Study area
This study was carried out in peri-urban areas of the city of Madrid (Spain) (Fig. 1). Madrid is located in the centre of the Iberian Peninsula, at an average altitude of 655 m.a.s.l. The average annual temperature is 15º C and the average annual precipitation is 421 mm (AEMET 2023). The bioclimate is Mediterranean semi-continental, with cold wet winters and hot dry summers (Rivas-Martínez et al. 2017). The soils in the study area can be classified as Anthrosols or Technosols (FAO 2014; Quintana et al. 2022). The city has almost 3,300,000 inhabitants in an area of 604.5 km2 (INE 2023).
Four greenspaces were selected on the outskirts of the city centre (Fig. 1), excluding the south of the metropolitan area to avoid the gypsum substrates. These greenspaces are established on wastelands, abandoned crop fields and landfills, where trees – often pine – and shrubs have been planted, and some areas are irrigated to maintain lawns. Nevertheless, the main vegetation matrix comprises spontaneous ruderal herbaceous plants, distributed in different habitats along gradients of soil disturbance (Molina et al. 2023; Molina et al. 2024). These non-irrigated green areas are subjected to annual mowing to prevent fires, which tend to occur in the dry summer season.
Sampling design and laboratory procedure
In each of the four greenspaces we studied four vegetation types to describe different ruderal habitats that are common in Mediterranean cities. We selected two perennial and two annual communities (Dana et al. 2002; Molina 2022). Sampling was carried out in the spring of 2023 (March-April) when the communities reached their maximum development. The following communities were studied: (1) tall perennial herbs (communities of Malva species, alliance Hordeion leporini); (2) annual medium-tall herb communities on roadsides (association Papaveri rhoeas-Diplotaxietum virgatae, alliance Hordeion leporini); (3) low annual grasslands (association Bromo scoparii-Hordeetum leporini, alliance Hordeion leporini; and (4) low perennial grasslands (community of Dactylis glomerata, class Lygeo-Stipetea) (RivasMartínez 1978; RivasMartínez et al. 2002). Hereafter these communities are called (1) perennial herb communities, (2) roadside communities, (3) annual grasslands and (4) perennial grasslands.
Three 1 m2 quadrats of each vegetation type at least 100 metres apart were surveyed in each greenspace, thus resulting in a study of 48 1m2 quadrats across four greenspaces. In each quadrat, plant species cover was estimated visually, always by the same observer to avoid bias, and the number of individuals (abundance) in each plant species were counted. To determine the productivity, the aboveground biomass of each plant species was collected, dried in an oven at 80°C until its mass was stable, then weighed.
Plant species were identified following Flora iberica (Castroviejo 2012). Each species was also assigned to a biotype (Raunkiaer 1934), identifying the following four biotypes: therophytes, hemicryptophytes, geophytes and chamaephytes. Species were also assigned to a functional group according to the biogeochemical classification established in Valverde et al. (2020). The following functional groups were identified: sulphur accumulators, N-fixers, N-compound-bearing, mucilage accumulators, terpenoid accumulators and silica accumulators. Species were also classified as native or naturalized alien species based on their geographical origin (Bot Mad et al. 2023). Nomenclature of plants used in the text is according to World Flora Online (WFO, 2023).
All the above-mentioned data were used to determine the following variables for each plant community: species density, abundance, alpha diversity (Shannon and Simpson indices), cover of biotypes and cover of functional groups in each plant community. Beta diversity (Sørensen index) was calculated for each pair of communities.
The quantitative content of carbon, hydrogen, nitrogen and sulphur was determined by combustion in a LECO CHNS-932, in leaves and roots of the most abundant plant species in each quadrat. Samples were processed in the Elemental Microanalysis Unit at the Complutense University in Madrid.
Statistical analysis
We used generalized linear models (GLM) to relate vegetation type, here considered as a habitat descriptor, to compositional and structural features related to ecosystem services such as biodiversity, productivity and nutrient cycling. All analyses were performed with R: a language and environment for statistical computing (R Core Team 2022), using the multcomp (v1.4-25, Bretz et al. 2023) and DHARMa (v 0.4.6, Hartig and Lohse 2022) packages for the analyses. The figures were prepared with ggplot2 package (v3.5.0, Wickham et al. 2024) for graphics.