Publication (Media): Taking the pulse of Australian rivers

Publication (Media): Taking the pulse of Australian rivers - using bugs and mathematics

Publication Type:	Media Release
Publication Name:	Taking the pulse of Australian rivers - using bugs and mathematics

Reference Information

Norris, R. and Markwort, K. (1998) Taking the pulse of Australian rivers - using bugs and mathematics - Sep 1 1998, CRCFE, Canberra - Media Release.

Optional Information

Other information:

September, 1998

CRCFE

Taking the pulse of Australian rivers - using bugs and mathematics
by Karen Markwort and Richard Norris

Australian rivers are dying. Toxic blue-green algal blooms, fish kills, salinity and bank erosion are just some of the symptoms of our 'sick' rivers. Other symptoms are being revealed through bugs and mathematics.

Just as a doctor might take your pulse to assess your health, Australian scientists and river managers throughout Australia are taking samples of aquatic bugs to check the health of our rivers. Information from these samples is then fed into a computer model to come up with a diagnosis.

Its all part of the National River Health Program, administered by Environment Australia and the Land and Water Resources Research and Development Corporation, which is aimed at monitoring, assessing and improving the health of rivers throughout Australia.

Prof Richard Norris from the Cooperative Research Centre for Freshwater Ecology explained that bugs, by their presence of absence, could reveal a lot about the condition of our waterbodies. Some bugs, such as mayflies, are particularly sensitive to pollution and will not be found in degraded rivers. Others like worms, however, can thrive in highly degraded sites because they don't require much oxygen.

So where does the mathematics come in?

Richard, a river ecologist, says that there are few things in the natural world that could be well described without at least some mathematics. River health is one example.

While the computer models involve fairly esoteric mathematics known as multivariate statistics, the result is a set of easy-to-understand probabilities—the probability, or likelihood of finding a particular bug at a particular river site, given the time of sampling and the characteristics of the site.

Assessing the health of rivers involves selecting a range of reference sites, or sites where there has been minimal disturbance or damage because of human activities. These sites provide a benchmark against which 'test' sites can be compared.

The reference sites are sampled to determine what types of bugs, properly known as macroinvertebrates, live in them. Habitat information is also gathered, such as the type and amount of surrounding vegetation, the type of material that makes up the streambed, the pH and temperature of the water and the latitude and longitude of the site.

Using multivariate statistics you can determine how closely a test site matches a reference site. A 90% match means that it if almost identical to the reference site. A 50% match means that it is not at all like the reference site.

"Once we've done this," Richard explained, "we can go to a new test site and measure its habitat attributes and based on those measurements determine the probabilities of the test site belonging to the same group as the reference sites.

"We can determine what type of bugs are likely to occur there because the same bugs should live in our test site as live in similar reference sites, if the test site is not damaged by human activities."

However, it is unlikely that any two sites will be exactly the same - which is where the probabilities come in.

Firstly, the probability of a test site belonging to each of the groups of reference sites is calculated using its habitat data. This ensures that we comparing apples with apples.

Then the probability of finding a particular animal at any one site within each reference site group is determined. Column 3 of the Table 1 indicates that there is a 90% chance of the mayfly occurring at Reference Group A.

To work out the likelihood of each animal occurring at a new 'test' site the two probabilities are multiplied. The Table 1 indicates that there will be a 75% chance that the mayfly will occur at our new test site.

Table 1 The probability of mayfly 2 occurring at test site X

Reference group	Probability that test site X belongs to group	Frequency of mayfly 2 in group	Probability of mayfly 2 occurring at test site X
A	0.4	90	36
B	0.3	70	21
C	0.2	70	14
D	0.2	40	4 Combined probability that mayfly 2 will occur at X = 75%

Column 3 calculates the frequency of a particular bug occurring. For example, if you sampled 10 sites in group A and found a particular bug at nine of those sites, you would have a 90% chance of finding that same bug at other sites within that group.

Based on the chemical and physical characteristics of a site, for example, there may have been a 100% chance of an animal occurring at a particular test site. If the animal wasn't detected at this site, you would know that there was something wrong. This would sound the alarm bell to look further.

"Biological monitoring can provide a better picture of what is happening in a stream over a long period," Richard said.

"During a recent sampling trip on Yarralumla Creek we found hardly any bugs at a particular downstream site.

"The chemical monitoring didn't detect a problem, yet our sampling team said that the site would have exploded if they had thrown a match in the water—the smell of petrol was so strong. So our biological monitoring certainly confirmed that there was a problem.

"We don't expect to get all the bugs that we predict to be at a particular site," Richard added. "So what we do is fiddle around with the probabilities again. We add up the probabilities of all those that had a better than 50% chance of occurring at our new site and that gives us the numbers that we would expect.

"When we sample the test site we look at what animals we actually got compared to the number we expected to get. If everything is okay and not damaged by human activities, we'd expect a ratio between what we measured there the observed number - to what we expected to get to be one. If it's much less than one, then we know we've got a problem."

Of course to make things that bit easier, much of these calculations are conducted using a computer model known as AUSRIVAS.

The beauty of the AUSRIVAS model, said Richard, is that you don't need a degree in statistics to use it.

"It's a bit like driving a car; you don't need to know how to build a combustion engine to use it," he said.

"The heart of AUSRIVAS is really quite simple; simple but elegant."

"Mathematics doesn't have to be complicated to work well," he added. "It just has to be clearly related to the things you're trying to resolve."

The bottom line in assessing water quality comes down to simple plus and minus calculation - how many bugs were expected to be present at a site and what is the difference between that and what were actually sampled.

This article was written for the ACT Mathematics Association and appeared in the Canberra Times during September 1996 to celebrate Mathematics Month.

Search classifications

Administration