Study area

The study area comprised three areas of olives orchards located in Cordoba province, Southern Spain. Two study plots, one with ground cover and the other with bare ground (mean ± SD plot size 5.9 ± 1.6 ha), were delimited in each of the three areas (Fig. 1). All the olive orchards selected for this study had undergone the same type of insecticide treatments (comprising patching treatments using 40% dimethoate (C₅H₁₂NO₃PS₂) plus hydrolyzed protein), none of which were organic. The elimination of the vegetation cover in plots with bare ground was performed by conventional tillage (consisting of three to four passes, 0.15 m deep, with a rotary tiller (5.5 h.p.) per year, starting after the first rain in late September or early October to control weeds in the streets of the olive groves (Gómez et al., 2009). In addition, for removal of vegetation under the canopy, glyphosate was applied (Piton, 0.36 kg a.e. (acid equivalent) L⁻¹, Dow AgroSciences, Indianapolis, Indiana) at 2.1 kg a.e. ha⁻¹ (in both study areas).

 The cover crop composition in plot with cover crop in study area 1 was mainly integrated by: Asteraceae (species as Conyza canadensis (L.) Cronq. or Leontodon longirostris (Finch & P.D. Sell)), Poaceae (species as Bromus rubens L. or Poa annua L.), Rubiaceae (Galium murale (L.) All.), and Fabaceae (Trifolium sp.); instudy area 2 was composed by: Asteraceae (species as Chrysanthemum coronarium L), Brassicaceae (species as Diplotaxis catholica (L.) DC.), and Poaceae (species as Bromus rubens L.); while in study area 3 was mainly composed by: Malvaceae (species as Lavatera cretica L. or Malva hispanica L.), Poaceae (Hordeum murinum L. or Poa annua L.), and Urticaceae (Urtica dioica L.). The climate in the region is characterised by warm dry summers and mild winters, which are typical of the Mediterranean climate (annual average ± SD temperature 17.6ºC ± 7.2ºC between the years 2010 and 2017). Study area 1 corresponded with an average maximum temperature of 26.2ºC, an average minimum temperature of 10.3ºC (the maximum temperature being 42.8ºC and the minimum being 5.1ºC) and a precipitation of 376.6 L in the study year. The average maximum temperature in study area 2 was 25.3ºC, while the average minimum temperature was 11.2ºC (the maximum temperature being 38.8ºC and the minimum being 8.8ºC) and the precipitation was 581.2 L. In study area 3, the average maximum temperature was 25.8ºC, the average minimum temperature was 11.1ºC (reaching a maximum temperature of 41.1ºC and a minimum temperature of 6.4ºC) and the precipitation recorded during the climatic year was 334.6 L. The study was performed in a region in which the predominant orchard is populated by traditional olive trees that are between 20 and 100 years old, although there are also areas where the cultivation of other species, such as crop plants or vines, takes place. The cultivated olives were of medium size (3–4 m tall) and their density varied between 100 and 200 trees ha^-1.

Figure 1. Study plots (coloured points) and distribution of olive groves (green areas) in Córdoba province (Andalusia, Spain).

Experimental design and arthropod sampling

 Three passive samplings were performed in each study plot (n = 6) to estimate the abundance and the species richness of Cicadomorpha. The distance between the study plots in each study area was 2.2±1.3 km, and the average distance between the three study areas was 30.2±7.3 km. Sampling was performed in the year 2017, during spring, the period in which this insect group is most active (La Spina et al., 2005). The three samplings were performed in the first weeks of May, June and July. In each sampling, eight yellow sticky traps (22 × 35 cm) (Pedigo & Buntin, 1993; Weintraub & Orenstein, 2004; Weintraub & Beanland, 2006) were randomly distributed in each study plot (n = 144 traps). The traps were placed in olive trees at a height of 1.5-2 m and in a vertical position (Prischmann et al., 2007). The position of each trap was georeferenced by means of GPS in order to calculate the average distance between them and the vegetation. The traps were deployed for a one-week period, after which the captured insects were unglued from the traps by employing ethyl acetate 99.8% and were then transferred to plastic bottles containing 90% of ethanol until identification. Specimens belonging to Cicadomorpha suborder (leafhoppers, froghoppers, treehoppers and cicadas) were identified to species level following the methods described by Della Giustina (1989) and Le Quesne (1965, 1969).

 Collection of environmental variables

Owing to the strong effect of environmental conditions on insect activity, the environmental conditions were recorded in each study area (1, 2 and 3) during the previous and the sampling week (PW and SW respectively) (Hay et al., 1996). The environmental data were obtained from agroclimatic stations located close to the three study areas, which were the agroclimatic stations of El Carpio, Córdoba and Baena for study areas 1, 2 and 3, respectively (distance between agroclimatic stations and study areas: 27.4, 25.6 and 18.4 km, respectively). The environmental variables obtained were: maximum, minimum and average temperature, maximum, minimum and average air moisture, precipitation, solar radiation and potential evapotranspiration (ET_o, expressed in mm day^-1). Averages per week were calculated for each environmental variable.

 Landscape variables

 According to Körösi et al. (2012) and Fereres (2017), the surrounding vegetation may enhance the presence and abundance of Cicadomorpha (Cicadellidae and Cercopidae) and we, therefore, recorded the distance between each sticky trap and vegetation. The GPS position of the traps was used to calculate the distance to the closest natural vegetation element (including boundaries, hedges and streams). Linear meters of natural vegetation were calculated in a buffer of 500 meters around every plot.

 Data analysis

 The total and average abundance of each Cicadomorpha species collected were calculated for all the study plots. A Spearman matrix was performed in order to exclude environmental variables with collinearity among them (Acevedo et al., 2005). Two variables were considered to be correlated when r > 0.8.

 Two generalised linear mixed models (GLMMs) were performed to determine the relationships between environmental variables (type of ground cover treatment, environmental and vegetation variables) with regard to two response variables: Cicadomorpha abundance (Model 1) and Cicadomorpha species richness (Model 2). In these analyses, the study area (3 levels) was considered as a random factor, while ground cover treatment (2 levels: cover crop/bare ground) and date (3 levels: May, June and July) were included as fixed factors. The distance from traps to natural vegetation, the linear meters of natural vegetation and the environmental variables (average temperature, average air moisture and precipitation from the PW and average temperature, average air moisture and ET_o of the SW) were included in the model as covariates. A Poisson function and log–link function were used in both models. The best model was selected by employing the forward-stepwise procedure of selecting the model with the lowest corrected Akaike information criterion (AICc) value. These analyses were carried out using the InfoStat software programme (Di Rienzo et al., 2011).

 Dissimilarity and differences in species composition among types of soil management (cover crops and bare ground) were tested using the permutational multivariate analysis of variance (PERMANOVA) (Anderson, 2001). PERMANOVA was employed in order to check significant differences between Cicadomorpha communities according to olive management. This analysis was performed by employing the farming system (2 levels: cover crop and bare ground) as a fixed factor and the environmental variables selected by the previous models (Model 1 and 2) as covariables. Type I Sum of Squares was used, and 9999 permutations were performed with the objective of increasing the power and precision of the analysis (Hope, 1968; Anderson et al., 2008). We also conducted a similarity percentages analysis (SIMPER; Clarke, 1993) so as to determinate which Cicadomorpha species explain the differences in the community composition observed in olive plots with cover vs. bare ground.

 The distance-based linear model (DistLM) is analogous to linear multiple regression and was employed to identify the relationship between environmental variables (climate and vegetation) and the biological Bray–Curtis dissimilarity matrix (Anderson et al., 2008). Two DistLM were employed to identify the relationship between samples and environmental conditionals. The first DistLM was performed with a forward procedure, 9999 permutations and R² as criteria. The sequential test obtained in this analysis made it possible to select those environmental variables that had a significant effect on variability (p<0.05). These variables were included in the second DistLM and were also performed with 9999 permutations. AICc was employed as a selection criterion (owing to the low ratio of samples/environmental variables), while “best” was employed as the selection procedure (Anderson et al., 2008). Finally, a distance-based redundancy analysis (dbRDA) was employed to visualise the DistLM results as principal components (McArdle & Anderson, 2001).

ET_o: potential evapotranspiration expressed in mm per day^-1. SW: sampling week.

* p < 0.05; ** p ≤ 0.01; *** p ≤ 0.001. SW = sampling week.

 Environmental factors affecting the Cicadomorpha communities

 The sequential test performed by means of the first DistLM indicated that eight environmental variables had a significant effect on the structure and composition of Cicadomorpha communities and that they should be included as possible candidates with which to build a model (moisture average SW, ET_o SW, moisture average PW, solar irradiation PW, linear meters of vegetation, ET_o PW, temperature average PW and precipitation SW) (Table 4).

 The second DistLM, which was performed with the eight environmental variables mentioned above, indicated that the best model excluded just one of these environmental variables: precipitation SW (probably owing to its low occurrence in Mediterranean hot-dry summers). The model composed of the seven remaining variables had an AICc value of 983.3 and its first two axes explained 86.5% of the fitted variation and 38.4% of the total variation, respectively (Fig. 2).

 With regard to the environmental variables, it is possible to see a negative relationship between the first axis and the average amount of moisture in the sampling week (Fig. 2a). Some environmental variables, such as linear meters of vegetation, average temperature PW and ET_o SW, meanwhile, had a positive relationship with the first axis. The second RDA axis was related to solar irradiation PW (positive relationship) and ET_o PW (negative relationship), as shown in Fig. 2a.

 The superimposition of the vector representing the relationship between the Cicadomorpha species and the dbRDA axes is owing to the strong relation between M. variatus and the first axis (Fig. 2b). The vectors that represent the species A. mixtus, Platymetopius spp and E. variegatus (Fig. 2b) have a similar direction and position to the vector of average moisture PW, linear meters of vegetation and the average of temperature PW (Fig. 2a),thus underlining possible relations between these variables and species.

DiscussionTop

 Of the 1379 specimens identified, only one is characterised as being a xylem-feeder (one specimen of Cercopis sanguinolenta, Kirschbaum, 1868) being phloem-sap feeding insects the remaining species. Olive groves with cover crops had the highest abundance of Cicadomorpha (phloem-fluid feeding), but this was not the case for species richness. Environmental variables, structural complexity (e.g. hedge or ditches), along with the presence of a cover crop increased the abundance of Cicadomorpha. Interestingly, the distance to vegetation showed a positive relation with the abundance (i.e.a lower abundance close to hedges with remnants of natural vegetation).

 Effect of cover crop and hedges and environmental factors on the abundance and species richness

 The relationship between plants and arthropod communities is complex. Bengtsson et al. (2005) found that organic farms (on which a large number of weed communities are still present) have a higher abundance of arthropod predators, while non-predatory insects and pests did not increase. Predatory and parasitic arthropod communities are often dependent upon the food that is available during different insect stages. Many predators and parasitic arthropods feed on non-prey foods, such as pollen, during their adult phase (Coll & Guershon, 2002), which provide some additional nutrients to those obtained from their host (Zhong-Xian et al., 2014). In our study plots, we found that the frequency of occurrence of Cicadomorpha was positively influenced by the presence of vegetation cover, which agree with previous studies carried out in agricultural landscapes (McClure, 1982; Altieri et al., 1985; Masters, 1998; Körösi et al., 2012; Helbing et al., 2017). Shelter such as bushes or shrubs have been identified as key habitat resources that lead to an increase in the abundance of Cicadomorpha (Redak et al., 2004). On the contrary in this study we interestingly found lower Cicadomorpha abundance close to edges and shrubs. Our results showed that the main factor to negatively affect the abundance of the Cicadomorpha species is the presence of remnants of natural vegetation. Traps located close to the edges captured fewer specimens than those placed in central areas of the olive orchards. The lower abundance of Cicadomorpha close to hedges may be the result of a higher abundance of predation on insects. Castro-Caro et al. (2015) found that the abundance of insectivorous songbirds in olive orchards was highly correlated with the percentage of hedge cover in a short radius (up to 50 m) around the sampling spots. Another possible cause of the low abundance of Cicadomorpha could be that the sticky traps were not located out in the vegetation cover, where the abundance of the cicadomorpha is high during spring (Morente & Fereres, 2017). 

 All Cicadomorpha species captured (except one specimen of Cercopis sanguinolenta) were phloem-sap feeding insects and it is, therefore, unlikely that they transmit the X. fastidiosa bacterium. Our results on the absence of X. fastidiosa vectors are noteworthy, since the use of herbicides to control X. fastidiosa vectors (which is currently mandatory in Apulia, where the olive infection is now f irmly established) is inadvisable, or at least when there is no empirical evidence of X. fastidiosa vectors. Moreover, previous studies have shown that olive groves with bare ground can reduce the diversity of arthropods owing to the explosion of certain dominant groups, while arthropod diversity is positively affected by cover crops (Carpio et al., 2019).

 The sampling date also had an effect on the frequency of cicadellids, with a high frequency of occurrence in June owing to the phenology of M. variatus, the most abundant species collected in that month (88% of the specimens were recorded in June) as also found in German vineyards by Riedle-Bauer et al. (2006), who mainly captured this species in June-August. What is more, La Spina et al. (2005) found a higher abundance of another species of the same genera, Macrosteles quadripunctulatus (Kirschbaum, 1868), during the same period.

 With regard to the environmental variables, we found that average moisture SW, average temperature SW and ET_o SW were negatively related to the abundance of Cicadomorpha.

 With respect to the insect-plant assemblages, our results coincide with those found previously as regards the abundance of both leafhopper species (Brown et al., 1992) and other groups of insects, showing the great influence of the composition and physical structure of vegetation at a local scale (Andrzejewska, 1965; Biedermann et al., 2005; Hollier et al., 2005). However, Altieri et al. (1985) found a higher abundance of leafhoppers in monospecific clover mulch when compared to those with a more diverse weed cover.

 The average moisture reduced the Cicadomorpha abundance; however, this environmental variable had beneficial effects to Cicadomorpha species richness. Our f indings showed only a positive relation of species richness with average moisture SW and ET_o SW. However, the results did not show any positive relation between species richness and cover crop or habitat variables. This may be owing to the fact that moisture and ET_o have a similar effect on species richness because ET_o is related to environmental variables such as temperature and moisture. In Mediterranean agroecosystems, which have extremely high temperatures in summer, high values of moisture may reduce the limiting influence of temperature on the development of some species. What is more, higher moisture values allow the development of more plant species that can provide shelter and a food source for a greater number of Cicadomorpha species. Nickel & Achtziger (2005) described the key role of moisture for the recovery of leafhopper communities in meadows with extensive land use. However, the same authors did not find any relationship between the number of leafhopper species and plant communities, as occurred in the present study.

 Effect of the cover crop and hedges on Cicadomorpha community assemblage

 Our results show that abundance and distance to remnants of natural vegetation, along with the sampling date and cover crop and their interactions, influenced community assemblage. However, only one environmental variable, the average moisture in SW, affected the insect community sampled. Olives with cover had twice the number of specimens and three species more when compared to olive groves with bare ground. The main species of Cicadomorpha communities indicated by the SIMPER procedure (M.variatus, Platymetopius spp, A. mixtus and E. variegatus) were found in higher abundance in olive groves with cover crops than in those with bare ground (see Table 1). These results highlight the benefits of the presence of plant communities for Cicadomorpha, as described in previous studies (Murdoch et al., 1972; Denno & Roderick, 1991; Denno, 1994). The strong and fast response of Cicadomorpha communities to the management of grasslands (such as cutting, grazing and fertilizing) led Nickel & Hildebrandt (2003) to employ them as indicators of disturbance in grasslands. The same authors state that species’ life strategies are correlated with the level of perturbation in ecosystems, with generalist species being more abundant in disturbed ecosystems, and the specialist species in low disturbed ecosystems. This observation is supported by the influence of vegetal variables on Cicadomorpha communities. The influence of the sampling date on the composition of the Cicadomorpha community is mainly related to the peak of the M. variatus population recorded in June for olive orchards with both cover crops and bare ground. Of the other species, E. variegatus and Platymetopius sp. also underwent an increase in June when compared to May and July. Changes in Cicadomorpha communities on a small-time scale were also observed by Brown et al. (1992), although they observed a peak of populations in the period late July-early August. They attributed these changes in abundance and community composition to a successive process. However, the main change to take place in the Cicadomorpha community in June occurs in the case of the M. variatus and coincides with half the flowering period of U. dioica (from May to July), its nutritional plant (Taylor, 2009). U. dioica is a species that is associated with an anthropogenic environment, typical of urban riparian habitats that are highly disturbed, and is commonly found in olive crops (Maskell et al., 2006; Perrino et al., 2014

In conclusion, Cicadomorpha richness and abundance were affected by structural complexity provided by cover crops (positive effect) and live hedges (negative effect), which may be owing to the higher food abunda ce and shelter when cover crops are present, whereas higher insect predation may occur close to hedges, probably owing to insectivorous song birds. However, further research should be done considering different weather conditions (humid or dry years), different cover crop species and managements or different sampling dates.