Ecological Indicators 124 (2021) 107380

Contents lists available at ScienceDirect

Ecological Indicators

journal homepage: www.elsevier.com/locate/ecolind

Sampling and selection of butterfly indicators for general surveillance of genetically modified maize in north-east Spain

Marina S. Lee *, Agn`es Ardanuy , Alejandro Juarez-Escario´ , Ramon Albajes

Universitat de Lleida, Agrotecnio Center, Av. Rovira Roure 191, 25198 Lleida, Spain

ARTICLE INFO ABSTRACT

Keywords: Genetically modified (GM) maize has been cultivated commercially in Spain since 1998. Although long-term Environmental monitoring environmental monitoring to detect unexpected environmental effects of GM crops (General Surveillance, GS) Farmland is compulsory in the EU, GS currently has a very low capacity to detect adverse effects on the environment. This Non-target Lepidoptera study aimed to increase the feasibility of GS of GM maize expressing insect-resistance (Bt) and herbicide toler­ Genetically engineered crops ance (HT) traits by using butterflies as models. Bt effects Herbicide effects Butterflies (Lepidoptera: Papilionoidea) were sampled using transect-counts in three differentiated maize- Mallow growing regions in north-east Spain. Five transects were established per region and sampled three times per season in two consecutive years. Transects were 300 m long, including 100 m sections in field margins, alfalfa (Medicago sativa) and non-crop vegetation. In addition, butterfly larvae were sampled during maize anthesis in field margins in Lleida region and distribution of larval host plants in maize agroecosystems was assessed in the three regions. Field data and literature were used to construct a step-by-step selection process to identify appropriate but­ terflyindicators for monitoring effects of GM maize cultivation. In addition, suitable multispecies indicators were constructed. The required sampling effort to detect effects using these butterfly indicators was estimated by prospective power analysis. We identified41 butterflyspecies, including three protected species. Most species were potentially exposed to GM maize cultivation effects because their larval host plants were present in maize fields, margins and neigh­ bouring habitats. We identifiedlarvae of four butterfliesin maize fieldmargins, the most abundant of which was Carcharodus alceae. It would be possible to detect a 30% population change by sampling its host plants Malva spp. in 35 to 95 site pairs. When we applied the selection procedure, the most appropriate species for monitoring depended on the region considered. Across regions, the sampling effort using selected indicators was lowest for multispecies groups (i.e. 15–32 site pairs for butterflyabundance) and for the single species Pieris napi and Polyommatus icarus (24–84 and 27–87 site pairs respectively). These indicators could be monitored through existing butterfly monitoring schemes as part of a wider environmental monitoring in agricultural regions to assess impacts of agri- environmental management.

1. Introduction adverse effects anticipated in the pre-release Environmental Risk Assessment (ERA), and General Surveillance (GS) which focuses on The cultivation of genetically modified (GM) maize (Zea mays L.) unexpected, delayed or cumulative effects that could not be detected by varieties may have effects on the receiving environment that only the ERA. GS is compulsory for as long as the GM crop is cultivated. become apparent after widespread or long-term cultivation. For this Despite the compulsory nature of GS, in its current form it has a very reason, Post-Market Environmental Monitoring (PMEM) of GM plants is low capacity to detect effects of GM maize cultivation on the environ­ mandatory in the EU (EC, 2001; EC (European Comission), 2018). There ment: GS is based only on farm questionnaires directed at the farmers are two types of PMEM (EFSA, 2011): Case Specific Monitoring (CSM) (Schmidt et al., 2008) and annual reviews of the scientific literature which is conducted on a case-by-case basis and focuses on potential (EFSA, 2011). Recognising this shortcoming, the European Food Safety

* Corresponding author at: Universitat de Lleida, Agrotecnio Center, Av. Rovira Roure 191, 25198 Lleida, Spain. E-mail addresses: [email protected] (M.S. Lee), [email protected] (A. Juarez-Escario),´ [email protected] (R. Albajes). https://doi.org/10.1016/j.ecolind.2021.107380 Received 6 April 2020; Received in revised form 3 January 2021; Accepted 7 January 2021 Available online 22 January 2021 1470-160X/© 2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license

(http://creativecommons.org/licenses/by-nc-nd/4.0/). M.S. Lee et al. Ecological Indicators 124 (2021) 107380

Authority (EFSA) recommends using data from environmental survey susceptible to Bt toxins and to changes in host plant availability due to networks (ESN) (EFSA, 2014) and integrating GS into a wider environ­ GMHT maize cultivation. They are good indicator organisms (Thomas, mental monitoring scheme but in 2020 this has not yet been 2005; EEA, 2013) and a valued conservation goal (Van Swaay et al., implemented. 1999). Furthermore, many countries in the EU have well-established A GS plan should be capable of detecting any potential effect butterfly monitoring schemes (Schmucki et al., 2016) which could be resulting from GM crop cultivation (EFSA, 2011) but such a plan would used for GS of GM maizes (Lee et al., 2020), as recommended by EFSA require intensive monitoring that would be too costly in terms of sam­ (EFSA, 2014). pling effort. Therefore, the most feasible option to date is a targeted GS focusing on measuring relevant assessment endpoints at critical mo­ 1.1. Objectives ments of exposure to the GM crop cultivation. This requires taking into consideration the potential adverse effects of the cultivation of each This study aims to outline a reliable and cost-effective general sur­ particular GM crop and the particularities of each receiving veillance plan for monitoring the effects of Bt and GMHT maize culti­ environment. vation on biodiversity in Mediterranean conditions. In order to achieve Currently, the only GM crop cultivated commercially in the EU this, we firstidentify the butterfliesand their host plants present in three (Spain and Portugal) is Bt maize MON810 which expresses the Cry1Ab different maize-growing regions in north-east (NE) Spain. Secondly, we toxin conferring resistance to Lepidopteran stemborers (ISAAA, 2019). It use fielddata and literature to develop a selection procedure to identify is very likely that in future GM maize varieties expressing other insect the most appropriate species for monitoring effects of GM maizes. resistance (Bt) or herbicide tolerance (GMHT) traits will be cultivated in Finally, we estimate the sampling effort required for effect detection the EU, particularly in the Mediterranean. Therefore, a sound GS plan using selected butterfly indicators. capable of detecting potential adverse effects of GM maize is needed for Mediterranean receiving environments. 2. Materials and methods In Spain, MON810 is prevalent in areas where pressure from the stemborers is high (Eizaguirre et al., 2006) and in 2018 MON810 2.1. Surveys covered 115,246 ha (ISAAA, 2019). This is over a third of the Spanish maize production area (MAPA, 2020). Worldwide, deployment of Bt In the first place, field surveys were carried out in order to identify crops is an effective control measure allowing farmers to reduce insec­ butterflies and their host plants in maize agroecosystems in NE Spain. ticide treatments (Naranjo, 2009) resulting in a reduction of adverse effects of pesticides on non-target organisms. However, Bt maize pollen 2.1.1. Study regions can be deposited around maize fields (Hofmann et al., 2016; Pleasants The field surveys were carried out in three different maize-growing et al., 2001) and have adverse effects on the larvae of non-target Lepi­ regions in NE Spain to account for variability due to differences in doptera. The adverse effects vary in magnitude depending on the climate, cultural practices or landscape among others. Studies to identify amount of Bt toxin ingested and the susceptibility of the species (Felke other taxa for GS of GM maize were carried out in the same regions (Lee et al., 2002; Hellmich et al., 2001; Lang and Otto, 2010; Lang and & Albajes, 2016; Ardanuy et al., 2018). ◦ ′ Vojtech, 2006; Sears et al., 2001; Zangerl et al., 2001). In the EU, non- The regions were Bujaraloz in the Monegros Badlands (41 29 N, ◦ ′ ◦ ′ ◦ ′ target Lepidoptera such as butterflies have been proposed for environ­ 0 9 O, 328 m.asl), Lleida agricultural plains (41 43 N, 0 26 E, 250 m. ◦ ′ ◦ ′ mental monitoring of Bt maize due to their susceptibility to Bt toxins (E. asl) and La Seu, in the Pre-Pyrenees (42 21 N, 1 27 E, 691 m.asl). g. Arpaia et al., 2018; Aviron et al., 2009; Lang et al., 2019, 2011; Lang Bujaraloz and Lleida are located in the Ebro basin in the same biogeo­ and Bühler, 2012). In particular, monitoring adults and larvae of Aglais graphical region, BSk Koppen-Geigen¨ climate type (Kottek et al., 2006). spp. that develop on nettles (Urtica) has been proposed as a cost-effective Maize is an important summer crop in the Ebro basin, occupying around method for monitoring environmental effects of GM maize because these 16% of the area of irrigated land (Farre´ and Faci, 2009). Pressure from butterflies are susceptible, common around maize fields and easy to Lepidopteran corn-borers is high in this region so it is common to use Bt sample (Lang et al., 2011; Schuppener et al., 2012). Nettle-feeding varieties (MON810). For control of other pests, seed dressing with imi­ species (Aglais sp. and Vanessa atalanta L.) have also been used as dacloprid was common at the moment of the study, but other in­ model species to study risks of Bt maize to non-target organisms (Bau­ secticides were rarely used (Eizaguirre, 2012). Maize is cultivated for drot et al., 2021; Fahse et al., 2018; Holst et al., 2013; Perry et al., 2012, grain, fieldsare ploughed and planting takes place from March to July, 2010). Specificinformation is lacking regarding non-target Lepidoptera pre-emergence herbicides are applied and fertilization is a mixture of in maize agroecosystems in Spain, although research has been con­ mineral NPK and manure. A key difference between the agricultural ducted on effects of both Bt and GMHT maizes on other non-target or­ regions of Bujaraloz and Lleida is that irrigation is fairly recent in ganisms (e.g. De La Poza et al., 2005; Ortego et al., 2009; Albajes et al., Bujaraloz (1970s). As a result, Bujaraloz maize fields are large (8.4 ± 2012; Comas et al., 2014; Arias-Martín et al., 2018). For instance, nettles 1.5 ha). Bujaraloz landscape is composed of arable crops and large are not common around maize fieldsin Mediterranean regions (Lee and patches of native dryland vegetation with little arboreal cover. In Albajes, 2013) and therefore alternative indicators must be identified. contrast, Lleida landscape is that of an intensive agricultural area The deployment of GMHT maize implies changes in the types and withrable crops and orchards covering most of the irrigated areas and application times of herbicides, which modifies flora abundance and very little semi-natural vegetation. Maize fieldsin Lleida were 5.4 ± 0.9 composition in maize fields and field margins (Albajes et al., 2014; ha. Maize field margins in Bujaraloz and Lleida are composed of her­ Hawes et al., 2003). The changes in availability or quality of host plants baceous plants because it is common practice to periodically cut, burn or can result in changes in populations of herbivores such as butterflies apply herbicide for weed control, particularly in Lleida. Recently, (Pleasants and Oberhauser, 2012) and other organisms (Albajes et al., intensification has reached a point that, in addition to the main maize 2009; García-Ruiz et al., 2020; Hawes et al., 2003). For this reason, crop planted in March-April, it is increasingly frequent to plant a second Lepidoptera (butterflies)have been considered to be good indicators for maize crop in May-June, straight after harvesting winter cereals. monitoring GMHT maize effects (Hilbeck et al., 2008). In addition to La Seu region is very different to the other two, with a cooler climate plant-mediated effects, some herbicides have been found to have direct and greater rainfall (Cfb Koppen-Geigen¨ climate type) but it is still toxic effects on non-target organisms, including Lepidoptera (Gill et al., necessary to irrigate maize. Maize is cropped for silage as part of a yearly 2018; Kutlesa and Caveney, 2001; Sz´ekacs´ and Darvas, 2018). crop rotation and agricultural practices include no-till and pre- Butterflies (Lepidoptera, Papilionoidea) are excellent candidates for emergence herbicide applications. Bt maize is not used because corn- monitoring environmental effects of GM maizes because they are borers are not important pests. Average maize field size was 3.0 ±

2 M.S. Lee et al. Ecological Indicators 124 (2021) 107380

0.7 ha and landscape is a mosaic of forage crops and pastures in the abundance per region (mean of 5 sites). valley and pastures and forest on mountain slopes. Field margins are often associated to stone walls and woody plants. 2.1.3. Larval host plant study Alfalfa (Medicago sativa L.) is one of the most common field crops In order to assess the distribution of butterfly larval host plants in present in the study areas in summer (Ardanuy et al., 2018; Clemente- maize agroecosystems information was collected from various sources; a Orta et al., 2020; Madeira et al., 2014). In Bujaraloz and Lleida alfalfa published study on maize weeds was used to determine abundance of cultivation practices are similar, the crop is grown for 4–5 years and cut larval host plants in maize fields, we conducted plant surveys in maize around 5–6 times per season, insecticides are applied for pest control. field margins and non-crop habitats and we searched literature to Alfalfa management in La Seu is coordinated for all fieldsand cuts take identify crop plants that can be larval hosts. place simultaneously. The crop is grown for around 5 years and cut 5 times per season. Fertilization is organic and no insecticides are applied 2.1.3.1. Butterflyhost plants in maize fields. Juarez-Escario´ et al. (2018) (Madeira and Pons, 2016). carried out surveys of maize weeds as part of a study on weed changes in irrigated crops. In brief, 392 maize fieldswere surveyed in the summer 2.1.2. Field survey of adult butterflies of 2009 in the Lleida province. 377 fieldswere surveyed in the counties ◦ ′ ◦ ′ ◦ ′ A two-year survey was carried out in the three maize-growing re­ of Segria,` Pla de l’Urgell (41 45 N, 0 36 E), and Noguera (41 54 N, ◦ ′ gions (Bujaraloz, Lleida and La Seu) to determine the abundance and 0 47 E), which correspond to the Lleida agricultural plains. 15 fields ◦ ′ ◦ ′ frequency of butterfly species present in the maize agroecosystems in were surveyed in Pallars Jussa` (42 5 N, 1 05 E), which is in the vicinity summer. of La Seu region. All plant species were recorded in a rectangle of 6 × 5 Sampling took place three times in each growing season (2012 and m (30 m2), at a distance of 5 m from the field margin. In order to esti­ 2013), July (maize growth stage V3-VT), August (V12-R1) and mate species abundance, an abundance-dominance score between ‘+’ September (VT-R6). The sampling dates were chosen in order to detect and ‘5’ based on the Braun-Blanquet (1979) scale was assigned to each the most abundant and/or frequent species during the period of higher species. The ordinal scores were transformed into mean cover percent­ risk from GM maizes. The moment of highest risk from Bt maizes is ages to allow numerical analysis (‘+’ = 0.1%, ‘1’ = 5%, ‘2’ = 17.5%, ‘3’ during maize anthesis when large amounts of pollen may be deposited = 37.5%, ‘4’ = 62.5% and ‘5’ = 87.5%). Plants were identifiedto species on host plants present around maize plants. The moment of highest risk using local flora (de Bolos` and Vigo, 2001) and plant nomenclature from GMHT maize is more diffuse because changes in flora due to dif­ followed the International Plant Names Index (IPNI, 2020). In this study, ferences in herbicide regime compared to conventional maize can occur we focused on butterfly larval host plants (García-Barros et al., 2013) at any moment of the cropping cycle. In this study the period around that were recorded in at least 5 sites. maize anthesis was selected because we expected that effects of both types of GM maize would be detected. Maize stage nomenclature fol­ 2.1.3.2. Butterfly host plants in the vicinity of maize fields. Butterfly lowed Ritchie et al. (1989), differing growth stages at the same sampling larval host plants were surveyed in fieldmargins and non-crop habitats date were due to differences between the three study regions. in the study sites located in Bujaraloz, Lleida and La Seu, during the Five transects (300 m each) were established per study region, summer of 2012. An additional host plant survey was conducted in field separated by at least 1 km (Map 1). Each transect was divided into three margins in Lleida in 2013 (details given in the following section ‘But­ linear sections (100 m per section) crossing habitat types associated to terfly larvae survey’). Also, literature was used to identify crops grown maize. The section length was established at 100 m to adapt to the small in the study areas that could be used as larval host plants (García-Barros size of maize plots and associated habitats in the study areas. Never­ et al., 2013; MAPA, 2020; Pujol i Palol, 2017). theless, in Lleida in 2013 we increased section length to 200 m to check Butterflylarvae developing on plants in GM maize fieldmargins are if this could contribute to reduce the sampling effort. highly exposed to the potential risks derived from the cultivation of both Butterflies are rarely observed in maize fields and for this reason Bt and GMHT maize. It is unlikely that herbicide will affect habitats not maize field margins, alfalfa and non-crop areas were sampled. Maize adjacent to maize fields such as other crops and non-crop areas but fieldmargins included the bands of vegetation found between fieldsand significant amounts of Bt pollen can be deposited on host plants at dis­ the margins separating the field from roads or waterways (Marshall tances of up to 20 m from maize fields (Messeguer et al., 2006). Flora et al., 1996). Alfalfa was sampled because it is an attractive nectar relev´es were carried out in maize field margins and nearby non-crop source and the larval food plant of various butterfly species (García- areas. Plants were identified within an area of 30 m2 (dimensions Barros et al., 2013). The third habitat sampled were non-crop areas of ranged from 1 × 30 m in narrow margins to 5 × 6 m in non-crop areas) in semi-natural vegetation because maize pollen maydustbutterfly host each site. The procedure for floraidentification and calculation of mean plants in natural areas (Lang et al., 2015) even at distances of 1000 m cover of larval host plants was carried out as described in the previous from the maize field( Hofmann et al., 2016). Sections in alfalfa and non- section. crop areas were usually located at a maximum distance of 20 m from the maize field because this is the area of maximal pollen deposition 2.1.4. Butterfly larvae survey (Messeguer et al., 2006). Butterfly larvae were sampled at maize anthesis in maize field Butterflies were sampled visually by the transect method (Pollard, margins in the Lleida region in 2013 (Map 2). We sampled larvae to 1977; Pollard and Yates, 1993) used by most European Butterfly identify which species were developing in field margins at maize Monitoring Schemes (Schmucki et al., 2016). The observer records all anthesis and to determine the sampling effort involved (Lang et al. × adult butterflies detected within a 5 5 m virtual area along a linear 2011). Two separate anthesis periods were sampled according to transect, obtaining an estimate of relative abundance (for simplicity, we planting date of the maize crops. The firstmaize crop planted around 15- will use the term “abundance” hereafter). Sampling only took place March to 15-April flowers in July (henceforth anthesis I, n = 10). The when meteorological conditions were favourable for butterfly activity. second maize crop planted around 15-May to 15-June following harvest Butterflies were identified to species level if possible based on García- of winter cereal, flowers in August (henceforth anthesis II, n = 12). At Barros et al. (2013) and Tolman and Lewington (2011), taxonomy fol­ each site, two fieldmargins were surveyed (in two sites only one margin lowed Van Swaay et al. (1999). was sampled). In each margin, 100 linear meters were searched, Butterfly abundance was calculated as density per km on a linear considering a one metre width. In total 4200 linear meters were sur­ transect. We first calculated the mean butterfly abundance per year in veyed (4,200 m2). Only larval host plants (García-Barros et al., 2013) of each habitat (mean of 3 sampling dates). This was the basic measure the butterfliesidentified in the fieldsurvey were sampled. The preferred used to calculate mean abundance per site (mean of 3 habitats) or mean

3 M.S. Lee et al. Ecological Indicators 124 (2021) 107380 sampling methodology was visual inspection, but in some cases, frap­ 2. Sensitivity. If the species is highly susceptible to Bt toxins or page was applied because it was more suitable for sampling grasses and sensitive to herbicide, we assigned the species one point in order thorny plants, similar as described by Lang et al. (2013). To quantify to prioritise the few species for which this information is known. sampling effort, we recorded plant number, dimensions, sampling time 3. Responsivenes. Species were prioritised according to ecological or and number of larvae collected. All Lepidoptera larvae (butterflies and biological attributes that could make them more suitable for moths) were collected, reared at the laboratory to adult and identifiedto monitoring: a) Mobility of species, because less mobile species are species if possible based on literature (e.g. Rougeot & Viette, 1980; expected to be the most impacted by onsite changes (Aviron et al., García-Barros et al., 2013). 2009; Hilbeck et al., 2008), giving sedentary species one point Prospective power analysis was used to determine the sampling and low mobility species 0.5 point. Mobility classes were assigned effort needed to detect a 30% population change in larval populations according to literature (Stefanescu et al., 2011a); b) Habitat between GM and non-GM maize field margins (see Section 2.4 for preferences: open-habitat species were given 0.5 point because further details). they would be expected to better reflect changes in areas domi­ nated by arable agriculture. 2.2. Selection criteria for butterfly indicators 4. Conservation value. Protected species according to the European IUCN red list (Van Swaay et al., 1999) or the Spanish red list We developed and applied a step-by-step selection process to identify (Verdú et al., 2011) were prioritised by assigning one point. the most suitable species for monitoring effects of GM maize cultivation 5. Availability of information on species distribution and abun­ in NE Spain. Selection criteria were based on similar studies (e.g. dance, ecology and biology. The biology and ecology of species Schmitz et al., 2003; Hilbeck et al., 2008). used as indicators should be well known. For instance, a) one In the firstplace, all species that were not present both years and in at point was assigned to species used as indicators for monitoring least two of the sites sampled per region and year were excluded. change in agricultural systems in the EU, i.e. European grassland Following this step, further exclusion and prioritisation criteria were indicator species (EEA, 2013) or species used for environmental applied to the widespread candidate species, explained in detail as risk assessment of Bt maize (e.g. Holst et al., 2013; Perry et al., follows: 2012, 2010); b) availability of information on each species was ranked by using the fraction of bibliographic references available A. Exclusion criteria. on the Web of Science (WOS) divided into the highest total 1. Exposure. The first consideration for selecting species for field number of references for any of the species; the search was done monitoring was the exposure of butterflylarvae to Bt maize pollen using the accepted name of the species and its previous taxonomic and/or direct and indirect exposure to herbicides. Species could synonims. be excluded if they were not exposed to Bt toxins and could not be affected by changes in herbicide regimes: a) species with no 2.3. Construction of multispecies indicators overlap between the larval stage and maize growth cycle; b) species not exposed to Bt pollen due to endophytic or below- When butterfly species pools’ are used for monitoring this usually ground larval stage; or c) species whose host plants are not usu­ increases statistical power compared to single species, resulting in a ally found in the vicinity (<20 m) of maize fields. reduction of the sampling effort required for effect detection (Lang, 2. Sensitivity. This can refer to the susceptibility of the species to Bt 2004; Lang et al., 2019; Lang and Bühler, 2012). In addition, it may be toxins expressed in the plant or the pollen (Felke et al., 2010; difficult to find single species that are present in all receiving environ­ Felke et al., 2002 Kjær et al., 2010; Lang and Otto, 2010; Lang and ments. For this reason, single species were aggregated into multispecies Vojtech, 2006; Schuppener et al., 2012) or to sensitivity to direct indicators; indicator composition is given in the supplementary infor­ and indirect effects of modifiedherbicide regimes (Pleasants and mation (Appendix A). The individual species integrating each multi­ Oberhauser, 2012). Species insensitive to glyphosate-based her­ species indicator could vary between sites. The indicator ‘All species’ bicides and Bt toxins could theoretically be excluded. It must be resulted from calculating the abundance of all butterflies recorded in noted, however, that there are no reported cases of non-target any given site and included butterflyspecimens not identifiedto species Lepidoptera being insensitive to the Bt toxins targetting Lepi­ level. “Mobile species” aggregated any species with high dispersion dopteran pests. capacity but that don’t migrate across the study area (Stefanescu et al., 3. Responsiveness. Species selected for monitoring should reflect 2011a). “Low mobility species” aggregated any species with low changes of the system. Crop pests or species with a strong dispersal capacity (sedentary species and species with a limited dispersal migratory behaviour in the study areas (Stefanescu et al., 2011b) capacity). “Open habitat species” aggregated species linked to open were excluded because it may be more difficultto establish causal habitats (from Herrando et al., 2016). “Grassland indicators” aggregated effects between fluctuations in the species’ abundance and any of the European grasslands indicator species (EEA, 2013). Finally, changes at the local scale. species number was included because it allows to further reduce sam­ B. Prioritisation criteria. In most cases there was insufficient informa­ pling effort (Lang and Bühler, 2012). tion regarding the direction of the possible impacts of GM maizes on butterflies so ranking criteria were applied to select the candidates 2.4. Prospective power analysis most appropriate for monitoring. Following the exclusion process, protected species (Van Swaay et al., 1999) and maize-feeding species Prospective power analysis was carried out to estimate the sample (critically exposed to Bt toxins) were reincorporated as potential sizes (always expressed as number of site pairs) needed to detect a candidates. The following aspects were considered: change in butterflypopulations between GM and non-GM sites using an 1. Exposure. Species most exposed to potential risks were ranked unpaired two sample t-test (Perry et al., 2009). The probability of positively. One point was assigned to a species if: a) it was committing a type I error (α) was set at 0.05 and type II error (β) was set recorded in both consecutive years; b) the species’ host plant was at 0.2, (statistical power = 0.8). The statistical power measures the present in or around maize fields; c) larvae of the species were chance of detecting an effect of a known magnitude using a specified identifiedduring the larval survey; and d) species could also feed experimental design. The effect size was established as a 30% change on maize. regarding the comparator population (non-GM sites), considered adequate for this type of studies (Comas et al., 2013; Perry et al., 2003). Abundance data were transformed by log10 (x + 1) for normalization

4 M.S. Lee et al. Ecological Indicators 124 (2021) 107380 and power was calculated with the (JMP Pro®, 2019) software. SD). In Bujaraloz, abundance was 79.4 ± 40.5 butterflies/km and 16 Prospective power analysis was used to determine sampling effort species were recorded. Three species represented 64% of the counts required to detect a population change in populations of butterflyadult (Pieris rapae (L.), Polyommatus icarus (Rottemburg) and Pieris napi (L.)). and larvae. For butterfly adults, comparator populations were approxi­ In Lleida, butterfly abundance was 46.8 ± 24.6 butterflies/km and 15 mated by calculating average annual butterflyabundance in each region species were detected. The same three species as in Bujaraloz repre­ (Bujaraloz, Lleida or La Seu). sented 64% of the total counts. Bujaraloz and Lleida shared almost all species, with three exceptions. Pararge aegeria (L.) was not recorded in 3. Results Bujaraloz where there is no arboreal cover. T. acteon was frequent in Bujaraloz (present in 40% of the samplings) but not recorded in Lleida 3.1. Field survey of adult butterflies where non-crop areas are much smaller. Gegenes nostrodamus (Fab­ ricius), a migrant butterflythat can feed on maize, was recorded only in A total of 41 butterflyspecies were recorded during the fieldsurvey Bujaraloz. of maize agroecosystems in NE Spain (Table 1); the dataset is available In La Seu, butterfly abundance was 61.1 ± 43.3 butterflies/km and at Mendeley Data (Lee, 2020a). We detected three protected species, 37 different species were recorded. Nettle butterfliesA. io and A. urticae according to the EU Red List (Van Swaay et al., 1999): Carcharodus were recorded, as well as other species common to cooler and more flocciferus (Zeller), Hipparchia fagi (Scopoli) and Thymelicus acteon Rot­ humid EU. temburg. Butterfly abundance (including identified and unidentified Only twelve species were shared across the three regions, and only specimens) was 62.4 ± 38.2 butterflies/km(mean ± standard deviation, fiveof these were present in at least 50% of the counts across the entire

Table 1 Adult butterflies(Lepidoptera: Papilionoidea) recorded in maize agroecosystems in NE Spain. Mean (m) number of butterfliesper km and standard deviation (SD) were calculated by averaging the mean number of butterflies recorded per site (5 sites per region). Three protected species were detected (Near threatened (NT) status according to the IUCN red list (Van Swaay et al., 1999).

Family/group Species IUCN Bujaraloz Lleida La Seu

m ± SD %Fr m ± SD %Fr m ± SD %Fr

Hesperiidae Carcharodus alceae (Esper) 1.2 ± 2.1 40 1.3 ± 2.1 50 0.4 ± 0.8 30 Carcharodus baeticus (Rambur) 0.1 ± 0.4 10 Carcharodus flocciferus (Zeller) NT 0.2 ± 0.7 10 Gegenes nostrodamus (Fabricius) 0.1 ± 0.4 10 Muschampia proto (Ochsenheimer) 0.2 ± 0.7 10 Pyrgus malvoides (Elwes & Edwards) 0.3 ± 0.7 20 Spialia sertorius (Hoffmannsegg) 0.1 ± 0.4 10 Thymelicus acteon Rottemburg NT 1.8 ± 2.7 40 0.1 ± 0.4 10 Thymelicus lineola (Ochsenheimer) 0.1 ± 0.4 10

Lycaenidae Aricia agestis (Dennis & Schiffermüller) 0.1 ± 0.4 10 Celastrina argiolus (L.) 0.1 ± 0.4 10 Cupido argiades Pallas 0.2 ± 0.5 20 Lampides boeticus (L.) 2.0 ± 3.0 60 0.8 ± 1.1 50 15.0 ± 33.6 90 Leptotes pirithous L. 0.4 ± 0.6 40 0.1 ± 0.3 10 0.3 ± 0.7 20 Lycaena phlaeas (L.) 0.3 ± 0.5 30 Polyommatus icarus (Rottemburg) 11.8 ± 11.7 100 10.3 ± 6.3 100 4.8 ± 6.7 80 Satyrium esculi (Hübner) 0.3 ± 1.1 10 Satyrium spini Dennis & Schiffermüller 0.1 ± 0.4 10

Nymphalidae Aglais io L. 1.4 ± 2.5 50 Aglais urticae (L.) 0.2 ± 0.5 20 Coenonympha pamphilus (L.) 0.7 ± 0.9 40 Hipparchia fagi (Scopoli) NT 1.0 ± 1.8 40 Lasiommata megera (L.) 1.7 ± 2.4 60 0.6 ± 1.2 20 1.6 ± 1.3 70 Maniola jurtina L. 0.7 ± 1.1 40 Melanargia lachesis Hübner 0.8 ± 0.9 50 Melitaea didyma Esper 0.1 ± 0.4 10 Pararge aegeria (L.) 0.1 ± 0.4 10 1.9 ± 2.3 60 Polygonia c-album (L.) 0.3 ± 0.5 30 Pyronia bathseba (Fabricius) 0.1 ± 0.4 10 Pyronia cecilia (Vallantin) 2.8 ± 3.8 60 1.3 ± 2.3 30 1.1 ± 2.1 30 Pyronia tithonus (L.) 0.8 ± 1.2 40 Vanessa atalanta (L.) 0.1 ± 0.4 10 Vanessa cardui (L.) 0.1 ± 0.4 10 1.8 ± 4.1 50 0.9 ± 1.4 40

Papilionidae Iphiclides feisthamelii (Duponchel) 0.2 ± 0.5 20 Papilio machaon L. 1.3 ± 1.9 40 0.2 ± 0.5 20

Pieridae Colias crocea (Geoffroy) 9.5 ± 8.7 100 3.9 ± 3.5 70 4.1 ± 2.2 100 Gonepteryx rhamni (L.) 0.1 ± 0.4 10 Pieris brassicae (L.) 0.6 ± 1.2 20 0.4 ± 1.1 20 2.1 ± 1.8 70 Pieris napi (L.) 10.8 ± 9.5 80 12.7 ± 13.9 90 5.6 ± 4.6 100 Pieris rapae (L.) 28.2 ± 20.0 100 6.8 ± 7.0 90 7.7 ± 6.0 90 Pontia daplidice (L.) 2.2 ± 2.1 70 1.9 ± 2.0 70 0.3 ± 0.5 30 Papilionoidea* All 9.4 ± 40.5 100 46.8 ± 24.6 100 61.1 ± 43.3 100 Species richness 16 15 37

*Papilionoidea includes identified species and butterflies not identified to species.

5 M.S. Lee et al. Ecological Indicators 124 (2021) 107380 study area (Table 1): the Lycaenidae Lampides boeticus (L.) and P. icarus; lucida (Hufnagel) (Noctuidae) and Pardoxia graellsii (Feisthamel) and the Pieridae Colias crocea (Geoffroy), P. napi and P. rapae. Most of (Nolidae). these species are migrants in the study area. The most numerous butterfly larvae were C. alceae and P. machaon. Although the affinityof the butterflyspecies for each habitat was not One of the host plants of C. alceae, M. sylvestris, was found in 91% of the analysed, some species were recorded across all habitat types in all re­ 22 sites sampled. A total of 23 larvae were collected in July and 57 gions, such as Carcharodus alceae (Esper) or P. icarus. Other species were larvae in August; the searching time to findone larva on M. sylvestris in frequently found in non-crop areas but rarely in maize field margins or any given margin was 9.7 min in July and 3.6 min in August (Table 2). alfalfa, such as Pyronia cecilia (Vallantin) and Pontia daplidice (L.) (Ap­ According to prospective power analysis, it would be necessary to pendix B). In alfalfa, L. boeticus, P. icarus, C. crocea, P. napi and P. rapae sample 35 to 95 site pairs (GM vs. non-GM site pairs) to detect a 30% were abundant in all regions, particularly when the alfalfa was difference in abundance of C. alceae larvae between sites at anthesis I flowering. and II, respectively (α = 5, β = 0.8). In the case of P. machaon, the host plant F. vulgare was recorded in 64% of the 22 sites. 17 larvae were 3.2. Larval host plant study collected in July but only 2 small larvae in August. This means that the time to findone larva on F. vulgare in any given margin was 5.2 min in 3.2.1. Butterfly host plants in maize fields July but it rose to 26.8 min in August (Table 2). According to prospective According to the results obtained by Juarez-Escario´ et al. (2018), 33 power analysis, it would be necessary to sample 62 to 787 site pairs (GM species from 10 plant families were recorded as weeds of maize. The vs. non-GM site pairs) in order to detect a 30% difference in abundance most abundant and frequent plant was the Malvaceae Abutilon theo­ of P. machaon larvae between sites at anthesis I and II, respectively. phrasti (Medik), recorded in 24% of the 392 sites sampled and with a mean cover of 2.5 ± 10.8% (SD). This plant is a regular larval host to 3.4. Selection of butterfly indicator species C. alceae in the region (García-Barros et al. 2013). Other frequent weeds belonged in families Poaceae, Asteraceae, Polygonaceae, Brassicaceae, From the initial 41 butterfly species, 25 species were recorded in at Rosaceae, Malvaceae and Plantaginaceae (Appendix C). Overall, the least two sites in any given region and year. Only 20 species were weeds recorded in maize fieldscould host larvae of 20 butterflyspecies, detected both years in any given region. After applying the exclusion according to literature (García-Barros et al., 2013). criteria (Fig. 1), eight species were excluded because they were migrants in the study area. Two of the migrants were also crop pests, and one had 3.2.2. Butterfly host plants in the vicinity of maize fields an endophytic larval stage. Thus, we were left with 12 candidate species. Most larval host plants were recorded mainly in field margins (Ap­ At this stage, we reincorporated the protected species and the single pendix C), dataset available at Mendeley Data (Lee, 2020b). However, maize-feeding species to the list of candidates resulting in a list of 15 there were also larval food plants in the non-crop areas, and sometimes species (seven species in Bujaraloz, six in Lleida and 14 in La Seu). We this was the only place where larval host plants were recorded. For applied the ranking process to select the most suitable species for instance, P. cecilia feeds on the grass Brachypodium retusum (Pers.) a monitoring GM maize and selected the six candidates with the highest plant native to the arid landscapes of Bujaraloz and Lleida. rank per region (Appendix D). For Bujaraloz and Lleida most candidates In addition, we identified 25 crops that can be used as larval host were shared (Pieris napi, Polyommatus icarus, Lasiommata megera, Pyronia plants across the study regions (Appendix C). cecilia and Carcharodus alceae) although the ranking differed between regions (Fig. 1). There were two exceptions: in Bujaraloz T. acteon was 3.3. Butterfly larvae survey and sampling effort the most suitable indicator (with the highest score) but this species was not recorded in Lleida. Similarly, in Lleida P. aegeria was a suitable in­ In July (Anthesis I), 1034 butterflyfood plants were searched and 49 dicator species but it was not recorded in Bujaraloz. Conversely, in La butterflylarvae and 35 moth larvae were found. In August (Anthesis II), Seu, the six most suitable species were the nettle butterfly Aglais io L., 774 host plants were searched and 60 butterfly larvae and 41 moth followed by Coenonympha pamphilus (L.), T. acteon, P. napi, P. icarus and larvae were collected. Although 34 different plant genus or species were Lycaena phlaeas (L.). The only candidate species common to all three sampled, butterflylarvae were only found on four plant species (Table 2, regions were P. napi and P. icarus. dataset available at Mendeley Data (Lee, 2020c)). C. alceae and Vanessa cardui (L.) were found on mallow, Malva sylvestris (L.); Papilio machaon L. 3.5. Required sampling effort of butterfly indicators for effect detection on fennel, Foeniculum vulgare Mill; and Leptotes pirithous (L.) on alfalfa. More Lycaenidae larvae were collected from alfalfa and Ononis spinosa The sampling effort required to detect a 30% population change was (L.) but larvae died so it was not possible to identify the species. Various very variable for each indicator across regions and years (Table 3). moth species were recorded; the most abundant on M. sylvestris: Acontia Regarding single species, required sampling effort (number of GM vs

Table 2 Butterfly larvae recorded in maize field margins in the Lleida region. The table shows the plant species on which butterfly larvae were found. Mean abundance, standard deviation (SD) and frequency (%Fr) of the butterflylarvae is given per site. Larvae were sampled during the floweringperiod of maize crops: July (anthesis I, 10 sites) and August (anthesis II, 12 sites). The sampling effort is shown as the number of minutes required to finda single larva on the host plant at any given site.

Anthesis I Anthesis II

larvae host plant larvae host plant

Host plant Butterfly mean SD % Mean cover SD Effort (min/ mean SD % Mean cover SD Effort (min/ Fr (%) larva) Fr (%) larva)

Foeniculum Papilio machaon 1.6 2.9 30 2.2 1.6 5.2 0.3 0.7 8 2.8 4.5 26.8 vulgare Malva sylvestris Carcharodus 1.2 1.5 40 1.7 1.7 9.7 2.9 5.6 58 1.6 1.9 3.6 alceae Vanessa cardui 0.1 0.3 10 101.5 0.1 0.2 10 144.6 Medicago sativa Leptotes pirithous 0.1 0.2 10 15.8 7.8 221.4 0.8 1.0 20 1.0 1.5 3.4 Lycaenidae sp. 0.3 0.5 20 55.4 0.1 0.2 8 39.6 Ononis spinosa Lycaenidae sp. 0.8 1.0 20 3.6 2.2 18.7 0 . . 0.5 0.3 .

6 M.S. Lee et al. Ecological Indicators 124 (2021) 107380

Fig. 1. Flowchart depicting the process followed to select suitable butterfly species for monitoring effects of GM maize and the final list of the highest ranking six candidates selected per region. The selection process was only applied to species that were recorded both years and present in at least two sites in any given region and year. non-GM site pairs) would be consistently lowest for P. napi (24–84) and regions. This finding shows that many butterflies are exposed to po­ P. icarus (27–87) across the three regions and both sampling years, using tential effects of GM maize cultivation in Mediterranean conditions and 300 m transects per site. When butterflies were aggregated into multi­ that effects on their populations should be taken into account. This species pools, the sampling effort was even lower. For instance, it would contrasts to the belief that butterflies are not common in the maize be possible to detect a 30% decrease in species’ number by monitoring agroecosystems in Spain. For instance, the mathematical models 7–27 site pairs, or a 30% decrease in abundance of all butterflies or developed to assess risks of Cry1Ab from MON810 to non-target Lepi­ mobile butterflies (by monitoring 15–32 and 16–41 site pairs, doptera in the EU (Perry et al., 2010), did not include any Lepidoptera respectively). representative for Spain, because they were not considered to be very The number of required sampling sites can usually be reduced by abundant during maize anthesis. This assumption was probably mostly increasing transect length. For this reason, we used 300 and 600 m due to a lack of data from butterfly surveys in Mediterranean maize transects in Lleida in year 2 but this did not appear to reduce sampling agroecosystems. Other authors that conducted field surveys also found effort (Table 3). butterflies to be relatively abundant and diverse in maize agro­ ecosystems, particularly in fieldmargins (e.g. Arpaia et al., 2018; Aviron 4. Discussion et al., 2009; Lang et al., 2019; Wallis de Vries et al., 2017). Although many butterflyspecies were shared between Bujaraloz and 4.1. Adult butterflies across maize agroecosystems Lleida, only a third of the species were shared across the three regions. This outcome could be expected given the divergent biogeographic and Butterflies were abundant in all habitats sampled in the study agricultural particularities of each maize-growing region (Dolezel et al.,

7 M.S. Lee et al. Ecological Indicators 124 (2021) 107380

Table 3 instance, in the US, at the beginning of field deployment of Bt maize Sampling effort, in number of site pairs, needed to detect a 30% change in there was considerable concern that this crop could have adverse effects butterfly abundance or species number between GM vs non-GM maize fields in on the monarch butterflypopulations (Pleasants et al., 2001; Sears et al., each region using an unpaired t-test. The sampling effort was calculated by 2001; Stanley-Horn et al., 2001). Nevertheless, it was found that expo­ α = β = prospective power analysis ( 0.05 and 0.2) on transformed data (log10(x sure of monarch larvae to Bt toxins in the field was relatively low + 1)). (Anderson et al., 2005) and they were not very susceptible to most Bt Bujaraloz Lleida La Seu toxins compared to other Lepidoptera species (Perry et al., 2012; Wolt Butterfly Year Year Year Year Year Year Year et al., 2005). However, when GMHT maize became widely cultivated indicator 1 2 1 2 2* 1 2 across the US corn belt, increasing glyphosate treatments reduced larval Species host plants, resulting in the decline of the monarch population (Pleas­ Aglais io 394 126 ants and Oberhauser, 2012). Carcharodus 567 103 126 191 232 222 567 alceae 4.3. Monitoring butterfly larvae Coenonympha 567 113 pamphilus Lasiommata 191 110 185 165 240 22 We recorded 15 butterflyspecies as adults in the Lleida region but we megera only found larvae of four of those species. The abundance and richness of Lycaena phlaeas 567 278 the butterfly larvae was low considering the relatively high sampling Pararge aegeria 567 143 567 40 Pieris napi 84 24 68 37 34 54 32 effort invested (1808 butterfly host plants sampled across 22 sites). It Polyommatus 54 35 27 42 41 81 87 was clear that the sampling effort required for effect detection was too icarus high for a monitoring plan. This outcome is not surprising, because Pyronia cecilia 185 68 80 62 319 242 larvae are more difficult to detect than adult butterflies. Thus, it is Thymelicus acteon 133 191 567 frequent to find low numbers of larvae (e.g. Arpaia et al., 2018; Gath­ Multispecies mann et al., 2006) resulting in a higher sampling effort compared to groups adult butterflies (Lang et al., 2011). All species 26 17 15 23 21 32 22 Different strategies can be used to reduce the sampling effort needed Mobile species 41 16 24 30 28 31 22 Low mobility 64 44 567 80 58 30 29 to detect larvae, such as mapping larval host plants and using a clear species sampling strategy. For instance, Lang et al. (2013), Lang et al. (2011) Open habitat 41 25 21 29 41 47 35 proposed the survey of nettle stands around the maize fields as a cost- species effective sampling strategy. However, in our study conditions nettles Grassland 41 24 19 43 42 72 33 indicator were not widespread across maize-growing regions. Thus, mallow species (Malva spp.) and Abutilon theophrasti could be a more promising group of Species’ number 27 7 13 19 17 9 13 plants to monitor because they occurred in all study areas and they are *In Lleida, 300 and 600 m transects were used in 2013. larval hosts to a number of Lepidoptera. For instance, in Lleida, C. alceae Note: some species were not observed in some regions and years so there is no larvae were abundant on mallow and the larvae were easy to spot data. because they fold the leaves to form a pouch. Arpaia et al. (2018) also sampled mallow around maize fields to detect larvae of V. cardui. 2018; Lang et al., 2019). This fact highlights the need to perform surveys Nevertheless, larval sampling for monitoring effects of GM maize in the in each differentiated maize-growing region because there can be broad Mediterranean region should be further studied to determine the most differences regarding the species most affected by GM maize cultivation suitable strategy across different regions. across receiving environments (Arpaia, 2021). In the arid Mediterranean, maize tends to be grown in intensive 4.4. Selection of indicator species for monitoring GM maize agricultural settings where non-crop areas are relatively small (Ardanuy et al., 2018; Clemente-Orta et al., 2020). This is probably why most of In this study we constructed and applied a selection process to the species found in Bujaraloz and Lleida were highly mobile habitat identify species appropriate for GS of GM maize in three differentiated generalists that concentrate in the humid environments resulting from maize-growing regions in NE Spain and then calculated the required irrigated agriculture whereas sedentary dryland specialists were present sampling effort. The selection process was based on the potential risks in lower numbers. Conversely, La Seu has a much more humid climate and the pathways through which they could be realised, similar to other and a broader range of habitats due to its biogeographical situation, authors (Hilbeck et al., 2008; Schmitz et al., 2003; Van Wyk et al., 2007) leading to a high number of species compared to the lowland regions. but we also took into account the capacity of the butterfly species to Indeed, La Seu was the only region where nettle butterflies A. io and reflect local impacts of GM maize. This is a highly relevant aspect that A. urticae were present during the period of maize anthesis. has often been overlooked in other studies. For instance, pest species and migrant species tend to be very abundant in farmland and are therefore suitable for statistical analysis (Comas et al., 2013). Nevertheless, link­ 4.2. Distribution of larval host plants ing measured differences in local butterfly populations to the effect of GM crop cultivation (EFSA, 2011) can be more difficult. For instance, When selecting butterflies for environmental risk assessment it is pest species’ abundance can depend on the host crop area and man­ essential to determine the distribution of their larval host plants in the agement. In the case of migrant species, their abundance may be linked maize agroecosystem. This allows to infer the exposure of butterfly to conditions at their place of origin (Stefanescu et al., 2011b). Never­ species to risks derived from the cultivation of GM maize. In this study theless, migratory species may be the most suitable option for a large- we found that almost all butterfliesrecorded had larval host plants in or scale monitoring plan in a future scenario when GM maize is grown close to maize fieldsso they would be exposed to GM maize cultivation across wide areas in the EU. Particularly, if the migratory species de­ to some degree. pends mainly on larval plants present in habitats associated to maize, as Weeds are not abundant within maize fields due to herbicide found in the case of the Monarch butterfly in the US (Pleasants and spraying, hence butterfly larvae would use mostly field margins where Oberhauser, 2012). However, considering that the only EU countries larval host plants are abundant (this study and Pywell et al., 2004; Lang that currently grow maize are Spain and Portugal (ISAAA, 2019), a GS et al., 2013; Arpaia et al., 2018; Wallis de Vries et al., 2017). For plan capable of detecting local impacts would appear most suitable at

8 M.S. Lee et al. Ecological Indicators 124 (2021) 107380 present. strongly recommend to carry out a field survey in each differentiated The most relevant result of our selection process was the clear dif­ receiving environment. The results of this study indicate that an ferences in candidate species between the three maize-growing regions improved GS could be implemented by monitoring selected butterfly studied, evidencing the need to conduct a specific selection process indicators such as Pieris napi, Polyommatus icarus and multispecies taking into account the particularities of each cultivation area, as groups. However, further research is needed in order to determine which already stressed by other authors (Arpaia, 2021; Dolezel et al., 2018). In and how butterfly indicators should be monitored across wider GM our study, the nettle butterflyA. io was selected for GS of GM maize in La maize cultivation areas. Seu, in agreement with the numerous studies focusing on these species for the risk assessment of Bt maize in the more humid EU (Arpaia et al., CRediT authorship contribution statement 2018; Fahse et al., 2018; Holst et al., 2013; Leclerc et al., 2018; Perry et al., 2010). However, it was clear that in the more arid regions the most Marina S. Lee: Investigation, Methodology, Writing - original draft. suitable indicators had to be species common in irrigated agricultural Agnes` Ardanuy: Methodology. Alejandro Juarez-Escario:´ Investiga­ environments (P. napi, P. icarus or C. alceae). tion. Ramon Albajes: Supervision, Funding acquisition. Considering sampling effort, it was consistently lowest for multi­ species indicators and the species P. napi and P. icarus in all regions considered. Monitoring P. napi is a good option because it is a wide­ Declaration of Competing Interest spread species in irrigated agricultural land in the Mediterranean and does not feed on common crops in the study area (García-Barros et al., The authors declare that they have no known competing financial 2013). Its larvae develop on plants common in irrigated field margins interests or personal relationships that could have appeared to influence throughout the maize growing season so it is exposed to effects of both Bt the work reported in this paper. and GMHT maize cultivation. Similarly, P. icarus could also be suitable for monitoring GM maize effects because it is widespread and common Acknowledgements in European farmland where it is monitored as part of the Grassland indicator (EEA, 2013). Nevertheless, P. icarus feeds on many Fabacea­ We thank M. Eizaguirre, C. Lopez,´ C. Stefanescu, O. Alomar, A. Lang, eous crops such as alfalfa so its abundance can depend on crop distri­ M. Wallis de Vries and three anonymous reviewers for their valuable bution and management. Thus making it difficult to link potential advice. We also thank colleagues who helped with fieldwork (Heather, butterfly population decline to GM maize cultivation. Melisa, Esther and JuanCarlos) and floraidentification (J. Recasens and Monitoring species’ pools allows to monitor butterfliesacross broad X. Sole-Senan).´ This work was supported by a grant from the Catalan geographical areas even when there are large differences in the distri­ Government (AGAUR FI-DGR) and the Spanish Government-funded bution of single species. In addition, it generally allows to gain in sta­ project AGL2011-23996. tistical power and thus reduce sampling effort (Lang et al., 2016; Lang and Bühler, 2012). In this study the sampling effort to detect a 30% Appendix A. Supplementary data decrease was lowest for species number (7–27 site pairs) and for overall butterfly abundance (15–32 site pairs) and mobile species (16–41 site Supplementary data to this article can be found online at https://doi. pairs). However, the grassland indicator group would be the most suit­ org/10.1016/j.ecolind.2021.107380. These data include Google maps able group despite a higher required sampling effort (19–72 site pairs). of the most important areas described in this article. The group of species that form this indicator are already surveyed for monitoring change in agricultural environments across Europe (EEA, References 2013) so GS of maize could be integrated into a community-wide monitoring plan, for instance to assess the effects of the measures Albajes, R., Farinos,´ G.P., Perez-Hedo, M., De la Poza, M., Lumbierres, B., Ortego, F., implemented through the EU’s Common Agricultural Policy (CAP) Pons, X., Castanera,˜ P., 2012. Post-market environmental monitoring of Bt maize in (Lefebvre et al., 2015). Spain: Non-target effects of varieties derived from the event MON810 on predatory fauna. Spanish J. Agric. Res. 10, 977. DOI:10.5424/sjar/2012104-691-11. Usually, sampling effort in terms of number of sites can be reduced Albajes, R., Lumbierres, B., Pons, X., 2009. Responsiveness of arthropod herbivores and by increasing transect length or sampling frequency (Brereton et al., their natural enemies to modified weed management in corn. Environ. Entomol. 38 2011; Lang and Bühler, 2012). This was not observed in this study but it (3), 944–954. https://doi.org/10.1603/022.038.0349. Albajes, R., Lumbierres, B., Pons, X., Comas, J., 2014. Changes in arthropod fauna from merits further research because transect length and sampling frequency weed management practices in genetically modified herbicide-tolerant maize. in the field survey were rather low compared to other studies (Lang J. Agric. Sci. 6, 67–78. https://doi.org/10.5539/jas.v6n10p67. et al., 2019, 2016, 2013). Nevertheless, Lang and Bühler (2012) found Anderson, P.L., Hellmich, R.L., Prasifka, J.R., Lewis, L.C., 2005. Effects on fitness and behavior of monarch butterfly larvae exposed to a combination of Cry1Ab that using transects of a similar length to those in our study (300 m) and -expressing corn anthers and pollen. Environ. Entomol. 34 (4), 944–952. https://doi. the same sampling frequency (3 visits) could still capture around 70% of org/10.1603/0046-225X-34.4.944. the species present. Ardanuy, A., Lee, M.S., Albajes, R., 2018. Landscape context influences leafhopper and predatory Orius spp. abundances in maize fields: Orius spp. and leafhoppers in The indicator selection process here described is broadly applicable arable landscapes. Agr. Forest. Entomol. 20 (1), 81–92. https://doi.org/10.1111/ to any Mediterranean maize-growing region. However, in each case a afe.12231. fieldsurvey of the butterfliesand their host plants is required to identify Arias-Martín, M., García, M., Castanera,˜ P., Ortego, F., Farinos,´ G.P., 2018. Farm-scale the most appropriate indicators. evaluation of the impact of Cry1Ab Bt maize on canopy nontarget arthropods: a 3- year study: Impact of Bt maize on canopy NTAs. Insect Sci.. 25 (1), 87–98. https:// doi.org/10.1111/1744-7917.12378. 5. Conclusions Arpaia, S., 2021. Environmental risk assessment in agro-ecosystems: revisiting the concept of receiving environment after the EFSA guidance document. Ecotoxicol. Environ. Saf. 208, 111676. https://doi.org/10.1016/j.ecoenv.2020.111676. Butterflies and their larval host plants were widespread and abun­ Arpaia, S., Baldacchino, F., Bosi, S., Burgio, G., Errico, S., Magarelli, R.A., Masetti, A., dant in the maize agroecosystems surveyed and therefore they could be Santorsola, S., 2018. Evaluation of the potential exposure of butterfliesto genetically exposed to effects of GM maize cultivation. In contrast, few butterfly modified maize pollen in protected areas in Italy: exposure of butterflies to maize pollen. Insect Sci.. 25 (4), 549–561. https://doi.org/10.1111/1744-7917.12591. larvae were recorded in maize field margins and in most cases the Aviron, S., Sanvido, O., Romeis, J., Herzog, F., Bigler, F., 2009. Case-specificmonitoring sampling effort involved would be too high. of butterflies to determine potential effects of transgenic Bt-maize in Switzerland. – A number of adult butterfly indicators appeared appropriate for GS Agric. Ecosyst. Environ. 131 (3-4), 137 144. https://doi.org/10.1016/j. agee.2009.01.007. of GM maizes. However, we found that indicator species most suitable Baudrot, V., Walker, E., Lang, A., Stefanescu, C., Rey, J.-F., Soubeyrand, S., Messean,´ A., for GS could vary between maize-growing regions. For this reason, we 2021. When the average hides the risk of Bt-corn pollen on non-target Lepidoptera:

9 M.S. Lee et al. Ecological Indicators 124 (2021) 107380

application to Aglais io in Catalonia. Ecotoxicol. Environ. Saf. 207, 111215. https:// crops. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 358, 1899–1913. https://doi.org/ doi.org/10.1016/j.ecoenv.2020.111215. 10.1098/rstb.2003.1406. Braun-Blanquet, J., 1979. Fitosociología: Bases Para el Estudio de las Comunidades Hellmich, R.L., Siegfried, B.D., Sears, M.K., Stanley-Horn, D.E., Daniels, M.J., Mattila, H. Vegetales (Pflanzensoziologie : Grundzüge der Vegetationskunde). Blume, Madrid, R., Spencer, T., Bidne, K.G., Lewis, L.C., 2001. Monarch larvae sensitivity to Bacillus Spain. thuringiensis-purified proteins and pollen. Proc. Natl. Acad. Sci. U. S. A. 98, Brereton, T., Roy, D.B., Middlebrook, I., Botham, M., Warren, M., 2011. The 11925–11930. https://doi.org/10.1073/pnas.211297698. development of butterflyindicators in the United Kingdom and assessments in 2010. Herrando, S., Brotons, L., Anton, M., Paramo,´ F., Villero, D., Titeux, N., Quesada, J., J. Insect Conserv. 15 (1-2), 139–151. https://doi.org/10.1007/s10841-010-9333-z. Stefanescu, C., Herrando, S., Brotons, L., Anton, M., Paramo,´ F., Villero, D., Clemente-Orta, G., Madeira, F., Batuecas, I., Sossai, S., Juarez-Escario,´ A., Albajes, R., Titeux, N., Quesada, J., Stefanescu, C., 2016. Assessing impacts of land abandonment 2020. Changes in landscape composition influence the abundance of insects on on Mediterranean biodiversity using indicators based on bird and butterfly maize: the role of fruit orchards and alfalfa crops. Agric. Ecosyst. Environ. 291, monitoring data. Environ. Conserv. 43, 69–78. https://doi.org/10.1017/ 106805. https://doi.org/10.1016/j.agee.2019.106805. S0376892915000260. Comas, J., Lumbierres, B., Pons, X., Albajes, R., 2013. Ex-ante determination of the Hilbeck, A., Meier, M., Benzler, A., 2008. Identifying indicator species for post-release capacity of field tests to detect effects of genetically modified corn on nontarget monitoring of genetically modified, herbicide resistant crops. Euphytica 164, Arthropods. J. Econ. Entom. 106 (4), 1659–1668. https://doi.org/10.1603/ 903–912. https://doi.org/10.1007/s10681-008-9666-9. EC12508. Hofmann, F., Kruse-Plass, M., Kuhn, U., Otto, M., Schlechtriemen, U., Schroder,¨ B., Comas, C., Lumbierres, B., Pons, X., Albajes, R., 2014. No effects of Bacillus thuringiensis Vogel,¨ R., Wosniok, W., 2016. Accumulation and variability of maize pollen maize on nontarget organisms in the fieldin southern Europe: a meta-analysis of 26 deposition on leaves of European Lepidoptera host plants and relation to release arthropod taxa. Transgenic Res. 23 (1), 135–143. https://doi.org/10.1007/s11248- rates and deposition determined by standardised technical sampling. Environ. Sci. 013-9737-0. Eur. 28, 14. https://doi.org/10.1186/s12302-016-0082-9. de Bolos,` O., Vigo, J., 2001. Flora dels Països Catalans IV. Monocotiledonies,` Vol.1-4. ed, Holst, N., Lang, A., Lovei,¨ G., Otto, M., 2013. Increased mortality is predicted of Inachis Flora dels Països Catalans. Barcino, Barcelona, Spain. io larvae caused by Bt-maize pollen in European farmland. Ecol. Modell. 250, de la Poza, M., Pons, X., Farinos,´ G.P., Lopez,´ C., Ortego, F., Eizaguirre, M., Castanera,˜ P., 126–133. https://doi.org/10.1016/j.ecolmodel.2012.11.006. Albajes, R., 2005. Impact of farm-scale Bt maize on abundance of predatory IPNI, 2020. International Plant Names Index [WWW Document]. URL https://www.ipni. arthropods in Spain. Crop Prot. 24 (7), 677–684. https://doi.org/10.1016/j. org/ (accessed 1.10.20). cropro.2004.12.003. ISAAA, 2019. ISAAA Brief No 54. Global status of Commercialized biotech/GM Crops Dolezel, M., Bartel, A., Heissenberger, A., 2018. Spatial analysis of the occurrence of 2018, ISAAA Briefs. ISAAA, Ithaca, NY. protected butterflies in six European biogeographic regions as a tool for the JMP Pro®, 2019. Version14.1.0. SAS Institute Inc.,1989-2019 Cary, NC. environmental risk assessment of Bt maize. BioRisk 2018, 31–52. https://doi.org/ Juarez-Escario,´ A., Sole-Senan,´ X.O., Recasens, J., Taberner, A., Conesa, J.A., 2018. Long- 10.3897/biorisk.13.20688. term compositional and functional changes in alien and native weed communities in EC (European Comission), 2001. Directive 2001/18/EC of the European Parliament and annual and perennial irrigated crops. Ann. Appl. Biol. 173, 42–54. https://doi.org/ of the Council of 12 March 2001 on the deliberate release into the environment of 10.1111/aab.12432. genetically modifiedorganisms and repealing Council Directive 90/220/EEC. Off. J. Kjær, C., Damgaard, C., Lauritzen, A.J., 2010. Assessment of effects of Bt-oilseed rape on Eur. Communities L106, 1–38. large white butterfly (Pieris brassicae) in natural habitats. Entomol. Exp. Appl. 134, EC (European Comission), 2018. Directive (EU) 2018/350 of 8 March 2018 amending 304–311. https://doi.org/10.1111/j.1570-7458.2009.00958.x. Directive 2001/18/EC of the European Parliament and of the Council as regards the Kottek, M., Grieser, J., Beck, C., Rudolf, B., Rubel, F., 2006. World Map of environmental risk assessment of genetically modified organisms. Off. J. Eur. Union Koppen¨ Geiger climate classification. Meteorol. Z 15, 259–263. https://doi.org/ L 67, 30–45. 10.1127/0941-2948/2006/0130. EEA (European Environment Agency), 2013. The European Grassland Butterfly Kutlesa, N.J., Caveney, S., 2001. Insecticidal activity of glufosinate through glutamine Indicator: 1990–2011, EEA Technical Reports. Publications Office of the European depletion in a caterpillar. Pest Manage. Sci. 57, 25–32. https://doi.org/10.1002/ Union, Luxembourg. DOI:10.2800/89760. 1526-4998(200101)57:1<25::AID-PS272>3.0.CO;2-I. EFSA GMO Panel (Panel on Genetically ModifiedOrganisms of the European Food Safety Lang, A., 2004. Monitoring the impact of Bt maize on butterflies in the field: estimation Authority), 2011. Guidance on the Post-Market Environmental Monitoring (PMEM) of required sample sizes. Environ. Biosafety Res. 3, 55–66. https://doi.org/10.1051/ of genetically modified plants. EFSA J. 9, 2316. DOI:10.2903/j.efsa.2011.2316. ebr:2003018. EFSA GMO Panel (Panel on Genetically ModifiedOrganisms of the European Food Safety Lang, A., Bühler, C., 2012. Estimation of required sampling effort for monitoring the Authority), 2014. Scientific Opinion on the use of existing environmental possible effects of transgenic crops on butterflies:lessons from long-term monitoring surveillance networks to support the post-market environmental monitoring of schemes in Switzerland. Ecol. Indic. 13, 29–36. https://doi.org/10.1016/j. genetically modified plants. EFSA J. 12, 3883. DOI:10.2903/j.efsa.2014.3883. ecolind.2011.05.004. Eizaguirre, Matilde, Albajes, Ramon, Lopez,´ Carmen, Eras, Jordi, Lumbierres, Belen,´ Lang, A., Otto, M., 2010. A synthesis of laboratory and field studies on the effects of Pons, Xavier, 2006. Six years after the commercial introduction of Bt Maize in Spain: transgenic Bacillus thuringiensis (Bt) maize on non-target Lepidoptera. Entomol. field evaluation, impact and future prospects. Transgenic Res. 15 (1), 1–12. https:// Exp. Appl. 135, 121–134. https://doi.org/10.1111/j.1570-7458.2010.00981.x. doi.org/10.1007/s11248-005-3998-1. Lang, A., Vojtech, E., 2006. The effects of pollen consumption of transgenic Bt maize on Eizaguirre, M., 2012. Cronología, danos˜ y m´etodos de control de las plagas del maíz. the common swallowtail, Papilio machaon L. (Lepidoptera, Papilionidae). Basic Vida Rural 340, 32–35. https://doi.org/ISSN 1133-8938. Appl. Ecol. 7, 296–306. https://doi.org/10.1016/j.baae.2005.10.003. Fahse, Lorenz, Papastefanou, Phillip, Otto, Mathias, 2018. Estimating acute mortality of Lang, A., Dolek, M., Theißen, B., Zapp, A., 2011. Are Adult Crambid Snout Moths Lepidoptera caused by the cultivation of insect-resistant Bt maize – the LepiX model. (Crambinae) and Larval Stages of Lepidoptera Suitable Tools for an environmental Ecol. Model. 371, 50–59. https://doi.org/10.1016/j.ecolmodel.2018.01.006. monitoring of transgenic crops? — Implications of a field test. Insects 2, 400–411. Farre,´ I., Faci, J.-M., 2009. Deficitirrigation in maize for reducing agricultural water use https://doi.org/10.3390/insects2030400. in a Mediterranean environment. Agric. Water Manage. 96 (3), 383–394. https:// Lang, A., Theißen, B., Dolek, M., 2013. Standardised methods for the GMO monitoring of doi.org/10.1016/j.agwat.2008.07.002. butterflies and moths: the whys and hows. BioRisk 8, 15–38. https://doi.org/ Felke, Martin, Langenbruch, Gustav-Adolf, Feiertag, Simon, Kassa, Adane, 2010. Effect of 10.3897/biorisk.8.3244. Bt-176 maize pollen on first instar larvae of the Peacock butterfly (Inachis io) Lang, A., Oehen, B., Ross, J.H., Bieri, K., Steinbrich, A., 2015. Potential exposure of (Lepidoptera; Nymphalidae). Environ. Biosafety Res. 9 (1), 5–12. https://doi.org/ butterfliesin protected habitats by Bt maize cultivation: a case study in Switzerland. 10.1051/ebr/2010006. Biol. Conserv. 192, 369–377. https://doi.org/10.1016/j.biocon.2015.10.006. Felke, M., Lorenz, N., Langenbruch, G.-a.A., 2002. Laboratory studies on the effects of Lang, A., Bühler, C., Dolek, M., Roth, T., Züghart, W., 2016. Estimating sampling pollen from Bt-maize on larvae of some butterfly species. J. Appl. Entomol. 126, efficiency of diurnal Lepidoptera in farmland. J. Insect Conserv. 20, 35–48. https:// 320–325. https://doi.org/10.1046/j.1439-0418.2002.00668.x. doi.org/10.1007/s10841-015-9837-7. García-Barros, E., Munguira, M.L., Stefanescu, C., Vives Moreno, A., 2013. Fauna Iberica:´ Lang, A., Kallhardt, F., Lee, M.S., Loos, J., Molander, M.A., Muntean, I., Pettersson, L.B., Volumen 37 - Lepidoptera: Papilionoidea. CSIC, Madrid, Spain. Rakosy,´ L., Stefanescu, C., Mess´ean, A., 2019. Monitoring environmental effects on García-Ruiz, E., Cobos, G., Sanchez-Ramos,´ I., Pascual, S., Chueca, M., Escorial, M., farmland Lepidoptera: does necessary sampling effort vary between different bio- Santín-Montanya,´ I., Loureiro, ´I., Gonzalez-Nú´ nez,˜ M., 2020. Dynamics of canopy- geographic regions in Europe? Ecol. Indic. 102, 791–800. https://doi.org/10.1016/j. dwelling arthropods under different weed management options, including ecolind.2019.03.035. glyphosate, in conventional and genetically modified insect-resistant maize. Insect Leclerc, M., Walker, E., Messean,´ A., Soubeyrand, S., 2018. Spatial exposure-hazard and Sci. 1744–7917, 12825. https://doi.org/10.1111/1744-7917.12825. landscape models for assessing the impact of GM crops on non-target organisms. Sci. Gathmann, A., Wirooks, L., Hothorn, L. a, Bartsch, D., Schuphan, I., 2006. Impact of Bt Total Environ. 624, 470–479. https://doi.org/10.1016/j.scitotenv.2017.11.329. maize pollen (MON810) on lepidopteran larvae living on accompanying weeds. Mol. Lee, M.S., Albajes, R., 2016. Monitoring carabid indicators could reveal environmental Ecol. 15, 2677–85. DOI:10.1111/j.1365-294X.2006.02962.x. impacts of genetically modified maize. Agric. For. Entomol. 18, 238–249. https:// Gill, J.P.K., Sethi, N., Mohan, A., Datta, S., Girdhar, M., 2018. Glyphosate toxicity for doi.org/10.1111/afe.12156. animals. Environ. Chem. Lett. 16, 401–426. https://doi.org/10.1007/s10311-017- Lee, M.S., Albajes, R., 2013. Butterfliesfor post market environmental monitoring of GM 0689-0. maize in Spain. IOBC/WPRS Bull. 97, 63–72. https://doi.org/ISBN 978-92-9067- Hawes, C., Haughton, A.J., Osborne, J.L., Roy, D.B., Clark, S.J., Perry, J.N., Rothery, P., 276-0. Bohan, D.A., Brooks, D.R., Champion, G.T., Dewar, A.M., Heard, M.S., Woiwod, I.P., Lee, M.S., Comas, J., Stefanescu, C., Albajes, R., 2020. The Catalan butterfly monitoring Daniels, R.E., Young, M.W., Parish, A.M., Scott, R.J., Firbank, L.G., Squire, G.R., scheme has the capacity to detect effects of modifying agricultural practices. 2003. Responses of plants and invertebrate trophic groups to contrasting herbicide Ecosphere 11. https://doi.org/10.1002/ecs2.3004. regimes in the Farm Scale Evaluations of genetically modified herbicide-tolerant

10 M.S. Lee et al. Ecological Indicators 124 (2021) 107380

(dataset) Lee, M.S., (2020a), “Butterflies in maize agroecosystems”, Mendeley Data, V1, Pywell, R.F., Warman, E.A., Sparks, T.H., Greatorex-Davies, J.N., Walker, K.J., Meek, W. doi: 10.17632/t8kjmg8szt.1 https://data.mendeley.com/datasets/t8kjmg8szt/draft? R., Carvell, C., Petit, S., Firbank, L.G., 2004. Assessing habitat quality for butterflies a=bac09fe8-f260-4b38-99ed-e298a8d2c77c. on intensively managed arable farmland. Biol. Conserv. 118, 313–325. https://doi. (dataset) Lee, M.S., (2020b), “Butterfly larval food plant distribution in maize org/10.1016/j.biocon.2003.09.011. agroecosystems”, Mendeley Data, V1, doi: 10.17632/kmh8kbbc4w.1 https://data. Ritchie, S., Hanway, J., Benson, G., 1989. How a corn plant develops. Sci. Technol. 48, mendeley.com/datasets/kmh8kbbc4w/draft?a=958a0ff7-fdaf-49cb-98ee-4be3f1ee 1–25. 51cc. Rougeot, P.C., Viette, P., 1980. Guía de campo de las mariposas nocturnas de Europa y (dataset) Lee, M.S., (2020c), “Butterfly larvae in maize field margins”, Mendeley Data, norte de Africa. Delachaux et Niestle,´ Barcelona, Spain. V2, doi: 10.17632/3wcks9kmzn.2. Schmidt, K., Wilhelm, R., Schmidtke, J., Beißner, L., Monkemeyer,¨ W., Bottinger,¨ P., Lefebvre, M., Espinosa, M., Gomez y Paloma, S., Paracchini, M.L., Piorr, A., Zasada, I., Sweet, J., Schiemann, J., 2008. Farm questionnaires for monitoring genetically 2015. Agricultural landscapes as multi-scale public good and the role of the Common modifiedcrops: A case study using GM maize. Environ. Biosafety Res. DOI:10.1051/ Agricultural Policy. J. Environ. Plan. Manag. https://doi.org/10.1080/ ebr:2008015. 09640568.2014.891975. Schmitz, G., Bartsch, D., Pretscher, P., 2003. Selection of relevant non-target herbivores Madeira, F., di Lascio, A., Carlino, P., Costantini, M.L., Rossi, L., Pons, X., 2014. Stable for monitoring the environmental effects of Bt maize pollen 2, 117–132. DOI: carbon and nitrogen isotope signatures to determine predator dispersal between 10.1051/ebr. alfalfa and maize. Biol. Control. DOI:10.1016/j.biocontrol.2014.06.009. Schmucki, R., Pe’er, G., Roy, D.B., Stefanescu, C., Van Swaay, C.A.M., Oliver, T.H., Madeira, F., Pons, X., 2016. Rubidium marking reveals different patterns of movement in Kuussaari, M., Van Strien, A.J., Ries, L., Settele, J., Musche, M., Carnicer, J., four ground beetle species (Col., Carabidae) between adjacent alfalfa and maize. Schweiger, O., Brereton, T.M., Harpke, A., Heliol¨ a,¨ J., Kühn, E., Julliard, R., 2016. A Agric. For. Entomol. 18, 99–107. https://doi.org/10.1111/afe.12141. regionally informed abundance index for supporting integrative analyses across MAPA, 2020. Agricultural production in Spain-Ministry of Agriculture, Fisheries and butterfly monitoring schemes. J. Appl. Ecol. 53, 501–510. DOI:10.1111/1365- Food [WWW Document]. URL https://www.mapa.gob.es/es/agricultura/temas/ 2664.12561. producciones-agricolas/ (accessed 2.10.20). Schuppener, M., Mühlhause, J., Müller, A.K., Rauschen, S., 2012. Environmental risk Marshall, E.J.P., Baudry, J., Moonen, C., Fevre, E., Thomas, C.F.G., 1996. Factors assessment for the small tortoiseshell Aglais urticae and a stacked Bt-maize with affecting floral diversity in European field margin networks. Spat. Dyn. Biodivers. combined resistances against Lepidoptera and Chrysomelidae in central European Towar. an Underst. Spat. patterns Process. landscape.Proceedings Fifth Annu. IALE agrarian landscapes. Mol. Ecol. 21, 4646–4662. https://doi.org/10.1111/j.1365- Conf. Stirling, UK, 9-12 Sept. 1996. 294X.2012.05716.x. Messeguer, J., Penas,˜ G., Ballester, J., Bas, M., Serra, J., Salvia, J., Palaudelmas,` M., Sears, M.K., Hellmich, R.L., Stanley-Horn, D.E., Oberhauser, K.S., Pleasants, J.M., Mele,´ E., 2006. Pollen-mediated gene flowin maize in real situations of coexistence. Mattila, H.R., Siegfried, B.D., Dively, G.P., 2001. Impact of Bt corn pollen on Plant Biotechnol. J. 4, 633–645. https://doi.org/10.1111/j.1467-7652.2006.00207. monarch butterfly populations: a risk assessment. Proc. Natl. Acad. Sci. 98, x. 11937–11942. https://doi.org/10.1073/pnas.211329998. Naranjo, S., 2009. Impacts of Bt crops on non-target invertebrates and insecticide use Stanley-Horn, D.E., Dively, G.P., Hellmich, R.L., Mattila, H.R., Sears, M.K., Rose, R., patterns. CAB Rev. Perspect. Agric. Vet. Sci. Nutr Nat. Resour. 4, 011. https://doi. Jesse, L.C., Losey, J.E., Obrycki, J.J., Lewis, L., 2001. Assessing the impact of org/10.1079/PAVSNNR20094011. Cry1Ab-expressing corn pollen on monarch butterfly larvae in field studies. Proc. Ortego, F., Pons, X., Albajes, R., Castanera,˜ P., 2009. European commercial genetically Natl. Acad. Sci. U. S. A. 98, 11931–11936. https://doi.org/10.1073/ modified plantings and field trials. In: Ferry, N., Gatehouse, A.M.R. (Eds.), pnas.211277798. Environmental Impact of Genetically ModifiedCrops. CABI Publishing, pp. 327–343. Stefanescu, C., Carnicer, J., Penuelas,˜ J., 2011a. Determinants of species richness in https://doi.org/10.1079/9781845934095.0327. generalist and specialist Mediterranean butterflies:the negative synergistic forces of Perry, J.N., Rothery, P., Clark, S.J., Heard, M.S., Hawes, C., 2003. Design, analysis and climate and habitat change. Ecography (Cop.) 34, 353–363. https://doi.org/ statistical power of the Farm-Scale Evaluations of genetically modified herbicide- 10.1111/j.1600-0587.2010.06264.x. tolerant crops. J. Appl. Ecol. 40, 17–31. https://doi.org/10.1046/j.1365- Stefanescu, C., Torre, I., Jubany, J., Paramo,´ F., 2011b. Recent trends in butterfly 2664.2003.00786.x. populations from north-east Spain and Andorra in the light of habitat and climate Perry, J.N., Ter Braak, C.J.F., Dixon, P.M., Duan, J.J., Hails, R.S., Huesken, A., change. J. Insect Conserv. 15, 83–93. https://doi.org/10.1007/s10841-010-9325-z. Lavielle, M., Marvier, M., Scardi, M., Schmidt, K., Tothmeresz, B., Schaarschmidt, F., Sz´ekacs,´ A., Darvas, B., 2018. Re-registration Challenges of Glyphosate in the European van der Voet, H., 2009. Statistical aspects of environmental risk assessment of GM Union. Front. Environ. Sci. 6 https://doi.org/10.3389/fenvs.2018.00078. plants for effects on non-target organisms. Environ. Biosafety Res. 8, 65–78. https:// Thomas, J.A., 2005. Monitoring change in the abundance and distribution of insects doi.org/10.1051/ebr/2009009. using butterflies and other indicator groups. Philos. Trans. R. Soc. B Biol. Sci. 360, Perry, J.N., Devos, Y., Arpaia, S., Bartsch, D., Gathmann, A., Hails, R.S., Kiss, J., 339–357. https://doi.org/10.1098/rstb.2004.1585. Lheureux, K., Manachini, B., Mestdagh, S., Neemann, G., Ortego, F., Schiemann, J., Tolman, T., Lewington, R., 2011. Mariposas de Espana˜ y Europa, 2nd ed. Lynx Edicions, Sweet, J.B., 2010. A mathematical model of exposure of non-target Lepidoptera to Barcelona, Spain. ˇ Bt-maize pollen expressing Cry1Ab within Europe. Proc. Biol. Sci. 277, 1417–1425. Van Swaay, C., Cuttelod, A., Collins, S., Maes, D., Munguira, M.L., Saˇsi´c, M., Settele, J., https://doi.org/10.1098/rspb.2009.2091. Verovnik, R., Verstrael, T., Warren, M., Wiemers, M., Wynhoff, I., 1999. European Perry, J.N., Devos, Y., Arpaia, S., Bartsch, D., Ehlert, C., Gathmann, A., Hails, R.S., Red List of Butterflies, Nature. Luxembourg: Publications Office of the European Hendriksen, N.B., Kiss, J., Messean,´ A., Mestdagh, S., Neemann, G., Nuti, M., Union, Luxembourg. https://doi.org/doi:10.2779/83897. Sweet, J.B., Tebbe, C.C., 2012. Estimating the effects of Cry1F Bt-maize pollen on Van Wyk, A., Van den Berg, J., Van Hamburg, H., 2007. Selection of non-target non-target Lepidoptera using a mathematical model of exposure. J. Appl. Ecol. 49, Lepidoptera species for ecological risk assessment of Bt maize in South Africa. 29–37. https://doi.org/10.1111/j.1365-2664.2011.02083.x. African Entomol. 15, 356–366. https://doi.org/10.4001/1021-3589-15.2.356. Pleasants, J.M., Oberhauser, K.S., 2012. Milkweed loss in agricultural fields because of Verdú, J.R., Numa, C., Galante, E., 2011. Atlas y Libro Rojo de Los Invertebrados de herbicide use: effect on the monarch butterfly population. Insect Conserv. Divers. 6, Espana.˜ Direccion´ General de Medio Natural y Política Forestal, Ministerio de Medio 135–144. https://doi.org/10.1111/j.1752-4598.2012.00196.x. Ambiente, Medio rural y Marino, Madrid, Spain. Pleasants, J.M., Hellmich, R.L., Dively, G.P., Sears, M.K., Stanley-Horn, D.E., Mattila, H. Wallis de Vries, M.F., van Deijk, J., van Alebeek, F., 2017. The importance of maize and R., Foster, J.E., Clark, P., Jones, G.D., 2001. Corn pollen deposition on milkweeds in oilseed rape field margins for Lepidoptera. Report VS2017.005 / CGM 2017-03, De and near cornfields.Proc. Natl. Acad. Sci. U. S. A. 98, 11919–11924. https://doi.org/ Vlinderstichting / Dutch Butterfly Conservation, Wageningen. 10.1073/pnas.211287498. Wolt, J.D., Conlan, C.A., Majima, K., 2005. An ecological risk assessment of Cry1F maize Pollard, E., 1977. A method for assessing changes in the abundance of butterflies. Biol. pollen impact to pale grass blue butterfly. Env. Biosaf. Res 4, 243–251. https://doi. Conserv. 12, 115–134. https://doi.org/10.1016/0006-3207(77)90065-9. org/10.1051/ebr:2006005 [doi]\rebr0529 [pii]. Pollard, E., Yates, T.J., 1993. Monitoring Butterflies for Ecology and Conservation: the Zangerl, A.R., McKenna, D., Wraight, C.L., Carroll, M., Ficarello, P., Warner, R., British Butterfly Monitoring Scheme. Chapman & Hall, London, UK. Berenbaum, M.R., 2001. Effects of exposure to event 176 Bacillus thuringiensis corn Pujol i Palol, M., 2017. Les Plantes Cultivades, 2. Farratges, I. Produccio´ i utilitzacio´ a pollen on monarch and black swallowtail caterpillars under field conditions. Proc. Catalunya. Romanya` Valls, Capellades, Spain. Natl. Acad. Sci. U. S. A. 98, 11908–11912. https://doi.org/10.1073/ pnas.171315698.

11