Not A Refuge

A curious tale of fire's lack

We follow fire. More than that, we are fuelled by it. We consume it, like some demonic ant colony on the trail of flaming crumbs. For us pyrovores, fire is our meal ticket. We use it to make science, to manage, to conserve, to care for Country.

Because we follow fire, we go everywhere. Well, almost everywhere. Forests, woodlands, and grasslands obviously. But even wet jungles and rainforests can burn when conditions are right (or wrong). Come to think of it, there isn’t really anywhere on land that can’t burn, thanks to the surfeit of flammable materials that fill our ecosystems, our cities, our agricultural and industrial lands and yes, our sewage treatment plants. Plunging underground through peat bogs, marshes, mires, moors and swamps, we find seams of gas and coal. We find magma.

You might think you’d be safe in the ocean, or air, or space, but have you heard of Deepwater Horizon? Deep sea hydrothermal vents? The Hindenburg? Nuclear weapons? Space experiments? Stars?

Fire reaches more places than you think. But the coolest place to evade fire is in fire. Fire-free zones inside a fireline are called refugia: unburnt islands in an ocean of fire. You’d be forgiven for not knowing about fire refugia, because it comes so naturally to talk about hectares and acres burned, razed to the ground, destroyed, gone. We count fire perimeters and footprints. It’s not so often we count the stuff that didn’t burn.

Fire refuge (Source: DALL-E. Prompt: An impressionist oil painting of a giant ice cube in the middle of a forest on fire)

The advent of fire severity mapping was like upgrading from black and white to colour TV, as we swapped out the old binary classifications of burnt and unburnt for endless shades of grey, from complete canopy consumption to partially perturbed understorey to barely bothered litter. Fire comes in flavours, and it’s always a buffet – never just a single dish.

New patterns emerged – patterns that had been visible from the ground, to be sure, but that had yet to enter the big data consciousness of modern science. We stopped just asking what drives fire, and asked what drives high severity fire. Similar answers though: weather looming largest, with a motley crue of also rans occasionally rearing their heads: terrain, soil moisture, vegetation type, time since fire.

Put the question on its head, and we find our oases. Turn the microclimate thermostat down, wet the soil, scoop out a valley, orient yourself away from the atom smashing machine eight light minutes down the road; there are your fire refugia. We can build an explanatory model, popping just a few of these variables in, and explain a respectable amount of the variation in fire refugia incidence. This is ecology so don’t get your R squared hopes too high. Chuck the whole kitchen sink of variables into your favourite machine learning algorithm and you can predict, rather than explain. The R squared creeps up, the causal connections recede. Depends what you’re trying to do.

Most scientists take some data, build some models, write it all up and move on. But what if that didn’t satisfy you? What if you kept going? What if you wanted, or even needed to know the story for every single patch of land identified as a fire refuge in every single fire you’d mapped? You might pore over all those variables, the climate, the topography, the whole enchilada, checking how they fit in each case. Maybe most of the time they line up, and you find refugia just where you expect to. They probably don’t all point in the right direction all the time, but enough of them do for it to make sense – there’s an unburnt patch on a north-facing ridge, but the fire passed through there just when it started raining. All good.

But then maybe you find a few cases where the predictors don’t fall into line, at all. Where fire was avoided despite bad weather, despite facing north, despite dry fuel, despite it all. What’s going on there? Just bad luck? It is a large sample size you got there, so big that you’d be surprised to not have a few peculiar cases like that.

But what if you find a couple of refugia that keep showing up, through multiple fires over the years, despite every fibre of your model telling you that these are areas that should burn, not just burn but burn at high severity? Do you go all Einstein on them, shouting hidden variables?

That might be when you decide to do some fieldwork, to see these refugia with your own eyes. So off you set, a mandatory colleague in tow, to investigate. This is deeply satisfying, until it’s not. You’ve marked off almost every anomalous fire refuge from your list, but there’s a problem with the last few. They’re hard to get to. Maybe they’re on private property, maybe there’s a military facility nearby. Maybe the vehicle breaks down, there’s an infestation of ticks or leeches or both. Maybe they’re culturally significant sites and you run out of luck with permits.

Maybe you don’t give up and you start bending a few rules. You go on your own, you go after hours. Maybe you get lost, maybe you fall, maybe you get stuck. Maybe the search party can’t find you, maybe they give up. Maybe time’s running out and that’s when you smell smoke. Maybe a part of you smiles.