EcoAssets:
Streamlining access to data from Australian environmental research infrastructures Shandiya Balasubramaniam
@ShandiyaB
Ecological Society of Australia 2023

I acknowledge the Traditional Owners of the lands on which we meet, the Larrakia people, as well as the Traditional Owners of the lands on which I live and work, the Wurundjeri people of the Kulin Nation. I pay my respects to their Elders past and present. I recognise the spiritual and cultural significance of land, water, and all that is in the environment to Traditional Owners, and their continuing connection to Country.

EcoAssets

Today I’m going to talk about EcoAssets, which is a project that brings together 3 research infrastructures, the Atlas of Living Australia, the Integrated Marine Observing System, and the Terrestrial Ecosystem Research Network, to provide data in forms that support environmental reporting, such as the 2021 State of the Environment report.

Image from the Biodiversity Heritage Library;, contributed by Missouri Botanical Garden, Peter H. Raven Library

The 3 research infrastructures are funded to collect, manage, and organise different types of environmental data: The ALA is Australia’s national biodiversity database, bringing together data from natural history collections, researchers, government departments, community groups, and individuals. TERN measures key terrestrial ecosystem attributes at representative locations and produces ecosystem datasets at a national scale. And IMOS collects data on all aspects of the marine environment, with extensive monitoring activities in coastal and open waters.

EcoAssets streamlines the flow of data from national research infrastructures into environmental reporting through versioned, aggregated datasets.

Historically, data from these activities have certainly fed into environmental reporting. Notably, the 2016 State of the Environment report referenced particular datasets from these infrastructures, but: it was difficult to access the data directly, and there wasn’t any option for revisiting those datasets at a later point in time if someone wanted to review the results.

EcoAssets addresses this by streamlining the flow of data from these infrastructures into reporting, by producing aggregated datasets that can be more easily accessed and used for analysis, and versioning these at regular intervals to support longer time-series analysis. This means that all the data underpinning these reports will be available for scrutiny or re-analysis, and users can have greater confidence in the summaries that are presented.

We’ve produced two types of datasets, and I’ll give an overview of both them and talk through some examples.

Monitoring

The first type of datasets relates to monitoring effort across the country

One of the challenges in compiling State of the Environment Report chapters is that studies are often place-based, meaning that compiling a truly national picture requires integrating separate studies, often conducted using different methods, and with gaps between them.

- Dr Dan Metcalfe, CSIRO and State of the Environment reports 2016 & 2021 chapter author ”

One of the points that was raised about state of the environment reporting, which you’ll see on the slide here, is pertinent to our understanding of national monitoring effort. Studies are often place-based, short term, and use different terminologies, so integrating them to get a national snapshot is challenging. EcoAssets has compiled monitoring datasets from the ALA, TERN, and IMOS, and mapped them to a common vocabulary to facilitate comparison of survey effort across time and space. These datasets go from 2010 to 2022, and include data from TERN monitoring stations across the country, IMOS marine surveys, and other systematic surveys from the ALA, where there was a clearly defined event ID and sampling protocol, as well as date and location.

Aggregated Monitoring Data

nri	datasetName	datasetURI	nriKeyword	decimalLatitude	decimalLongitude	year	stateTerritory	ibraRegion	imcraRegion	featureID	featureName	featureFacet1	featureFacet2	featureFacet3
ALA	SA Fauna (BDBSA)	https://collections.ala.org.au/public/show/dr365	observed-no method stated: 884 samples	-34.92498	139.0421	2018	South Australia	Kanmantoo	NA	https://biodiversity.org.au/afd/taxa/65625205-db74-4a87-b566-ca387b119974	Biodiversity | Animals | Vertebrates | Birds	Biological Classification	Animals	Birds
ALA	SA Fauna (BDBSA)	https://collections.ala.org.au/public/show/dr365	observed-no method stated: 24 samples	-34.89952	138.5954	2010	South Australia	Flinders Lofty Block	NA	https://biodiversity.org.au/afd/taxa/65625205-db74-4a87-b566-ca387b119974	Biodiversity | Animals | Vertebrates | Birds	Biological Classification	Animals	Birds
TERN	TERN Surveillance monitoring site NSAMDD0006	http://linked.data.gov.au/dataset/ausplots/site-nsamdd0006	soil profile	-33.66430	143.1670	2014	New South Wales	Murray Darling Depression	NA	https://gcmd.earthdata.nasa.gov/kms/concept/7a16aa40-c74b-4a69-a230-1edd1b453332	Land Surface | Soils | Soil Horizons/Profile	Land Surface	Soils	Soils
TERN	TERN Surveillance monitoring site QDAEIU0010	http://linked.data.gov.au/dataset/ausplots/site-qdaeiu0010	plant occurrence	-19.89340	146.1820	2021	Queensland	Einasleigh Uplands	NA	https://id.biodiversity.org.au/taxon/apni/51414459	Biodiversity | Plants	Biological Classification	Plants	Plants
TERN	TERN Surveillance monitoring site SAAEYB0008	http://linked.data.gov.au/dataset/ausplots/site-saaeyb0008	plant community	-35.04930	137.2140	2020	South Australia	Eyre Yorke Block	NA	https://id.biodiversity.org.au/taxon/apni/51414459	Biodiversity | Plants	Biological Classification	Plants	Plants
IMOS	IMOS SOOP-XBT Upper Ocean Thermal Data collected in the Tasman Sea	https://catalogue-imos.aodn.org.au:443/geonetwork/srv/api/records/55b6a183-e942-424e-b3bd-476dfb284fc0	oceans | ocean pressure | water pressure	NA	NA	2010	Victoria	NA	Tweed-Moreton	https://gcmd.earthdata.nasa.gov/kms/concept/dd025312-0d27-44e0-ae05-7cfcc1aa17f0	Oceans | Ocean Pressure | Water Pressure	Oceans	Ocean Pressure	Ocean Pressure

This is a sample of what the aggregated dataset looks like: on the left is information about where the data is from and some associated keywords, in the middle we’ve included information about survey data and location, and cross-referenced this against spatial layers that tell you which state or territory the survey was in, and whether it intersected with terrestrial or marine bioregions, and on the right is where we’ve mapped it all to a common vocabulary. As you progress towards the right, the values in the columns get more specific, so it lets you look for information at different hierarchical levels. For instance, if I were only interested in bird surveys, I could just filter on this column for birds.

For example, here I’ve chosen biological monitoring events only, and mapped the number of surveys in each terrestrial region, for every year. These maps highlight the vast difference in monitoring effort across regions - some areas are consistently surveyed every year, like parts of South Australia, whereas other areas appear to be data-deficient, whether due an actual lack of monitoring or because EcoAssets doesn’t have that information. So that’s the first type of datasets we’ve produced, …

Biodiversity

… the second type summarises biodiversity. These are based on species occurrence records - so, records of a taxon detected in a particular time and place. In aggregate, these data represent our understanding of biodiversity

Aggregated Biodiversity Data

year	basisOfRecord	stateTerritory	ibraRegion	imcraRegion	forest2018Status	forest2013Status	capadStatus	epbcStatus	griisStatus	speciesID	speciesName	occurrenceCount
1900	HUMAN_OBSERVATION	New South Wales	Australian Alps		forest	forest	PA	Not listed	Native	https://biodiversity.org.au/afd/taxa/70b2fe09-3113-4167-bb3d-6f579419b747	Antechinus mimetes	1
1900	HUMAN_OBSERVATION	Australian Capital Territory	South Eastern Highlands		non-forest	non-forest	not protected	Not listed	Native	https://biodiversity.org.au/afd/taxa/5254fe03-630b-44b2-9233-df51a7b8f25f	Pardalotus punctatus	1
1900	HUMAN_OBSERVATION	South Australia	Channel Country		non-forest	non-forest	PA	Not listed	Native	https://biodiversity.org.au/afd/taxa/cbef5be9-829a-4691-9de4-125356441b1f	Calidris subminuta	1
1900	HUMAN_OBSERVATION	Western Australia	Avon Wheatbelt		non-forest	non-forest	not protected	Not listed	Native	https://biodiversity.org.au/afd/taxa/031b2b69-e9fc-44c6-9df9-03c1470d5ec3	Sericornis frontalis	1
1900	HUMAN_OBSERVATION	Tasmania	Ben Lomond		forest	forest	PA	Not listed	Native	https://biodiversity.org.au/afd/taxa/ac61fd14-4950-4566-b384-304bd99ca75f	Ornithorhynchus anatinus	1
1900	HUMAN_OBSERVATION	Queensland	Brigalow Belt North		non-forest	non-forest	not protected	Not listed	Native	https://biodiversity.org.au/afd/taxa/112d02bc-0878-49e1-b605-d971d27698e3	Eudynamys orientalis	1
1950	HUMAN_OBSERVATION	South Australia		North Spencer Gulf	non-forest	non-forest	PA	Not listed	Native	https://biodiversity.org.au/afd/taxa/cf17c9bb-09d1-47ce-adfe-dc060ba0b04c	Morus serrator	1
1950	HUMAN_OBSERVATION	Victoria		Otway	non-forest	non-forest	PA	Not listed	Native	https://id.biodiversity.org.au/node/apni/2916905	Atriplex billardierei	1

So we’ve produced a dataset that comprises counts of species from 1900 onwards, and a number of derived datasets that describe the state of threatened species, invasive species, and the use of protected areas in both terrestrial and marine regions. This is a sample of what the aggregated dataset looks like - there’s information on …. This makes it possible to ask questions at different spatial, temporal, and taxonomic levels. I’ll start with an example of a single species.

Channel-billed Cuckoo

Scythrops novaehollandiae

Image from the Biodiversity Heritage Library; contributed by Smithsonian Libraries

This is the channel-billed cuckoo, which has a fairly wide distribution across northern and eastern Australia. It’s also the world’s largest brood parasite.

Channel-billed Cuckoo

There are just over 69,000 records of CBCs in the dataset, and this figure shows the number of records each year since 1970. There’s a clear upward trend across the years

Channel-billed Cuckoo

We can also look at where the records are from - the figure on the right shows the number of terrestrial bioregions with records, over the same time period. The fact that it’s being seen in an increasing number of regions suggests that the species could be expanding in range.

Channel-billed Cuckoo

If we map this across the five decades, we can get a better idea of where the expansion is occurring - here it looks like the cuckoo is spreading southward. It’s not unusual for individuals to move around, but changes in the range of a species may indicate that species are responding to disturbances like climate change, and these data have the potential to provide some insight into these sorts of species-level changes. We can also use these data to understand what’s happening with groups of species…

Invasive and Introduced Species

… here I’ve pulled out information on species classified as introduced or invasive. The heatmap shows terrestrial regions in columns, and 5 year time periods in rows. The colours represent the proportion of introduced and invsive species for each segemnt of time and space, relative to the total number of species. The darker colours represent higher proportions. And this can be extended to other regions, like marine bioregions, or other types of species, like threatened species.

Protected Areas

not shown: Heard Island and McDonald Islands Marine Reserve

These datasets also cross-reference occurrences with regards to the protected area network, which includes the National Reserve System and Marine Protected Areas. This shows the extent of the network, with marine and terrestrial protected areas shown in darker shades.

Comb Wrasse

Coris picta

Image from the Biodiversity Heritage Library; contributed by Harvard University

To show how we could use this information, I’ve pulled out records of the comb wrasse. It’s found in coastal and offshore rocky reefs, between southern Queensland to southern New South Wales, and there are just under a thousand records of it in our dataset

Comb Wrasse Records in Protected Areas

Over 60% of those records come from protected areas. To put this in context, the protected area network only covers about 20% of the Australian landmass, which suggests that the wrasse may be dispropotionately using protected areas. More broadly, 12% of all species in our datasets have only ever been recorded in protected areas. So this can be a useful facet to include when we think about how to manage species.

Considerations

Aggregating millions of records from a diverse range of sources is inevitably going to result in some degree of error, with uneven levels of coverage, completeness, and methodology. Where possible, we’ve tried to compensate for this. There are 2 things in particular that I wanted to highlight today: the first is that we’ve excluded records if an organism was not identified to at least the rank of species, but this biases the data towards taxonomic groups that are well studied. It’s impossible to eliminate this but aggregating records means that the bias should be consistent across regions and time periods.

Record Types

And the second thing to note is that our evidence for the occurrence of a species has also changed over time. For much of the 20th century, this evidence came from museum specimens.

Record Types

More recently, this has changed in favour of human observations, which includes data collected by ecologists, field monitoring programs, and, increasingly, citizen science projects. This has changed the likelihood of a species being recorded. Some species are only ever recorded by specialists working in collections, whereas common species are probably going to be under-recorded in collections.

Record Types

Another record type is machine observations, which includes images from camera traps and acoustic recordings. These currently make up a small proportion of the overall records, but it’s likely that they will soon become more prevalent. Records in the aggregated datasets we’ve created are separated by the basis of record field, which allows users to make judgements about which types of records to include in any analysis.

ecoassets.org.au

That was a whirlwind tour of many months of work - there’s much much more to the datasets, and if you’d like to check them out yourself, this is the website where they can be downloaded.

Shandiya Balasubramaniam
Data Analyst, Decision Support Program Lead
Science & Decision Support, ALA
shandiya.balasubramaniam@csiro.au
@ShandiyaB
shandiya

Slides: shandiya.quarto.pub/esa2023

And if you’d like to take another look at anything, the link to the slides is there. Thank you.