Recent advances that combine remote sensing tools with drones provide the capacity to monitor and evaluate landscapes at a precision, accuracy and scale that are now relevant for large ecosystem restoration projects. Although many of these tools are becoming available as commercial turnkey solutions for environmental assessment, successfully operationalising these platforms still presents a range of challenges. The forthcoming rehabilitation of Ranger uranium mine (approximately 1000 hectares) offers an opportunity to implement landscape-scale drone-derived ecosystem assessment tools. A high resolution assessment of the restored ecosystem will allow for measurements of plant species diversity, abundance, patchiness, function and ecophysiology that are nearly scale independent for the purposes of large restoration projects. Using this approach, census-based data sets are now feasible, reducing the need for data extrapolation and interpolation.
While we have had success deriving landscape-scale metrics to inform ecosystem restoration, we have also had to manage legislative, logistical and analytical challenges. These challenges include adapting to the legal framework for operating large drones, an ever increasing data management capacity, the development of automated tools for processing and analysing very large datasets and the adoption of machine learning analytical techniques. There is an ongoing need to scale from plot level, ground-derived metrics to our established multirotor drone fleet through to long-range fixed-wing drones. A key driver for developing this approach is the need for effective tools that can confirm the restored Ranger mine site develops into an ecosystem similar to the surrounding environment of the UNESCO World Heritage Listed Kakadu National Park.

Biodiversity Information Standards (http://tdwg.org), has a ‘Data Quality’ Interest Group with Task Groups to develop a Data Quality Framework (TG1), Tests and Assertions (TG2), Case studies (TG3) and Vocabularies (TG4). The framework has been published (see https://tdwg.github.io/bdq/tg1/site/). TG2 has spent 3 years developing 100 core tests with associated assertions (https://github.com/tdwg/bdq/projects/2). TG3 has summarised 28 biodiversity-related use cases in evaluating ‘data quality’ - https://tinyurl.com/yyeb73bq and TG4 is identifying existing relevant vocabularies and gaps, and efficient processes for addressing those gaps.

I lead TG2 with the of Arthur Chapman (Australia), John Wieczorek (USA), Paula Zermoglio (Argentina) and Paul Morris (USA). ‘Tests’ classify as Validations, Amendments, Notifications or Measures. Validations test values within known limits and it is the lack of controlled vocabularies that lead to establishing TG4. An example validation is “geodetic datum not supplied”. Amendments may be possible if an adequate context is available. An example is “Taxon rank has been standardized”. A notification alerts potential issues, for example “the location of the observation has been generalized”. A measure summarizes issue status for each record, for example “Validation tests that result a status of Not Compliant”.

TG2 has developed a template defining key parameters for each test: a unique ID, a label, classification, Darwin Core elements used (see https://dwc.tdwg.org/terms/), expected test response, data dimension (name, space, time, other), warning type, parameters (default source authorities or values), an example, references and notes: See https://github.com/tdwg/bdq/issues/141 as an example test.

Functional traits capture core features of an organism's physiological function. Previous work, in North America, suggests traits can predict the climatic ranges occupied by different species, thereby connecting species function to the environment. However, the extent to which the same predictive rules apply across different continents remains unknown.

Backed with a large compilation of trait measurements (AusTraits) and species occurrence data (the Atlas of Living Australia), we investigated the potential of five key plant functional traits - height, seed mass, wood density, specific leaf area, and leaf area - to describe the climate ranges of 9,940 woody plant species in Australia. Using quantile regression, we quantified the centre, and lower and upper limits of observed distributions across key environmental variables. We then examined whether the relationship between trait and climatic-niche in Australia was consistent with the previously-reported patterns from North America.

Broadly, we found that, in Australia, all traits correlated with a species climatic niche, as measured by precipitation and temperature seasonality. However, the direction of these relationships contrasted to the patterns in North America. Moreover, the relationships between traits and the centre of a species climatic niche were much stronger in Australia compared to North America. In particular, the strong link between traits and the centre of the climate niche in Australia suggests a stronger role for biotic controls on species distributions. Combined, this cross-continental offers a new view on which “assembly rules” may be general and which are continent specific.

Worldwide, the number of threatened species increases with inadequate resources available to stop the biodiversity crisis. Today more than ever, we urgently need conservation dollars to go further. We present a novel conservation planning tool: the integrated spatial prioritisation (ISP). The ISP provides optimal solutions to both problems of maximising the number of species saved for a given budget or minimising the cost of saving a target number of species. We applied the ISP to a record number of 491 species, 45 actions, and covering the entire state of New South Wales (800,000km2) to provide recommendations for the Saving our Species program (DPIE, NSW). The optimal solution for securing all species included in the analysis would cost AUD$15.335 million annually. Our network analysis revealed that only five species did not benefit from complementarity – on average, management of one species co-benefits an additional 22.2 species with 14% of the cost shared between connected species. The fruit bat large-footed myotis was the champion species with the highest connectivity to other species (353). We also identified 31 ‘umbrella species’ - where investment in their priority sites and actions results in the security of other species for no additional cost. ISP uniquely solves very large conservation planning problems and provides “a la carte” solutions for private investments or solutions that complement existing investments. ISP paves the way to a new wave of agile conservation planning tools that offers best deals for conservation investors and on-ground conservation.

With more than a million users globally, iNaturalist.org is now one of the world’s most popular nature applications for recording and sharing species observations. A joint initiative between the California Academy of Life Sciences and National Geographic, the open source platform has steadily grown since its beginnings in 2008.

The Atlas of Living Australia (ALA) along with its host organisation CSIRO has become a member of the iNaturalist network and now manages iNaturalist Australia, a localised portal into iNaturalist.org. Even without a local presence, iNaturalist has already attracted 10,000 users who have contributed 600,000 observations of around 23,000 species in Australia.

At its core iNaturalist is a social network focused on building a community of naturalists who produce scientifically valid data by recording observations and identifying species correctly. In the past two years, automated image recognition has been deployed to help identify species, which assists even novice users with classifying a species and directing their observations towards expert users who can refine the identification.

Crowd sourcing the curation of observations mitigates against misidentifications and inaccuracies. Building such an online community and the software user experience to support it is no small achievement for which iNaturalist is to be commended. This session will examine some of the data quality issues the ALA has encountered with species misidentifications, and how we expect the Australian uptake of iNaturalist will help improve citizen science data in the national biodiversity record.

Climate change threatens many species though vast environmental changes and extreme weather events. To mitigate climate change impacts, one broad-scale management action is protecting climate refugia. In this study, we mapped malleefowl (Leipoa ocellata) predicted breeding activity during and after the Millennium drought to highlight climate refugia. We fitted a Poisson regression model to malleefowl breeding counts from 144 sites with vegetation productivity, moisture and landscape characteristics: elevation, slope, aspect, ruggedness, soil clay content, patch size, time since fire and topographic position index. We tested the predictive ability of vegetation productivity using the Normalised Difference Vegetation Index, soil moisture and rainfall across multiple spatial and temporal scales. The best model included NDVI at 1 km and the cumulative rainfall, both over May to September, with all landscape variables except slope. Using the best model, breeding counts were predicted across Australia to vegetated areas. During a drought in south-eastern Australia, areas outside the current range seem to maintain high breeding activity, including south of Little Desert and north-central Victoria. Post-drought breeding activity is high in the central parts of the NSW-Victorian boarder. These areas highlight possibly important areas for malleefowl conservation currently and into the future. Similarly, our results indicate that high and stable vegetation productivity at broad-scales is essential for malleefowl. This impact should be taken into consideration when making decisions regarding vegetation management in arid Australia.

Since 2016 we have seen a rapid increase in the range of ecosystem variables mapped and monitored at global scales using very high resolution satellite data. In addition, data collected from drones at very high spatial resolutions are being presented as alternatives to traditional field measurements. The use of various forms of satellite for measuring, modelling, mapping and monitoring ecosystem: a) structure (e.g. tree height , biomass), b) composition (e.g. species, community type) and c) function (e.g. primary production) is becoming increasingly common across both terrestrial and marine environments. This presentation will show that accurate measurement, modelling, mapping and monitoring of structure, composition and function requires careful integration of ecological data and knowledge, with earth observation data acquisition and processing techniques. Examples are presented across a range of terrestrial and marine ecosystems in Australia, stretching from chenopod shrublands and mulga woodlands, through to eucalypt woodland, coastal forest and mangroves, and from small atoll reefs to entire barrier reef systems. The TERN supersites (https://supersites.tern.org.au/) are used to demonstrate best practice for combining field ecology and earth observation for accurately mapping and monitoring vegetation structure, composition and function. The Allen Coral Atlas (https://allencoralatlas.org/) and Great Barrier Reef mapping programs are used to demonstrate how to map reef structure and composition by combining ecological data and knowledge with various satellite image data sets. The work concludes by specifying an approach to integrate ecological and earth observation knowledge to ensure accurate mapping and monitoring of ecosystem structure, composition and function can be completed.

Accounting for the effects of fire on ecological values in fire management planning has been limited by the ability of end users to integrate and assess complex and dispersed data and models. We built a user-friendly Fire Analysis Model for Ecological values (FAME), to enable a more transparent and values focused consideration of ecological assets in fire management. FAME enables land managers and stakeholders to visualise and interpret the consequences of alternative fire strategies on multiple ecological objectives. It brings together inputs of spatial bushfire and planned burn histories, future fire projections, native vegetation maps, species habitat distribution models and models of relative abundance of fauna with time since fire. FAME models and undertakes geoprocessing to output annual changes in: post fire age of vegetation; inter-fire intervals, times burnt outside ecologically tolerable fire intervals; and relative abundance of hundreds of vertebrate fauna species. Outputs are generated as annual raster maps, and aspatial summaries across the area of interest. The code has been optimised to allow rapid calculation of these data for regional or statewide scales, which enables the comparison of the effects multiple alternative burn plans – or future bushfire scenarios. We discuss the development and implementation of FAME, which is already being used to compare the consequences of alternative planned burning strategies in bushfire management planning in Victoria, and to examine the impacts of altered bushfire regimes under climate change on the abundance of threatened forest species, and location of potential refugia.

Meta-analysis plays a crucial role in the synthesis of quantitative evidence in ecology and biodiversity conservation. The reliability of estimates in meta-analyses strongly depends on an unbiased sampling of primary studies. Although earlier studies have explored potential biases in ecological meta-analyses, biases in reported statistical results and associated study-level characteristics published in different languages have never been tested in environmental sciences. Here we address this knowledge gap by comparing effect sizes between English- and Japanese-language literature included in existing meta-analyses. Of the 40 ecological meta-analyses published by authors affiliated to Japanese institutions, we found that only three actively searched literature in both English and Japanese and involved sufficient numbers of English- and Japanese-language literature. In two of the three cases, effect sizes differed significantly between the English- and Japanese-language literature included in the meta-analyses, causing considerable changes in overall mean effect sizes and even the direction of effects when Japanese-language literature was ignored. The observed effect size differences are likely attributable to systematic differences in reported statistical results as well as associated study-level characteristics, particularly taxa and ecosystems covered between English- and Japanese-language literature. Our findings therefore suggest that ignoring non-English-language literature may bias inferences derived from ecological meta-analyses.

Australia is a party to the United Nations’ Convention on Biological Diversity and to attain the terrestrial protected area target of 17% by 2020 set through the Convention, the Australian, state and territory governments have adopted the Australia’s Biodiversity Conservation Strategy 2010–2030. The strategy provides a nationally agreed approach for conserving biodiversity in the face of climate change and other pressing threats in Australia. Working towards this goal, Queensland Government is developing a Protected Area Strategy, which aims at expanding the protected area system (now covering 8% of the state) to secure and conserve representative and resilient samples of Queensland’s biodiversity. According to the strategy, selection and management of new protected areas will be achieved in accordance with best practice standards and principles, including climate change resilience, landscape connectivity, and principles of comprehensiveness, adequacy and representativeness.
Within this framework, this project aims at modelling the current and future (under climate change) distribution of threatened and representative species of Queensland’s biodiversity, and at identifying suitable areas for expanding the current protected area network through an objective spatial prioritisation process. Areas that are expected to support the persistence of the species under climate change and that are spatially connected will be recognised as a priority for inclusion in the network.

For the past year I’ve been involved in a project for the Academy of Science, looking at (big) data in Australian research, and our capacity to respond well to the challenges and opportunities of a future that increasingly has large amounts of data available. In this talk I’ll present a few key thoughts from that project – I hope these will be useful for ecologists across organisations as they interact with data, and for academics in both teaching and research.