Urban meadows as an alternative to short mown grassland: effects of composition and height on biodiversity

Abstract There are increasing calls to provide greenspace in urban areas, yet the ecological quality, as well as quantity, of greenspace is important. Short mown grassland designed for recreational use is the dominant form of urban greenspace in temperate regions but requires considerable maintenance and typically provides limited habitat value for most taxa. Alternatives are increasingly proposed, but the biodiversity potential of these is not well understood. In a replicated experiment across six public urban greenspaces, we used nine different perennial meadow plantings to quantify the relative roles of floristic diversity and height of sown meadows on the richness and composition of three taxonomic groups: plants, invertebrates, and soil microbes. We found that all meadow treatments were colonized by plant species not sown in the plots, suggesting that establishing sown meadows does not preclude further locally determined grassland development if management is appropriate. Colonizing species were rarer in taller and more diverse plots, indicating competition may limit invasion rates. Urban meadow treatments contained invertebrate and microbial communities that differed from mown grassland. Invertebrate taxa responded to changes in both height and richness of meadow vegetation, but most orders were more abundant where vegetation height was longer than mown grassland. Order richness also increased in longer vegetation and Coleoptera family richness increased with plant diversity in summer. Microbial community composition seems sensitive to plant species composition at the soil surface (0–10 cm), but in deeper soils (11–20 cm) community variation was most responsive to plant height, with bacteria and fungi responding differently. In addition to improving local residents’ site satisfaction, native perennial meadow plantings can produce biologically diverse grasslands that support richer and more abundant invertebrate communities, and restructured plant, invertebrate, and soil microbial communities compared with short mown grassland. Our results suggest that diversification of urban greenspace by planting urban meadows in place of some mown amenity grassland is likely to generate substantial biodiversity benefits, with a mosaic of meadow types likely to maximize such benefits.

Treatments D2 and D3 (grass/forb mixes) plots were initially sown at 4g/m 2 , and bare patches were over-sown at the end (autumn) of the first growing season at the same density. Treatment D1 (grass only) plots were initially sown at 4g/m 2 and most (two plots were excluded as they established adequately: H3 at Goldington Green and Bramingham) were sprayed and re-sown at 20 g/m 2 at the end (autumn) of the first growing season.  Biomass estimates for invertebrates sampled in the meadow experiment were obtained using data from invertebrate samples taken across Bedford, Luton and Milton Keynes in the summer (July and August) of 2013. Invertebrates were sampled in 243 plots (each 25 m 2 in area) across 78 urban green spaces using a sweep net and vacuum sampler, identified to order level primarily (see table for details) and stored in separate vials for each taxon. Of this collection, samples from 18 sites that reflected a range of vegetation types and site sizes were selected to estimate biomass. From this subsample, up to 48 individuals (based on availability of specimens) were selected from each taxon, excepting Acari, which had not been separated from the samples. Length of each individual was measured under the microscope with a graticule to the nearest 1/10th mm. Specimens were then oven-dried overnight (16 hours) at 60 o C. Each individual organism was weighed on a microbalance (in mg to 4 dp). Collembola were too light to measure individually, and were instead weighed in groups of 24 and an average weight derived. Where there were too few specimens (< 5) to derive an average weight for a particular taxon, an average weight was derived as a surrogate, using the weights of taxa with similar physical properties.
Dry weight per individual organism in a taxon used to estimate biomass across the treatment plots. The number of specimens is the number used to derive the dry weight. Where this is blank, a surrogate was used, indicated in the final column. ix Section S1: Extended method details: supplement to main manuscript Methods: Invertebrates

Phylum
Aboveground invertebrates were sampled twice in all plots: first in the summer (June/July) and then the autumn (early September) of the second year post-sowing. Sampling was scheduled to occur at least a week after mowing, but unplanned changes in the council maintenance schedule meant that a subset of plots (n = 4, September) had been recently mown prior to sampling. Sweep netting (primarily targeting taxa that can fly) was undertaken along a 20 m transect through the middle of the plot, using a white net (60 cm diameter). Vacuum sampling (for invertebrates on the ground and vegetation) was undertaken at two points in each plot using a cordless leaf blower/vac (airspeed = 320 km/h; 11 cm diameter nozzle). Sampling locations were away from the sweep net transect and at least 2 m apart.
Sampling was conducted at all vegetation heights in a sampling container (0.5 m in diameter and 1.5 m high) that prevented the escape of aerial insects.
The potential of meadows to support overwintering invertebrates was assessed through sampling conducted in February 2016 at sites at the end of their second-year post-establishment (Brickhill Heights, Chiltern Avenue, Bramingham Road and Goldington Green). At Goldington Green only three plots were available for sampling as four plots had experienced unscheduled rotovation by greenspace management teams. Sweep netting was conducted as in summer but vegetation was too compacted and damp for effective use of the blower/vac. Hand-searching was thus conducted in two, 1 m 2 quadrats in each plot, off the sweep net transect and at least 2 m apart. Searches involved a sequence of beating, cutting, collecting and sieving vegetation. A pooter/aspirator was used to extract invertebrates from the ground surface and collected vegetation, by orally sucking them into collecting tubes. Each quadrat was surveyed for a maximum of 50 minutes, with equal time allocated to each quarter of the quadrat and a stopping rule used for each survey method.
All invertebrate specimens were stored in 80% industrial methylated spirits prior to identification to order (class for Chilopoda, Diplopoda and Gastropoda) using a binocular microscope. Order is a coarse resolution, but sorting to this level allowed an assessment of all collected taxa simultaneously at a level directly comparable between sites and studies. The numbers of each taxon were counted precisely except where one taxonomic group (usually Acari or Collembola) exceeded approximately 200 individuals, when they were estimated by precisely counting two or three groups of 10 or 100 individuals (depending on abundance) and then estimating the number of such groups in the entire sample. Adults and larvae from each taxon were combined to order level for analysis. Coleoptera are a taxonomically and functionally diverse order of insects, and were ubiquitous across plots. Coleoptera adults from the summer and autumn samples and from the three sites with a full set of treatments (Chiltern Avenue, Bramingham Road, Jubilee Park) were identified to family to provide a taxonomically and functionally more finely resolved assessment of differences in diversity and composition across the plots. Finally, invertebrate biomass was estimated based on an average oven-dry weight of approximately 30 individuals for each order, or the closest taxonomic grouping if too few specimens were available, using samples collected from greenspace in the surrounding urban landscape (Table S3).
Total DNA was extracted from 250 mg soil using the PowerSoil DNA Isolation kit (MoBio, USA) according to the manufacturer's instructions. For each sample, 10 ng of DNA was used to amplify either the ITS region using the ITS fungal primer pair ITS3 and ITS4 (White et al. 1990), or the 16S rRNA gene using the 16S bacterial primer pair 515f and 806r (Caporaso et al. 2011). The two primer sets were modified at the 5′ end with Illumina Nextera Index Kit v2 adapters. PCR reactions were performed in a reaction volume of 25 μl, containing Q5® Hot Start High-Fidelity 2X Master Mix (New England Biolabs) and 0.5 µM of each primer. Thermocycling consisted of an initial denaturation at 98 o C for 30 s followed by 30 cycles of 98 o C for 10 s, 50/57 o C (16S/ITS) for 15 s and 72 o C for 20 s with a final extension at 72 o C for 5 min. The libraries were sequenced using the Illumina MiSeq Reagent Kit v3 (600-cycle). Following sequencing, Trimmomatic v0.35 was used to remove low-quality bases from the sequence ends (Bolger et al. 2014).
The following steps were then performed using USEARCH and UPARSE software (Edgar 2010(Edgar , 2013. Paired-ends reads were assembled by aligning the forward and reverse reads, trimming primers and quality filtering (-fastq_maxee 0.5). Unique sequences were sorted by abundance and singletons were discarded from the dataset. Sequences were clustered to Operational Taxonomic Units (OTUs) at 97% minimum identity threshold. Taxonomy was assigned using Quantitative Insights into Microbial Ecology (QIIME 1.8) (Caporaso et al. 2010) and the Greengenes reference database for 16S (McDonald et al. 2011), or the UNITE database for ITS (Kõljalg et al. 2013). Table S4: Total abundance of each aboveground invertebrate taxon collected in summer, autumn and winter in the second growing season of the meadow treatments. Collection methods in winter were different, from those in summer and autumn (see main manuscript Methods: Invertebrates and Appendix S1: Sect. S1) and sampling was undertaken on a subset of sites (Table S1).  xiii Figure S2: Boxplot of total abundance per plot by height treatment (H1 = short, H2 = medium, H3 = tall) of invertebrates collected during winter surveys (February 2015). Figure S3: (a) Total nitrogen, and (b) total carbon (both measured as percent dry weight), by treatment and by depth. Bars are organised from short (left) to tall (right) treatments, with small gaps between the height groups, and diversity treatment is indicated by grey shading; light grey = low diversity, medium grey = medium, black = high diversity. White bars represent the unmanipulated control. Bars are the mean per treatment combination with standard deviation bars. Table S6: Non-sown plant species detected in treatment year 2 and their lifecycle and strategy according to Grime's CSR theory (Grime 1979, Grime et al. 1995. Each species is denoted as C = competitor, S = stress tolerator, R = ruderal, or some combination of these. ND indicates no data are available. Life cycles are denoted as per = perennial, ann = annual, biann = biannual. xviii Section S3: Bare ground cover extended results details: supplement to main manuscript Discussion: Plants Figure S4: Percent of bare ground cover (mean and standard deviation) in the experimental treatments and unmanipulated control. Bare ground cover was recorded as part of the plant surveys, using Domin cover estimates (for details see main manuscript Methods: Botanical surveys).
Linear mixed effects models were constructed in lme4 (Bates et al. 2015) with height (three levels) and diversity (three levels) treatments included as fixed effects and site as a random intercept, using maximum likelihood parameter estimation. P-values for the main fixed effects were extracted using the package lmerTest (Kuznetsova et al. 2017), with degrees of freedom based on Satterthwaite's approximation. Within-treatment pairwise comparisons were determined with post-hoc tests using least-squares means in package lsmeans (Lenth 2016).
Section S4: Vegetation clippings biomass extended results details: supplement to main manuscript Discussion: Plants

Method
We estimated biomass of vegetation clippings (arisings) from a complete cut of the medium and tall plots prior to termination of the experiment (3 November 2015). Samples were taken from Chiltern Avenue (year 3 post-sowing) and Jubilee Park (year 2 post-sowing). These sites were selected as they retained a full set of un-mown treatments at this point of sampling.
In each medium and tall treatment plot at each site, all vegetation within two 0.5 m x 1 m quadrats was cut off at the base and collected in to large, plastic bags. Quadrats were placed to capture remaining areas of non-trampled vegetation. Samples were stored in freezers for two days prior to processing.
All samples were weighed wet, cut into pieces and thoroughly mixed up. A subsample of each was ovendried at 60 o C for 60 hours. Subsamples were weighed before and after drying to measure moisture content of the sample. The moisture loss from the subsample combined with the final dry weight was used to back-calculate the total estimated dry weight of each sample. The two plot replicates were combined to estimate dry weight per 1 m 2 per plot. Table S7: Estimated dry weight in kilograms per 1 m 2 of vegetation clippings from medium (H2) and tall (H3) experimental meadow plots. D1 to D3 represent the low to high diversity treatments ( Figure 1).