Evaluation of the City Biodiversity Index
While commendable for pioneering the measurement of urban biodiversity, the City Biodiversity Index is not flawless due to its underemphasis on ecosystem services and its equal-weight approach. This essay proposes the use of Principal Components Analysis to more accurately weight the core components.
Given the significance of biodiversity to human well-being (Taylor & Hochuli, 2015), the burgeoning rate of urbanisation has engendered increasing pressure to measure biodiversity conservation efforts in cities. The City Biodiversity Index (CBI) is a pioneering tool designed to measure such urban biodiversity. It strives to benchmark biodiversity conservation efforts across cities, in order to pinpoint existing lacunae in conservation efforts. The CBI ‘serves as a self-assessment monitoring tool’ for cities to improve the management of their urban ecological resources and conservation of their urban biodiversity (Tok, 2011). The Index was formulated by Singapore’s National Parks Board, alongside a committee of international experts. In contrast with other existing indices measured at the national level, the CBI is the first that aims to measure biodiversity specific to urban areas (Rodricks, 2010).
The CBI comprises two segments: a ‘Profile of the City’ providing background information on the city, and 23 indicators that measure three dimensions of urban biodiversity. These dimensions are: native biodiversity in the city, ecosystem services (ES) provided by biodiversity, and governance and management of biodiversity (Table 1). Each indicator ranges from 0 to 4 points, with a maximum possible of 92 points for a city (Figure 1). Primarily designed to be a self-assessment tool, the CBI will take the first year of the city’s score as the baseline for comparison against future years’ development in biodiversity conservation. Statistical normalisation was employed for data of indicators 2, 3, 9, 11, 12, 15 and 16 to provide a scientific basis for objective and fair scoring (Chan et al, 2014).
The index is an additive one comprising equally-weighted, interval-level indicators. As such, it imposes false equivalences on what constitutes the same single-point increase across different indicators. For example, one might not be inclined to believe that having 6 more governmental partnerships with nature-related organisations (a single-point increase from 0 to 1 point in Indicator 21) would be directly equivalent to a decrease of 10 percentage-points in invasive alien species (a single-point increase in Indicator 10). By weighting every indicator equally, the index purports that each increase in interval is equivalent across indicators, in spite of how some indicators measure drastically different qualities. Furthermore, points within some indicators are not coded at the interval scale but rather ordinal instead. Whilst the process of assigning points to each ordinal level has converted the indicator into an interval one, the distances between each interval do not reflect consistent increases in the measured value. For example, for Indicator 3 (Native Biodiversity in Built-Up Areas), 1 point is assigned to a range of 8 species (19–27), while 2 points correspond to a range of 18 (28–48) which is a noticeably larger jump.
To combat this, a possible solution would be standardisation, based on the Mahalanobis distance, of existing raw data from cities for all quantitative indicators, in order to equalise the magnitude of each indicator. This ensures that every indicator would be on the same scale, enabling numerical comparability across indicators which would otherwise have varying units. After which, a single-point increment will be affixed to each quantile. This ensures consistency across single-point breaks within indicators, through the statistical basis provided by standardisation. Nonetheless, this can only be applied to quantitative indicators. Additionally, the standardised indicators are comparable only in terms of their distribution within the 23 cities over the time period in which data has been collected. As such, the external validity of the measurement may be jeopardised.
Each dimension is unequally weighted, resulting from the unequal number of indicators within each. Defined as the benefits provided by ecosystems to humans (FAO, 2020), ES are arguably as important as the other dimensions, which are roughly weighted the same at a maximum score of 40 and 36. However, with only 4 indicators measuring it, this dimension of ES is weighted about 25% as much as each of the other indicators. Since a plethora of key ES functions are provided and regulated by biodiversity (ibid), this dimension is equally as critical in measuring biodiversity. The First Expert Workshop held in 2009 revealed that this paucity of indicators can be attributed to the difficulty of designing measurements for this novel field of study (Chan et al, 2014). However, extensive research has since been published that substantiates the value of ES in service delivery, risk reduction, direct economic value, and intrinsic value (Brown et al, 2014), across various categories of provisioning, regulating, habitat and cultural and amenity services. At the same time, various indicators have been proposed to measure functions within respective categories — such as ‘soil infiltration capacity’ or ‘market value of carbon uptake’ (Gómez-Baggethun & Barton, 2013). With the development of ecosystem service indicators since CBI’s inception, this dimension should justifiably receive similar weightage to the other two dimensions.
Proposed solutions
To increase this dimension’s weightage, one solution would be adding more indicators to measure ES, to reflect this subset of biodiversity more accurately. However, a few challenges lie in this endeavour. ES also reflect the social aspect, as they are products of an interconnected socio-ecological system, rather than physical biodiversity in silos (Reyers et al, 2013). As such, additional social and economic data, atop of biophysical data, are required to holistically measure ES. However, this is a non-issue in the context of the CBI, which already incorporates socio-political measures in the ‘Governance and Management’ dimension. Secondly, the various components of ES tend to be correlated with indicators of biophysical structures and processes, as reflected in the ‘Native Biodiversity’ dimension, which may lead to double-counting.
To mediate these issues, a proposal would be to apply Principal Components Analysis (PCA) on the data to identify the most pertinent indicators of this dimension. The aim of applying PCA is twofold: identify indicators of ES that would describe the most variation in the data, whilst weighting each indicator in a manner that best explains variation in the data (rather than the current equal weights). PCA combines all indicators such that maximum variance in the data is captured. In this process, PCA decomposes the original set of indicators into principal components, each uncorrelated with one another. Within each principal component is a linear combination of the original indicators, with a weight assigned to each.
Generally, correlations amongst indicators are sufficiently high that the first few principal components can account for most of the variance in the data. If this is true, ‘parsimony and clarity in the structure of the relationships are achieved’ (Nardo et al, 2005: 17). PCA is highly applicable to the CBI as typically, ‘environmental variables are…closely related’ (Jha & Murthy, 2003: 11). With a relatively large number of indicators, the CBI estimates produced are vulnerable to the vagaries of measurement errors, to which environmental data are highly prone (ibid). The multivariate analysis performed by PCA will be used to choose the most relevant indicators, by measure of loading in the first few principal components, relative to biodiversity. By applying it to the newly picked indicators of ES, measures of ES most pertinent to biodiversity can be captured.
Since indicators hold varying levels of importance to the measurement of biodiversity, differential weights derived from PCA can more holistically reflect the differences across indicators. As such, PCA is an appropriate statistical approach to determine these indicators’ contribution to biodiversity measure, whilst retaining objectivity. However, a significant trade-off of using PCA would be the ease of applying the Index. The CBI is designed to primarily be a self-assessment tool for use by city and municipality governments rather than technical experts. The added layer of complexity from differential weights, compared to the initial linear additive nature of the Index, may discourage further uptake of CBI. Furthermore, PCA may not be able to measure the exact concept of biodiversity, even when indicators are specifically picked to this end; rather, it merely recovers dimensions that explain the most variation in the simplest way (Lauderdale, 2020).
Indices, when accompanied by targets and demonstrable outcomes, are a desideratum in measuring progress. The CBI is one such laudable effort at measuring urban biodiversity, reminding urban governments to prioritise environmental goals in the master planning of a city. The construction of such an index is a dynamic, iterative process in which periodic evaluation of the methodology is warranted. This essay has identified some conceptual and methodological pitfalls of the CBI, such as equal weightage and sidelining ES. The amendments proposed on the basis of developments in research, whilst adding complexity to the Index, will hopefully reduce measurement error and capture the target concept of biodiversity more precisely. Accurate, holistic and updated measures of biodiversity, through the CBI, will indubitably be a tool for urban environmental discourse, and can greatly influence urban policy analysis.
References
Brown, C., Reyers, B., Ingwall-King, L., Mapendembe, A., Nel, J., O’Farrell, P., Dixon, M. & Bowles-Newark, N. J. (2014). Measuring ES: Guidance on developing ecosystem service indicators. UNEP-WCMC, Cambridge, UK.
Chan, L., Hillel, O., Elmqvist, T., Werner, P., Holman, N., Mader, A. and Calcaterra, E. (2014) User’s Manual on the Singapore Index on Cities’ Biodiversity (also known as the City Biodiversity Index). Singapore: National Parks Board, Singapore.
FAO (2020). Plant Production and Protection Division: Biodiversity and Ecosystem Services. Retrieved 9 December 2020, from http://www.fao.org/agriculture/crops/thematic-sitemap/theme/biodiversity/en/
Gómez-Baggethun, E., & de Groot, R. (2010). Natural capital and ecosystem services: The ecological foundation of human society. In R. E. Hester & R. M. Harrison (Eds.), Ecosystem services: Issues in environmental science and technology (Vol. 30, pp. 118–145). Cambridge: Royal Society of Chemistry.
Gómez-Baggethun, E., Gren, Å., Barton, D. N., Langemeyer, J., McPhearson, T., O’Farrell, P., Andersson, E., Hamstead, Z., & Kremer, P. (2013). Urban Ecosystem Services. In T. Elmqvist, M. Fragkias, J. Goodness, B. Güneralp, P. J. Marcotullio, R. I. McDonald, S. Parnell, M. Schewenius, M. Sendstad, K. C. Seto, & C. Wilkinson (Eds.), Urbanization, Biodiversity and Ecosystem Services: Challenges and Opportunities: A Global Assessment (pp. 175–251). Springer Netherlands. https://doi.org/10.1007/978-94-007-7088-1_11
Jajuga, K., & Walesiak, M. (2000). Standardisation of Data Set under Different Measurement Scales (pp. 105–112). https://doi.org/10.1007/978-3-642-57280-7_11
Jha, R., & Murthy, K. V. B. (2003). A Critique of the Environmental Sustainability Index. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.380160
Lauderdale, B. E. (2020). Pragmatic Social Measurement. N.p
Nardo, M., Saisana, M., Saltelli, A., & Tarantola, S. (n.d.-a). Tools for Composite Indicators Building. 134.
Rodricks, S. (2010). Singapore City Biodiversity Index. 4.
Stern, S., & Epner, T. (n.d.). 2019 SOCIAL PROGRESS INDEX. 37.
Taylor L, Hochuli DF. Creating better cities: How biodiversity and ecosystem functioning enhance urban residents’ wellbeing. Urban Ecosyst. 2015; 18: 747–762. https://doi.org/10.1007/s11252-014-0427-3
The Singapore Index (SI) | Urban Ecology LSM4265. (n.d.). Retrieved 9 December 2020, from https://blog.nus.edu.sg/urbaneco/2017/04/07/the-singapore-index-si/
Tok, C. Y. H. (2017). City Biodiversity Index | Infopedia. Retrieved 4 December 2020, from https://eresources.nlb.gov.sg/infopedia/articles/SIP_1765_2011-02-11.html
