## It’s make or break time for Australia’s national parks

This post was published originally at The Conversation. It is authored by Euan Ritchie, Bill Laurance, Corey Bradshaw (re-blogged on his site too), David Watson, Emma Johnston, Hugh Possingham, Ian Lunt and me, and arose from a conversation that Euan initiated on Twitter.

National parks on land and in the ocean are dying a death of a thousand cuts, in the form of bullets, hooks, hotels, logging concessions and grazing licences. It’s been an extraordinary last few months, with various governments in eastern states proposing new uses for these critically important areas.

Australia’s first “National Park”, established in 1879, was akin to a glorified country club. Now called the “Royal National Park” on the outskirts of Sydney, it was created as a recreational escape for Sydney-siders, with ornamental plantations, a zoo, race courses, artillery ranges, livestock paddocks, deer farms, logging leases and mines.

Australians since realised that national parks should focus on protecting the species and natural landscapes they contain. However, we are now in danger of regressing to the misguided ideals of the 19th Century.

Parks under attack

In Victoria, new rules will allow developers to build hotels and other ventures in national parks. In New South Wales, legislation has been introduced to allow recreational shooting in national parks, and there is pressure to log these areas too.

Is harvesting trees for timber really appropriate management of national parks?

Late last year, NSW announced a new trial to re-instate grazing in the new Millewa National Park and other reserves, following Victoria’s unsuccessful attempt to allow grazing in the Alpine National Park. And just this week, the Queensland government passed new laws that allow graziers to feed their stock in national parks during droughts.

It’s not just the land that’s under assault. NSW recently lifted bans on shore-based recreational fishing in most state marine sanctuaries. Coastal marine parks in Australia are mostly young, small (particularly the sanctuary zones), and poorly resourced. But they are vital for regulating human activities and making coastal ecosystems resilient to pollution, invasive species, resource extraction and climate change.

The picture is grim and set to get worse. In Queensland and Victoria, land clearing laws (outside of national parks) are being “relaxed”, with at least two major impacts. First, it will place an even higher value on our reserves, as more land is cleared and further degraded. Second, it will decrease connectivity between remaining patches of native vegetation, further threatening species that require large, connected habitats.

Set against a background of rapidly changing climates (and associated changes in storm and fire frequency, droughts and floods), many imperilled species will face range contractions at best, and full extinction at worst.

Orange-bellied parrots – despite their breeding habitat being protected, they are still in decline. But they don’t need any more impacts, inside or outside national parks.

Why should all this matter?

It’s widely acknowledged that our current reserve system and efforts to conserve our native biodiversity are eminently praiseworthy, but hopelessly inadequate. Indeed, prolonged government failure to implement existing environmental laws and draw up plans for threatened species has recently resulted in court action.

The problem isn’t limited to Australia.

Biodiversity in many protected areas around the world is declining due to encroaching threats from surrounding areas. There is no free lunch; as parks suffer, their biodiversity suffers too.

Management interventions such as feral animal control, fire management and at times, grazing management, can be useful tools to achieve conservation goals in some circumstances. However, these need to be based on the best available ecological knowledge and practice and be aimed at conserving biodiversity. This is not the motivation for any of the recent changes.

Exploitation of our parks, without scientific evidence for positive biodiversity outcomes, will hasten losses. These areas need to be in the best shape possible to cope with the intensifying pressures imposed by a disrupted climate and likely increases in the frequency of species invasions.

Australia’s rich biodiversity is one of the few things our country has that is truly, globally unique. It is worth billions of dollars to our economy, and provides crucial natural services that are not easily replaced. Beyond their value to plants and animals, our national parks, wild places, and nature in general, are “good for us“.

Desperate times call for smarter measures

To illustrate how we, and the governments representing us, are failing to make use of the best available science to aid park management, and the consequences this has, we draw attention to two issues: grazing and pest animal control.

Nobody questions the stress graziers face when their stock begins to starve as drought intensifies in parts of southern and eastern Australia. But simply opening the fences to national parks is a dangerous precedent that provides, at best, a Band-Aid solution to a recurring problem.

Moving stock from pastures to parks increases the risk of spreading weeds and further [degrading natural habitats for birds and mammals, as well as sensitive water resources on which we and our livestock depend.

The current trend in drought-relief programs is helping farmers prepare for droughts. We no longer rely on emergency measures to cope with droughts that are expected and recurrent. Opening parks to grazing does not fit this model.

Australia needs to get smarter. We should do more to encourage flexibility in our agricultural and aquaculture systems. Why don’t we produce animals better suited to the unique Australian conditions?

Making better use of Australian species could also help us deal with the pest animals, overabundant herbivores (goats, camels, buffalo, deer and kangaroos) and introduced predators (cats and foxes) overrunning our parks. We have been using bullets and poison for a long time, with little evidence for an overall gain. In some cases, this approach has generated new problems.

In many regions, the best available weapon to control pest animals is the dingo. Abundant research now demonstrates that dingoes strongly limit goat, kangaroo and fox populations. Dingoes are an unrelenting and ultimately free service.

Dingoes therefore provide the perfect example of how we can start making better use of our native species to protect biodiversity more broadly, build more resilient landscapes and shift our approach from the reactive, ineffective, costly and interventionist approaches we often see at present, to more proactive, longer-term, integrated and effective conservation and management solutions.

Our parks are the last vestiges of Australian nature – a final refuge for our irreplaceable biodiversity and ecosystems. A return to the outdated views of the 19th century – when parks were little more than playgrounds for city dwellers to escape the urban _malaise_ – would run counter to everything that Australians have learnt about environmental conservation in the last 150 years.

## Talk to the Mathematical Association of Victoria

I’m talking to The Mathematical Association of Victoria this evening at the Royal Society of Victoria. The topic is “Mathematics for Conservation Decisions”. I have put my slides here for anyone interested in obtaining a copy.

Edit: some of the metapopulation modelling research that I mentioned in this talk has just been published. You can get a copy via the links here:

http://mickresearch.wordpress.com/micks-publications/#GrowlerMetapop2013

## Considering uncertainty in environmental management decisions

This is a post about a new paper, which forms part of the PhD thesis of Yacov Salomon. Yacov is jointly enrolled in the School of Botany and the Department of Mathematics and Statistics at The University of Melbourne.

Salomon, Y., McCarthy, M.A., Taylor, P., and Wintle, B.A. (2013). Incorporating uncertainty of management costs in sensitivity analyses of matrix population models. Conservation Biology 27: 134–144.

If managing koalas, should we target fecundity or survival rates? And if we are unsure of the effectiveness of management, should we do a bit of both? If so, how much should we spend on each?

The idea behind this paper arises from two strands. Firstly, we have the paper that Peter Baxter led on incorporating costs into sensitivity analysis. The essence of that paper was to take standard sensitivity analysis of matrix population models and incorporate costs. Previously, recommendations from sensitivity analysis typically suggested targeting the transitions to which the growth rate was most sensitive. Peter’s paper, however, simply noted that those recommendations should account for possible differences in costs. So rather than thinking about changes in population growth as a function of changes in parameters, the analysis should focus on changes in population growth as a function of how much money is spent attempting to change parameters (e.g., a manager might compare the efficiency of managing fecundity or survival). A key gap of this paper was that uncertainty in the efficiency of management was ignored.

The second strand arose from thinking about environmental management decisions as a problem of efficient allocation of limited resources, possibly where the outcome of management is uncertain. In this case, the optimal solution can be obtained using theory derived for financial investment. Elegant solutions can be obtained under particular assumptions (e.g., assuming that uncertainty is well described by normal distributions). An important insight of this paper was that attitudes to risk and the aspiration of managers drove the optimal decision. A key gap was that uncertainty might not follow a easily defined distribution, such as a normal.

This is where Yacov’s work sprang from. Does the choice of distribution matter, especially in cases when the normal distribution assumption is likely to be violated? The answer is, unfortunately, yes. The new paper shows that the optimal allocation depends on the model of uncertainty that is assumed. The paper uses two cases studies: control of over-abundant koalas and protection of olive ridley sea turtles.

I’d argue that the effectiveness of management actions is rarely measured. Uncertainty about this measured benefit of management, perhaps measured by a standard error, is rarer still. Yet even rarer is any assessment of the distribution of the uncertainty. Yet this new paper shows that assumptions about the distributional assumptions matter when aiming to find the best way to spend limited resources. However, defining appropriate models for uncertainty is likely to be difficult. I’m not sure if there is a simple answer to that problem.

## Why did the squirrel glider cross the road?

Why did the squirrel glider cross the road? The answer: because now it can, thanks to glider poles and rope bridges that have been installed across the Hume Freeway.

A squirrel glider crossing the Hume Freeway on a rope bridge. Photo by Kylie Soanes.

The first paper from the PhD thesis of Kylie Soanes has been accepted for publication in Biological Conservation. This is the research for which Kylie recently won a student award for best spoken presentation at the Ecological Society of Australia meeting.

Kylie’s research shows that squirrel gliders are able to use glider poles and rope bridges to cross a freeway that they previously didn’t cross. She has written a piece for The Conversation about her research. You can read a submitted version of the paper in Biological Conservation here.

## Effects of timber harvesting on water yield from mountain ash forests

A stream in Melbourne’s water catchment, carrying water that has flowed from mountain ash forest.

The effect of timber harvesting on water yield from mountain ash forest has been studied for decades. It is topical because mountain ash forests supply a large amount of water to Melbourne, a city of more than 4 million people.

Mountain ash forests are also the main source of wood for the timber and pulp industry in the state of Victoria. There seems to be an enduring debate about the extent to which timber harvesting influences water yields.

Let’s start with an aspect of the topic that is reasonably well resolved. Because of different rates of evapotranspiration, the amount of water that arrives in streams depends on the age of the trees. When the trees are old, their rate of growth is low, the trees are widely spaced, and water use by the forest is low; streamflow is at it highest. Immediately after a fire that kills the trees or after a timber harvesting event, water use is also low; again streamflow if high.

However, as the trees grow in dense stands, their water use increases and streamflow declines. However, some trees die as the forest ages, and the rate of water use of the survivors does not increase to compensate, and eventually water use of the forest declines again; streamflow begins to increase.

Water yield (as a proportion of the maximum achievable) versus the time since fire for mountain ash forest. This assumed curve is based on the “Kuczera curve”.

The change in streamflow with forest age can be represented by a curve – it is commonly referred to as the “Kuczera curve”, named after one of the researchers who first documented the relationship between forest age and water yield. Such a curve can be represented by the mathematical equation:

$y(x) = 1-e^{-b x}(1-e^{-c x})$

Here the water yield , y(x), is expressed as a proportion of the maximum possible yield (the yield achieved in an old growth forest or immediately after timber harvesting or a fire). The age of the forest stand is x, and the parameters b and c control the shape of the curve. I’ll use b = 0.022 and c = 0.07 for the remainder of this post*. This results in the yield curve shown above.

In this case, the water yield reaches its smallest level when the forest is about 20 years of age, at which point the yield is almost 50% less than the maximum achievable.

So, this shows that if we were able to keep the forest as old growth (or at zero years of age), then we could maximize the water yield. However, as noted in my previous post, it is overly optimistic to assume that mountain ash forest won’t burn.

The chance that a forest will survive for x years given fires occur randomly in time with a mean interval of m years is:

$S(x)=e^{-x/m},$

or equivalently the cumulative distribution function of the time of fire is:

$F(x) = 1-S(x) = 1 - e^{-x/m}.$

From this we can get the expected age structure from the derivative of F(x), which is defined by the probability density function:

$f(x) = \dfrac{e^{-x/m}}{m}.$

The above is all basic survival theory assuming a constant rate of fire with forest age (McCarthy et al. 2001, email me for a copy of the paper).

Now, we can get the expected water yield E(Y) by integrating the product of the expected age structure f(x) and the water yield curve y(x). The range of the integration is zero to infinity, which are all possible ages for the forest (OK, an infinite age is impossible – I am about to impose an upper limit on forest age via timber harvesting).

$E(Y) = \int_0^\infty \! f(x)y(x) \, \mathrm{d} x.$

This might look a bit complicated, but it is simply the average water yield, but it is a weighted average based on the amount of forest that is expected to be of different ages.

For the water yield curve I have used (y(x)), solving this integral leads to:

$E(Y) = 1-\dfrac{1}{1+bm}+\dfrac{1}{1+(b+c)m}.$

Plotting this expected yield versus the mean fire interval, we have:

Expected water yield from mountain ash forest (as a proportion of water yield from an old growth forest) versus the mean fire interval.

So, we see that as mean fire intervals increase above about 20 years, the expected water yield increases. If we assume the average fire interval in mountain ash forests is 100 years (assuming tree-killing fires), then the expected water yield is about 20% less than that obtained for an old growth forest.

Now, the main question was about how timber harvesting might influence the water yield. The influence will occur via the effect on the forest age structure. If we assume that forests are harvested when they reach the rotation age R, then the age structure of the forest becomes truncated at R. The probability density of f(x) above the value of R needs to “redistributed” to values below R. Hence, the truncated probability density function is given by:

$f_R(x) = \dfrac{e^{-x/m}}{m(1-e^{-R/m})},$

Using the same logic as above, the expected water yield given a mean fire interval of m and a rotation age of R is:

$E(Y_R) = \int_0^R \! f_R(x)y(x) \, \mathrm{d} x.$

One can solve this integral. I’m not going to bore you with the solution – it is not particularly simple. And rather than focusing on this, I calculate the ratio:

$p=\dfrac{E(Y_R)}{E(Y)}$

This is the expected water yield from a forest under a rotation age of R years relative to the expected water yield of an unharvested forest. In both cases I am accounting for unplanned fires that occur with a mean interval of m years.

When p is less than 1, timber harvesting reduces the expected water yield from the forest. When p is greater than 1, timber harvesting increases water yields. Plotting the relative yield p versus the rotation age R gives:

Expected water yield from a harvested forest as a proportion of the expected water yield from an unharvested forest versus the rotation age. This assumes that the average fire interval is 100 years. The dashed line shows where the water yield from a harvested forest matches that of an unharvested forest.

This shows that as the rotation age increases above approximately 40 years, the water yield increases. Increasing the rotation age from 50 to 100 years increases expected water yield by approximately 10% in those areas exposed to timber harvesting. A further ~10% increase would be obtained by increasing the rotation age from 100 years to 200 years. At a rotation age of 200 years, the water yield is only about 5% below what would be expected in the absence of timber harvesting (but in the presence of fires).

Of course, these increases in water yield need to be weighed against the other possible benefits (e.g., more large trees for wildlife, possibly more sawlogs) and costs (e.g., fewer stands reaching rotation age due to fire) of increased rotations. Also, note that this analysis is limited to those parts of the forest exposed to timber harvesting. Effects will be moderated proportionally depending on how much of the forest is harvested.

Interestingly, water yields can be increased by reducing rotation ages below 40 years. In fact, when the rotation age is 8 years or less, timber harvesting actually increases water yield above that obtained in the absence of timber harvesting. This is because it keeps the water yield curve near it maximum at x=0. Such short rotation ages might not be feasible because sawlogs would not grow within this time. Water quality problems due to frequent harvesting might also be problematic, as might regeneration due to paucity of seed.

Given that nominal rotation ages in mountain ash forests are greater than 50 years, the right hand section of the graph is probably more relevant to the question about effects of timber harvesting on water yields. This simple model shows that it is reasonable to claim that clearfall timber harvesting reduces water yield in mountain ash forest.

*If you would like to investigate different parameters for yourself, I have created an Excel spreadsheet that does the various calculations displayed here.

In case you are interested in trying different values for the parameters b and c, note that the minimum yield occurs at x = ln[(b+c)/b]/c. At this point, the reduction in streamflow below the maximum is:

$b^{b/c}c (b + c)^{-(b + c)/c}$

## Planning for unplanned fires, and the response of biodiversity

Old growth mountain ash forest (for scale, note the Toyota Landcruiser and people near the base of the tree)

There is a report in today’s Age about the decline of Leadbeater’s possum in the face of fires and timber harvesting. Professor David Lindenmayer of the Australian National University notes that the situation is dire, and that timber harvesting should be ceased in mountain ash forests. It is worth noting here that these forest are the main source of wood for native forestry in Victoria. The forest industry representative, Lisa Marty, says that fire is the problem, not timber harvesting. Who is correct?

Leadbeater’s possum relies on old forest for its survival. While it can persist in recently burnt forest, it only does so where large hollow bearing trees that existed before the fire remain present. However, if these trees are killed by the fire, most will collapse within decades. As a consequence, in a forest that is exposed to unplanned fires, only a fraction of that forest will be high quality habitat – it will be those areas that are currently old, and those that are young but were previously old.

It is possible to do a back-of-the-envelope calculation to determine that proportion. Let’s assume for simplicity that high quality habitat only occurs in forest where the previous fire occurred more than 200 years ago. The average fire interval in these forest is approximately 100 years. If fires occur at any site randomly in time, then the probability that a site will escape a fire and reach 200 years of age will be exp(-200/100), which is approximately 14%.

This means if an area of forest is conserved from timber harvesting, in the long term we would expect that only ~14% of it would be old enough to support high quality habitat. The actual proportion would fluctuate around that number over time, but it is a realistic (albeit rough) assessment of the amount of forest we would expect to occur in an old state at some time in the future.

So, if we set aside 10,000 ha for conservation, we would expect less than 1,500 ha to be suitable. The more that is set aside, the greater the amount we expect to be suitable in the future. Further, a larger conservation reserve also provides a larger buffer against fluctuations caused by irregular large fires.

Therefore, in a world exposed to unplanned fires, any harvesting reduces the expected area of forest that will become old because harvesting reduces the age of trees. In this light, timber harvesting certainly reduces the production of older forest, thereby contributing to declines of old-growth dependent species.

This is basically an issue about how we should plan to conserve biodiversity in the presence of unplanned fires or other random (but expected) events. It is a topic I studied during my PhD, leading to the following paper:

McCarthy, M.A., and Burgman, M.A. (1995) Coping with uncertainty in forest wildlife planning. Forest Ecology and Management 74: 23–36. [Email for PDF]

In fire-prone environments, forest managers should assume that fires will occur, but at a time they cannot predict. Thinking of a fire regime as a stochastic process provides a framework for predicting impacts.

## Optimal monitoring when detectability varies – my talk at #ESAus2012

I’m looking forward to the Ecological Society of Australia conference this week. I’m speaking in the second time slot (2:15 p.m.) of the last session on Thursday (the last day). Check out the other QAEcology talks here.

I am sandwiched between two Bayesian talks. I’ve heard a rumour that Gerry Quinn and Ralph Mac Nally will present an entertaining double act (prior to me), while Brendan Wintle will be talking about Bayesian detection (posterior to me; sorry about that. I couldn’t resist…).

How many surveys should be done to account for variation in detectability?

My talk will discuss how to optimize the number of times one should survey a site to maximize the chance of observing a species. This might be the number of nights to survey a stream to find a frog species, or the number of plots to search to detect a species in a region, or even the number of searchers to send to a site.

The key is that because of variation in detectability (over time, over space, or among people), a single visit to a site (or a single site within a region) might have, by chance, a low probability of detection. So there is an incentive to search more than once – that will increase the chance of having a higher detection rate at least once. However, going more than one incurs greater travel costs, which eats into the time available to actually search.

The tension between more shorter searches (with extra travel time) and fewer longer searches (with less total travel time) creates a trade-off. We developed and analyzed a model to optimize this trade-off. And we used a field experiment to evaluate it.

If you want a sneak preview, a pdf file of my slides (minus the best jokes) is posted here.

I’ll also be auto-tweeting this talk using the hash-tag #micktalk in case you can’t make it to Melbourne and want to follow along with the slides. For overseas folk, my talk is on Thursday 1415 Melbourne summer time, which is Thursday0315 UTC (I’m not expecting too many folks following along in London!), Wednesday1915 in Los Angeles, and Wednesday 2215 in New York.