Volume 25, Issue 1 p. 139-151
Open Access

Sown wildflowers between vines increase beneficial insect abundance and richness in a British vineyard

Janine Griffiths-Lee

Corresponding Author

Janine Griffiths-Lee

School of Life Sciences, University of Sussex, Brighton, UK


Janine Griffiths-Lee, School of Life Sciences, University of Sussex, Brighton, UK.

Email: [email protected]

Search for more papers by this author
Balin Davenport

Balin Davenport

School of Life Sciences, University of Sussex, Brighton, UK

Search for more papers by this author
Bradley Foster

Bradley Foster

Agriculture and Environment Department, Harper Adams University, Newport, UK

Search for more papers by this author
Elizabeth Nicholls

Elizabeth Nicholls

School of Life Sciences, University of Sussex, Brighton, UK

Search for more papers by this author
Dave Goulson

Dave Goulson

School of Life Sciences, University of Sussex, Brighton, UK

Search for more papers by this author
First published: 22 September 2022

Funding information: CB Dennis Trust; UK Research and Innovation Future Leaders Fellowship, Grant/Award Number: MR/T021691/1; University of Sussex


  1. Traditional vineyards are generally intensive monocultures with high pesticide usage. Viticulture is one of the fastest-growing sectors of English agriculture, although there is currently limited research on habitat management practices.
  2. In a vineyard in East Sussex, England, we tested five inter-row ground cover treatments on their potential in supporting beneficial insects: two commercially available seed mixes (meadow mix and pollen and nectar mix), a wild bee seed mix (formulated based on pollinator foraging preferences), natural regeneration, and regularly mowed grass.
  3. Over two years, from May to August, we conducted monthly floral surveys and insect surveys using transect walks and pan traps.
  4. The abundance and richness of flowers in the natural regeneration treatment were twice that of the regularly mown inter-row treatment. By year 2, the abundance of “total insects” sampled was significantly higher in the wild bee mix compared to mown. Likewise, there was a significant effect of treatment type on pollinator richness, with a higher mean richness found in wild bee mix. Solitary wasp family richness was highest in the natural regeneration treatment and lowest in the mown treatment.
  5. Given the rapid growth and lack of specific environmental recommendations for British viticulture, we demonstrate a simple and effective approach for supporting beneficial insects and ecosystem services. Promotion of perennial wildflowers through sowing or allowing natural regeneration in inter-row ground cover in vineyards has the potential to boost biodiversity in vineyards on a large scale if widely adopted.


Land-use change due to intensive agricultural practices is a major driver of global biodiversity loss (Newbold et al., 2016). Habitat conversion to vineyards currently threatens biodiversity in many of the world's top wine-growing regions, including South Africa (Fairbanks et al., 2004), California (Merenlender, 2000) and Chile (Armesto et al., 2010). Globally, approximately 7.4 million hectares of land are under vine (OIV, 2019), and these landscapes are generally intensively managed monocultures with high pesticide usage (Urruty et al., 2016). In Great Britain, vineyard coverage has more than quadrupled since 2000 (WineGB, 2021), and currently there is 3800 ha of land under vine. Ninety-eight per cent of British vineyards are in England (WineGB, 2021), and viticulture is considered one of the fastest-growing sectors of English agriculture (South Downs National Park Authority [SDNPA] 2021).

European agri-environmental schemes (AES) include recommendations for the sowing of wildflowers to provide resources for pollinators (e.g., DEFRA, 2020). Wildflower strips can provide important resources for pollinators in agricultural environments (Blaauw & Isaacs, 2014), aiding pollinator conservation and promoting pollination services (Korpela et al., 2013). However, there are currently no vineyard-specific recommendations under the UK AES. Grapevines are not dependent on pollinators, yet the positive effects of wildflowers in vineyards for pollinators have received attention in many traditional wine-growing regions. For example, studies on inter-row floral plantings in vineyards in Europe, California, and South Africa conclude that increased wild bee richness, abundance and functional traits (Kehinde & Samways, 2014; Kratschmer et al., 2019; Kratschmer et al., 2021; Wilson et al., 2018) regardless of organic or conventional practices (Kratschmer et al., 2021).

Habitat management in agroecosystems also provides essential resources for natural enemies (Landis et al., 2000), increasing the abundance of beneficial insects such as hoverflies, lacewings and ladybirds (Tschumi et al., 2016) and reduces pest damage (e.g., Tschumi et al., 2015). Such ecologically-based pest management as a part of integrated pest management is increasingly considered an alternative to pesticide use (Wilson & Danne, 2017). Agro-ecological approaches to vineyard habitat management also promote natural pest control, with inter-row wildflower strips increasing the abundance of insect parasitoids (Judt et al., 2019). Natural regeneration of floral communities in a vineyard has also been found to promote hoverfly diversity (Pétremand et al., 2017). Adult hoverflies are effective pollinators (Doyle et al., 2020), and zoophagous hoverflies have predatory larvae that are also pest control agents (Wotton et al., 2019). However, some studies have found that although floral plantings attract natural enemies and increase parasitism, this does not necessarily translate into effective pest reduction (Berndt et al., 2006; English-Loeb et al., 2003; Pétremand et al., 2017).

Reduced mowing regimes have a positive effect on insect abundance and diversity (Wastian et al., 2016) and indeed on the abundance of parasitic wasps in vineyards (Zanettin et al., 2021). Due to high levels of disturbance, the abundance of natural enemies can be low in agroecosystems (Landis et al., 2000), and perennial grasses may provide refuge for natural enemies during such disturbances (Daane et al., 2018). In an Australian vineyard, the abundance and diversity of parasitoids were higher in vines surrounded by perennial grasses, and predation of the pest Epiphyas postvittana was greater (Danne et al., 2010). Likewise, in Mediterranean vineyards, the abundance and richness of parasitoid species were higher in the natural regeneration or “managed weed” treatment areas (Möller et al., 2020). In addition to supporting pollination and pest control services, there are numerous other ecosystem service benefits associated with sowing wildflowers. Wildflowers can provide soil protection, weed suppression, biodiversity enhancement and increased aesthetics (Fiedler et al., 2008) and enhance soil fungal networks, leading to increased nutrient availability for grapevines and increasing tolerance to abiotic stress (Trouvelot et al., 2015). For those crops not dependent on pollinators, such as grapevines, wildflowers can support biodiversity and the pollination of wild plants and other insect-pollinated crops in the wider landscape.

The viticulture industry in Great Britain is experiencing rapid growth (SDNPA, 2021) with thus-far limited research into agroecological management. To the best of our knowledge, this is the first published study on the effects of wildflowers on biodiversity in a British vineyard. Here we focus on the intrinsic value of beneficial insects in supporting a healthy and biodiverse ecosystem, considering changing habitats in the South Downs National Park, England. There has been a 90% increase in the coverage of vines in the South Downs National Park since 2016 and an estimated further 40,000 ha of the park is suitable for viticulture under climate change projections (SDNPA 2021). Considering these projections, appropriate environmental and sustainability management of this sector is vital. By experimentally increasing floral plantings in inter-row spaces, we evaluate their potential to increase the abundance and richness of beneficial insects in a British vineyard. We aimed to determine: (1) If inter-row sowing of wildflower seed mixes increases insect, pollinator and solitary wasp abundance and richness; (2) Which inter-row ground cover treatment interventions best encourage beneficial insects and (3) The effectiveness of natural regeneration in encouraging beneficial insects and floral establishment, compared to mowing inter-row spaces.


Study site and inter-row treatments

The study took place at a vineyard estate in East Sussex, UK (Lat/long 50.797, 0.125). The vineyard is located in the South Downs National Park, on lime-rich chalk soil and land previously used as conventionally managed arable farmland. The experimental site within the vineyard contains 37 rows of established vines, allowing 36 rows of inter-row ground cover treatments. Prior to the study, the treatment rows were regularly mowed. Five different inter-row treatments were tested, including three different wildflower mixes, natural regeneration, and control of mown grass. Supplementary information S1 lists the wildflower and grass mix compositions, indicating which of the flowering species germinated and in which year of the study.

Treatment rows were 140 m in length and were paired, so two rows of the same treatment were placed together and the 18 pairs were then randomly allocated to one of five different inter-row ground cover treatments as replicates (Supplementary information S2 shows the arrangement of treatment rows). The “meadow mix” treatment was based on a wildflower mix recommended under UK AES for the establishment of flower-rich margins and plots. Meadow mix contains both perennials and grasses, and is ideal for chalky and limestone soils. It was sown on four replicates at a rate of 4 g/m2. The “wild bee mix” treatment was formulated based on existing literature and personal communications identifying biennial and perennial flowers that attract a range of wild pollinator species and provide flowering cover over the longest season. We chose to create mixes with mostly perennial species as they tend to produce more pollen and nectar than annual plants (Hicks et al., 2016), and last multiple seasons. A wild bee mix was sown on four replicates at a rate of 4 g/m2. The “pollen and nectar mix” treatment is based on a mix recommended under the Countryside Stewardship Scheme for the planting of nectar-rich flowering species under its AES. The mix was grass seed-free and contained six nectar-rich flowering species. It was sown on four replicates at a rate of 1 g/m2. The “natural regeneration” treatment strips were permitted to regenerate naturally from flowering plant species already present at the vineyard site. Natural regeneration was allocated to three replicates. The “mown” treatment strips were mown approximately every two weeks through spring and summer, in line with the management of the vineyard outside of the experimental site. The mown treatment was allocated to three replicates.

Except for the “mown” and “natural regeneration” treatments, grass was removed with a disc cutter, and wildflower mixes were sown in May 2016 by mechanically broadcasting along the inter-rows. Seeds were supplied by Agrii (United Agri Products Ltd & Masstock Arable Ltd, Cheltenham, UK). Growth was cut back in August and cuttings removed.

Insect surveys

Insect surveys consisted of transect walks and pan trapping. Surveys took place monthly from July to August 2016 and May to August 2017 between 10:00 and 16:00, on days with a minimum temperature of 13 °C, low wind and no precipitation.

One inter-row of every treatment pair was randomly chosen for a transect walk and walked at a pace of approximately 10 m/min. At approximately 30 m intervals on the transect walk, a butterfly net was swept into the inter-row ground cover foliage for 20 s (collecting wild bees and hoverflies only) and a pooter was used to collect smaller species from the net. Insects were collected in jars containing ethyl acetate.

A pan trap set was placed on the ground under the vine in between each treatment pair at the same halfway point of each row, to collect insects actively foraging in the flowers. A 24-hour period was chosen each month with a low chance of rain, and daytime temperatures above 13 °C. Pan traps were spray-painted by hand and a set consisted of four 750 ml plastic food containers (Go Packaging Products Ltd, UK), one sprayed white, one yellow, one pink (Rust Oleum spray paint Direct to Plastic White/Sun Yellow Gloss/Berry Pink Gloss) and one blue (Plasti kote Pacific Blue Gloss). An asterisk was drawn in permanent marker pen (Sharpie, Sanford L.P, US) on the inside of the pan traps as a “nectar guide”. Pan traps were ¾ filled with water and a squirt of natural-fragranced washing-up liquid (Ecover, Malle, Belgium).

Identification of samples

Using pan trap samples, the abundance of bumblebees, solitary bees, honeybees, solitary wasps (including parasitoid wasps), social wasps, beetles, hoverflies and “other” (non-syrphid) flies were counted. No butterflies or moths were caught. In this paper, our definition of “solitary bees” includes non-corbiculate bees that are solitary or eusocial, and those that do not fall under the bumblebee (Bombus) or honeybee (Apis) groups. From net and pan trap samples, all hoverflies, bumblebees and solitary bees were identified to species level (10 specimens were only recorded to a broad group as they were unidentifiable). Solitary wasps were recorded to family level for July 2016 and 2017 pan traps only (identification for this group is very time-consuming, so a month with a high occurrence of wasps was chosen as a representative sample).

Floral surveys

During the transect walk, floral surveys were also conducted at 30-m intervals, using 1x1m quadrats. All blooming inflorescences were identified to species and percentage coverage of floral units was estimated (the floral unit could comprise of a single flower, or cluster of flowers within a head, such as Trifolium spp.). Grasses and non-flowering plants were not identified. Using the percentage coverage of each species present in the 1x1m quadrat, the average for each species was calculated over three quadrats per row and recorded as average species to the nearest integer + 1 (this allowed us to account for the rarer flower species only occurring in one quadrat).

Data analysis

Data analysis was conducted in R (R core team, 2020). Data from 2016 and 2017 (henceforth year 1 and year 2) were analysed separately due to differences in flower abundance and diversity that occurred between years due to the establishment of perennial flowers. Data from the two sampling methods (pan trap and transect walk) was also analysed separately. “Total insect abundance” included counts of solitary bee, bumblebee, honeybee, hoverfly, solitary wasp, social wasp, beetles and “other” (non-syrphid) flies. “Total insect abundance” was only available for pan trap methods, as transect walks only recorded bees and hoverflies. Hoverfly and bee richness considered the number of species, and data from both insect groups were combined into a single measure of “pollinator richness” for analysis. Solitary wasp (including parasitoid wasp) richness was analysed at the family level and for July in years 1 and 2. For flowering plants, Shannon's diversity index was calculated for each row for each month and included both sown (i.e., included in the wildflower mix) and unsown flowers (spontaneous).

Effects of inter-row treatment on total insect abundance

A Shapiro–Wilk normality test was conducted to test for parametric data. Generalized Linear Mixed Models (GLMMs) were constructed using the lme4 package, zero-inflated models using glmmTMB and graphs were created using ggplot2. Models of best fit were chosen based on AIC values followed by diagnostic residual plots to ensure they conformed to underlying model assumptions. ANOVAs were then performed, comparing full and reduced models, and results were reported as chi-square and p values. Then Tukey's Honest Significant Difference Test was used post-hoc to compare inter-row treatments. To investigate the effects of inter-row treatment on total insect abundance, GLMM's with a negative binomial family was constructed with treatment, month and diversity of flowers as predictor variables. Row number was included as a random variable (to account for repeated measures).

Effects of treatment on richness of hoverflies, bees and solitary wasps

To investigate the effects of inter-row treatment on richness of pollinators (number of species of bees and hoverflies), treatment, month and diversity of flowers were set as predictor variables and row number as a random variable. Year 1 transect walk and year 2 pan trap were both analysed using GLMM with Poisson distribution, whereas zero-inflated GLMM with Poisson distribution was constructed for year 1 pan trap and year 2 transect walk data. To test the effects of inter-row treatment on the richness of solitary wasps (number of families), a Kruskal-Wallis H test was conducted.

Community dissimilarity analysis

Community dissimilarity analysis was performed to assess the (i) floral, (ii) bee and hoverfly and (iii) solitary wasp communities of the inter-row treatments over the two years of study. Jaccard dissimilarity was performed on the floral community matrix, and Bray-Curtis was performed on bee and hoverfly matrix, and also on the solitary wasp community matrix (Vegan package), followed by Non-metric Multidimensional Scaling (NMDS) using the MASS package to create an NMDS matrix. The significance of key species/families was tested with 999 permutations and adjusted using Bonferroni corrections. To analyse “Treatment” and “Year”, a Permutational Multivariate Analysis of Variance (PERMANOVA) was performed on the interaction between the two variables. A PERMANOVA tests differences in similarities, and a significant result suggests that groups differ in their location and/or relative dispersion (Assis et al., 2013). When PERMANOVA results were significant, a Permutation Analysis of Multivariate Dispersion (PERMDISP) was performed on the community matrix (Jaccard/Bray-Curtis), determining if there was variability in dispersion, possibly accounting for significant results seen in the PERMANOVA.


Wildflower establishment

Over two years, 50 species of flowering plants spanning 17 families were identified (Figure 1), of which 24 were sown species and 26 were established via spontaneous natural colonisation. The floral diversity for each of the five treatments increased from year 1 to year 2 (Figure 2c), as did the abundance and richness of the sown flower species (Figure 2a,b), which would be expected with the establishment and flowering of perennials in the year following sowing. In year 2, the diversity of wildflowers was greatest in inter-rows sown with meadow mix, followed by wild bee mix and pollen and nectar mix (Figure 2c) and greatest sown species richness and abundance were seen in meadow mix and wild bee mix (Figure 2a,b).

Details are in the caption following the image
Flowering plant species occurrence by treatment. Heatmap presenting all flowering plant species recorded at the study site within the five inter-row ground cover treatments, combining data from years 1 and 2. Those flowering species presented in bold with an asterisk are unsown and colonised naturally. Based on mean species abundance, square root transformed for visualisation purposes.
Details are in the caption following the image
Mean (a) abundance (b) richness and (c) diversity of flowering plant species. Species mean abundance and species mean richness of sown and unsown flowering plant species are measured. Diversity (Shannon's diversity index) includes of both unsown and sown flowers. Flowers were recorded in five inter-row ground cover treatments, in both year 1 and year 2 of the study.

The floral diversity, richness and abundance of flowers in natural regeneration were greater than in the mown treatment (Figure 2a–c). In year 1, the mean richness of the flowering species found in natural regeneration was 3.17 species (SE ± 0.27), compared to 1.50 species (SE ± 0.18) for mown. By year 2, species richness of natural regeneration was a mean of 6.33 species (SE ± 0.43), compared to 3.42 species (SE ± 0.37) for mown (Figure 2b). Mean abundance of flowering species in year 1 was very similar between natural regeneration (mean 7.00 species SE ± 0.65) and mown (mean 7.83 species SE ± 1.45), but by year 2, species abundance had doubled to mean 14.58 species (SE ± 0.81) in natural regeneration, whereas mown strips remained roughly the same as in the first year of the study (mean 7.33 species SE ± 0.59) (Figure 2a). The diversity of flowering species in natural regeneration in year 1 was mean 1.05 species (SE ± 0.15) and in year 2 was mean 1.52 species (SE ± 0.18; Figure 2c). This is compared to year 1 mown's mean 0.31 species (SE ± 0.25) and year 2 mean of 1.01 species (SE ± 0.19; Figure 2c).

NMDS analysis showed that inter-row ground cover treatment floral communities differed significantly (Figure 3; PERMANOVA: F4,76 = 4.84, p < 0.001), and analysis of dispersion suggested that this was due to variation between treatments rather than within treatments (PERMDISP: F4,81 = 2.065, p = 0.101). All mixes showed high levels of overlap with other inter-row treatments in terms of floral composition (Figure 3). Likewise, NMDS analysis showed that year 1 and 2 floral communities differ significantly (PERMANOVA: F1,76 = 9.24, p < 0.001), and analysis of dispersion suggested that this was due to variation between years 1 and 2 rather than within years (PERMDISP: F1,84 = 3.713, p = 0.061). Nine flowering plant species showed significant presence within the ordination and were significantly associated with specific inter-row treatments and years of the study (Figure 3). Four of these nine were sown as part of the wildflower mixes: Centaurea nigra, Leucanthemum vulgare, Daucus carota and Lotus corniculatus, the remaining five species were spontaneous.

Details are in the caption following the image
NMDS plot using Jaccard dissimilarity distances of flowering plant species amongst different inter-row treatments. Nine of the flowering plant species identified at the vineyard showed significant presence associated with year/treatment after Bonferroni correction, with black lines representing the direction and strengths of their gradients within ordinate space. Ellipses show the 95% CI of multivariate t-distribution for each treatment.

Beneficial insect abundance

Over two years, 77 bumblebees, 215 hoverflies, 20 honeybees, 844 solitary bees, and 920 solitary wasps were collected. Eighteen families of solitary wasps were identified (July 1 & 2). The majority of the wasps identified were parasitoids, the only exceptions being the crabronid sample and some pompilids (Table 1). Thirty-six species of bee were identified, spanning 9 genera, including Apis mellifera and five Bombus species. The most common wild bee was Lasioglossum minutissimum, a solitary mining bee that may benefit from the undisturbed soil in the vineyard for nesting. Thirteen species of hoverfly (including Sphaerophoria sp., which could not be identified to species level) were identified, spanning 9 genera. The most common hoverfly, Eupeodes corollae, is an aphidophagus hoverfly that could contribute to the pest control of aphids in a vineyard landscape. The top 20 most abundant bees (solitary bees, honeybees and bumblebees) identified are listed in Table 1. Also listed are all hoverfly species identified and all solitary wasp families. (A full species list of bees and hoverflies is available in Supplementary information S3).

TABLE 1. (i) Twenty most abundant bee species, (ii) all hoverfly species and (iii) all solitary wasp families sampled across two years in all five inter-row treatments
(i) Bee species Count (Ii) hoverfly species Count (Iii) wasp family Count
Lasioglossum minutissimum 397 Eupeodes corollae 102 Pteromalidae 54
Halictus tumulorum 114 Sphaerophoria scripta 42 Figitidae 32
Andrena flavipes 94 Melanostoma mellinum 29 Platygastridae 31
Lasioglossum calceatum 47 Sphaerophoria taeniata 10 Braconidae 27
Lasioglossum morio 44 Episyrphus balteatus 9 Ceraphronidae 22
Bombus lapidarius 39 Syrphus ribesii 8 Diapriidae 14
Bombus terrestris 29 Eupeodes luniger 3 Pompilidae 11
Lasioglossum malachurum 24 Sphaerophoria sp. 2 Ichneumonidae 10
Halictus rubicundus 21 Platycheirus manicatus 2 Eulophidae 9
Apis mellifera 20 Melanostoma scalare 2 Megaspilidae 7
Lasioglossum pauxillum 18 Cheilosia vernalis 1 Mymaridae 7
Andrena minutuloides 14 Eristalis tenax 1 Tetracampidae 2
Lasioglossum leucopus 13 Syritta pipiens 1 Proctotrupidae 1
Halictus eurygnathus 6 Aphelinidae 1
Lasioglossum parvulum 5 Torymidae 1
Lasioglossum leucozonium 5 Cynipidae 1
Lasioglossum lativentre 5 Crabronidae 1
Lasioglossum xanthopus 4 Encyrtidae 1
Osmia bicornis 4
Bombus hypnorum 4
  • Notes: Bee and hoverfly data includes that sampled by transect walks and pan traps across all months of the study. Solitary wasps were captured by pan trap in July only year 1 and 2.

There were no significant differences in the overall abundance of all insects between treatment groups in year 1 pan traps (X2 = 4.88, df = 4, p = 0.30; Figure 4). In year 2 pan traps, however, significant differences were detected in the overall abundance of all insects between treatment groups (X2 = 10.31, df = 4, p < 0.05; Figure 4), with posthoc Tukey tests indicating the abundance of “all insects” was significantly higher in the wild bee mix compared to the mown inter-row treatment.

Details are in the caption following the image
Abundance of “all insects” caught in pan traps by treatment. Mean (±SE) abundance of insects caught in five different inter-row ground cover treatments in year 1 and year 2. Letters indicate significant differences in abundances between treatments (Tukey's honest significant difference).

Generally, the most abundant bees were evenly distributed across inter-row treatments, although fewer bees were captured in the mown treatment (Figure 5a). The most abundant bees, Lasioglossum minutissimum and Halictus tumulorum, were abundant across all inter-row treatments. Certain species, such as Bombus lapidarius and Bombus terrestris, were most abundant in the three wildflower mix treatments compared to the mown inter-row treatment or natural regeneration. Fewer hoverflies were captured in the mown inter-row treatment compared to the other four treatments (Figure 5b). The majority of hoverfly species were recorded in the pollen and nectar mix and wild bee mix. The most commonly sampled hoverfly, Eupeodes corollae, was abundant across all treatments. Solitary wasps were also evenly distributed between treatments (Figure 5c). The most abundantly observed wasp family, Pteromalidae, was abundant in all treatments, although several families showed greater abundance in the natural regeneration treatment.

Details are in the caption following the image
Abundance heatmaps of (a) twenty most abundant bee species (b) all hoverfly species and (c) all solitary wasp families. Based on average mean abundance of sampled insects by inter-row ground cover treatment, combining year 1 and year 2 data. Bee heatmap (a) presents the 20 most abundant bee species sampled by pan trap and transect walk, log-transformed for visualisation purposes. Hoverfly heatmap (b) presents species sampled by pan trap and transect walk. Solitary wasp heatmap (c) presents family-level based on pan trap samples from July only.

Pollinator (bee and hoverfly) and solitary wasp richness

In year 1, inter-row treatment did not have a significant effect on the richness of pollinator species for either pan trap (X2 = 2.53, df = 4, p = 0.64; Figure 6) or transect walk (X2 = 5.8, df = 4, p = 0.21; Figure 6) sampled insects. In year 2, however, analysis of transect walk data indicates a significant effect of treatment on pollinator richness (X2 = 9.87, df = 4, p < 0.05; Figure 6), although post-hoc tests did not identify the driver of this effect. However, wild bee mix had the highest pollinator richness (mean ± SE: 2.38 ± 0.34) and mown treatment had the lowest (mean ± SE: 0.50 ± 0.32). In year 2, inter-row treatment did not have a significant effect on the richness of pollinator species for pan trap sampled insects (X2 = 4.92, df = 4, p = 0.30; Figure 6).

Details are in the caption following the image
Richness of pollinators (pan-trap and transect walk data) and solitary wasps (pan trap only) collected across five different inter-row ground cover treatments. Pollinator species (bees and hoverflies) are from monthly samples collected between May and August each year, and solitary wasps are from July only each year.

The majority of the solitary wasps identified in family were parasitoid wasps (Section 12). There were no differences between treatments in the richness of solitary wasp families in year 1 (X2 = 1.93, df = 4, p = 0.75; Figure 6). There were also no differences between treatments on the richness of solitary wasp families in year 2 (X2 = 8.92, df = 4, p = 0.06; Figure 6). However, this is a marginal result. Again, the average richness of wasp families sampled from mown was lower (mean ± SE: 5 ± 0.26) than all other inter-row treatments (Figure 6), most noticeably natural regeneration (mean ± SE: 8.67 ± 0.6).

NMDS analysis showed that pollinator communities (bee and hoverfly species) did not differ significantly between inter-row treatments (PERMANOVA: F1,34 = 0.98, p = 0.51). Solitary wasp family communities did not differ significantly between inter-row treatments (PERMANOVA: F1,34 = 0.77, p = 0.8).


In an experimental manipulation of vineyard inter-row ground cover management, we found that insect abundance, pollinator richness and solitary wasp richness respond positively to the sowing of wildflowers in a British vineyard. While this result is not unexpected, it is, to the best of our knowledge, the first published study on the role of wildflowers in increasing insect biodiversity in a British vineyard. It confirms that wildflowers have the potential to support biodiversity in these typical monoculture landscapes, boosting biodiversity and potentially enhancing pest control management.

Sowing wildflowers increased both floral abundance and diversity. By year 2, the diversity of the floral community had increased for all five treatments, which would be expected with the establishment of perennials in the year following sowing and dispersal of seeds amongst rows in the experimental site. The diversity, richness and abundance of flowers were highest for the mixes that contained grasses (meadow mix and wild bee mix), followed by the pollen and nectar mix and natural regeneration. Unsurprisingly, the mown treatment had the lowest floral richness and abundance. By the second year of the study, floral diversity, richness and abundance of flowers in the natural regeneration treatment were twice that of the mown treatment. Two-thirds of British vineyard managers currently maintain frequently mown grass as inter-row ground cover (with a mowing regime ranging from 10 days to monthly in the spring and summer). Of the remaining vineyards, 28% allow natural regeneration of the existing seedbank, and just 6% currently sow wildflower seeds between rows (Griffiths-Lee et al., 2022).

In year 2, a total of 23 species of flowering plants were recorded in natural regeneration inter-row treatment, 18 of which were naturally colonised (not included in any of the sown mixes). A total of 13 flowering plants were recorded in the mowed inter-row treatment, 10 of which were naturally colonised. Certain flowers are traditionally considered “weeds”, yet they contribute valuable food resources for pollinators with high nectar/pollen rewards. For example, dandelions (Taraxacum agg.) produce high quantities of pollen and nectar (Hicks et al., 2016), and Taraxacum officinale, along with three species from the Sonchus genus, (commonly known as “sow thistles” from the dandelion tribe) were present in the natural regeneration inter-rows (compared to just one of these Sonchus species being present in the mown inter-rows). Likewise, Cirsium arvense, a top nectar producer, and Papaver rhoeas, a top pollen producer (Hicks et al., 2016) germinated in the natural regeneration treatment but not the mown. Existing seedbanks can provide a diverse range of flowers that are visited regularly by hoverflies and bees (Warzecha et al., 2018). Therefore, natural colonisation and simply reducing mowing could enhance pest management and biodiversity without the agronomic, management and resource challenges of adding floral plantings. The site of the study vineyard was previously arable farmland and has a varied seedbank of perennial flowers that reappear readily after mowing, so even the mown inter-row treatment produced flowers during this study. Other vineyards may have a more limited seedbank, perhaps due to herbicide application or a more frequent mowing regime, and perhaps more significant differences between wildflower mix and mown inter-row treatments would be seen in these vineyards.

Once wildflowers were more established in the second year of the study, the abundance of all insects was significantly higher in the wild bee mix treatment compared to the mowed treatment. There was also a significant effect of inter-row treatment on pollinator richness in year 2, and although posthoc comparisons could not determine where this significance lay, again the wild bee mix had the highest average richness of pollinators and mown inter-row strips had the lowest average richness. Certain plant species are more or less beneficial for increasing pollinator abundance (Warzecha et al., 2018), and in our study, floral communities differed significantly between the treatments. This suggests that key species to benefit pollinators are found in the wild bee mix. Our findings are consistent with previous studies conducted in wine-growing regions throughout the world (Kehinde & Samways, 2014; Kratschmer et al., 2019; Kratschmer et al., 2021; Wilson et al., 2018), that inter-row floral plantings increase the richness of bee species and beneficial insect abundance. In our study, significant results were limited to year 2, when wildflowers were established, and differences in significance were also seen between the two sampling methods. Indeed, Templ et al. (2019) recommend a combination of both sampling techniques to obtain more information on wild bee species populations.

The majority of the solitary wasps identified as a family were hymenopteran parasitoid wasps, a group of insects with a very important role in pest control. The effect of inter-row treatment on the richness of wasps in the second year of the study was marginally significant, despite data being limited for this particular analysis. As was found for pollinator richness results, mown inter-rows had the lowest average wasp richness compared to all other treatments. However, in contrast to the results for pollinator species richness (which was highest in the wild bee mix treatment), wasp richness was actually highest in the rows permitted to naturally regenerate. Similarly, previous studies report that introduced floral resources are beneficial for parasitoid wasps in vineyards (Judt et al., 2019; Nicholls et al., 2000), as is natural regeneration (Möller et al., 2020) and simply a reduced mowing regime (Zanettin et al., 2021).

Previous research regarding habitat management to enhance natural pest control has been dominated by cultivation of a limited number of plant species. Indeed, one or more of the plant species Phacelia tanacetifolia, Fagopyrum esculentum, Lobularia maritima and Coriandrum sativum were used in 79% of studies included in a review of habitat management for natural predators by Fiedler et al., 2008. The authors state that as these particular species were effective in earlier studies, they have become influential in later research. Interestingly, flowers considered “weeds” were as effective in increasing parasitoid longevity and fecundity when compared to “flowers” commonly used in parasitoid studies (Araj & Wratten, 2015). It would be beneficial to investigate the potential of a more diverse range of flowers in providing dual resources for pollinators and parasitoid wasps, as we have shown in this research.

The presence of parasitoid wasps or hoverfly larvae has been associated with reduced pest populations (Ramsden et al., 2017) and limited pest damage (Tschumi et al., 2015) although other studies question whether this increased abundance of natural enemies translates into effective pest reduction (Berndt et al., 2006; English-Loeb et al., 2003). Pest-natural enemy interactions and quantifying pest reductions are certainly a natural extension of the current study. Seventy-four per cent of British vineyard owners use synthetic chemical treatments for pest control (Griffiths-Lee et al., 2022) and in 2020, UK vines received on average of 10 fungicides, five sulphur, two insecticide and two herbicide spray rounds (Ridley et al., 2020). Therefore, further research into effective natural biological control would contribute to the sustainability of this sector.

The spatial scale in the current study is the main limitation, which may have been too small to detect significant differences in certain analyses. A greater distance between inter-row wildflower strips, for example, would limit cross-over of flowers and insects between treatments. Future research should ideally be expanded to incorporate multiple vineyards on different soil types and over a longer temporal scale and include landscape-scale factors, given this has been shown to affect wild bee diversity in various ways (e.g., Kratschmer et al., 2019, Kratschmer et al., 2018, Uzman et al., 2020, Wilson & Danne, 2017). Comparison of crop yield data of different inter-row ground cover treatments within the vineyards is also necessary before making definitive recommendations to landowners. The potential role of wildflowers on other taxa, such as soil-dwelling arthropods and birds, should also be explored in future studies.

One of the perceived obstacles in the creation of floral plantings in agroecosystems is the loss of space for crops (Landis et al., 2000). However, the approach tested here utilises inter-row spaces with no loss of cropped land. Headlands around the vines are commonly used for wildflower planting, with 80% of British vineyards utilising this space for sown or unsown flowers (Griffiths-Lee et al., 2022). However, headlands are generally much smaller than the swathes of land converted to ground under-vines. Additionally, inter-row plantings could act as corridors, encouraging natural enemies which would spill over onto the vines (Woodcock et al., 2016).

Research on inter-row floral plantings in cherry orchards found that active management in keeping floral cover height at 20 cm increased floral abundance and provided pest regulation services comparable with floral plantings cut at the end of the summer season (Mateos-Fierro et al., 2021). The authors suggest this management might encourage more landowners to plant floral resources as it reduces humidity for the crop and facilitates management activities. Difficulties in accessing the vines are a reason for resisting the planting of inter-row wildflowers in Californian vineyards (Wilson & Danne 2017). Indeed, alternating management of inter-row ground cover by having wildflowers every other row would benefit biodiversity whilst also facilitating movement around the vines. Additionally, patches of bare ground as part of a habitat mosaic could also benefit insectivorous birds (Schaub et al., 2010).


Promotion of perennial wildflowers through sowing or allowing natural regeneration in inter-row ground cover in vineyards has the potential to boost biodiversity in vineyards on a large scale if widely adopted. Here we report that total insect abundance and pollinator richness benefited from increased floral resources, and the creation of more diverse insect communities results in more resilient pollination services (Woodcock et al., 2019). We also found that a wild bee wildflower mix attracted more insects overall and specifically more pollinator species than any other inter-row treatment, probably due to key floral species present in the mix. We found that simply allowing the recolonisation of floral species by decreasing the mowing regime in the natural regeneration treatment increased the diversity of flowers. Furthermore, although significance was marginal, average solitary wasp family richness was greater in naturally regenerated inter-rows than in that of the mown treatment. Therefore, natural regeneration of inter-row space could benefit biodiversity without requiring significant resources. UK agri-environmental schemes have yet to make specific recommendations to support biodiversity in viticulture. Here we demonstrate a simple, low-cost and effective approach for maximising beneficial insects and supporting key ecosystem services. Given the rapid growth of the vineyard industry and its potential impact on habitat change, further investigation of the potential for enhancing biodiversity in British vineyards is essential.


JGL and DG conceived the methodology and site design; JGL and BD conducted fieldwork and processed samples; BF identified wasp samples; JGL conducted data analysis and led the writing of the manuscript; EN and DG critically evaluated the manuscript for intellectual content. All authors approved the final draft. The authors declare no conflict of interest.


The authors are extremely grateful to Rathfinny Wine Estate for sowing, managing and financially contributing towards the wildflowers in the experimental site. A huge thanks to Steven Falk for helping with bee identification. Thanks to Richard Brown at Emorsgate and Thomas James Wood for discussions on wildflower mix design and wild bee flower preferences. Thanks to two anonymous peer reviewers for their valuable comments. This work was made possible thanks to funding from the C.B. Dennis Trust and a UK Research and Innovation Future Leaders Fellowship awarded to EN (MR/T021691/1).


    The data that support the findings of this study are openly available in Figshare at http://doi.org/10.25377/sussex.20052056 (DOI).