Bacterial symbionts as mediators of ecologically important traits of insect hosts

Authors


Heike Feldhaar, Behavioural Biology, University of Osnabrück, Barbarastr. 11, D-49076 Osnabrück, Germany. E-mail: feldhaar@uni-bayreuth.de

Abstract

1. Bacterial symbionts play a prominent role in insect nutritional ecology by aiding in digestion of food or providing nutrients that are limited or lacking in the diet. Thereby, endosymbionts open niches to their insect host that would otherwise be unavailable.

2. Currently, several other ecologically relevant traits mediated by endosymbionts are being investigated, including enhanced parasite resistance, enhanced heat tolerance, and influences on insect–plant interactions such as manipulation of plant physiology to the benefit of the insect.

3. Traits mediated by endosymbionts are often identified by correlative studies where traits are found to be altered in the presence of a particular symbiont. Recent developments in genomic tools offer the opportunity for studying the impact of bacteria–insect symbioses under natural conditions in a population and community ecology context. In vivo experiments specifically testing putative functions of endosymbionts in parallel to population-level studies on the prevalence of endosymbionts allow disentangling host versus symbiont contribution to phenotypic variability observed in individuals. Effects of symbionts on host phenotype are often large and relevant to host fitness, e.g. by significantly enhancing survival or fecundity in a context-dependent manner.

4. Predominantly vertically transmitted endosymbionts contribute to the heritable genetic variation present in a host species. Phenotypic variation on which selection can act may be due to differences either among host genomes, symbiont genomes, or genotype × genotype interactions. Therefore the holobiont, i.e. the host including all symbionts, should be regarded as the unit of selection as the association between host and symbionts may affect the fitness of the holobiont depending on the environment.

Introduction

Symbiosis, the persistent and intimate association of organisms belonging to different species, is a ubiquitous phenomenon in nature. In his seminal book Endosymbiose der Tiere mit pflanzlichen Mikroorganismen published in 1953 Buchner described hundreds of different symbioses of insects with microorganisms with an emphasis on anatomy (Buchner, 1953). Symbiotic associations may lead to ad hoc novelty in the host organisms as the incorporation of an entire functioning organism, with all its metabolic pathways, may at once confer a suite of novel traits to the host organism. Symbiotic bacteria have long been known to facilitate colonisation of new feeding niches of insects, allowing specialisation on a broad range of diets (Douglas, 2009). Bacterial endosymbionts may either provide essential nutrients that are lacking in the diet or aid in digestion and detoxification of food (Cardoza et al., 2006; Adams et al., 2009). Symbiotic bacteria have been shown to be major players in defence of the insects themselves by enhancing pathogen and parasitoid resistance (Currie et al., 2003a,b; Oliver et al., 2003, 2009, 2010; Kaltenpoth et al., 2005; Hedges et al., 2008; Teixeira et al., 2008; Brownlie & Johnson, 2009; Kaltenpoth, 2009) or may aid in prey preservation and nest hygiene (Currie et al., 1999, 2003a; Kaltenpoth et al., 2005; Kaltenpoth, 2009). In addition, endosymbionts have been shown to mediate thermal tolerance of their hosts (Dunbar et al., 2007) and to facilitate use of novel hosts (Tsuchida et al., 2011).

In the last decade the function of more and more of these symbionts to their hosts has been uncovered following two different approaches. First, in vivo tests were conducted, measuring reduction of host fitness either after eradication of the endosymbionts from the hosts or when providing food lacking substances putatively provided by the endosymbionts. Second, the putative function of endosymbionts has been deduced from their genome sequence and subsequent gene expression studies. Recent developments in genomic tools (next generation sequencing allowing massive metagenomic approaches and cheap genome sequencing) have now opened the door for studying the impact of bacteria–insect symbioses under natural conditions in a population and community ecology context. An increasing number of genomes of endosymbionts has been published, allowing a first interpretation of the potential biological role of an endosymbiont in the association with the host. In vivo experiments specifically testing potential functions identified by genome sequencing can provide more definite evidence for the function of an endosymbiont (Feldhaar et al., 2007; Gunduz & Douglas, 2009). The first comparative studies are emerging that attempt to disentangle host versus symbiont contribution on phenotypic variability of insects at a population level (Hosokawa et al., 2008; Oliver et al., 2008; Vorburger et al., 2009).

In this review, I will give a broad overview of the diverse functions of bacterial symbionts on insect host ecology and the magnitude of effect of symbionts on ecologically relevant host traits is compared. Due to the wealth of recent discoveries on how bacterial symbionts contribute to the insect host's phenotype this review is restricted to bacterial symbionts, while fungal and eukaryotic symbionts are excluded here. I will discuss future directions such as the potential role of endosymbionts in invasiveness of insect hosts, on endosymbiont-mediated capacity to enter new niches (termed ‘ecological opportunity’ in Janson et al., 2008), and potential impact of endosymbionts on host behavioural ecology.

The sum is more than its parts: the holobiont and hologenome concept

In recent years, the awareness of the impact of mutualistic symbiotic relationships with bacteria on diverse aspects of insect host biology has steadily increased. With very few exceptions, all animals have incorporated microorganisms that became organelles such as mitochondria (Embley & Martin, 2006). Then, aside from the ubiquitous gut microflora in animals, numerous invertebrates harbour endosymbiotic microorganisms inside their body cavity. Thus, what we generally perceive as a single eukaryotic organism – such as a pea aphid – is rather an aggregation of a few to many different organisms, each equipped with its own genome and metabolism. Our emphasis is therefore currently changing from studying the form and function of host individuals as a derivative solely of their own genotype and phenotype towards a more holistic view. Thus, hosts are increasingly studied as holobionts, i.e. as an organism whose phenotype is determined by the combined genotype of the host's genome and genome(s) of all symbionts carried by the host (Zilber-Rosenberg & Rosenberg, 2008; Rosenberg et al., 2010). In turn, the sum of genetic information carried by the holobiont and the microbiota is termed hologenome (Zilber-Rosenberg & Rosenberg, 2008).

Evolution of the holobiont

Genetic variation and the entailing phenotypic variation among holobionts that selection can act upon may be due to differences either among host genomes, symbiont genome(s), or genotype × genotype interactions within the holobiont. Additional variation based on the microbiota can result from the acquisition of novel symbionts from the environment or differences in the proportion of symbionts, i.e. an increase of a particular symbiont in relation to others (Rosenberg et al., 2009). As the holobiont comprises a suite of independently replicating units it experiences multilevel selection; similar to intragenomic conflict within the host genome, competition between symbionts or genotype × genotype interactions within the holobiont may result in conflicts within the hologenome and, as a result, the holobiont. The conflict between host and symbionts can be high and may reduce host fitness substantially, e.g. when symbionts manipulate the host's sex ratio (Engelstädter & Hurst, 2009). The holobiont should be regarded as the unit of selection in evolution, rather than the host without its symbionts, as the association between host and symbiont may affect the fitness of the holobiont depending on the environment.

The traditional view of the evolutionary process by descent with modification is that a trait in an organism is inherited from its ancestors and changes in a stepwise fashion over time due to the accumulation of mutations. Likewise, phenotypic traits induced by heritable endosymbionts that are transmitted vertically underlie selection at the host individual level (Ferrari & Vavre, 2011). Variation among holobionts on which selection can act can arise additionally via the uptake of a new symbiont from the environment and subsequent transfer to the next generation. Novel complex traits can be acquired by the holobiont in a single step as a whole organism with a suite of metabolic capabilities is incorporated. When lateral transfer between hosts is possible, then beneficial symbionts – and the ecologically important traits they confer such as enhanced parasitoid resistance – can spread faster within a host species as well as among host species than beneficial mutations within the host genome (Hurst & Darby, 2009; Hurst & Hutchence, 2010; Himler et al., 2011). However, as carrying an endosymbiont will usually entail at least maintenance costs they may be lost again due to selection on the host if they do not confer a fitness benefit (Oliver et al., 2008; Ferrari & Vavre, 2011).

Bacterial endosymbionts associated with insect hosts

Endosymbionts have traditionally been grouped into primary and secondary symbionts. Associations between hosts and primary endosymbionts are often ancient, with an estimated age of 30–250 million years (Baumann, 2005). Following presumably a single infection event, primary endosymbionts are transmitted exclusively vertically via the germ line. Primary endosymbionts are generally considered to be mutualistic since they are usually required for host survival and reproduction, often by contributing to the hosts' diet by providing nutrients that are otherwise lacking in food. Associations based on metabolic needs between primary endosymbionts and their hosts are thus generally obligate for both partners. Primary endosymbionts reside in specialised host cells that may constitute a larger organ-like structure, the bacteriome. An estimated 15% of insects harbour this type of endosymbiont (Buchner, 1953; Baumann, 2005). In contrast, secondary endosymbionts are often facultative symbionts from the host's perspective and usually have a shorter coevolutionary history with a single host species (Dale & Moran, 2006). Their incidence may be from sporadic only to fixation in a host species (Simon et al., 2003; Gueguen et al., 2010). Secondary endosymbionts do not necessarily reside in specialised host tissues but may occur extracellularly in the haemocoel or in other body tissues like fat body, muscle, nervous tissue, or gut and occur at lower titres in comparison with primary endosymbionts (Dobson et al., 1999; Moran et al., 2008). They are primarily transmitted vertically but horizontal transmission among hosts occurs (Russell et al., 2003; Dale & Moran, 2006; Oliver et al., 2010), although the frequency of such transmission events is yet generally not known. Therefore, related hosts may harbour distantly related strains of an endosymbiont or even different strains within a single host, while the same strain may be found in distantly related hosts. Whereas, primary endosymbionts have adapted to the host and vice versa, stable associations between secondary endosymbionts and hosts can be formed readily in some cases as exemplified by successful experimental transfer among individuals intra- and interspecifically (Tsuchida et al., 2004; Vorburger et al., 2010), perhaps implying that no prior host adaptation (e.g. the presence of bacteriocytes) is required for chronic infections – a feature that secondary endosymbionts share with pathogenic bacteria (Dale & Moran, 2006; Feldhaar & Gross, 2008; Moya et al., 2008). Relationships of secondary endosymbionts can range from beneficial to detrimental for the host. Secondary symbionts residing in the body cavity may be maintained in a population through four different factors (Hurst & Darby, 2009): (i) If vertical transmission is ineffective, then sufficiently high rates of horizontal transmission can balance the loss of symbionts from matrilines. (ii) Host sex ratio may be manipulated to produce more female offspring carrying the symbiont, or (iii) fitness of uninfected females may be reduced due to incompatibility with infected males. (iv) Endosymbionts may provide direct fitness benefits to the host and thus they are maintained by selection. The high prevalence of several secondary endosymbionts without obvious parasitic phenotypes can sometimes be linked to such direct fitness effects, such as in bedbugs where Wolbachia has evolved into an obligate endosymbiont that provides its host with vitamins of the B group that are lacking in blood meals of their hosts (Hosokawa et al., 2010). Thus, categorisation into primary and secondary endosymbionts may not always be clear-cut as the endosymbionts form a continuum in function from obligate to facultative associations as well as from pathogenic to mutualistic.

Microbes of the gut microflora are also considered endosymbionts (Moya et al., 2008; Brune & Ohkuma, 2011), albeit they are localised extracellularly within the gut lumen. Microorganisms of the gut microflora can vary in their persistence within the gut, ranging from transient visitors to permanent inhabitants. Nonetheless, a specific gut microflora may be a necessary component of the insect's microbiota since its absence may have considerable negative effects on development and immune function (Zurek et al., 2000; Buchon et al., 2009). However, the distinction between bacteria that are contained extracellularly within the gut and those harboured within the body cavity may be blurred. Recently it was shown that the endosymbiont of the olive fly Bactrocera oleae was localised intracellularly in the larval gut tissue but then moved into the gut lumen in adult flies (Estes et al., 2009). Likewise, numerous insects harbour bacteria in specialised gut compartments such as pouches or caecal evaginations as stable gut microflora. In spite of the extracellular existence, vertical transmission and thus strict coevolution can be attained by passing the bacteria to the next generation via the egg surface, bacteria-containing capsules, or faecal droplets (Kaltenpoth et al., 2009; Kikuchi et al., 2009; Ohkuma & Brune, 2011). Such specific gut microbiota may play a substantial role in host nutritional ecology by providing essential nutrients or aiding in digestion of resources, such as lignocellulose in termites (Brune & Ohkuma, 2011).

Ecologically important traits mediated by endosymbionts

To date a suite of traits of the holobiont mediated by symbiotic microbiota has been identified. Whereas primary endosymbionts are quite well characterised and often play a role in aiding in digestion of otherwise unsuitable diets or provisioning of essential nutrients, other ecologically important traits are frequently conferred by facultative secondary endosymbionts. To date, most of these traits were characterised in correlative studies on model organisms in symbiosis research such as the pea aphid or Drosophila. More and more traits can now be correlated with the presence of the often less well characterised secondary symbionts (Teixeira et al., 2008; Brownlie & Johnson, 2009; Oliver et al., 2010). Interestingly, even seemingly well characterised symbionts such as Wolbachia (Werren et al., 2008; Engelstädter & Hurst, 2009) are frequently found to be responsible for wider range of holobiont phenotypes than formerly assumed (Teixeira et al., 2008; Hosokawa et al., 2010).

Adaptation to the abiotic environment

The effect of endosymbionts on temperature tolerance of their insect hosts is the sole abiotic factor that has received substantial attention so far. The range and variability of temperatures that an organism can tolerate is an important factor in determining its geographic range. Temperature can either have direct effects on the insect hosts or indirect effects by changing the abundance of symbionts within the host or their efficiency of transmission to the offspring. Thus, it was already observed two decades ago that Wolbachia-induced phenotypes such as parthenogenesis or cytoplasmic incompatibility are attenuated by exposing insects to heat (e.g. Hoffmann et al., 1990), presumably due to reduced survival of the bacteria under heat stress.

In the pentatomid bug Nezara viridula it was shown that the transmission rate of symbionts decreased from 100% at 20 °C to below 10% at 30 °C. Loss of the symbionts did not shorten lifespan of the insects but resulted in strongly reduced fertility at lower temperatures (Prado et al., 2009). Likewise, aphids cannot tolerate high temperatures well (Dean, 1974; Oliver et al., 2010). Although the aphid host itself may not be adapted to higher temperatures, here again the endosymbionts may limit tolerance of higher temperatures. The number of bacteriocytes, the cells containing the primary endosymbiont Buchnera that supplies essential nutrients to the host, has been shown to decrease dramatically at higher temperatures or heat shock (Montllor et al., 2002). The secondary endosymbionts Serratia symbiotica and Hamiltonella defensa can ameliorate aphid fitness under heat stress (Montllor et al., 2002; Russell & Moran, 2006; Harmon et al., 2009) presumably by enhancing retention of bacteriocytes and survival of Buchnera under heat stress. Selection for heat-shock tolerance results in higher prevalence of the secondary endosymbionts in aphid populations after periods of summer heat or in hot desert sites (Harmon et al., 2009). Presumably, the bacterial chaperone groEL, which is constitutively overexpressed in primary and secondary endosymbionts (Wilcox et al., 2003; Stoll et al., 2009), may also protect host proteins from heat degradation when circulating in the haemolymph.

Variation in the genome of Buchnera itself adds to variation in heat tolerance of the pea aphid. A point mutation in the promoter of a heat-shock gene in Buchnera aphidicola eliminates expression of the heat-shock gene even under heat stress. When the promoter is impaired, aphids lack heat tolerance and produce hardly any offspring after a short exposure to heat stress. The polymorphism may be maintained in the population, as reproductive rates of those aphids are higher under constantly low temperatures (Dunbar et al., 2007).

Rickettsia secondary endosymbionts may confer heat tolerance to the whitefly Bemisia tabaci. When Rickettsia are scattered in the tissues of the insect host they induce expression of stress-related genes under normal temperature regimes. While the number of Rickettsia decreases strongly when insects are confronted with higher temperatures, the stress-related gene products already present in the host may confer thermotolerance (Brumin et al., 2011).

Influence of endosymbionts on plant–insect interactions

The most well known ecological trait conferred to insect hosts by endosymbionts is their role in providing nutrients essential to their hosts. Especially phloem- or xylem-feeding insects and cellulose-digesting insects such as termites can survive exclusively on strongly imbalanced food sources by microorganisms synthesising essential nutrients (Baumann, 2005; Douglas, 2009). As several excellent recent reviews have treated the role of obligate primary endosymbionts in insect nutritional ecology (Baumann, 2005; Douglas, 2009; Clark et al., 2010), I will concentrate here on the role of facultative secondary endosymbionts in mediating plant–insect interactions.

In many phytophagous insects, different populations of a species may specialise on different food plants locally, potentially resulting in the formation of host races or biotypes (Mopper & Strauss, 1998; Funk et al., 2002). As host race formation may be an initial step towards speciation, this topic has received much attention from evolutionary biologists. Generally it has been assumed that the traits enabling a shift to a new food plant are mediated by genes encoded in the insect genome (Via et al., 2000; Funk et al., 2002; Stireman et al., 2005; Via & West, 2008). Several recent studies on food plant use of herbivorous insects revealed that fitness of insects can be enhanced by secondary endosymbionts.

The host spectrum utilised by aphids has been shown to be correlated with the presence of the secondary endosymbiont Regiella insecticola (Tsuchida et al., 2004, 2011; Ferrari et al., 2007). A beneficial effect of Regiella can even be transferred interspecifically. The vetch aphid Megoura crassicauda occurs sympatrically with pea aphids. The former species does not use clover as a host under natural conditions and does not harbour Regiella. Nonetheless, upon infection with the symbiont, survival of M. crassicauda was significantly enhanced on clover, albeit still significantly reduced in comparison to controls on vetch (Tsuchida et al., 2011). Transfection experiments in pea aphids revealed that Regiella enhances reproduction on clover (Tsuchida et al., 2004; Ferrari et al., 2007). However, only some pea aphid lineages showed increased fitness, pointing towards genotype × genotype interactions between host and facultative endosymbiont (Leonardo, 2004; Ferrari et al., 2007). Performance on vetch, the alternative host, was reduced in the presence of Regiella. Moreover, artificial infection with the secondary endosymbiont reduced acceptance of aphids of both vetch and clover (Ferrari et al., 2007). Thus, although Regiella influences performance as well as host acceptance behaviour in aphids, the impact of the endosymbiont is not necessarily positive and seems to be context dependent.

As in aphids, the endosymbionts of Megacopta stinkbugs are correlated with food plant use. The endosymbionts are transmitted to offspring via symbiont-containing capsules deposited with the egg masses. When symbionts of Megacopta punctatissima, a pest species performing well on soybean and other crop legumes, were exchanged between egg masses with the closely related non-pest species Megacopta cribaria that suffers low egg hatching rates on the respective crop legumes, egg hatch rate was inverted between the two species (Hosokawa et al., 2007). It is not clear though how the symbiont facilitates usage of the crop legumes. The symbiont of the pest species M. punctatissima may either perform better on the crop plants, e.g. by aiding in detoxification of a plant secondary compound, or may provide nutrients lacking on potentially suboptimal crop plants (Hosokawa et al., 2007).

Endosymbionts may thus indirectly limit use of food plants. For example, aphids reared on hosts that provide only low amounts of amino acids in the phloem have elevated levels of secondary endosymbionts. However, as the secondary endosymbionts do not contribute to amino acid nutrition of the aphids, negative effects of low quality phloem on aphid performance are exacerbated (Wilkinson et al., 2007; Chandler et al., 2008).

The secondary endosymbiont Wolbachia confers fitness benefits to the leaf-mining moth Phyllonorycter blancardella (Kaiser et al., 2010). The presence of Wolbachia in the larvae of the leaf-miner is positively correlated with high levels of cytokinins in the mined tissue of leaves. High levels of cytokinin result in a local delay in leaf senescence, thus inducing the ‘green-island’ phenotype, where photosynthetically active green tissue patches are maintained among senescent yellow leaf tissue. These green tissue patches allow prolonged growth or an additional generation in the leaf-miner moth and thus enhance its fitness (Kaiser et al., 2010). Currently the mechanism of how Wolbachia manipulates plant physiology during the plant–herbivore interaction is not known.

Endosymbionts have been shown to facilitate the transmission of plant pathogens by herbivorous insects. The whitefly Bemisia tabaci is a major pest species on a large diversity of economically important plants. Most of the damage inflicted to plants by this homopteran is due to the transmission of a large variety of plant viruses (De Barro et al., 2011). Virus particles can establish chronic infections in the whiteflies by binding to the bacterial chaperone groEL circulating in the haemolymph (Gottlieb et al., 2010). The virions accumulate in the salivary glands of the whitefly and are released from there during feeding (Gottlieb et al., 2010).

Invasiveness

Endosymbionts may play an important role in facilitation of invasions of insects. However, to date studies linking endosymbiont-mediated traits to invasive potential are surprisingly scarce, given that a large number of phytophagous species – including invasive species – harbour endosymbionts (Douglas, 2009). When different insect species or different populations of an invasive species stemming from different source populations and harbouring different endosymbionts come into contact, the opportunity for horizontal transmission of symbionts arises. The acquisition of an endosymbiont from the environment results in novel genotype × genotype interactions (host–symbiont recombination) and may result in ad hoc acquisition of new traits, including enhanced stress tolerance and resistance to parasites. This in turn may enhance the invasive potential of species.

A prominent example is the whitefly B. tabaci. This species comprises several biotypes that differ strongly in host range and invasiveness. Only few biotypes are responsible for worldwide invasions, specifically biotypes Q and B. The biotypes can not only be identified by mitochondrial DNA but also show biotype-specific associations with their endosymbionts (Gueguen et al., 2010). Hybridisation between the invasive B biotype and an indigenous biotype on La Réunion revealed that hybridisation among the different whitefly populations is possible but occurs asymmetrically (Thierry et al., 2011). The endosymbiont community was largely transmitted maternally as expected, however, hybrids carrying Arsenophonus were found significantly less often than expected, pointing either towards loss of this endosymbiont in viable hybrids or lower viability of hybrids carrying it. Associations of symbiotype and certain nuclear host alleles were maintained in hybrids, suggesting genotype × genotype interactions between host and symbionts (Thierry et al., 2011). Thus, endosymbionts could favour the spread of invasive host lineages that they are associated with in order to benefit their own reproduction.

Immunity and resistance towards pathogens and parasites

Variation in resistance towards pathogens and parasites has been shown to be associated with the presence/absence of secondary symbionts in a number of insects. Currently most studies have been correlational; however, an understanding of the mechanisms of how enhanced resistance is mediated by endosymbionts is steadily increasing.

Wolbachia, the widespread secondary endosymbiont of diverse arthropods, is usually considered a reproductive parasite (Hilgenboecker et al., 2008; Werren et al., 2008). As a number of studies failed to show negative effects on host reproductive biology (e.g. Reynolds & Hoffmann, 2002; Bouwma & Shoemaker, 2011), it was suggested that the high prevalence of Wolbachia in such insect species may be explained by direct fitness benefits conferred to the host by the endosymbiont. Recently Wolbachia has been shown to confer enhanced resistance towards a variety of RNA viruses in Drosophila (Hedges et al., 2008; Teixeira et al., 2008; Glaser & Meola, 2010) and other dipterans that are important vectors for viral diseases such as Culex quinquefasciatus or Aedes aegypti (Moreira et al., 2009; Glaser & Meola, 2010). Either survival of the flies was significantly enhanced when infected with Wolbachia (Hedges et al., 2008) or viral load was significantly reduced (Teixeira et al., 2008; Glaser & Meola, 2010), which may be of importance in controlling vector-transmitted viral diseases such as the West Nile Virus or Dengue. The enhanced resistance mediated by Wolbachia may result from the host's innate immune system being primed by the endosymbiont or alternatively from competition for key host cell components required by both Wolbachia and a virus (Moreira et al., 2009).

Spiroplasma, another male-killing bacterium widespread in insect hosts (Duron et al., 2008), may turn into a mutualist by conferring protection against nematodes (Jaenike et al., 2010) or parasitic wasps (Xie et al., 2010).

The most studied case of symbiont-mediated resistance towards pathogens as well as parasites are aphids that harbour a suite of secondary endosymbionts. Alongside Buchnera, the primary endosymbiont that plays a role in nutritional upgrading, seven different secondary endosymbionts have been identified that occur regularly in aphid hosts, namely the Enterobacteriacae Hamiltonella defensa, Regiella insecticola, and Serratia symbiotica as well as Rickettsia, Rickettsiella, Spiroplasma, and Arsenophonus (Oliver et al., 2010; Tsuchida et al., 2010). Hamiltonella defensa and to a lesser extent S. symbiotica and R. insecticola have been shown to enhance resistance towards endoparasitic wasps in several aphid species (Oliver et al., 2003, 2005, 2010; von Burg et al., 2008; Vorburger et al., 2010). While oviposition rate of the parasitic wasp was not altered by the presence of secondary endosymbionts, rates of successful parasitism were reduced by increased death rate of wasp larvae developing in aphid hosts (Oliver et al., 2003, 2005). Recent studies strongly suggest that the protective effect of H. defensa is dependent on the presence of a bacteriophage within the genome of the symbiont. Different variants of the bacteriophage encode several toxins that are known or suspected to target eukaryotic cells (Oliver et al., 2009, 2010). Secondary symbionts may also enhance resistance towards bacterial and fungal pathogens. The secondary endosymbiont R. insecticola in pea aphids (Scarborough et al., 2005) and, again, Wolbachia in Drosophila have been shown to be effective against entomopathogenic fungi (Scarborough et al., 2005; Panteleev et al., 2007).

Aside from the enhanced resistance towards endoparasites, insect hosts may benefit from the production of toxins or antibiotics by endosymbiotic bacteria to reduce palatability or prevent pathogens entering the insect host. An inherited Pseudomonas symbiont produces the polyketide pederin in several species of rove beetles of the genus Paederus (Kellner, 2003; Piel et al., 2004). While this toxin does not alter the behaviour of insect predators it strongly deters wolf spiders and thus reduces palatability of the beetles as prey (Kellner & Dettner, 1996). Endosymbionts producing antibiotics and thus aiding in pathogen resistance of the hosts are known from digger wasps of the genera Philantus and Trachypus (Kaltenpoth et al., 2005, 2010) as well as fungus-growing insects such as attine ants (Currie et al., 1999) and pine beetles (Scott et al., 2008). In digger wasps, a streptomycete is harboured in antennal segments of female wasps. Prior to oviposition, female wasps secrete actinobacteria from antennal reservoirs to cover the inside of the subterranean brood cells. Subsequently, larvae actively transfer the bacteria onto the surface of their cocoons in order prevent fungal infestation (Kaltenpoth et al., 2005). As the bacteria Candidatus Streptomyces philanthi produce a mixture of nine antibiotic substances they provide protection against a range of opportunistic fungi (Kroiss et al., 2010). Bacterial endosymbionts of fungus-growing beetles produce antimicrobial substances effective against opportunistic fungi that parasitise and overgrow the mutualistic fungus that is farmed as food by the insects (Currie et al., 1999; Scott et al., 2008). At least for attine ants the specificity of antimicrobial substances towards the parasitic fungus is disputed; although the fungus-garden parasite is inhibited, other fungi including the mutualistic fungus are also suppressed (Sen et al., 2009).

Population-level impacts of endosymbionts

Endosymbionts can strongly impact population dynamics and demography as well as the genetic diversity of the host species. Reproductive manipulators such as Wolbachia, Spiroplasma, and Arsenophonus are widespread among arthropods with an estimated 70% of all species being infected (Duron et al., 2008; Hilgenboecker et al., 2008). Such symbionts manipulate host reproduction in order to increase their own vertical transmission via matrilines by two different strategies. Either reproductive output of uninfected females is reduced relative to infected ones due to cytoplasmic incompatibilities (CI), or the sex ratio of offspring is shifted towards females. The latter can be attained by killing male offspring, inducing feminisation in genetic males or thelytokous parthenogenesis (Werren et al., 2008; Engelstädter & Hurst, 2009). Cytoplasmic incompatibilities result in sterile crosses between uninfected females and infected males, leading to a spread of the infection in a population. However, when different strains of CI-inducing bacteria are harboured, both reciprocal crosses will be sterile. Thus, gene flow either between uninfected and infected sub-populations or those infected with different strains is reduced on the scale of a meta-population due to a reduction in effective migration rates. Sex-ratio distorters may decrease genetic diversity within a population and reduce effective population size, entailing strong negative effects such as an increased rate of fixation of deleterious mutations and stronger genetic drift effects. When prevalence of bacteria inducing male-killing and feminisation of genetic males is sufficiently high, only few males will be left to mate with a relatively large number of females. Thus, the effective population size is reduced to the uninfected part of the population (Engelstädter & Hurst, 2007; Ferrari & Vavre, 2011). Due to the high selection pressure exerted by cytoplasmic sex-ratio distorters, nuclear resistance genes are selected to counteract symbiont effects (Ferrari & Vavre, 2011). Resistance alleles may thus spread rapidly in a population. In the butterfly Hypolimnas bolina the selective sweep of a resistance allele has been shown to shift the sex ratio from extremely female biased (100 : 1) to equal numbers of males and females in fewer than 10 generations (Charlat et al., 2007). Thus, the spread of a male-killing bacterium can theoretically drive small populations to extinction.

Parthenogenesis-inducing (PI) bacteria may result in reduction or abandonment of sexual reproduction. As PI bacteria infected host populations consist of clones that selection acts upon, genetic diversity may decline rapidly (Ferrari & Vavre, 2011).

Population dynamics of insect hosts may also be altered by the impact of endosymbionts on traits relevant to dispersal. Pea aphids harbouring the facultative endosymbiont Regiella insecticola produced fewer winged morphs in response to crowding than those lacking this endosymbiont. In addition, in two out of three aphid lineages tested, the timing of sexual reproduction was altered by the presence of Regiella (Leonardo & Mondor, 2006). Thus, this facultative endosymbiont may limit gene flow. In contrast, presence of Rickettsia in the spider Erigone atra enhanced long-distance dispersal via ‘ballooning’, i.e. aerial dispersal (Goodacre et al., 2009).

Adaptive body coloration

As a large variety of predators and parasitoids detect their prey based on visual cues, body colour is an ecologically relevant trait of a prey insect. The pea aphid Acyrthosiphon pisum displays a colour polymorphism, which influences their susceptibility to natural enemies. Ladybird beetles tend to consume red beetles that are easier to detect on green plants, while parasitoid wasps preferentially attack green aphids (Losey et al., 1997). An infection with the facultative endosymbiont Rickettsiella changes the body colour of aphids from red to green by increasing the amounts of blue–green polycyclic quinones within the aphid (Tsuchida et al., 2010). Thus, susceptibility towards ladybird beetles should be lower in these aphids, but higher towards parasitoid wasps. Interestingly, Rickettsiella is often found in concert with Hamiltonella or Serratia, those aphids that confer protection against parasitoid wasps (Oliver et al., 2010; Tsuchida et al., 2010).

Behavioural manipulation – do only parasites do it?

Many parasites are known to modify host behaviour in order to enhance their probability of transmission (Poulin, 2010). It is assumed that the behavioural manipulation is adaptive to the parasite whereas the host may suffer fitness losses. While this phenomenon has drawn much attention from parasitologists, the impact of mutualistic primary and secondary endosymbionts on host behaviour has received relatively little attention. Nonetheless, alterations in behavioural traits by such mutualistic endosymbionts may be common as the examples above on host choice in herbivores or dispersal behaviour show (Ferrari et al., 2007; Goodacre et al., 2009) or increased wandering behaviour of stinkbug nymphs when uptake of symbionts from the egg mass was experimentally hampered after hatching (Hosokawa et al., 2008). Thus, symbiont-mediated alterations in host behaviour may be far more common than previously thought. Similar to comparative studies on interactions of pathogenic and mutualistic bacteria with the host immune system, comparative studies of how symbionts and parasites influence host behaviour mechanistically may yield important insights in the physiological integration of the holobiont.

How large are the effects of symbionts on host phenotype?

A range of beneficial effects has been attributed to the presence of such facultative secondary endosymbionts (see above and Table 1). Generally, effects of facultative endosymbionts on survival or fecundity published to date are often large, i.e. individuals possessing the facultative endosymbiont had on average more than three times as many offspring or survived three times longer (effect size of studies covered in Table 1 mean ± standard deviation = 1.25 ± 0.75). However, this large effect size may be due to a bias in publication of significant effects of endosymbionts on host phenotype, with only few studies reporting non-significant results as well (e.g. Russell & Moran, 2006).

Table 1.  Overview of effect sizes of symbiont-mediated traits on host fitness, measured as production of offspring or survival.
Type of traitHostSymbiontSymbiont typeEffect size*Study
  1. *Effect size was calculated as a log response ratio to quantify the effect of an endosymbiont being present on fitness of the host insect: inline image/inline image, where si is the effect size on fitness of the host insect measured as increased survival or increased number of offspring. Thus, an effect size of 0.69 means double the number of offspring or doubled survival time.

  2. Number of offspring or fitness of offspring measured.

  3. Survival measured within the same generation directly (larval stages or adult insects).

  4. ? Endosymbiont classification ambiguous.

Nutritional upgrading
 Provisioning of amino acids lacking in phloemAcyrthosiphon pisum (Hemiptera)Buchnera aphidicolaObligateRequired for survivalAkman Gündüz and Douglas (2009)
 Provisioning of B vitaminsCimex lectularius (Hemiptera)Wolbachia (supergroup F)Obligate (?)1.25 (eggs), 1.47 (larvae)Hosokawa et al. (2010)
 Provisioning of essential amino acidsCamponotus floridanus (Hymenoptera)Blochmannia floridanusObligate1.22Zientz et al. (2006)
 Nutritional upgrading when reared on low-iron dietDrosophila melanogaster (Diptera)Wolbachia pipientisFacultative0.3Brownlie et al. (2009)
Temperature tolerance
 Reproductive advantage after heat stressAcyrthosiphon pisum (Hemiptera)Serratia symbioticaFacultative2.72Montllor et al. (2002)
 Fecundity after heat shockAcyrthosiphon pisum (Hemiptera)Serratia symbioticaFacultative−0.06 to 0.52Russell and Moran (2006)
 Survival after heat shock (37 °C and 40 °C)Bemisia tabaci (Hemiptera)RickettsiaFacultative0.88 (37 °C)Brumin et al. (2011)
    2.01 (40 °C) 
Defence against pathogens and parasitoids
 Secretion of antimicrobial compoundsPhilanthus triangulum (Hymenoptera)Streptomyces philanthiObligate2.52Kaltenpoth et al. (2005)
 Defence against fungal parasiteAcyrthosiphon pisum (Hemiptera)Regiella insecticolaFacultative1.44Scarborough et al. (2005)
 Defence against parasitoid waspAphis fabae (Hemiptera)Regiella insecticolaFacultative0.33Vorburger et al. (2010)
 Defence against parasitoid waspAcyrthosiphon pisum (Hemiptera)Hamiltonella defensaFacultative0.8Oliver et al. (2005)
 Defence against RNA virusesDrosophila melanogaster (Diptera)Wolbachia (strains wMelCS; wMelPop)Facultative>0.69Hedges et al. (2008) and Teixeira et al. (2008)
 Defence against predatory spiders (rejection as prey)Paederus fuscipes and P. riparius (Coleoptera)Pseudomonas?0.77Kellner and Dettner (1996)
 Defence against parasitic nematodeDrosophila neotestacea (Diptera)SpiroplasmaFacultative2.46Jaenike et al. (2010)
Insect–plant interactions
 Survival correlated with green-island phenotype on senescent leavesPhyllonorycter blancardella (Lepidoptera)Wolbachia (supergroup A)?1.89Kaiser et al. (2010)
 Utilisation of clover as hostAcyrthosiphon pisum (Hemiptera)Regiella insecticolaFacultative1.61Tsuchida et al. (2004)
 Acceptance of clover as hostAcyrthosiphon pisum (Hemiptera)Regiella insecticolaFacultative−1.16Ferrari et al. (2007)
 Hatching rate on new food plantMegacopta cribraria (Hemiptera)Ishikawaella capsulataObligate0.58Hosokawa et al. (2007)
Body coloration
 More greenish body colourAcyrthosiphon pisum (Hemiptera)RickettsiellaFacultativeNot measuredTsuchida et al. (2010)
Life-history traits
 Production of winged dispersal morphs under crowded conditionsAcyrthosiphon pisum (Hemiptera)Regiella insecticolaFacultative−1.5Leonardo and Mondor (2006)

Whereas obligate primary endosymbionts are by definition required for reproduction and growth of the insect host and thus will have a large effect on host fitness, the beneficial effect of secondary endosymbionts may vary quantitatively and in a context-dependent manner. For example, a facultative symbiont mediating resistance towards a particular parasitoid may only be beneficial to the host when the parasitoid is present in the population but may otherwise incur maintenance costs only. As a result, such facultative symbionts may be lost from the host population (Oliver et al., 2008). Only a few studies have shown trade-offs, i.e. decreased fitness of hosts when carrying facultative secondary endosymbionts under some circumstances (Russell & Moran, 2006), while negative effects were not detected in some studies (Russell & Moran, 2006; Oliver et al., 2008; Himler et al., 2011). Thus, when facultative endosymbionts provide a large fitness benefit at no or relatively little cost, these symbionts can spread quickly within the host population (Oliver et al., 2008; Jaenike et al., 2010; Himler et al., 2011). With more population-level studies on the prevalence of bacterial symbionts in relation to ecological traits more subtle effects may be uncovered in the future.

Concluding remarks

The rising awareness of the important roles that endosymbionts play in host ecology has led to a steep increase in the identification of ecologically important traits being attributed to facultative secondary endosymbionts. Metagenomic approaches are becoming more feasible in community ecology and population ecology studies, allowing the identification of widespread endosymbionts as well as an estimation of their prevalence in an insect population. Due to these technical advances it is now becoming feasible to study the impact of relatively little-known but seemingly widespread symbionts on insects such as yeasts and other fungi (Gibson & Hunter, 2010).

In the past, bacteria without obvious effects on host fitness but commonly associated with an insect host species were quickly labelled as commensals. However, the increasingly broad range of phenotypic effects mediated by endosymbionts rather suggests that many of the insect-associated symbionts may influence host ecology in some way, often in a condition-dependent manner. The good news for field ecologists is that it will be up to them to test those potential effects in the natural environment.

Acknowledgements

I am grateful to Rebecca Schulte as well as four anonymous reviewers for helpful comments on the manuscript.

Ancillary