Most species of ladybirds (Coleoptera: Coccinellidae) are carnivorous predators that feed on a variety of phytophagous arthropods including aphids, coccids and mites (Giorgi et al., 2009; Weber and Lundgren, 2009) and therefore render important services to agricultural and forest ecosystems through biological control of those pests. As such, non-native coccinellid species have been introduced around the world. Biological control using coccinellids started in the late 1800s with the successful introduction of Rodolia cardinalis Mulsant from Australia to control populations of the coccid Icerya purchasi Maskell infesting citrus orchards in California. Since then, several other ladybird species have been introduced to new regions for classical biological control (Roy and Migeon, 2010). The objective of the imports and releases was the establishment of self-perpetuating populations in order to exert natural control over specific pests (Michaud, 2012). Besides the benefits, this process also bears risks, including the spread and establishment of ladybirds in new geographic ranges. The monetary costs of managing alien species once established are often very high. The environmental costs (e.g., effects of an alien biological control agent on species other than the pest species) may also be high (van Lenteren et al., 2008). Thus preventing invasion in the first place is highly desirable, and in order to do this, knowledge of the pathways of invasion are needed (Hulme, 2015).

Whilst there have been numerous intentional coccinellid introductions since the end of the nineteenth century (Hokkanen and Sailer, 1985), the negative impacts were only recently acknowledged (van Lenteren et al., 2003, 2006). The need for intensive study of the consequences of alien coccinellids has been realized due to the negative effects of two species: Coccinella septempunctata L. in the Nearctic and Harmonia axyridis Pallas worldwide. However, initially there was little (or no) concern that these species would have negative impacts on “non-target” indigenous species, through species interactions such as competition and/or intraguild predation.

The review of non-native ladybird species in Europe to 2008 was provided by Roy and Migeon (2010) and updated to include the data of the period 2009–2017.

The first prediction was tested with aphidophagous species of the subfamily Coccinellinae originating in the Palearctic that spread to the Nearctic. This is because of the availability of information of species distribution in the source area and presence in the recipient area. Seven species that became established in the Nearctic were selected for analysis: C. septempunctata, C. undecimpunctata L., H. axyridis, H. dimidiata (Fabricius), H. quadripunctata (Pontoppidan), Hippodamia variegata (Goeze), and Propylea quatuordecimpunctata (L.) (Hemptinne et al., 2012). The size of native area in the Palearctic was determined using data for Coccinellidae in the Catalog of Palearctic Coleoptera (Kovář, 2007). This catalog divides the Palearctic, mainly according to political boundaries, to 51 districts of Europe and 85 districts of Asia. The size of area occupied by a species was classed as the total number of districts occupied in Europe and Asia. This area size was determined for the above species that invaded the Nearctic and for each of their non-invasive congener species that occur only in the Palearctic. Within each genus the data on species distribution were standardized as SAi = (MAg–Ai)/SDAg where SAi is standardized area of the species i, MAg is mean area of the genus, Ai is area of the species i and SdAg is standard deviation of the mean area of the genus. After standardization, data of all genera were tested for difference between size of area of invasive species (N = 7) and their congeneric non-invasive species (N = 40: Coccinella 27 spp., Harmonia 7 spp., Hippodamia 3 spp., Propylea 3 spp.) using the Mann-Whitney test.

Concerning the second prediction, the relationship between body-length and the number of ovarioles (representing reproductive potential) was tested using morphological and biological data published in Nedvěd and Honěk (2012), using linear regression y = a + bx, where x is body length (mm) and y is ovariole number. The differences in the number of ovarioles were tested using the Mann-Whitney test.

To test the third prediction, we focused on aphidophagous species that invaded from the Palearctic to the Nearctic, and that are widely distributed in the new area (Hemptinne et al., 2012) and for which the relevant thermal characteristics are available. Thermal constants, lower development threshold (LDT) and sum of effective temperatures (SET), were established for a number of coccinellid species. The constants were calculated from published data (Supplementary Material) using linear regression R = a + bT of development rate (R) on temperature (T) as LDT = −a/b and SET = 1/b. Differences in either characteristic were not useful for comparison. This is because LDT and SET are dependent, which means that errors of estimation of LDT are paralleled by errors in estimating SET (high estimates of LDT are compensated by lower estimates of SET and vice versa). Using LDT and SET for eggs, pupae, and total development, we calculated development duration at a range of ecologically relevant temperatures (16, 18, 20, 22, 24, 26 and 28°C) and tested for differences between invasive and non-invasive species. For each stage and temperature the differences between invasive and native species were compared using the Mann-Whitney test.

Regarding the fourth prediction, native and exotic species were characterized in terms of their dry body mass. The power function W = 0.032 L^2.62 was used to calculate dry body mass (W; mg) where L is the body length (mm) (Rogers et al., 1976; Sabo et al., 2002); the median body length data for particular species were retrieved from the literature (Dobrzhansky, 1926; Bielawski, 1959; Leeper, 1976, 2015; Gordon, 1985; Raimundo and Alves, 1986; Rathour and Singh, 1991; Stathas et al., 2002). The potential reproductive rate was expressed as half the number of ovarioles, with data retrieved from the literature (e.g., Hemptinne et al., 2012; Nedvěd and Honěk, 2012; and personal observations). Predicted dispersal abilities of native and alien aphidophagous ladybirds were estimated from the values of dry body mass raised to the power of 0.5618, the sum of individual effect of potential reproductive rate (0.3718) and dispersal ability (0.19). We found a significant positive correlation between the logarithm to base e of adult biomass (lnAM) and that of the potential reproductive rate (half the number of ovarioles–lnPRR) and this was expressed as lnPRR = 0.3718 lnAM + 2.6085 (r = 0.50 p = 0.013). The value of the power of dispersal ability takes into account the maximum rate of oxygen consumption, which increases with body mass raised to the power 0.85, and the cost of running, which increases with the exponent 0.67 (Schmidt-Nielsen, 1984) (for further methodological details, see Hemptinne et al., 2012).

Finally, we tested the significance of the values of potential dispersal ability of the aphidophagous alien species (H. axyridis, H. conformis, and H. convergens) against the average potential dispersal ability of native species (one sample t-test; p = 0.05) (SPSS, v. 24).

The 12 alien ladybird species recorded in Europe were introduced to control coccids, aphids, psyllids and whitefly, with the majority of the species being coccidophagous. In 104 species-country records a ladybird species was released in a country as a biological control agent (further called “direct” cases), whereas in 25 species-country records a ladybird species spread from a country in which it was released to a neighboring country (“indirect” cases). In two cases the source of the alien ladybirds is unknown. These figures include species-country records in which the ladybird did not establish, and non-target effects are not evident in many further cases. Hence, only in a relatively small proportion of cases did the alien coccinellid have widespread non-target effects (and these mostly involve H. axyridis). Within the “indirect” biological control cases, the exact mechanisms of spread are often unclear, and for each case of an alien arriving in a particular country, it may do so by various means. For example, in Great Britain some H. axyridis arrived by natural dispersal from mainland Europe, but others are known to have been introduced with products (e.g., imported cut flowers, vegetables and fruits), and yet others arrived with people, transported inadvertently in cars, trains and ships (Roy and Brown, 2015). Additionally, the source populations were not restricted to Europe: some H. axyridis arrived in Great Britain in packages originating from Canada (Majerus et al., 2006). The examples for which the exact pathway is known are rare: in most cases the mechanisms by which a particular insect or group of insects arrived to a particular place is unknown. As such, the best that we can do is to classify all such arrivals under a broad banner such as “introduced as an indirect result of biological control.”

Perhaps as would be expected, food relationship seems to be a good indirect predictor of a successful coccinellid introduction: worldwide, from the total attempts to control aphids (155) and coccids (613) using classical biological control approaches, only one case (0.65%) of aphid control turned out to be satisfactorily successful and complete success has never been achieved (Dixon, 2000). On the contrary, the control of coccids has been considered complete in 23 cases (3.75%) and substantially successful in 30 attempts (4.89%) (Dixon, 2000).

According to our initial prediction, invasive ladybirds tend to be more dispersed geographically in their native areas. An ideal picture of coccinellid distributions would be via high resolution digital maps in which each pixel represents a small area of the Palearctic marked by the presence or absence of the species. However, reaching this goal seems impossible in the near future because of the difficulties associated with recording species presence in most parts of the world. In this study, we have chosen a compromise solution of dividing the Palearctic into political districts and using data on species presence within these districts. This approach has some pitfalls, particularly since the districts are of unequal landscape diversity and size. Small districts with e.g., great altitudinal differences may provide an array of diverse local conditions greater than large districts comprised of flat landscapes. Consequently a species living in a small district with a diversified landscape may be adapted to a wider range of environmental conditions than a species living in an extensive area with uniform landscape conditions. However, the correlation between landscape diversity and range of species adaptation to environmental conditions is not as important as it may seem. Regardless of the range of available options of habitat choice, coccinellid species prefer particular habitats determined by hostplant, microclimate, prey and seasonality (Honěk, 1985; Nedvěd, 1999; Honěk et al., 2015). A consequence of this habitat specialization is a limited choice of utilizable sites, even in areas of high landscape diversity. However, some variation in environment cannot be eliminated even by a strong individual habitat choice: this is the variation associated with climatic differences among districts that fill a large geographic area. Thus a species distributed in a large geographic area, despite its narrow habitat selection, is likely to meet, cope with and be adapted to a greater range of environmental conditions than a species distributed in a small geographic area and this adaptation may be, in our opinion, a basis of its success in colonization of new geographic areas. Our results thus confirm that invasive species are, in their native areas, more prone to adapt to a wide range of conditions compared to non-invasive species in their native ranges.

For generalist predators, flexibility and breadth of habitat use and diet are direct underlying factors of a successful invasion. In general, successful invaders in their new ranges tend to occupy the same habitats as in their native ranges. This is the case with C. septempunctata and H. axyridis, which are most abundant in herbaceous and arboreal habitats, respectively. However, despite occurring in many semi-natural and natural habitats in their invaded ranges, C. septempunctata and H. axyridis, as well as the successful A. bipunctata and P. quatuordecimpunctata, tend to be abundant in agricultural habitats and urban landscapes (e.g., Lucas et al., 2007; Brown et al., 2008; Roy and Migeon, 2010; Evans et al., 2011). In the USA, H. axyridis became a dominant species in agricultural landscapes, thriving in disturbed and man-made habitats such as orchards and urban gardens, feeding on a wide range of aphid species (Koch, 2003; Evans, 2004). In Europe, H. axyridis is more common in northern and central countries, where it can be found in trees, shrubs and herbs (Brown and Roy, 2017; Honěk et al., 2017; Soares et al., 2017). Coccinella septempunctata, H. axyridis and A. bipunctata have a broad and flexible diet range, being quite responsive to new prey species (Evans and Toler, 2007). However, the mechanism by which integration of a new prey species into the diet of a generalist predator is able to drive adaptation to a new geographic range, is unclear. To explore this, it is important to evaluate whether life-history traits of organisms are responsive to selection, and what constraints and trade-offs may limit the overall possible set of traits. However, there are only three studies addressing prey adaptation, two using the generalist H. axyridis as a model (Ueno, 2003; Fukunaga and Akimoto, 2007), and one using A. bipunctata (Rana et al., 2002). Over the course of several generations the aphidophagous species were able to improve their fitness when fed on a single diet of lower prey quality. However, the genetic and epigenetic mechanisms that promote this difference remain to be studied.

The absence of significant differences between developmental rates among aphidophagous ladybirds agrees with previous results from laboratory experiments. For instance, in some studies the overall pre-imaginal development time of H. axyridis and C. maculata were similar (Labrie et al., 2006) with other studies demonstrating a slower developmental time of H. axyridis in comparison to other species (Adalia bipunctata L., H. variegata, Lanzoni et al., 2004; P. quatuordecimpunctata, Mignault et al., 2006; Olla v-nigrum Mulsant, Michaud and Olsen, 2004). An expected consequence of faster developmental rate is the possibility of the invasive species achieving more generations per year. However, this does not occur even with successful ladybirds spreading into novel ranges, partly because of lower average temperatures. For instance, within much of the invaded range, H. axyridis only achieves two generations per year (Roy et al., 2016), whereas in the native range up to eight generations have been observed (Osawa, 2011), this difference being associated with environmental temperatures. In conclusion, our results do not corroborate our initial prediction because the invasive species may not have an advantage against natives in using the thermal environment.

Although fecundity is influenced by abiotic (e.g., temperature) and biotic (e.g., prey quantity and quality) factors, plus factors such as age and life-history (Nedvěd and Honěk, 2012), body size is a primary determinant of the number of ovarioles (Hemptinne et al., 2012). The scatter of data for large aphidophagous species was greater than for small non-aphidophagous species. This may lead to invasive species having higher potential fecundity than non-invasive species. In fact, whilst large invasive species have, in some cases, more ovarioles than predicted by the overall relationship, this was not strongly evident in our results. In the case of H. axyridis fecundity seems to be similar to that of some other aphidophagous species (Soares et al., 2001, 2004, 2008; Bazzochi et al., 2004; Lanzoni et al., 2004; Michaud and Olsen, 2004). For instance, Italian specimens of H. axyridis presented a mean fecundity of 783.8 eggs per female, only slightly higher than that of the indigenous species A. bipunctata (720.2 eggs per female) (Bazzochi et al., 2004). Additionally, fewer eggs were laid by H. axyridis (560.5 eggs per female) than by H. variegata (841.7 eggs per female) in a laboratory study by Lanzoni et al. (2004). Soares et al. (2004) reported an even lower fecundity of 314 eggs per female for the aulica phenotype of H. axyridis. Michaud and Olsen (2004), using psyllids as a food source, showed that H. axyridis laid as many eggs as O. v-nigrum (242.8 vs. 224.5 eggs per female). Despite these similarities in fecundity, the differences in ovariole number do not always mirror clutch size or fecundity. This is because daily and lifetime egg production are influenced by other environmental factors. However, we predicted that H. axyridis has higher dispersal ability than H. conformis and H. variegata. To test this prediction, we compared the potential rate of spread of the three species based on their potential reproductive rates and adult body sizes. In short, invasiveness in aphidophagous ladybirds is associated with their large body size, a good surrogate of reproductive potential and dispersal ability, which determines their efficacy in exploiting aphid outbreaks in extensive forestry and agro-ecosystems (Honěk et al., 2008; Kajita and Evans, 2010; Hemptinne et al., 2012; Soares et al., 2017).

Our data point out trait differences between native/alien and invader/established aphidophagous ladybirds and some traits seem to explain successful establishment. However, data of species that have been introduced from the same region but have not become established in the new region would be important in order to better understand invasion processes.

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.