

For a long time, grasslands were thought to be ecosystems that wanted to be forests, but had had their dreams dashed by fire or by humans who had harvested all the timber.

Certainly, grasses and trees have engaged in a millennia-long tug-of-war, with trees alternately advancing into and retreating from grasslands as environmental conditions shifted and changed. Indeed, in the modern era, encroachment of woody plants into grasslands is a subject of major concern to conservationists, and a major subject of research for ecologists the world over (and the subject of a future blog).

Despite all the uncertainties, we have enough knowledge now to be able, with a reasonable degree of certainty, to determine whether a particular area of grassland has a long evolutionary history as a grassland, or whether it is a degraded ex-forest that should be rehabilitated by planting trees. We can survey the vegetation and determine whether there is a high proportion of grassland endemics; we can examine historical records and aerial photos; we can compare the physical environment to well-studied ecosystems with similar rainfall, soils and topography. And ultimately, we can make a judgment based on evidence and experience about what our vision for the ecosystem should be.

Intact, natural grasslands play a vital role in our biosphere, far more so than most people realise. Forest rehabilitation programs are hugely important for restoring degraded forest ecosystems, but a lot more consideration needs to be given by activists, policymakers and funders to the role of grasslands and to ensuring that forest rehabilitation efforts are focused where they are needed - in forests - and do not encroach into grasslands.

Photographs are an important tool for consultants and researchers. They are part of the data collection process and provide evidence of the state of vegetation or of specific features of importance in the landscape, like a rare species or illegal dumping.

Geotagging photographs is the process of adding a latitude and longitude to the photograph. All photo files come with an associated metadata file; the most common is the EXIF format, although there are others. If you look at the properties of one of your photographs, under the "details" tab, you will see a range of information about the photograph, right down to the focal length, f-stop and shutter speed, make and model of the camera, and of course the time the photo was taken (for some reason, in Windows, the time the photo was taken is labelled under "Date modified" not "Date created").

If your camera has a built-in GPS or you have a GPS accessory for your camera, you can automatically geotag your photos. Modern cellphones come with a geotagging option (you need to turn on the Geotagging in the camera settings). But what if you don't have that tool, and you have a batch of several hundred photos that need geotagging? You can't manually geotag them all in Picassa or Google Earth.

There are several software tools available online that can make the process a lot easier. The one that I've been using is a free tool called ExifTool. I'm going to explain how I use Exiftool to geotag several hundred photographs at a time, using separate GPS trackfiles.

The tool is a command line tool, meaning that you'll have to go back to good old typing instructions into the command line window of your computer (just go to the Windows search and type in "CMD", and you'll be able to open the command-line interface). For those of you old enough to have forgotten how DOS worked, too young to have heard of DOS, or not savvy enough for Linux (like me), you'll need to do a bit of scratching around online for the procedures.

The ExifTool website describes exactly how to use the tool, but you might need a bit of a primer before you can get started. I'm going to do my best to put together that primer.

Note: the advice below only applies to Windows.

1) Download the ExifTool.exe file and place it in your C:\windows directory
2) Open the Command line (go to windows search, type "CMD" and the window will open)
3) In the command line, change directory to your path. That should be something like C:\users\alan (to change directory, type cd "C:\users\name". In my case, it's cd c:\users\alan)
4) Now you're ready to use the ExifTool in the command line.

 
To geotag your photos, you need a GPX track file from your GPS. If you've been working in the field with a GPS, and the track log is running, you can download the GPX file. Depending on the date that you saved the file, it should have a filename like TRACK_2015_11_07 123423.gpx (the date and time it was saved). You also need at least one or two reference photographs - photos where you know the precise location of the image. You need to apply a GPS time stamp to the photo (which is separate from the camera's time stamp), which I'll explain below. Remember the filename of the reference photo (e.g SAM_1603.jpg). In my case, I renamed the reference photo to GPS_234.jpg for simplicity.

Your GPS time stamp records the time in UTC (formerly Greenwich mean time). In South Africa, that means that your GPS time will be two hours ahead of the local time. Don't worry too much about that.

For some time now, I've been taking the occasional picture of my GPS itself, with a waypoint marked. I usually do this to quickly provide a reference for a photograph of a key location or object. So if I see a rare plant, I'll photograph the plant and then mark the location of my GPS and photograph the face of the GPS immediately. This, completely by accident, provided a handy tool for synchronising my GPS points with my photographs using ExifTool, which I'll come to in a moment.

The other way you can synchronise your photos is by picking one image where you know the location, and then finding that location on your GPX tracks or waypoints (just upload the GPX file to Google Earth, pick the spot where the photo was taken, and pick the nearest trackpoint or waypoint. By nearest, I mean "on the spot where the photo was taken"). Record the waypoint name or trackpoint number.

Now:
1) Move your GPX file to the same directory where you are working in the command line (in my case, c:\users\alan). Change the name of the file to something simple like "track_log.gpx"
2) Move your folder with the images to the same directory. We'll call the folder "project_photos" (the underscore keeps things simple in the command line)
3) Open your GPX file in a text editor, like Notepad. You will see a wall of text describing all the points.

All you need to know is that each waypoint opens with <wpt> and closes with </wpt>. If you are using trackpoints, the tags will open and close with <trkpt> and </trkpt>.

Handy tip: if your GPS data is stored in more than one file (e.g. if you worked over more than one day and saved your waypoints and trackpoints as separate GPX files) you can open all the files in a text editor, then copy and paste all the trackpoints or waypoints from the second and subsequent files to the first file, after the last </wpt> or </trkpt> tag and before the final </gpx> tag.

4) So find the name of your reference point. It will be between the "name" tags, like this: <name>234</name>.
To avoid confusion, you might want to copy and paste the entire waypoint into a blank text editor:

<wpt lat="-23.126471" lon="27.980537"><ele>826.274414</ele><time>2016-02-23T12:13:21Z</time><name>234</name><sym>Pin, Green</sym></wpt>

If you look at the line, you will see that there is a wealth of data. The latitude and longitude, the elevation, the date and time, the name of the waypoint, and what symbol is used to represent the waypoint on your GPS (in this case, a green flag). The one piece of information that is relevant to the next step is the time; this waypoint was recorded on 23 February 2016 at 12:13:21 UTC.

5) In the command line, type exiftool -gpstimestamp="12:13:21" c:\users\alan\project_photos\GPS_234.jpg

The Exiftool has applied the GPS timestamp to the image. Now your image has two important data: the time recorded by the camera, and the time recorded by the GPS at the same location and time. Apart from the time zone difference, there will be a difference between the accuracy of the camera and gps times. This time difference is used to apply the correct GPS time stamp to every image in your folder, and interpolate the location of the photograph from the times in the GPS track log.

6) In the command line, write
exiftool -geosync="c:\users\alan\project_photos\GPS_234.jpg" -geotag track_log.gpx "c:\users\alan\project_photos"

This tells the tool to use your reference image to synchronise the camera and GPS times (-geosync) and then interpolate locations from the track log to all the images in the folder (-geotag)

Now Exiftool will apply the Geotag to every image in the folder.

7) Check the accuracy of one or two reference photos, where you know the exact GPS coordinates. Right-click on the relevant image, click "properties" and click the "Details" tab. Scroll down until you see the GPS coordinates. They will be presented in Deg min sec format. Refer to the locations of those images (on Google Earth, in your notes or on your GPS waypoints) and check whether they are accurate.

You also need to beware of gaps in the GPS log. If you switched your GPS off at some point (say while you were driving to the next location) and took some photos en route, those photos may be inaccurately geotagged. Scan through your photos with Picassa, with the Geotag window open, and check if they seem to be correct. If not, you may have to do some troubleshooting or remove the incorrect images from the folder.

If they are accurate, then you are done. You can move all your project files and folders back to your project folder. However, there is one more step that you may find useful - providing a map of all of your images.

1) Create a new folder for the images you want to include in your report or on your map, and copy or drag your chosen images to the new folder
2) Place the new folder in the c:\users\name directory
3) Copy and paste the following text into a text editor (from the ExifTools geotag page):

#------------------------------------------------------------------------------
# File:         kml.fmt
#
# Description:  Example ExifTool print format file for generating a
#               Google Earth KML file from a collection of geotagged images
#
# Usage:        exiftool -p kml.fmt FILE [...] > out.kml
#
# Requires:     ExifTool version 8.10 or later
#
# Revisions:    2010/02/05 - P. Harvey created
#               2013/02/05 - PH Fixed camera icon to work with new Google Earth
#
# Notes:     1) All input files must contain GPSLatitude and GPSLongitude.
#            2) For Google Earth to be able to find the images, the input
#               images must be specified using relative paths, and "out.kml"
#               must stay in the same directory as where the command was run.
#            3) Google Earth is picky about the case of the image file extension,
#               and may not be able to display the image if an upper-case
#               extension is used.
#            4) The -fileOrder option may be used to control the order of the
#               generated placemarks.
#------------------------------------------------------------------------------
#[HEAD]<?xml version="1.0" encoding="UTF-8"?>
#[HEAD]<kml xmlns="http://earth.google.com/kml/2.0">
#[HEAD]  <Document>
#[HEAD]    <name>My Photos</name>
#[HEAD]    <open>1</open>
#[HEAD]    <Style id="Photo">
#[HEAD]      <IconStyle>
#[HEAD]        <Icon>
#[HEAD]          <href>http://maps.google.com/mapfiles/kml/shapes/placemark_circle.png</href>
#[HEAD]          <scale>1.0</scale>
#[HEAD]        </Icon>
#[HEAD]      </IconStyle>
#[HEAD]    </Style>
#[HEAD]    <Folder>
#[HEAD]      <name>Waypoints</name>
#[HEAD]      <open>0</open>
#[BODY]      <Placemark>
#[BODY]        <description><![CDATA[<br/><table><tr><td>
#[BODY]        <img src='$directory/$filename'
#[BODY]          width='$imagewidth' height='$imageheight'>
#[BODY]        </td></tr></table>]]></description>
#[BODY]        <Snippet/>
#[BODY]        <name>$filename</name>
#[BODY]        <styleUrl>#Photo</styleUrl>
#[BODY]        <Point>
#[BODY]          <altitudeMode>clampedToGround</altitudeMode>
#[BODY]          <coordinates>$gpslongitude#,$gpslatitude#,0</coordinates>
#[BODY]        </Point>
#[BODY]      </Placemark>
#[TAIL]    </Folder>
#[TAIL]  </Document>
#[TAIL]</kml>

Save the file in the same directory as your photos, as kml.fmt (you need to change the .txt extension to .fmt)

Personally, I don't like the big clunky Google Earth flag marker (in the original code), so in the code above I replaced it with a circle icon.

4) Type

exiftool -p kml.fmt "c:\users\alan\image_folder" > out.kml

The tool will generate a kml file in the same directory that you can show in google earth with the filenames of each image as labels.

Firstly, your filenames all have the .jpg extension, which makes them unnecessarily messy.

1) Go to the kml file, right click, click open with... and choose a text editor (e.g. Notepad).
2) Do a find and replace. Find ".jpg</name>" and replace with "</name>". Save
Now all the .jpg extensions have been removed from the labels. Remove your kml from Google earth and then replace it.

But what if your filenames are clunky and inefficient, with lots of "SAM_1702 Themeda triandra possibly with some hyparhennia in the background.jpg"?

There's a simple way to rename all your photos simultaneously. This is where I will usually work with a copy of the photos for the report, rather than the original files, because I use my clunky filenames for searching for images.

1) Open the folder with the photos
2) Order them by the "date taken" column (in the menu, click on the "view" tab, choose "Details" then click "add columns" then check the "Date taken" column. NOT the "Date Created" column)
3) Sort in ascending order by Date taken.
4) Select all (Ctrl+A)
5) Right-click the first image, click "Rename" and rename them all (I just use a space). All the photos will be renamed with the same filename followed by a number in brackets. With a space as the name, the photos become (1).jpg, (2).jpg, and so on.

Now run the KML format tool as described above, and remove the .jpg tags from the labels, as described above. You will now be able to produce a google earth image with all your photos labelled as points (1)-(n).

Your photos can then be archived, included in the report with tag numbers, or sent to the client along with the kml file so that each photo has a reference point. In addition, the photos themselves are geotagged, and you can use Picassa or any number of other tools to view and locate your photographs.

The ExifTool website provides detailed instructions for a range of functions that can be performed with the tool - the methods I've outlined were the ones that, by trial and error, worked best for me. But there are many other ways of achieving the same goals described on the website, and my methods may not even be the most efficient. Read through the site and the instructions, and see what else is possible.

It's official - 2015 was the worst rainfall year on record in South Africa.

The current drought is hitting farmers hard, particularly in the already-dry interior of the country. And as usual, we wait until the disaster has hit before making any plans for how to avert the next one.

South Africa is already a dry country, and is subject to substantial fluctuations in rainfall from a variety of drivers, including the current El Nino. One recent paper by Justin du Toit of the Grootfontein Agricultural Research Institute in the Karoo and Tim O'Connor of the South African Environment Observation Network showed how rainfall at one site was subject to cycles laid upon cycles laid upon cycles. They found evidence of an approximately 20-year cycle, superimposed on another cycle of between 44 and 70 years. How these cycles will be influenced by climate change is difficult to predict, but most researchers agree that we will see more extreme events - more severe (and longer) droughts interspersed with more intense floods.

More importantly, droughts often last several years, and the cumulative effect of a multi-year drought can be far more devastating than a drought lasting just one year.

So, with or without anthropogenic climate change, southern Africa is a drought-prone region. And yet we always seem to be taken by surprise when a drought hits.

The most severe impact of drought on livestock farmers is, of course, fodder shortages. The lack of moisture causes substantial reductions in forage production, and animals starve to death in their thousands. I've seen plains covered with  bare soil and carcasses and the stench of rotting flesh is not an experience one forgets.

So how does a farmer avoid this situation the next time around?

Creating a fodder reserve is one of the most important principles of livestock farming. It goes without saying that animals need to eat year-round, but food doesn't naturally grow all year.

The fodder reserve is, simply put, a bank of stored food put aside for the lean times. That bank can take the form of veld set aside for winter, as well as hay, pastures, and other forms of intensively managed fodder. For this discussion we'll focus on veld reserves.

Veld reserves come with their own problems. In the sourveld parts of the country (high-rainfall areas with acidic soils where the protein content of the winter forage drops below animal maintenance levels), a range of strategies are needed to allow animals to cope with the poor quality of the forage, including protein supplements and additional feed. And of course, there's the ever-present threat of unplanned fires devastating the ungrazed veld. In the sweetveld areas of the country, where animals can be grazed year-round (but at much lower stocking rates), winter veld is more useful.

One of the most important principles of veld management is that veld requires adequate rest in order to recover from grazing. By adequate, most researchers advocate a year or more. In fact, depending on the environment and the amount of grazing applied, veld may need 400 days or more to adequately restore nutrient reserves to the roots of the plants.  With good rest, the productivity of the veld in the following seasons will increase, relative to veld that hasn't been rested.

The simplest way to apply this system is to set aside a portion of the farm to be rested for the entire year. Dividing the farm into multiple camps (by fences or by herding) allows the farmer to systematically graze each camp only as much as the veld requires, before moving on to the next area. The trick, however, comes in moving back to the first camp when the grass is tall enough to be grazed, rather than blindly following a schedule of moving the animals around all of the grazing camps.

In good years, where the forage production is greater than the animal requirements, the last camps in the grazing schedule will be almost untouched. In bad years, all the camps in the grazing schedule will be grazed and the farmer might start eyeing the reserve camps long before winter.

In either case, by mid-season, the farmer should have a good idea of his fodder budget for the remainder of the year, and can adjust accordingly. If there is more grass than the animals require, the farmer has a range of options. He can buy in more stock to take advantage of the surplus, or lease grazing to neighbours, or simply leave the additional rested veld to provide extra productivity in the following season. If the animals are outstripping the supply of grass, the farmer can destock or buy in fodder early, before the rush for food sends prices sky-rocketing.

A good summary of the system can be found on page 119 of the 2015 National Wool Growers' Association Guidelines for Livestock Farming by Aumie Aucamp and Arno Moore (disclosure: I also contributed a small article to the publication). But the system is something that has been advocated for many years, including by farmers in sourveld such as Clive Buntting in Dundee, KwaZulu-Natal.

Drought is a reality in southern Africa, and is likely to get worse.  Rainfall is beyond a famer's control, but the farmer can control stocking densities and fodder banks and plan for the future to ensure that when disaster strikes, he is prepared for it.

EDIT: Corrected the link to the Clive Buntting paper

“Grazing capacity” is not the most click-bait phrase that one would find on the internet. It probably comes somewhere between “Donald Trump’s hair” and “exciting new developments in toaster manufacturing” in public interest.

But for millions of square kilometres of the earth’s surface, grazing capacity is critically important. Grazing capacity, very simply, is the number of animals that can graze on a certain area of land for a certain period of time, generally without dying of starvation in the process, and without denuding the soil of every blade of grass.

The concept of grazing capacity, therefore, implies a balancing act between the needs of the veld and the needs of the animal. The veld does usually need a certain amount of grazing, but too much and the perennial, palatable grasses start to disappear and unpalatable grasses and bare soil take their place. Weeds start to encroach, erosion gullies appear and rainwater no longer infiltrates into the topsoil to feed the roots of the remaining plants. Overgrazing, in other words, causing grazing capacity to decline in the long term, and nobody benefits. Yet we still struggle to actually measure grazing capacity.

But at the same time, farmers have to make a living. They need to produce enough calves, or lambs, or wool or milk to turn a profit (whether they are commercial farmers farming for profit, or traditional herdsmen farming for multiple cultural reasons, they still want new calves). Which means that the farmer needs to have enough animals on the land to keep his operation successful. So he needs to find that optimal number of animals that keeps animals fat, calves appearing every season, and the veld in good working condition. The same principle applies to “natural” ecosystems – fenced off game reserves where animals are restricted from migrating to greener pastures. Overpopulation can cause a mass-die off of herbivores, and at the same time damage the veld so much that it might take years to recover.

And that balance is what we refer to as the grazing capacity of a particular area of veld (we also talk about browsing capacity, for those animals like goats, kudu and giraffe that prefer to browse on shrubs and trees rather than graze on grass, but for this discussion I’m confining myself to grazing. Pretty much everything I can say here about grazing capacity applies in spades to browsing capacity, which is far more complicated to estimate). If we have a good handle on grazing capacity, we can manage the veld better by knowing when the veld is overstocked, or when to add more animals.

The concept of grazing capacity is mostly used for planning and regulation, which is what makes good grazing capacity estimates even more important – can the department of agriculture take action against a farmer who is overstocked compared to departmental regulations, if we can’t demonstrate that the department’s estimates are reasonable?

So how do we calculate grazing capacity?
Many different ways of estimating grazing capacity have been developed over the years, and they differ widely in their assumptions and processes.

The first, and oldest, category of methods works on the principle that the long-term productivity of the veld is directly related to the species composition of the veld – that is the proportions of different species that occur in the community. Generally, the species composition of a particular site is determined by a field survey, and compared to the (often hypothetical) species composition of an “ideal”, or benchmark, site for that particular veld type. The principle is well-founded with a lot of empirical evidence to support it.

Then there are methods that look directly at the productivity of the grass – that is, how much grass can be predicted to grow on a particular area, based on rainfall and soils. Remote sensing (satellite) technology is an important development here, where the productivity can be estimated using sophisticated analytical tools.

And finally, there are methods that combine both approaches – that is, they estimate both the species composition of the veld, and the amount of grass produced, and use various models to convert those data to grazing capacity.
All of these models will take some measure of rainfall into account – usually the long-term average for the area, or the long-term growing-season average.

All of these methods have been peer-reviewed, tried and tested in the field, sometimes for decades. However, they also have the veneer of accepted wisdom. They are rarely questioned and have the authority that comes from the respected scientists that developed them and the artificial precision of their results (I’ve seen students express the results from these models with several decimal points of precision). The underlying assumptions are rarely questioned, and often are based on ecosystem models that are increasingly out of step with our current understanding of how veld ecosystems work. Most of these methods were early forays, and even their original authors described them as that – early attempts at improving our models. First attempts require second and third and fourth iterations, otherwise they simply become outdated.

For the field worker, a major challenge comes when trying to decide which method to use on a particular area. In principle, it should be straightforward – pick the method that was developed for the area in which you’re working, do the field work, run the calculations and viola! – you have a grazing capacity. But there are two problems. Firstly, some areas have not had simple grazing capacity models developed for those conditions, so people will pick one developed in another area (for example, using a model developed for the Eastern Cape in the central highveld). And secondly, for any one area there may be two or three competing models available, each with its own set of assumptions, giving different results. Mostly, the person doing the work picks the grazing capacity model that they learnt at varsity, or the one that happens to be more accessible (some models are tucked away on obscure software packages in government institutions), or the first one that they found when looking up “grazing capacity models”.

I first became aware of these competing models several years ago, working for a provincial department of agriculture. At the time, the national department of agriculture was updating its national grazing capacity map (a critical part of the agricultural regulations) using remote sensing technology and a set of analytic tools to develop a map of grazing capacities for the whole country. In KwaZulu-Natal, our department had already developed a grazing capacity map over many years of research and experience, and we weren’t satisfied with the national map. There appeared to be quite a few contradictions. So my colleagues and I extracted hundreds of field surveys from a central database, surveys which had been conducted over many years all over the province, and compared the grazing capacity estimates from those field surveys to the grazing capacity estimates from the satellite survey, for each site.

There was no correlation between the two sets of results. None. Nada.

That results were pretty surprising. Even if the two datasets didn’t agree exactly, one would expect sites with low grazing capacity on one scale to have low grazing capacity on the other scale, and the same for high grazing capacity. But that wasn’t the case.

It wasn’t that one model was better than another. It was that they were two datasets using different models, different assumptions and mapped at different scales, but both purporting to be able to predict, for a particular patch of land a few hectares in size, how many animals could be sustainably grazed.

Since then, as a consultant and advisor, I’ve had the same problem. Which model to use? Which benchmark to use as a comparison? Which assumptions to plug into the model? Should animals leave seventy percent of the grass, or sixty percent, or fifty percent, in order to allow the grass to recover after grazing? How many kilogrammes of grass are you likely to get for every millimetre of rain? How long is the growing season?

And these competing models can result in very different recommendations – say, 500 breeding cows versus 300 breeding cows on one particular farm. That can be the difference between bankruptcy and profitability. It can also be the difference between a sustainable farming enterprise, where the natural resources remain productive for decades to come, and an unsustainable farming enterprise where the natural resources are depleted within a few years.

Is it time to ditch the concept of grazing capacity?

Some commentators have suggested that trying to estimate fixed, long-term grazing capacities which apply for all situations is meaningless.

First of all, drier parts of the world – the semi-deserts and low-rainfall savannas – are known for one thing, and that is that nothing is stable. The rainfall varies enormously from year to year, from crippling droughts to abundant water, and the veld responds to that rainfall like a four-year-old to a sugar rush. One year there’ll be almost no grass, and the next there’ll be more than you know what to do with (I’m simplifying, but that’s the basic principle). So the grazing capacity will vary enormously from season to season. (In the wetter parts of the world - in southern Africa, the eastern escarpments and coastal areas - the rainfall doesn’t vary as much, and the grass production is relatively constant from year to year).

Secondly, you can’t talk about grazing capacity without taking into account the objectives of the land users. So commercial farmers want fatter animals producing more milk and meat and calves, whereas traditional herdsmen are not as concerned about productivity of the animals and are more concerned with keeping reasonably stable herd numbers over time. Game farmers might want large numbers of animals for eco-tourism, or small numbers of trophy animals for hunting. All of these objectives have different optimal grazing capacities.

But that doesn’t help the planner. Drawing up policies, or a plan for a particular community or game farmer or cattle rancher, without having some idea of how many animals can be carried without causing long-term damage to the veld and long-term decline in animal production, makes it very difficult to make any plans.

“What sort of turnover can be expected from this operation?”
“Well, it depends on how many animals you can keep”.
“OK. How many animals can I keep?”
“Weelllll. That depends”.

Not very helpful to anyone, least of all the advisor’s reputation.

Roughly right

Barry Smith is a farming advisor and extension officer with decades of experience in southern Africa. He published a fantastic set of guidelines, rules of thumb and advice for farmers and advisors, which I use as one of my standard references. His most important rule was that it is “better to be roughly right than precisely wrong”.

That’s a sentiment with which I heartily concur. You have to start somewhere. As long as the land-user follows basic common-sense rules of adaptive management (keeping an eye on operations, and his veld, and adapting his management if his goals are not being met) then the initial grazing capacity estimation becomes merely a starting-point in a process of improvement.

But at the same time, the proliferation of models and rules for estimating grazing capacity is not helpful. It’s time we had a good hard look at what models and approaches actually work, and which ones we need to abandon.

It’s about the grass

First of all, we need to ditch the “grazing capacity” (number of animals) aspect of these models and start focussing on the forage production side – the amount of grass. Once we have an idea of forage production, we can decide what the appropriate grazing capacity will be for a specific objective. We can do this because we have a pretty detailed understanding of animal nutrition and physiology.  We also have a pretty good understanding of plant physiology. At the moment, the productivity of the grass and grazing capacity of the veld are entangled in some of the models – that is, the final result is expressed in grazing capacity, without any way to separate out the productivity part of the equation.

I’m fairly optimistic about the use of technology to assist this process. We’ve made huge strides in the last two decades in remote sensing technology, as well as in the tools and algorithms used to analyse the raw data and convert it to something useful. But most of these tools remain inaccessible to the public – they require sophisticated software, vast amounts of raw data (some of which is not cheap) and years of training to develop the skills needed process the data. So if we can package the results of these tools into useful apps and simple decision-support tools for farmers and advisors, we could start standardising the many disparate models that currently exist.

Of course, the problem then becomes, how accurate are these models? As the story I gave earlier illustrates, the department of agriculture tried to do exactly that, and developed a grazing capacity map for the country which disagreed with the results from a different model.

Nonetheless, the user-friendly app approach has the advantage that models can be constantly tweaked and updated behind the scenes, as new research improves our understanding of system processes. The department of agriculture map was a starting point, but it can be continually improved. Using high-resolution satellite imagery, we can start providing better estimates of grass production at much finer scales – scales that are useful to the farmer as much as to the policymaker. And once we have an estimate of grass production, we can set our grazing capacity according to our objectives, rather than working from a fixed, one-size-fits-all approach to grazing capacity.

Grazing capacity has been a staple of grassland science research and thinking for decades. As I’ve said before, though, livestock and game farmers are ultimately grass farmers. Think of grass as the crop. Every other crop in agriculture is expressed in productivity units – tonnes per hectare. Saying that your grass production has declined from 2000 kg per hectare to 1600 kg per hectare is far more understandable than saying the veld grazing capacity has declined from 3.7 hectares per animal to 4.6 hectares per animal (yup, the larger number of hectares needed to feed each animal means a decline in grazing capacity).

In the example I just used, the calculated grazing capacity depended on at least three hidden underlying assumptions, and tweaking any of them would cause significant changes to the result. I could change the amount of grass required by each animal every day (set at 10 kg per day). I could change the proportion of available grass grazed by each animal (set at 50%). And I could change the number of days that I want to graze the animals on the veld from the whole year (365 days) to a few months.

In fact, let’s try that out. I’ll change the amount of forage required by each animal daily to 12 kg (used in some literature). I’ll change the proportion of forage consumed to 40% (used in some literature). And I’ll keep the grazing period the same. The result? On 2000 kg per hectare grass production, and a 2000 hectare farm, the first set of assumptions says the farmer can carry 548 animals. The second set of assumptions says the farmer can carry 365 animals. That’s a difference of 183 animals, or one third fewer animals. On the same amount of grass.

If farmers are given direct estimates of grass productivity, then any further assumptions about how to use that grass become explicit and specific to the situation of that farm. But if the model simply spits out a grazing capacity in numbers of animals, it should raise a lot of questions.


I think farmers can easily and intuitively understand the tonnes per hectare of veld and incorporate that into their planning, and it will give them the flexibility that is required to adjust animal numbers rapidly as conditions change.

Agricultural colleges have been teaching trainee farmers a precise, powerful, repeatable and completely inappropriate method of veld monitoring for decades.

One of the first things any serious livestock farmer will tell you is that they are not farming livestock, they’re farming grass.

The grass is the primary crop, converted into cash via the medium of a cow or a sheep or a springbok. But farming wild grass – veld in southern Africa - is not like farming other crops. The farmer has very little control over productivity. He cannot plough the fields and plant precisely the variety of crop that he wants, while carefully weeding out the undesirable ones. He has to deal with the normal vagaries of nature. And finally, he has the grazing itself, which affects the composition of different grasses and weeds in ways which, after more than a century of research, we barely have a handle on.

Simple, prescriptive formulae for how to manage nature don’t work in the long run. Farmers need to read the subtle signs of the veld and adapt, constantly tweaking their management, week to week, month to month and year after year. They need to decide when to put cows and calves into the veld based not on the calendar but on the grass itself. They need to move their animals following the signals of the veld, not the traditions of their fathers. And they need to decide when and where to burn following the same cues.

So how does a farmer, given the constraints within which he must operate, make decisions?

Well, obviously the farmer needs to be constantly aware of the changing circumstances of his grass crop, as well as the condition of the animals. And the only way to be aware is to observe – monitor – the veld.

And this is where four decades of agricultural colleges and training courses have got it completely wrong.

You see, in the first half of the twentieth century, we hadn’t yet developed “proper” veld monitoring techniques. So we took the methods of the farmers themselves and formalised them. They looked the grass, the bare soil, the weeds, and a bunch of other cues to understand the state of their veld at any given time, and we, the scientists, simply wrote those cues down on scoresheets with scales from 1-10 and gave definitions for each score. The methods were simple, quick and intuitive.

Then, in the 1970s, new methods were developed. Scientists were, understandably, dissatisfied with the unscientific and imprecise measurement techniques of yore. For research and scientific monitoring, they needed something more precise and repeatable. So they built on the scientific methods developed previously, simplified them, and came up with Veld Condition Assessment (note the capital letters). Veld Condition Assessment is a generic name for a family of formal, quantitative, field-based (boots on the ground) methods of surveying the vegetation of a sample patch of land.

I won’t go into details of the techniques here (I'll leave that for another blog), except to say that they do all that they’re supposed to do. They’re quick, precise, and very powerful for measuring change in the composition of grasses in the veld and for calculating grazing capacity (the number of animals that can be carried on a given piece of land). And they’re completely wrong for farmers.

Let me explain.

Firstly, a Veld Condition Assessment, depending on the type of vegetation and the method, generally takes from twenty minutes to an hour to do, for a small patch of land on the veld. By small patch, I mean a line about a hundred metres long (known as a transect), or perhaps a little square of around twenty by twenty metres (a plot). So, to get a really good picture of the veld, you should ideally place dozens of sites on your farm. Which no-one ever does. More to the point, in the time that it takes you to survey a tiny sliver of veld, you could have walked an entire paddock and gained a much broader understanding of the state of the veld. In fact, one glance at the sliver that you’re about to survey often gives you more useful information than you would gain from conducting the formal survey. Dominant plants – check. Bare soil and amount of erosion – check. Weeds – check. Amount of grass – check.

Secondly, you need some experience and knowledge to do both the formal and informal survey, so there’s not much difference there. In fact, I would say that the Veld Condition Assessment requires training, while the informal assessment requires knowledge and experience. And knowledge and experience often provide richer information than training.

Thirdly, the Veld Condition Assessment is intended to be performed occasionally – usually once a year or once every two years. It’s the audited annual report of the veld – precise and quantitative. Many pretty graphs can be made.

But farmers’ main priority is making good decisions from week to week. Leaving animals in a paddock a few weeks too long in a bad year can harm both the performance of the animals and the long-term condition of the veld. Conversely, putting animals into a paddock too late can result in animals grazing a large amount of low-quality grass. Which will have little effect on the veld but will reduce animal performance.

And for those reasons, no farmer that I have ever met has consistently applied classic Veld Condition Assessments on their property. Some have done it once or twice and then abandoned it. Others have had consultants and advisers do it for them, sometimes for a few years. But when budgets are cut or the adviser moves on, the Veld Condition Assessment ceases. It is simply too much trouble for too little return.

But good farmers do monitor their veld, all the time. They monitor their veld the way traditional herdsmen have for millenia – the greenness of the grass in one paddock versus another, the amount of poisonous plants in a wetland in spring, the amount of grass from month to month and year to year. They just don’t record it. Until now.

Frits van Oudsthoorn, the author of the ubiquitous Guide to the Grasses of Southern Africa, has revised an old and simple monitoring method, for the bushveld. Also in the bushveld, a set of tools  has been adapted for measuring and recording – quickly and easily – the most important variables for a livestock farmer: grass, animal condition and numbers, rainfall, erosion and bush. The Australians and Americans have made great strides in creating simple and effective monitoring tools for farmers and advisors. I call these methods Scoresheet methods, because they work like a customer satisfaction survey – asking a few questions and asking you to rate the answers on a scale.

But the best example of real veld monitoring for farmers that I’ve seen is in a pair of papers in the journal Rangelands co-authored by an American rancher. The rancher used the Eyeball method, the method that farmers have used for millennia, combined with specific goals that they wished to achieve on the ranch.

In one paper (see here for PowerPoint), the rancher, Grady Grissom, and his co-author, Tim Steffens, described the changing management styles on a 14,000 acre ranch in Colorado over two decades. After initially starting with classic livestock farming goals, which resulted in financial difficulties, they shifted gradually to ecological and economic goals which were measured with simple monitoring – the Eyeball method. And Grady Grissom and his ranching partners recorded their observations in notebooks.

In the second paper, they described in more detail exactly what sort of cues they looked for on a week to week basis when making decisions about moving animals, and the ecological basis of those cues.

The full story of their shift from one management style to another is fascinating in its own right. But to me the interesting part of Grady Grissom’s story can be summed up in this passage:

In other words, eyeballing the veld gave them more valuable information more quickly than doing a Veld Condition Assessment, and as reliably for their purposes. The purposes is the key. They needed to measure things that were relevant to their goals, their management, their knowledge.

So what did Grady Grissom and his partners measure?

Of course, they kept detailed records of their livestock performance, numbers, condition, calving and supplementary feeding, and detailed financial records. But more importantly for this discussion is their monitoring of the veld.

They estimated the amount of grass cover left in a pasture after grazing on a scale of “low, medium or high”. They estimated the amount of litter cover on the ground (the dead plant material that eventually decomposes to form humus) on the same scale. And they estimated the amount of two common, and important, grasses on a rough percentage scale. They also mapped, using photos and GPS, the locations of important shrubs and weed patches every now and again, to see whether they were spreading or shrinking. They eyeballed important grasses on a regular basis to decide whether they had had enough grazing and needed a rest, or whether they had had enough rest and were ready to be grazed again.

The financial payoff for this (minimal) effort is described in their paper. The ranch became more profitable despite difficult economic circumstances, because they shifted their focus from purely livestock farming to grass farming via livestock, and adopted a few simple goals and means of measuring their achievement.

The experience of Grady Grissom and many other successful farmers underlines that we need to stop teaching agriculture students Veld Condition Assessment (capital letters) with its formal rules and analysis, and teach them how to assess the condition of the veld using knowledge and experience, based on a specific set of goals. Those students who go on to institutions where formal Veld Condition Assessments are part of the job will be taught the techniques on the job. For the remainder, teaching them how to set goals for the veld, a simple set of cues to help decision-making, and some in-depth knowledge of the way the veld works, will encourage far more livestock farmers to become grass farmers.