Since Apollo is protected by national and international law [4,45,46], necessary permissions for rearing and breeding it were obtained from the Polish Ministry of Environmental Protection and State Nature Conservator. All P. apollo ssp. frankenbergeri feeding preferences tests were carried out on individuals from the semi-natural open-field breeding colony set up for a recovery plan. The colony shares genes pool with the wild population due to the bi-directional flow of genes. Emerging first-instar apollo larvae, after over-wintering in natural conditions, were placed in large glass insectaries (1.5×0.8m) covered by soil layer with various growing plants, including Sedum telephium and Sempervivum sp. Such conditions were suitable and provided young larvae with the appropriate food amounts. Starting from the third instar, freshly collected sprouts of S. telephium were provided as food. Larvae were exposed to natural weather conditions. Only in case of heavy rains, the insectaries were covered with glass lids to avoid larvae drowning. In these conditions, larvae developed even slightly faster than in their natural biotopes, probably due to more stabilised thermal conditions. For feeding preferences tests, all commercially available cultivated Sedum species were purchased. These were the following: S. floriferum, S. ewersii, S. kamtschaticum, S. spectabile (species with flattened leaves); S. album, S. reflexum, S. rupestre, S. rosea and S. spurium (species with terate leaves). Sedum taxonomy is still problematic, so the names given by the purveyor were confirmed according to Flora Europea and Royal Botanic Garden Edinburgh (RBGE) database (Table 1) [47,48]. The former source was also used to unify nomenclature of ‘telephiophagous’ apollo subspecies food-plant, since earlier authors often used various Latin names (Table 2) [48]. All purchased plants grew in natural weather conditions, because previous studies had shown that the plants with artificially accelerated growth (glasshouse cultivation) were unsuitable as the food-source for the Apollo larvae [3,34]. Potted plants were transported to the Apollo colony in the Pieniny Mts and placed in a separate insectary for acclimation. The main experiment was preceded by a preliminary test, in which samples of all the plants were offered to the group of fourth instar (L4) Apollo larvae. This preference test allowed selection of edible plant species for further experiment, carried out on the last-instar (L5) larvae. To assess the effects of selected plants on larval vitality and further development to adult stage, late fourth-instar larvae similar in size and weight were selected from the colony. At the end of April, three experimental groups (I–III; 15 individuals in each) were set up on different menu options. Larvae of group I were fed on Sedum kamtschaticum var. floriferum and S. rupestre, larvae of group II on S. album and S. ewersii, and the larvae of group III (being the reference group) on S. telephium ssp. maximum (the same as that provided to the stock colony insects) only. The larvae from the onset of the last instar were provided every day with fresh Sedum leaves and the remnants were removed. Fresh weight of larvae in each experimental group was recorded until the insects began building the cocoon shell. The measurements as well as observations of development were carried out in 1–3 day intervals, depending on the weather conditions, except for one rainy episode, when larvae almost ceased feeding for 4 days and were almost inactive. To minimise the stress caused by the manipulation, the larvae were quickly weighed without anaesthesia. The dimensions of stretched forewings were measured in the newly emerged adults. Fecundity was estimated by collecting and counting the eggs laid by survived females. Obtained results were analysed by ANOVA/MANOVA with STATISTICA 5.0^® package for PC. Post hoc T Tukey test was applied to estimate significant differences among experimental groups (p<0.05).

The preliminary test confirmed that food plants spectrum of P. apollo larvae inhabiting the Pieniny Mts is narrow. Sedum kamtschaticum, S. spectabile, S. rupestre, S. spurium and Rhodiola rosea were even not bitten, although caterpillars actively searched for food. The main experiment revealed that successful development of the last-instar larvae until pupation depended strongly on the plants available as food. Only larvae fed on S. telephium ssp. maximum all developed into pupa. Average duration time of the L5 instar was also the shortest (18 days) in this group (Figs. 2 and 5). Feeding on other Sedum species increased L5 life span by about two weeks. Moreover, a high mortality was stated – 60% in group I and 87% in group II (Figs. 3 and 4). It should be noted that larvae fed exclusively on S. kamtschaticum var. floriferum and S. rupestre, almost did not develop. Hence, pupation of the larvae in this group took place only due to including, after 12 days of experiment, S. telephium ssp. maximum, as an additional food source (Fig. 2).

The fastest growth of the larvae was observed in group fed on S. telephium. After nine days, L5 larvae doubled their initial fresh body weight (0.391 g), and during following days they reached 1.458 g on average (Fig. 2). The maximum individual fresh body weight was 1.925 g. Larvae fed S. album and S. ewersii needed almost three weeks to double their initial fresh body weight. Despite further slow growth only two heaviest individuals in the group pupated (Figs. 2 and 4). In group I, young L5 larvae tasted the offered plants but wandered all over the insectary in search for other food. Before S. telephium was added, body mass increased only by 27%. Despite further significant growth on S. telephium 50% mortality had been observed during pupation, mainly among larvae with low body weights (Figs. 2 and 3). One day before they started to produce cocoon, larvae from group I and II had also significantly lower fresh body weights compared with those of group III (Fig. 6). Decreased consumption of food in the last larval instar affected consecutive developmental stages. One of effect was increased pupal mortality – 55.6% in group I and 50.0% in group II, compared with 13.0% in the control group. Adults that emerged from pupae in group I and II were also smaller than the individuals fed on S. telephium leaves. The latter had significantly longer and broader forewings. The smallest wings occurred in adults from the group fed on S. album and S. ewersii (Fig. 7). Unsuitable food offered to larvae during the last instar resulted also in small number of laid eggs (Table 3).

Three plants species among those offered to Apollo caterpillars S. telephium, S. spectabile and S. ewersii represented the Hylotelephium group. Closely-related S. telephium ssp. fabaria, recorded as a food plant of P. apollo ssp. niesiolowskii, inhabiting the northern part of the High Tatra range [25], rarely occurs in the Pieniny range and in its vicinities. As it was also unavailable commercially, it was not included in the study. Among European Sedum representatives there were S. rupestre, S. album and S. spurium, of which S. album has widest distribution range in Europe, grows at very contrasting altitudes and in a great variety of soils [49]. Sedum kamtschaticum and its variety floriferum, represented far eastern, and Rhodiola rosea, inhabiting the European mountain ranges are circumpolar stonecrops species. Apollo larvae inhabiting the Pieniny Mts grow and develop successfully only when they have sufficient amounts of S. telephium ssp. maximum. Slower and prolonged larval development on S. album and its almost complete inhibition on S. kamtschaticum var. floriferum and S. rupestre, led to high larval and pupal mortality and low fecundity of the adults. These findings as well as the rejection of the eastern Asiatic S. spectabile point out that this Apollo subspecies belongs definitely to ‘telephiophagous’ group, and has strict food preferences. Deschamps-Cottin and co-workers [11] discovered that Apollo strains, originating from various regions, developed into healthy adults both on S. album and on S. telephium, although the individuals fed on S. telephium were generally larger. Moreover, they have shown that fourth- and fifth-instar apollo larvae collected at various localities in France and Sweden attacked sprouts of almost all (7) offered Sedum, including S. altissimum, and S. ochroleucum, as well as Sempervivum arachnoideum and S. montanum. It is noteworthy that neither S. altissimum nor S. ochroleucum was previously cited among food-plants of P. apollo. Vegetative forms of these European species are difficult to distinguish between each other and from S. rupestre, used in the present study [49]. These findings led French authors to conclude that Apollo larvae are oligophagous, depending on the plants available in the feeding place. However, they also observed that larvae fed on S. ochroleucum developed much longer and slower and finally all individuals died [11]. This coincides with our findings on S. rupestre, which, together with S. kamtschaticum var. floriferum appeared unsuitable food source for the Apollo larvae. In the French study [11], Apollo strains from different origin showed various preferences toward Sedum and Sempervivum plants. The broadest spectrum had strains from the Gotland Island and the Southern Alps near Briançon, while strains from the French Massif Central were the most selective [11]. Apollo larvae from the Pieniny Mts appear to be even more specific. A question that remains is whether they could use S. telephium ssp. fabaria as a food source, but certainly S. album or S. ewersii are much less suitable than S. telephium. Our results give further evidence that P. apollo in Europe falls into ‘telephiophagous’ and ‘albophagous’ trophic groups. This division probably reflects two main routes of its westward migration from Central Asia in the Tertiary and subsequent changes in distribution following glacial and postglacial periods in Europe [6,17,25], with S. telephium being its prime food-plant. Populations that colonised mountain ranges in western Europe had to change its feeding into the available ‘lower’-quality food, and became ‘albophagous’. This explains why they still develop better and grow larger on S. telephium [11,34], while ‘telephiophagous’ forms cope much worse on S. album [17]. This raises a question about the reason of such ‘trophic asymmetry’ among these groups. It has been suggested that the larval performance may depend on the composition of secondary metabolites in their host plant [11,34]. Phytochemical data point out that S. album and S. telephium contain different amounts and composition of phenolic and flavonoid compounds [50–53]. Effect of a specific set of these allelochemicals on Apollo larvae performance and development remains so far unexplained. It may be concluded, that albophagous strains or subspecies of P. apollo from mountain ranges in Southern and Western Europe remain oligophagous, but telephiophagous forms from central Europe should be considered as monophages.