 Well, welcome everybody to Hydrotera's latest webinar series. Today we've got David Dettrick, who's going to present to us about managing our waterways and aquatic ecosystems through multivariate river health assessment. Before we get started I would like to begin by acknowledging that we conduct our work across this great land and for that privilege we would like to thank the traditional owners. Hydrotera respectfully acknowledges the Boon Wurrung, people of the Kulin Nation, where we are located today. We pay our respects to their elders past, present and emerging. So David Dettrick is our presenter today and he's a principal environmental engineer and a specialist in numerical modelling. So thanks very much, David, for presenting today. A little bit about David. So Dr David Dettrick has over 25 years experience in river health and aquatic ecosystem water quality monitoring assessment and management. He is an accredited Oz Rivers practitioner in Tasmania, Victoria and Northern Territory. And has worked for Rio Tinto, Tasmania EPA, Northern Territory EPA and Earth Systems. And has helped develop the Tropical Oz Rivers River Health protocol amongst other guideline development for the Australian, Northern Territory and Tasmanian governments. He has been involved in several complex system wide river health assessment programs, including the Dewan History Program, the Melville Bay Marine Health Monitoring Program, the Darwin Harbour Water Quality Protection Plan and Mixing Zone Management at more than 100 major industrial outfalls around the world. He has worked extensively throughout Australia, Africa and Southeast Asia. And how did David get to this position for some of you youngies on here today? It wasn't an easy journey, lots of study as well. So David's background in education was he studied initially at Monash University doing civil and environmental engineering. Then at UTES he did a master of science in which was focused on aquatic ecology and environmental technology. And finally he did his PhD in soil gas transport at the University of Melbourne. So a long hard slog of education there to compliment what's a pretty impressive range of hands on experience. So without further, oh no, I nearly forgot. A couple of other things before we get started. We love your questions. So please use the Q&A button at the top of the screen to raise those questions. And I will read those questions out at the end. And David will do his best to answer them. What does Hydrotera do these webinars? We like to share knowledge. We like to facilitate education and we like to lead industry. Speaking of such things, we've started an education program. And our first training course which we're doing in collaboration with ALGA, which is the Australian Land and Groundwater Association, is on groundwater sampling. So please feel free to reach out if you've got some graduates or experienced field techs who want to be trained formally in groundwater sampling. And we can run them through that course. So without further ado, I will pass over to David and many things David for sharing some of your knowledge today. No problems, Richard. Can you hear me? Loud and clear. Excellent. Thank you so much for being on board, everyone. Today we're going to talk about one of the most complex issues that humanity has to grapple with, which is managing our aquatic ecosystems with multi-variate river health assessment. Aquatic ecosystems are very interesting on planet Earth because they're probably the most biodiverse habitats on the planet. And they're also the smallest in area and generally the most impacted because all people need water everywhere around the world. So next slide, Richard. Okay, so this is a graphic that just sort of tries to outline the various major subjects in a classic multi-variate river health program. We've got fish sampling. We do fish community sampling. So we try to get an assemblage of species from each water course. We also have them filleted and send fish fillets off for toxicant tissue sampling analysis in various labs around the world. We do sediment and substrate sampling as well. And again, that data goes off for ICPMS in various labs around the world to collect metals, mages and trace elements, which are important for toxicity assessment. We also do water quality both field and laboratory. So we're doing metals again and trace elements and nutrients. We also do an assessment of riparian and aquatic vegetation. Riparian assessment is done by a terrestrial ecologist and the aquatic vegetation is usually done by an aquatic ecologist. Then we do some macroinvertebrate sampling, which generally follows the OzRivers style framework for people who are familiar with OzRivers sampling protocols. Then we also look at climate and stream flow, usually for the 12 months before the program, just to get a feeling for what kind of climate and stream flow effects may have impacted on the aquatic ecosystem that we're managing. Okay, this presentation is going to be in two main parts today. We've got a bit of a discussion about multivariate program design and data sampling, and then we've got the engine room of it, which is data testing and multivariate analysis. Okay, part one, program design and data sampling, arguably the most important aspect to get it right before you do it. So multivariate sampling design usually invokes Baki, which is the classic before-after-control impacted site selection process, which is essential for statistical analysis of impact. As you can see on the graphic, before is basically before anything occurs. It's a more natural system upstream of the generally point source activity, point source discharge activity, such as a mine site factory, city waste management facility, etc. Then you have the after point, which is downstream of that activity, and you also have the impact point, which is monitoring directly at the impact. So we can have a transition between a mixing zone at the impact and the after site, and then arguably most important of all is the site in purple there, which is the control site. So when we're trying to assess potential impacts from drought or other climactic impacts, or even land use impacts, which might be upstream at the before point, we can look at the control site. Selection of the control site usually is done in such a way as that it's an area not going to be impacted by human activities, which in these days is becoming increasingly hard to find, but it's still, when you've got a good control site, it's good to hang on to it. I've got the 4S's here, my 4S's, which I find important in this kind of multivariate sampling design, statistical program design, how to remove external bias from the data, and of course that involves the selection of a great control site and doing all of the same tests at the control site as we do at the other 3 sites. Site selection is very important to be done based on a whole bunch of stuff, but the macroinverterate habitat types, riffle pool and edge habitat are quite important, and there's also 4 velocities, slow, fast, slow, deep, fast, deep, etc. Very important to grab in that section. Access is really important. You've got to carry quite a bit of equipment and people. Quite often we do these studies with more than 10 people in the field, so you need a couple of four-wheel drives, and if the four-wheel drives can't get there, you've got to be prepared to walk. And some of the stuff we do in places in Africa gets quite hot, so you really need to think about hydration and keeping people cool. And then finally, which is also interrelated, is the safety issue, flash flooding, flow velocities, flow depths, dangerous animals working in and around water issues. So they're all important. Next slide. Just before you do, so how often do you actually have an adequate control sign? We try to have one in every program, Richard. The key word there is adequate. And look, sometimes the control is adequate for one year, and then we've literally, in places like Laos, we've had local communities come in and do shifting agriculture straight through our control site, and the control site's literally gone from a site that scores 7s and 8s, and then 12 months later scores 4, and in one program actually scored lower than all of the before-after-impacted sites. So we try to have a control site every single time, but in some countries it's very difficult because of land ownership, land use styles. Laos is a very communist tribal country, so any piece of land can be used almost at any time. It's controlled, for instance, by the seven-year farming cycle, Richard. So you've really got to try and work around that. The other way to do it is to also, if your client does actually own other pieces of land around the place, which they can control access to, it's quite often a good idea to try and grab a site which has been fenced or something, which has got a small creek or a piece of a river in it, and try and use that control to control the control. But yeah, it's a difficult thing to do, and I think as the human race becomes more and more popular, it's going to get more and more difficult. Thank you. Okay, climate and stream flow data, essential to really figuring out what's going on and to assess inter-annual variability in all of your data sets. Annual rainfall provides runoff for river flow, obviously, and can be affected by drought and wet years. You can see on the graph to the right, we've marked on the 10th percentile dry, 90th percentile wet years, and the median, and you can see there's quite a few years, which are below a one in 10 drought. There's quite a few years that are quite wet, and these kind of things do impact. Variations in flow depth create variations in aquatic habitat area, so it can change the position of edge habitat, and change the amount of sediment in pools and riffles. Variations in flow velocity can create different sediment deposition rates. So if you, for instance, got very low flows during drought, the low velocities can mean that you've got a lot of sediment depositing in riffles that would not normally deposit in riffles, because the velocity in the riffle is much lower, and velocity in riffles is usually one of the highest velocities in a small stream. And large rivers. Climate and flow records are therefore very useful for assessing this inter-annual variability, and it's very important to apply them particularly to your control site data. If you've got an evident signal in your control site data, then you can use that to modify your impacted and upstream downstream sites. Next slide. OK, riparian habitat assessment is quite complex. We really need a pretty experienced terrestrial and aquatic ecologist to undertake this well. We always make sure we get local experts in as field team contractors, because they always have the best knowledge of sites. You can very quickly and easily get some detailed data about, you know, change in your riparian habitat, and that's one of the most crucial things to know about in relation to what's happening to bug communities in the river. And you can see you can make quite good graphics, like we've got on the right there. We can see slight shifting change over the years from habitat loss, and also in some cases from quite large droughts. We've been sampling sites recently that have had some pretty big droughts, one in 50 year droughts that have spanned over two to three years. Next slide, Richard. OK, the visual aquatic habitat assessment is focused around the actual aquatic ecosystem of the river itself. We always use the Osreva's visual habitat assessment protocols, and that's derived from various US EPA visual assessment protocols, FabSims and a few others. There's a long history of development and research in this area. US EPA has been doing this since the 70s. Australia has been doing it since the 90s. I think Europe's finally catching up now in the northeast and a few other places, Africa and Asia, are tending to look around and use whatever seems available for their particular use. We do a lot of Osreva's in Southeast Asia and Africa. Assessor training in this is essential. So I train up our earth systems guys, and we have to show reproducibility in their scores when they're assessing the ecology visually. And I think you need to do a lot of checking of their scores each year. And you need to look at things like drift, because we've found that different assessors, one thing that comes out of the statistics actually, is that you can actually pick up different assessors from their different scores, and you may need to actually correct some scores. Basically, it's in the personality of a person. Some people are very optimistic, some people are less optimistic, and their scores kind of reflect that. Some people also even like different colors, Richard, which is extraordinary. And their score is that have a particular color regime differently to another area that has a different color regime. So there's all sorts of stuff to look out for in that. Next slide. So before we do this, if someone wanted to get trained up in that methodology, do you know where they can go? Yeah, eWater is running various Osreva's training courses for Australia. And it is, I think that's pretty much one of the only training sources for Osreva's. If you Google Osreva's training, something should pop up. I'm certainly not current in that area, but the last time I looked, which was a few months ago, eWater was coordinating a lot of the Osreva's training packages for Australia. And there's kind of a national training package, and each state will have its own Osreva's protocols. You'll need to make sure that you're also getting your local states accreditation as well, which might be another training course entirely. You said there's bias in the way they collected between individuals. It sounds like kind of tricky for you to really determine that bias, like unless you're duplicating up on the same plots you're assessing. Yeah, and this is where having a ton of photography really helps, Richard. I encourage the teams, I mean, I'm always with the teams, of course, but I encourage the teams to take a lot of photos. So one of the data sources, which I haven't talked about many times in this short webinar, is photography. Generally, we will have something between 400 and 500 photographs of, say, 10 sites to look at. Those photographs will be across rivers, upstream, downstream from each site, detailed shots of substrate, detailed shots of any kind of habitat differences, woody debris levels, riffles, pools, all the habitat types. And what I can do is jump into those photos and then check a score from a person, and then I can actually make, I can actually do like a desktop scoring myself, and then I can actually compare my desktop score to their actual field score, and then we can use that to assess, drift, and correlate where we need to. So essential to collect field photography. And in the era of mobile phones with GPS locked into the picture, there's no excuse for not taking enough photos, really. Okay, fish community sampling. Again, you need a fish ecologist, really, to get this done well. We've done it, there's two photos there. You can see there's two ways of doing it. One is electrofishing, which is a bottom image. And the above one is same fish net collection. We basically choose one or the other based on conditions. We found in some of the remote areas of the world that electrofishing is almost impossible because high temperatures are very bad for batteries. You have a range of different salinity in the rivers and streams, and that really impacts on setting the thing up so they can become very complex and unwieldy at times to use. And if you've only set up for an electrofisher and it breaks down, then suddenly you don't get any actual fish, which is something to really think about. So I would definitely, if I was going to do electrofishing, I would be taking nets just in case you have a tech failure. But in a lot of areas, we've actually moved towards same fishing because that's basically what local people would use. So if you're essentially trying to work for a client or a government department and you're trying to assess river health, one of the key issues is how much of the local fishery is being consumed by the local inhabitants. And so if you're using nets and other similar methods to catch the fish, that's replicating what the locals would be doing in their spare time and possibly to get food. So it's a very good way to look at catch per unit effort and other things like that that we would use to assess the fish community. We do need to look at fish for food safety requirements. If you've got industries which are potentially discharging toxicants which can bioaccumulate, you do need to look at food safety requirements. In many places in the world when we work in the mining sector, we find that artisanal miners are actually illegally attracted to the functional mine site and they quite often mine upstream and downstream. And increasingly our job is to actually separate the large scale miners impact from the small scale messy impacts of the artisanal miners. Some of the artisanal miners use a lot of mercury to process gold, et cetera. So we're finding increasingly mercury levels in water courses again which is very unhappy for me because I've had a lot of experience in picking up trace mercury in the west coast rivers of Tasmania for instance which has come from gold mining in the 1860s which is very sad to see it popping up again. So food safety is huge. And because of fish migration however their ability to move away from pollution we can't really use fish community data to assess impact because as soon as there is an impact the fish are basically gone within hours. So it's not a great way to assess impact. I mean one of the things you can look at is presence absence. If there's no fish there when there was fish there then obviously something's happened. But as pollution impacts in rivers are generally a matter of hours and then they're washed downstream the fish will generally come back so you'll just miss the pollution entirely. But there is great relationships between water quality, sediment quality and fish tissue which need to be followed through with because if you find something in a fish tissue it will trigger a bit of research in both sediment water quality and other things. Next slide Richard. Do you use E-DNA? Do we use what? Sorry Richard? E-DNA? Oh we've thought about using DNA. One of the big issues is cost for us and the other big issue is actually customs importation of biological material. There's a lot of messing about to get live stuff across borders and because we're doing a lot of work in Southeast Asia and Africa it's just easier to basically get bugs and things like that. Fish and vegetation samples actually sampled by local experts. They're quicker, faster. It's also creating local employment which I think is really important and giving local education and research opportunities for university people. So we have a lot of PhDs and people generally supporting us in the field in the fish and in the macroinvertebrate area and we try to do a lot of capacity building locally as well. So we don't do a lot of DNA work. If I was in Australia and doing it in Australia I would consider using it for specific targeted issues but it's not officially part of an AusRivers protocol. In an AusRivers protocol you're doing a live pick in the field and you're actually picking out the bugs as you see them from a white tray in a timed period which is often 20 minutes and then that is actually part of the statistical sampling stratum for the macroinvertebrate procedure. So I guess the short answer is no. The long answer is yes I'd love to and the longer answer is but it's very difficult to arrange. Okay, next slide. You're on it. Okay, macroinvertebrate sampling. So again we've got several protocols available for macroinvertebrate sampling. I prefer to use AusRivers because it's a really good protocol. It has holistic training and you get the people who are trained up in it get a certificate saying I can do AusRivers training. We've got the three key habitats there on the right-hand side. Edge habitat, pool habitat and riffle habitat. You're basically using a kick net to do the sampling and the way we do it is we reduce our 100m stretch of river which is the site that we're trying to at the back of the system. We try to take that 100m piece of river and we break it up into 10m representative section. So if there's two pools in that 100m section we try to do 5m in one pool and 5m in another pool. If there's three riffles we try to do 3m in one riffle 3m in another riffle and 4m in a bigger riffle to try and represent those different habitat types within our 100m representative patch of river. The bugs as they move very little they're demersal and immobile generally quite often bugs will only move 1-2m within their riffle or edge habitat during their larval phase. They're seasonal so they tend to breed in either spring or autumn planetarily which is interesting and they're also at the family level there seems to be very similar species globally which is also a really interesting thing that I've realised in the last 5 years when I've re-looked at some of my bug samples from Africa and Asia and Australia they're at a family level the niches in ecology are all filled by very similar bug types so it's interesting. As a result of them not moving very much at all during their life cycle they're very useful for detecting water quality related impacts for up to 6 months after a pollution event which is fantastic because as you would know if you do water quality sampling within hours a pollution event can be gone from a river so if you're not there and it's occurring exactly when it's occurring you're not going to be able to pick it up with a conventional field water quality test or a laboratory water container because it's going to be gone and I've seen this happen we've seen some artisanal mining where a bulldozer pushes a whole range of dirt into a river and it will be read for 10 minutes and then suddenly 10 minutes later it's clean again so it's quite extraordinary but an event like that can actually completely change in a macro invertebrate community entirely and you can actually look at it and then figure out what's happened in terms of what was in that sediment cloud elementally using these kind of multivariate analysis systems. Next slide. Okay pool, riffle, run. For people who don't understand aquatic ecology very well it all comes down to these aspects of a stream you've got the two images there I'll look at the cross sectional one first you can see pools kind of form as rocky barriers appear in rivers and everybody will have noticed that you can get pools in rivers where you have rock bars popping up so that's your pool sequence and then in the 2D plan view you can see that you've got the riffles popping up where rocky outcrops are coming from the geology and also possibly where sediments have been shifted or even some geomorphologies occurred where you run where you've got water moving along with a no riffle and a moderate depth and then you then form a pool you can also see edge habitat which will basically be near trees and riparian vegetation and that's basically how we form the concept of the three habitat types for an effective sound planning site. Next slide Richard. Okay now there's various risks with habitat types and I think these are important to bear in mind when you're about to begin a program because this will all be interrelated with upstream land use, upstream waste discharges, all sorts of types of things that we need to take into account when we're looking at the program because those things can affect the program more heavily than the industry or point source that we're actually trying to sample and it's very important to understand that at this stage. So pool habitats are very heavily impacted by sedimentation rates because the pool can fill up small macroinvertebrates which are often smaller than one millimetre can actually be buried by high levels of sedimentation rates which can kill some species off entirely and it appears that some of the macroinvertebrate types that are degraders and browsers in the bottom of pools actually can only handle certain sedimentation rates before they can't handle a level of change and they will leave. This is very much impacted by upstream land use. So if you've got poor practice agriculture or poor practice forestry for instance forestry roads with no stormwater controls when it rains all that sediment can end up in a pool and I think the summary is captured by usually most impacted from human activity the pools unfortunately. Edge habitat is a reasonably stable habitat if it's not damaged by poor riparian land uses. So in this situation you could have for instance unrestricted stock access to a river. The stock with their hard hooves can walk across down to the river, smash down the banks and kill a lot of vegetation around the river which immediately degrades the edge habitat quite strongly. You can also have weeds that can affect edge habitat as well if the weeds overwhelm the edges. Edges are classic disturbance ecosystems so weeds are very attracted to them and edge habitat can also be affected by water depth. If you've got long-term drought or long-term floods you can have very interesting impacts on edge habitat and macroinvertebrates communities. Rifle habitats is usually the most biodiverse habitat that we sample after edge. Rifle is often the most stable because it's less impacted by land use. So riparian changes don't really impact riffles because they're in the middle of the river and vegetation removal from upstream and in the riparian zone doesn't tend to affect riffles too much either. Occasionally you can remove a large tree and then a shaded riffle can become a sunny riffle and that water temperature change will change the community somewhat but even with that sort of impact it will generally be a percentile change rather than a major change to the macroinvertebrate community found in that riffle. But they are impacted by elevated sedimentation rates so if the riffles I mean I always think of riffles as like multi-story apartments if you fill up the multi-story apartments with rubble there's not going to be many people living in them and I guess the riffles just like that. Once the sedimentation rate goes up too high you've got a problem for the riffle. Next slide Richard. Sediment sampling so we do composite sampling across the 100m stretch of the river usually 20 to 30 sites. Again this is based on success, velocity, safety, etc. We have to preprocess that for a laboratory in the field so we generally put it through a 1m or 2m stainless steel sieve or something equivalent to try and separate out sand grains from what's being carried in the sediment. Different depths of sampling are relevant to depending on what you want to pick up so if you want to know what happened last year you're probably sampling shallow. If you want to know what's happened 10 years ago you're sampling a bit deeper and it's quite important to try to determine your sedimentation rates in a river to figure out what those sampling depths are and that's an art form all until itself. Definitely gives you lots of good data on seasonal deposition and metals precipitation. Water quality and sedimentation sediment quality is always interrelated in a river but sometimes with the spot sampling of water quality the statistical relationship is not that great and again water quality can change quite a lot at any time and the great thing is the sediment is a bit like the bank balance so if things are settling out of poor water quality events that will generally be picked up in sediments which is a very good reason to sample sediments at all times and again parameters, metals, major cations and things like that it depends what you're looking for. Definitely focus your program based on risk so if you know there's a specific risk from your performance source discharge or land use activity you try to sample the sediment for those kind of issues. Next slide Richard. Water quality sampling, another key component of a multivariate program. Again everybody's probably familiar with the key parameters we use for water quality sampling. Field water quality is very important to take at the time of our sampling and we're looking at typically temp, pH, ORP, EC, turbidity. Field alkalinity is very important for the AusRivers protocol as field alkalinity was discovered to be one of the major pivot points for the statistical analysis. I think the sensitivity analysis of field alkalinity placed it in one of being the top one of the top water quality parameters actually that really does change the macroinvertebrate community structure so it's a very important thing to pick up. Not so important if you're not using the AusRivers scoring system there's a bunch of indices you can use for macroinvertebrate community assessment. If you're not using AusRivers field alkalinity which involves usually field titration of water and stuff which is complex and you need to train people how to do it may not be required for all types of protocols. Lab water qualities very useful to get nutrients, major trace metals, specific point source toxicants if required very important to again structure that based on what's likely to be in the discharges and it's very important to select our indicators which are relevant to our potential impacts for a really good statistical effects analysis and then do them at every site in the program. One of the things about multivariate analysis is you really do have to have large amounts of data and you have to do everything at every site so you can do cross comparison. Next slide Richard. Before we move on there's a few presentations at Ecoforum this year that Olga runs and one of them was all about these emerging contaminants right and I just wonder like when you're selecting which toxicants to actually analyse for do you choose the organic stream I suppose like it seemed to me that this chap was saying that we're really selecting them based on what a cost to analyse and you know where we've got a methodology that's fairly straightforward to analyse and he was really concerned that there's a whole wave of these emerging contaminants that are just not on our radar even in terms of that. I'll look Richard all I can actually do is agree with that person. I mean when I was at TazEPA Chemwatch database used to come on like 5 3.5 inch floppy disks then it was 1 CD then it went up to about 1 DVD and then I think it was about 19 I think it was about 2002 Chemwatch sent us 6 DVDs and they sent us a letter saying hi we're going to continue to look at about 50,000 chemicals 50,000 new chemicals a year in the Chemwatch database however the human race is currently making 500,000 new chemicals every year and that's exponential so I mean I don't know the answer to it I think the human race really has to control its chemical production one of the things places like Chemwatch and the National Water Quality Management Strategy does is they have models for looking at dangerous molecules Richard and you can take the molecular structure of something which might be from an upstream activity you can pump it into a model and then it will actually give you a pretty quick and dirty high risk, low risk moderate risk from the structure of the molecule I would really recommend people get into that sort of analysis before they start spending lots of money on very complicated testing which is quite often if it's not done properly going to give you garbage results anyway so I would always say if you're going to test for something complicated make sure you do it right because there's nothing worse than getting the wrong answer when you're not doing it right unknowingly so yeah definitely do it properly and it's an invidious issue I think there are so many new chemicals now and there's no real way of actually being able to figure out what's going on apart from doing very decent upstream research so I think when you're looking at managing a river you really do have to as a first step analyze all of the upstream stakeholders and contributors to that river in terms of discharge and even just activity because dust being blown off a stockpile can sometimes be a major issue I'm thinking of the Derwent in Hobart where you've got the EZ electrolytic zinc stockpiles with a whole bunch of heavy metals which blow into the Derwent regularly as dust from their old stockpiles and that dust is in the bottom of the Derwent and it's going to be there for a long time because the Derwent is now a controlled river and where the EZ factory is in a part of the drowned river system so it doesn't really get high flows anymore it has tidal influence but it's quite deep so you've got issues like that which people might not immediately associate with water quality monitoring and you have to take into account all those issues when you're looking at upstream industry it's a complicated issue and it really requires a very high level of expertise and research to really figure out what the likely risks are and then you need to carry those risks through to your program and make sure you've got appropriate parameters in it to pick those risks up so it's very complicated and I really don't know the answer because there's a whole bunch of stuff we can't even test for Richard so I'm not even sure how we deal with that I might have derailed the presentation that's okay, it was a good derailment I think we need to realise that the modern technology is really a big thing and we can't even monitor for everything it's an issue that I think our young people in particular really need to have deep things about okay, part two, data testing and multivariate analysis this is really the engine room of a multivariate program of course next slide Richard sorry it all comes down to statistical testing of your data so you've been out in the field for two weeks, you've collected ten sites, you've got all of the above for everything you've got you know you've got say 12 macro and vertebrate filer you've got probably 5 to 20 sub family species, you've got these giant spreadsheets of data, you've got all this water quality lab water quality, you've probably got 60 parameters in water quality you've got 60 parameters in soil sediment you've got all your field parameters and you've got at least probably 5 to 6 parameters of field parameters field water quality parameters at each site you've got all of your riparian vegetation community assessment you've got all this packs and packs of data and you basically then have to split it up turn it into a big table of summary data and then start doing some statistical testing on it to make sure you've got sufficient variation in that data to justify that it's actually going to stand up statistically to testing and these statistical tests are essential and you've really got to look at things like minimising your possibility for the null hypothesis we've got to look at relative predictive power of all the ecological indices that we're using for the description of all of the different ecological communities that we're testing and that's macro and vertebrates and riparian vegetation and sometimes aquatic plant vegetation some people also use macrophytes some people use attached growth there's a whole bunch of crazy ecology which you will need to drag into this test to make sure that you're actually getting valid data and there's quite a few relationships between the ecological indices you're using water quality, land use and habitat so again you've got to think very closely about all of those relationships when you're looking at these tests next slide Richard so to do this data variation testing we're using F and P tests the F test is probably the classic critical test for the confirmation of rejection of the null hypothesis and if you've got a good test you should have a really good F test which means you're going to have valid variation for the calculated degrees of weather skies so if we can get David back seems to have dropped out large numbers of sites but fewer numbers of parameters you can also have difficulties with making sure you've got the appropriate levels of variation so high relative values for the F tests usually shows that the data is good for multivariate statistical analysis and the P value is just giving you a probability estimate the calculated F value is going to be lower and if you've got a very low P value and in this table below you can see we've got a P value of 2.8 by 10 to the minus 9 which is a pretty low probability that it's wrong so in that one we're looking happy and we've got appropriate degrees of freedom within groups and between groups and the F test is higher than the F critical which is great to see so whatever data we've applied that to that's good data and we can be happy knowing that it's statistically valid for our testing for multivariate analysis next slide David you did drop out for a couple of minutes there what did you miss out on probably the first part so maybe if you just introduce the F and P test again the F test is critical for confirmation of the rejection of the null hypothesis in your data set and if you get the appropriate F test you're going to have a valid variation level for your calculated degrees of freedom and that means that issues such as having too few sites and too many variables and too many sites and too few variables can be picked up within this testing and to some extent this has to be dealt with by an experienced practitioner when they're setting up the program in the original setting up of the program you can test it when you've done the program because you can get crazy situations where you have seasonal drought and suddenly you don't have a great F test and you have to then adapt your program to deal with that sometimes you have to drop data in order to make the multivariate assessment work because you don't have a significant enough variation in the data to actually do multivariate testing on it and get an actual result a high relative value for the F test and the data is generally showing you that you've got a good result and that you can do lots of multivariate and statistical analysis effectively on the data set and that will be accurate the P value gives you the probability estimate of the calculated F value being lower than what's estimated and low P values provide the good confidence in an F test so in the table below you can see the P value there is 2.9 by 10 to the minus 49 so that's a very low probability the test is higher than the F critical which basically means that this data set we've tested here checks out as being good quality and we're going to get great levels of analysis for the degrees of freedom you've got degrees of freedom between the groups of 17 and within groups 214 so that's generally a pretty good sign of a good program next slide okay choosing appropriate indices is super important for our ecology analysis and you can apply them to life, riparian life fish life, macroinvertebrate life there's many many indices around there's diversity abundance, richness, ASPT which is good for sewage and landfill BMW tie we use in Asia quite a lot that's great for nutrient impacts and also picks up sewage impacts very well and then of course we've got our osrevers systems which are good for toxicants good for all round impacts but they're only really set up for Australia and Indonesia and you can adapt them for other environments which is what we tend to do at earth systems we contest these indices by looking at linear regression models and you can see here we've got one model down the bottom there which is richness versus signal that's got an r squared of about 0.42 which is good whereas other things such as abundance versus richness richness versus filer and abundance versus ASPT for this data those indices weren't as good so we're suddenly thinking richness and signal is probably not a bad method of analysis for this particular data set next slide Richard and another great test is similarity testing to look at the suitability of various parameters for assessment as well so we apply them to all of our multivariate parameters this is very complicated this method and sometimes requires large amounts of processing power to do depending on how big your data set is but in the one to the right we see a stripped down system where we've got quite a few indices richness, filer, etc set up with field water quality so temperature, pH, ORP, DO and AC and you can see from that that data set that we've got our pool, riffle and edge indices coming up as being extremely interrelated with the water quality indicators which is exactly what you do want to see whereas BMW, PTI, ASPT TrueDiversity, Shannon and Simpson, they're off to one side so they are similar but they're only at a similarity of about 0.1 whereas our edge pool and riffle indices are at a similarity of about 0.8 which is extremely similar and that means you've got a great data set in those indices so you'll tend to use the pool riffle indices that you've got to do the analysis on the major system effects lots of tests for similarity are available like Bray Curtis which we'll use here on the right next slide. Spatial mapping is extremely important because we've got upstream downstream land uses we've often got multiple backies in a situation you might have upstream and downstream at multiple sites like we do in this picture and you might have a reference site as well which we do in this picture so it's incredibly important to see what's going on you can quite often see increases and decreases in water quality as you go down the river system which you've got here as you can see in the middle of this river system you've got a score of about 5.09 then it declines because of particular human impacts in the creek including quarrying and sand mining drops down a bit further as a result of some downstream artisanal mining and then pops back up later on to 5.1 because the river's had long enough to recover and the tributary to the north has delivered some good quality water with a score of 5 so you can see that it's really essential to look at spatial mapping and and I really can't recommend that enough especially where you've got complex land uses so in southeast Asia as you probably are aware any catchment you can point a stick out will have so many different activities going on there'll be upstream grazing, agriculture forestry, heavy industry, all sorts of stuff and you need to really look at these scores carefully because at each point in the river you might have a particular upstream impact which is causing that result next slide Richard really should have got that glass of water detailed impacts and effects analysis now this is one of the really strongest tools that comes out of the multivariate macro invertebrate style and ecology style analysis of a river you can see on the right hand side there we've got a way of providing the indices there we've got file a mapped against indices scores and when we use that 2D framework we can actually break impacts down into these what we call the quadrant system you can see where we've got near-natural conditions we've got high levels of file and high levels of indices scores you're looking at a near-natural system which is a relief you're sort of thinking okay this river is within some sort of natural tolerable range but then once you start having file are dropping off or indices scores dropping off you're jumping into such as specific toxic and impacts we've got low levels of file but high scores so you've got specific toxic and only impacting specific species which is a really interesting signal to come out of the whole system or you can have low scores and low file which means you've probably got all-round impacted water quality or even riparian zone massive land use impacts which is something else that needs to be managed differently where rather than worrying about water quality you're really having to go back to step one people okay you need to fix your riparian zone habitat because there's so few trees left there that the ecology is suffering fish numbers are dropping bug numbers are dropping and then of course you can have the fourth possible outcome which is high file low scores and that's a classic situation you get from salinity and habitat impact so if someone has really messed up the edge habitat by having stock access for instance that that can very frequently push a stream section into that zone the grey zone as we call it where you've just got scores much lower than they should be but you've still got a lot of bugs hanging on in that habitat type. Next slide Richard okay now one of the penultimate things that you can do with this kind of detailed multivariate analysis is actually work out things like upstream land use relative contribution to index scoring and you can see here we've prepared a pie chart which gives you the relative impacts of each of the upstream impacts you've got things like you know being able to put an actual percentage on the amount of impact being caused by things like industry you know townships and their landfills agricultural impacts things like sand mining in this in this particular river was found to be one of the biggest issues for their local area we were monitoring in so it's incredibly useful because when you're giving this data to a government they can then say okay this is how we're going to have to split up our budget this is how we're going to have to apply the effort this is where we're going to be pushing the wheelbarrow for instance rather than immediately going to see industry and saying hey we need a million bucks to fix this problem they're probably going to be saying hey wow township and agriculture are big big stakeholders in this problem so they need to be at the bargaining table as well. So it gives the ability to do some really interesting follow-up recommendations for clients and for government departments sorry about that just coughing a bit and it's a very powerful technique and I can't think of any other way that you can really back this kind of information up statistically without using multi-variate river health assessments anyway next slide Richard Do you want to grab a glass of water? No I think I'll make a thing I'll make it through the end Richard Okay I could have a bit of a break in a minute Alright so maybe I'll do the key take-homes while you grab a glass of water if you want. Okay, shall do thank you Richard So what have we learned from David today? Well it's certainly learned a lot in terms of program design and data sampling well clearly it's very important to get that bit right this concept of before after control and impacted site selection is essential for statistical analysis of impact secondly the statistical program design is essential to remove various external bias from the data so obviously a lot of potential for that to occur like David's mentioned things like dust blowing in from stockpiles and that sort of thing for visual aquatic habitat assessment the Oz Rivers visual assessment protocols provide a good approach and it sounds like eWater can provide training in that methodology so if you're interested in that I'm just reaching out to them multi-parameters need to be sampled in the field simultaneously so you need to have a broad range of people of different skills to get this right was something that I really picked up in this so sediments, water quality, macro invertebrates, fish as well as the riparian vegetation side of things so a lot of different skills required to do it right in terms of data testing and multivariate analysis it's always a thing that's underdone in this you know in a lot of studies which is just having enough data to statistically assess your dataset for variability and QAQC so really important because that really does underpin everything that you then conclude it's important to check ecological indices for their applicability and accuracy it'll be very easy to use the wrong one and as David said there's lots and lots of them there are some very useful statistical analysis procedures such as linear regression and similarity testing so finally when the program is built and analysed it's a very powerful tool for detecting effects and impacts in aquatic ecosystems that is it allows us to determine what action causes what impact and I thought that was really powerful that pie chart that showed the various industries and their percentage impact because it'd be very easy to spend money in the wrong spot for rectifying sites so I thought that was really powerful now we better move straight to the questions so in terms of our early bird questions we've got four first one has this work been or is it being developed as a method for regional water accounts under the accounting for nature framework David yeah so we thought it's probably not I don't know for sure but I would certainly recommend that it is being used for regional water accounts under the accounting for nature framework we very probably collect enough data to use these kind of approaches to do this work but I can't be sure that it's going on question number two I think we might have covered this in the presentation how statistical methods can distinguish upstream downstream differences in water quality yeah so really I think the importance of a good backie design is really the answer to that question if you've got a good control site you can really measure the differences in upstream and downstream seasonally through time and also individually which is quite important because the upstream impact will be close to what the control site is if something changes in the upstream site that can be quite often very responsible for the downstream changes as well and that's the key point is to have a control site which is in a completely separate river not impacted by any of the same land uses or other issues point source discharges etc in your monitoring program and again that's what that really is the rub and it's also the tough thing because control sites can be impacted themselves so selecting a good control site that has little human potential impact is essential and difficult so where can this go wrong I suppose what are the pitfalls in this method yeah look I'd say the pitfalls are really Richard that it is very complicated there's a lot of different people involved there's a lot of specialist inputs so I think the pitfalls are not employing experienced people to do the work you really need people who are top of their game in all these areas you need to have very good communication I think if you've got people off in their ivory towers and they're all siloed into different functions within the project you're gonna possibly have pitfalls because people aren't talking to each other about what each different team needs so communication is extraordinarily important getting the right experts is extraordinarily important and then following good procedures in the fields having established protocols well trained people knowing the limitations of your people and knowing the limitations of your laboratories is also incredibly important so working very closely with your home laboratories so finding out the limitations of your lab finding out things like detection limits and stuff are all essential and then feed back up into the major pitfalls and they're pitfalls if you don't undertake those those if you don't have an answer to those issues they're major pitfalls that can occur other things that can happen for instance I was talking about before you can set up a spectacular program one year and then next year you can come back and there's been an avalanche a landslide through one of your key sites and it's no longer a key site that can create a huge issue if you're using a control site for inter-annual variability so yeah there are potential pitfalls I mean the answer is you can't control everything but having a backup plan is always good so having two reference sites is sometimes a great idea and we try to do that as much as we can having two or three reference sites is very very useful but again getting harder and harder to be possible in this day and age okay the last one of the early bird how much data do you need to make this work well the answer to that is as much as you can get the more data you have the better the method works but again it needs to be selected within a budget so you can't have an infinite amount of data and again it comes down to choosing what the monitoring program needs to pick out of the system so again I think one of the key pieces of data you need is what are the upstream land uses and potential points or stitch charges and other impacts that the anthropocentric parts of the aquatic habitat are being exposed to if you've got that data then you can flow that data down into the construction of the entire program including the parameters but the short answer is as much data as possible and I think to get it to work properly you can pare down the system you can for instance not do sediment if you want to you can potentially not do lab water quality but some of the data is non-negotiable field water quality, field alkalinity and macroinvertebrates if you want to do a really good multivariate analysis you really really need them and without having lab water quality you are missing a major possible source of data that can give some absolutely illuminating and targeted answers for your client or your government department that you might be doing this work for so I think yeah the short answer is as much as you can get and the long answer is but you can strip it down a little bit thank you now we've got quite a few questions in the Q&A so you've sparked some interest well done I'm going to read those out if you could answer those first question is from Andrew Thexton sampling methods such as netting may affect water quality how is this taken into account we do all of the water quality sampling before we get into the river to make a mess that's probably the first way we take into account secondly if we're talking about localised impacts on species because the sampling being undertaken so quickly when the disturbance is occurring all of the bugs and all of the fish are still there after we've done the water quality and sediment sampling we then collect the macroinvertebrates sometimes we collect macroinvertebrates fish and sediment simultaneously just by moving to different sections of the river and using the river to flush itself as it needs to and for instance for fish quite often we're saying that across the bottom of the run that we're testing and the field fish ecologists will walk up the reach looking for the fish and then they'll go and check the lower net to make sure that they've got all of the fish they're looking for so there's a variety of techniques to make sure that you're not actually dirtying your sample in inverted commas and that's basically the way we do it so you do your water quality first and then you're less sensitive if you do first and then you work through into the less sensitive parts of the methodology as you go down the hierarchy of sensitivity. Okay, good. Dan Evans another problem with using eDNA might be a lack of DNA lobby data the species that David is focused on in other countries. Yeah, that's possible because we're working at family level it's probably not as bad as what we would think if we're working at species level I think we'd be in big trouble but family level is probably not too bad because I think there's pretty strong genetic similarities at family level but look one thing that I think would be very difficult is again just moving that biological sample around the place because I think the number of institutions that can do really decent DNA sampling from aquatic macroinvertebrates it's a pretty short list of people and there's no real established methodology for doing it that I've seen yet Richard either like I've seen methods varying from like throwing all of the samples into a blender and then sending it off for RNA analysis and I've also seen where they preferentially sample the macroinvertebrates and then put them into a blender and then do the same thing so I don't know I think before it's a useful technique we need to have a protocol we need to have somebody to sit down look at the statistics of the situation model it make some decisions about the statistical significance of the DNA sampling methodology itself then finalize a methodology and then we do it but at the moment it's more of a I don't know academic interest style thing in the future it will probably be becoming something where it's very quick like you can imagine in a bright future you have some sort of field DNA test that you can just, you know, vitaminize your bugs into throw it into this thing and a whole bunch of bars come up saying here's your species list that would be fabulous but I think that's a few years away yet we did have a webinar on this must be 12 months ago they did have some methodologies and they did have a very comprehensive library of the DNA that they referred to so I might hunt that out for you it's a good one to look at we do look at it from time to time but it's a very specialist area and again, I think we just have problems importing material from other countries to the complex labs that can do the work I think the person who asked the question is probably correct too there's probably some issues in transferring species between continents like I'm not sure the DNA will be exactly the same so that would be something we should definitely look for in that presentation it might be an Australia based thing all right we'll move on from that slightly contentious topic next one's from an anonymous attendee how would you quantify whether harm is trivial or non-trivial from the water quality perspectives that you assist well luckily that just comes naturally out of the program the program is extremely subtle and has multiple levels of ability to detect change and I think in the indices scores there's quite a lot of academic research which has been done over the last 30 to 40 years where you can relate a particular indices score to being low risk, moderate risk or high risk and probably even shades of grey in between that so yeah the system is actually extremely easy to use to judge those impacts whether or not humans want to do anything about those impacts that's a completely different thing you can say to a government department wow you've got this problem and then literally nothing will be done about it one of the issues for me is not really the science it's actually the next step which is taking the science to the stakeholders and then making sure that the budget gets cut up correctly from all of the various people who own risks and are causing risk and then it gets actioned and to be honest that definitely is the Pareto effect I think the monitoring of this stuff is 10% of the work, 90% of the work recommendations that come from the science where people are fixing stuff I just ran out of water sorry about that you might be the first presenter we've killed yeah I only had Covid 3 weeks ago Richard so I'm still coming back probably should have your mask on David next questions from Jean Meeklen for unknown chemicals try looking at R-G-A-R-C-H-I-E database D-C-C D-C-C D-C-C-E-E-E-W that's this top assessment looking at chemicals by structure similarity and other factors excellent that's more of a note rather than a question isn't it yeah I think that's a really good bit of input I didn't write down obviously I didn't prepare for that answer so I think that's if people are interested in that specific issue that's a great great model there are other models which also do it and I'm very sorry but they've all slipped out of my mind I could probably Google quickly but I think we'll just leave it to everybody else to Google and that comment I think can give a good start Google okay how high is high value like well in OzRivers the system is A or X which is super abundant which quite often occurs by a sewage outfall and then we have A, B, C, D so D to get a D with OzRivers you really are looking at a situation where you've got something like 50% of all the macroinvertebrate species impacted missing with presence, absence, data recorded to get that kind of level of problem you'd have to have either some massive land use upstream which is really impacting sedimentation and water quality levels such as poor practice agriculture, poor practice forestry poor practice road design things like that so it would be something that's highly impacted you'd also have to have multiple missing parts of the riparian zone the bugs quite often have a life cycle which includes macrophytes in the stream edge include some of the trees, they lay their eggs all over these places and then sometimes the eggs hatch fall into the water sometimes the eggs hatch into a larval phase which then falls into the water so removing the riparian habitat and riparian vegetation is often the key causing big problems for a stream and there's plenty of streams that I've monitored where the water quality is actually quite good like it's possibly at a 95th percentile 99th percentile level of toxicants based on the Osriva sorry the National Water Quality Management Strategy and ZEC guidelines but then for some reason the macroinvertebrates are missing and when you look carefully at the surrounding land uses it's because the surrounding land uses are very poor but yeah to get a D we have to have a bunch of stuff missing from the ecology alright next question another anonymous attendee do you have any general go to transformations for biotic and abiotic datasets prior to assessing similarity and dissimilarity yeah look there's quite a few transformations I think I tend to follow the protocols for various indices and I think throughout the talk you saw me throwing back a few times to things like BMW, PTI ASPT we've used the Osriva stuff we've used Signal there's a whole bunch of standard ecological indices which we'd use for transforming data we have used things like twin span style analysis which is a statistical ecological function for looking at community similarity and variance you've got things like similarity analysis with Bray Curtis there's a swath of other similarity and difference style functions which you can look and apply to your data and then I think you can actually select parts of your data after you've done that kind of analysis and create sub data sets which you can then do even more detailed variation assessments on to get much more focused results and when you've got huge data sets that's a very very important process so yeah there's a whole bunch of ways of doing things I tend to use data transformation based on the data itself so it's very difficult to say look here's one way of doing it because as you investigate the data in nearly every data set you can even be looking at the same river from like 2022 and 2023 and you'll see different indices, different trends in the data actually result in different systems being useful for really picking up the precise, fine tuned aspects of the analysis like if you've got scoring shifts where you've got scores of great difference that's always the thing that you want to pull into the program, variation is really important if you've got a scoring system which is giving you zero variation there's not a lot that you can say about it in any real relative term so it's really about sorting through your data, checking it for QAQC looking for variation looking for the issues and the indices which have the most variation in them and then really nutting down on those and figuring out why Thanks for that Are you happy to keep going for another 10 minutes Yeah, for sure, assuming my throat Subject to that If you see me fall over Richard just end the thing quickly please A little eulogy for you Next questions from David Bacetto Is it possible to get the copy of the presentation and the video recording Yes, if you log in to our website on Monday afternoon we will have the recording up there there's all the recordings of all the webinars it's under the webinar heading and they're all there with the video so feel free to grab that Next question from Rain Ready Rainy Has this been done for the Murray-Darling River where it is really needed across states That's a good question and the answer is I'm going to give it pretty much a yes not exactly the same system that we've used at Earth Systems because our system is very complicated and includes probably one of the largest collections of simultaneous data that I can think of There's a lot of programs like the Derwent Estuary program springs to mind where they collect sediment at a different time to when they collect macroinvertebrates they collect water quality but they collected at a different time to when they collect sediment so I think it's the simultaneous collection of data that we do at Earth Systems which provides a really great snapshot of what's really going on when we apply that to the client's data and the government data which is long term and probably regular like monthly or similar time steps to that the data that you've got on the day into that data and it gives it a great context and a great level of analysis that's not possible in any other way and the Murray-Darling does have Osreva's programs done on it and it is part of the National River Health Project I can Richard sorry you cut out there for a minute Victoria has its own Osreva's protocol so Victoria used to have its own kind of like Victorian river health assessment based on bugs and macroinvertebrate assessment as well so you can Google for those kind of things to see this kind of work being done in Australia but again I don't think you're going to see programs where they've taken into account as many issues as what we do the stuff we do at Earth Systems is quite often based in the mining sector so the discharges we have in the mining sector are highly technical and highly complex whereas a lot of the work that's being done in the Murray-Darling Basin is aimed towards agricultural and other land use styles to different types of impacts which can be picked up different ways so you would probably not need the type of complexity that we have in our systems I'm not sure for instance that they take fish tissue analysis into account in the Murray-Darling work and I'm not sure how complicated they undertake sediment sampling in that program either I think they really are just looking at riparian zone habitat and bug scores for that sort of national river health stuff hopefully that answers that and I could be wrong so if anybody knows more about that feel free to chime in you better not be wrong that's a quite answer David next question from Rainie ready have you observed any clear impact of climate change in the river ecosystems you have worked on yeah I won't say climate change but I will say climate variability as I said before we've got clear signals of drought we've got clear signals of flash flooding we've got horrendous levels of flash flooding that actually modifies stream beds and takes out entire sites that have been monitored successfully in previous years and there's literally missing pool habitat and things like that you've got vast ecosystem change yeah I think we have climate variability impacts that's why we need control sites because the issue is the world is a dynamic place you can have a one in a thousand level storm up catchment of your target river there's no doubt the river will be a different river after that one in a thousand year rainfall than what it was before and hopefully your control river is not as impacted by that rainfall event or even better you have three control sites one is impacted by the rainfall but the other two aren't and then you can actually compare within your controls to see what the impact of that one in a thousand year storm was and then apply that to your upstream downstream and impacted site analysis data to figure out what's really going on so yeah I think the short answer is no I haven't done any climate related specific stuff but we definitely have got clear evidence of the fact that climate variability impacts on these work and if climate variability impacts on the work then there's no doubt that climate change will shift results through time gradually and we need to start looking at that as we move into these SSP 8.5 times that we're living in do you ever do that? do you have any projections based on the long-term climate change modelling that's been done? I do Richard for many clients I actually derive statistically downscaled future climate data from anywhere between 2050 to 2200 very complicated stuff and we've just moved from CMIT 5 to CMIT 6 so instead of having 40 general circulation models to do this work on we now have about 170 and we don't just have I think we don't have 3 or 4 scenarios anymore now we've got about 6 or 7 climate scenarios it starts at about SSP 1.1 or something rather and goes up to 8.5 in the old system CMIT 5 there was only a few fewer ones I think it was RCP 2 up to RCP 8.5 one thing I'd like to actually ask the group rhetorically is should we have SSP 10 because do we have evidence that we're even going to achieve 8.5 and 8.5 is our current trajectory so it's not very optimistic but maybe we need an SSP 10 to come out of the IPCC modelling they're going to have trouble answering that unless they write in but let's move on if there's anyone from the IPCC on the call we'd love to hear from you next question from Dan Evans can David please elaborate a bit on the relative contribution to aquatic ecology effects approach guessing some form of mixing model cumulative impact analysis would be required because localised observed river inside impacts might not necessarily be due to localised land use or contaminant sources golly that's a complicated question to pull apart I think again the short answer is you really need to take into account every single piece of data that you can collect to the particular site so you need to look at immediate land uses in the riparian zone you need to even look at change such as shading of river systems and things like that then you need to look beyond the river banks into what's happening immediately outside there so farming, agriculture industry, whatever's going on and then if you are aware of the fact that industry is using a mixing zone at a site I probably would avoid the mixing zone we know the mixing zone is going to be an area of reduced environmental values what we really want to know is is that mixing zone sustainable and what size is the mixing zone so I would probably recommend doing this kind of work at the edge usually mixing zones have an agreed size and let's put a site at the bottom edge let's put a site at the top edge and then let's try and actually figure out is that mixing zone causing an impact but I don't think I'd use this sort of work to detect impacts in a mixing zone per se you can easily do that but I think once you're in a mixing zone you have continual toxicants arriving in the ecosystem and I think probably the better way and the more I guess simple way to look at these impacts would be just to use the ANZAC 2000 and ongoing guidelines because we've got specific toxicant levels we've got specific levels of protection within the mixing zone and you can easily make a decision what levels of protection you're going to have in your mixing zone you could do an AusRivers program to validate those levels so like if you're saying hey outside my mixing zone there's a 99% level of protection you could easily go downstream do AusRivers and then figure out if you've got 99% protection and things like that and you could also do the AusRivers program within the mixing zone and if there was a level of protection within the mixing zone like say they were using 80% you could probably use it for that but I know some mixing zones are set up so they're actually very much below the ANZAC guidelines like you might have a 50th level of protection just extrapolating down in terms of guideline levels or extrapolating up depending on which way you're thinking about it so yeah you could do it but it would be different strokes to different folks if I'm in a mixing zone I'm going to be looking a lot more at toxicants I think a framework in the ANZAC guidelines is very well established and very complex and useful in its own right Thanks for that A couple more questions Margaret Peale why does sand mining have such a major impact on river health? That's a great question Margaret and the answer is basically sediment when sediment is released into a river the river stops being clear and sunlight in ecology where sunlight can get to the bottom of the river you've nearly always got happier bugs I think it's the analogy I've got is why do people move to Queensland when they retire and I guess it's because it's warmer and sunnier up there than it is in Melbourne similar thing in rivers bugs really like it when they've got sun they've got appropriate conditions if we're using the analogy of multi-storey apartment again in a river that's clear the multi-storey apartment all of the rooms in the multi-storey apartment are nice and big and lots of bugs can live in those rooms once you start having sediment arrive from things like a sand mine the rooms in the apartment fill up with sediment and you've got less room for the bugs you've also got less diversity in the bugs because most bugs have gills and sediment loading and turbidity levels on gills are only survivable to a certain point so very sensitive species like dragonfire larvae mayfly nymphs, things like that their gills are going to be impacted by sediment very quickly indeed and they will be not in that ecosystem they'll be gone from sediment pollution so sediments and sand mining are interrelated and unless the river has natural levels of sediment in it already higher levels of sediment are going to cause immediate change I think there's been a couple of papers written in Australia about things like major stock crossings in major river systems where they've done osrevers upstream downstream of major stock crossings and they've discovered that a major stock crossing will actually influence the seasonal osrevers scores so yeah sediment is a really big player and therefore appropriate land use and good quality or not good quality good practice land uses are essential to maintain the health of our streams but one of the main things is removing sediment and that's why there's been a lot of work done in the last 20 years on stormwater design and stormwater quality because stormwater carries a lot of sediment which impacts on our ecosystems in a very big way probably more so than point source discharges because there's far fewer point source discharges than there are stormwater discharges in most rivers very good okay last four questions I think we'll call it a day after that Will McBeth in classic Baki BA refers to before after in the temporal context but your BA David refers to levels of the spatial comparison additional to impact and control broadly how would you then monitor through time maybe dookie would be a more accurate name for your spatial design yeah it could well be that's true I always think of Baki and Mbaki as being in that particular annual sequence either in spring or autumn and I think we analyse stuff interseasonally and interannually through time because that's incredibly important to see if there's drift in our control data particularly and also to see if we need to apply any sort of drift from controlled data into the actual effects data having said that I think yeah all I can do is probably agree with that assessment that our the levels of analysis that we've got in some of our sites where we have multiple years of analysis is definitely a different level of temporal analysis but having said that I've done many jobs like this where we're just in that river once and so there is no real temporal aspect to it we're just giving a government department or a natural resource management agency or an industrial client a single data point which then they're taking into their management systems so yeah sometimes yes sometimes no thank you Gene Meeklin has sent through a link to the Australian regulatory chem informatics engine fantastic another great link yeah thank you I will circulate that through to you David Paul Webb Marie Darling Basin has done what they call a sustainable rivers audit which has included Oz rivers bugs sampling has been intermittent and is currently being redeveloped to also include non biophysical indicators so there you go David they're following in your fine footsteps there by the sounds of it I think one of the things they're doing Richard is they're trying to bring in more human activities I think so land use style things into their assessment system which I think is great we need to do that definitely in Australia there's so many people using the river for so many different things yeah last ones an anonymous one how do you account for lab holding times when working remotely in Tanzania the short answer is well you can't really quite often we will be in a situation where the lab the preferred lab holding times are exceeded the ice has melted in the the samples have probably got up to the same ambient temperature as the air temperature but you know when I talk to lab managers about what this really means 99% of the time they're not too worried because the tests we're trying to do are not specifically impacted by that change in holding time so again it gets back to my comments about talking to your lab if you know the program you're going to do you need to talk to the lab and say okay I'm going to exceed the holding times of this this this and this what do I need to do should I monitor for that parameter that lab manager will then give you an extremely detailed response about the potential impacts on your data the potential impacts on the reliability and quality of the result that they're going to give you and you then need to make a decision about whether you include or not include that parameter in your sampling program and because sampling programs are expensive if there's any doubt at all where there's a mismatch between what you're trying to find impacts from and the procedure to do it you you really are better off dropping that test because if it's not if it's not value for money if you can't demonstrate QAQC you better off not doing the test so I think working really closely with your lab manager on it literally every parameter for water quality and sediment and fish tissue is essential but we've found over the years that because of the things we've been looking at we really able to adjust our programs based on withholding periods sample preservation times and all those kind of things to make sure we still get great results one of the things for instance we don't do because it's just too much for us acidifying samples in the field we just don't do that it's too dangerous we've got so many things on our mind we don't want to splash HCL on ourselves in the field so we just get predosed lab bottles we change our approach to try and make everything as easy as possible by working really closely with our lab guys and it's astounding when you start working closely with really experienced lab people they generally have a solution to pretty much every problem you throw at them so yeah, just communicate tell them what the problems are and they'll generally have answers the second part of that question was I'm interested in local macro specialists and keen to find out if you're aligned with the university there so that's in Tanzania oh wow, Tanzania I'll have to take that on notice Richard we do have a guy in Tanzania but I can't remember who it is because it's been about five years since I've been to Tanzania so yeah, I'll have to take that one on notice the anonymous attendee will have to reveal themselves if they can just send us an email we can stay anonymous they'll just have to send me an email so if you can send them my email Richard I'm happy to respond to Tanzania thank you yeah, EarthSystems has an office in Kigali in Rwanda so I think we have a guy there who'll be able to tell me this answer pretty quickly because I've completely forgotten and it would take me a bit of messing about to remember no worries well that's it for today's webinar thanks very much everyone for attending and a really big thank you to David David that was excellent and obviously you're a real specialist in this area and a big thank you to EarthSystems that's the second webinar in this webinar season so many thanks for your contribution and sharing your knowledge no problems, did we go very far over time Richard well we always can move at a time this time we've only gone half an hour over comprehensive, excellent work great, I'm happy I hope everybody else is happy and if anybody has follow up questions you can email them to me and I'll try to deal with them as they come through thanks David thanks everybody for coming we'll see you later