Beyond the urban-rural gradient: Self-organizing map detects the nine landscape types of the city of Rome
Introduction
Urban biodiversity regards the whole set of plants and animals, as well as their interactions, constituting the biological communities of an ecosystem (Puppim de Oliveira et al., 2014). The standard framework for the structure of biodiversity in cities is the urban-rural gradient (McDonnell and Pickett, 1990): A heavily settled centre, a more loosely settled suburban belt with many open areas, and the rural/natural outskirts, with a progressive increasing gradient of naturality. Recently Cadotte et al. (2017) have questioned the effectiveness of this traditional model of urban structure, since the situation can be much more complex. The authors underscore that this issue should be investigated with better designed data, able to take into account the complexities of the urban structure and function. In fact, although urban biodiversity has been widely studied (Schwartz et al., 2006; McDonnell and Hahs, 2008; Nilon, 2011; Malkinson et al., 2018), fewer studies deal with the patterns of species distribution within an urban environment. In Rome, due to its local authorities’ interest in developing a system of protected areas, for fostering the collection of data on the city’s plant and animal species, very good basic data (atlases) are available (Cignini and Zapparoli, 1996; Zapparoli, 1997; Capotorti et al., 2013). The data for plants (Celesti-Grapow, 1995) have been thoroughly analysed regarding determinants of species’ richness and number of aliens (Ricotta et al., 2001; Celesti-Grapow et al., 2006). However, a study linking the pattern of composition of the flora with the city’s urbanistic structure and environmental variables is still lacking. Rome has a complex urbanistic structure, characterised by many shortcomings in urban planning, a high number of open areas (from abandoned fields to archaeological sites, even in the inner areas) and a heterogeneity of habitat patches. Thus, the ultimate definition of the patterns of species distribution in this context may be difficult. In this study we resorted to a machine learning technique, which is particularly suited for classifying our complex dataset: Artificial Neural Networks (ANNs), more in particular a Self-Organizing Map (SOM). First applications of SOMs in ecology date back to Chon et al. (1996) in freshwater ecology. Since then, SOMs have been applied to represent or classify observations and ecological data (Park et al., 2006; Astel et al., 2007; Mele and Crowley, 2008; Chon, 2011), but the application to urban ecosystems is very rare (Bergerot et al., 2011).
This study aims to analyse the structure of Rome’s flora, relating it to environmental and urbanistic features. In particular, we aim to answer to these questions:
- i)
What are the flora’s distribution patterns?
- ii)
Do these patterns follow an urban-rural gradient, a centreless mosaic structure or a composite pattern?
- iii)
Which predictors are able to explain these patterns?
Section snippets
Study area
Rome is situated in the centre of the Italian peninsula, roughly halfway between the Tyrrhenian sea, to its West, and the Apennine mountain chain, to its East. The metropolitan area of Rome extends for about 1285 km2 (Comune di Roma, 2018). We considered only the inner part of this area (>300 km2), which includes most of the city’s inhabitants; that is, those dwelling inside the so-called Grande Raccordo Anulare (henceforth: GRA), the motorway ring surrounding the city, with an average diameter
Clusters of squares and indicator species
We selected 9 clusters as optimally representing the city’s structure on the diagram obtained with the method of the Fusion Level and on the base of ecological definiteness (Fig. A1 in Annex 1). SOM results are largely comparable with those of Cluster Analysis. Nonetheless, clusters’ output returned by SOM appear more significant compared to Cluster Analysis from an ecological perspective (Fig. 2). Clusters are better defined, while intermediate squares are assigned more logically to their
Discussion
Floristic data are particularly suited for classificatory purposes but have drawbacks when used to infer ecological processes at community (productivity, photosynthesis, etc.) and at landscape level (i.e. fragmentation). For these purposes they represent only a guide and other types of research (permanent plots, long-term studies, etc.) are necessary. Identifying homogenous landscape units is a preliminary step in any study concerning each of these processes. In other words, classification on
Conclusions
Previous researches highlight the complexity of Rome’s urban structure (Frondoni et al., 2011; Salvati, 2015; Salvati et al., 2016). Rome expanded exponentially after becoming capital of Italy in 1870, with a huge influx of immigrants and dramatic problems regarding the building of a proper network of infrastructures (Krumholz, 1992). As if its rapid expansion were not enough, Rome has been characterised by lack of urban planning, extensive speculation and a corrupt political system. This model
Declaration
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Acknowledgments
We thank Andrea Palmeri for helping in managing the data base, Marco Martinoli for providing valuable tips in the use of QGis software, Paolo Sarandrea (Settore Pianificazione Urbanistica) for his availability during maps consultation, two anonymous reviewers for the useful comments, and Martin Bennett for the English revision and editing of the manuscript.
References (83)
- et al.
City profile: Berlin
Cities
(2014) - et al.
Comparison of self-organizing maps classification approach with cluster and principal components analysis for large environmental data sets
Water Res.
(2007) - et al.
Paradoxical environmental conservation: Failure of an unplanned urban development as a driver of passive ecological restoration
Environ. Dev.
(2017) - et al.
How can wastelands promote biodiversity in cities? A review
Landsc. Urban Plan.
(2014) - et al.
The vanishing landscape of the Campagna Romana
Landsc. Urban Plan.
(1993) Self-organizing maps applied to ecological sciences
Ecol. Inform.
(2011)- et al.
Patternizing communities by using an artificial neural network
Ecol. Modell.
(1996) Urbanization in developing countries: current trends, future projections, and key challenges for sustainability
Technol. Soc.
(2006)- et al.
A landscape analysis of land cover change in the Municipality of Rome (Italy): spatio-temporal characteristics and ecological implications of land cover transitions from 1954 to 2001
Landsc. Urban Plan.
(2011) Temporal analysis of the Brussels flora as indicator for changing environmental quality
Landsc. Urban Plan.
(2001)
Roman impressions: contemporary city planning and housing in Rome
Landsc. Urban Plan.
From rural-urban gradients to patch – matrix frameworks: plant diversity patterns in urban landscapes
Landsc. Urban Plan.
Application of self-organizing maps for assessing soil biological quality
Agric. Ecosyst. Environ.
Application of a self-organizing map to select representative species in multivariate analysis: a case study determining diatom distribution patterns across France
Ecol. Inform.
Topological analysis of the spatial distribution of plant species richness across the city of Rome (Italy) with the echelon approach
Landsc. Urban Plan.
Informal urban green space: a trilingual systematic review of its role for biodiversity and trends in the literature
Urban For. Urban Green.
Urban containment in action? Long-term dynamics of self-contained urban growth in compact and dispersed regions of southern Europe
Land Use Policy
Lost in complexity, found in dispersion: ‘Peripheral’ development and deregulated urban growth in Rome
Cities
Biotic homogenization of the California flora in urban and urbanizing regions
Biol. Conserv.
Urbanophob — urbanoneutral — urbanophil Das Verhalten der arten gegenüber dem lebensraum stadt (urbanophob - urbanoneutral – urbanophil behaviour of species concerning the urban habitat)
Flora
Urban development and insect biodiversity of the Rome area, Italy
Landsc. Urban Plan.
Urban spatial structure
J. Econ. Lit.
Flora e vegetazione dei muri di Roma (Flora and vegetation of the walls of Rome)
Annali di Botanica
Plant invasion along an urban-to-rural gradient in northeast Connecticut
J. Urban Ecol.
Biodiversity in the city: key challenges for urban green space management
Front. Ecol. Environ.
An updated checklist of the vascular flora native to Italy
Plant Biosyst. - Int. J. Deal. With All Asp. Plant Biol.
Landscape variables impact the structure and composition of butterfly assemblages along an urbanization gradient
Landsc. Ecol.
Fitoclimatologia del Lazio (Phytoclimatology of Lazio)
Fitosociologia
Numerical Ecology with R
A geological model of the buried Tiber River valley beneath the historical centre of Rome
Bull. Eng. Geol. Environ.
The growth of the city: an introduction to a research project
The determination of gradients in the growth of the city
Publ. Am. Soc. Soc.
Are urban systems beneficial, detrimental, or indifferent for biological invasion?
Biol. Invasions
Exploring biodiversity in a metropolitan area in the Mediterranean region: the urban and suburban flora of Rome (Italy)
Plant Biosyst. - Int. J. Deal. With All Asp. Plant Biol.
Conflict between insect conservation and public safety: the case study of a saproxylic beetle (Osmoderma eremita) in urban parks
J. Insect Conserv.
Determinants of native and alien species richness in the urban flora of Rome
Divers. Distrib.
The vascular flora of Rome
Plant Biosyst. - Int. J. Deal. With All Asp. Plant Biol.
Temporal floristic variations as indicator of environmental changes in the Tiber River in Rome
Aquatic Ecol.
Cited by (15)
Integrating biodiversity as a non-human stakeholder within urban development
2023, Landscape and Urban PlanningAn analysis of multiple ecosystem services in a large-scale urbanized area of northern China based on the food-energy-water integrative framework
2023, Environmental Impact Assessment ReviewCitation Excerpt :Thus, it can map ESs that occur together in space explicitly. The advantages of SOM in visualizing clustering patterns of complex data have been widely recognized in environmental and ecological sciences (Dittrich et al., 2017; Juntunen et al., 2013; Levers et al., 2018; Fratarcangeli et al., 2019; Zhang and Guo, 2020). SOM consists of 2 layers: the input layer and the output layer (or competition layer).
Muscle pigmentation in rainbow trout (Oncorhynchus mykiss) fed diets rich in natural carotenoids from microalgae and crustaceans
2021, AquacultureCitation Excerpt :The optimal number of clusters for the SOM grid was obtained by performing a hierarchical cluster analysis using the Unweighted Pair Group Method with Arithmetic Mean (UPGMA) and detecting clusters in the SOM, according to the similarity of the weight vectors (Franceschini et al., 2019; Park et al., 2003). The optimal number of clusters (ranging from 2 to 10) was chosen by the Silhouette Width (SW) index (Fratarcangeli et al., 2019; Rousseeuw, 1987). Basic functions of R environment were used to compute the SW index.
Promoting wildflower biodiversity in dense and green cities: The important role of small vegetation patches
2021, Urban Forestry and Urban GreeningCitation Excerpt :The biodiversity value of patches of herbaceous urban vegetation depends on a range of factors such as their size and connectivity (Liu et al., 2019; Beninde et al., 2015), management and land use history (Lerman et al., 2018; O’Sullivan et al., 2017), and the socioeconomics of their neighborhood (Hope et al., 2003; Leong et al., 2018). There is a growing research interest in understanding the interplay of these factors and uncovering the mechanisms which determine urban plant diversity patterns along urban gradients and in heterogeneous urban landscapes (Bretzel et al., 2016; Fratarcangeli et al., 2019; Malkinson et al., 2018; McDonnell and Hahs, 2013). However, most analyses focus on coarse-resolution landscape level patterns or relatively large urban green spaces (Hope et al., 2003; Hoyle et al., 2018; Kühn et al., 2004; Pysek, 1998).