Skip to main content
Loading Explore guides…
Biodiversity Data to Actionable Insight · Acedra
Home
Explore
Biology
Biodiversity Data to Actionable Insight
Sign in to save
Chat
If YouTube asks you to sign in, open the video on YouTube.
Watch on YouTube
Reload player
Transcript
71 segments
Copy
Hide video
0:17
Welcome and thank you for joining me. I think the topic of of my talk today is quite timely and urgent. We have to move from fragmented inaccessible biodiversity data more towards a system of actionable insight and high integrity information that helps us meet the growing demand of nature finance, corporate sustainability and also global biodiversity targets simultaneously. Why does nature matter? First of all, we care for biodiversity many reasons but also their
0:50
international commitments. One of the major one that is being discussed at the moment is the 30 by 30, where governments have committed to conserving 30% of their lands on sea by 2030. I think the latest figure I read today in the UK is that effectively we only conserve 13% at the moment. So there's a long way still to go. Then we have increasing recognition of biodiversity dependencies and risks for continued business operations along supply and value chains. A recent
1:25
analysis by one of our colleagues, Nicola Ranger, found that there's really strong microciticality of nature related risk including a 12% risk of agricultural output related to pollinator decline. I'm just going to steal the words by the European Central Bank board member Frank Elderson who said that nature is not just a domain of scientists. It matters to central bankers, financial supervisors,
1:52
entrepreneurs, workers and policy makers. Why does it matter to all of us? Well, it's to meet expectations and requirements. So, for corporates it's meet their stakeholder expectations or also adhere to regulatory and compliance requirements. For example, under the Taskforce on Nature-related Financial Disclosures or the EU CSRD directive. Then it's to meet targets and progress against those kind of standards, like the science-based target network framework or in line with the
2:26
global biodiversity framework. Increasingly also to address the needs of biodiversity credits either with involuntary or regulatory frameworks, for example the biodiversity net gain in England or for value chain monitoring, such as corporate risk and projection analysis along value chains. But do we currently live in a world of data desert or data deluge? it's not actually such an easy question to answer because every second a lot of biodiversity data are generated through all
2:59
kind of mechanisms, through site surveys by nature consultants to adhere to some legal requirements, for example prior to developments for corporate prospecting purposes. For example, the mining industry collects loads of data regularly - really high resolution, high value data. NGOs are monitoring key species also in remote areas of the world, academics gathering nature data, citizen scientists recording sightings. So, a huge amount of data being generated all the time, and
3:31
some recent assessments found that, for example, for each ecological assessment conducted before some kind of developmental project on average about 500 to 1,000 species occurrence records are produced. And if we look at for example the World Bank, the French Development bank, the KFW group, the European Asian and African Development Banks. A recent estimate found that they collectively collect up to 300,000 occurrence records each year. So that's not just valuable
4:09
in terms of the data, but it's also really costly to collect it. But despite all of that data being generated increasingly there's a call for more data availability better data availability and when I've been to different meetings and different fora again and again that kind of call for more data and issue of having lack of data or data insight keeps coming up. So I'm just going to go into that kind of data gap for a little bit. Now those of you who are not ecologists are maybe
4:42
not familiar with this concept but we generally talk about the seven shortfalls of large scale biodiversity data and they address those seven concepts. I'm not going to go into too much detail but just to say that we are very aware that we lack data on different concepts. For example, understanding organisms responses to environmental changes, understanding u morphology and function of organisms, having information on species interactions or also challenges of sampling
5:14
large and and uncertain fogynies. All of those are common across different fields and they also then apply to different conservation applications including planning for 30 by 30 tipping points, ecosystem resilience and so on. So all of them are really highly relevant when we think of biodiversity data for biodiversity finance and and corporate reporting standards. Before I go any further, just to say that also in terms of terminology when we talk about detecting or
5:45
assessing biodiversity change that word detection is actually really important and in that context it refers to the identification of directional change in the system of interest. In biodiversity science, we talk about detection when we talk about estimating of generally annual or long-term trends and biodiversity or biodiversity metrics and also to assess the degree of support relative to historical baselines, counterfactual states or long trend scenarios. Usually we also focus
6:17
on the estimate of magnitude and rate of change rather than binary classification. So change or no change. And there's a whole literature just on how we can do detection in biodiversity science. So how can we do this detection? What kind of data do we need? How does the ideal data look like? We'll need a whole seminar, I think another seminar maybe to talk about the scientific basis for
6:41
selecting different measures of biodiversity for the purpose of reflecting impact or dependencies. So I'm not going to go into any detail on that now just to mention that many of them go beyond simple measures of for example habitat or ecosystem state indicators also considering for example the maintenance of ecological and evolutionary potential but generally to reflect bodiversity impact risk or dependency there are a lot of metrics available many theoretical approaches
7:12
as well generally they address either or all of specific indicator of species, genetic species or population, ideally abundance information, sometimes also on characteristics of species, so-called traits and or ecosystem health or state indicators. Ideally, they should fulfill biodiversity credit or or nature market assessment requirements. they should be scalable, accessible, granular and evidenceable. In other words, they should not be limited in availability or
7:45
efficiency for global use. They should be suitable for various project spatial extents. They should not be limited to data collection by experts or only available at high costs. They should be available at sufficient spatial, temporal and taxonomic resolution. And they should be evidenceable. So readily giving evidence of an organism's presence and allowing third party verification and auditing. Additionality and leakage assessment is also really important.
8:14
So generally there should be data on and off site ideally some kind of baseline information across the different range of indicators as well. And then also ideally they should fulfill the modeling requirements. So it should be of high spatial temporal structure and coverage which means that ideally the same kind of methodologies should have been used in collecting the data and ideally
8:40
they should also cover a broad range of taxa and a broad range of ecosystems and habitats. The data should be location specific again covering a range of different taxa and also the appropriate amount of time which is challenging depending on what what kind of data and indicators you're looking at. So there are a lot of requirements for the kind of ideal data set for detection and on top of that if you think of nature finance they should also be cost effective. Ideally they
9:13
should also adhere to policy recommendations. So, for example, for biodiversity credits we have the recently published high level principles to guide the market but there are also other standards, for example, around ecological restoration and so on. Ideally they should do all of that along the nature value data change. So it's not just about the kind of about the collection of the data and the storage of the data. It's also around the publication the uptake connecting influencing
9:45
and then the use reuse and storage of the data. How does the state of nature data look like at the moment? If we just think of the requirements that I mentioned to fulfill the detection modeling requests. here is just an outline of some of the common biodiversity databases. they're all global databases where we did an assessment just to look at the different structural coverage and quantity requirements. So ideally we would want to have all of those going high on all of the
10:20
scales. But as you can see that's of course not the case. So we face trade-offs where either you have very high sampling and spatial structure but then you have generally low data quality, low feature coverage and low spatial coverage or vice versa that applies to all data bases that we have available at the moment. So there's a choice to be made what is important in a in a certain application and choosing the kind of data ading to standards. Again, just a a quick example here
10:51
looking at biodiversity credits where we compared 11 major biodiversity credits and certificate suppliers and we scored them according to whether they meet the high credibility standards that are published by the international advisor panel for biodiversity credits economic foam and BCA. And so where they score zero that means that none of the indicators were met. where they score three, it
11:17
means that they meet them at an advanced stage. So you can see the progressive level of fulfillment here and the horizontal dotted line in red shows the average value across all six criteria and you can see that none of them scores three. So there's room for improvement and some of the criteria are not met very well yet. So for example, criterion three which is is around verification of biodiversity positive outcomes by independent and transparent means. Okay. So why do we have
11:52
these data and knowledge gaps? Firstly there's no systematic biodiversity monitoring across realms and countries. Again as I said earlier ideally we want to have highly structured data. So the same kind of monitoring approaches the same methodologies and that's just not possible and and non-existent at the moment. There's some political geopolitical issues. So for example the global biodiversity information facility which stores and collects vast amounts of species distribution
12:23
data. They face huge issues to mobilize data from certain parts of the world. Then we have limited capacity and the high cost of storing and managing those vast amount of dates are really prohibitive. A lot of initiatives were originally initiated by academia and as I think you appreciate hopefully we are usually cash strapped. So it's very difficult to get funding for continuous management of biodiversity databases. And I know some of the databases I've personally been
12:54
involved in that they were literally run by two people who were just really committed and put a lot of their personal time into containing and and maintaining those databases. Then there's still limited cross-sectional collaboration. I mentioned some of the data that is collected by different actors in this in this field and generally there's just not a lot of collaboration or sharing going
13:18
on. Although I will mention some positive leaders and examples later on. Then there are some privacy and legal requirements. For example, for mining corporates, it's actually very difficult to share their data because of competition laws. So even if they want to, they can't easily provide the data that they collect to other actors. Then there are also some concerns potentially around threatened
13:43
species that you don't really want to publish the exact locations and again there are solutions around that but just a lot of thinking to be done before data can be shared and a lot of mechanisms that have to be developed to enable that. And then we have some historical and legacy issues where again some of the databases were developed on quite low budgets and so it's very difficult to turn them into larger databases that are more suitable for the technology that we have nowadays
14:15
and for the huge amount of data that we see coming in. So I just want to pick out two of those aspects. The limited cross- sectoral collaboration and then the privacy and legal requirements issues. So as I mentioned before data does exist but it's often not available. For example, data collected at the government level, military, there's a lot of biodiversity data collected globally that's also not made publicly available. Industry national parks and so on. There's also
14:46
issues around long-term storage of all data. So for example, I've talked to a lot of biodiversity and nature consultancies also one of the largest one across Europe and they say that they don't store their raw bio acoustics or or edna data because it's just too much data too expensive to store. they have to provide the consolidated information that they draw from the data like for example species lists to their clients but then once that is done they delete the raw data and so
15:20
that's problematic because we continuously develop the methodological capabilities to make sense of the data and throw out a lot of information content. So losing that raw data can be really problematic in like 5 10 years time. And then maybe most importantly there's also still limited adherence to data and reporting principles. So we have the fair and the care data principles. Just briefly in case you're not familiar with those fair refers to being findable,
15:52
accessible, interoperable and reusable. So ideally all data should adhere to those principles which would then make it either now or in future possible to reuse the data by different actors and through different platforms. And then the care principles refer more to the rights of indigenous peoples and local communities where they highlight importance that data of collective benefits that they are collected and used responsibly that the indigenous peoples and local communities
16:28
remain as some kind of authority to control and that the ethics are considered at all stages. they often the data collected does not actually adhere to those principles. saving files is also very difficult. So you have to make sure that they also still be reusable in 10 years time. And there are some shocking examples of data being stored in formats that just made them unusable within a few
16:57
years or decades also on on governmental levels. And then there's often a lack of proper taxonomic naming or taxonomic name checking. Again, that's something that that I've heard repeatedly in different for on different platforms, which again makes it difficult to reuse and combine data. Okay. So, so a lot of negativity here. So, I want to move on more to to the solutions or at least thinking about some of of the solutions that have been proposed. First of all, looking
17:28
at the raw nature data gap, technology is often proposed as a potential solution to a lot of of our issues and indeed technology is amazing. We've had so much progress in the in the last decade and it's really progressing very very rapidly. So there are a lot of means to measure nature either directly or or indirectly now including full remote sensing for example through EDNA or meter barcoding bio acoustics and eco acoustics. Then we have camera trapping. We also have
18:05
structure promotion and proof systems and then proximal and satellite remote sensing as well. But there are still again tradeoffs. so there's no one solution to fit them all at the moment. So here's just an example and apologies it's it's a little bit small but I'll just talk you through some some aspects here. So you you see here an assessment that we did of whether they adhere to the requirements I mentioned earlier for nature finance market ready data. so they should be
18:40
scalable, accessible, granular, evidencable and direct. And wherever they are read we say that they generally adhere to those principles and where they are light brown they don't. So you can see with satellite remote sensing for example we say that this technology is highly scalable. You can use it across different spatial extents and large spatial coverage. Temporal coverage can still be an issue because a lot of the high resolution data don't go back that many years but
19:10
that's also improving. Cost varies widely. So if you want a really high resolution data maybe also more complex data such as hyperspectral then that is still very expensive and you generally need quite a lot of expertise to use it well. Then there are also issues around weather conditions. For example, if you have a cloud cover, then it can be very difficult to get any information apart
19:35
from sensors like like radar. Then if you look at camera trapping, it's it kind of looks a bit the other way around. So they are not as scalable because obviously you need to have a a person go out put up the camera traps and there's a lot of manpower involved in maintaining them as well. the cost is generally lower and they are more evidencable and more direct because you have the physical image of an animal that you can use to evidence your claims. Then if you look at
20:09
meter barcoding which is also a technology that has seen rapid improvements recently maybe not so scalable yet and also some concerns around cost contamination that we can't really know the optimal sampling distances yet so you might detect edna in your sample but it's difficult to trace that back to the personal differences and and origins of the DNA because DNA is sample DNA travels through space and time and again getting information on abundance needs a lot more work as
20:44
well. So very promising but still those trade-offs exist. What about robotics and automated systems that's another kind of technological advancement that is often being seen as as really revolutionizing by diversity assessment. also hear the ideas that ideally in kind of the bright future, we don't need people to go out and sample anymore. We can have automated systems and networks of sensors to measure things and provide us data in real time. In this analysis, we looked
21:18
at interviewing different experts on their views of the key challenges and and key promises of robotic and automated systems. those experts are from across the the globe. So we had usual kind of gaps in parts of Africa but generally we had input from across the world and also from experts across different taxonomic groups and different kind of sector applications and the key opportunities and challenges I identified. I just want to highlight a few. So the most obvious
21:55
opportunity is serving it across large spatial scales in real time. So again that would address two of the key requirements I mentioned earlier for biodiversity data for detection. It would allow us to survey over longer temporal periods as well and potentially also remote areas which are far from infrastructure. Then some of the challenges power availability. So at the moment all of those robotics or sensors still require power supply and ideally also sensor network
22:28
access so that the data can be stored in the cloud and accessed remotely then there's still a risk of misidentification and serving rare elusive species is still very difficult. And then we we looked at where we should focus technological developments in future. So the color shading here indicates the main areas in which technological developments are needed and could bring kind of the the largest
22:58
benefits to enable rest to perform each monitoring task. So ideally we have the scope here required and then the different aspects that all need to be fulfilled in order for this system to work well. so ideally you would want to have each of those lines gray which means that we have field tested technology that can be applied anywhere in the world at least theoretically. Green means that we have the technology but there are still substantial limitations. So that could be for
23:31
example around access to power or access to some kind of internet or data uploading system. Then in dark green we have working prototypes but they have not really been field tested well. Pink means that the technology is still in the research and developmental phase and then blue we have major technological breakthroughs that are required. so we see potential there but it will still need a lot of developmental work and you can see it doesn't actually look too bad. So there are a
24:03
lot of applications here where we do have a lot of the aspects already covered and we just need kind of a small step towards developing those as a routine tools for by diversity monitoring. So what about the nature data and sharing and and access gap? I just want to highlight a few examples here which I think are quite exciting. So one is the data for nature program where basically it addresses some of that nature data that I mentioned earlier that is being collected
24:38
by other actors but not shared particularly around public development banks which play a surprisingly important role in biodiversity conservation. And so through the projects that they fund, they help to generate really valuable biodiversity data, often really detailed on kind of species level abundance data as well. And that information of course is really important and can be a real asset for other studies. Often those reports and studies funded by the banks are confidential and
25:09
limited, but the raw data is not. So, GBIF has developed this framework that allows basically development banks and their contractors to upload their raw biodiversity data and make it accessible to others as well. So, it encourages banks and their partners to share their biodiversity data and it made it as easy and as cost-ffective as possible to clean and share the data. They have developed also really good guidelines on how to do that and and all kind of different major
25:42
languages and have a great pipeline that allows non-experts to get the data in a format that is adhering to the fair and care data standards. Another example that has just been published is around the unified global soil biodiversity observatory. So when you think of biodiversity data, we often don't really think about soil. Even though remarkably it holds more than half of the world's biodiversity and supports really critical ecosystem services including purifying water
26:13
substrate of course for for growing plants and our food, cycling nutrients, storing carbon and so on. So really important to think about soil as well. But so far there have been quite little efforts to protect soil biodiversity and share data on soil biodiversity. so this initiative aims to collect data and create standardized indicators to help countries globally share the data and strengthen
26:41
monitoring systems as well. So the goal really is to enable evidence-based decision making and and feeding into global assessments and making sure that measures and indicators of soil by diversity are considered in nature assessment frameworks as well. Moving on to addressing the nature data insight gap. So if you've solved access to data and we've solved sharing of the data, you still
27:08
don't have decision ready information. so making sense of the data is is another big challenge and again there are a lot of really great initiatives and I just have time to go into some of them now. One is the SEAF the shared environmental analytics facility which has been developed by the Western Australian Bodiversity Science Institute and the Western Australian Marine Science Institution.
27:33
And here again the idea is that you have a lot of different users of data and different contributors of data who often don't have the expertise needed to collect clean data and then provide information ready outputs from that data. So including calculating indicators and so on. And so the idea behind that facility is that science knowledge and data analytics can be provided to streamline decision making and highlight major gaps and and data and information content. So again
28:09
that facility is originally focused on Australian actors but the framework is openly available, and they're really happy to share their learnings and the insight they gained. I think it's also really great because it considers terrestrial and marine systems when often we actually find that marine or aquatic systems in general are not considered explicitly. Another initiative which I think is is really great is the BON in a box which is developed by GEOBON, the Group on Earth
28:42
Observations Biodiversity Observatory Network - quite a mouthful -- and it's a really powerful open access tool. It was created initially to help tackle one of the major challenges in biodiversity monitoring to collect all that raw data we have and then turn it into meaningful indicators for decision-making. What BON in a box does, it's a very busy figure here – I'm aware of that. So I just tease it out a little bit. BON in a box provides everything that is here in
29:16
the green box. So it uses data from biodiversity observation networks and from all kind of major biodiversity data providers and then provides access to a pipeline engine. So it allows users to build their own indicators and turn all that biodiversity data into the indicators that they need. For example, essential biodiversity variables or indicators of biodiversity change or metrics that are used for the global biodiversity framework evidencing and so on. And
29:53
those pipelines are really customizable. They are community-built and they can be run either locally on your own machine or remotely in the cloud. Basically, it allows also non-experts to some extent to access this data and make most of the data available. And so what about AI and all of that? That's something that comes up a lot in discussions as well. There's a really nice recent analysis of the potential of using AI in in advancing biodiversity science and
30:27
assessments. So far most of the attention around AI for biodiversity has been around wildlife tracking and monitoring, also some analysis side of things, so especially around detection and the attribution of changes to different drivers, but we've seen some really exciting progress in using AI also to, for example, build phylogenetic trees or improve species distribution models. But there's a lot of untapped potential of AI that could potentially help us tackle bigger ecological
30:57
questions, especially the ones that involve pulling together really diverse and complex types of data that have been collected for different means. For example, combining citizen science data with remote sense and data provided by development banks, and so on. AI can be really useful in helping combine information and insight from images, videos, audio recording, DNA sequences and and structured text, and so on. Those are the kind of applications where AI can help us reframe or
31:30
even rethink the way we approach ecological data and research. And in that paper they show specifically how AI can help address some of those seven shortfalls in biodiversity data science that I mentioned earlier. But they also highlight very rightly that AI tools are not a panacea. So they have to be used effectively by experts and require really careful integration into by the scientific
31:58
knowledge and decision making. So it's not a plug and play solution at all and there's a lot of risk in using AI as a kind of quick win solution to some of those challenges. Okay. So to conclude, are we living in a data desert or data deluge or is it something else that we should really think about? Kind of all of it together, I would say. So the importance of location specific data comes up again and again. So we definitely need more data through time and space. We have to keep
32:33
in mind and try to address those seven shortfalls in biodiversity data that I mentioned a few times and also put a lot of effort in improving data access and data sharing. Again, I mentioned a few initiatives that have worked really hard to make that possible, but it's definitely some space there to improve that and and also make sure that those initiatives are taken up more broadly. TNFD recently said that the lack of understanding of how the information derived
33:04
from data can be used by decision makers is more of a challenge than the shortage of data. So they argue that the data desert is not the issue, it's more getting information out of the data, and again that's something that I've also seen a lot in discussions especially with banks and insuranceances. So, again, that's something where a lot of tools exists and where collaboration of the scientists who use and develop those tools is really key in making progress, and then just to
33:32
point out that the challenge is not just getting and distributing the data but storing it as well. So I mentioned all that raw data on bio acoustics and EDNA for example there are huge amounts of data that require really good infrastructure, database management and maintenance, but they also, importantly, have a really large carbon footprint. So we have to think about which data is really worth saving and which format we can save them in to reduce the carbon footprint
34:02
as much as possible. And then just as a kind reminder, it's a huge list here, but nature is important for so many things and I think it's really important to just sometimes take a step back and remember why we work on this and why we have to improve some of the things that are done at the moment. Biodiversity underpins everything, from clean air and water to soil health pollination, climate regulation and so on. So it's really the infrastructure of our economy
34:32
and survival. Nature loss destabilizes global economy. Again, there are loads of studies that have shown how important nature is for the economy. So, even if you don't care about nature by itself, financially, it's really important to consider that the losses in nature are accelerating. Huge declines in global wildlife populations, we are breaching planetary boundaries, especially the one around the biosphere is already severely transgressed.
35:00
We have some ethical responsibility as well. So nature should be seen as an intrinsic value as well beyond of the kind of sheer utility it has for either our businesses or ecosystem services it provides because we have responsibility to other species and future generations. Restoring nature is really cost effective. Difficult to put numbers on there, but just one assessment said that for
35:26
every dollar invested in ecosystem restoration, we yield $30 in economic benefits. Again, those numbers could be debated, but it's very clear that there is definitely a cost effectiveness in protecting nature. Now, and delaying action is not neutral, it's actually destructive. So, that's really important to keep in mind as well. The longer we wait, the more costly, complex and incomplete solutions will become. What should we focus on? Data management and sovereignty are really key
35:57
aspects, and it's good not to underestimate the kind of efforts that have to go into doing this. Well, we need to always uphold fair and care data standards within all frameworks of data being collected and shared for nature finance as well. Solutions do exist but they either need wider applications, further development or more support. Cross-disciplinary and cross- sector learning and collaborations are really essential and there's so much potential still out there to
36:28
to improve that and make really large leaps in some of those fermenting challenges. Technology is not yet the solution to all of our issues, but we know where to focus efforts and we can see that the the potential is really enormous to help advance biodiversity nature collection for finance as well. So, thank you very much for your attention. I would also like to thank my many collaborators across the world who have informed a lot of this thinking and who I've been
36:59
working with a lot on trying to find solutions to some of those issues as well. Thank you.
Here to help you learn
Sign in to keep a private conversation with your course tutor.
Ask me anything about the material…
Sign in to chat