It’s such a simple question, everyone knows what a forest is. Or do they? I thought I knew, until I started my masters project. It is on deforestation hotspots, using a global satellite-derived dataset available from Google Earth. Unsurprisingly, I came across the issue of how to decide what would be considered a forest.
Is a forest defined by a clump of trees? Three trees huddled together in a field would not a forest make. How big does the clump have to be then? If the trees are spaced out, with metres in between each one, that would not be considered a forest, more of a savannah-type ecosystem. A community/urban park with some trees interspersed over an area wouldn’t be considered a forest either. Does it matter how tall the trees are? What about how long the trees have been there for? If the ground was recently razed and the seeds of forest plants are sprouting, would that still be a forest? Does it matter if these trees harbour animals which are reliant on the ecosystem, does that make it more of a forest? Yet despite time lags in extinctions, we are increasingly getting the empty forest syndrome.
According to Wikipedia,
The second hit that turns up on my Google search ‘what is a forest’ seems to do a pretty thorough explanation of the different definitions (if you’re interested). I’m not trying to be philosophical or pedantic, but trees, which make up forests, occur in so many different ecosystems, at different densities, and in different forms, resulting in varying definitions of what forests are. It is tricky trying to pinpoint an exact definition for a forest, but when you’re doing it traditionally, at the site itself, most people do have some kind of gut notion about what forests should be like and what the vegetation in front of them would be considered. Though perhaps that also depended on cultural contexts (see Box 1).
Box 1. Forests in the UK
I had always thought forests referred exclusively to natural forests, but soon learned otherwise. In the UK, places referred to as ‘Forests’ are often planted or maintained by humans (e.g. New Forest), usually for deer-hunting purposes (for royalty, at least historically). Otherwise, they would be plantation forests (managed by the Forestry Commission) . In contrast, woodlands is the term for natural forests, and ancient (or old-growth) woodland for forests that have been around since the 1600s (but possibly still not primary).
With advances in satellite technology and other remote-sensing methods though, forests are often reduced to just a bunch of numbers. Most definitions use some percentage of canopy/crown cover (over a given space). The UN’s Food and Agriculture Organisation FAO (2000) for example, defines a forest as perennial woody plant >5m tall, with a crown cover of >10% and an area >0.5ha. It might not matter to the average individual on the streets perhaps, but the vague definition has repercussions on biodiversity conservation, climate change agreements and ultimately on humanity. Satellite-based estimates for global forest cover range from 32 M km2 to 41 M km2, and the huge discrepancy is mainly due to the ambiguous definition of forest (Sexton et al. 2016). Technical discussions aside, almost everyone would agree that forests provide immense benefits to mankind. Exactly what benefits though, which is what politicians and decision makers want to know, would depend on the definition used. Sexton et al discovered that the difference between using a >10% tree cover and a >30% tree cover definition for forests within the tropics alone would incur a difference of 45.2 Gt C of biomass, which is valued at US$1 trillion.
The other issue with using satellite-based products to monitor forests, is that the definition used would often include areas that are not of particular conservation importance (e.g. plantations) but cannot be easily differentiated. Conservationists are also often more concerned with the deforestation of primary forests, and less so with secondary (regrowth) forests (although secondary forests are increasingly perceived to be of conservation importance too), but the distinction cannot be made (easily) from satellite products. If you’re staring at the images, perhaps you would be able to pick out the neat rows, or the distinct crown of oil palms. Often though, especially if one is covering a huge geographic area, you would just be looking at numbers and coding language.
Forests are great in themselves, trees are cool, but often the biological value of forests lie not so much in the sole presence of the trees, but the presence of animals too. Remote-sensing technology has not quite levelled up to being able to identify the presence of animals in landscapes (with a few exceptions like ecosystem engineers perhaps), particularly those that like to hide in vegetation.
Camera traps and drones are great, of course, and many forest species thought to have disappeared have been found to still dwell in that forest with the use of these technology. They don’t make themselves amenable to large-scale processing though, and there are projects that make use of the public to help identify the animals (e.g. Snapshot Serengeti). And still, they require people to be on the ground (to set up the camera traps or fly the drones).
So what is a forest? I nearly went into an existential crisis (a bit of an exaggeration) thinking about this (a few months ago, but it took me this long to formulate a blog post) with regards to my project. Cos how I choose the definition of forest could affect the results I get, which I not only want to be ‘true’, but also want to have some policy-relevance for (looking at how much time I have left though, getting to this stage is looking increasingly unlikely). It is really difficult deciding what to conserve when one cannot define the target well. Perhaps conservation is better left to locals and those who work on the ground, and know the site well, rather than someone like me sitting at my desk far away, typing a bunch of code.
Yet conservation as a field is underfunded (and understaffed), and it’s way cheaper and faster to identify areas computationally than through fieldwork. They work at different scales too, and examining global datasets can throw up areas of rapid change that conservationists were not aware of. Even so, I think ground-truthing needs to be done, and fieldwork is still an integral part of conservation (yeah okay that’s partly because I want to get outside and not just stare at my screen). This post has kind of degenerated into a rant about some of the limitations of a computational project relying on satellite-based data. It’s not all been bad though, and my next post will be about what I’ve learned, broadly, from my project thus far.