Factors determining local and seasonal variation in abundance of Harmonia axyridis ( Coleoptera : Coccinellidae ) in Central Europe

To determine the causes of the variation in the seasonal dynamics of Harmonia axyridis (Pallas) in Central Europe, numbers of adults and larvae of this invasive species were recorded on trees (Acer, Betula, Tilia) throughout the growing seasons from 2011 to 2016. Each year beetles were collected every two weeks, using a standardized sweeping method. The seasonal dynamics was expressed as plots of abundance (number of individuals per 100 sweeps) against time (Julian day) and these plots (seasonal profi les) were compared in terms of their size (area under the seasonal profi le curve), range, timing and height of the mode (maximum abundance). Timing and size of seasonal profi les varied among hostplants, years and sites. Abundance of larvae paralleled aphid occurrence and peak abundance of adults followed that of larvae 10 to 20 days later. Population dynamics before and after the peak were determined by dispersal. Adults arrived at sites before the start of aphid population growth and persisted there long after aphid populations collapsed. The abundance of H. axyridis decreased from 2011 to 2013 and then increased, achieving the previous levels recorded in 2015 and 2016. The variation in seasonal profi les revealed that H. axyridis, in terms of its response to environmental conditions, is a plastic species and this fl exibility is an important factor in its invasive success. * Corresponding author; e-mail: jirislavskuhrovec@gmail.com INTRODUCTION Harmonia axyridis (Pallas) is an invasive ladybeetle (Coleoptera: Coccinellidae) that arrived in the Czech Republic in 2006 (Sprynar, 2008). In three years, between 2007 and 2009, H. axyridis became a dominant species (Nedved, 2014), at times making up more than 90% of the adult ladybird species in coccinellid communities (Honek et al., 2015). The massive occurrence of H. axyridis decreased the diversity of coccinellid communities, not only by virtue of numerical dominance (Kindlmann et al., 2017), but also because of its likely detrimental effect on the abundance of some native species (Honek et al., 2016). Due to competition for prey and/or intraguild predation (Gagnon & Brodeur, 2014), this non-native species may negatively infl uence not only coccinellids but also other aphidophagous taxa (Brown et al., 2015). Since well-founded worries exist about H. axyridis devastating natural coccinellid Eur. J. Entomol. 116: 93–103, 2019 doi: 10.14411/eje.2019.011


INTRODUCTION
Harmonia axyridis (Pallas) is an invasive ladybeetle (Coleoptera: Coccinellidae) that arrived in the Czech Republic in 2006 (Sprynar, 2008).In three years, between 2007 and 2009, H. axyridis became a dominant species (Nedved, 2014), at times making up more than 90% of the adult ladybird species in coccinellid communities (Honek et al., 2015).The massive occurrence of H. axyridis decreased the diversity of coccinellid communities, not only by virtue of numerical dominance (Kindlmann et al., 2017), but also because of its likely detrimental effect on the abundance of some native species (Honek et al., 2016).Due to competition for prey and/or intraguild predation (Gagnon & Brodeur, 2014), this non-native species may negatively infl uence not only coccinellids but also other aphidophagous taxa (Brown et al., 2015).Since well-founded worries exist about H. axyridis devastating natural coccinellid pling was done by the same person (AH) using a standard sweep net (35 cm in diameter, 140 cm handle), on sunny and calm days, between 08:00 and 18:00 h.Each sampling session involved collecting coccinellids at a particular site from a particular species of tree and lasted 15-30 min depending on the number of trees.Successive sampling at the same site was done using the same number of sweeps.Adults and larvae (3rd and 4th instar, which can be quickly and correctly identifi ed) of H. axyridis were counted and immediately released at the site.In total, there were 727 sessions with a mean 127 ± 2.3 sweeps per session (range 50-301 sweeps).To compensate for the differences in the intensity with which the adults and larvae were sampled on each sampling session, the numbers were recalculated to numbers per 100 sweeps (further "n/100"), which is referred to as "abundance".

Sampling aphids
Tree stands in this study were populated by host specifi c and holocyclic species of aphids.Tilia stands were regularly populated by Eucallipterus tiliae (L.), which has a single peak of abundance, the timing and the size of which vary from year to year (Dixon & Barlow, 1979;Dahlsten et al., 1999).The dynamics of E. tiliae populations are determined by the size of the overwintering population, hostplant condition, predator activity (Dixon & Barlow, 1979) and aphid behaviour (Kidd, 1976).Betula stands were populated by Euceraphis betulae (Koch) whose winged females do not aggregate (Dixon & Thieme, 2007) and infest young as well as senescent foliage (Johnson et al., 2003), or by Calaphis fl ava Mordvilko which prefers young foliage (Dixon & Thieme, 2007).Acer stands were populated by Drepanosiphum platanoidis (Schrank), a holocyclic species whose winged females live dispersed on the undersides of leaves (Dixon & Thieme, 2007) and the seasonal dynamics of which is determined by hostplant quality and population density (Dixon et al., 1993).Periphyllus spp.(not identifi ed to species) was also present on maple.The term "hostplant" thus includes plant species, its architecture and microclimate, and behaviour (distribution on hostplant, escape behaviour) of the aphid fauna.Aphid abundance on Tilia was classifi ed on a scale of four degrees (hereafter referred to as 'degree of aphid abundance ' [daa]).These degrees were based on the aphid counts on 50-500 leaves, with (1) indicating no aphids, (2) < 0.002 aphids per leaf (aphids in swept material but no individuals found on 500 leaves), (3) ≥ 0.002 to 1.0 aphid per leaf, and (4) > 1.0 aphid per leaf.cessful propagation of H. axyridis in recently invaded areas.
The factors that shape the timing and extent of the seasonal variation in abundance of H. axyridis are likely to be similar in both its native and invaded areas (Osawa, 2011).Aphids are essential for H. axyridis.Due to H. axyridis's great ability to locate prey (With et al., 2002) and quickly start reproducing, this species is able to establish sub-populations in temporary habitats, patches of hostplants populated by aphids and persist there for a long time (Osawa, 2005).In urban and suburban areas there are trees, which are populated by abundant populations of aphids (Fluckiger & Braun, 1999;Mackos-Iwaszko et al., 2015).Adult H. axyridis prefer to stay within sub-populations on such trees although the beetles that move away may have a good chance of reaching a patch with a higher aphid density (Osawa, 2000).This habitat fi delity is supported by readiness to accept, besides essential aphid prey (Hodek & Evans, 2012), a wide range of alternative animal and plant food such as Sternorrhyncha (McClure, 1987;Michaud & Olsen, 2004), Heteroptera (Tillman, 2013), Thysanoptera (Zhang et al., 2014), Coleoptera (Stuart et al., 2002), Lepidoptera (Koch et al., 2003), immatures of other predators (Pell et al., 2008), mites (Lucas et al., 1997), pollen, nectar and fruit (Koch et al., 2004;Lundgren et al., 2004;Mathews et al., 2016).
Although many aspects of the life history of H. axyridis in Central Europe are well known, the seasonal dynamics of local populations in Central Europe need further study.Particularly unexplored are the variations in local and annual abundance of sub-populations (populations on groups of trees at particular sites) within a metapopulation (a group of partially isolated sub-populations) of H. axyridis (Lincoln et al., 1998).This study summarizes six years of records of the abundance of larvae and adults of H. axyridis on different species of trees in a suburban area.We address the seasonal variation in abundance of larvae and adults of H. axyridis on particular trees and analyze the relationship of this species with site, hostplant and aphid abundance.We tested the hypothesis that the timing and the size of the peak in adult abundance are determined by the size of larval populations that develop at a site rather than the number of adults that arrive at and/or leave the site.

Sampling H. axyridis
Harmonia axyridis were sampled in the western part of the Czech Republic, at 8 sites situated within a 5.0 × 1.5 km area, centred at 50.0860N, 14.2954E.The sites with groups of lime (Tilia cordata Mill., 7 sites), maple (Acer platanoides L., 2 sites) and birch trees (Betula pendula Roth., 2 sites) were at least 200 m apart from each other (Table 1).
Each year sampling started after budburst, in late April or early May, and terminated at leaf fall, in late October to mid-November.The coccinellids were sampled at intervals of two weeks, between the 5 th to10 th and 20 th to 25 th each month.The beetles were swept at particular sites from the lower canopy of the same trees and from heights up to 3 m (measured from the ground).Sam-Table 1. Characteristics of the sites sampled: Site number (used throughout this study), coordinates of the sites, tree species growing at each site (Species: A -Acer platanoides L., B -Betula pendula Roth., T -Tilia cordata Mill.), number of trees growing at each site (n), their trunk diameter at 1 m height (Diam.)and height (Height).In the text, localities are indicated by a combination of tree species and site number, such as T1 = Tilia stand at site 1.

Data analysis
For each hostplant, site and year, the abundance of H. axyridis established for particular sampling sessions was plotted against calendar time (Julian days) and the plots (further called "seasonal profi les") were described by four characteristics: (i) "Size" of the area below the profi le of the seasonal curve i.e. a product of abundance and time given in "individual*days" units [IDU], (ii) "Range" equal to number of days that elapsed between the earliest and the latest date of species presence, (iii) "Mode position" corresponding to the Julian day when maximum abundance of a species was attained in a given year.(iv) "Mode height" equal to maximum abundance given as number of individuals per 100 sweeps (further n/100).
Analyses were carried out using the seasonal profi les of three hostplants, Betula (two sites), Acer (two sites) and Tilia (seven sites) (Table 1), and six (adults) or fi ve (larvae) years.Variation in seasonal profi les was fi rst investigated using ANOVA (ANOVA), followed by Tukey's HSD (multiple-comparison-corrected) test (HSD) with the characteristics of each profi le (profi le size, range, mode position and mode height) as response variables and hostplant or year as factors.
To investigate the relationship between the seasonal profi les of adults and larvae two methods were used: (i) The coeffi cients of regression of the population characteristics (defi ned above) of adults on those of larvae.To compare the seasonal dynamics of abundances of larvae and adults a regression of maximum adult abundance on maximum abundance of larvae was calculated.To compensate for differences between adults and larvae and for annual variation in their abundances, the data for a particular developmental stage (larva or adult) and year were standardized within years using the formula xs = (xo -xa)/sx, where xs is the standardized value of abundance, xo is the original (observed) value of abundance, xa is average abundance and sx is the standard deviation of the abundance values.(ii) Plotting the seasonal course of adult abundance against time scaled according to the course of development of the larvae.For each site and year, the data were normalized as follows: On the abscissa (x-axis representing time) are the dates on which sampling occurred, the date of the peak (mode position) number of larvae was set to 0 and the dates on which sampling occurred were designated by the length of the period (number of days) prior to the population peak (negative values) or by length of period (number of days) after the population peak (positive values).On the ordinate (y-axis), the relative abundance of adults was expressed for each site as a percentage of the maximum (mode-value) attained in a particular year.The data were fi tted using the Asymmetric Double Sigmoidal (ADS) function, which explained maximum variance in the data, using TableCurve 2D (Systat Software, 2002 ).

RESULTS
Seasonal profi les and changes in the abundance of H. axyridis in calendar time were unimodal and their characteristics varied among hostplants, years and sites.

Variation in the seasonal dynamics of adults of H. axyridis
For adult H. axyridis (Fig. 1), the size of the seasonal profi les was signifi cantly (p ANOVA = 0.0007) affected by hostplant.Abundance on Tilia (N = 40, mean = 3982 ± 372.1 IDU) was signifi cantly greater than that on Acer (N = 12, 2159 ± 461.9 IDU, p HSD = 0.  Annual variation in the seasonal profi les of adults was as important as variation among hostplants.There were signifi cant annual differences (Fig. 2 There was no interaction between hostplant and annual effects for any of these characteristics. Local (among sites) variation in the size of the seasonal profi le (p ANOVA < 0.0001, F = 10.259) and maximum abundance (mode height) (p ANOVA 0.0012, F = 5.090) was important on Tilia.The ranking of sites based on mode height was different in particular years (Fig. 3A).At some sites, the values of these characteristics were consistently low as e.g. at site T4 where mean size of the seasonal profi le was 2140 ± 355.7 IDUs (range 1326-3721 IDUs) and mean mode height (maximum abundance) was 37 ± 4.7 n/100 (range 24-53 n/100).In contrast, at site T5 mean profi le size was 6658 ± 1036.4 IDUs (range 4780-10638 IDUs) and mean mode height (maximum abundance) was 125 ± 23.3 n/100 (range 57-188 n/100).

Variation in the seasonal dynamics of larvae of H. axyridis
For larvae, the effect of hostplant was less conspicuous than for adults (Fig. 4).Comparison of the sizes of the seasonal profi les was on the borderline of formal signifi cance (p ANOVA = 0.0643, F = 2.911).Size of the seasonal profi le was greatest on Tilia (638 ± 34.0 IDUs) and smallest on Acer (308 ± 8.0 IDUs) and Betula (133 ± 8.0 IDUs).The ranges in the seasonal profi les (90 ± 6.4 d for Tilia, 85 ± 10.5 d for Acer and 69 ± 20.5 d for Betula) did not differ signifi cantly (p ANOVA = 0.4210, F = 0.882).Mode height (maximum abundance) (21 ± 3.9 n/100 for Tilia, 11.0 ± 4.0 n/100 for Acer and 4.3 ± 1.2 n/100 for Betula) was not signifi cantly affected by hostplant (p ANOVA = 0.0863, F = 2.583).But the position of the mode (date of maximum abundance) was signifi cantly different among tree species (p ANOVA = 0.0046, F = 6.047).The population mode occurred later on Tilia (Julian day 215 ± 6.1 d, which is 2 nd August) than on the other two tree species, Acer (Julian day 175 ± 13.7 d, which is 23 rd June, p HSD = 0.0277) and Betula (Julian day 174 ± 18.5 d, which is 22 nd June, p HSD = 0.0256).
There was a signifi cant hostplant x year interaction for range characteristics (p ANOVA = 0.0027, F = 3.793) and position of the mode (p ANOVA < 0.0001, F = 10.130).In 2014 and 2016 the ranges of the seasonal profi les were broad and the mode (maximum abundance) was attained later, while the reverse was true in 2012, 2013 and 2015 (Fig. 5).In contrast, sizes (area) of the seasonal profi les (p ANOVA = 0.0582, F = 2.478) and mode height (maximum abundance) (p ANOVA = 0.1641, F = 1.715) did not vary significantly among years.

Factors determining population dynamics
Adult peaks (total n = 50) coincided or immediately (on the next sampling date) followed larval peaks in 28 (56%) cases (combinations of year x site).In 19 cases (38%) adult peaks followed larval peaks 1 to 3.5 months later and in only 3 cases (6%) did the adult population peak precede the larval peak, by 1 to 2 months.The extreme asynchrony was recorded in cases when adults assembled before (Betula) or  1).A -adults, B -larvae.Despite the rare asynchrony in occurrence, abundance of larval populations and abundance of adult populations, there were signifi cant correlations between the seasonal profi les of adults and larvae, the sizes (p < 0.0001, r = 0.7474), range (p = 0.0159, r = 0.3394), mode position (p < 0.0001, r = 0.5788) and height of mode (p < 0.0001, r = 0.7127) (Fig. 6).The match of larval and adult seasonal dynamics was best on Tilia (Fig. 7) where the larval peaks occurred when aphid abundance was highest and adult peaks followed the larval peaks.The patterns in the coincidence of larval and adult population peaks differed with the timing of aphid and H. axyridis occurrence.In 2015 when H. axyridis populations peaked early in the season (Fig. 7A), adults assembled and immatures developed in synchrony with the increase in the aphid populations.Larval populations peaked early, 1 to 1.5 months after the arrival of the adults.Abundance of adults peaked 8 d after the larval peak (ADS function: R 2 = 0.5233, F = 18.4451).Adults then stayed at sites long after the decline in aphid abundance (3-5 months).In 2016 when H. axyridis peaked late in the season (Fig. 7B), adults started to assemble, in small numbers, long before the increase in aphid abundance (at some sites > 50 days earlier) and the larval population peak (4 to 5 months), apparently feeding on alternative prey.Abundance of adults peaked 23 d after the larval peak (ADS function: R 2 = 0.7062, F = 38.4549).The adults did not persist long after the peak since they left the trees and fl ew to hibernacula.

DISCUSSION
Seasonal dynamics of coccinellids have been studied over a long period of time by many researchers (e.g.Skuhravy & Novak, 1957;Radwan & Lövei, 1983; Hoff-    , 1997;Vandereycken et al., 2013;Kawakami et al., 2016).Evaluation of their results reveals that for the safe interpretation of the seasonal dynamics of coccinellids observations should be replicated in space (sites, hostplants) and over time (years), as in our study.

Factors determining the seasonal variation in abundance
The six year census of abundance revealed constant and variable characteristics in the seasonal dynamics of H. axyridis.The seasonal profi les of larvae and adults were unimodal and their size, timing and shape varied among sites and years.Throughout the growing season, the adult population consists of mainly overwintered individuals that produce up to three generations of offspring in Central Europe (Honek et al., 2018a).Thermal conditions enable the overwintered adults of this species to start reproducing at the end of April, the fi rst generation, which may start laying eggs in late June, the second generation in late July and the third one as late as in early September (Honek et al., 2018a).Oviposition by the second and third generations (i.e.under shortening day length) is possible because they only have a weak tendency to winter diapause (Reznik & Vaghina, 2013;Reznik et al., 2015).Since adult life is long, individuals of all generations born during a growing season may survive into autumn and overwinter (Honek et al., 2018a).It is unknown in what proportions the individuals of the successive generations contribute to adult seasonal profi les in particular years and sites.Determining the age structure of natural populations requires information not yet available for H. axyridis; such as the length of their reproductive period, time of dormancy induction and percentage of individuals that enter diapause in each generation (Hagen, 1962;Hodek, 2012).
Essential prey are a crucial factor limiting reproductive activity.It is a prerequisite for attracting adults and development of larval populations.This is not surprising considering the food specialization of coccinellids (Hodek & Evans, 2012), but we expect that H. axyridis is limited less than native species because it is very polyphagous.Harmonia axyridis arrival before aphids was observed in years when aphid abundance on Tilia peaked late (Fig. 7), however, they did not reproduce.Threshold aphid population densities necessary for oviposition under natural conditions are unknown but may be quite low, as in Coccinella septempunctata L. (Honek, 1980).The adults thus aggregate in large numbers at the very beginning of aphid population development.Towards the end of a season low numbers of adults (Fig. 1) may persist for long periods because they can feed on alternative prey.In contrast to adult seasonal profi les the range in larval profi les was narrow, usually ~ 1 month, probably because oviposition is limited to a "window" (Doumbia et al., 1998;Hemptinne & Dixon, 2000), which ensures there is usually suffi cient aphids for the larvae and for H. axyridis to lay eggs during the "window" just before aphid numbers peak (Kajita & Evans, 2010).

Timing of reproduction
In many potentially polyvoltine species of native coccinellids including Adalia bipunctata (L.), Hippodamia variegata (Goeze), Ceratomegilla undecimnotata (Schneider), Coccinella septempunctata L. and Propylea quatuordecimpunctata (L.) breeding does not occur late in the season.Reproduction in these native species is limited to the fi rst half of the growing season probably because reproduction mainly occurs in stands of herbaceous plants (Hodek, 1960).In contrast, that of some native species developing on trees such as Calvia decemguttata (L.), C. quatuordecimguttata (L.) and Oenopia conglobata (L.), may extend into autumn (Honek et al., 2019) and their larvae may feed on populations of monoecious species of aphids that persist until late in the season.Seasonal profi les of larvae of H. axyridis, however, extend to as late as November (Honek et al., 2018a).A prolongation of larval development into autumn is hazardous if winter comes early.However, at least in warm years, late larval development may be successful.This extension of the period available for development, although risky, may contribute to the success of invasive species such as H. axyridis in extending the area colonized.

Demographic processes
Seasonal profi les of coccinellids are shaped by demographic processes, natality and mortality, and movements of individuals (dispersal/migration).A signifi cant positive correlation between larval and adult seasonal profi les (Fig. 7) indicates that local reproductive activity is an important determinant of adult population size.At least two facts indicate that dispersal (movement of adults between sites) is also important in determining adult population size.First, temporal disconnection between peak larval and adult occurrence, which was recorded in a few cases, when the timing of the mode in adult occurrence was delayed by > 1 month after the larval peak.These late peaks may be composed of "autochthonous" individuals that developed locally plus immigrants from surrounding areas.The source populations of immigrants is likely to be stands of herbaceous plants (Honek et al., 2019) on which the abundance of H. axyridis declined because of plant senescence.Abundance of H. axyridis in stands of herbaceous plants and crops is low but their area in agricultural landscapes is much bigger than the area covered by stands of trees, their preferred habitat.The role of migration in the seasonal dynamics of H. axyridis is important.In native populations in East Asia, adults fl y continuously between sub-populations (Osawa, 2000).Adults in invasive populations also fl y regularly (Nalepa, 2013) perhaps in search of new hostplant stands and thus increase the probability of fi nding habitats populated by aphids.Flights supported by wind may cover several kilometres (Jeffries et al., 2013).This high mobility results in the opportunistic location and exploitation of small patches of prey (With et al., 2002) and contrasts with the sedentary behaviour of native species of coccinellid, which tend to remain at sites where they reproduce until the prey become extinct (Kokubu, 1986).
Although it was impossible to survey all the habitats where this species occurs, those studied included the hostplants they prefer (Adriaens et al., 2008;Viglasova et al., 2017).On Tilia, abundant assemblages of adults followed peaks in aphid abundance, whereas the adults that assembled on Acer and Betula bore no relation to the numbers that bred there.Adults are able to exploit aphid outbreaks whose timing is unpredictable by surviving in relatively high numbers at sites where aphid prey abundance is insuffi cient for breeding, or even at sites where there is only alternative prey.Occupation of such transient habitats depends on a high dispersal capacity and is no less important than breeding locally.

Short-and long-term population trends
The results contradict intuitive expectations concerning seasonal and long-term trends in H. axyridis abundance.Numbers of H. axyridis adults in particular sub-populations on trees (this study) and other hostplants (Honek et al., 2019), which make up its metapopulation, should increase over the course of a season as a consequence of polyvoltinism and an increase in the number of individuals that reproduce (Honek et al., 2018a).Adults, therefore, should be more abundant in autumn than in spring.However, this trend was not refl ected in our data (Fig. 1) as there was no systematic difference in the modal height (maximum abundance) of the early (2012,2015) and late (2011,2016) adult peaks (Fig. 2C, D).Our six year study of the population dynamics of H. axyridis encompasses more than half of the period of its occurrence in the Czech Republic, where this species became established in 2006 (Sprynar, 2008).During this period we expected this species to increase in abundance if it was still continuing to spread or no trend if it had reached the "carrying capacity" of this newly occupied area.This analysis did not reveal a monotonic increase in the size of the seasonal profi les (Fig. 2A).
The results point to the importance of long-term studies for understanding the mechanisms of population development in H. axyridis.Annual variation in seasonal profi les revealed that H. axyridis is plastic in its response to environmental conditions.This fl exible response to trophic and climatic conditions (Honěk et al., 2018b) is apparently an important factor in the invasive success of this species.Other factors determining the abundance of H. axyridis, e.g.metapopulation size and landscape quality, await clarifi cation.

Fig. 1 .
Fig. 1.Annual variation in seasonal profi les of the abundance (number of individuals per 100 sweeps) of H. axyridis adults on Tilia (fi gures on left-hand side, sites T1-T8), Acer (fi gures on right-hand side, sites A3 and A4) and Betula (fi gures on right-hand side, sites B3 and B6).
Fig. 2. Annual variation in mean (± SE) values of characteristics of the seasonal profi les of H. axyridis adults on Tilia.A -size of the area below the seasonal profi le curve (individual*days units IDU), B -range of the seasonal profi le curve (days that elapsed between the earliest and the latest date of species occurence), C -mode position (Julian day when maximum abundance of adults was attained), D -mode height (maximum abundance, number of adults per 100 sweeps).
) in the area of the seasonal profi les (p ANOVA = 0.0204, F = 2.932) and their ranges (p ANOVA < 0.0001, F = 14.156), which were narrowest in 2013 and broadest in 2016.There was signifi cant variation (p ANOVA < 0.0001, F = 12.849) in the position of the mode (timing of maximum abundance), which was early in 2012 and 2015 and 40-60 d later in 2011 and 2016.Mode height (maximum abundance) varied signifi cantly (p ANOVA = 0.0024, F = 4.255) and was highest in 2015 and 2016.

Fig. 4 .
Fig. 4. Annual variation in seasonal profi les of the abundance (number of individuals per 100 sweeps) of H. axyridis larvae on Tilia (fi gures on left-hand side, sites T1-T8), Acer (fi gures on right-hand side, sites A3 and A4) and Betula (fi gures on right-hand side, sites B3 and B6).
after (Acer) hibernation, without any relation to breeding at the site.

Fig. 5 .
Fig. 5. Annual variation in mean (± SE) values of characteristics of the seasonal profi les of H. axyridis larvae on Tilia.A -range of the seasonal profi le curve (days that elapsed between the earliest and the latest date of species occurrence), B -mode position (Julian day when maximum abundance of adults was attained).

Fig. 6 .
Fig. 6.The regression of modal values (maximum abundances) for H. axyridis adults on modal values for the larvae.Data for Tilia in 2012-2016.To compensate for the annual variation in abundance the data were standardized within years.The graph thus shows abundance as a population mean 0 ± SD.R 2 = 0.5030, p < 0.0001.

Fig. 7 .
Fig. 7. Seasonal variation in the abundance of adult H. axyridis on Tilia.Adult abundance (percentage of maximum at each site) is plotted against the time (days) that elapsed from the date when larval abundance was at its maximum (0).A -data for 2015 when population dynamics peaked early in the season, B -data for 2016 when population dynamics peaked late in the season.The symbols indicate different degrees of aphid (Eucallipterus tiliae) abundance at the time of sampling: □ no aphids ▲ < 0.002 aphids per leaf ♦ ≥ 0.002 to 1.0 aphid per leaf ■ > 1.0 aphid per leaf.The data were fi tted using the Asymmetric Double Sigmoidal (ADS) function.