To support sustainable food production and the delivery of ecosystem services through ecological intensification, wildflower strips have become a popular strategy. Despite their success in temperate orchard systems, they remain understudied in Mediterranean ecosystems, which poses a significant barrier to uptake. In order to further promote their adoption, seed mixes must be optimised for commercial orchard systems and for the Mediterranean climate. Plant species should be selected for their consistent performance, while the availability of resources for ecosystem service providers determines the quality of the wildflower strip. In this study, the suitability of 12 native perennial forbs and two tussock-forming grass species for wildflower strips in commercial Citrus orchards was assessed over a three year period. Distinct resources for natural enemies according to the different plant growth stages was used an indicator of wildflower strip quality. The wildflower strips were managed under two different cutting strategies; i) standard management, in which wildflower strips were cut once annually in February, ii) active management, in which wildflower strips were cut two additional times each year.  The establishment and success of the sown species were compared. The influence of wildflower strips and their management on plant species richness, community structure, and the provision of resources was compared with iii) a control treatment, in which alleyways were managed conventionally by cutting any naturally occurring vegetation to a height of ≤5 cm, four to five times annual. For the first time, the performance of native perennial plant species has been assessed in Mediterranean orchard systems and a seed mix developed targeting pest regulation services. The wildflower strips were successful in increasing plant species richness and the available resources expected to support natural enemies.  However, only wildflower strips managed with cutting once annually enhanced vegetation cover relative to the control, whilst extending the flowering period. This study therefore provides crucial tools for the further development of sustainable approaches to food production in Mediterranean orchard systems. 

1.1.  Site Description

The three year study was conducted in three commercial navel orange (Citrus sinensis) orchards in Huelva, Andalusia (Appendix A1, Figure A1): Madre del Agua (37°26'27.80"N  7° 9'55.73"W) and La Calvilla (37°24'10.95"N  7° 3'42.67"W) in southern Huelva and Montepinos (37°47’43.21”N 6°56’21.11”W) in northern Huelva.  All orchards were managed under Integrated Pest Management (IPM) guidelines (Llorens Climet and Martín Gil, 2014) and contained naturally occurring vegetation in alleyways, managed with regular cutting to a height of ≤5cm, four to five times annually (Figure A2).

1.2.  Study Design

A randomised block design was established with four complete replicate blocks in November 2016.  The experimental treatments were applied in 0.5 ha plots, separated by at least 150 m.  Each plot measured 100 m in length and 50 m wide and consisted of eight rows of orange trees and seven alleyways between rows (Figure 1).  The wildflower strips were sown in November to take advantage of seasonal rainfall and the associated increase in germination success (Ramírez and Lasheras, 2015).  Competition from existing vegetation was eliminated by applying glyphosate (RoundUp, Monsanto, Missouri) (Natural England, 2013).  After seven days, the alleyways were then cultivated to create a fine seedbed (Westbury et al., 2017).  The seeds were mixed with sand to ensure even sowing by hand, immediately after which the seedbed was rolled to firm the seeds with the substrate.  The novel seed mix consisted of all native perennial species, including twelve forb and two tussock-forming grass species.  Species were selected to provide floral resource in succession across the length of the year and a diversity of plant traits (such as phenology, growth forms, height, and floral traits) to increase sward structure, provide a diversity of plant growth stages, and support natural enemies (Table 1).  The seed mix was sown in alternate alleyways at a rate of 5.66 gm^-2 to create wildflower strips measuring 2 m wide and 100 m long (Table 1).  During the establishment year of 2017 (year one) all wildflower plots were managed by cutting once in March and once in April to ≈ 10 cm to promote the establishment of the sown species (Woodcock et al., 2008). 

To reduce edge effects, the outermost alleyways within the 0.5 ha plots were excluded from sampling and a 20 m buffer region was established at either end of the alleyways.  As such a 60 m long central sampling area consisting of two alleyways between four rows of orange trees was established (Figure 1) (Englund and Cooper, 2003).   

1.3. Botanical Surveys

Botanical surveys were conducted in May of each year to determine plant species richness and community composition according to alleyway treatment.  Early spring ensured unsown spring ephemeral species were recorded prior to the application of the cutting regimes.  As all eight wildflower plots were managed the same in year one, four replicate wildflower plots were randomly selected and sampled, equivalent to the same number of control plots.  From year two, the two different management strategies were randomly allocated, and all plots were surveyed.  Six replicate 0.5 m  0.5 m quadrats were randomly placed in each of the two wildflower strips within the survey area so that a total of 12 randomly placed quadrats were sampled from each plot at each sampling date. 

All plant species present within quadrats were identified to species where possible, except for seven plants which were identified to the lowest rank possible (genus or family) and unsown Poaceae which were identified to family.  Each species from within the quadrat was assigned a percentage cover score according to an eight-point scale (1 = <1%, 2 = 1-5%, 3 = 6-10%, 4 = 11-20%, 5 = 21-40%, 6 = 41-60%, 7 = 61-80%, and 8 = 81-100%).  Bare soil and alleyway leaf litter (unattached) were recorded using absolute percentage cover values.  The reproductive status of each species was also recorded as i) vegetative only or had ii) flower shoots present or budding, iii) flowers open in bloom, or iv) seeds in formation, ripe or dehiscent.  A four-point scale assigned the proportion of individuals at each stage (1 = 1%-25%, 2 = 26-50%, 3 = 51-75%, 4 = 76-100%) (Westbury et al., 2017).

To calculate the percentage cover scores for the resource classes (grasses and forbs), a reproductive percentage cover score was first estimated for each species surveyed within the quadrat.  The reproductive percentage cover score was estimated from; i) the reproductive scores per species and ii) percentage cover score per species (as previously described).  Firstly, reproductive scores for each species were back-transformed to their mid-point values to estimate the percentage of each of the four reproductive stages for each plant species (1 = 12%, 2 = 38%, 3 = 63%, and 4 = 88%).  Secondly, the estimated percentage cover each species occupied within the whole quadrat (0.5%, 2.5%, 8%, 15.5%, 30.5%, 50.5%, 70.5%, or 90.5%) was multiplied by the estimated reproductive scores, to calculate the reproductive percentage cover scores for each species at each reproductive status.  Once the reproductive percentage cover score for each species had been calculated, these scores were summed across all grasses or all forbs to give the percentage cover scores for the reproductive resource classes for each of the ten groups (vegetative forbs, vegetative grasses, budding forbs, budding grasses, flowering forbs, flowering grasses, dehiscent forbs, dehiscent grasses).  The non-reproductive resources classes (bare soil and alleyway leaf litter), were scored using absolute percentage cover values.

1.4.  Vegetation height and structural heterogeneity

The height of the alleyway vegetation was measured and the structural heterogeneity then determined by calculating the coefficient of variation for each alleyway sampling area.  For this, a wooden disc of a standard diameter (30 cm) and weight (200 g) was dropped down a 1 m rule and the height it rested on the sward was recorded (Stewart et al., 2001).  Twenty drop disc measurements were taken monthly from each alleyway from April to October during all study years.