The path to scientifically sound biodiversity valuation in the context of the Global Biodiversity Framework

The 16th Conference of the Parties (COP16) for the Convention on Biological Diversity will be held in the fall of 2024 and is expected to focus on measurement, monitoring, and valuation of biodiversity. There is a need to coalesce around a coherent system of biodiversity measurement that can support valuation. Numerous biodiversity indices exist, but these generally do not support unique and policy-meaningful valuation. Most of the components of a science-based system of measurement exist. Here, we discuss how to think about the criteria for a measurement system and how to organize the existing pieces by framing biodiversity as a specific summary statistic of a subset of natural capital accounts.

Abstract

Successful implementation of the Kunming-Montreal Global Biodiversity Framework requires identifying a process for measuring and valuing changes in biodiversity that build on the recognition that economics and valuation must play a key role in “halting and reversing” biodiversity loss. Here, we discuss considerations for a practical path to valuing changes in biodiversity. Framing changes in the value of biodiversity as a summary of changes in certain natural assets enables leveraging existing approaches and international standards associated with environmental-economic accounting. We discuss why an approach that builds from individual species, evolutionary groups, or functional groups into a practical, hierarchical statistical classification system is better than the development of any one biodiversity index. We merge techniques from ecology and other natural sciences, national and environmental-economic accounting, and economics, which are all on the cusp of making measurement of the change in the value of biodiversity possible. The focus should be on scaling and integrating these approaches. The path forward appears to begin with imperfect but useful measures, grounded in robust concepts, while establishing ambition to further scale-up measurements—just like the past evolution of many other official statistical series.

The Convention on Biological Diversity (CBD) (1) Kunming-Montreal Global Biodiversity Framework (GBF) recognizes that economics must play a key role in “halting and reversing” the loss of global biodiversity by 2030. Building on the 2011 to 2020 global biodiversity goals, known as the Aichi Targets, the GBF repeatedly calls out the importance of valuing and accounting for biodiversity as a critical tool for halting and reversing biodiversity loss. GBF target 14 specifically calls for integrating economic valuation of biodiversity into policymaking throughout all levels of government and across sectors. Despite this recognition—as well as the landmark Dasgupta Review on the Economics of Biodiversity (2), the rapid development of biodiversity disclosure frameworks such as The Taskforce on Nature-related Financial Disclosure (TNFD), and the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services (IPBES) Assessment on the Diverse Values and Valuation of Nature (3) — environmental policy and governance does not seem anywhere close to employing a standard protocol for measuring, much less valuing, biodiversity. Indeed, in the final Aichi report, 110 out of 192 countries made no attempt to value changes in biodiversity (4). As the GBF moves from framework to implementation at the 16th Conference of the Parties (COP16) in the fall of 2024, the need to identify an appropriate means for valuing biodiversity will be critical for the successful achievement of the targets.

Measuring and valuing any change in biodiversity requires transitioning the measurement standard from a research concern to a policy concern. Accountability for approximately 14 of the 23 GBF 2030 targets requires measurement, valuation, or both of changes in biodiversity; and even for targets where valuation is not required, successful implementation of the GBF requires measurement approaches that would ideally integrate with valuation approaches. How such integration among measurement, monitoring, and valuation is to be accomplished was intentionally left vague in the adoption of the GBF Monitoring Framework. The issue is expected to be addressed at COP16 in 2024, when the Monitoring Framework (CBD 2022) will be reviewed and finished. This gives precious little time to coalesce around a workable, evidence-based, measurement regime that can gain acceptance far beyond the biodiversity conservation community.

As part of the approval of the monitoring framework, an Ad Hoc Technical Expert Group on Indicators was established to provide technical advice on remaining and unresolved issues relating to the post-2020 monitoring framework and to prioritize work on addressing critical gaps, particularly under Goals B, C, and D and Targets 2, 13, and 14 to 22, which relate to the value of biodiversity. Within the mandate of this group was a call for support in methodological improvements that can help in the identification of important aspects related to disaggregation and aggregation for each headline indicator. The aggregation and disaggregation of a headline indicator is reflected by the three indicators for each goal or target—headline, component, compliment. While most of the targets that are based on spatial data have sufficient methodology to set the indicators, targets that rely on economics lack specificity in setting an appropriate indicator.

Biodiversity can fit into an economic measurement system, but not in the way biodiversity is often discussed. Most economists start by considering changes in wealth, with changes in the value of natural capital being an important component of changes in wealth (e.g., ref. 5). Biodiversity can be considered part of natural capital (6), where biodiversity is measured as a subset of natural capital. This, however, begs the question whether it is helpful to develop a biodiversity statistic prior to valuing the constituent parts (e.g., species, functional groups, genotypes, etc.). The evidence suggests that it is not; a point we formalize in this article. In fact, those charged with providing guidance on valuing biodiversity have pivoted their charges to valuing changes in nature broadly, in part because they appear to have found the question of valuing biodiversity poorly defined. For example, the 2022 IPBES Report on the Diverse Values and Valuation of Nature (3) reframed valuing “biodiversity” to valuing “nature’s contributions to people” (NCP) (7).^* Similarly, Dasgupta (2) focused on connecting the value of nature for people rather than on connecting the value of biodiversity to people. Rapid progress is being made on accounting for the changing value of critical elements of nature (8), but this leaves a gap for biodiversity and the GBF.

Many valuation approaches have focused on using multidimensional biodiversity indices as complements with other headline indicators of nature’s value (9). However, such approaches will likely lead to greater confusion than clarity because ecological science has long acknowledged that appropriately quantifying biological diversity is extremely controversial (10). These challenges are likely one reason that the U.S. National Strategy for Natural Capital Accounting (11) states, “biodiversity cannot be included as its own environmental sector, but it may still be informative about the condition and extent of certain ecosystem assets.” As a counterpoint, the strategy does provide a path to measuring, valuing, and accounting for other components of nature including species.

Here, we discuss how people should think about a path toward economic measurements that support the GBF goals. We argue that the change in biodiversity wealth, as a subset of the change in natural capital wealth, can be a feasible option for the GBF Monitoring Framework. This requires thinking of biodiversity as a shorthand for the assemblage of life in a specified area that includes an accounting for specific ecological relationships, and not as a physical measure itself. We show why this should be by providing an “impossibility theorem” that states that first measuring biodiversity precludes capturing the heterogeneous contributions, dynamics, and interactions among the constituent parts (e.g., species) in the value of biodiversity. This means that a change in the measure of biodiversity can be linked to changes in value only by first measuring the changing value of the constituent parts. Starting the process with an aggregate measure of biodiversity cannot lead to useful economic values of biodiversity for policy. This requires approaching the biodiversity valuation problem differently from approaches that are customarily used, such as for climate change, where policy decisions regularly include globally agreed-upon measures, such as mean global temperature, tons of carbon, and the social cost of carbon. Rather, the wealth lost from biodiversity decline, or gain from recovery, may need to serve as the aggregate headline measure of biodiversity change.

Making Biodiversity an Economically Meaningful Question

Economists have long derided the notion of assessing a total value for the biosphere (12). Most agree with Toman (13) that any value would be a “gross underestimate of infinity”, by which the infinite value of the biosphere would be what humanity is willing to pay to avoid a shift from the current state of the biosphere to a biosphere ceasing to exist entirely. A similar rationale applies for biodiversity valuation. Biodiversity is the diversity of life that underpins human existence. Biodiversity is closely associated with the charismatic animals one sees in a nature documentary, but biodiversity is much more. Biodiversity includes soil microbes that facilitate plant growth by breaking down and recycling waste products. It includes the algae and trees that sequester carbon, the variety of plants that stabilize soils on steep slopes, and coastal areas that attenuate wave action. It includes food sources, agricultural pests and their control agents, and more. A common theme across these services is that they are not one input to one output, but rather involve a suite of interactions among living and nonliving things. Humanity undoubtedly ought to be willing to sacrifice all it has to avoid a situation where all life, including humans, ceases to exist. Accurately quantifying the magnitude of this total value is not a well-posed question. Still, smaller changes in biodiversity certainly can be valued. Therefore, any notion of “the value of biodiversity” must characterize the value of a change in the condition of biodiversity or the change in wealth resulting from a change in the composition of biodiversity (14).

A tractable and standardizable process for measuring the value of a change in the condition of biodiversity requires considerations of ecology and economics together. The nature of measurement, and the required precision to chart a path forward, requires consideration of the mathematical underpinnings and of measurement theory (15). One of the many commonalities between ecological and economic systems is that nonlinear relationships abound within them. The existence of nonlinearities in the mathematics of ecology and of economics implies that the order in which considerations are addressed matters because of Jensen’s inequality. Jensen’s inequality, applied to this case, implies that deriving an aggregate measure of biodiversity and then its value, leads to a different result than if value is derived from the constituent parts of biodiversity, whether species, functional groups, genotypes, guilds, whole ecosystems, or other classification regimes summarized into biodiversity and then aggregated. Hence, it may not be possible to develop a measure of biodiversity stock that has a well-defined value, in stark contrast to the process of measuring carbon dioxide emission flows and valuing them at a single social cost of carbon (16). This basic, but typically unattended, fact leads to substantial confusion among experts and policymakers alike seeking to measure the value of a change in the condition of biodiversity. For instance, a narrative perspective would hold that the intuitive steps to valuing are to, in sequence: measure biodiversity in an initial time period; value the extant biodiversity in that time period; repeat the measurement and valuation in a subsequent time period; and subtract one from the other to calculate value change. However, considering Jensen’s inequality, this process cannot lead to a correct value for the change in biodiversity. Worse, it risks causing ill-defined quantities and contradictions that make the problem of measuring the value of changes in biodiversity intractable.

Properties for a Measure of the Value of a Change in Biodiversity

A biodiversity measure, or index, $D(·)$, may best be thought of as a specific summary of many, but perhaps not all, natural assets within the wealth accounting framework (5). Thus, an economically meaningful definition of biodiversity is nested within the concept of natural capital. Wealth accounting seeks to measure the change in value over a comprehensive or inclusive set of assets, A, that is usually divided into produced (human-made), human, and natural capital. This change in wealth is proportional to a change in economic welfare (17). Gains in natural capital are valued like other changes in capital, as the change in the net present value of the flow of future services (2, 18). In the case of natural capital this depends on service flows to people, institutions that guide human interactions and those human actions, the discount rate, and ecological interactions that alter the net growth of the asset (19).

A biodiversity index describes, in one form or another, the assembled species composition within a fixed area. The mapping $D(·)$ must operate over the constituent elements, $S\subset A$, that make up biodiversity to take the many constituent elements and return a single scalar index, e.g., species richness or the Shannon–Weiner Index. To fix ideas, we refer to the constituent elements as species. However, biodiversity could be defined over any taxonomy or classification system, so it may, for example, be defined over lineages, functional groups, or ecosystem types. Furthermore, the function $D\left(·\right)$ may capture multiple dimensions and interactions among the constituent parts through weighting terms. For example, Weitzman (20) proposes that $D(·)$ include genetic distance between species, and generally emphasizes the importance of including “distinctiveness” in measures of biodiversity (21). To further fix ideas, let, $\mathit{s}(t)$ be the vector of species $s_i\forall i\in S$, at time $t$, in the area within which we wish to measure and value biodiversity. Furthermore, let there be a cardinal measure associated with each $s_i$, again to fix ideas, think of that as the population size, but other cardinal measures such as biomass would work equally well.^† The core question that we address is can the change in value be defined using a biodiversity index that uses the same species list, $D(\mathit{s})$ and the dynamic changes in the system of those species $dD(\mathit{s})/dt$, in a way that replicates using a list of species and their interactions, $\mathit{s}$ and the changes in those species over time, $\frac{d\mathit{s}}{\mathit{dt}}=\dot{\mathit{s}}$, where $\dot{\mathit{s}}$ must include interactions that determine changes in each element of $\mathit{s}$.

In order to focus only on the changes in biodiversity, assume all assets that would not be considered as contributing to biodiversity (e.g., machines and buildings), $A\backslash S$, are fixed. Indicate them with the parameter $a$.^‡ Now, consider a measure of the change in value using the list of species rather than a biodiversity index. Define the net present value from the system at time $t$ as

$$V\left(\mathit{s}\left(t\right);a\right)={\int }_t^{\infty }U\left(\mathit{s}\left(\tau \right);a\right)e^{-\delta \left(\tau -t\right)}d\tau \mathrm{subject}\mathrm{to}\dot{\mathit{s}}\left(\mathit{s}\left(t\right);a,\delta \right),$$ [1]

where $a$ and $t$ are suppressed when doing so does not cause confusion, $U$ is the value of services provided to people at time $t$, conditional on the populations of all species in $S$, and $\delta$ is the discount rate.^§ The calculation of $V\left(\mathit{s}\left(t\right);a\right)$ depends on the potentially stochastic dynamics of $\mathit{s}$, indicated by the dependence on $\dot{\mathit{s}}$, which includes ecological interactions and human feedbacks that drive the dynamics of the ecological and economic system. The dynamics for any single $s_i$ generally depend on the entire vector $\mathit{s}$.

The value of a small change in the population of species $i$, i.e., its price, is defined as $p_i\left(s_i,\mathit{s}\right)\equiv \frac{\partial V\left(\mathit{s}\right(t\left)\right)}{\partial s_i\left(t\right)}$. The marginal value of species $i$ depends on the species’ own population dynamics and its ecological interactions with other species in the community (22, 23). This structure does not impose that $V$ is separable in species. In other words, the marginal value of a contributing species to a biodiversity value cannot be assessed independently of its relationship to other species, and more generally, to other capital stocks that may not be species. For instance, in physical terms, a decline in the abundance of a focal herbivore species could cascade through the ecosystem leading to a decline in the abundance of its carnivorous predator species, as well as increase the abundance of the plant species eaten by the herbivore, and in turn decrease the stock of soil nutrients used by the plant. In economic terms, the decline in the herbivore’s abundance leads to changes in the marginal values of the predator, plants, and soil nutrients because a change in the herbivore population changes the marginal value of adding an additional unit of predator, plant, or soil nutrient to the system.

The change in the value of the system over time, $\mathrm{\Delta }V$, is a total derivative that encompasses all interactions. Therefore, as long as price depends on the population $s_i$, then the change in value using equation 1 is not the simple difference of prices times quantities at two different time periods, $\mathrm{\Delta }V\ne p_i\left(s_i\left(2\right), \mathit{s}\left(t\right)\right)s_i\left(2\right)-p_i\left(s_i\left(1\right), \mathit{s}\left(t\right)\right)s_i\left(1\right)$ (24). This inequality follows from the fundamental theorem of calculus, because the change in value becomes the difference in populations multiplied by a convex (i.e., weighted average) combination of the prices (25), e.g., $\overline{p}=\overline{p}\left(p_1,p_2\right)$. In the rare case that only a single $s_i$ changes and the price of $s_i$ is linear in $s_i$, or changes are sufficiently small to justify a linear approximation, then $\overline{p}$ is exactly the arithmetic mean of prices. When the populations of many species are changing, ecological or economic relationships are changing, or both, then the sort of index numbers used to convert from nominal to real value changes need to be used to measure value changes (25).

Now, consider the implications of using $D(\mathit{s})$ and $\frac{dD\left(\mathit{s}\right)}{\mathit{dt}}$ in place of $\mathit{s}$ and $\dot{\mathit{s}}$. Assume that for a specific set of species and parameters $a$ that $D(\mathit{s})$ can be normalized to $\stackrel{\sim }{D}(\mathit{s})$ such that $V\left(\tilde{D}\left(s\left(t^{\text{'}}\right)\right);a\right)=V\left(s\left(t^{\text{'}}\right);a\right)$, where $t^{\text{'}}$ is a specific point in time and the normalization does not need to hold through time. Consider what happens with a small increase in $s_i$. This can also be thought of as going from absence, $s_i=0$, to the presence of a species, $s_i=ϵ$. It is often said that the value of biodiversity is greater than the sum of the parts, which implies that any reasonable measure of biodiversity must yield $\frac{\partial V\left(\stackrel{\sim }{D}\left(\mathit{s}\right)\right)}{\partial s_i}\ge \frac{\partial V\left(\mathit{s}\right)}{\partial s_i}.$ This “greater than the sum of the parts” relationship holds as an equality when the biodiversity index, $\stackrel{\sim }{D}$, and model of the system, $\dot{\mathit{s}}$, account for all the interactions among all species.

When all ecological interactions are accounted for, the chain rule from calculus can be used to expand the “greater than the sum of the parts” condition $\frac{\partial V\left(\stackrel{\sim }{D}\left(\mathit{s}\right)\right)}{\partial \stackrel{\sim }{D}(\mathit{s})}\frac{\partial \stackrel{\sim }{D}(\mathit{s})}{\partial s_i}=\frac{\partial V(\mathit{s})}{\partial s_i}$. The chain rule from calculus ensures that this condition holds for small changes in any species $i$ contributing to biodiversity measure $\stackrel{\sim }{D}$ or $D$, since the normalization would not affect this property.

The chain rule reveals important features necessary for an index of biodiversity that could provide tractable and accurate ways to integrate measures with valuation. First, if species $i$ and species $j$ provide different current or future services, then changes in species $i$ and $j$ must be reflected differently in the measure of biodiversity, and $D$ must map changes in each species uniquely. This is partly because people derive different values from different species (6) and partly because of species’ properties and interactions. For a biodiversity index to account for species attributes in a unique fashion it must account for changes in species in a unique fashion. Specifically, if species $i$ changes, leading to a change in the biodiversity index, then there cannot exist any potential change in species $j$ that could have yielded the same change in the biodiversity index. The second left-hand side factor in the greater than the sum of the parts condition implies that any observed change in a biodiversity index needs to be uniquely attributable to a specific species. Biodiversity indices that do not account for species attributes in such a unique fashion cannot replicate the change in value based on the species upon which the biodiversity measure is built. Second, the usefulness of a biodiversity index is that it reduces the dimensionality of the system. The goal is that a biodiversity index can be valued using the marginal value of the biodiversity index $\frac{\partial V\left(D\right)}{\partial D}$ alone. For an observed change $D\left(2\right)-D\left(1\right)$ the real, as opposed to nominal, marginal value, ${\overline{p}}_D$, is computed as a convex combination of $\frac{\partial V\left(D\left(\mathit{s}\left(1\right)\right)\right)}{\partial D\left(\mathit{s}\left(1\right)\right)}$ and $\frac{\partial V\left(D\left(\mathit{s}\left(2\right)\right)\right)}{\partial D\left(\mathit{s}\left(2\right)\right)}$, so that the marginal value must be assessed before and after the observed change. Thus, the nature of the change matters. In order for the elements of the convex combination ${\overline{p}}_D$ to be the same as prices that account for individual species changes, i.e., $\frac{\partial V\left(\stackrel{\sim }{D}\left(\mathit{s}\right)\right)}{\partial \stackrel{\sim }{D}(\mathit{s})}\frac{\partial \stackrel{\sim }{D}(\mathit{s})}{\partial s_i}$, and so that there is a price of biodiversity alone, the second factor, $\frac{\partial \stackrel{\sim }{D}(\mathit{s})}{\partial s_i}$, would have to equal 1 for all species $i$. Requiring $\frac{\partial \stackrel{\sim }{D}(\mathit{s})}{\partial s_i}$ to equal one is the only way that the marginal value (i.e., implicit price) of a change in the biodiversity index associated with a change in species $i$ is equivalent to the marginal value of a change in species mapped through the biodiversity index and accounting for the full set of interactions, i.e., $\frac{\partial V\left(\stackrel{\sim }{D}\left(\mathit{s}\right)\right)}{\partial s_i}=\frac{\partial V(\mathit{s})}{\partial s_i}$. This second condition generally contradicts the first. Treating all changes in species in an equivalent fashion implies that there likely exists at least one species $j$ for which $\frac{\partial V\left(\stackrel{\sim }{D}\left(\mathit{s}\right)\right)}{\partial s_j}<\frac{\partial V\left(\mathit{s}\right)}{\partial {\mathrm{s}}_j}$. Therefore, any biodiversity index $D$ that does not uniquely treat changes in all species cannot be used to accurately value the changes in biodiversity. This result implies that computing a biodiversity index cannot be the first step in valuing changes in biodiversity because it will mask important differences in the change in value associated with changes in different species. Starting with valuation of key species will lead to a more robust measure of change in value, and it seems likely that it is easier to capture important interactions among species, or with other forms of capital, using system specific $\dot{\mathit{s}}$ than using a general index $D(\mathit{s})$.

Biodiversity Indexes with Heterogeneity are also Insufficient

The fact that no index for measuring biodiversity that treats species equally can satisfy the desirable properties for valuing changes in biodiversity is not surprising considering that measures of biodiversity vary widely across disciplines (26), and few have been designed with valuation in mind. This proliferation of biodiversity metrics arose in part because the set of living things can be defined in many dimensions, and different researchers aimed to describe different aspects of ecological communities, often with little thought about value to humanity.

Biodiversity measures range from species richness, i.e., the count of species, to evolutionary and functional distance measures (10). For some applications, it is important to consider not only extant species (lineages or functional groups), but the relationships of extant groups to extinct ones (27). Chao, Chiu, and Jost (10) show that most phylogenetic or abundance-based, parametric biodiversity indices within the ecology literature are merely variations of each other because they are nested within the Hill number concept for biodiversity, which provides a measure of the “effective number of species, lineages, or functional groups.” In the simplest case, Hill number diversity indices, $D_H$, are defined up to an arbitrary elasticity of substitutability within the index, $\sigma =\frac{1}{1-q}$, among the species $s_i\in S$

$$D_H={\left({\sum }_{i=1}^S{\alpha }_i{s}_i^q\right)}^{\frac{1}{1-q}},$$ [2]

where $q\ge 0$ and $q\ne 1$, and ${\alpha }_i$ are attribute values assigned to species $s_i$ that enable the Hill number approach to apply to lineages or functional groups.^¶ The coefficient ${\alpha }_i$ can also be viewed as an importance weight, which begins to account for the heterogeneity of species and the services they provide. By allowing $\alpha$ to vary by species, Hill numbers can be generalized to evolutionary and functional diversity measures.

In the cases where ${\alpha }_i=1\forall i$, which is the case of equal importance, then if $q=0$, all species are perfectly substitutable for each other. In this case, $D_H$ is a measure of species richness. If $q=1$ taking the limit implies the Shannon Diversity index after a logarithmic transformation. Chao, Chiu, and Jost (10) provide connections to other common biodiversity measures.

A key feature of Hill numbers is that $D_H$ does not define a unique assemblage of species. Consider the case where there are two species in the group of species. For a given value of $D_H$, choice of $q$, and level of $s_1$, there is a unique level of $s_2$. As the number of species increases so too do the combinations of species in the group that lead to the same value of $D_H$. If there are only two species, then the relationship $\frac{d{s}_j}{d{s}_i}=-\frac{{\alpha }_j}{{\alpha }_i}{\left(\frac{s_i}{s_j}\right)}^{1-q}$ maps all the combinations of species $s_i$ and $s_j$ that yield a single value $D_H$. Mathematically, there are an infinite number of combinations of $s_i$ and $s_j$ for any value of $D_H$. A finite subset will be ecologically feasible, but that subset is still likely to be large. It is unlikely that all these combinations of species provide the same value to society.

For a biodiversity index to be useful for connecting biodiversity to value for people, it must at a minimum account for the assemblage of species and their ecological relationships and interactions as determined by their functional roles. It is through these functional roles that species provide the ecosystem services, support life, and provide value to people. To this end, the concepts such as trophic level (TL) index (28) represent useful information and that implicitly measure species functional interdependencies. Recent developments expand the TL idea to include attributes of species that determine how they function together with their interdependencies within regional assemblages (29, 30). Moreover, these indices can also be represented using the Hill number approach (29). Hence, even for these recent index innovations, the ${\alpha }_i$ value could be a marginal economic value weight.

Using the relatively flexible Hill number approach as a basis for valuation, however, still imposes unsatisfactory constraints and requires imposing assumptions that do not come from economic or ecological theory. First, the value of $\alpha$ must be measured and known prior to forming the biodiversity index. However, if this information were known, then one could compute the change in value without first calculating the level or change in the biodiversity index. Second, the value of services from biodiversity at any point in time is the biodiversity index itself, up to an affine transformation, but it is not clear that the same transformation will hold in the next period. To use the Hill number approach, the future services from the ecological assemblage would need to be provided according to the constant elasticity of substitution function, which, while a common functional form assumption in economic analysis, still seems unlikely. Moreover, the elasticity of substitution for biodiversity in terms of social preferences or production would necessarily be the elasticity of substitution used in the Hill number biodiversity index. It would mean that the units of change in biodiversity would need to be the same units of change in value with respect to the change in the underlying species assemblage that leads to a change in the biodiversity index. Finally, one must conclude by the time we had all the information to implement a Hill number index suitable for valuation, we would have already valued the underlying species, accounting for their interactions, and thus already valued biodiversity.

In contrast to the Hill number approach, economists, e.g., Weitzman (21), have focused on service flow, including direct service flows and “distinctiveness,” where distinctiveness is thought of as how much unique (genetic) information a species contains. Such analysis tends to abstract from the joint dependencies in the system and has been primarily used to rank conservation programs ex ante rather than measure or value changes in biodiversity ex post.

Discussion

In adopting the Kunming-Montreal Global Biodiversity Framework, global policymakers have determined that valuing the contribution of biodiversity to humans is necessary for halting and reversing the loss of global biodiversity. However, by leaving open the methodology for how best to perform that valuation, a discordant set of valuation measures and approaches are being undertaken globally - leading to incongruous outcomes and a likelihood that the goals of the GBF will not be achieved.

As a practical approach to support near-term implementation of the GBF, we propose that valuing changes in biodiversity, as a summary statistic of a subset of natural capital, would leverage existing approaches and international standards associated with environmental-economic accounting to produce a measure of change in biodiversity in units that can reflect the change in the value of biodiversity. Changes in the value of biodiversity should be thought of as changes in wealth. It is easiest to develop this idea holding all other stocks constant, but changes in nonbiodiversity stocks, such as boats, houses, and human knowledge, may also affect the value of biodiversity. A biodiversity index is intended to collapse the multifaceted aspects of a biodiverse community into a single numerical value. In parallel, valuation leads to a single numerical value that conveys information about relative scarcity in terms of value to society (31). Extending the price system to nonprivate natural capital is essential for wealth-based assessments of sustainable development (32) and resolves the many-to-one question in a manner consistent with other contributors to changes in human welfare. It is important to value biodiversity to connect changes in biodiversity to changes in human welfare, particularly to support policy decisions regarding sustainable development.

Valuing changes in biodiversity needs to be based on the valuation of the underlying species or other component parts, accounting for their ecological interdependencies and relationships, how people respond to changes in the species assemblage, and the service contributions to people from the species. Given the challenge this poses, policymakers need to choose between doing a poor but seemingly comprehensive job or a good, but partial job. The latter approach lends itself to improvements in methods and expansion of coverage over time.

Approaches are well on their way to meeting the conditions necessary to develop useful first approximations. For example, Kvamsdal, Sandal, and Poudel (33) and Yun, Hutniczak, Abbott, and Fenichel (22) measure the marginal value of changes in fish populations within multispecies fishery systems. Both show how marginal values change over time as populations of the species being measured change and as their competitors, prey, and predators change. Bond (34) undertakes a similar analysis in a fishery system, showing how the marginal value of a fish population changes as the availability of nursery habitat changes. Maher, Fenichel, Schmitz, and Adamowicz (23) study wildlife in a forested ecosystem under human development pressure. Similarly, they find that the marginal value of a species depends on its own population and trophic relationships, all of which are dynamic. They also show how the marginal value of each species depends on human influence on habitat. These studies present their work as changes in wealth over a relatively small portfolio of species presented as natural capital. It is necessary, and likely possible, to scale up such approaches to provide measures of the value of changes in biodiversity. Effort should be focused on improving such empirical methods and associated data collection rather than continued debate over what it means to value biodiversity.

Complementing localized measurement activities is the development of regional and national species accounts within the System of Environmental-Economic Accounting (SEEA) (35). King et al. (36) explore these possibilities and find initial promise with respect to taxonomic species, while flagging difficulties related to including migratory species. They also point to the early phases of experimentation with accounting procedures and emphasize the need for standardized species and biodiversity measures to be regularly included in economic measurements. The use of an accounting framework that can provide a nested hierarchy for increasing detail overtime has worked well for developing statistics on economies, which today provide international comparisons and support a wide range of policy decisions. The need to understand interactions and a wide range of connections is not that different from measuring the interactions in the traditional economy. While developing economic accounts has taken time, and they have improved and become more refined overtime, they also started out relatively simple.

Developments in community ecology enabling enhanced understanding of food webs, such as stable isotopes (37), proteomics (38), and dynamics (39), are helping develop trophic connections that are likely of first-order importance. The underlying structure of modern food web models often is very close, and likely can be aligned with, the supply-use table structure used in national economic accounts and SEEA (40).

Focusing on a few important components of biodiversity and their first-order connections is the important first step. Many services from biodiversity have been identified and even more hypothesized, but nearly none are currently included in practice in national accounts. This is not a conceptual failure; the internationally agreed guidance for national accounting (41) states in paragraph 1.46 “many environmental assets are included within the SNA [System of National Accounts]”. The components of biodiversity are formally considered nonfinancial, nonproduced assets. Rather, their omission is the result of the inability of the expert community to agree on where to start, reflected by continued debate on how to measure and value biodiversity. The focus on every dimension of biodiversity is a reason for doing less than has already been agreed to. And, debates of the “correct” biodiversity index feed further delay.

Taking all this together, there is a clear path forward for measuring the value of changes in biodiversity. First, recognize that biodiversity is not something that can be rolled up into an arbitrary index; rather it is shorthand for the assembly of life in a well-defined geography that accounts for ecological and human interactions. The best way to measure biodiversity is to measure the marginal value of specific species, or groups of species belonging to a community, and condition valuation on ecological interactions and interactions with people. Measured species can be added to balance sheets as nonfinancial, nonproduced assets. Then changes in biodiversity are changes in the wealth contribution of those species.

A set of priority species could be chosen based on specific geographies and social and ecological conditions. The international community should focus on the process for countries to choose a first set of species relevant for local conditions. This might involve focusing on keystone species that are viewed as ecologically, socially, and economically important. The international community should also focus on guidance for accounting for ecological connections and valuing species, and clear ambition to expand the set of species over time, with expansion opportunities built into the classifications systems. Our proposed approach contrasts with first attempting to measure biodiversity or changes in biodiversity and then valuing those changes. Creating a physical biodiversity index introduces an overly complicated step that cannot replicate measurement based on more detailed data about ecological communities, except under highly contrived conditions. It also introduces nonuniqueness in the ensuing value—something that is untenable for policymakers.

The necessary approach to value biodiversity loss stands in stark contrast to the path for valuing climate change, where mean global temperature is understood as a sufficient first-order physical measure (16). In climate change, first measuring the aggregate physical change and then valuing that change is a sufficient measure for many uses. Despite the appeal of a 1.5 °C-like target for biodiversity as a communication tool (42), it is likely to distract from obtaining workable solutions. It will invariably lead to measures that introduce arbitrary decisions that are untenable in the policy space. Instead, it is important that valuation of loss of biodiversity happens, in a way aligned with the change-in-wealth concepts underpinning the Dasgupta Review (2), so that policymakers can better understand what changes in biodiversity mean for national budgets, risk and insurance, bank and industrial regulation, sectors with the power to shift trends, as well as for changes in national wealth and human well-being in the long-term.

In the context of the Kunming-Montreal Global Biodiversity Framework, measuring biodiversity as a summary statistic of a subset of natural capital provides a solution to many of the remaining indicator gaps in the monitoring framework. By adopting the proposed methodology as an indicator, particularly for Goals B and D and Targets 14 to 17, there would be a more transparent and responsible approach to facilitate monitoring and review of progress at all levels, as well as provide a means of comparison on country-level progress.

Data, Materials, and Software Availability

All study data are included in the main text.