 Good morning and welcome everyone to the Koala Coalab Conference Series for 2021. My name is Geoff Lundy Jenkins and I'm the Director of Southern Wildlife and Koala Operations within the Department of Environment and Sites. And I'll be your emcee for each of the six separate theme sessions that make up this year's Koala Coalab Series. I'd first like to acknowledge the traditional owners on the land on which each of us attends today's virtual event and pay my respects to their elders past, present and emerging and acknowledge their ongoing connection to country and the biodiversity it supports. Today's event is the first of a series of six themed conference sessions split over the next six weeks and builds on the success of the inaugural Koala Coalab event that was convened at Lone Pine Sanctuary back in 2018. This year's Koala Coalab Conference Series brings together a diverse range of industry professionals, researchers and community groups to share their knowledge, experiences and research outcomes to establish a basis for more effective collaboration on the threats facing wild Koala populations in Queensland. The Queensland Government is committed to hosting further Koala Coalab events every two years to ensure we continue to share our knowledge and experiences to inform and direct our future conservation efforts. This conference series delivers on a key action under the Community Engagement Action Area in the SEQ Koala Conservation Strategy and is founded on the principle that successful Koala Conservation relies on a collaborative approach across all sectors and communities who have a critical role to play in protecting local Koalas. The strategy prescribes that a coordinated and collaborative approach to habitat protection, restoration and threat mitigation is imperative to achieving the targets and halting the decline in Koala populations within the life of the strategy and to growing the Koala populations in the long term. Today's first session explores the theme of Koala mapping, remote sensing and habitat modelling. All of the six sessions run for about two and a half hours. There will be opportunities to ask questions of the speakers by the live Q&A box that you'll find on your virtual portal screen. These questions will be moderated by me as the emcee and any event that we have more questions and can be accommodated in the allocated question time will arrange for the presenters to be provided with any unanswered questions related to their presentation so that responses can be provided out of session. Before we kick off today's session, there are a couple of important housekeeping matters you need to draw your attention to. When you're in the portal, don't push your back button as this will take you completely out of the on-air platform. Go to the return to timeline to go back to the timeline to view the full agenda and join the various sessions that are available or to go to the meeting hub or to the exhibition. The timings for all the events are in Australian Eastern Standard Time. There is live support available. So just like in a live event, the conference organisers have provided a live concierge service. So just click on the live support red person icon in the top right hand corner of your screen to live chat to one of the support team. This will be monitored by the events team throughout the conference. As I said before, there's also the option for live question and answers. We want to hear your questions. Feel free to ask questions via the Q&A box. Please make sure you enter your questions here rather than in the discussion forum. Otherwise, they may not be seen by me and hence may not be included in the conversation. There is also a like feature on the live Q&A. If someone has asked a question that you are keen to hear the answer to, show your support and it will be moved up the rankings. There's also a networking overview. So there's an opportunity to meet people at the virtual event. So the meeting hub is available to all delegates and the place where you can connect with other attendees to arrange instant or scheduled meetings. Make sure your profile is up to date and don't forget to accept your connection requests. In order to kick off the event, I now have the pleasure of introducing Megan Scanlon, MP, the Minister for Environment and the Great Barrier Reef and the Minister for Science and Youth Affairs. Minister Scanlon wasn't available today due to Parliament sitting. She's prepared a pre-recorded message to launch the Koala CoLab 2021 conference series. Minister Scanlon was first elected to the Queensland Parliament in 2018 and was previously served as the Assistant Minister for Tourism Industry Development. In her current portfolio, Minister Scanlon has responsibility for climate change policy, environmental planning and protection policy. The Great Barrier Reef, Pollution and Waste Management, Marine and National Parks Management, Science Strategy and Youth Affairs. The implementation of the SEQ Koala Conservation Strategy is a key component of Minister Scanlon's portfolio. I know she was disappointed she couldn't join us personally today, but I'll hand over and you can listen to her recorded message to open the conference. Thank you. Hello everyone. Let me begin by respectfully acknowledging the Yagura and Terrible people, the traditional owners and custodians of the land I'm personally on today. And as we're connecting from different locations for today's virtual event, I'd also like to acknowledge all traditional owners and custodians across Queensland. I'd also like to acknowledge the many individuals and organisations, government departments that contribute to Koala Conservation in Queensland. Of course, koalas are one of our country's most iconic native species, but we know they face a range of challenges. Those include impacts from high-intensity bushfires, climate change, disease, dog attacks, habitat changes and car strikes. And our Palaszczuk government is committed to the protection of koalas and koala habitat. We have introduced the strongest koala protections Queensland has ever seen. These protections have increased both the area and level of protection given to koalas in southeast Queensland. That's been joined by an increased $7.5 million commitment by our government to wildlife hospitals in this year's budget, as well as a focus on designing and delivering infrastructure that emphasises wildlife safety. And that's all underpinned by a strategy backed by science and aimed at ultimately reversing the decline in koala populations. Using science and knowledge to generate new ideas and develop new technologies and approaches that can support koala conservation from high-quality mapping, monitoring, research programs that can help measure changes in population and threats over time. But obviously, we all need to work together. Successful koala conservation relies on a collaborative approach across all sectors. Government landholders, koala carers, First Nations people, researchers in the community all have a part to play. The Koala CoLab is a great initiative, one that will foster cross-section collaboration and strengthen our partnerships to protect koalas. And I want to thank all of you for your involvement in the program, for your passion, action to help protect this much-loved creature. Thank you for taking part in this event, and I look forward to hearing about the outcomes of this valuable discussion today. Thanks. Our first presentation in the Koala CoLab event for 2021 will be provided by Tim Ryan. Tim leads the Queensland State Government's Ecosystem Survey and Mapping Unit, based in the Queensland Herbarium. He has over 26 years of ecosystem survey and mapping experience from both public and private sectors. Tim's team is responsible for producing Queensland's foundational ecosystem-related datasets used in a wide range of planning and regulatory frameworks by all levels of government. The datasets include statewide remnant and pre-clearing regional ecosystem mapping, Queensland wetlands mapping, groundwater dependent ecosystems mapping, and high-value regrowth mapping. Today, Tim will be presenting in relation to regional ecosystem and high-value regrowth mapping. So just welcome Tim for his presentation. Thank you very much for that introduction, Geoff. Good morning, everyone. It's great to be here at this year's Koala CoLab, and I'm very happy to be presenting to you today on regional ecosystems and high-value regrowth mapping. I'd like to just start by pondering a problem quickly. So Queensland's really big. It's over 1.85 million square kilometres in the area. We know it's got high biodiversity that includes over 14,000 native plant species and a huge number of fauna species to match. It also is suffering from high developmental pressures with an ever-increasing population base. So it brings about the question, how do we capture biodiversity-related information in a way that facilitates businesses, governments, landholders, and organisations to plan for and manage the natural environment? And in Queensland, the answer to that question at least begins with the regional ecosystems framework. So in Queensland, the regional ecosystems mapping and framework provides foundational ecosystems related mapping and information that underpins environmental planning and regulatory processes, including the implementation of state and federal legislations, including the Planning Act, the Veg Management Act, the EP Act, now with Koalas on board, the Nature Conservation Act as well. And at a federal level, the Environmental Protection and Biodiversity Conservation Act. So derivatives of the RE mapping can be constructed to target specific conservation priorities or objectives, such as wetland management for reef water quality priorities, or as has been done recently in South East Queensland, koala habitat, and that will be the subject of the next talk by Stephen Howe. The RE mapping or derivatives of it are also often used to inform local government planning schemes. So the take-home message here is that all three tiers of government rely on the RE mapping framework in some form for the implementation of environmental policy and legislation. So what are regional ecosystems? So regional ecosystems are vegetation communities in a bioregion that are consistently associated with a particular combination of geology, landform and soil. So the classification of vegetation communities as regional ecosystems recognises the interaction between geology, landform, soils and vegetation patterns. And hence, the way the landscape is broadly functioning. The regional ecosystems also serve as a very good surrogate overall for biodiversity. So many of you in Queensland would already have experienced the RE classification system, but I will just breeze through it quickly. So REs are made up of three components, the first being the bioregion, the second is land zone, and the third is the vegetation community. Just looking at the first component, bioregion. Bioregions are large geographically distinct areas of land with common characteristics such as geology, landform patterns, climate patterns, ecological features and plant and animal communities. And in Queensland, we recognise 13 different bioregions. I'm sitting down here in southeast Queensland at the moment, ranges right to the Channel Country out west and north to Cape York Peninsula. The second component to a regional ecosystem is the land zone or sometimes referred to as the substrate. So land zones describe the major geologies and associated landforms and geomorphic processes in Queensland. There are 12 land zones in total. And land zone 1 to 6 are our unconsolidated land zones, which are made up primarily of deposits of sand, silt, clay and gravel. I won't go into the detail of each one of these specifically, I'll just whizz over them. But land zone 1, areas subject to marine tidal inundation, land zone 2, coastal sand masses, which include our lovely sand islands. Alluvium and River Creek Flats, land zone 3, tertiary clay plains, land zone 4, sand plains and deeply weathered landscapes. So land zone 5 and inland dune fields are land zone 6. You can see this picture up the top right here is an eroding river bank, which is typical of land zone 3, Alluvia. And here we have land zone 5, some old sandy plains, which occupies a lot of central and western Queensland. Just continuing on to land zone 7 and 12, these are soils formed in situ on older consolidated rocks such as land zone 7, adic jury crusts, eight are our igneous rocks, nine are fine grain sedimentary rocks, 10 is our medium to coarse grain sedimentary rocks, 11 is our metamorphose rocks and 12 are our older igneous rocks. So land zone 10 gives us our lovely sandstone range national parks that go right up through central Queensland. Moving on to the third component, which is vegetation. The vegetation classification focuses on consistent variation in the ecologically dominant layer, or maybe a more simplistic way of putting that is that it focuses on the dominant canopy species. Looking at this example over here on the right, that would be referred to as Brigolobala, and the actual grasses or shrubs in the understory which vary depending on annual conditions or management practices and typically not used to describe the vegetation. It stays with that dominant upper strata. So now bringing these three components together in this example of 11-4-3. The buyer region is 11, which is Brigolobalt. The land zone happens to be land zone 4, which is gently undulating clay plains. The vegetation as we said is Brigolobala. So 11-4-3, bring that together. 11-4-3 is Brigolobala shrubby open forest on clay plains. Simple. So scale is critical to both the classification and the mapping. You see the diagram on the right here, the sandy colour. That indicates the area that's mapped at a 1-100,000 scale. And as you can see, it covers most of Queensland. The yellow areas are mapped at 1 is to 50,000 scale. And then you can see a few local government areas down here in the southeast corner are mapped as 1 is to 25,000 scale. So most of the state is mapped at 1 is to 100,000 scale. Just to give you an idea, 1 millimetre on the map at 100,000 scale equals 100 metres on the ground. And that imposes certain map limits. We have minimum remnant patch sizes of five hectares at 100,000 scale and minimum width for linear features of 75 metres. Now that's reasonably coarse and it's fine for Western Queensland. But as you might imagine, when we get into the more populous eastern parts of the state where we're trying to manage delicate or specific conservation aspects, koala habitat, for example, we're actually looking for better scale of mapping. And that's the reason why some of those southeast Queensland LGAs are mapped at 1 is to 25,000. Because at 1 is to 25,000, 1 millimetre on the map equals 25 metres on the ground. And its map limits are much reduced in that a minimum patch size can be a quarter of a hectare. And we actually map minimum widths of linear features down to 20 metres. So if you take a large mature eucalypt tree, you know, often that has 10 to 15 metre crown widths. So we're really getting down to almost one to two crown widths that we're mapping. The RE mapping itself largely centres around the pre-clearing extent. So the pre-clearing extent of regional ecosystems is based on old aerial photographs and other supporting resources. The pre-clearing vegetation or regional ecosystem is defined as the vegetation that is present before it was cleared or what would have been there without clearing. So here we have an example of an old aerial photograph that the pre-clearing boundaries have been manually marked up on. And that gets scanned and digitally transferred into a GIS that's overlaying over ortho rectified high resolution satellite imagery. And you can see the common features here from the manually marked up photo now in the completed pre-clearing regional ecosystems map. The delineation of the veg patterns on historical aerial photographs is done using stereoscopic pairs of photographs, which gives us a 3D view of the landscape. There's me with my trusty stereoscope, this stereoscope. It's very old technology, but it's just tried and proven. It's extremely reliable and works very well. Then of course, the vegetation patterns that we delineate on those air photos are used in conjunction with field survey, geology and soils information, as well as ecological and historical knowledge to define and map the pre-clearing regional ecosystems. And really in these last two slides, I've just shown you there's an entire mapping methodology based around this. I'm really just skimming the surface for the purposes of this talk. There is a methodology document, which I have links for at the end of the presentation if you want to go into that in further detail. The other thing that we have in our bag of tricks, and this is used specifically for areas that may have been cleared for 100 years or more, you know, prior to our earliest historical aerial photography. And that is the old survey plans from when Queensland was first broken up by surveyors. This particular example I have here is from 1889. It's situated down the Gold Coast along the Narang Creek. And you can see what the surveyor is marked here, dense scrub. And then gum and apple, black soil forest. And then they very nicely delineated the boundary between the two for me. So that boundary is, you know, you wouldn't treat that as gospel, but it's still a very useful guide when reconstructing the landscape for regional ecosystems. Obviously this is a very time consuming process to look at every parcel of land. So we haven't done it for all of Queensland. It's mainly reserved for those areas where we have no other sources of information available because it's been cleared for so long. Okay, moving on to remnant area mapping. So this is a very mouthy definition. Remnant woody vegetation is defined as vegetation that has not been cleared or vegetation that has been cleared. But where the dominant canopy has regrown so that 70% of the height and at least 50% of the cover relative to the undisturbed height and cover of that stratum and is dominated by species characteristic of the vegetation's undisturbed canopy. So as I said, that's a bit mouthy. So let's let's work through that with a diagram. So first of all, on the left here, we have an intact forest that's never been cleared. So regardless of its canopy height and cover, because it's never been cleared, it is remnant. The example on the right here, it's the same vegetation type. You can see the trees are shorter and less there's more there's more spacing between them. So it's been cleared in the past, but it has regrown to the point where there is greater than 50% of the original ecosystems canopy cover. And more than 70% of the original ecosystems canopy height. So that vegetation would also be classified as remnant. Now looking at this example down here on the bottom left where this vegetation has been cleared. It's regrowing. It has the canopy cover, but it has not yet attained 70% of the original canopy height. So that would be non remnant. And this example here, it has not attained 50% of the original canopy cover. So that is also non remnant. Okay, so once we've got our pre clearing, the remnant boundaries are mapped and effectively cookie cut into the pre clearing to produce a remnant RE map. So here's our completed pre clearing map with its REs allocated to each polygon. Then we construct this remnant non remnant cover layer. The remnant bits here being in green and the yellow is the areas that are non remnant. That's then intersected or cookie cut with the pre clearing to then give us our remnant regional ecosystems mapping. So all these yellow bits here on our cover layer translate into non remnant regional ecosystems over here. There's no RE allocated to those areas. And of course, by having the pre clearing and remnant RE extents to compare with, we can then generate all sorts of tree clearing statistics. And this is one of the real strengths of the RE framework. Many jurisdictions both nationally and internationally have some really great remnant vegetation mapping, but none have reliable pre clearing mapping to do this sort of detailed analysis with. So we can do comprehensive breakdowns of all regional ecosystems and they're available currently on the DES website by sub region, catchment, natural resource management area, local government area, electoral district and that just to name a few. Of course, you can do whatever breakdown you like once you've got those base input data sets. And then the other strength of this with having that sort of information and tree clearing statistics, we can then allocate what's called vegetation management act class and biodiversity status, which is used under the various state government planning here in Queensland. So RE is individually assigned a class for vegetation management act purposes based on how much of their mapping of the map pre clearing extent has been cleared. For example, if there's less than 10% of pre clearing area remaining, that would be considered an endangered regional ecosystem. If there's 10 to 30% of a pre clearing area remaining, it would be of concern and greater than 30% remaining least concern. So looking at these two maps here on the right, we have a pre clearing RE map based around the Brisbane area and a corresponding remnant RE map based around Brisbane. You can see the pink areas of the endangered, oranges of concern and green areas of least concern. And logically, you can see all, well, most of the pink areas that were mapped on the pre clear have been cleared on the remnant map. And hence the reason why there's less than 10% of those RE's remaining moving on to biodiversity status. It uses these same tree clearing based concepts, but it introduces information about degradation or threatening processes that may be going on within each of these regional ecosystems. So biodiversity status for use under the Queensland's Environmental Protection Act, it can never be less than what the Veg Management Act class is. But if there is a threatening process going on, it can in fact be elevated in status. And we do our remnant extents every two years, and we have done so since 1997. So we can generate all sorts of clearing trend over time. This is the average annual clearing rate of remnant regional ecosystems from 97 to 2019. You can see here back in the bad old days before the introduction of the Vegetation Management Act in Queensland, there was some pretty high tree clearing rates. It was 2000 when the VMA, the Veg Management Act was actually enacted. And you can see the resultant drop in tree clearing over time, right down to a low from in 09 through to about 2013. About 2013, things kicked up a little bit, and that's often referred to as the Campbell Newman bump. And then in recent times, it's declined again, which is a good sign. This is another way of looking at clearing over time. The darker colours here represent greater tree clearing rates. So this is the rate of clearing since 1997 through to 2019. And I mean, you can really observe how overall there's been significant clearing in the south-eastern quarter of the state, including Brigolo Belt, South East Queensland, New England, Tableland and eastern parts of the Mughal and Bayer regions. Here's another way of looking at it. Percentage of remnant regional ecosystems remaining overall in Queensland by subregions. New England, Tableland has the lowest extent of remnant vegetation out of Queensland's 13 Bayer regions at 36%. The Brigolo Belt Bayer region has the second lowest remnant extent, 41%. And the Tarraghounds and Tarum subregions, which are these two small subregions, have the lowest subregional remnant extent in Queensland with only 5.9 and 6.95 respectively. So very highly cleared landscapes. So I've spoken a lot about remnant regional ecosystems, but what happens to that vegetation which doesn't quite make it to remnant status? That's where high value regrowth mapping comes in. So here's another very mouthy definition under the Veg Management Act of what high value regrowth is, but it can basically be summarised as non-remnant native vegetation that hasn't been cleared for greater than 15 years. And in actual fact, that can be quite reliably mapped from remote imagery, often without the need for ground surveys, as unlike the remnant criteria, there's no height criteria. So the only problem we face with applying this definition is the identification of exotic vegetation. Most of the time we can see exotics with high resolution imagery, but not always when high resolution imagery is not available. So here's an example. You can see all these very well vegetated areas. They're all remnant and that's captured under our remnant RE mapping. But then we've got all this additional high value regrowth throughout the landscape captured now as well. And this is the sort of thing we're capturing. You can see we've got trees which are greater than 15 years old, but you can see this doesn't have the intact nature of remnant vegetation yet. However, that still has a lot of habitat value and it is returning to a broadly functioning ecosystem. And that's what high value regrowth is really trying to capture. There is one minimum requirement for high value regrowth and that is a minimum crown cover. So if an ecosystem was originally a sparse regional ecosystem, and we know that from our pre clearing regional ecosystems mapping, then you would need at least a minimum of 10% cover to qualify as high value regrowth. So just for example, looking at these photos to our right here, this top example you can see there's reasonable canopy cover coming in, still very young juvenile trees. But if they were over 15 years old, that would qualify as high value regrowth. This example below it where we have well spaced scattered trees. Well, that probably wouldn't. So what's next for the REs and HVR program? Well, we're currently trying to improve both remnant and HVR extent by reconciling with a new product produced by the Queensland Remote Sensing Centre, which is a new woody vegetation layer. That relies on the 30 year Landsat satellite history and reliably picks up woody vegetation over all of Queensland. We will be reconciling our remnant and HVR layers with that additional information. We're always looking to improve the RE map scale, as I mentioned, particularly in coastal areas and in what we might describe hotspots such as the South East Queensland koala plan area. We're using high resolution imagery and other information to reduce the number of REs per polygon. You would have seen in some of my example slides through this presentation and RE can have—sorry, a polygon can have more than one RE in it. That's okay in some instances, but it's not useful in others. So we are endeavouring to divide those up to have more homogenous mapping of REs. And then we also have a focus in the Great Barrier Reef on wetlands at the moment for the purposes of the Reef Water Quality Improvement Plan. So that's basically where we're heading just at present. If any of you are interested in any of what I've spoken about in this presentation, here's a whole bunch of useful links here, which you can grab from the presentation. And yeah, thanks very much for listening. Thanks very much, Tim. Really appreciate the presentation in the background with regards to the technical detail that goes into mapping vegetation in Queensland as an important underlay for the development of the koala habitat mapping. We've got a number of questions that have come through on the Q&A app, and a lot of those are focused actually on the koala mapping itself rather than on the vegetation mapping. So I'm proposing not that we don't address those questions but potentially hold those questions over a bit to be answered by Stephen because a number of those relate to why certain areas aren't reflected in the koala mapping and why some areas aren't protected. What I might in fact do is I'll address a couple of the questions myself. One of the questions relates to the mapping has been changed to accommodate development over years. For example, 800 hectares of Oki Flat Road in Narengba was classified as KMA, never to be developed in the koala management plan in 2006 and three years later there was not a tree left due to development. So what is the value of the mapping? So I think my response to that is very much that the Queensland Government based on the SCQ koala population study identified that there were continuing declines in koala populations despite the measures that were in place, and that was the basis for establishing the koala expert panel. One of the key recommendations of the koala expert panel was for the need for more comprehensive and consistent mapping of koala habitat across south-east Queensland. And that was the impetus for the current mapping, which has now was introduced in February in 2020. And that mapping, which is based on the regional ecosystem mapping that Tim and his team produce, now identifies and maps the best quality koala habitat in south-east Queensland in a consistent manner. And at the same time that that mapping was introduced, there was also introduced new planning controls, which now provide protections for koala habitat identified in koala priority areas, where there's koala habitat and locally refined koala habitat. So yes, there was recognition by the Queensland Government that under previous arrangements that koala habitat continued to be lost and koala habitats declined. That's been addressed through the expert panel recommendations and the subsequent development of the new state of the art koala mapping, which Stephen will discuss in the next presentation. There's a couple of questions with regards to areas of koala habitat that are excluded or haven't been covered in local government areas, despite the fact that koalas have been identified in those sites. What I can, I guess, address in response to those particular questions is that, again, local government across south-east Queensland have been very vocal in representing to the Queensland Government that there are areas within their jurisdictions that they were previously protected that may not be protected now and other areas that they believe warrant protection because of the presence of koalas or the presence of habitat they believe is important to support the persistence of koalas in their local government areas. In recognition of that concern, the Department of Environment and Science is currently working with local governments on a project to develop some technical guidance around other mechanisms to protect koala habitat at a local scale. So, for those that are aware of the planning framework, currently the koala habitat mapping represents a matter of state environmental significance, but planning regulations don't allow us to represent the same or essentially the similar matter as a matter of local environmental significance. So there's currently work between the Department and local governments to look at other mechanisms to identify and protect locally significant koala habitat in those local government areas. So in the most recent update of koala habitat mapping, a number of locally refined koala habitat areas were recognised and converted to core koala habitat in the new mapping. Again, this was through a process of engagement with local governments collection and analysis of data collected by local governments to support that. So I'll also provide those questions to Stephen so that he can address some of the issues in those. I've got a question here, which I think is related to Tim's area. So there's a question here that says, the system of conservation classification is highly flawed. Is it effectively allows particular ecosystem to be cleared to a point where in some instances, especially in highly urbanised areas, it becomes unviable and thereby only sort of 10% remaining before it is afforded protection under the legislation. Ironically, we are attempting to provide greater protection to high value non-reminant as opposed to in situ remnant. So I'm not sure if Tim has a thought in relation to that. Well, what I think that might be getting to is the urban exemption of regional ecosystems. It's not true to say that something needs to be endangered before protection mechanisms kick in. The Vegetation Management Act actually regulates broad-scale clearing of all vegetation management class ecosystems. So you need permits to clear. And there are only very limited circumstances in which a permit will be given. However, when it comes to urban areas, there is an exemption that only endangered regional ecosystems are protected under the Vegetation Management Act. The point I'd like to raise there is that the Vegetation Management Act is only one mechanism to protect that vegetation. And as we now have the koala plan and the mapping that Stephen will talk about, that protects habitat regardless of status. And all it needs to be is a koala habitat RE and either be high-value regrowth or remnant vegetation. The actually endangered of concern or least concerned status is not pertinent or relevant to that argument. It's more specifically to Tim's habitat mapping and the question is you discuss regrowth habitat that had reached a couple of thresholds. 50% canopy cover, 70% canopy height. Would you be able to... where the vegetation would be classified as remnant? Could you talk through the process by which that occurs and whether that new remnant vegetation is included in your clearing rates data? Yeah, okay. So the latter part of that is new remnant in our clearing data. Yes, it is, but it is not as thorough currently as what removals or losses of vegetation is. So in other words, over the last 20 years, there's been a lot of effort put into the mapping and capture of the loss of remnant vegetation. Not a commensurate amount has been put in to the re-establishing vegetation as it re-grows in the landscape, but there is a whole new program called the Enhanced Slats Program, which has been designed and built around tackling that particular aspect. And we're already looking at... Well, we already do add some remnant or newly established remnant back in, but we will be doing it more thoroughly in a statewide manner when we have the new Remotely Sense products coming out of that Enhanced Slats Program, which actually sort of maps and monitors the re-establishment of vegetation over time in Queensland. So there will be considerably... There will be a balanced effort then that all vegetation change, both the losses and the gains will be accurately reflected in the regional ecosystems map. Now, how we actually do that, as I said, we will use that Remotely Sense product. Determining whether something meets that 50-70 cut-off, it can be problematic because to be absolutely certain about the height and criteria, that's not something we can get yet from Remotely Sense data across the state, so you do need on-ground information, and that's usually provided by a PMAV sort of process. If an area is inadequately mapped, that information can be provided by a property map of accessible vegetation, which can then address that newly established remnant on any given property. Yeah, welcome back everyone. The second presentation for today's session of the Koala CoLab 2021 will be provided by Stephen Howell. Stephen is the manager of the biodiversity assessment team within the Queensland Herbarium. Teams responsible for the objective spatial assessment of terrestrial and aquatic conservation values across Queensland, using the best available science and methodologies to inform decision-making for a range of purposes and at a range of scales. The teams focused on habitat modelling for a number of threatened species, and in particular koalas, for which they have mapped koala habitat areas across southeast Queensland using modelling based on the latest regional ecosystem, high-value regrowth mapping, as discussed by Tim Ryan in the previous presentation, koala sightings and biophysical measures. Stephen's a landscape ecologist with a focus on spatial ecology and an interest in rainforest ecology. So welcome, Stephen, to present this morning on koala habitat model and results. Thank you. Great, thanks very much, Geoff. So as Geoff mentioned, Stephen Howell, look after the biodiversity assessment team. We're part of the Queensland Herbarium within the Department of Environment and Science. So I'm going to talk about the koala habitat areas, KHAs, give you a little bit of history, then go through the methodology of how these areas were developed, and then have a look at the results. I'll also talk about locally refined koala habitat areas, LRKHAs, and then, depending on where you're going with time, I'll touch on the koala priority areas, KPAs, and the koala habitat restoration areas, the KHRAs. So sorry about all the abbreviations. So a little bit of history about habitat mapping. The development of the new mapping was recommended by the koala panel in 2017, and the recommendation was to develop consistent mapping for koala habitat across SQQ at a fine resolution that addressed some of the issues of the previous mapping and also implement the systematic mechanism for updating this mapping for accuracy and for tracking of changes over time. So the recommendation was supported and it was implemented by the government. And that's what I'm going to be talking about today is the habitat mapping. So just a little bit about habitat mapping, just to give you a bit of background to how we've done this. For a lot of the work that we do and the habitat mapping slash modelling that we have, we take the buffered species points, and we assign that as being habitat. So we take the species records, and we get those from corporate databases and a range of other databases, and I'll talk specifically about the koalas, data sources and the tech, but we get the records from various databases and we apply a number of validation of filtering rules. So we get rid of old records, imprecise records, cultivated records, sightings of birds that just flown over when it's not really habitat, it's just a bird flying around. So we get rid of a lot of those types of records. Then we put a buffer around it, and we say any vegetation within that buffer is habitat for that species. And that's not too bad. In the absence of any other information, at least we're identifying and hopefully protecting the habitat surrounding those particular records. But our preference definitely is to create these habitat suitability models. And so that's where you have core habitat that's based on vegetation, but also other biophysical data. These are more ecologically accurate, but they are more time consuming. They do take longer to create, longer to run, especially when you're creating these models that it's getting into something statutory that's actually going to say to someone you can't clear there because this model says that. So we need to make sure that they're robust enough to be able to feed into those types of processes and be suitable for that. So what we'd like to do, so you just see this little point there. So say that there was a record found in this particular location, so once again not quite as, but there's a record there. So under this, the buffered species points we put a buffer around that and then we would select all the vegetation that falls within that buffer. So this particular species actually likes the riparian vegetation associated with this stream going through here. And so by doing this method it's picked up the habitat close to that species record, but it's also picked up habitat that's quite a fair distance away. And when we're looking to identify core habitat to protect that core habitat then that gets a little bit harder to justify. So our preference really is to identify that as being the core habitat for that species. And that's what those habitat models give us is a way of identifying those. Then we're able to remove that point because we're just trying to avoid duplication doubling up. But one of the other real strengths of this is we're also able to identify that that's core habitat. So that additional area up here, that additional stream and perhaps even these other streams over the other area are also core habitat even though there were no records found in that particular area it has all the right attributes. So the vegetation, the biophysical attributes, etc. that we will consider it to be core habitat for that particular species. So this habitat suitability model that's what I'm going to go through now. So for the koalas so it's built on the existing we did already have a habitat model for the southern part of SEQ. Based on contemporary methodologies, the best available science and data based on a regionally consistent high-value regrow. So it's based on the excellent work from Tim and his team that he went through previously. It's tenure independent. Optimizes the use of existing data and information transparent and repeatable. And that's really important. So we really need to be able to make sure that people understand how the methodology and how the information came together. And it's also a decision support tool. So when I'm starting off talking about this habitat model it is just about a habitat model then it turns into something that is applied through the VMA or through matters of state environmental significance. So it is a decision support tool that feeds into those planning, that planning and regulatory environment. So just having the koalas so these are koala records across Queensland outside the SEQ by region. These are those buffered filtered records and they just have a circle placed around them just to identify habitat and show where they are. Northern part of the SEQ by region we have an existing model but I'm really going to focus on the southern part of the by region which is where we've updated and run this new model. So that's the area that we're looking at. So I'm just highlighted by the local government areas just to give a bit of context. This is the SEQ shaping SEQ the SEQ 20 planning area that we've done this habitat model and honestly we don't actually like doing habitat models based on planning boundaries. We would really prefer to do a habitat model based on ecological boundaries at least the by regional boundary and preferably the full extent of the koala of a particular species and obviously this case koalas. But unfortunately that wasn't possible at this time and certainly to do it for the northern part of the by region and all of SEQ is on our agenda but there's a significant amount of work involved. The methodology that we've developed and I'll go through in a sec we're hoping actually can be applied to other areas across Queensland with just changes to the input parameters and things that become important as you move out west or as you go north. So this is the habitat model has the components maxing the buffer records and the ecosystems get classified and create the remnant and pre-clearing and then turns into the koala areas. So just teasing that apart a little bit and quickly going through this. So we rank the region leaker systems based on the presence and relative dominance of trees that are important for koalas so it's based on the expert knowledge published literature etc etc and we actually do this for pre-clear and current and remnant vegetation so that gives us a way of identifying how much there used to be and how much there is now. So just an example, so this is from the technical reports available on the web where regional ecosystem 1233 is used for tree corners woodland and there's a whole stack of information Tim would have alluded to that goes associated with that description. That's been rated as high whereas something like complex notable for us will bring on rain forest that's been ranked as being very low in terms of its importance for koalas. So we have that ranking for all the region leaker systems across southeast Queensland from high to medium low very low and non-habitat. Then we bring in the buffered koala records and that comes from existing corporate databases that there's systematic survey data koala based local governments ALA Atlas of the living Australia etc. We apply those filters and in this case we lost about lost is a bad way but about 20% of the records didn't pass the filters in this case we had the date is there and the precision got rid of duplicates and a range of other filters were ecological filters were applied. So we ended up with 88,000 odd records. That's 98,000 koalas in southeast Queensland. How many records we have passed all those filtering rules during that time period. Just gives you a bit of an idea of the distribution of where the koala records are and the data sources. Once again this is from the technical report. Okay so then we have within the matrix where there is or is not a koala record. Final component of the matrix is MaxSend. So this is a commonly used statistical tool looking at species distribution modelling and I guess the easiest way to explain MaxSend is that we gave the program a number of variables like soil, nitrogen, phosphorus, slope, elevation rainfall etc and we gave those koala records you saw before and we told MaxSend to look for associations between those two. So what best explains where the koala records were found and so what this table is once again in the report also says that elevation was the thing that most explained where the koala records were across the landscape. So elevation wasn't weighted but it just says that elevation is most likely going to be a proxy for some of the other drivers of the distribution so it was considered a useful variable. So we've got that MaxSend and we've got that from low, medium and high. We have the koala records where there's one there or not and then we have the regional ecosystems as part of we then classified that into a range of categories and we can then map that because we've attributed that to the vegetation so we've got the remnant and we've got the pre-clearing. So having a look in the top left hand corner here in the 10s, so these are areas where it has the right environmental variables, it has records and it also has the right regional ecosystems. Then down in the bottom right hand corner where there's ones so it doesn't have the right sort of environment available through MaxSend. There are no koala records and it's non-habitat. So Sands or main groves or something like that so it's non-habitat for koalas. Okay so we've got that matrix that's our starting point that fit into the SQQ koala conservation strategy. We've reclassified that to identify core non-core and non-habitat and there's definitions associated with those so I won't go through those but that just gives you the breakdown of that core, non-core and non-habitat and the core component is what's fed into the SQQ strategy. So it's that final step with that map at the end so koala habitat areas version 2. Okay so this is what the map looks like and results and we can actually break it down by looking at remnant and regrowth so regrowth is a little bit hard to see but it's just that lighter coloured green and it just shows you that's the full amount that we've identified currently for koala habitat areas. It's worthwhile just having a look at that's how much there used to be of that core koala tap which is a bit of a scary amount having a look at seeing how much there is now first how much there is there used to be. The other useful thing to look at is that we can then within that sort of matrix there we can break it down a little bit more and we can have a look at what are the best of the best areas so even within those areas of core those ones up in the top right left hand corner have the best environmental variables. They have a record and they have Euclidus tree corns all the best re's considered important for koalas. So just having with the numbers so easiest one to have a look at someone on the right hand side so how much there used to be and then the next one shows how much there is how much has been cleared how much is left of regrowth and then how much is left of remnant. So it's been peer reviewed the methodologies so by sorry Land and Water and the review of supportive of the modelling approach and they said that the method improves on current methods for identifying koala habitat by incorporating different knowledge sources into useful tool it made repeatable mapping and that repeatable mapping is quite important because we repeat this at the moment every year and it gets released and that's what this was so version 2 was released on the 8th of September along with a range of other updates to other products based on RE version 12 we did a quick review of the re's that changed between versions and new re's that might have come in high value regrowth did a review of high value regrowth just to make sure it was better attributed we included the latest koala sightings we had up until the time we didn't update Maxent so those environmental variables like soil and elevation that sort of stuff won't change that much so we're looking to actually run Maxent every five years and that will then get updated and included in the model but the model at the moment and the release into these products is running on a yearly schedule so quickly the local refined koala habitat areas so the transition responsibility from koala conservation to state government from local government sorry to state government these would provide continuous regulatory protection for these areas for two years for two year period local governments were asked to provide these areas for part of the kind of conservation efforts and where they were outside what we'd already identified as being koala habitat areas they were included and they were clipped to any remnant and high value regrowth so it doesn't matter whether it was koala or remnant or high value regrowth it was any remnant or high value regrowth so just looking at the combination of the two the koala habitat areas and the locally refined koala habitat areas so it sees an apparent reduction in the area although there was an increase in the amount that was in KH8 and that reduction was really about continuing refinement to the habitat model and the source mapping for the regional ecosystems and high value regrowth but over time it's expected that development restrictions will result in reduced habitat clearing as well as allowing areas to regenerate so they regenerate, they become high value regrowth and if they meet those other components of the model then they get included in the model so results are included in SEQ conservation strategy they're included in the vegetation management active central habitat and they're included in matters of state and environmental significance and these also get into a range of other planning systems and I won't go through those just there results are available from the website from Queensland Globe from the VMA report at the separate section of the VMA report to get that but you can also download the GIS results from Qspatial you get all the GIS results and you get all the information and the attributes associated with it how to colour it metadata etc associated with that so that covers off the Coal Habitat areas and the locally refined Coal Habitat areas I've only got a few minutes left so I'll just quickly touch on the priority areas and the restoration areas so priority areas were about identifying management areas containing core habitat but minimal threats so they had to have the best habitat that we could identify but the least amount of threats and so we had the Habitat model and that's the same one you saw before just broken down by those individual categories we then had some threats constraints opportunities and resilience mapping so range of criteria here that we mapped and we identified ecological cost so we had Habitat ecological cost ran it through MarkSAN so it's not MaxSAN but MarkSAN software and that tries to identify the most amount of habitat for the least amount of threats and the most amount of opportunities tries to find that balance between the two and that's what you're looking at the right hand side from that one on the right hand side it turned into that map once we consolidated some of the boundaries aggregated the parcels, snapped the cadastra etc and that's where the Coal Priority areas were identified so the other component of that is the Coal Habitat Restoration areas, KHRAs so this is about cleared highly suitable habitat so this is where it used to be really important Coal Habitat but it no longer is important Coal Habitat because it's been cleared and if you were looking to maybe do some restoration activities this might be a good place to start so this isn't a statutory layer this doesn't say you have to do this we have to do that but it does give you a bit of a guide as to if you had to choose between two different areas this would probably be a good one to start with because it used to be really important Coal Habitat so it's only an indicative layer won't go through this but it's a similar process we had how much there used to be the ecological cost layer based on threats etc through Marksand and then identified those areas that were perhaps suitable for Coal Habitat Restoration areas so just finally just like to say thanks so the Coal Expert panel thanks for integrating and guiding us through this at the time when we were developing the model we formed a separate Coal Advisory Group that just helped us with the development of the model but especially thanks to the biodiversity assessment team so Harriet, Courtney, myself and Lindsay so current members of the team and who have been involved with the model from the start but also there's a range of staff that have been involved with the team and specifically with development of the Coal Model over time and that's everything I think on nature some of a broader nature the ones I might just touch on that arose during Tim's session was a question with regards to why it would be that certain coastal habitats and the example was some of the lowland areas in Redlands local government area why might they be not included in the Coal Habitat mapping even though there's knowledge that Coal has used those areas and there's records to support that So there's definitely a lot of records and I say a lot of knowledge and information about areas that Coal has used outside our Habitat mapping but we needed to feed our mapping into statutory products like the Vegetation Management Act and the state and environmental significance and we were bound by the base imports into that so the regional ecosystems and the high value regrowth that fed into those is what we had to attribute and include and identify as Coal Habitat So there's two possible reasons why areas might not get picked up it could be that there's no regional ecosystem or high value regrowth mapping even though there's Coalers and there are scattered trees and Coalers do move from one area to the other independent of where the vegetation is or it could be that the vegetation isn't the right vegetation to be included as that high, medium or low that we've identified as being important for Coalers There's a couple of technical questions around the use of Maxent so one is even though Maxent is the most widely used model for species distribution modelling is it worth using multiple models and averaging based on the our key coefficient or is it really much of a muchness so I guess the question is Yes so it's a good question and it's an interesting question we did, you write there are other models around and even within Maxent there are a number of variables and things that you can apply to Maxent so we did explore a number of those models before we settled on using Maxent but we just felt that that gave us the best output based on the information and the scale and the type of work that we were doing so certainly when we rerun Maxent in that five year period which is I think only about two years away we will rerun it and we will look again at the input parameters and some of the variables and also stuff that we chose when we ran it so Yeah Okay another Maxent related question was did you do any sensitivity analysis comparisons with other species distribution modelling algorithms also the regional ecosystem classification the comment from the question is it's pretty subjective for instance I would have thought the presence of the robusta would make 12.3.4 pretty popular for koalas so surprised you've only listed it as low ranking is there a full list of these classifications published somewhere so there's there's two parts to that did you do sensitivity analysis and then the question around the subjective nature of the RE and how those classifications were derived and whether there's a list of those so in terms of the second part the if you have a look at the technical document so that's available on that koala website that I talked about but it's also available through the Q spatial download package but in that document there's an appendix there's a section there explaining how we identified those regional ecosystems and there's an appendix listing at the bottom listing all those 200 something odd RE's that are found so it's the root RE's and all those individual sub areas and the rankings that we applied to them so all the information's listed in the technical report so and as I said that's available just from that koala website you don't have to download the GIS package we did do a bit of a comparison with other models and we did look at so there's an online system where you can load up your data and it uses max in type programming to do it and we did have a look at some of the results of that we did do a sensitivity analysis and we did actually have some records just as insights where we had some records that weren't included in the model and we were able to the maxing component and we were able to use that to help sort of back cross validate the outputs from the maxing results and we were quite confident that those records weren't used in the model that when we ran the model the model actually identified those areas covered those records as being important habitat as being high in terms of maxing we did use the high and the medium categories of maxing so it meant that if there was any sort of uncertainty in what the results were coming from maxing I guess we were a little bit more more precautionary by using those top two categories okay again there's a lot of interest in some other questions there and as I said in the introduction we'll seek to provide responses to those other questions offline really grateful for Stephen taking the time to give that presentation today good morning everyone and welcome back to the third session for today's Koala CoLab conference series our third presentation today will be done by Harriet Priess Harriet is a conservation analyst with the biodiversity assessment team within the Queensland department of conservation science she's got extensive experience in wildlife ecology, threatened species monitoring spatial modelling and links between science and policy Harriet's a spatial ecologist currently working on the science that underpins koala conservation and played a very instrumental role in developing the new approach to habitat mapping that's been utilised for the Seq koala conservation strategy Harriet's been a ranger with national parks and monitoring koala populations to support the first state planning policy back in 1995 she was a co-author of the University of Queensland report on the status of koala populations in Seq which used more than 20 years of monitoring data to quantify declines in the koala populations of up to 80% this morning Harriet will be presenting in relation to the koala threat hotspot mapping so thanks very much Harriet Hi everyone, thanks Geoff Geoff and I go way back known him for a long time so today we'd like to talk about the threat mapping component that follows on from the work that Stephen has just presented in terms of the koala habitat modelling and this presentation there's a lot of based on a lot of the work that the only Seabrook has done for us so we're going to start off with a bit of background about the koala population decline and how that's led to the koala conservation strategy and then moving on to how much progress we've made to date in reviewing the key threats and what our next steps might be in terms of modelling the threats including koala occurrence how we're going to combine the threat layers how we're going to link that to other programs and then finally any questions soon as we start off with a brief background and that's the recognition that koala populations are in decline across south-east Queensland and in fact across all of New South Wales and Queensland and our report that I worked with Jonathan Rhodes at the University of Queensland and others showed that we have this 80% decline in the population between 1996 and 2014 in the koala coast and then a 50% decline in the Pine Rivers local government area as it was back then so that report precipitated a number of state government initiatives including the koala expert panel which morphed I suppose into the koala advisory council and that led to the habitat suitability modelling that Stephen has just talked about and of course the new SEQ conservation strategy so in that strategy that has just been released last year there's a number of action areas six action areas all together first one being habitat protection second one is habitat restoration the mapping monitoring research and reporting community engagement and then partnerships and strategic coordination the threat management component of course links to all of those action areas so the action area four is where we're undertaking this threat mapping and the idea here is to identify the koala threats and developing and enable us to develop a mapping methodology and then this would enable us to track changes in the threats over time and hopefully to enable other groups to institute some mitigation strategies action area three which is the threat management component you can see from the highlights here that a large number of components of threats have been discussed here so getting into why we conserve koalas in South East Queensland good question I hear you say so South East Queensland of course is a very large koala population it's home to in excess of 20% of Queensland's koalas has a large amount of suitable habitat it's got a suitable climate and of course above all of these things we've definitely got a moral responsibility to conserve them but there's a large number of challenges in South East Queensland in particular the development pressure the development associated threats that come with those developments and urbanisation and of course the ongoing threat increasing threat of climate change so we've got to so far with our threat mapping to undertake a literature review and we've got an internal report where we've investigated these threats we looked at the criteria and the different approaches we've been able to review the existing threat criteria which we've used previously in the koala habitat mapping to identify the koala priority areas and the koala habitat restoration areas we've looked at some additional criteria that we might be able to incorporate into this mapping and we've investigating we're starting to investigate the options for combining the threats into a combined risk layer and identify the koala threat priority areas so this slide gives you our framework for threats, constraints, opportunities and resilience so we've got a number of criteria here along the top line and then we've got a number of indicators a second row here and there's various numbers of measures that we've using to track the threats so in our framework we've identified threats as being those things that directly impact the koalas and can cause harm to them constraints are those things that limit an area's ability to supply habitat and constrain koala populations themselves on the other side of the ledger though are opportunities and resilience so opportunities benefit koalas and we can leverage off that to increase koala habitat or viability and then resilience are those things that enable koala populations to bounce back or provide sanctuary in times of crisis or climate crisis or fire that type of thing so if we just look at some of the threats and constraints at the moment so these are the negative things that are impacting on koalas here's an example of one of the data layers this is heat stress and it's made up of actually four data sets that represent the number of consecutive hot days hot nights and consecutive hot nights and we've combined these together to come up with a measure that represents heat stress and we know that koalas suffer greatly if temperatures are too hot particularly during the night which is when koalas tend to feed and if this heat goes on for several days several nights this can lead to starvation loss of body condition and ultimately death of the koalas so the measures enabled us to look at the data that represents hot nights going back back to 1950 to get a baseline data came from silo and when we did a statistical analysis it said it's a large proportion of the study areas already experiencing statistically significant increases in temperature which is not good so another variable that we've used is dog this is a domestic dog density based on the number of dog registrations with the local governments throughout South East Queensland so again the red areas threat hot spots for in terms of dog density another constraint here is the urban constraints urban development it's made up of two criteria so the dwelling densities and the urban zoning so this is the SEQ regional plan zoning gives us a combined threat of urban development so this gives you some idea of what it might look like when we start to look at combining these layers so this is the heat stress element that we showed you just a second ago and you can see these western areas where heat stress is dominating urban development the the threat is concentrated around Brisbane Gold Coast Sunshine Coast urban development areas then when we look at the conservation layer which is an opportunity that we have follows along the daigula range for example but unfortunately when we look at climate change when we're seeing a contraction in the amount of habitat and it's actually affecting places like the daigulas and of course main range so that limit you can see how this is going to start to limit our ability to conserve koalas in a number of areas so this map represents extractive industries and a protected area estate in these examples the red is used to designate areas which have got the high threats and blues areas where we've got low threats so what we wanted to do is have a look at the relations between the threats so we came up with this causal diagram which is a different way of looking threats because we're actually looking at cause and effect and it's a way we're using to actually make sense of the relationship between different elements so here we have the endpoint which is basically the koalas their populations and the resilience of those populations and that's greatly influenced of course by the amount of habitat so that's our endpoint and the effects things that affect those greatly are things like habitat loss habitat degradation fragmentation those elements and death or injury of koalas impacts the koalas themselves so what we're starting to see here is a web of interactions and the modes in which they apply so for example changes to habitat influence that could directly cause the death or injury of koalas or things like fire could do that fire could also degrade the habitat influence the habitat itself and lead to declines in the koalas so the usual suspects you know the cars, dogs and disease so we've built those into into our model so disease for example can result in reduced fertility and fecundity which means koalas having fewer offspring and that directly impacts the size and persistence of these koala populations so the main impact cause of course comes from land management aspects so urban development, linear infrastructure resource extraction tentative agriculture and other land uses and of course climate change so climate change can increase the amount of fire, increase the amount of the number of droughts and severity of droughts or cause heat stress and then that directly causes death to koalas and population declines so that's our causal diagram and it enabled us to look at the impact mode and then hopefully gives us the ability to intervene or mitigate some of these threats and do something about it ultimately so this helps us to monitor whether interventions ultimately are working so the next steps in our threat project is to look at incorporating additional threats so we'd like to get data on wild dog densities across South East Queensland areas of disease hotspots fragmentation of habitat and some of the drought impacts if we could get that data from the long paddock or silo we've also looking at the koala currents data fragmentation and mapping so we can do that things like frag stats and we want to develop a combined threat layer and move on to developing links with other programs so just have a quick look at some of our next steps which is this is some of the occurrence data so in this diagram here you can see the blue dots representing koala occurrence so koala reports the dark green is our core remnant habitat that Stephen spoke about in an earlier presentation and the bits that are not mapped are smaller fragmented habitats so it's non remnant habitat it's not high valley regrowth habitat as mapped by the Queensland herbarium and what this is showing us is if the clearing of these patches occurs what we can end up with severe bottlenecks limited ability for koalas to disperse and move across these landscapes and these populations of koalas could become locally extinct and as we know we've got a very large number of koala records that occur in the non remnant habitat and we're keen to identify these pinch points where the habitat is really vital to maintain in the landscape so that we don't further fragment and lose these populations so we could use some type of fragmentation analysis to do this also looking at the links between koala occurrence and the habitat so on the right we've got the map of the habitat which is ranked from lowest suitabilities in the green to the highest suitability in the red so you can see areas like the koala coast with large amounts of highly suitable habitat so this is the remnant and high value regrowth that's left then on this map we're showing the koala occurrence hotspots so areas like the koala coast or pine rivers out through Ipswich Esk the Lockia up to Nusa Malani to Gouloua down to the Gold Coast and other types of areas and so these are based on the density of koala sightings and what it means is that the koala occurrence doesn't always coincide with where we've mapped suitable habitat so that's that linkage that we were saying before between the koala habitat and the koala themselves what we want to move on to next is to develop a methodology so that we can actually create this combined threat layer we're looking at using weighted overlays which is very important to work at how to choose our weights and the importance and this is often done using expert opinion to reduce the bias or any differences in that or subjectivity in that we can use things like the local hierarchy or the Delphi method and the other thing we'd like to do is consider the threat interactions by developing something like a threat web which can also use expert opinion and discuss these interactions so this is an example of a threat interactions this is a bivariate example so where you have urban development for example influencing habitat to cause it to be lost and then influencing koalas or urban development directly impacting on koalas by increasing the number of car strikes so there's different types of threat interactions so this is a threat web and you can start to see that this can start to get quite complicated so this is what we're looking at at the moment so the other thing we're doing is risk analysis so the elements here that are in black are the ones that they're our group the biodiversity assessment team looking at undertaking so developing this risk matrix the ones in blue are the programs within department like undertaking so implementing the threat abatement or mitigation strategies themselves so what we're hoping to do is identify the likelihood and consequence of the different threats and how these might be interacting and do that as part of a structured decision making process so I think the final element is really looking at the links to other partner organisations so we feel that the threat mapping could help inform other abatement programs by local government contribute to foreign sensitive design guidelines identify locations for community restoration and revegetation programs assist with public education campaigns for koala conservation and help coordinate research and fill knowledge gaps that might be apparent some of the things that we like to keep in the back of our mind are these threat questions based on the fact that we know the main threats to koalas but we have less understanding of interactions and tipping points in illustration here on the right is based on some of the work done by Jonathan Rhodes and Clive McCalpin's groups and it shows that at a certain certain amount of habitat in the environment we often see this tipping point a threshold is reached where koala habitat koala numbers crash when a certain threshold is reached knowing what that threshold is is one of the key elements of landscape ecology and this relationship isn't always linear so what we need to always bear in mind is how much habitat can we keep in a landscape to ensure the long term viability of koalas high numbers of koalas in the environment we're also interested in this idea of empty habitat what's causing this is it that the habitat is not suitable or is it just that it's not accessible koala populations have died out or been killed off by other components disease as an ongoing issue with koala populations what causes this disease expression does fragmentation affect it for example we've got a fantastic example from the Morton Bay rail link study it was published by Hawthorne Bayer it showed that dog control and vaccinations led to increasing viability of that population so that's thanks to the work of John Hanger and his group able to look at turning that population around and that's about that's about me I think so it's moving on if anybody's got any questions thanks thanks very much Harriet you certainly generated a number of questions and unfortunately we haven't got a lot of time for those but quickly there was a question there about considering the grazing, agriculture road strikes, climate change and invasive species also occur outside of SEQ and the question is how is the SEQ focus justified and are we focusing on an area that is preferable for humans to live and work instead of areas that contain koalas so it's not a technical question but I guess it's a question of the approach that the government's taking so if you hope to tackle that or I can provide a response I can do a little bit which of course goes to that slide where we know that a very large proportion of the koala population is actually in SEQ so as Tim said before Queensland's very large, koalas are distributed across a large proportion of Queensland but the biggest concentration of koalas or wild koalas in Australia is in SEQ so by no means would we say that there are no threats outside of SEQ or no koalas outside of SEQ but yes our current focus is in this SEQ but you might like to follow up on that Jeff? Yeah happy to add a little bit extra I guess both the expert panel report and the government's response to the expert panel report indicated that the types of mapping that's being developed and applied in SEQ certainly has scope for expansion into other areas of Queensland. Obviously the focus is very much on managing threats to the high density population that exists in SEQ but that's to say at no means at the expense of koalas elsewhere in Queensland the timeframe for that obviously will be dependent on how successful we are in implementing these things in south-east Queensland so it's not to say that that focus ignores what's happening elsewhere in the state certainly some of these threats and the information from the habitat mapping is also used in essential habitat mapping which again is used to identify and protect koala habitat outside of SEQ Got one more question I think that we can have time for and that's have the wild and feral dogs been included as a threat layer or consideration particularly given the instance of the wild dog impact on koalas that was identified in the Morton Rail project Yes so wild dogs is one of the data sets we'd like to incorporate in this next phase we weren't able to easily get a data, well it's difficult to get a data set that actually represents wild dogs so we're definitely looking at incorporating that into this threat mapping because absolutely right I mean the wild dogs just devastated the koala population and around the Morton Rail link for sure we've got to include that in the analysis Thanks Thank you Welcome back to the fourth presentation of today's Koala CoLab 2021 conference series our fourth presenter today is Professor Amando Opan Amando is the Professor of Geographic and Information Systems and Remote Sensing at the University of Southern Queensland His research interests focus on the applications of remote sensing and GIS to observe terrestrial ecosystems and their responses to environmental climate change He's applying geospatial technologies to map, monitor and model forest vegetation biodiversity land cover and use agricultural crops, floods and droughts He has over 180 published papers in international refereed journals and conference proceedings He was also the recipient of the Queensland Spatial Science Excellence Award for Education and Professional Development in 2006 and was elevated as a fellow of Australia's Surveying and Spatial Sciences Institute in that same year So today Amando will be presenting in relation to GIS and remote sensing in southwest Queensland Thanks very much Amando Thank you Jeff and good morning to everyone My presentation is about modelling and mapping of wild habitat and threats in southern Queensland I would like to acknowledge my co-investigator Utam Srestha and Kate Reardon-Smith and basically we are involved in this project together although the modelling majority of those was made by Dr Srestha Basically something different or some aspects that's not mainstream in our study is number one we focus on southern inland Queensland and that's an area we're concerned part of our region as well and another aspect that's more or less different is we would like to focus or we would focus on the impacts of climate change There has been a study on this but the resolution that they did is relatively coarse and the final aspect that we considered something new or innovative is we used the Biomode tool which is an ensemble of modelling algorithms in addition to the max and software algorithm that's being used The presentation outline covers the following I'll be presenting a brief information about the rationale of the project the objectives and the study area My slides cover two major parts first one is habitat suitability mapping and the second one is on threats analysis and mapping and then I'll end up with conclusion and a few acknowledgement So the rationale of the project we know that the maps of quality distribution and the habitat suitability are one thing especially in the southern inlands Queensland and the limited information exists about threats to quality populations and distribution For climate change impact analysis the existing course scale which is about 10 km by 10 km has limited use so that gives us the objectives for this course for this presentation is to generate the quality suitability maps using SDM, the previous presentation they already talked about that but we focus on climate change or current and future climate The second one is to identify, prioritize and map especially explicit threats to koalas and also to analyse the cumulative impacts of the threats in relation to the distribution of koalas and the last one is to engage with stakeholders to better inform strategic planning and management So our study area basically covers the catchment boundary the QMDB Queensland Maridarling Basin you can see in this area we have here Tobumba and then major towns or cities includes Chinchilla, Roma here and then Charleville, Kanamala and it covers approximately 259 sq km vegetation types include Eucalypt and Brigalo woodlands on the east, grassland with scattered acation on the northwest and Malga acations in the southwest and there are many river flood plains and drainage channels that support many koala feed trees So I would like to present an overview of the methods that we have used for habitat suitability modeling So first we collected data and pre-processed including assembling and filtering of koalakaris data a typical procedure in many SDM studies and then we collected bio-climatic variables as well as environmental variables and then we have done correlation analysis, the purpose of those is to exclude or include variables In addition to this we also used methods to reduce sampling bias and other pre-processing techniques to ensure that we will get a valid result. So after that we have done modeling and validation we use species distribution modeling but we did some schemes where we analyzed first climate only variables second part we analyzed environment only variable and then we can combine that and we did the analysis for the current climate 2030 and then 2070 and then output generation we have maps of probability distribution to generate some information about area and you know percentages we categorize that probability distribution into different categories like suitability classes as low, medium, high, very high and then we have summary statistics okay so data sets used we have used lots of data but we ended up in the final model we have this so in the species distribution modeling part we obviously we use location data occurrence data from these different sources we have bioclimatic variables we use narclim data so our study attempted to really like we would like to have a high resolution, relatively high spatial resolution with regards to climate modeling because if you will probably you're aware that the best resolution we could get outside narclim data is the 1000 meter or 1 kilometer and that has a severe limitation with regards to resolution so narclim climate data was developed by New South Wales and ACT climate modeling project and it so happened that this part of Queensland was covered by that data set and then we have also used different environmental variables from different sources including elevation, altitude, soil and use vegetation types etc from different data sources or mapping threats we have initially considered 12 mappable threats during the workshop we have conducted I will discuss more of that but we ended up incorporating these 6 variables or threats because we have to be able to map or make at least surrogate maps out of those identified threats so with regards to quality occurrence points as I mentioned we obtained those points close to 1000 but because we have to remove duplicates and points with dubious locations and also we have to ensure that there's only one point inside this 250 meter by 250 meter cell size we have to use 123 unique occurrence points and that's a typical requirement for the modelling for the bioclimatic data as I mentioned we used the data from NARCLEAN project it's a relatively high resolution 250 meter data we used this CSROMK30 GCM and R2 configuration for the current climate we have this data set or this sub-period here and for the projected climate we have this 2030 and 2070 so initially we have 35 variables but we have to reduce it to 10 bioclimatic variables these are those variables here from annual mill temperature to lots of these including moisture indexationality etc and the reason we have to do that is in SDM modelling we have to remove those variables with high multi-colinearity to make the model modelling dependable so this is just a more detailed representation of the modelling tools I already mentioned about data preparation analysis but more specifically for the modelling and validation so we use ensemble models different models available in biomode 2 package in R where we use 70% data sets for training and 30% for testing which is pretty much standard in many modelling methodology we use different measures of accuracy for model validation and we selected those models at least with greater than 0.6 TSS score and then we use the different models we did an ensemble using weighted mean and then for the output we have the distribution and then of course the suitability classes with very high high moderate and low and now for the environment only variables because we are familiar with max and just like in any modelling so for the environment only variables we use max and probably we can also use biomode but because you know this is something that is available and easily to manipulate many of our programs are in max and as well so we use this so basically same thing we use different environmental variables as I mentioned here so altitude, soil, land use, vegetation as the final environmental variables use and I believe many of you are familiar with SDM if not basically SDM is also called niche modelling, bio climatic envelopes habitat suitability modelling and it assumes that the current distribution of the species is a good indicator of ecological requirements it estimates the probability of a species occurring in a place as a function of environmental conditions of that place and this is a mature modelling technique in fact if you google search SDM there is actually 2.79 million results in google scholar about SDM so for our algorithm for modelling as I mentioned we use biomode 2 in R so we basically used 8 algorithms out of 17 so the first four models are regression based and the last four are machine learning and we conducted a workshop attended by over 35 people from different government agencies organizations, local community groups private individuals and also academies so basically we have first workshop discussion with the local community groups about data and information needs and also workshop number 2 presented preliminary results that we discussed about habitat threats etc so results this is a map of the suitable habitat for Kuala under current climate so basically we have this output which is what we call in GIS terminology it's a continuous variable continuous surface and this value is from 0.07 to 1 basically so that's the output and you will note here that here near the dividing range great dividing range this is close to Togumba Darling Downs area and as you move west of the catchment you will see that the suitability is low and we sort of expected that and that's really the modelling and the modelling produce good a higher than 0.60 where 1 is the best and negative 1 is the worst so now we converted this probability distribution into suitability classes now this is a methodology or a part of the analysis where we just want to convert this into classes for the purpose of basically we would like to categorize as low moderate high very high and the good thing about this is we can generate some statistics like area for each class and percentages of each class but the true value of any modelling really is the one here on the left because I can apply many different schemes on how to categorize in a GIS environment you can apply at least four or five methods from manual method to interval yanks optimization and many things but in this case we use geometrical interval which is the one suitable for our data set now for 20-30 you can see here that there's been a slight decrease with regards to the map and if you compare for example with this one in current climate you can see that this area here or near to the great dividing rates it's now reduced I show you some statistics and after that 20-70 that's the output and I'll be able to quantify the different areas and percentages of the suitability classes in a minute so if we will combine everything here from current to 20-30 to 20-70 these are the different like total area is square kilometers of current 20-30 and 20-70 and we can see here that for current suitability like 25% is called very high category in 20-30 there is a reduction from 21% to 17% of the total area and interestingly in 20-70 it increased a bit also like to 20% so with high similar trend but something that's probably I would like to highlight here is the moderate class there is a consistent decline from 28% to 27% to 26% and correspondingly the low suitability is increasing with a bit of change here so we can quantify this and we can sort of understand not just the statistic the data in area and percentages but also the variation of those pixels or grids now for the environmental variables these are the original variables that we included like altitude, slope soil, vegetation, distance of water and this is separate from climate only so this is for the environment only and we use different data set from different sources so what we want to do is to really be able to look at the most important environmental variables and ultimately we use these four soil altitude, land use, vegetation and we base it on the area under the receiver operator carb of 0.85 so these are the variables we use for environment only input for climate only so this is the output of that modeling and this is the probability distribution map it's more crisp because we were able to use 30 meter cell size compared to climate only the best we could get for climate only modeling is the 250 meter so the AUC value is 0.88 which is the indicator of good model performance so with regards to result this is a combination now also we have the climate only here on top we have the environment only here at the bottom and we have the ability to combine this as an option to see if we have those two schemes of modeling so this is the climate and environmental environment variable combine now so again from the final suitability habitat map we have categorized that into four so climate plus the environment variables now with regards to comparison we would like to compare what happened with the performance of the climate only and the environment only we can say for example for very high suitability class 19% but for the environment environment only model is 11% and the reason is of course with the environment only model it's more restrictive because you know we have more like factors that govern the distribution including land use, vegetation types et cetera et cetera but if we can combine that we can have this different statistic so threats analysis similarly we have followed the different procedure here so first threat identification and ranking from literature and workshop I just mentioned and then we collected the data the main challenge in any sort of mapping is if you have a variable you have to find the right expression of that variable in mappable form or something that you can map because if you cannot map it you cannot include in this sort of modeling and yeah we have done that we processed the data now the question then is how would you combine all those threats to come up with a single map and there are many ways to do it but for our purpose we use the weighted overlay approach, weighted linear combination I mean we can use many methods but for the purpose of our work we use this and yes after that we combine the threats and the existing and the habitat suitability map with the one we obtained from the previous habitat modeling I'm sorry and when you combine those you will be able to identify the different combinations between threats and suitability and to come up with some strategies we did not develop actual strategies but that could be done with regards to say prioritization which the previous speakers have mentioned okay so with regards to threats so as I mentioned we have 12 variables and finally we have 6 okay so this was done during the workshop so different experts they have to individually identify and rank the different threats and we have a procedure where we could identify the different like the most important so to speak of those variables and we have habitat conversion loss of habitat presence of dingos or wild dogs we have done that using another SDM modeling we also have traffic collision as important variable here but we use raw density as a proxy we have fire okay and then urbanization we have the population greed habitat loss and then you know part of that data is from global forest watch so these are the maps and based on our ranking we use the workshop the expert ranking we have these following weights for forest loss 25% for habitat conversion 20% for fire 20% population 15 bingo 10 so we use these different weights for GIS based weighted analysis and also we have these results that we obtained where we were able to put some values you know it's scaled from 1 to 9 where 1 is lowest and 9 is highest so this is the number of pixels with the log transform so basically we were able to output this using the GIS best method now this is a quite a busy map but the idea here really is we want to know the different threats in the different suitability classes because you know we would like to know for example what's the threat situation in very high suitability areas so here we were able to look at for example area with at least one threat and then the 1 in percent the percent of total area and then another column in the table is the area with threat intensity score greater than 5 okay and of course the percentage so basically we can see here that you know in very high and high suitable area the threats are also high okay at least 1 which is probably low but also you can say relatively there are also big threats for relatively for those greater than 5 threat intensity score so that's the situation and we can identify that on a map when we combine those so I think if we will combine the threats and also the suitability class similar with other presenters they would come up with something like combination like high threat and high suitability moderate threat and moderate suitability and then low threat and low suitability and this is something that could be useful for many conservation work in terms of prioritization resource allocation etc etc and of course you know this is a relatively high area the extent is relatively high if you look here at the scale you know that's 100 kilometers and you can say you know with a map with this kind of map with 250 meter cell size looking at you know let's say local level planning probably you can say is it really useful so we look at example like this like a pitch work locality which is here Tomumba so for example we were able to look at mapping those combination of threats and habitats so we have this green here as low combination of threats and habitats and high so that's the sort of use possible use of the data sets that we have produced generated from this project so in conclusion quality habitat suitability including current and projected climate and threats were successfully mapped with good model performance and from here climate change will have a significant impact on the area of habitat the potentials of habitat potentially suitable for koalas in the study area and 83% so this big of the total area has at least one threat intensity level of koala populations and the distribution of threats areas of more suitable habitat are also having a higher threat level okay so I just want to end this presentation it's a bit fast by thanking the Queensland Department of Environment and Science for funding this and also for the many people who participated in the two workshop we held here at USQ and also online and we have produced a report of these and data sets I'm happy to share the report and data sets for those who are interested, thank you thanks very much Amanda again I think your presentation again demonstrates that there's interest in looking at koala distribution and threats beyond South-East Queensland and the work that you and your team have done is certainly taking us into a new area of exploring how we might model and map koala habitat in those broader areas in western and central Queensland there's a number of questions that have come up some that are technical related to your presentation some that are broader which I'll probably seek to address either in written answers or otherwise one key question here was the questioner was just interested in how you might account for standing water and drainage lines particularly given the observation that these are really important to koalas in the western parts of their range so yes that's a good question yes actually we included that in the initial modelling in the environment only variables we have a data set called distance from drainage and water sources so we included that but in the final model because of you know we have to select the most important variables to get the best modelling output so we didn't include that in the final model but in the initial screening of the variables we included that distance from water or water sources thanks for that there's a couple of questions with regards to using different inputs into the mapping I think you covered up on the use of including predators and the life cycles of koalas as part of that modelling I'm not sure if you wanted to add anything further to that yeah we attempted to look at predators there has been a number of discussion during the workshop and the problem really is data set we could not really find the best variables with regards to predator and other related the issue so we ended up with dingo data set because we have a colleague working with dingo or Ben Allen and we did the kind of modelling but yeah that's a challenge and hopefully in the future we'll be able to develop some more like a better technique to incorporate the predator variable okay another question here which is it's about whether the modelling and mapping you're using is dynamic so I guess the question is if the scope or scale of threats changed is the modelling and the mapping approach you've adopted dynamic such that those outputs from that mapping would change okay yes well with regards to the use of bio mode and max and there are facilities or options to incorporate like new data sets that will say portray changes like for example we use 2030 climate and 2070 climate so it means that there will be changes in land use land cover we have to do the modelling again in other words another run of the model so that's the only way but when you say is it like when we do some sort of sensitivity analysis by just changing or sliding some values or thresholding it's not possible at this stage but it means we have to run again and probably not a difficult thing because in bio mode we have the code to do that it's a matter of preprocessing the new like new sets of land cover like land cover updated land cover land use and that's the only way to do it unfortunately but we can run it again okay there's a couple of questions here which I'm trying to respond to which actually relate more to the broader protection mechanisms around koalas so again as we've done in previous sessions we'll seek to provide some written response to people in relation to those questions um just trying to access that final question um the you're obviously the question that one particular person has asked a question about what variables you're using both biotic and abiotic variables and in terms of climate are you using only temperature or are you using rainfall humidity and other climate related factors okay yes so with the narclim bioclimatic variables so it's a combination of temperature and precipitation but there are also a combination of those to come up with bioclimatic expression I don't know if yeah probably I don't have time to show this but we have uh now let's see yeah yes because I cannot remember everything we have many of those in fact they're 35 so we ended up with you know like annual mean temperature biowan means the urinal temperature range etc so yeah it's these bioclimatic variables that uh that presented as important in modeling uh 10 of those and we use that but if the question is something about you know like what sort of scenario that we have used uh the scenario that we use is CSRO MK3.0 Global GCM so it's a scenario where it's a warm dry scenario relative to 1999 to 2009 and I believe it's R2 configuration so they say that it's probably the best of the best among the three so it's a scenario where it's a warm dry see uh playmate in the future okay thanks very much uh Amando again um really complimentary presentation to the other presentations that we've had this morning so thanks very much for that input um that presentation wraps up today's um koala co-lab uh event uh we'll be having the other events on the following Wednesdays over the next five weeks uh certainly invite you to participate in those sessions um the information uh from uh today's session will now also be available to review uh via the portal the next session as I said is on Wednesday the 20th of October commencing at 9 30 a.m going through till 11 45 um certainly encourage you to um invite colleagues and other people to attend uh these sessions are free so we're very keen to get as many people involved as we can apologies to those people who haven't yet got answers to their questions as I said we do intend to provide those responses um so that we can address the the various questions and comments that have come up through the forum today so thank you to all the presenters and thank you to everyone for attending