In 2008, field surveys were carried out in three different regions in Germany: South Upper Rhine (Südlicher Oberrhein), Upper Franconia (Oberfranken) and Jülich Boerde (Jülicher Börde). The three sampling locations of the Upper Rhine were near the villages Rheinweiler, Blansingen and Efringen-Kirchen (about 10–20 km north of Basel, Switzerland), the two locations of Upper Franconia were Bindlacher Berg and Benk (north of Bayreuth), and the two locations in the Jülich Boerde were near Koslar and Stetternich (between Aachen and Cologne). All sampling locations were within arable land and were situated in field margins along maize fields in the regions South Upper Rhine and Upper Franconia, and along winter wheat and sugar beet fields in the Jülich Boerde.

Along crop fields, transect routes were established, walked in regular intervals and all observed adult butterfly and Crambid snout moth specimens recorded, i.e., species of Papilionoidea, Hesperioidea as well as Crambinae. The monitoring approach followed a strictly standardized design according to the VDI guidelines: defined favorable weather conditions were a precondition, monitoring was to be conducted between 10 am and 5 pm, walking pace and spatial observation radius was fixed, transects were divided into 50m sections and additional parameters were recorded such as land use and flowering aspect. For the transect counts, the field recording sheets given by the VDI guidelines were used (for more details see [7]). The transect counts were carried out between 9 June and 9 September 2008. Transect lengths as well as recording dates and intervals differed between the study regions (Table 1).

The larvae of Macrolepidoptera were recorded by two different methods: visual searching and beating samples.

A directed visual search for recording lepidopteran larvae has been described by Hermann [20]. In the VDI guidelines this method was adapted for standardized monitoring purposes [7]. The general approach is a plant-oriented search, i.e., the search is directed to the known larval host plants. A crucial point of the directed search is to check only suitable host plants, i.e., suitable for the larvae due to the microclimatic condition as well as nutritional situation and growth state of the plant. The VDI standardization included a defined number of plants to be searched, defined searching times per host plant (populations), and adequate weather conditions. For the visual searches, the field recording sheets given by the VDI guidelines were used (for more details see [7]). Visual searches were carried out between 9th June and 9th September, and differed among regions due to differing climatic conditions and species occurrence (Table 2).

Beating samples is a widely applied approach to collect arthropods from bushes and trees [14]. The standardized use for sampling lepidopteran larvae occurring on bushes and trees is described in the VDI guidelines [7]. The general approach is to beat branches with a stick and to collect and count the specimens caught on a collecting sheet placed below. The VDI standardization of the method included favorable and dry weather conditions during sampling events, beating technique (two hits per branch), and sampling of 100 branches (if possible) per location (e.g., a hedgerow, or a row of trees). For the beating samples the field recording sheets given by the VDI guidelines were used (for more details see [7]). Beating samples were carried out between 9th June and 11th August, and differed among regions due to differing climatic conditions and species occurrences (Table 2).

Overall, 30 lepidopteran species were recorded by the transect counts: 25 butterfly species and five species of Crambinae (Table 3, Appendix 1). Most individuals and species were recorded in the South Upper-Rhine, reflecting the higher number of sites, the longer transects and the general higher biodiversity in this region. Crambid Snout Moths were recorded in four out of the seven sampling locations, i.e., in 57% of the field margins. On average, 1.6 species and 16.7 individuals of Crambid Snout Moths were recorded per field margin, the proportion of the total Lepidoptera count being 10% for species number and 16% for individual abundance. Especially in the species-poorer and intensively managed regions Upper Franconia and Jülich Boerde, the fraction of Crambid Snouth Moths was considerably high (Table 3). The most abundant species of the Crambinae was Crambus perlella (Table 4), occurring in all study regions and making up a considerable part of the overall Lepidoptera count (compare Table 4 and Appendix 1). In the Upper Rhine, occurrence of Lepidoptera was analyzed in relation to the flowering aspects, i.e., in relation to the number of flowering plants present in the respective field margins. In general, butterflies were more abundant when flowers were present in field margins (Figure 1). Crambid Snout Moths showed no positive association with the number of flowering plants, in contrast, they were more abundant on field margins without flowers (Figure 1).

Overall, visual searching of Lepidoptera larvae detected the larvae of nine species. On Urtica dioica host plants, three species were detected: Inachis io, Vanessa atalanta and Pleurotypa ruralis (Table 5). The larvae of I. io and P. ruralis were found in all three regions. The number of larvae of I. io in communal nests ranged from 40–160 larvae per web. The searching times for webs of I. io were 18.6 minutes/web (South Upper-Rhine), 8.8 minutes/web (Upper Franconia) and 26.4 minutes/web (Jülich Boerde).

Of plants other than U. dioica, 17 potential host plants were searched (Table 5), which yielded 27 larvae of six species: Pieris rapae (Pieridae), Emmelia trabealis (Noctuidae), Tyta luctuosa (Noctuidae), Tyra jacobaeae (Arctiidae), Phragmatobia fuliginosa (Arctiidae) and an unidentified moth larva on Sedum telephium. The searching times per larva were 73.5 minutes/larva (South Upper-Rhine), 109.4 minutes/larva (Upper Franconia) and 3.25 minutes/larva (Jülich Boerde). The low searching time in the Jülich Boerde was caused by a single recording event of 20 aggregated larvae of T. jacobaeae at one occasion.

During searching of larvae, eggs of Lepidoptera were also detected: 18 eggs of Cyaniris semiargus (Lycaenidae) on Trifolium pratense in the South Upper-Rhine, and 128 eggs of Pieris rapae/brassicae (Pieridae) on Brassica napus in Upper Franconia.

In addition, it was noted that larvae of the Ermine Moths (Yponomeutidae) were occurring occasionally, feeding on blackthorn bushes (Prunus spinosa). Most larvae had already pupated due to the late sampling date, but their communal nests could still be counted. In the South Upper-Rhine, 91 yponomeutid webs were recorded by searching an area of 67m² densely covered with blackthorns (searching time of 15 minutes), while in Upper Franconia 17 yponomeutid webs were counted by checking 50 single blackthorn bushes (searching time of 30 minutes).

Overall, the beating samples yielded 30 larvae (Table 6) which belonged to nine species: Eilema sp. (Arctiidae), Notodonta ziczac (Notodontidae), Clostera pigra (Notodontidae), Craniophora ligustri cf. (Noctuidae), Catocala sp. (Noctuidae), Lomaspilis marginata (Geometridae), Chloryclysta siterata (Geometridae), Cabera exanthemata (Geometridae) and Yponomeuta sp. (Yponomeutidae). Of the overall catch, 33% of the individuals could be attributed to the species or genus, respectively, while 66% of the individuals could not be identified. Handling time of one double-hit was roughly one minute including sorting and collecting the catch (but excluding species identification), and collection time per larva was on average 20 minutes.

Crambid Snout Moths in partim proved to be a suitable extension to the transect counting of butterflies, requiring not much additional effort. When occurring, they were abundant and made up a substantial part of the individuals. Also, they were widespread, occurring in all three regions studied, in particular C. perlella. By including Crambid Snout Moths in GMO monitoring, the data set would be increased, thus improving the power of subsequent statistical analyses. This may especially hold true for very short transects along small fields, where recorded Lepidoptera abundance in total will be lower [21]. Crambid Snout Moths are not very mobile and effects on them could be more attributed to the site. Adults do not feed on flowers, and the larvae feed on grass [16,17], hence the moths can be still abundant in grassy habitats or agricultural field margins without flowers (this study), where butterfly numbers typically decrease [11]. However, Crambid Snout Moths were not detected in each field margin and their species richness was relatively low. Although this might be due to the limited scope of the study, it may also indicate that Crambid Snout Moths are only suitable as a supplementary monitoring module, but not as a stand-alone tool.

There are some minor handicaps involved with the monitoring of Crambid Snout Moths. The moths must be captured with a net for species identification. But in contrast to the Skippers (Hesperiidae), which must be netted too, catching Crambid Snout Moths is much easier. In the field, the moths themselves can easily be spotted and identified as Crambid Snout Moths by their typical head downwards posture in the vegetation (Kolbeck, pers. comm., personal observation). There exists no field identification guide for the Crambinae. Identification must be done by consulting several books, pictures from the internet, or newsgroups of experts. However, species number in arable land is low, and experienced butterfly experts should be able to cope with the respective species set of Crambinae after some practice. Adult Crambid Moths will not necessarily indicate a reduction of flowering plants like butterflies, and larvae may possibly not indicate aerial pollutants as they can live quite hidden in the vegetation, e.g., in stems of grasses [17].

It appears possible to include the recording of Lepidoptera larvae for GMO monitoring purposes. However, the results and efficiencies differ depending on the applied methods. Nettles, Urtica plants, are quite abundant in arable land and field margins [22], consequently larvae of the Peacock Butterfly (I. io) were easy to find and to monitor. However, it has to be noted that Peacock larvae of earlier stages spin communal nests and occur aggregated in batches, which can also influence the distribution of older larvae even after separation of the caterpillars (personal observation). The occurrence of further larvae of common lepidopteran species on nettles adds to the usefulness of monitoring a group of “nettle Lepidoptera”, i.e., V. atalanta, Aglais urticae, Araschnia levana, Polygonia c-album, Hypaena proboscidalis, and P. ruralis. It has to be kept in mind that the different species differ in detectability and required monitoring effort, e.g., while communal nests of I. io and A. urticae are quite conspicuous, the single larvae of P. ruralis, which are hidden in nettle leaf bags, are more difficult to spot. Amazingly, larvae of the Small Tortoiseshell (A. urticae) could not be recorded in this study, although it is a common species in agricultural land [22]. However, in 2008, numbers of Small Tortoiseshells were generally low in Germany [23]. Further, a monitoring program more extensive than the present study will yield a much larger data set and more robust results. Very similar to the Peacock Butterfly, the communal nests of larvae of Ermine Moths (Yponomeutidae) were easy to detect in bushes and hedgerows along fields.

Generally, visual search of host plants other than Urtica was laborious. Searching time was higher and recorded number of larvae lower in comparison with the other studied approaches (i.e., visual search on nettles and blackthorn, and beating samples). In two regions, searching time for one larva was longer than one hour, while the low searching time in the Jülich Boerde was unusual and due to a single recording event of aggregated larvae. Therefore, visual search of host plants appears less suitable for a general monitoring, and may be used in a more case-specific and supplementary way, e.g., if certain species (groups) are to be monitored for protection purposes or if some species are known to occur abundantly. By all means, species-specific knowledge on larval biology and occurrence is paramount, or must be acquired beforehand.

Beating samples performed intermediate in terms of invested sampling effort in relation to recorded results. However, sampling success of this study suffered from the fact that some common and abundant species could not be recorded due to the later season sampling date (e.g., Operophtera brunata, Orthosia gothica, Orthosia incerta, Orthosia cerasi, Cerastris rubicosa). The method is convenient to apply, and especially suitable to sample bushes, trees and hedgerows where a visual search is often impractical. Beating samples collected many species of Geometrid Moths (Geometridae), which can sometimes not be identified to a species level. These and other unidentifiable specimens must be reared to adulthood for identification, which causes additional effort, and is sometimes not successful as the larvae may die before reaching adulthood. On the other hand, sampling and identification efficiency will improve if monitoring persons are instructed and gain experience (cf. [7]).

As a general rule, the monitoring of Lepidoptera larvae must be planned and analyzed much more on a regional level. The occurrence of larvae is strongly affected by the available and suitable host plants, habitat fragmentation, regional weather, micro-climatic conditions or management practice, to name just a few relevant factors. These factors are unlikely to be the same over a wider geographic area, and/or are difficult to standardize and to harmonize among study sites.