Introduction

Alfalfa (Medicago sativa L.) is one of the most economically important forage crops worldwide (Michaud et al. 1988; Frame et al. 1998). In Spain, it covers an area of 250,000 ha, accounting for ~ 20% of European production. In the Ebro Basin (Fig. 1a and b), alfalfa is a common crop in irrigated areas, representing ~ 60% of Spanish production (Delgado and Lloveras 2020).

Fig. 1
figure 1

a Localization of the Ebro Basin in Spain. b The four counties from which the 65 fields were selected during 2018, 2019 and 2020: (1) Urgell (n = 27); (2) Segrià (n = 21); (3) Cinca Medio (n = 10); and (4) Monegros (n = 7)

Full size image

The alfalfa weevil Hypera postica Gyllenhal (Coleoptera: Curculionidae) is a highly destructive pest specific to alfalfa (Goosey 2012; Saeidi and Moharramipour 2017; Pons and Nuñez 2020). It is native to Eurasia but has a global distribution (Hoffmann 1963). Both adult weevils and larvae feed on alfalfa leaves, but the larvae cause most of the damage, leading to economic losses by impeding plant growth and reducing biomass accumulation (Berberet and McNew1986; Alfaro 2005). In Spain, the alfalfa weevil inflicts most damage during the initial alfalfa cutting in spring (March–April), although they occasionally have an impact on the second cutting too (Pons and Núñez 2020). One or two complete generations may occur, depending on the temperature (Levi-Mourao et al. 2022c). Female weevils begin laying eggs in October after a period of summer estivation, depositing clusters of eggs within alfalfa stems (Domínguez 1989; Alfaro 2005; Pons and Nuñez 2020; Levi-Mourao et al. 2021, 2022b). The resulting larvae hatch between the end of winter and the beginning of spring, and feed on leaves and new plant buds. At the end of the fourth instar, the larvae pupate between the litter and the plant leaflets in white cocoons (Levi-Mourao 2022c).

As a perennial crop and one of the traditional components of arable crop rotations in Spain, alfalfa provides a favorable habitat for many natural enemies of the alfalfa weevil (such as predators and parasitoids) and is much more stable (it remains 3–6 years in the field) than other extensive field crops such as cereals, sunflower or ryegrass (Summers 1998; Núñez 2002; Pellissier et al. 2017; Rand 2017). This natural enemy complex helps to minimize primary and secondary pest outbreaks in alfalfa and its surrounding crops (Summers 1998; Madeira et al. 2019, 2021). The most important natural enemies of alfalfa weevil larvae are solitary larval endoparasitoid wasps of the genus Bathyplectes (Hymenoptera: Ichneumonidae) (Hussain 1975; Flanders et al. 1994; Radcliffe and Flanders 1998; Pellissier et al. 2017). Although Ribes (2012) reported the presence of eight Bathyplectes species in the Iberian Peninsula, only two are known to be associated with the alfalfa weevil: B. anura and B. curculionis (Thomson) (Pons and Nuñez 2020; Levi-Mourao et al. 2022a).

Pest outbreaks and pest control by natural enemies depend not only on local field conditions but also on landscape patterns (With et al. 2002). Some of the literature on the effects of landscape complexity on insect abundance is related to the natural enemies of insect pests in the context of pest control (Symondson et al. 2003; Bianchi et al. 2006; Tscharntke et al. 2012; Rusch et al. 2016). Most insects need to move across the landscape to search for resources (e.g., change of host, feeding or mating), and scaling up from the field to the landscape therefore appears necessary to understand pest control services in crops (Landis et al. 2003; Rand et al. 2006; Tscharntke et al. 2007, 2012; Bianchi et al. 2013). Agricultural landscapes are more dynamic than other landscape types (Petit 2009) because most crops are frequently disrupted by agronomic practices, changes in crop phenology, harvest and/or crop rotations, making them periodically unsuitable to support pests. The impact of landscape patterns on pest abundance and pest control by natural enemies is thus expected to change during the year and between years (Menalled et al. 2003; Bianchi et al. 2006; Rand 2013). Moreover, landscape composition has inconsistent effects on pests and natural enemies, especially when different landscapes are considered (Karp et al. 2018). Studies considering the effect of landscape structure should therefore cover multiple years and should consider the different characteristics of the area, such as crop cover changes among seasons.

When considering a specialist pest and the specific crop it attacks, a positive correlation would be anticipated between pest abundance and the crop area at the landscape scale (Rand et al. 2014). However, this relationship may be nonexistent or even negative in certain cases. For example, some pests require alternative habitats and may live elsewhere during an important part of their life cycle (Östman et al. 2001; Thies et al. 2005, 2008). Furthermore, the abundance of some pests may depend on the presence of non-disturbed elements in the landscape, such as non-treated areas, margins, and woody areas (Ricci et al. 2009). Similarly, a positive correlation would be anticipated between the abundance of specialist natural enemies in the landscape and a larger area of the host crop. However, natural enemies may use alternative food resources or refuges beyond the area covered by the crop. For example, Bathyplectes spp. use alternative food sources such as aphid honeydew and flower nectar, which could have a significant impact on their parasitism levels at both the field and landscape scales (Jacob and Evans 1998, 2000; Evans 2018; Rand and Lundgren 2019).

Pests and their control strategies in alfalfa have mainly been studied at the field scale in Spain. Recognizing the need for larger scales, some studies in the Ebro Basin have reported the relationships between landscape characteristics and insect abundance in crops such as maize and alfalfa (Madeira et al. 2014, 2021; di Lascio et al. 2016; Clemente-Orta et al. 2020). However, none of these studies focused on the alfalfa weevil and its natural enemies. Little is known about the motility of this pest, and how its abundance may be affected by other alfalfa fields or by the surrounding landscape structure. Recently, a parallel study using a different approach but with a similar goal was conducted in the USA to evaluate the effects of landscape structure and local factors on the abundance of the alfalfa weevil and B. curculionis (Pellissier et al. 2022). However, there are significant differences in agricultural practices and landscape structure between the Northern Great Plains of the USA and the Ebro Basin, with the latter being much more heterogeneous and composed of many more small fields per unit of area. Additionally, alfalfa weevil populations can diverge significantly across different geographical areas, including USA and Europe (Sanaei et al. 2019), which could result in different adaptations to local and landscape factors.

We investigated the influence of the surrounding landscape structure (composition and configuration) on the abundance of alfalfa weevil and the abundance and parasitism rates of Bathyplectes spp. We tested five specific hypotheses. First, following the resource concentration hypothesis (Root 1973), the abundance of alfalfa weevil (a specialist pest) should be positively correlated with the abundance of alfalfa in the landscape. Second, the abundance of alfalfa weevil should be positively correlated with the proportion of natural edges (field margins) in the landscape because natural edges facilitate dispersal to new fields and serve as overwintering and estivation areas (Manglitz 1958; Dennis and Fry 1992; Holland and Fahrig 2000; Prokopy et al. 1965, 1967). Third, the abundance of alfalfa weevil larvae (a low mobility insect stage) should be influenced more by local field characteristics and crop management than the surrounding landscape patterns (Blodgett et al. 2000; Goosey et al. 2004; Rand 2013). Fourth, the abundance of Bathyplectes spp. and the parasitism rate should be related to the surrounding landscape characteristics rather than local factors of the alfalfa field, due to the dispersal capacity of these parasitoids in the search for food (Jacob and Evans 1998, 2000; Evans 2018; Rand and Lundgren 2019). Finally, parasitism rates by Bathyplectes spp. should decrease with the higher abundance of alfalfa weevil larvae as previously reported in North America (Rand 2013).

Materials and Methods

Study area

The study was performed in the Ebro Basin (Fig. 1a), at the northeast of the Iberian Peninsula, western Mediterranean area (Fig. 1b). The mean temperature ranges from 1 °C in winter to 30 °C in summer. Annual rainfall is variable, ranging from 200 to 800 mm, and is mainly concentrated in spring and autumn. The mean altitude is 200 m above sea level (asl). In this area, agricultural landscapes are traditionally dominated by arable crops that are managed by the rotation of winter cereals, such as wheat and barley, summer cereals, mainly maize, and alfalfa. Winter cereals are grown from October to June and maize is grown from the end of March/April to October, although the current tendency in non-rotation crop systems is to sow it earlier, during March (Cantero-Martínez et al. 2006; Albajes et al. 2022). Non-cultivated areas, such as forests and woody areas, older fallows, natural habitats, field margins, roads, and irrigation canals are interspersed in these agricultural landscape mosaics. Forests are typically dominated by replanted Pinus halepensis (Mill.) and Mediterranean bushes.

We selected 65 commercial alfalfa fields located in four different counties of the region (Urgell, Segrià, Cinca Medio and Monegros) (Fig. 1b) during 2018, 2019 and 2020 (field characteristics and geographic locations are summarized in Table S1). The fields in each county were separated by at least 2 km within years to avoid potential spatial autocorrelation. The landscape studies were selected along a gradient of high to low alfalfa dominance (Table S2). The selected alfalfa fields were 2, 3 or 4 years old, and were sown with the Aragon variety, derived from the ecotype Aragon, which has been cultivated in the Ebro Valley for decades. As well as tolerating temperatures below –15 °C, it has a short dormancy period, fast development in spring, and regrows after cutting, allowing 5–6 cuts per season under irrigation (Delgado 2020; Delgado and Lloveras 2020). No insect resistance traits are known for this variety. The field sizes ranged from 1 to 7 ha, a common range in this area, and were irrigated by sprinkling or flooding. No pesticides were applied during the study period. The parasitism rate was estimated in 35 fields selected from the original 65, representing the 2019 and 2020 campaigns.

Data records

For each selected alfalfa field, we recorded crop management and field characteristics. We also recorded the landscape coverages at buffer radii of 250, 500 and 1000 m (Fig. 2) from the center of each sampled field, based on our practical constraints in terms of work and time, and following previous studies (Clemente-Orta et al. 2020; Madeira et al. 2021). Each field was sampled for insects once during winter (eggs) and twice (larva and adult) during the first alfalfa intercut, between the beginning of vegetative growth and the first cutting (Pons et al. 2005) in March and April. When possible, sampling was repeated in the same fields in all 3 years of the study (n = 3), but due to crop rotations some fields were sampled only twice (n = 41) or only once (n = 21).

Fig. 2
figure 2

Examples of 1000 m cover type descriptions at the three buffers with low, medium and high alfalfa gradients (a, b, and c respectively). Dark green shows alfalfa patches (AL). Different colors indicate the different cover types initially defined in the landscape (AC: arable crops; AG: water areas; ED: Buildings and urban areas; FO: forest; NC: non-crop; OR: orchards; and VI: roads)

Full size image

Local variables

Local variables included the perimeter, area, alfalfa age, and irrigation system (sprinkling or flooding) in each field. In the analysis of Bathyplectes spp. (adult abundance and parasitism rate), we included the abundance of alfalfa weevil larvae and aphids as explanatory variables. The perimeter and the area of each selected alfalfa field were calculated using ArcGIS 10.3.3 (Environmental Research Institute—ESRI, Redlands, CA, USA 2015).

Landscape variables

Landscape variables were selected based on previous work in the region (Clemente-Orta et al. 2020a, 2020b; Madeira et al. 2021), and their potential effect on the populations of alfalfa weevil and its larval parasitoid, Bathyplectes spp. ArcGIS software was used to quantify land cover types. Landscape composition was determined by direct field inspections during the first alfalfa cutting (March–April) each year; by using orthophotos from the Plan Nacional de Ortografía Aérea (PNOA) and other reference geographical information maps of the Instituto Geográfico Nacional de España (https://www.ign.es); and by consulting Declaració Agrària (DUN) from the Departament d’Acció Climàtica, Alimentació i Agenda Rural (http://agricultura.gencat.cat/ declaracio-unica-agraria/). The DUN is an annual declaration that must be submitted by the person in charge of the holding, defining the crop and its cultivated area. Thirteen cover types were categorized at the 250 and 500 m buffer radii (Table S2). However, only eight cover types were considered at the 1000 m radius due to the large number of fields within this radius, making it impossible to visit all of them and distinguish between some cover types with the geographical information available (Table S2). Fruit orchards were grouped into one category. They were initially recorded by species due to the higher variability between sampled areas, but the most prevalent ones belonged to the same family (Rosaceae) and in any case they represented a low proportion of the Monegros and Cinca Medio landscapes. Cover types representing a negligible proportion of the landscape (mean < 2%) were not included in the analysis (Table S2). Cover types with similar functions in our system, for example apple and peach orchards, were grouped together (Table S2). For data analyses, the initial cover types were combined into six variables at 250 m and 500 m buffers and five variables at the 1000 m buffer (Table 1).

Table 1 Local and landscape variables surrounding the sampled fields within the three different buffer radii (250, 500 and 1000 m) included in the model analysis
Full size table

Landscape configuration was characterized using FRAGSTAT v.14.2 (McGarigal et al. 1995). We selected configuration metrics that capture different information about landscape structure (Hann et al. 2019; McGarigal et al. 1995) and were relevant to the focal species studied: alfalfa edge density, total edge density, patch-Euclidean nearest neighbor, alfalfa-Euclidean nearest neighbor and Simpson’s diversity Index (Table 1).

Insect sampling

The abundance of alfalfa weevil eggs was determined on samples of 200 similar length (30–40 cm) alfalfa stems per field (from the middle of December to the second fortnight of January). Stems were cut using scissors and stored at 5 °C until dissection (within 7 days). The abundance of weevil larvae, weevil adults, Bathyplectes adults and aphids was determined by conducting 180° sweeps with a net 38 cm in diameter. Bathyplectes adults were distinguished from other parasitoids following Ribes’ (2012) identification key. Fields were sampled on two different dates during the first intercut period (from mid-March to the end of April). Twelve samples (five sweeps per sample) were taken in each field on each date. Each field was divided into four sectors, and three samples per sector were collected following the central part of one of the main diagonals. Samples were stored at –20 °C.

On the same sampling dates, we collected an additional sample consisting of 20 sweeps per field, along the main diagonal of the field, to the estimate of larval parasitism in 35 of the selected fields (14 from 2019 and 21 from 2020). Alfalfa weevil larvae were kept in 500-ml polyethylene rearing cages (maximum 50 larvae/cage), covered with mesh to facilitate aeration. Fresh alfalfa was provided every day. Larvae were maintained in a climatic chamber at 22 °C, with a 8:16 (L:D) photoperiod and 50% relative humidity until pupation. Parasitoid puparia were used for morphological identification (Levi-Mourao et al. 2022a).

Data analysis

Spearman rank correlations (Dormann et al. 2013) were used to assess the correlation between landscape structure and field variables (Table S3), to prevent multicollinearity in the statistical models (Legendre 1993; Wagner and Fortin 2005). When strong or very strong correlations between variables were found (Spearman’s rho > 0.59) (Campbell and Swinscow 2009), variables with lower biological relevance were excluded from the models. Moderately correlated variables (Spearman’s rho 0.4–0.59) (Campbell and Swinscow 2009), were not excluded to build the models (Schmidt et al. 2019). For the dependent variables, spatial autocorrelation among fields was tested based on mean values for the 3 year study period, using Moran’s I statistic (Paradis 2019). No significant autocorrelations were detected (Table S4).

The effects of landscape structure and local variables on the abundance of alfalfa weevil (eggs, larvae and adults) and Bathyplectes spp. adults were evaluated by sampling the abundances in two samples collected during the first alfalfa cutting and presenting the average value. For the adult weevils, one outlier (nine times higher than any other value observed, and caused spatial autocorrelation problems) was discarded from the dataset. The data were not distributed normally even after transformation, so gamma-family generalized linear mixed models (GLMMs) were applied because they fitted well the distribution pattern of our data. For the analysis of Bathyplectes spp. parasitism rate, binomial-family GLMM models were used in which we included the weight of the variable (initial number of reared H. postica larvae for parasitism rate estimation). Field and year were always included as random factors. Models were fitted for each spatial scale (250, 500 and 1000 m) using the glmer() function of the lme4 package (Bates et al. 2015). Landscape and local metrics for each model were standardized (mean centered and scaled) before analysis using the caret package (Kuhn 2022).

A multi-model inference approach was used to obtain robust parameter estimates. The dredge() function of the MuMin package (Bartoń 2020) was used to fit all possible combination models, describe the effects of independent variables on each dependent variable, and calculate their associated Akaike information criterion corrected for small sample sizes (AICc). We created an arranged list of models based on the comparison of AICc with the full model values. Model averaging was applied to the model set with ΔAiCc < 2 (Burnham and Anderson 2004) using the model.avg() function from the MuMIn package. To avoid multicollinearity, variance inflation factors (VIFs) for the covariates in the averaged models were calculated using the vif() function from the car package (Fox and Weisberg 2019). Covariates with VIF > 5 were discarded and the analysis was repeated. Lastly, the effects package (Fox et al. 2016) was used to represent the effects in partial residual plots. The function sigtest() from the influence.ME package (Nieuwenhuis et al. 2012) was used to identify influential data points but none were detected. Statistical analyses were conducted using R v4.0.3 (R Development Core Team 2022).

Results

Insect abundance and parasitism rates

We collected 25,024 H. postica eggs, 171,808 larvae and 1,952 adults in the 65 sampled alfalfa fields during the 2018, 2019, and 2020 seasons. We also collected 863 adult Bathyplectes spp. Yearly parasitism rates per field ranged from < 2% to > 30%, with mean values of 15.4% in 2019 and 5.1% in 2020.

Effects of local variables

The abundance of H. postica eggs was highest at all scales when the field was sprinkler irrigated (Fig. 3a–c). Likewise, the abundance of H. postica larvae increased with sprinkler irrigation at the 500-m and 1000-m scales (Fig. 3e, f). The field perimeter had a positive association with the abundance of H. postica eggs at the 1000-m scale (Fig. 3d) and larvae, at all scales (Fig. 3g–i). In contrast, local variables had no significant effect on the abundance of H. postica adults.

Fig. 3
figure 3

Partial residual plots showing the significant local variables (irrigation system and perimeter) that affected the abundance of H. postica eggs (a–d) and larvae (e–i). Y-axes are on the log-scale to preserve the linear structure of the model. Blue shaded areas represent 95% confidence intervals

Full size image

The abundance of Bathyplectes spp. adults was positively associated with the abundance of alfalfa weevil larvae at all scales (Fig. 4a–c) and so was the parasitism rate (Fig. 4j–l). Furthermore, the abundance of parasitoid adults was positively associated with the abundance of aphids (Fig. 4d–f). Age was found to be a significant factor affecting parasitism rates, with lower rates observed in older fields (Fig. 4g–i). Table 2 shows the significant effects of alfalfa field variables on H. postica and Bathyplectes spp. (adults and parasitism rate). The total effects of all local variables are shown in Table S5.

Fig. 4
figure 4

Partial residual plots showing the significant local variables (H. postica larvae, aphids and age) influencing the abundance of Bathyplectes spp. adults (a–f) and parasitism rate (g–l). From figures a–f, the Y-axes are on the log-scale, and from figures g–l, they are on the logit-scale to preserve the linear structure of the model. The blue shaded areas represent 95% confidence intervals

Full size image
Table 2 Significant variables (p ≤ 0.05) in the best models (ΔAiCc < 2) showing relationships between local or landscape variables with the abundance of H. postica (eggs, larvae, and adults) and Bathyplectes spp. (adults) and the parasitism rate. The sub-tables represent the three radii buffers of 250, 500 and 1000 m. Variables were mean-centered and scaled. Relative importance is the sum of Akaike’s weight associated with the variables in the best models. AIC (Akaike information value) values indicate differences in support from data model. The different statistical values come from the output generated using the model.avg() function from the MuMIn package
Full size table

Effects of landscape variables

The landscape structure had a relatively weak effect on the abundance of H. postica and Bathyplectes spp. (adult abundance and parasitism rate). None of the composition variables had a significant effect. Only one configuration metric (alfalfa edge density) was positively associated with the abundance of H. postica larvae across all landscape scales (Fig. 5a–c). Conversely, the landscape structure had no significant effects on H. postica adults and eggs. The most parsimonious models are shown in Table S5, and the significant effects of landscape variables on H. postica (abundance) and Bathyplectes spp. (adult abundance and parasitism rate) are shown in Table 2.

Fig. 5
figure 5

Partial residual plots showing the significant landscape variable (alfalfa edge density) and its effect on the abundance of H. postica larvae. Y-axes are on the log-scale to preserve the linear structure of the model. Blue shaded areas represent 95% confidence intervals

Full size image

At the 250-m scale, there was a negative relationship between the abundance of Bathyplectes spp. adults and Simpson’s diversity index (Fig. 6a) and a positive correlation with alfalfa edge density (Fig. 6b). However, no effects were observed at the 500 m and 1000 m scales, and the landscape structure also had no impact on parasitism rates.

Fig. 6
figure 6

Partial residual plots showing the significant landscape variables (Simpson’s diversity index and alfalfa edge density) and their effect on the abundance of Bathyplectes spp. adults. Y-axes are on the log-scale to preserve the linear structure of the model. Blue shaded areas represent 95% confidence intervals

Full size image

Discussion

Over the last two decades, several studies in the Ebro Basin have described the composition, ecological role and abundance of insect populations that live in alfalfa, which is a prevalent forage crop in the region. These studies concluded that alfalfa is an important reservoir of natural enemies that provides biocontrol services to alfalfa and other neighboring crops (Núñez 2002; Pons et al. 2005; Madeira et al. 2014, 2019; Di Lascio et al. 2016; Batuecas et al. 2022). However, these studies did not consider the effect of landscape structure or field characteristics on major pests such as the alfalfa weevil, which causes extensive damage at the larval stage in most areas where alfalfa is cultivated (Goosey 2012; Saeidi and Moharramipour 2017; Soroka et al. 2019; Levi-Mourao et al. 2022b). Indeed, little is known about the factors that regulate the abundance of this pest in Europe. Here, we show for the first time the effects of field characteristics and landscape structure on the alfalfa weevil and its larval parasitoids in Mediterranean alfalfa crops.

Contrary to our first hypothesis, we observed no association between the abundance of alfalfa weevil and the proportion of alfalfa in the landscape. The abundance of specialized herbivores tends to correlate with the extent of their host crop, typically being more numerous in simple environments characterized by low plant diversity (O’Rouke et al. 2011; Root 1973). Consistent with this expectation, the abundance of alfalfa weevil in North America was recently shown to correlate with the proportion of alfalfa in the landscape, particularly at scales ≥ 2000 m (Pellissier et al. 2022). Our contrasting results may be due to the greater heterogeneity of the agricultural landscape in the Ebro Valley, which consists of smaller fields (Clemente-Orta et al. 2020; Madeira et al. 2021) compared to the more uniform landscape of larger fields in North America (Pellissier et al. 2022). Dispersal mortality and fitness costs become significant factors when specialist herbivores are compelled to relocate across greater diversity of habitats in pursuit of their host crop (Chaplin-Kramer et al. 2011). Altogether, these results illustrate how the availability and distribution of resources can influence pest dynamics in different agricultural landscapes (O'Rourke and Peterson 2017; Fahrig and Jonsen 1998; Samaranayake and Costamagna 2018). In our case, we hypothesize that alfalfa weevil adult populations that must cross fields of other crops or orchards treated with insecticides (Clemente-Orta et al. 2020; Madeira et al. 2021) to reach the nearest alfalfa fields may suffer high mortality during dispersal.

According to our second and third hypotheses, alfalfa weevil populations are associated with the occurrence of natural edges in the landscape, and larval abundance in the field mostly depends on the field characteristics. We found that larval abundance was associated with greater alfalfa edge densities and longer field perimeters, the latter showing also a positive correlation with the number of H. postica eggs at 1000 m. In this context, both variables may be considered as a proxy of the margins surrounding an alfalfa field, which act as natural refugees for alfalfa herbivores including the alfalfa weevil, supporting their population growth (Landis et al. 2003; Madeira et al. 2021; Werling and Gratton 2008). Several studies have reported that adult weevils aestivating in the summer become concentrated along the borders of alfalfa fields, allowing rapid re-infestation for feeding and reproduction in the autumn (Manglitz 1958; Prokopy and Gyrisco 1965; Prokopy et al. 1967; Saeidi and Moharramipour 2017; Pellissier et al 2022). This re-infestation by adults is indirectly supported by the abundance of eggs and larvae recorded in winter and spring, respectively. Alfalfa was not irrigated during the time of our sampling, but the irrigation system also influenced the abundance of H. postica eggs and larvae, with higher numbers of both stages found in fields irrigated by sprinkling rather than flooding. This difference may be caused by the absence of cutting along the route of the sprinkler supply pipe, ensuring that some plants effectively function as field refuges, continuously available for oviposition (Levi-Mourao et al. 2022b).

None of the local or landscape structure factors influenced the abundance of H. postica adults. However, this may not be relevant to the generation of adults arising from the spring larvae collected in our samples, which are expected to stay in the field for aestivation during the summer (Prokopy and Gyrisco 1965; Prokopy et al. 1967; Manglitz 1958; Levi-Mourao et al. 2022c). Instead, landscape structure and field characteristics may affect adults recovering after aestivation because they are responsible for mating, oviposition and field colonization during the autumn months (Manglitz 1958; Prokopy and Gyrisco 1965; Prokopy et al. 1967; Levi-Mourao et al. 2022c).

In contrast to our fourth hypothesis, landscape structure variables had only a limited explanatory value concerning the abundance of Bathyplectes spp. adults and none concerning the rate of parasitism. Instead, our results indicated that local variables have a greater influence on both outcomes. However, the abundance of Bathyplectes spp. adults correlated with the abundance of aphids and alfalfa weevil larvae, which probably reflects the resources available in the fields that support parasitoids: weevil larvae hosts and aphid honeydew (Tscharntke et al. 2016; Pellissier et al. 2022). Several studies have shown that the presence of aphids in the field can benefit Bathyplectes spp. by providing food (honeydew) thus increasing parasitoid longevity and their control over the alfalfa weevil larval population (Jacob and Evans 1998; Rand and Lundgren 2019). However, we are the first to demonstrate that aphid abundance favors Bathyplectes spp. adults on a local scale.

The abundance of Bathyplectes spp. adults also correlated positively with alfalfa edge density and negatively with Simpson’s diversity index, but only at the 250-m scale. The positive association with alfalfa edge density may result from the search for food resources near the field borders, where several nectar sources from wild flowers are likely to be found. Most of these spring weeds belong to the family Asteraceae, including the common dandelion (Taraxacum officinale Weber), and have been recognized in other studies for their ability to enhance the longevity and reproductive performance of Bathyplectes spp. (Maingay et al. 1991; Jervis et al. 1993; Jacob and Evans 2000). This result is consistent with the negative association with Simpson’s diversity index and it can be assumed since higher alfalfa edge density corresponds to a larger percentage of the area being occupied by alfalfa crop, indicating a reduction in the surface area of other crops.

We found that the Bathyplectes spp. parasitism rate was negatively correlated with alfalfa age. Given that alfalfa is a pluriannual crop, the density of weeds increases as the crop age and are overexploited (Taberner 2020; Fahrig and Jonsen 1998; Zumoffen et al. 2012), thus reducing the availability of host plants for weevil larvae. Since we did not quantify weed density in our focal fields, further research is needed to determine the role of weeds in the system. In our study the parasitism rate was positively associated with the abundance of weevil larvae, contradicting earlier reports of a negative association (Rand 2013; Pellissier et al. 2022). In these two previous studies, spanning one or two seasons, this negative association was attributed to the inability of B. curculionis to keep up with heavy alfalfa weevil infestations. However, the host–parasitoid relationship can change between years, depending on the population dynamics (Al Ayedh et al. 1996; Costamagna et al. 2004). The differences between our results and the earlier studies (Rand 2013; Pellissier et al. 2022) may also be due to the different proportions of Bathyplectes species that we found. Both previous studies only considered B. curculionis, the major parasitoid of alfalfa weevil larvae in many regions of North America (Rand 2013; Berberet and Bisges 1998; Radcliff and Flanders 1998). In contrast, B. anura accounts for 90% of Bathyplectes specimens collected in the Ebro Basin (Pons and Núñez 2020; Levi-Mourao et al. 2022b). B. anura also has a higher reproductive capacity than B. curculionis, is faster at searching and host-handling, and globally has a more aggressive behavior (Harcourt 1990; Levi-Mourao et al. 2022b).

Conclusion

In conclusion, pests, natural enemies and biological control services do not always respond to the complexity of landscape structure (Rusch et al. 2016; Tscharntke et al. 2016; Karp et al. 2018). Our 3-year study of H. postica at three development stages, and its larval parasitoid Bathyplectes spp., provides the first evidence that H. postica and Bathyplectes spp. respond more to local factors than landscape characteristics under Mediterranean agricultural conditions in Europe. The abundance of alfalfa weevil eggs and larvae was positively associated with longer field perimeters and a sprinkling irrigation system, whereas the landscape composition had no effect and only one configuration metric (alfalfa edge density) explained the larval abundance. The abundance of Bathyplectes spp. correlated with the abundance of alfalfa weevil larvae and aphids, highlighting the role of aphids as a driver, probably by providing honeydew as food. Finally, Bathyplectes spp. parasitism rates mainly depended on the age of the alfalfa plants and the abundance of alfalfa weevil larvae.