Introduction

The European paper wasp Polistes dominula is a highly invasive species recently recorded in South Africa (Eardley et al. 2009; Veldtman et al. 2012; Benadé et al. 2014). Its native range includes parts of Europe, Asia, Russia and northern Africa, but it has now successfully spread to every continent except Antarctica (Cervo et al. 2000; Miller et al. 2013) via human mediated introductions (Buck et al. 2008; Miller et al. 2013). On many of these continents, it is intruding into the niche of native paper wasp species (Gamboa et al. 2002, 2004) which leads to direct competition for food, foraging habitat and nesting sites (Gamboa et al. 2002, 2004; Benadé et al. 2014). In South Africa, P. dominula was first recorded in 2008 (Eardley et al. 2009), but has rapidly expanded its range (Veldtman et al. 2012; Benadé et al. 2014). It shares this invaded distribution range with the native P. marginalis, but has become much more abundant than the native species (Veldtman et al. 2012; Benadé et al. 2014). Determining the specific traits that make P. dominula successful over the ecologically and taxonomically similar P. marginalis in its newly acquired habitat may provide vital information for its control and for the prevention of future invasions (Jones and Lawton 1995).

The success of P. dominula over native congener species in North America has been attributed to multiple factors including shorter brood development times, natural enemy release, and greater reproductive abilities (Cervo et al. 2000; Gamboa et al. 2004). Determining the success of this same exotic species over a different native congener, and on a different continent, provides opportunity to identify the most important factors that explain its success in general. Invasive success is often attributed to lack of regulation by natural enemies (Jeffries and Lawton 1984; Liebert et al. 2006), hence the drive behind biological control efforts are usually to reconnect exotic species with their natural predators and parasites, i.e. classical biological control (Irvin and Hoddle 2010). Recently, Madden et al. (2010), Miller et al. (2013) and Benadé et al. (2014) reported parasitism of P. dominula outside of its native range in North America and South Africa respectively. In North America, P. dominula is parasitized by Dibrachys cavus (Hymenoptera: Pteromalidae), Chalcoela iphitalis (Lepidoptera: Pyralidae) and a Sarcophaga sp. (Diptera: Sarcophagidae) (Miller et al. 2013). In South Africa, nests often contain larvae and puparia of the parasitic fly Anacamptomyia sp. (Tachinidae) (Benadé et al. 2014). This fly probably moved onto P. dominula from the native P. marginalis (Benadé et al. 2014). This provides the opportunity to assess parasite loads on colonies of both wasp taxa, and to evaluate the biotic release hypothesis (Jeffries and Lawton 1984; Liebert et al. 2006) as explanation for successful invasion by P. dominula (i.e. parasite load is expected to be much lower on the novel exotic species than the native species).

A smaller body size of an exotic competitor wasp can contribute to its invasive success, as it requires less energy to maintain a smaller body than a larger one (Armstrong and Stamp 2003). For social Hymenoptera, a smaller body size denotes a greater number of individuals produced and maintained (e.g. workers in ant colonies) for the same amount of energy consumed (e.g. a limited food resource) when compared to larger counterparts (Bourke and Franks 1995). This results in superiority through numbers, and a greater degree of colony productivity (McGlynn 1999). The advantages of greater numbers include better foraging efficacy, better brood care, and more success against competitors (Bourke and Franks 1995).

In North America, P. dominula has multiple advantages over the best studied of the native congener species P. fuscatus (Armstrong and Stamp 2003), including smaller body size, greater foraging efficacy, and a much higher reproduction rate. P. dominula has a shorter brood development time than P. fuscatus, allowing for greater colony productivity, and as a result, greater competitive fitness (Cervo et al. 2000; Gamboa et al. 2002; Armstrong and Stamp 2003). The greater colony productivity manifests as nests that contain three times as many cells in the case of P. dominula compared to P. fuscatus (Gamboa et al. 2002). This link between smaller body size, greater number of nest cells, greater nest productivity, and consequently, increased fitness of P. dominula over P. fuscatus (Gamboa et al. 2002) may also explain the apparent competitive advantage of P. dominula over P. marginalis in South Africa.

Here we determine whether the exotic P. dominula has characteristics that result in a performance advantage over the native congener P. marginalis in South Africa. We specifically compare these two species in terms of (1) overlap in foraging habitat, (2) differences in seasonal activity periods, (3) possible role of natural enemy release (i.e. parasitism rates by shared parasitoid species), (4) differences in body size, and (5) differences in colony productivity (in terms of cell numbers in nests). Because there is currently no effective way for surveying Polistes individuals using traps in South Africa (Van Zyl 2016), nests of P. dominula in its invaded range in North America are most often encountered on man-made structures in urban areas (Gamboa et al. 2002, 2004; MacKenzie et al. 2006; Benadé et al. 2014) and finding nests of paper wasps in natural areas are notoriously difficult (Liebert et al. 2006), we relied on an active search technique to enumerate numbers of foraging wasps.

Materials and methods

Habitat preferences and foraging activity

To determine habitat preferences and seasonal foraging activity periods for both species, the numbers of actively foraging wasps (wasps flying close to the vegetation at ground level that were seemingly searching for prey; i.e. not flying at high altitude and in a straight line) were determined at 40 sites over a period of 12 months (July 2013 to July 2014) in and around the town of Stellenbosch. This area was selected as it contains various habitat types within a relatively short distance from one another, and both P. dominula and P. marginalis are common in the region (Benadé et al. 2014). Sites comprised four habitat types, each with ten replicates, that were > 1 km apart to minimise pseudo replication. Urban sites included botanical gardens, schools, sport grounds and parks. Agricultural sites (rural) were set up in grassy areas on wine farms surrounding the town. Fringe sites contained transitional vegetation types between urban and natural sites usually with high cover of invasive alien plant species. Natural sites were indigenous fynbos vegetation and were > 1 km from nearest anthropogenic influences.

Visual surveys were undertaken for 20 min at each site during the first week of every month. The number of foraging individuals of P. dominula and P. marginalis observed was recorded. Active visual surveys consisted of continuous walking at a constant relaxed pace through the site in a randomised fashion by a single observer for ca. 300 meters and counting foraging wasps. Care was taken not to record the same individual more than once by not crossing already surveyed parts of transect. Surveys were conducted only on windless and cloudless days and between 11:00 and 15:00.

Body size, parasitoid load, and colony productivity

Polistes dominula and P. marginalis often nest only centimetres from one another on buildings in and around our study area (Benadé 2015). This allows for direct comparisons between the two species in terms of biology and ecology while partially controlling for environmental conditions. To compare body size, parasite loads and colony productivity between these species, nests and individuals were collected for physical measurements and nest-cell counts. Nests of both P. dominula and P. marginalis were collected at the end of the activity season in May 2014 (i.e. the start of winter), as this is when colonies would likely be mature. Nests were only collected from urban and rural sites, as no nests were observed in fringe and natural habitat types. Collection took place either at night or early morning, as this is when wasp activity was low and the majority of individuals were present on their nests. Nests and accompanying wasps were removed by dislodging nests from the substrate using a honey bee-hive tool and placing these in separate re-sealable plastic bags. Bags were frozen at − 20 °C to euthanize the wasps and for storage until further processing.

We determined body mass (as individual dry mass) of P. dominula and P. marginalis wasps collected with their nests after removal of foundress females when present (assumed to be the largest bodied, female individual). Wasps were oven dried at 60 °C for a period of 24 h before determining dry weight of individual wasps using a digital scale (Sartorius Model BP 110 S, Capacity: 0.1 mg–110 g).

Benadé et al. (2014) identified three parasitoids and/or hyper-parasitoids on nests of both the exotic P. dominula and the native P. marginalis. A fly from the genus Anacamptomyia (Tachinidae) was most prominent, allowing comparisons of nest parasitism rates in terms of number of cells affected per nest. Personal observations (FR) indicate that the only other common parasitoid (Eurytomidae) is likely a hyperparasitoid on the Anacamptomyia fly as emergence holes are often seen on fly puparia and cells containing these, and never on wasps or cells containing wasps. Unparasitized nest cells can easily be differentiated from those parasitized by Anacamptomyia as the latter contain brown fly puparia or their remnants. Parasitoid load for each nest was calculated using the number of cells containing fly puparia as a percentage of the total number of cells per nest.

Colony productivity, as an estimate of the potential number of offspring reared, can be assessed using the number of cells in a nest (Gamboa et al. 2002). To determine the potential colony productivity of P. dominula and P. marginalis, we therefore determined the total number of cells of each of the collected nests for both P. dominula and P. marginalis. It has to be noted that Polistes species often reuse cells in a nest for successive broods (Karsai et al. 1996). Therefore, actual colony productivity may be an underestimation of true colony productivity, and the degree of this may differ between different species.

Statistical procedures

Data on numbers of foraging wasps, habitat preferences, and monthly counts were statistically compared using generalized linear mixed models (GLMMs) calculated using the lme4 package in R (Bates et al. 2015). Habitat type (Urban, Rural, Fringe, Natural) and species (P. dominula, P. marginalis) were used as fixed effects, while observation sites nested within the observation month were used as random effect. GLMMs were calculated using a Laplace approximation, and data fitted to a Poisson distribution (Bolker et al. 2009). Two models were evaluated, the first testing for an additive effect between the fixed effects, and the second, testing for an interaction effect between these. Body size differences between P. dominula and P. marginalis were also compared using generalized linear models (GLMs) in R. Models tested included dry weight of individuals partitioned by species, the latter variable with the additive effect of wasp sex (male, female), or the interactive effect between these two variables. Similarly, data on total number of nest cells were compared between P. dominula and P. marginalis using generalized linear models (GLM) in R. Best-fit models were chosen based on the Akaike information criterion (AIC). Models with the lowest AICc scores were considered as best. Percentage data on parasitoid loads for each nest of each wasp species contained numerous zeros, and so we statistically compared parasitoid load data using a Pearson’s Chi squared test of independence in R.

Results

Habitat preferences and activity period

During the first 2 months of observations in winter (July and August), no foraging wasps of either species were observed. The first P. dominula individual appeared in September in the urban habitat type at the onset of spring as temperatures started to rise. This was more than 2 months before the first foraging P. marginalis individual was observed during November. Only one individual of P. dominula was observed in September while 12 individuals were observed during October. Abundance of both wasp species was highest in late summer/early autumn (February to April). However, from late autumn (May) onwards into winter (data not shown), their abundance score was zero.

A total of 809 foraging wasp individuals was observed during the entire survey period. This constituted 758 P. dominula and only 51 P. marginalis individuals respectively. The model that best explained wasp foraging abundance incorporated the interaction of the fixed effects habitat type and species with the random effect of observation site, nested within observation month (AIC = 1231.9; residual deviance/degrees of freedom = 1.177). In the best-fit model, P. dominula was significantly more abundant than P. marginalis (Table 1, Fig. 1). Habitat type also significantly influenced wasp numbers with greater abundance of foraging wasps of both species within or close to high-density human-settlements (urban and fringe habitats, Table 1, Fig. 1). P. dominula was especially abundant in these environments, with far fewer observations in natural and rural areas. In addition to urban and fringe habitats, abundance of P. marginalis was higher also in rural areas compared with no wasp individuals encountered in natural habitats (Table 1, Fig. 1).

Table 1 Best-fitting linear mixed-effects model showing the response of abundance of P. dominula and P. marginalis foraging wasps to the interacting effects of habitat type and species identity
Full size table
Fig. 1
figure 1

a numbers of individuals of P. dominula (n = 758 individuals) and bP. marginalis (n = 51 individuals) observed at four habitat types (n = 10 sites per habitat) for the entire observational period (n = 12 observational events during 1 year per habitat type). Note the difference in the scale of the y-axes

Full size image

Body size, parasitoid load, and colony productivity

In total, we collected 132 paper wasp nests. Of these, 101 nests were those of P. dominula and 31 nests were P. marginalis. Of the nests collected, 46 were still occupied by wasps at the time of collection (27 P. dominula and 19 P. marginalis nests). From these, a total of 858 P. dominula and 265 P. marginalis adult wasps were collected, and individual dry mass determined. The best-fit model for evaluating differences in the dry weight of collected individuals included only the difference in species (AIC = − 5875.4). According to this model, there was a significant difference in the dry mass of the two species (estimate = 0.004, standard error = 0.0012, z value = 3.341, P < 0.0001). However, the model explained only 1% of the total variation in the data. The median dry mass of the two taxa differed only slightly, and the two taxa had largely overlapping data ranges (Fig. 2). We therefore conclude that this statistical difference is likely due to the large sample size and that it is not biologically meaningful. We therefore consider the two taxa to be similar in size as measured by individual dry weight.

Fig. 2
figure 2

Median dry weight (g) of Polistes dominula and P. marginalis individuals collected at the end of the active season

Full size image

Overall percentage of cells per P. dominula nest parasitized by the Anacamptomyia fly was significantly higher than that of P. marginalis nests (x2 = 124, degrees of freedom = 96, P < 0.05) (Fig. 3). Most P. marginalis nests collected did not show any evidence of parasitization by the fly (only 13% of all nests collected showed any evidence of parasitization). In contrast, most nests (85%) of P. dominula collected showed some evidence of parasitization by this species. The numbers of cells parasitized in individual nests of P. dominula often exceeded 15%, and reached > 50%. Even so, P. dominula nests generally contained significantly more cells, and therefore had a greater colony productivity potential than those of P. marginalis (AIC = 1833.5,  % variation explained = 16.7%, residual deviation/degrees of freedom = 1.19, estimate = − 259.36, standard error = 50.80, z value = − 5.105, P < 0.0001) (Fig. 3). Nests of P. marginalis rarely contained > 100 cells, whereas nests of P. dominula in some cases contained 10 × more than that.

Fig. 3
figure 3

a Percentage of P. dominula and P. marginalis nest cells parasitized by Anacamptomyia sp. b Total number of cells in nests of P. dominula and P. marginalis collected at the end of the activity season

Full size image

Discussion

Polistes dominula was far more abundant in its newly invaded habitat in South Africa than was the native P. marginalis, even though they have similar habitat preferences. Foraging workers of P. marginalis were most common in association with human activity (i.e. urban, fringe, and rural habitat types). In contrast, P. dominula was most abundant in urban habitats, while significantly less at the urban fringe, yet far more abundant than in the natural and rural habitats. This suggests that resources such as prey and nesting sites in urban settings result in high P. dominula abundance. Although rural and urban habitat types supported the most nests, the main difference between urban and rural sites was the number of human structures available for constructing nests. For both P. dominula and P. marginalis, the natural habitat type was least preferred, most probably also due to lack of adequate nesting sites. These results agree with those of Gamboa et al. (2004), who showed that feeding and habitat preferences of the exotic P. dominula and native P. fuscatus in the USA are similar. As both P. dominula and P. marginalis probably use similar resources such as food (Cervo et al. 2000) and nesting sites, a high level of indirect competition may exist between these two species as in the USA (Liebert et al. 2006). However, we did not sample sites without P. dominula and therefore we cannot enumerate possible competitive effects on the numbers of P. marginalis here (Pilowsky and Starks 2018).

When compared with the USA, the great abundance of the exotic P. dominula in South Africa may have resulted from a longer seasonal activity period, natural enemy release, smaller body size and/or greater colony productivity (Cervo et al. 2000; Gamboa et al. 2002; Armstrong and Stamp 2003). For example, a smaller body size of an exotic species over a native competitor can lead to a competitive advantage in the exotic species (McGlynn 1999), because a smaller body size can denote a greater number of individuals produced (Bourke and Franks 1995) for the same amount of food resources (energy) available. This leads to a feedback system where greater numbers improves competition for resources (Bourke and Franks 1995). However, we found that P. marginalis and P. dominula in South Africa are very similar in size and the numerical dominance of the exotic species is therefore not coupled to a smaller body size as may be the case for the interactions between P. dominula and P. fuscatus in the USA.

Under predictions of the enemy release hypothesis, the exotic wasp would be less affected by parasitoid species than the native wasp species (Colautti et al. 2004). This would therefore give a competitive advantage of the exotic species over the native species in terms of release from top down population regulatory effects. Benadé et al. (2014) found the same taxa of native parasitoids in the nests of P. dominula and P. marginalis in South Africa. Contrary to expectations, here we show that the nests of P. dominula suffer significantly higher parasitism rates (median ca. 7% cells) than the nests of the native species (median 0% cells). Release from natural enemies is therefore unlikely to contribute to the success of P. dominula in the region. However, parasitoids, such as D. cavus in the USA, may help stabilise the populations of P. dominula, especially in the later stages of the invasion process (Miller et al. 2013). It is therefore expected that the parasitoids of P. dominula in South Africa may slow the invasion process and prevent populations of the wasp attaining ever greater numbers than at present in the invaded areas. However, Miller et al. (2013) made use of data on the percentage of nests containing signatures of total parasitoids as measure of parasitism levels. They found that ca. 60% of P. dominula nests were parasitized, while only ca. 20% of P. fuscatus nests showed evidence of parasitoids. When considering parasitism levels at this ‘nest’ level, and only for the Anacamptomyia sp. fly here, 85% of P. dominula nests contained signs of parasitism, while only 13% of P. marginalis nests were parasitized. Parasitism levels at the nest level for P. dominula is therefore close to a maximum in South Africa, but despite this, numbers of P. dominula are much higher than P. marginalis in the region.

Polistes dominula started actively foraging almost 2 months before the first foraging individuals of P. marginalis appeared. This is in sharp contrast to the situation in the USA, where there were no significant difference in the timing of colony initiation between P. fuscatus and P. dominula (Gamboa et al. 2002, 2004). Despite the similar colony initiation times in that study, workers of P. dominula appeared on average a week earlier than P. fucatus workers. This faster developmental time was linked to more ‘condensed’ larval and pupal developmental periods in P. dominula, that may be related to its smaller size (Gamboa et al. 2004). However, the early production of P. dominula workers in South Africa is probably due to its shorter larval and pupal periods (as nest initiation is earlier for P. dominula than P. marginalis) bearing in mind that the comparative body size of individuals of the two species are similar. Reasons for the much earlier activity of P. dominula in South Africa may therefore relate more to physiological differences between the native and the exotic species, such as greater lower thermal tolerances of the exotic species (Weiner et al. 2010, 2011, 2012). This aspect, together with studies on brood developmental times, should be investigated further. Some anecdotal evidence for the role of physiology in P. dominula success is provided by recent range expansion studies in Europe where nest productivity of P. dominula is very flexible and closely linked to climatic conditions (Höcherl and Tautz 2015). The much earlier activity of P. dominula likely results in a competitive advantage over P. marginalis as this allows earlier nest establishment, more advantageous nest placement (in terms of protection from water and predation) and earlier production of workers (Cervo et al. 2000; Gamboa et al. 2002).

At the end of the activity season, nests of P. dominula were significantly larger than those of P. marginalis and contained 3.8 times more cells (median of 273 cells) than those produced by P. marginalis (median of 72 cells). Our results are therefore very similar to those comparing colony size of P. dominula to the native P. fuscatus in the USA. There, the exotic species exhibited 3–5 times greater colony productivity (as measured by cell counts of nests) than the native congener (Gamboa et al. 2002, 2004). It has to be noted that, although every effort was made not to include nests containing multiple foundress females an re-used nests from previous seasons as was documented for P. dominula (Liebert et al. 2008), we cannot be certain that all nests in our study were founded by single females and were constructed in the current season. Therefore, a few of the larger nests may well over-estimate colony productivity. However, superiority in colony productivity by P. dominula over native congeners was also detected in its native range in Europe (Cervo et al. 2000), and when compared to P. metricus in the USA (Pickett and Wenzel 2000). Evidence is therefore mounting that superior colony productivity, linked to faster brood-developmental times, is one of the most important factors leading to the success of this species as an exotic competitor with native wasp species. As we show here, this can be independent of body size. However, other factors that may also contribute to its invasive success, such as those that contribute to colony productivity (e.g. better foraging rates and resource use) and factors associated with survivorship (e.g. lower usurpasion and predation levels) (Liebert et al. 2006), need further study.

To conclude, the success of P. dominula in its newly acquired habitat in South Africa is likely closely linked to its ability to produce large numbers of offspring earlier in season, allowing it to outnumber the native P. marginalis, which increases its chances of being a successful invader. Colony productivity of this species is more than three times higher than the native species, even though parasitism levels were significantly higher. The presence of parasitoids on P. dominula, although more pronounced than on the native P. marginalis, therefore does not reduce their numbers to levels lower than that of the indigenous species. Factors positively affecting colony productivity and survivorship, combined with a preference for human-altered environments, likely significantly contribute to the invasive success of P. dominula in the country. As previously indicated (Liebert et al. 2006), when this species successfully establishes itself in a new environment such as in South Africa, P. dominula will likely be a permanent member of human-altered environments with possibly a large impact on trophic levels in these environments. It could be either positive when it acts as an effective biological control agent or negative when it removes food sources for organisms at higher trophic levels. Its possible impact on native biodiversity is less clear, as it currently largely avoids these environments. Nevertheless, numbers of the exotic wasp was still much higher than that of the native species in natural environments, and its possible impact on native biodiversity should be monitored closely, especially as this area is a global biodiversity hotspot.