North vs. South: Contrasting patterns in the phenotypic plasticity of the firebug Pyrrhocoris apterus (Hemiptera: Pyrrhocoridae) at the latitudinal extremes of its distribution range

In widely distributed insects, some life-history traits are conserved across the whole distribution range and are considered species-specifi c while other such traits differ geographically. This interplay of geographic variation and phenotypic conservatism is poorly understood even in relatively well-studied model species. Furthermore, a careful study may reveal that conventionally stable traits, such as the lower temperature threshold for development and the sum of degree-days, are both geographically variable and environmentally plastic. We studied how photoperiodic conditions and temperature jointly affect immature development, adult body size and wing polymorphism in two populations of the fi rebug from the opposite latitudinal margins of this species’ range. All the three traits rarely clearly differ under short-day and long-day conditions or between north and south. Instead, we fi nd prevalent temperature-by-photoperiod and temperature-by-origin interactions, which emphasize that it is not only the absolute values of these traits but the degree of their temperature-dependence, or thermal plasticity, that varies in time (in response to seasonal changes in day length) and in space (along latitudinal gradients). These results indicate that caution should be exercised when extrapolating any life-history traits in P. apterus beyond the season when and the location where these were measured. In particular, the use of a constant lower temperature threshold coupled with a constant sum of degree-days is likely to oversimplify the diversity of current and projected phenological patterns in this species. * This paper was contributed to a virtual special issue in memory of Ivo Hodek, a long-time editor of the European Journal of Entomology, who died on June 11, 2021, shortly after his ninetieth birthday. INTRODUCTION Many Eastern European insects are widely encountered from the boreal-forest zone to forest-steppes and even penetrate farther north and south. The vastness of their distribution ranges is at least partially explained by the relatively fl at terrain in this area, which, together with a moderate Atlantic infl uence, has resulted in a gentle climatic gradient that spans a vast stretch of land from the Mediterranean to Fennoscandia. These widespread insect species have long been used as models for studying local adaptation and continuous latitudinal clines (Maslennikova & Mustafaeva, 1971; Vinogradova, 1975; Druzhelyubova, 1976; Imasheva et al., 1994; Elmes et al., 1999). Even after decades of dedicated research, adaptive differences observed across geographic space remain a recurrent theme in ecology because these differences ultimately shed light on the selective pressures that have acted over the evolutionary time. One such common European insect is the fi rebug Pyrrhocoris apterus (Linnaeus, 1758). This is a Palaearctic Eur. J. Entomol. 119: 454–465, 2022 doi: 10.14411/eje.2022.048


INTRODUCTION
Many Eastern European insects are widely encountered from the boreal-forest zone to forest-steppes and even penetrate farther north and south. The vastness of their distribution ranges is at least partially explained by the relatively fl at terrain in this area, which, together with a moderate Atlantic infl uence, has resulted in a gentle climatic gradient that spans a vast stretch of land from the Mediterranean to Fennoscandia. These widespread insect species have long been used as models for studying local adaptation and continuous latitudinal clines (Maslennikova & Mustafaeva, 1971;Vinogradova, 1975;Druzhelyubova, 1976;Imasheva et al., 1994;Elmes et al., 1999). Even after decades of dedicated research, adaptive differences observed across geographic space remain a recurrent theme in ecology because these differences ultimately shed light on the selective pressures that have acted over the evolutionary time.
One such common European insect is the fi rebug Pyrrhocoris apterus (Linnaeus, 1758). This is a Palaearctic which this incidence is highest decreases: for an Israeli population, up to 35% of fi rebugs reared under the shortday photoperiod of 12L : 12D are macropterous (Socha & Šula, 1996;Socha, 2001). The effect of thermal conditions on wing polymorphism in a southern population of P. apterus remains to be tested as the cited experiments were only carried out at 26°C.
While the qualitative photoperiodic responses in P. apterus (i.e., induction and termination of diapause and percentage of macropters) are studied in great detail, much less attention has been paid to the parallel quantitative responses, namely, the effects of photoperiod on immature developmental rate and adult body mass at eclosion (but see Honěk, 1983Honěk, , 1987Saunders, 1983). In a previous study , the simultaneous effects of temperature and photoperiod on immature development in four P. apterus populations from Central Russia, i.e., from the interior of its geographic distribution, are reported. This study revealed a strong photoperiod-by-temperature interaction whereby short-day conditions accelerated nymphal development at low temperatures and long-day conditions at high temperatures.
This communication presents the results of eco-physiological laboratory experiments with two more populations of the fi rebug that were sampled at the opposite latitudinal ends of this species' range three thousand kilometres from each other. We measured immature development time, adult body mass at emergence and the frequency of the wing-morphs under various combinations of constant temperature and constant photoperiod. This multifactorial design allowed the comparison not only of the mean trait values between the two populations but also the plasticity of these traits in response to temperature and photoperiod.

Sources of material and maintenance of parental individuals
The experiments were carried out on two geographically distant populations of P. apterus, one from the village of Podlipye (58°27´N, 27°50´E, Pskov Oblast, the extreme northwest of the European part of Russia) and the other from the city of Tel Aviv (32°4´N, 34°50´E, Israel). The former locality has a temperate humid continental, or hemi boreal, climate with warm summers and freezing cold winters. The latter locality has a hot-summer Mediterranean climate. These two populations are hereinafter referred to as northern and southern, respectively.
Adults of the northern population were collected by hand from leaf litter under linden trees on May 13, transported to St. Petersburg the same day and the experiment started immediately. Groups of bugs were maintained in the laboratory in half-litre plastic containers with a paper sheet on the bottom and pieces of accordion-folded paper as shelter. The bugs were supplied with food (fruit of linden Tilia platyphyllos Scop., which were split to facilitate access to the seed) and water (Eppendorf tubes fi lled with water and plugged with cotton wool), which were replenished as needed. The containers were kept in an environmental chamber at a temperature of 25°C under a 20L : 4D photoperiod.
The southern population resumes activity earlier in the season and, accordingly, the experiment was also carried out earlier in order to ensure that the parental individuals of both populations were in more or less equivalent physiological states. Adults and litter, under bark, and in crevices of buildings. Copulation and oviposition begin early in spring. Nymphs pass through fi ve instars. During the latter half of their nymphal life, fi rebugs become sensitive to photoperiodic cues that control reproductive activity in the adult (Hodek, 1971;Volkovich & Goryshin, 1979). Such sensitivity is also retained in the adult fi rebug (Hodkova, 2015). Short-day conditions induce reproductive arrest, or diapause, in the adult stage whereas long-day conditions promote oviposition (Hodek, 1968(Hodek, , 1971Volkovich & Goryshin, 1979;Saunders, 1983). The critical daylength for diapause induction increases northward (Saunders, 1983;Socha et al., 2001), which is typical of temperate insects in the Northern Hemisphere and ensures a timely onset of dormancy before winter (Danilevskii, 1965). This qualitative photoperiodic response depends on the thermal conditions such that chronic or periodic exposure to low temperatures increases the critical photoperiod and, effectively, bolsters the tendency to diapause (Numata et al., 1993). Reproductive arrest in P. apterus is thus facultative and the species is potentially multivoltine (Socha, 1993). In the central parts of the fi rebug's range, such as the South Bohemian Region of the Czech Republic (Socha & Šula, 1992;Ko štál & Šimek, 2000) and Belgorod and Ryazan regions of Russia (Saulich & Musolin, 1996;Orlova et al., 2009), females from early-season clutches commence reproduction in midsummer and so at least a partial second generation is possible, especially in warmer years and at better-insolated sites. The current rise in global temperatures will likely increase the opportunity for bivoltinism in P. apterus populations (Honek et al., 2020). We are not aware of any formal phenological studies on the fi rebug in the south of its range but extensive GBIF observations of this species (GBIF Secretariat, 2021) show that, e.g., in Israel, nymphs have been photographed in the fi eld in every month from April through to December, indicating a multivoltine cycle with winter diapause.
Despite the species name, the degree of wing development in P. apterus is variable and depends on both genetic and environmental factors (Socha, 1993). The two most common morphs are macropters with both pairs of wings fully developed, albeit non-functional (Socha & Zemek, 2000), and brachypters with reduced forewing membranes and rudimentary hindwings (Socha, 1993). In the fi rebug populations from the Czech Republic, the highest proportion of the macropterous morph, up to 11-14%, can be obtained by rearing the nymphs under long-day conditions (> 16 h of light per day, or > 16L : 8D) at moderately high temperatures (24-27°C), as opposed to short-day photoperiods and cooler temperatures, which promote brachyptery (Honě k, 1995). From an eco-physiological standpoint, these macropters are characterized by slightly longer immature development times at 24-26°C, larger body size, enhanced walking activity, decreased male mating propensity, and delayed oviposition as compared with their brachypterous counterparts (Honěk, 1985(Honěk, , 1987(Honěk, , 1995Socha & Zemek, 2003;Socha, 2004). The incidence of macroptery increases southward while the daylength at nymphs were collected in urban parks in early November, transported by air to St. Petersburg and kept in the same manner as the northern population except that the environmental chamber was set to 25°C and a 12L : 12D photoperiod and linden seed was supplemented with dry seed of hollyhock Alcea rosea L. The remaining nymphs were allowed to complete development and accumulate reserves for overwintering. Then the rearing temperature and daylength were gradually decreased over several weeks and the bugs were eventually transferred to a cool dark room with a temperature of 4°C where they were stored until late January when the experiment began. Overwintered adults were kept in an environmental chamber in which the temperature was gradually increased to 22°C and the daily light-dark cycle, to 16L : 8D.
At the start of the experiment, fi rebugs were divided into malefemale pairs. Each pair was confi ned in a 9 cm Petri dish with a paper disc on the bottom and provided with food and water. In total, there were 28 northern and 42 southern pairs. Eggs laid by females of both populations were collected daily and transferred to small plastic Petri dishes (40 mm in diameter) which were kept in larger dishes (100 mm in diameter) on a layer of damp cotton wool to prevent desiccation. Thus, the experiment in both cases used the fi rst laboratory generation obtained from overwintered fi eld-collected parents. Eggs were laid by 25 of 28 northern and 24 of 42 southern females. Lower oviposition rates in the latter population were presumably due to incomplete reactivation from diapause.

Experimental procedures
Clutches of eggs (whole or divided into two, depending on size) were randomized among several combinations of constant temperature and photoperiod, with one environmental chamber per treatment. With the northern population, ten such combinations were used, comprising fi ve temperatures (20, 22, 24, 26 and 28°C) and two photoperiods (short-day 12L : 12D and long-day 22L : 2D). With the southern population, the same fi ve tempera-tures were used in combination with one of three constant photoperiods (short-day 10L : 14D, long-day 16L : 8D and extremely long-day 22L : 2D). Hatching was recorded daily. As hatchlings emerged synchronously within each clutch, all eggs that remained after mass emergence were considered non-viable. The initial median number of egg batches per treatment was 11.5 (minimummaximum: 8-17) for the northern and 8 (5-17) for the southern populations, with fewer clutches incubated at higher temperatures where survival to the adult stage was expected to be better. None of the females laid suffi cient eggs for its progeny to be included in all of the temperature-photoperiod combinations; the median number of treatments per parental female was 5 (minimum-maximum: 2-8 for the northern and 1-14 for the southern population). However, the distribution of clutches across the treatments was random and every treatment contained a diverse subsample of either population. The incompleteness of our experimental design with respect to parentage was later taken into consideration in the statistical analyses.
Newly hatched nymphs were transferred to 90 mm Petri dishes on the day of hatching and to 250 ml containers after moulting to the third instar. Nymphs were supplied with food and water in the same manner as the adults. Newly emerged adults were recorded daily and weighed on a digital analytical balance (Gosmetr VL-210 with 0.1 mg precision in the earlier experiment on the southern population and a Discovery DV215CD with 0.01 mg precision in the later experiment on the northern population). Wing morph was determined by examining the posterior abdominal tergites, which were exposed in brachypterous individuals and hidden under velvety-black forewing membranes in macropterous ones. Sexes were distinguished by the shape of the last abdominal sternite.
Temperature in the environmental chambers was maintained within ± 0.4°C of the desired level via a software-controlled balance of heating and cooling (RLDataView 1.03) and was automatically recorded every 10 s. Average rearing temperatures slightly deviated from the set values and are given in Tables 1 and  2. The chosen long-day photoperiods of 22L : 2D and 16L : 8D approximately corresponded to midsummer daylength, including civil twilight, at the collection sites. Although the southern population never experiences more than 16 h of light per day, the experimental design with this population did include the 22L : 2D photoperiod so that the southern reaction norm could be directly compared with its exact northern counterpart. The short-day photoperiod for the southern population had a shorter photo phase than that for the northern one because the former was expected to experience a broader range of daylengths in nature owing to a mild, frost-free winter in its Mediterranean habitat.

Statistical analyses
Statistical analyses were done using R version 4.2.1 with RStudio 2022.07.1+554 (RStudio Team, 2022;R Core Team, 2022). The infl uence of temperature and photoperiod on the characters that could be expressed as binary variables (survival rate, sex ratio and wing morph ratio) was tested with lme4 package (Bates et al., 2015) using generalized linear mixed-effects models with a logit link and binomial error structure. Maternal identity was included in the models as a random intercept term. The two wing-morphs were compared in terms of development time and body mass using one-way ANOVA followed by Tukey's HSD pairwise comparisons test.
A full analysis of the effects of temperature and photoperiod on developmental rate and body mass was only done for brachypterous bugs because these were present in all of the experimental treatments. In this analysis, actual incubation temperature was treated as a continuous predictor and photoperiod as a categorical predictor. The durations (D) of the egg and nymphal stages were transformed into rates (R = 1/D). The effects of temperature, photoperiod and sex on developmental rate and adult body mass were tested for signifi cance by fi tting maximum-likelihood linear models implemented in the R package nlme (Pinheiro et al., 2022) as this package allowed the specifi cation of the variance associated with each temperature level in order to overcome heteroscedasticity. Egg data were analysed using fi xed-effects models because nymphs hatched synchronously and thus there was only one developmental rate for the egg stage per group reared. For nymphs, individual developmental rates were available, and so mixed models were fi tted to nymphal data. In the mixed models, the random effects term was specifi ed as rearing density nested in maternal identity. Density was expressed as the fi nal, not initial, sample size (number of adults that emerged per group), because most growth took place during the later nymphal instars, whereas most mortality occurred during the earlier instars (pers. obs.). In addition to including rearing density in all models as a random effect, Pearson's correlation coeffi cients of the analyses of the number of adults reared per group and temperature were computed. Model assumptions of linearity and normality of residuals were verifi ed by visual inspection of residuals plots. The signifi cance of differences was determined with F-tests in fi xed-effects models and log-likelihood ratio (LLR) chi-squared tests in mixed models. Regardless of Akaike's criterion differences and p-values, terms were not removed, except for the factor sex (see below) and the reported statistics refer to full models. However, to visualize the effects of photoperiod and geographic origin on the thermal reaction norms for developmental rate, a separate linear mixed model was fi tted for each combination of origin and photoperiod, with temperature as the only fi xed effect. Thus, temperature-dependent development was described assuming a linear regression between the rate of development (R) and temperature (T): The parameters a and b with their standard errors were taken from the model's output and used to calculate the sum of degree-days (SDD) as 1/b and the lower temperature threshold (LTT) as -a/b (Campbell et al., 1974).
Overall, the survival rates of nymphs were lower under short-day conditions, but the difference was small and not consistent across temperatures (Fig. 1B, C).

The macropterous morp h
This morph occurred disproportionately with regard to all of the three factors studied: rearing temperature (more common at high temperatures), photoperiod (more common under either short-or long-day conditions), and geographic origin of the fi rebugs (more common in those of southern origin). To simplify the analysis and interpretation of the results, the fi ndings for the two morphs are presented separately.
In general, macropterous individuals took somewhat longer to develop than brachypterous individuals ( Tables 1 and 2). This difference in development time between the two morphs was signifi cant, e.g., in the northern population at 28°C (ANOVA, effect of morph: F 1,256 = 28.3, p < 0.0001; Tukey's HSD test: p = 0.04 under the 12L : 12D photoperiod and p < 0.0001 under 22L : 2D). An interesting exception was the effect of morph on development time in the southern population under 10L : 14D at 24-28°C. This effect was not signifi cant per se (F 1,162 = 0.02, p = 0.9) but the interaction of the factors of morph and temperature was highly signifi cant (F 2,162 = 11.4, p < 0.0001) and the Tukey's test showed that, at 28°C, macropterous individuals had a signifi cantly longer developmental time (p = 0.004); at 26°C, the development times of the two morphs were not signifi cantly different (p = 1.0), while at 24°C, macropterous individuals developed signifi cantly faster than brachypterous individuals (p = 0.004).
The macropterous morph also tended to attain a slightly greater body mass than its brachypterous counterpart (Fig.  3). However, this difference was not consistent throughout the experimental conditions and was rarely signifi cant. In the northern population at 28°C, there was no signifi cant difference in body mass between the wing-morphs (F 1,256 = 0.8, p = 0.4). In the southern population under the 10L : 14D photoperiod, there was a signifi cant interaction of the factors morph and temperature (F 2,162 = 6.9, p = 0.002) and the Tukey's HSD test indicated that macropterous individuals were only signifi cantly heavier than brachypterous ones at 26°C (p = 0.001). On average across all the rearing treatments, mean body mass (± SD) of southern macropterous individuals was 39.2 ± 5.48 mg and of southern brachypterous individuals 33.9 ± 5.48 mg.
Unfortunately, it proved diffi cult to carry out formal comparisons in many cases due to very different sample  sizes and patchy distribution of macropterous individuals across the treatments. The sections below deal with the temperature and photoperiodic responses of eggs and nymphs that developed into the brachypterous morph.
The effect of sex on developmental rate was unclear. Overall, northern females developed slightly but consistently faster than males (Table 1). However, the signifi cance of this effect varied greatly depending on how variances were modeled and whether (and which) interaction terms were included or not. In the fi nal model, which contained all main effects and double interactions, the effect of sex alone was non-signifi cant (LLR χ 2 (1) = 1.1, p = 0.3), but weakly signifi cant in some interactions. Similarly, there was no difference in body mass between freshly emerged males (mean ± SD across all treatments: 41.3 ± 6.61 mg) and females (41.5 ± 8.46 mg) (LLR χ 2 (1) = 0.03, p = 0.9). As the main focus of this analysis was the photoperiodic responses in a geographic context, the preferred option was to disregard the minor difference in developmental time between males and females, as it did not alter the overall picture.

Development and body mass recorded for the southern population
Developmental rates of southern fi rebugs signifi cantly depended on rearing temperature (egg batches: F 1,127 = 3492.8, p < 0.0001; nymphs: LLR χ 2 (1) = 183.2, p < 0.0001; Table 2, Fig. 4). There was no signifi cant effect of photoperiod on developmental rate of eggs when the data for the 22L : 2D treatment were removed from the dataset (F 1,85 = 0.02, p = 0.9). However, inclusion of these data resulted in a signifi cant effect of photoperiod (F 1,127 = 6.5, p = 0.01) and signifi cant temperature by photoperiod interaction (F 1,127 = 33.1, p < 0.0001). Compared with the ecologically relevant 16-h photoperiod, in the very long-day treatment, egg development was slightly faster at the three low temperatures and slightly retarded at the higher two. Photoperiodic conditions also signifi cantly affected the rate of nymphal development, both as a main effect (LLR χ 2 (1) = 21.1, p < 0.0001) and in interaction with temperature (LLR χ 2 (1) = 29.1, p < 0.0001). However, after the removal of the 22L : 2D data, the effect of photoperiod became nonsignifi cant (LLR χ 2 (1) = 0.01, p = 0.9) and the interaction disappeared (LLR χ 2 (1) = 0.4, p = 0.5), likely due to a reduction in sample size because otherwise the response to both long-day treatments was very similar (Fig. 4C). In general, southern nymphs developed more slowly and had a shallower thermal reaction norm under the 10L : 14D photoperiod than under either of the long-day photoperiods (Tables 2 and 3). All three regression lines for developmental rate on temperature crossed near the LTT (Fig. 4C). As with the northern fi rebugs, there was no difference in body mass of freshly emerged males (mean ± SD across all treat- ments: 34.0 ± 5.35 mg) and females (33.9 ± 5.63 mg), and so the data for both sexes were pooled. Adult body mass in the southern population signifi cantly depended only on developmental temperature (LLR χ 2 (1) = 5.6, p = 0.02): the fi rebugs tended to attain a larger size in warmer conditions (Fig. 3C).

Geographic differences
Eggs of the northern population developed signifi cantly faster (F 1,221 = 331.8, p < 0.0001) than those of the southern population and their low temperature threshold was lower (Fig. 4A, Table 3; the southern 22L : 2D data were not included in the regression analysis). In particular, mean egg development times at the temperatures from 20 to 28°C were 15.1, 10.1, 8.5, 7.2, and 5.9 d for the northern population and 16.3, 11.7, 9.2, 7.6, and 6.7 d for the southern one. Under short-day conditions, nymphs from the northern population developed signifi cantly faster than southern ones at temperatures below 26°C and slower than those at the highest temperature (main effect of geographic origin: LLR χ 2 (1) = 49.3, p < 0.0001; interaction with temperature: LLR χ 2 (1) = 42.3, p < 0.0001). There was no signifi cant difference in the temperature-dependent development of the two populations under long-day conditions, regardless of how they were compared: under 22L : 2D only or under 22L : 2D vs 16L : 8D. Firebugs from the northern population had a signifi cantly greater body mass (LLR χ 2 (1) = 13.6, p = 0.0002), especially at high temperatures (interaction: LLR χ 2 (1) = 10.2, p = 0.001) and under the long-day photoperiod (interaction: LLR χ 2 (1) = 41.3, p < 0.0001).

DISCUSSION
The fi rebug P. apterus is an extraordinarily diverse species with broad inter-and intrapopulation variation (Socha, 1993;Lopatina et al., 2007;Pivarciova et al., 2016;Ditrich et al., 2018). The results presented show that the northern (temperate) and southern (subtropical) populations of P. apterus studied differ not only in the absolute values of wing-morph occurrence, rate of immature development and adult body mass, but also in the thermal and photoperiodic plasticity of these traits.
Both populations are clearly thermophilic and perform better at high constant temperatures of 24-28°C, which is manifested in higher survival rates and bigger body size than at 20-22°C (Figs 1 and 3). There is a marked tendency for both northern and southern fi rebugs to be macropterous at high temperatures (Fig. 2). In both populations, there is no pronounced sexual size dimorphism, sex ratio is independent of temperature and photoperiod and both sexes respond to these factors in a similar way. While sharing these features, by and large the northern and the southern populations differ markedly in their responses to temperature and photoperiod, which was anticipated because the climate in their habitats is very different.
The northern population is less prone to be macropterous and long-winged individuals primarily emerge under longday conditions at high temperatures. The southern population had a higher proportion of macropterous individuals, especially under short-day conditions, and also produced these at low temperatures (Fig. 2). There is no consensus on the adaptive signifi cance of wing polymorphism in P. apterus (Socha, 1993;Honěk, 1995). Macroptery may be associated with greater dispersal capacity (Socha & Zemek, 2003) or, alternatively, may merely represent a neutral vestigial trait in the process of transition to complete brachyptery (Seidenstücker, 1953;Honěk, 1995). Our results corroborate previous fi ndings (Honěk, 1987;Socha & Šula, 1996;Socha, 2001) that geographic populations of P. apterus differ not only in the incidence but also in the environmental control of macroptery. For example, Honěk (1987) reports an unusually low percentage of macropterous individuals in a population from Istanbul, Turkey, reared under a long-day photoperiod at 26°C; comparing his result with ours (Fig. 2B), it is likely that the fraction of macropterous individuals in that population would have been higher under short-day conditions. However, the high percentage of macropterous individuals (about 23%) reported by this author for a population from Almaty, Kazakhstan, reared under the same conditions, clearly indicates that not all southern populations are similar in this regard. Thus, if there is an adaptive scenario for the occurrence of macroptery in P. apterus, it is likely to be complex and incorporate the thermal and photoperiodic plasticity of wing-morphs.
We do not have a good explanation for why the sex ratio at adult eclosion is slightly female-biased in the north and male-biased in the south. One possible mechanism for this could be sex-biased mortality during the immature period. Overall, the southern population survived worse under our experimental conditions and were smaller, which might have been due to feeding them on the seed of linden, a plant genus common in Europe but not native to Israel.
Egg development recorded for the northern population is faster and more tuned to low temperatures (Figs 1A, 4A). However, when egg batches of the southern population were kept under the 22L : 2D photoperiod, hatchlings emerged relatively earlier at low temperatures, i.e., at least superfi cially they appear to be similar to their northern conspecifi cs. It is unlikely, however, that photoperiod could directly affect embryonic development in P. apterus. The signifi cant effect of the 22L : 2D photoperiod on egg development in the southern population is presumably due to the disruption of a circadian hatching rhythm. As the postglacial northward expansion of P. apterus implies adaptation to low temperatures and long days, it would be interesting to examine in detail whether there are any latitudinal differences in the fi rebugs' daily hatching patterns. Our results hint at a possibility of such geographic variation in this species.
The photoperiodic responses of the nymphs of the brachypterous morph are markedly different in the two populations. The northern population has a strong photoperiod-by-temperature interaction controlling the rate of development (Fig. 4B) and in this respect it is very similar to the partially bivoltine, more southerly populations from Ryazan and Belgorod . In all these populations, long-day conditions accelerate nymphal development at high, but not low, temperatures. Together with basking behavior, which is well documented in P. apterus (Honek & Martinkova, 2019), this response may expedite maturation in fi rebugs in summer and facilitate bivoltinism. Conversely, short-day conditions accelerate nymphal development when combined with low temperatures. Even at the expense of a smaller fi nal body mass (Fig. 3A), this response is likely to be advantageous late in the season in terms of the timely completion of development before winter. The southern population exhibits only a weak response to photoperiod such that long-day conditions slightly accelerate nymphal development over the whole temperature range studied (Fig. 4C). The fi rebugs from Israel share this type of quantitative photoperiodic response with many other subtropical and tropical insects, including the sympatric pyrrhocorid bug, Scantius aegyptius (Kutcherov et al., 2018).
The phylogenetic tree for P. apterus reported by Pivarciova et al. (2016) consists of three distinct and wellsupported clades: Western European + (Israeli + Eastern European). Thus far, the emerging geographic pattern in the photoperiodic responses of nymphs is consistent with this topology as the P. apterus population from Israel has a markedly different response compared with that of Eastern European populations (Lopatina et al., 2007 and the present study). Unfortunately, the absence of such data for Western European populations prevents a broader generalization. What can be stated is that the short-day thermal reaction norm exhibits more geographic variation than its long-day counterpart as the former becomes considerably shallower northward and intersects the temperature axis at a comparatively low threshold value of 13.7°C (Fig. 4B, C; Table 3). Effectively, under short-day conditions, nymphs of northern origin develop signifi cantly faster than southern nymphs at low temperatures (the sign of the difference is opposite at 28°C due to a strong genotype-by-environment interaction, i.e., an intersection of reaction norms). In contrast, under long-day conditions, southern nymphs at low temperatures outperform northern ones, but only slightly.
From a physiological standpoint, photoperiod does not infl uence the rate of development, but rather the thermal reaction norm for development, including both the SDD and LTT. It should be stressed that SDDs calculated from different LTTs are not a proxy of development time and, generally, are not comparable. For example, in our experiments, nymphs of the northern population have a shallower yet more elevated reaction norm under short-day conditions, i.e., they develop on average faster than those under the long-day photoperiod, despite having a greater SDD (Table 3). Although our experimental design is static (the photoperiods are kept constant throughout development), previous results  indicate that this response to photoperiod may be more or less gradual. In other words, the LTT and SDD in P. apterus are not constant and change as the season progresses in the fi eld, which is especially conspicuous in the northern population. Similarly, it is not possible to say which of the two populations, the northern one or the southern one, develops more rapidly during the nymphal stages as the answer will depend on the combination of temperature and photoperiod under which the comparison is made. A possible confounding factor that may have contributed to the geographic disparities in our experiments is different parental overwintering history (natural overwintering in the northern population and simulated in the southern one). Such transgenerational effects are rather small in magnitude in P. apterus , but add another dimension to the variation in thermal reaction norms. Also, our experiments were not designed to explicitly test for the effects of rearing density (which is a nuisance variable here, treated as a random-effect term), and yet, temperaturedependent mortality resulted in larger groups at higher temperatures where nymphs might have developed faster, achieved a bigger size (Schmuck, 1995) and switched to the macropterous phenotype more frequently than under the infl uence of high temperature alone. The precise contribution of the factor group density remains to be quantifi ed in future studies.
Arguably, caution should be exercised when extrapolating any life-history traits in P. apterus beyond the season when and the location where these were experimentally measured. For example, the LTT and SDD can vary both in time (seasonally) and in space (geographically) ( Table 3). The use of a constant LTT coupled with a constant SDD likely oversimplifi es the diversity of current and projected phenological patterns in this species. Taking this variation into account, however daunting it may seem, will undoubtedly improve the accuracy of phenological forecasts.