TERN Surveillance Training 2019 - Day 5, Final Lectures
1. Training Session 5 โ Almost There!
๏ง TERN Surveillance โ Data and Sample Curation (Christina Macdonald)
๏ง TERN Surveillance specimen and sample loads (Ben Sparrow for Sally OโNeill)
๏ง Woodlands Protocols (ben Sparrow, for Sally OโNeill)
๏ง Condition Protocols (Ben Sparrow, for Sally OโNeill)
๏ง Future Protocols โ Invertebrates (Ben Sparrow, for Sally OโNeill)
๏ง TERN Surveillance โ New tech (Andrew Tokmakoff)
๏ง Remote Sensing Validation (David Summers)
๏ง The Role of Citizen Science (Katie Irvine)
๏ง TERN Surveillance Research and Applications (Samantha Munroe)
2.
3. Data Curation
Data comes back from the field in components
โข The Tablet (requires little curation)
โข Camera- Photo panoramas- JPG and CR2 files
โข Soil analysis data โ Soil site spreadsheet (.xlsx)
โข Leaf Area Index wand- text file
4. The Field Data Collection App โ The Tabletโ
โข Developed in house by Andrew Tokmakoff running on
consumer grade Android tablets.
โข Collects data according to TERNโs protocol
โข An efficient data collection method that avoids double-
handling and transcription errors
โข Maximises integrity of data with drop down menus to
record site attributes and characteristics.
โข Scans barcode numbers, provenancing all plant, DNA,
soil and metagenomic samples.
โข Functions without a network (remote locations) then
uploads data to safe storage when coverage is
available.
5. The Tablet Modules
โข Site description
โข Vegetation Vouchering โ field names & barcodes
โข Genetic Vouchering โ DNA sample barcodes
โข Point intercept data- 1010 points of observation
โข Basal wedge recordings
โข Vegetation structural summary
6. Soil Site Spreadsheet
โข Soil data is recorded on paper then
entered to an excel template.
โข The soil site sheet has 33 dropdown
menus. Records soil surface attributes,
disturbance and analysis results. Back in
the lab, barcodes are added.
โข Uploaded to the Database.
โข Looking to move this process to the
Field App.
8. Pick lists control the quality of data and eliminate
transcription error.
Surface Disturbance
0 - No effective disturbance
1H - HEAVY grazing by hoofed animals
1M - MEDIUM grazing by hoofed animals
1L - LIGHT grazing by hoofed animals
2 - Limited clearing
3 - Extensive clearing
4 - Cleared for pasture, but never cultivated
5 - Cleared for pasture, cultivated at some stage
6 - Cultivated, rain fed
7 - Cultivated, irrigated, past or present
8 - Highly disturbed, e.g. Mining, urban
NC - not collected
Surface Coarse Fragment Abundance
0 - No coarse fragments
1 - Very slightly or very few (< 2%)
2 - Slightly or few (2-10%)
3 - No qualifier or common (10 - 20%)
4 - Moderately or many (20 - 50%)
5 - Abundant (50 - 90%)
6 - Extreme abundance (> 90%)
NC - not collected
Surface Coarse Fragment Size
1 - Fine gravelly or small pebbles (2-6mm)
2 - Medium gravelly or medium pebbles (6-20mm)
3 - Coarse gravelly or large pebbles (20-60mm)
4 - Cobbly or cobbles (60-200mm)
5 - Stony or stones (200-600mm)
6 - Bouldery or boulders (600-2000mm)
7 - Large Boulders (>2000mm)
NA - not applicable
NC - not collected
9. Pick lists control the quality of data and eliminate transcription error. Pick lists control the quality of data and eliminate transcription error.
Surface Coarse Fragment
Lithology
AD - Adamellite
AG - Agglomerate
AC - Alcrete (bauxite)
AM - Amphibolite
AN - Andesite
AH - Anhydrite
AP - Aplite
AR - Arkose
BA - Basalt
BB - Bombs (volcanic)
BR - Breccia
KA - Calcarenite
KM - Calcareous mudstone
KL - Calcilutite
KR - Calcirudite
KC - Calcrete
CH - Chert
CO - Coal
CG - Conglomerate
CU - Consolidated rock
(unidentified)
SD - Detrital sedimentary rock
(unidentified)
Erosion
Severity
X - Not apparent
0 - No erosion
1 - Minor or present
2 - Moderate
3 - Severe
4 - Very severe
NC - not collected
Erosion
Type
W - Wind
C - Scald
M - Mass movement
S - Sheet
R - Rill
G - Gully
T - Tunnel
B - Stream bank
V - Wave
E - (see site notes)
NC - not collected
NA - not applicable
Erosion
State
A - Active
P - Partially stabilised
S - Stabilised
NC - not collected
NA - not applicable
Pick lists control the quality of data and eliminate
transcription error.
10. Data modules in the tablet pushed to a cloud based staging
warehouse then ingested into the Postgres Database
Site Description
Veg Vouchering
Genetic Vouchering
Point intercept
Basal Wedge
Structural Summary
Uploaded
Staging
Warehouse
Checks for
anomalies
ingested
Site description
Veg vouchering
Genetic Vouchering
Point intercept
Basal wedge
Structural summary
Database
SITE SAAVVP0001
11. Text and .xls files curated and added to Database
Herbarium idโs
Taxonomic name check
Soil analysis data
Soil samples assigned barcodes
GPS coordinates
uploaded & linked to site
data
Herbarium idโs
Taxonomic name check
Soil analysis data & barcodes
GPS coordinates
Site description
Veg vouchering
Genetic Vouchering
Point intercept
Basal wedge
Structural summary
Database
Site SAAVVP0001
12. Curatorial checks are made and itโs OK to Publish
Content curated
OKโed to Publish
Content curated
Herbarium idโs
Taxonomic name check
Soil analysis data
GPS coordinates
Site description
Veg vouchering
Genetic Vouchering
Point intercept
Basal wedge
Structural summary
Database
OK to Publish ๏ผ
Site SAAVVP0001
Available to
TERNโs Data Portal
TDDP (AEKOS)
R-Package
S2S
Contains all data from 660 sites
14. Photo panorama images
โข 3 x photo panoramas set in an equilateral
triangle around the plot centre pole.
โข 75 -100 images stored as .JPG and CR2
files at TERN on a managed filesystem.
15. Leaf Area Index measures plant canopy leaf area or
density
โข LAI data is downloaded to conversion
software.
โข Results stored at TERN.
โข Available for access, not yet
integrated within DB.
16.
17. Sample Management
When samples arrive back from the field
โข Vouchered plants are dried in a plant press,
information tags added to each specimen and
then they are sent as a batch to the State
Herbarium for id.
20. โข Leaf tissue samples (linked to each vouchered
plant by a corresponding barcode) are dried
and stored in Sistema boxes containing silica
and vacuum sealed.
Sample Management
21. Labelled with adhesive
barcode label and scanned
with app
silica granules dry the
sample Boxes are vacuum sealed with some silica in
them. Labelled and stored.
6. Samples can then be
used for isotope and
DNA analyses
+
22. โข Soil samples are dried in tins, assigned barcodes,
tubbed, labelled and stored at the Roseworthy
campus.
Sample Management
23.
24. โข Metagenomic samples are assigned barcodes,
their wet silica replaced with dry until all
moisture has been removed.
โข Stored at the TERN facility.
Sample Management
25.
26.
27. Samples available for researchers as of May 2019
From 691 Long Term Monitoring Sites
Herbarium Verified Plant Specimens 34,846
Herbarium Verified Leaf Tissue Samples 30,000
Field analysed Soil Pit Samples 2,912
Soil Sampled from 3 Depths 14,878
Metagenomic (genetic material) samples 6,371
28. Example : All Drosera species collected in WA available as DNA and Voucher samples
29.
30. 674+ plots 36,837
plant vouchers
7317 soil metagenomic
samples
56,272 leaf
tissue
samples
Data from
559+ plots
16,276
500 g soil
samples
1* method
44. Woodlands are commonly considered to be
ecosystems with 10-30% projected foliage
cover, and trees ranging in height from 10-30 m
45.
46.
47.
48.
49.
50.
51.
52. โa relative measure of the status of the biota at
an assessment site against a natural reference
state, described by key attributes that represent
the structure, function and composition for the
relevant type.โ
71. Up/Down App
โข Taking images of
canopy and substrate
alternately, along a path.
โข Potential for leveraging
existing Photopoints
Analysis system (+dev)
72. Camera Traps
โข Standards for Camera Traps Metadata:
โProjects and data management (TERN, CSIRO, Aust.
Museum)
โReady for trials/validation
73. Soils (in App)
โข A new App module (+ re-factoring of existing)
โข Collect characterisation, subsites, metagenomics and
bulk density
โBarcode scanning, Pictures, Drop-downs, number
entry, etc.
โข Why?
โCompleteness and correctness
โข Less curation effort
74. Multi-Tablet App (exploratory)
โข Within each field team, there are two tablets
โข Right now, they are not synchronised
โSite definitions (are not synched between tablets)
โVouchers within a site (are not synched between
tablets)
โข Potential to connect and synch tablets via WiFi-Direct
(an adhoc WiFi network)
75. IGSN (International Geo Sample Number )
โข CSIRO and ARDC
โข Designed to provide an unambiguous
globally unique persistent identifier for physical samples.
โข Facilitates the location, identification, and citation of physical
samples used in research.
โข Impetus largely from the earth science community
โ IGSN are assigned to geological and environmental samples such as
rocks, drill cores and soils, as well as related sampling features such
as sections, dredges, wells and drill holes.
78. Photopoints: Scalable Cloud Analysis
โข Current:
โAnalysis was developed on a PC
โProduction system runs on a VM in the cloud
โIsnโt fully automated, from file collection to results
โข Solution:
โMigrate to a scalable container-based approach with
automation from file upload in Dropbox, Google Drive,
etc.
79. Our Focii:
โข Less curation effort
โข Easier to access data
โข New and richer data
sources
Looking forwardโฆ
82. Remote sensing validation
โข Trade-off in data collection
โ Scale and extent
โ Accuracy and precision
โข Remote sensing collects
spectral information (EM
Radiation) about the land
surface
โ Fine scale and large extent
โ Low accuracy and precision
โข This spectral information is
interpreted/processed into
other derived products
These products
require validation!!
85. The use of TERN Surveillance Plots for
remote sensing validation
โข Provides Remote Sensing
compatible data
โข Point intercepts convert
to pixel
โ Homogeneous
โ 1 hectare
โ Aligned with grid
โข Image classification
โ Vegetation communities
โ Vegetation cover
Landsat
~ 30m
Spot
~ 10m
MODIS ~250m to ~ 1000m
TERN Surveillance Plot
1 ha (100m x 100m)
Not to scale!!
86. The use of TERN Surveillance Plots for
remote sensing validation
โข Collaboration with TERN Landscape regarding field
methods
โ Will use our data to validate many of their
products
โ LAI Meter, Cover Measures etc
โ Potential for Spectrometer/operator to join in
some of our field trips โ โTag along tourโ concept
โข Data particularly useful for validating cover
products....
87. SLATS
โข Statewide Landcover
and Trees Study
โข Queensland tree cover
mapping
โข Star Transect Protocol
โข Methodology for
quantitative measurement
of vegetation cover
โข Used to validate remote
sensing image
classification
S1
NEN5N4N3N2N1NW
W5
W4
W3
W2
W1
SW S2 S3 S4 S5 SE
E1
E2
E3
E4
E5
100 m
100 m
88. โข 1 metre samples
โข Three vegetation categories
โข Non-woody
โข Ground cover
โข Woody >2 m
SLATS
โข Site description
โข Topography
โข Vegetation structure
โข Erosion characteristics
โข Tree basal area
89. Why SLATS?
โข National standard for vegetation
cover field survey
โข Collected for different purposes
โ SLATS purpose build for remote sensing
validation โ highly refined
โ Surveillance Plots method designed for
maximum widespread utility of data,
including ecological research AND remote
sensing validation
โข Surveillance Plots can be
summarised into SLATS form
90. Unmanned Aerial Vehicles - UAV
โข Revolutionary data source
โข Remote sensing at the site scale
โข <1 m pixels
โข Spectral information
โข Multispectral
โข Hyperspectral*
โข Surface/Structural information
โข Optical 3D point cloud
โข Lidar*
*Research only
91. Spectral and structural information
โข Small pixels less likely to be mixed
(made up of a mix of materials)
โข Non-mixed pixels easier to
interpret
โข Small pixels can be โup-scaledโ or
resampled to larger pixels
UAV Landsat Modis
92. Spectral and structural information
โข Overlapping flight paths provide
stereo image
โข 3D image from numerous 2D images
โข Provides enhanced information on:
โข Structure
โข Density
โข Height
โข Spectral properties
95. The Role of Citizen
Science
katie.irvine@adelaide.edu.au
info@wildorchidwatch.org
96. 1.Quality science
2.Linked with community
3.Makes the world a better place
Dr. Alan Finkel, Australiaโs Chief Scientist,
Australian Citizen Science Association Conference
2018
Citizen Science isโฆ
107. The extent of forest in dryland biomes
Poor estimates in dryland biomes
Dryland forest cover ex.
Baobab Trees
TERN tree cover data
Very high spatial resolution and
Google tools
108. The extent of forest in dryland biomes
Increase current estimates of global forest
extent by 9% (400-500 Mha)
Plots with forest are in green
109.
110. Soil and Landscape Grid of Australia
Estimated % clay content
0-5 cm depth
5-15 cm depth
< 10 %
10 โ 20%
20 โ 30%
30 โ 35%
35 โ 45%
>45%
111. ~400 samples
Variables
โข Latitude
โข Longitude
โข Altitude
โข Soil pH
โข MAP
โข MAT (maximum)
Eucalypt
Woodlands
Chenopod
Shrublands
Acacia Woodlands
Tussock
Grasslands
Eucalypt
Open
Woodlands
Other
ShrublandsMallee
Woodlands
112. Geographic trends in metabolite composition Starting points for future discoveries!
The botanists (Emrys / Michael) search the entire plot, collecting a sample of all plant species present
Herbarium quality samples, ideally grabbing diagnostic material, flowers, seed pods, bark if necessary
Full newspaper sized
Also for every species present in the plot, one tea bag sample is collected, or for the dominant species of the plot, 4 samples are collected to enable access to within population dna
Around 10 cm of high quality leaf material, ideally fresh green stuff โ which isnโt always available
Placed into the synthetic tea bag, labelled with a barcode, scanned into the app, dried on silica in the field, and maintained until dry
Stored in air tight containers
Once collected, samples are dried back in our facility, tubed up, barcoded, scanned into our database
Stored at our Roseworthy facility
These figures are a little out of date and donโt include data/specimens/samples collected so far in 2019
Essentially, we have a lot of specimens and samples to share with researchers
Just a reminder โ there is no application process to obtain and use our data
Just a reminder โ there is no application process to obtain and use our data
Covers all these topics
Key components of the application
Agreement includes these key areas
Woodlands update
Background
One component of the Enhancing long-term surveillance monitoring programme is a project to develop field survey protocols (an โAusPlots Woodlandsโ method) that will provide a consistent national approach to fixed surveillance site monitoring of Australian woodlands, focussed on vegetation and soils. Any approach must be meaningful but still simple and affordable, and the Department also indicated that this work should build on existing approaches and capacity, and specifically identified the AusPlots Rangelands and Forests methods as existing methods to use as a basis for the Woodlands work. With this in mind, a working group from various government agencies, Universities, and private industry have provided expert input to the development of this method, recommending a trial protocol that builds on existing AusPlots methods and adds new protocols specific to woodlandsโ ecosystems to provide a cohesive woodlands survey method.
What we did
The trial woodlands method includes two entirely new protocols not used by AusPlots before, to assess trees and coarse woody debris (CWD).
The new woodlands protocols for CWD and trees were trialled at two 100 m x 100 m AusPlots, one in a black box woodland and one in a mallee woodland:
CWD was sampled along 4 x 100 m line intercepts. Any CWD that intercepted the line, and that had diameter >10 cm was measured for diameter at widest point, diameter at narrowest point (down to 10cm), length, and a decay class was assigned.
Large trees (DBH >10 cm) were surveyed in the full 1ha plot. For each large tree the species, alive/dead status, DBH, and height were recorded.
Small trees (DBH<10 cm, height>2 m) were measured in 2 x 5 m wide x 50 m long belt transects arranged in a cross at the centre of the plot. Species, alive/dead status, DBH and height were recorded.
Seedlings/saplings of each tree species were counted in the belt transects
Based on what worked well and not (see notes below), edits to the method were suggested by the field team and then trialled at another 100 m x 100 m AusPlot in black box woodland.
What worked well
The CWD method was easy to implement in the field and integrated well with existing AusPlots methods that are recommended for the woodlands protocols. This is because the CWD method aligned with existing transect lines established in the plot for use in the Point Intercept.
Marking trees and CWD with bright soluble paint once they had been measured provided an easy visual cue to avoid remeasure and assisted in navigation through the plot for the 1ha large tree survey.
Use of a vertex hypsometer to measure tree height was quick, easy, and accurate, and can be done by one person.
Edited โversion 2โ (see โSuggested editsโ below) of the tree survey worked incredibly well when trialled โ it was logical and simple to carry out in the field, and aligned well with existing AusPlots methods that will be included in the woodlands protocols.
Suggested edits
Suggested changes to the tree method that were subsequently tested in the field were:
large trees measured in 4 x 5 m wide x 100 m belt transects, aligned with the existing 30 m and 70 m transect lines laid out for the point intercept method
small trees and seedlings/saplings measured in 4 x 2.5 m wide x 100 m belt transects, aligned with the existing 30 m and 70 m transect lines laid out for the point intercept method
location of large trees, small trees, and CWD along the transect line recorded to the closest metre to give a better indication of spatial distribution in plot.
Take callipers for CWD and small trees โ quicker/easier than wrapping a tape around, and sometimes you canโt get tape around CWD on the ground.
Australian woodlands are both culturally and ecologically important to the nation, playing home to a significant portion of Australiaโs biodiversity (Lindenmayer et al. 2010). Woodlands are commonly considered to be ecosystems with 10-30% projected foliage cover, and trees ranging in height from 10-30m (Specht 1970, Yates and Hobbs 1997). Woodlands make up a significant portion of Australian ecosystems, encompassing ten of the major vegetation groups defined under the National Vegetation Information System. They are present in at least 80 IBRA bioregions, encompassing climate ranges from the monsoonal tropics of the north, through to temperate zones of southern Australia.
Woodlands habitats have been severely affected by a range of processes since European settlement. For example large areas of woodlands habitats โ especially in temperate zones - were cleared for agriculture and settlements, or heavily modified by grazing (Lindenmayer et al. 2010).
Ongoing threats to Australian woodlands include:
๏ท Clearing, result in both loss and fragmentation of habitat
๏ท Grazing
๏ท Removal of timber and debris (e.g. for firewood)
๏ท Climate change
๏ท Changes to nutrient cycles (e.g. due to agricultural fertilisers in the environment)
๏ท Invasive and exotic species
๏ท Habitat modification
Environmental monitoring in woodlands habitats has been varied. A range of state/territory-based programs have included woodlands habitats in broader monitoring programs. Other long-term research projects have investigated dynamics of specific woodlands ecosystems in defined locations (e.g. the Great Western Woodlands of WA, Cumberland Plain Woodland Restoration Plot Network in NSW, PPBio woodlands research plots at Karawatha Forest in Queensland, and Nanangroe Plantation Plot Network in NSW; see Lindenmayer et al. 2014 for an overview of some long-term woodlands research). Undertaken by universities or other research groups, these projects are designed to investigate site-specific issues or particular species and processes of interest. Thus, availability of consistent information about woodlands habitats across a wide geographic range is lacking.
Components of the AusPlots Woodlands survey method
The modules included in the AusPlots Woodlands survey method are:
1. Plot and physical descriptions
2. Plot layout and positioning
3. Photo-panoramas
4. Vegetation species survey and vouchering
5. Point intercept
6. Tree survey
7. Coarse woody debris
8. Structural summary and homogeneity
9. Soils and Landscapes
10. Soil metagenomics
11. Post field trip procedures
Note โ the manual defines what to do with โproblemโ trees with atypical trunks / trunks half within the plot/half out of plot etc.
Note โ the manual defines what to do with irregularly shaped CWD = multi-stem measurements as per pic
The AusPlots Condition Protocols were developed as part of the Enhancing Long-term Surveillance Monitoring Across Australia Programme funded by the Commonwealth Department of the Environment (through the 2014/5 National Environmental Research Program) to enhance the breadth and depth of Australiaโs terrestrial ecosystem condition monitoring and reporting at national and regional scales. This is hoped to be achieved by increasing the range and type of AusPlots field sites and monitoring, and through providing guidelines and protocols manuals that will enhance environmental data quality.
. The attributes are:
age class
clearing
cover
coarse woody debris
fauna
fire
floristic composition
groundcover
land use
landscape configuration
phenology
soil
vegetation health
vegetation structure
woody cover.
Invertebrates are the largest component of global biodiversity, play a major role in herbivory, nutrient cycling and maintaining soil structure and are an important food source for many vertebrate species
Incorporating information about invertebrates into surveillance monitoring makes sense!
For now โ so we can implement soon-ish, it will be an add-on activity for the field team to conduct whilst doing the current veg and soils methods, therefore something quick and easy โ something that could be set-up in less than 30 minutes, with pitfalls left in place for just a few hours, and then 20 minutes of active work, sampling by tree and shrub beating.
Especially ANTS as they are the dominant terrestrial invertebrate group in the Australian environment, and are by far the most commonly used invertebrate indicators in land management
Abundant, diversity and functional importance, sensitive to disturbance, so we know they will be of research interest
Baited tubes with a meat slurry attracting ants โ 20 vials placed along the ground, the slurry attracts all ant species but prone to interference by the most dominant species (researchers will be aware of this). Vials will be kept separate, not pooled together
Avoiding the hottest weather,
Shrub beating - beating 6 individual plants of the dominant shrub species (when itโs a woody species) with a standardised number of whacks, catching what falls out
Some pretty photos
It wonโt look like this on the right! But rather jars of preserved un-sorted samples
It wonโt look like this! But rather jars of preserved un-sorted samples
Comms and engagement officer for the Wild Orchid Watch project based with TERN at UoA.
I also run citizen science projects with the Department for Environment and Water, and coordinate the SA chapter of the Australian Citizen Science Association.
Citizen Science is the practice of community volunteers participating in the scientific process. Citizen scientists may be retired or student scientists or amateur naturalists who work in a completely different field. Citizen scientist are supported by professional scientists to ensure that data collected is of a high quality.
The growing field of citizen science provides us with a unique opportunity to harness community skills and passion
We can collect information at far greater temporal and spatial scales with community participation than with trained scientists alone
Education and engagement for greater scientific literacy and conservation messaging. People feel empowered when given the opportunity to participate in a project that they are passionate about
Citizen science has the potential to cover large temporal and spatial scales in ways that it is impossible for traditional science projects. While many volunteers contribute to TERN data collection and processing, there is scope exciting ways that citizen scientists might be able to get involved in TERN in the future.
Dr Alan Finkel, Australiaโs chief scientist gave an inspiring presentation at the national citizen science conference held in Adelaide last year. He described the 3 pillars of good citizen science:
Quality Science -
-Citizen science projects must be designed using robust scientific methodology and must aim to answer important scientific questions.
- We use technology such as apps and online databases to streamline information management and ensure accuracy of information collected.
-The beauty of citizen science as a method for scientific research is that it allows us to collect and share information at temporal and spatial scales that are otherwise unachievable.
Linked with community
- Participation in citizen science enriches people and communities.
-It fosters greater scientific and technological literacy and provides an avenue for people to get involved in meaningful scientific research and share their skills and knowledge.
Makes the world a better place
- ย Citizen science volunteers learn about their local conditions, and then can use the information to inspire collaborative action and influence planning and policy decisions.
-Information gathered by citizen scientists is relevant to local problems and conditions and can be re-used in research on global issues, such as climate change.
-Citizen science is a valuable tool for scientific empowerment, education, and action, and it is growing in momentum in Australia.
The variety and beauty of orchids is what draws people to his family. These are photos from the WOW Instagram account which are donated by citizen scientists. The online community of orchid enthusiasts is very active and is a great way to communicate with project participants.
The data collection is based on photos taken and submitted via the WOW app. Participants will be prompted by the app to enter additional phots and info on habitat etc to provide context and valuable details.
This is a citizen science bat monitroing project that I have been involved with for the Department for Environment and Water. Bat monitoring is part of the fauna protocols for TERN.
Social survey
Anabat recorders - image
BioCollect App for field data
Species ID from sonograms - image
Data analysis โ species distributions
PARTNERS โ very important for the success of citizen science projects, must be properly resourced. Best outcomes when government, community groups and universities work together.
State and national projects. Some have apps and some submissions are via website. A lot of choice!
The first example I want to talk about is a research project that is a a great example of how to apply PI data, which in this case was used to look habitat variation in the rangelands.
The Australian rangelands, is a exceptionally diverse are home to a wide range of species and habitats. But much of this diversity is not well categorised, which is a problem for conservation or research.
Hence the goal of this study, was, using PI data, to develop a habitat classification scheme. The authors clustered or grouped plots based on the species diversity and structural traits of each plot to identify similar habitat types, which could be the basis of a classification scheme. In additional, they correlate the plot structure to \environmental variables to understand the abiotic context for variation in rangeland vegetation. These variables would included things like MAP and MAT, and soil attributions, which were gathered for the slga, like soil ph, and total N and P.
After quantifying species diversity and structure in roughly 400 plots, the team initially identified 3 โsuperclustersโ or groups of plots, which were primary distinguished by latitude and associated climate variables. These three groups were Mediterranean, savannah, and desert. And as you can see from this multivariate analysis, where each point is plot, these three groups cluster together fairly well, indicating this a logical initial classification.
Within each of these superclusters, the authors identified smaller clusters based on differences in species and growth form diversity. This figure showing the number of different clusters that were identified, and the area of the chart slices represent the proportional contribution of the importance of each growth form. potential categories within a classification scheme. Our results offer a tentative classification scheme that is novel, ecologically sound and coherent in terms of floristic composition and structural attributes.
Been introduced this week to our photo-panoramas.
Not going to discuss that in detail
But what it is important to known when working with these photo-panoramas, is are they accurate?
Can photo panoramas faithfully record structural attributes of a plot
In order to answer that question, we compared structures attributes including stem count, diameter at breast height (DBH), and basal area as determined by field ground observations, PP, and high cost survey grade Terrestrial Laser Scanner (TLS) RIEGL VZ-400. Assuming field observations were correct
Stem Count
Assuming that the ground obs are the best most accurate value we can achieve
Neither method did a great job
Blockage caused by stems in the foreground
In both cases these methods need to be refined
Mean DBH
Pretty similar for all three
Basal Area
much better with photopnaorams
Good tool for photopanaorams, good alternative ground observations
And considering that PP is cheaper ( it takes a little longer, but it far most cost effective).
Switching gears a bit now, because the first two examples are a bit more localised
But our plot data is not just used for research studies, it also contributes to large global survey programs
For those of you who are not familiar, GEDI (Global Ecosystems Dynamics Investigation), a High resolution laser which takes globally representative measurements of vertical structure in the worldโs forests. It located on International Space Station (ISS) These measurements are used to quantify aboveground biomass density (AGBD) at the scale approximately a 25 m diameter circle on.
TERN ES supplies GEDI with basal wedge data from our plots, which in turn can be used to calculate biomass, which is then used to validate the GEDI imagery.
But recently, an example of our data being used that we are particularly proud of is TERN Plot data, which was used in a global analysis of the distribution of forests and woodlands in dryland areas,
dryland forest extent and cover, in the past, has not been accurately measured, by the scarcity of large-scale studies of dryland biomes
But, as we have discussed formally and informal this week, dryland ecosystems contain great deal of biodiversity, in fact the first paper I discussed really drives that point home, with all the different clusters
And that includes, in some areas forests
And so understanding their extent has substantial implications for understand this like carbon flux and storage capacity
To accurately determine how much forest and tree cover remains in dryland biomes, a large team associated with the Food and Agriculture Organization of the United Nations established a global initiative to undertake a Global Drylands Assessment of forest. high spatial resolution imagery and Google geospatial tools to determine tree cover in dry land biomes all around the globe. They did this by targeting a large sample of 0.5-ha plots.
So you might be wondering what TERN ES have to do with this.
Well, similar to GEDI, TERN was able to provide our tree cover data for 100s of plots across the Australian ranges land, which in turn were used to help validate remote sensing data on this project
TERN staff were also heavily involved in that process.
TERN data for crucial on-ground verification of satellite-based analyses in Oceania
So in the end, what did they find?
- โfoundโ 467 million hectares of previously unreported forest around the world
The discovery increases the known amount of global forest cover by around 9%, and will significantly boost estimates of how much carbon is stored in plants worldwide
Global drylands contain 45% more forest than has been found in previous surveys
Also by revealing that drylands a greater capacity to support trees and forest than previously perceived and understood, a unique chance to mitigate climate change impacts through large-scale dryland conservation and afforestation actions is presented
This has particular relevance for Australia considering much of our country is arid semiarid land.
Look at the applications of our samples
Continuing the theme of ES making contributions to much larger programs. Our samples, similar to plot data, also help to validate and develop models of soil and landscape attributers.
We have not discussed this much this week, but some of you will be familiar with the slga, this platform also includes data and mapping products of roughly 30 different attributed, modelled for the entire country. It provides easy access to nationally-consistent and comprehensive soil and landscape attribute data at a finer resolution than ever before in Australia, particularly now that it has its own r package.
But similar to gedi, even the greatest national data sets, need a ground support. In this case ES soil samples are scanned, and the result data for a wide range of attributions, such as estimate clay content, and used to ultimate build and validate slga models. TERN samples are critical to this landscape product landscape product.
But our soil samples have been use for a number of different research projects, not just validation, but for new research as well. And a great example of thisis a study that use our samples to help create new and improve drugs derived from secondary metabolites. Secondary metabolites are organic compounds produced by bacteria that are not directly involved in growth, development, or reproduction but help them interact with their environment. These include things like toxins and signalling agents that allow bacteria to obtain nutrients as well as defend against, and communicate with, other organisms. But these metabolites can also be extremely valuable from an economic or more anthropocentric perspective because these natural products can have pharmaceutical applications. In fact bacterial bioactive metabolites have been one of the most prolific sources of small-molecule therapeutics, things like antibiotics, antitumor drugs and immunosuppressive drugs. So there is an enormous incentive to identify areas or environments that may be significant sources for these metabolites.
However, that is a big gap in our understanding of the best way to find key sites to look for the metabolites. We know that environmental factors contribute to differences in the composition of natural products, but there is a limited understanding of how bacterial biosynthetic diversity varies from one environment to the next, and what environmental factors correlate with changes in biosynthetic diversity.
TERN soil samples provide a unique opportunity to investigate the relationship between metabolite and bacterial diversity and the environment, Using TERN Soil samples, a team from The Rockefeller University (scientific research, primarily in the biological and medical sciences, that provides doctoral and postdoctoral education. Rockefeller is the oldest biomedical research institute in the United States.)
Used ~400 top soil samples from ecologically and geographically diverse plots across Australia including samples from the major vegetation groups we often survey, For each sample, eDNA (environmental DNA) was extracted from the soil to look for gene clusters associated secondary metabolite production. They then compared eDNA diversity and composition at each site to the latitude, longitude, altitude, soil pH, MAP and MAMT.
Didnโt detect any correlations with these factors and diversity, changes in latitude and to a lesser extent changes in pH were found to correlate most consistently with changes in composition on a continent-wide scale. Although further studies are needed to better understand the underlying causes driving this relationship, these findings provide insights into how best to direct future natural product drug discovery efforts.
They were also able to identify the sites that had key metabolies often used in current drug therapies, so this is a good clue to look in these areas for future , and thus indentify key hotspots for drug discovery. These data can be useful for identification of environments that are potentially productive starting points in the discovery of novel members of these families of biomedically relevant NPs.ย
Moving on to plant samples
In fact, just this past year, one of the veg voucher samples we collected turned about to be species never before recorded in Austrlaia.
Working with plant samples collected duringย TERNโs ecosystem survey of Australiaโs alpine region, Neville Walsh of Royal Botanic Gardens Victoria identified Australiaโs newest weed species: Golden Clover (Trifolium aureum). Native to eastern and central Europe, Golden Clover is not known to occur elsewhere in Australia except where it was found by TERN on Victoriaโs Dargo High Plains south-west of Mt Hotham in January 2018.
Similarities to other clover species could have led to Golden Clover being overlooked in the past. The discovery is the result of the systematic monitoring of all species at TERNโs monitoring sites, followed up with retention and formal identification of voucher specimens.
The identification of this new species reinforces the value of theย TERN AusPlots methodologyย for the surveillance of Australiaโs ecosystems
Itโs not known how common Golden Clover is in the Australian alps, but surveys at the TERN ecosystem surveillance alpine plots this summer by Royal Botanic Gardens Victoria staff are expected to determine if the population is large enough to be of concern to native vegetation. Future repeat TERN surveys will also detect any significant increases in frequency or abundance of the species, a finding that could trigger more targeted monitoring or management.
Now one of the main things that samples are collected for with plants is N and C, and isotope analysis. And this next example really shows the potential applications of the plant samples in this regard.
Nitrogen (N) is an essential nutrient for primary production and plant growth. Nitrogen content per unit leaf area (Narea) is a key variable in plant functional ecology. is important for making perditions about growth, productivity and resilience.
In this study, the authors set out to test the predictability of Narea using measurements carried out on dried plant material collected by the Terrestrial Ecosystem Research Network (TERN) ES. Tested and quantified the effects of irradiance, ci V ca ratio (from 13C), temperature, LMA, and N-fixation ability (26% of the species sampled were Nfixers) on variation in Narea.
Analysis used 442 leaf measurements representing all species found in a 100m100m plot at each of 27 sites on a broad northโsouth transect across Australia.
Canโt go through all aspects of their findings, but in brief, regression analysis revealed significant partial relationships Narea vs.
ci V ca, MAT, and ln IL (Table 1, Fig. 2).
The relationship was also negative for MAT, as expected, because there is an inverse relationship between temperature and the quantity of leaf proteins
Our application of trait gradient analysis also points out a way towards process-based treatments of functional trait diversity in next-generation models
And finally, I wanted to highlight a good example of where plot and sample data can be used together, in a project that as I said early, I am not sure when samples were getting collected that people were thinking paleoclimate applications, but creative people saw the potential.
This example actually comes right out of the Uni Adelaide from a phd thesis which was interested in using microfossils in soils to study past environments.
Microfossils, are exactly what they the sound like, (they are fossils you can only see with a microscope). Organic microfossils you might find in soils includeย pollen or spores. Scientists can make interpretations about past vegetation composition and environmental change by extracting microfossils because we can assume these trends and relationships are relatively consistent between the past and present. If we know what factors control modern day pollen composition or distribution, we can assume that the processes control pollen composition and distribution in the past.
Leaf waxย n-alkanes are one type of compound made by plants that can be preserved in soils and sediments for millions of years. It is a microfossil. But it was unclear how variation in n-alkane composition, structure, and concentrations can be interpreted relative to vegetation structure
ย
So the authors took soil samples from about TERN from about 20 sites extracted theย n-alkanes and compared them to variables such grass and tree cover at TERN plots to see if alkanes composition changes predicably with either.
So what did they find,
Well Norm33 is a specific group of n-alkanes, they have a certain composition, and the n-alkane values from TERN soils appear to correlate with the general vegetation found over a broad area. This finding suggests that these molecular fossils provide a faithful record of large-scale ecosystem structure such as the predominance of grasses versus trees, which can be very useful in reconstructing past vegetation change.โ
So in ancient soils if see high Norm33 values, this could be used to indicate the area used to have a lot grass cover, rather than trees.