Introduction

The global process of urbanization has consolidated cities as the main human habitat. The primary reason behind this trend is the role of cities as gravity centers attracting population due to their role as hubs of higher access to health and education, economic growth, infrastructure, and other opportunities compared to non-urban areas (Bettencourt et al. 2007; Elmqvist et al. 2021). Despite spatially covering a small portion of the Earth’s surface (~ 3%), the urban processes have consequences far beyond their physical limits (Liu et al. 2014). To maintain their accelerated pace of function and growth, cities require constant regional and global flows of materials and energy (Elmqvist et al. 2021; Grimm et al. 2008; Inostroza and Zepp 2021).

The changes imposed by urbanization are a force of landscape transformation, which alters the functioning and structure of local ecosystems, fragments and degrades natural habitats, and disrupts ecological cycles (Alberti 2015; Lemoine-Rodríguez et al. 2020). This results in the loss of biological diversity and triggers several sustainability challenges in cities. Additionally, urbanization produces a complex matrix of land cover spatial arrangements at diverse scales with differential social and ecological effects (Forman 1995). Therefore, the assessment of urban form patterns (i.e., the spatial structure of cities) and their evolution is key to understand the spatial efficiency and the environmental impact of cities (Grimm et al. 2015; Wentz et al. 2018).

Diverse indicators have been proposed to capture the spatial complexity of cities. Such metrics have primarily aimed to aid decision makers with key components, offering them substantial evidence on which to base their actions (Andreasen et al. 2001; Kabisch et al. 2016; Ferreira et al. 2020). Indices often concentrate on different facets of the urban environment and population, spanning lifestyle, economy, ecosystemic health, and biodiversity, with the goal of evaluating urban environmental conditions and resilience (Mori et al., 2012). Of these indices, the Ecological Footprint and the City Biodiversity Index (also known as the Singapore Index on Cities’ Biodiversity) stand out as the most frequently utilized. The Ecological Footprint index gauges urban sustainability by integrating ecological and economic principles (such as carrying capacity and natural capital). It achieves this by accounting for the biologically productive land and sea area necessary to produce the renewable resources demanded by the urban population and to assimilate the generated waste (Wackernagel et al., 2006). In contrast, the Singapore Index focuses on assessing biodiversity at a local level, providing cities with the means to monitor their conservation efforts over time (Niemelä 2014).

Additionally, the increasing availability of open access remote sensing data has prompted numerous local, regional, and global spatial characterizations of cities (Güneralp et al. 2020; Huang et al. 2007; Lemoine-Rodríguez et al. 2020; Taubenböck et al. 2020). In the last decades, the inclusion of landscape metrics has allowed to describe spatiotemporal urban patterns quantitatively (Alberti et al. 2001; Schneider and Woodcock 2008; Uhl et al. 2010). Most efforts to improve the representation of the spatial complexity of cities have been focused on their built component. Diverse indicators such as land surface temperature (LST), building footprint, volume and height have been used to describe in detail the physical aspect of cities and as proxies of urban metabolism (Inostroza 2014; Lemoine-Rodríguez et al. 2022b; Li et al. 2020; Mahtta et al. 2019). Nevertheless, the representation of the biological dimension of cities has predominantly relied on the use of land cover data or radiometric indices such as the NDVI or EVI to describe the presence or greenness of urban greenspaces (Lemoine-Rodríguez et al. 2022b; Von Thaden et al. 2021). While these indicators help to identify vegetation in cities, they do not allow to measure the ecological value of urban areas.

To tackle this, the application of the concept of ecosystem integrity in urban systems was recently proposed (MacGregor-Fors et al. 2021). Briefly, urban ecosystem integrity refers to the similarity of a site compared to a reference system (often a natural ecosystem similar to that assumed to exist prior to the settlement of the city; Karr and Dudley 1981; MacGregor-Fors et al. 2022). This concept is operationalized in the Urban Ecosystem Integrity Index (UEII), which has been proposed as a spatially explicit indicator to quantitatively measure the ecological integrity of cities. The index was initially applied to the city of Xalapa, a Neotropical Mexican urban center, showing to be effective in capturing the interplay between the urbanization (i.e., built-up and land surface temperature; LST) and the biological (i.e., plants and birds’ richness) components and to quantitatively represent the spatial heterogeneity of the city (MacGregor-Fors et al. 2022). Moreover, the UEII allowed to quantify the gradient of ecological integrity values of the urban greenspaces of the city in a spatially explicit way.

Aiming to test the applicability of the index in a contrastingly different environmental scenario, we assessed the suitability of the UEII to measure the ecological integrity in a boreal city: Lahti, Finland. We then evaluated the performance of the UEII of both Lahti and Xalapa (MacGregor-Fors et al. 2022) compared to the widely used NDVI, which has often been proposed as a proxy of biodiversity (Benedetti et al. 2023). Given the nature of the UEII, we expected it to perform informatively for Lahti. Also, due to the amount of multivariate high-resolution information collated by the UEII, we also expected important variation between it and the NDVI, making the UEII a more reliable proxy of the ecological condition of urban settings on a citywide scale.

Methods

Study areas

We tested the applicability of the UEII in Lahti and Xalapa, two cities located in contrasting ecosystems.

Lahti is a boreal city located in Southern Finland (60°59’01” N, 25°39’23” E). Lahti is the sixth most populated city of the country with ~ 120,000 inhabitants and its urban continuum covers an area of 54 km2 (Official Statistics of Finland 2020). The vegetation of Lahti is boreal forests, mainly characterized by spruces, pine trees, birches, and aspens (Pfadenhauer et al., 2020), which covers over half (~ 51%) of the extent of its urban continuum. Xalapa is a Neotropical city located in East-central Mexico (19°32’38” N, 96°54’36” W). The city of Xalapa houses ~ 550,000 inhabitants and its continuous urban area covers 65.5 km2 (INEGI 2020; Lemoine-Rodríguez et al. 2019). The original vegetation of the area where Xalapa is located consisted of cloud forest, tropical dry forest, and riparian vegetation, distributed in subtropical highland climates (Castillo-Campos 1991; Kottek et al. 2006; Rzedowski 1978). The urban vegetation of the city decreased during the last decades due to urbanization and it is currently composed of a high number of exotic species (Falfán and MacGregor-Fors 2016; Lemoine-Rodríguez et al. 2019).

UEII calculation of Lahti

The UEII represents two urban form dimensions: the physical (describing all possible aspects of urbanization intensity) and the biological component (describing all possible aspects of urban biodiversity; MacGregor-Fors et al. 2022). The urbanization intensity affects ecosystem integrity negatively and it is related to the human footprint in the landscape (e.g., built-up cover, pollution, traffic, noise). The indicators that make up the biological component of the UEII represent the biodiversity of the reference system (e.g., plant and bird native species of the ecosystem where a city is located). The index ranges from − 1 to 1, with negative values describing an intensively urbanized system (i.e., low ecosystem integrity) and positive values related to high urban biodiversity (i.e., high ecosystem integrity).

The calculation of the UEII of Lahti consisted of the same procedure as described in MacGregor-Fors et al. (2022). Briefly, we delineated the urban continuum of Lahti (employing a detailed manually generated delineation of its urban continuum based on high resolution 2020 satellite imagery, following the clustering and communication criteria suggested by Lemoine-Rodríguez et al. 2019). Afterward, we focused on four variables: (1) land surface temperature (LST), (2) built-up area, (3) native woody vegetation species richness, and (4) bird species richness. To quantify the urbanization intensity (i.e., built cover) we used freely available remote sensing data. We acquired a 30 m resolution cloud-free Landsat 8 image from Lahti (28/06/2019) and a 10 m resolution Sentinel-2 scene (19/05/2019). We estimated the per-pixel LST in Celsius degrees for Lahti based on the Landsat image, following the approach described in Lemoine-Rodríguez et al. (2022a, b) and in MacGregor-Fors et al. (2022). To represent the built-up extent inside each Landsat pixel, we: (1) geographically adjusted the Sentinel-2 pixels to the limits of the Landsat pixel grid, (2) estimated the NDVI from the Sentinel-2 image to identify and exclude pixels with vegetation cover, and (3) manually removed water and bare soil pixels. Finally, we computed the total built-up area inside each 30 m resolution LST pixel (following MacGregor-Fors et al. 2022).

Native plant and birds species richness information used to represent the biological component of Lahti was collected through a standardized citywide survey (217 sites; 50 m radius). The surveys took place in June and July 2021. To transform the field data into spatially continuous information, we produced interpolated raster layers of plants and birds’ richness applying the inverse distance weighting (IDW) method in R (R Core Team 2020). The cells of the Landsat image were used as base spatial grids for the interpolation to spatially match the LST and built-up layers for the subsequent steps. In order to make our data comparable, we re-scaled the LST, built-up, native plant and bird species richness spatial layers into values ranging from 0 to 1 using the Min-Max transformation formula (MacGregor-Fors et al. 2021; So et al. 2020). Finally, we computed the UEII for each 30 × 30 m pixel applying the formula proposed by MacGregor-Fors et al. (2022), where each component corresponds to its re-scaled version:

UEII = {[((LST + Built cover)/2)×(-1)]+[(Plant richness + Bird richness)/2)]}.

Comparison between the UEII and the NDVI in Lahti and Xalapa

To assess the suitability of applying the UEII in Lahti and determine, together with the previously published UEII of Xalapa (MacGregor-Fors et al. 2022), if they provide more valuable information than the NDVI itself for assessing the ecology of cities, we conducted a comparison between the UEII and the NDVI in both Lahti and Xalapa. We estimated the NDVI at 30 m resolution for each city based on the NIR and Red bands of the Landsat 8 images and compared the UEII and the NDVI fitting linear and general additive regression models between the two indices based on a random sample of 10,000 pixels from each city. Finally, we performed a two-sample t-test using the residuals from the linear models to test if their distribution differed between the two cities.

Results and discussion

The UEII is aimed to spatially represent the interplay between the urbanization and biological dimensions across cities (Fig. 1). Despite the use of few variables as inputs to compute the UEII of Lahti in this paper (i.e., LST, built-up, plants and birds’ richness), it showed to provide relevant spatially explicit insights for the city, as well as it did for Xalapa (MacGregor-Fors et al. 2022). The UEII assessment for Lahti showed higher ecological integrity (mean: 0.10 ± 0.28; min–max: -0.75–0.86) when compared to that of Xalapa (mean: -0.34 ± 0.32; min–max: -0.93–0.85; Fig. 1). The UEII showed to be effective in capturing the spatial heterogeneity of the ecosystem integrity of both cities, showing a gradient of values related to the presence of built-up and green cover, but also related to their biodivesity (through the measurement of plant and bird species richness; Fig. 2). This assessment reaffirmed the importance of including field data for the greenspace network of cities (i.e., largest urban greenspaces) in providing further detail to the UEII values (MacGregor-Fors et al. 2022). This is in line with previous studies showing that the greenspaces of Xalapa exhibit considerably different ecological integrity in terms of their biodiversity, size, and spatial configuration (e.g., Falfán et al. 2018; Lemoine-Rodríguez et al. 2019).

Fig. 1
figure 1

Histograms of the UEII values of Lahti and Xalapa

Full size image
Fig. 2
figure 2

Spatial representation of the UEII (upper and lower left) and the NDVI (upper and lower right) of Lahti and Xalapa

Full size image

Despite Xalapa being considered a green city with high ecological value since the biodiversity it supports is remarkable (González-García et al. 2014; Falfán et al. 2018), its UEII values were significantly lower when contrasted to those of Lahti. Most of the UEII values were below zero in Xalapa, with the highest values of the index distributed mostly in three of the largest greenspaces, as the urbanized area is highly dense (Escobar-Ibáñez et al. 2020). Contrastingly, in Lahti most UEII values were above zero, showing a more homogeneous spatial distribution and higher ecosystem integrity scenarios in the city. Thus, this example shows that the UEII is dependent to the context (i.e., contrast system) and therefore was successful in showing Lahti’s higher ecological integrity regardless of the lower biodiversity it harbors in relation to Xalapa. As shown in this example, the spatial representation of the UEII values together with their derived density curves have the potential to become important tools to produce cross-city comparisons, as well as multitemporal ones for the same urban system (MacGregor-Fors et al. 2022). Moreover, the UEII might be useful to inform local and regional spatial planning in diverse urban environments (MacGregor-Fors et al. 2022). Notably, in order for UEII values to be comparable, similar data feeding the index are required, with careful interpretations when reference systems (native ecosystems) are not alike.

The comparison between the UEII and the NVDI showed that, although the indices exhibit a positive relationship for both cities, there is a considerable amount of variation, indicating that the UEII is providing further information on the ecological condition of sites, which may or not have similar NDVI values (Fig. 3). This is particularly noticeable for NDVI values ≤ 0. Since the NDVI only captures the vegetation greenness, it is not able to fully describe other aspects of the urban greenspaces nor other land covers/uses (Jones and Vaughan 2010). Notably, NDVI is an indicator of “greenness and productivity,” relying on the unique characteristics of plant chlorophyll that strongly reflect near-infrared light (0.7 to 1.1 μm) and absorb visible red light (0.4 to 0.7 μm) amidst the diverse wavelengths captured by satellite sensors (Yengoh et al. 2015). It is essential to bear in mind that satellite sensors detect only the outermost vegetation layer, situated farthest from the ground, emphasizing the ecological significance of this observation; while satellite sensors may capture similar chlorophyll reflections from various urban greenspaces, these areas often harbor distinctly different avifaunas and overall biota (Jones and Vaughan 2010).

Fig. 3
figure 3

Linear and general additive fits between the UEII and the NDVI of Lahti a and Xalapa b

Full size image

Compared to land cover data or radiometric vegetation indices such as the NDVI, the added value of the UEII is its capacity to describe the ecological value of the urban greenspaces (MacGregor-Fors et al. 2022). This adds crucial information to identify priority areas to focus on the provision for local species inside cities, as well as the ecological value of recreation areas. Although the NDVI and other radiometric indices are effective to detect green cover in cities (Lemoine-Rodríguez et al. 2022b; Von Thaden et al. 2021), they do not allow to differentiate between types of greenspaces in ecological terms. This represents an important limitation for urban ecology studies and for informing urban planning. Interestingly, when contrasting the residuals of the linear models between the UEII and the NDVI in both cities (t19867 < 0.001, P > 0.999), we found evidence that the UEII values are not biased in relation to the NDVI, but rather inform in higher detail about the ecological condition of each site across a city. Hence, urban greenspaces that exhibit comparable NDVI values, yet differing in their biodiversity, would be discernible through the UEII if data from these greenspaces were incorporated into the interpolations. Such a differentiation would enable managers, planners, and policymakers to identify such areas for targeted interventions, as necessary.

It is noteworthy to highlight that since the UEII is a customizable index, it offers the possibility to include as many variables as available. Nevertheless, if comparative approaches are to be employed, it is advisable to measure the same variables consistently across cities. Therefore, the UEII can be computed for all cities, irrespective of their size or shape, as long as continuous spatially explicit data can be generated for both their physical (urbanization) and biological (biodiversity) dimensions. However, it is crucial to emphasize the significance of the resolution scale of the raw data employed to produce these continuous spatially explicit layers of information. Moreover, in future assessments, 3D spatial indicators such as built-up height and volume might also be included to improve the representation of the urbanization component of cities, as well as multitemporal models (Lemoine-Rodríguez et al. 2022; Li et al. 2020; Mahtta et al. 2019). One of the main limitations of the current spatially explicit urban ecology is the lack of representation of the social dimension of cities. Novel flexible indicators such as the UEII allow to test how the inclusion of geospatial social data (e.g., census data, human health; income; Rautjärvi and MacGregor-Fors 2024) can improve the representation and understanding of cities in a more integral fashion. On the other hand, a practical limitation of the UEII is the collection of spatially continuous data. While this limitation can be avoided by employing the citywide survey approach (Turner 2003), this type of method remains to be scarce in the urban biodiversity field (Rega-Brodsky et al. 2022). Nevertheless, there is an increasing access to crowd-sourced and official data, which could be interpolated to obtain spatially continuous information based on sample cites.

Conclusions

Findings of this study show that the UEII is effective as a valuable indicator for measuring the ecological integrity of urban areas in diverse environments, exemplified by a Neotropical and a boreal cities. By using readily available remote sensing data and conducting field surveys to encompass the array of environmental conditions and biodiversity present in cities, it is clear that UEII describes the intra-urban environmental heterogeneity related to built-up and greenspaces. Unlike the widely used NDVI –often suggested as a proxy of ecological quality– the UEII provides a more comprehensive understanding of the ecological value of both built-up and greenspaces, making it a promising tool for identifying priority areas for habitat preservation within urban settings, as well as those in need of further improvements. If calculated properly, and considering the limitations given the array of variables used, the UEII can become a useful tool for academic purposes, but also those related to the monitoring of the environmental condition of cities, translating into decision-making and resource allocation at the citywide level. For example, city managers can pinpoint greenspaces of significant biodiversity relative importance to reinforce their development and maintenance efforts. Simultaneously, they can identify greenspaces that may appear sizable but lack substantial species diversity. This could lead to the formulation of management and enhancement plans for the greenspace networks. Furthermore, UEII measurements across cities can provide urban planners with insights into the environmental conditions of both grayspaces and greenspaces in areas intended for urbanization. They can also draw valuable lessons from previously assessed zones, contributing to the ongoing sustainable development of cities. Additionally, cities could calculate their UEII over time, creating a time series that graphically illustrates the city’s ecological development. This approach would enable the measurement of environmental initiatives and budget allocations aimed at establishing more ecologically friendly, biodiverse, and livable urban environments.