Anelosimus (Agnarsson, 2004)

Agnarsson, Ingi, 2006, A revision of the New World eximius lineage of Anelosimus (Araneae, Theridiidae) and a phylogenetic analysis using worldwide exemplars, Zoological Journal of the Linnean Society 146 (4), pp. 453-593 : 472-473

ANELOSIMUS

Strikingly, the preferred phylogeny implies that sociality evolved independently within five Anelosimus species groups, and apparently twice within the studiosus group for a total of six origins within the genus ( Fig. 61 ). Permanent sociality therefore originates at least 7–8 times in the lost colulus clade (see Agnarsson, 2004). This relatively small clade (∼1200 species, about 3% of spider diversity, calculated from Platnick, 2006) thus accounts for about half of the origins of spi- der sociality. The clustering of 7+ origins of sociality in the lost colulus clade suggests a common cause, perhaps preadaptations to sociality that are unique to this group of spiders (see Agnarsson, 2002, 2004). One such modification might be the prolongation of cohabitation by juveniles due to maternal care (e.g. Shear, 1970; Kullmann, 1972; Brach, 1975, 1977; Burgess, 1978; Krafft, 1979; Cangialosi & Uetz, 1987; Avilés, 1997; Jones & Parker, 2002; Schneider, 2002; Agnarsson, 2002). The origin of sociality from maternal care and intermediate subsociality seems strongly corroborated ( Fig. 61 , see also Agnarsson, 2004). A threedimensional web has also been frequently considered as a preadaptation for sociality (e.g. Shear, 1970; Krafft, 1979, 1982; Buskirk, 1981; D’Andrea, 1987; Cangialosi & Uetz, 1987; Avilés, 1997). On the preferred phylogeny sociality is indeed concentrated where maternal care and a three-dimensional web overlap (see Agnarsson, 2004). Interestingly, corroborating data come from the other group of spiders with multiple origins of sociality (the distantly related, non-orbicularian Stegodyphus , see Kraus & Kraus, 1988, 1990) where maternal care and threedimensional webs also overlap. However, no doubt additional, as yet undiscovered, features have facilitated the evolution of sociality in both Stegodyphus and the lost colulus clade.

Although sociality evolves repeatedly from subsociality, thereafter social lineages apparently fail to diversify: all social clades are small (one or two species, see Fig. 61 ). The pattern repeats: social species succeed in the short term (ecological time) but fail to diversify in the long term (over evolutionary time). Sociality seems to be responsible for a dramatic shift in population structure from outbred panmictic to strongly inbred, subdivided populations. Social individuals are thus comparatively homozygous and intra- populational variation is slight compared with subsocial or solitary individuals or populations ( Smith, 1986, 1987). Group living and co-operation benefit social species in diverse environments (see Avilés, 1997, for review). On the other hand, low genetic variance can be maladaptive during rapid environmental change. These fleetingly well-adapted populations may be quite vulnerable in environments where change is episodic or fast, or where unexpected challenges arise (disease, parasites, specialized predators, etc.). Subsocial populations may not be able to compete with the more fecund social species, but their survival rate over the long run may be higher. Testing such ideas and investigating the potential disconnection between ecological and evolutionary time scales will require research in several areas. Phylogenies can corroborate or falsify patterns by addition of morphological and molecular data. Population genetics can compare gene trees within and among species to indicate whether slow divergence or frequent turnover of lineages best explains the phylogenetic patterns, and computer modelling can clarify how inbred social genotypes can out-compete outbred subsocial genotypes in the short run, but are unable to track environmental change over evolutionary time.

It should be noted that ‘maternal care’, ‘subsociality’ and ‘sociality’ are not three well-defined categories linked in a simple two-step road to permanent sociality. The categories themselves (derived from Wilson, 1971), although useful as summaries for discussion, are faulty and should be used with care. First, if sociality requires co-operation, only non-territorial web-sharing spiders are truly social in my view ( Agnarsson, 2002, 2004). Second, defining classes of sociality by the duration of cohabitation is ambiguous at best. Third, the categories contain a mosaic of primitive and derived characters and overlap considerably. Sociality requires multiple behavioural mechanisms and the real world presents a continuum rather than discrete classes of sociality; which behaviours precede others phylogenetically thus needs more detailed scrutiny. Nevertheless, social species are generally quite similar. The entire range of social behaviour displayed by all species that leave their natal nest prior to or just after mating is then described as ‘subsocial.’ Agnarsson (2004) remarked that ‘[s]ubsociality is maternal care that spans several, rather than few, juvenile instars. Sociality is another point on the continuum in which maternal care never ceases.’ In such a continuum the transition from ‘maternal care’ to ‘subsociality’ to ‘sociality’ is unclear; two subsocial species may differ more than some subsocial and maternal care species. Other components of social behaviour such as co-operation in web building, attacking prey, regurgitation feeding, intra- and interspecific tolerance mechanisms, sex ratio, communication, brood care, inbreeding, generational overlap and per female reproductive output will enrich the comparison of social behaviour among species. Table 2 attempts to compare the levels of sociality shown by a selection of relatively well-studied species. All the species differ and, interestingly, increasing sociality generally occurs by ‘terminal addition’. In other words derived behaviour B requires A, and C only occurs in the presence of A and B. Egg sac guarding precedes posthatching maternal care, which precedes co-operation in web building and prey capture, and so on. In addition, more social species exhibit greater sex ratio bias (see also Avilés, 1986; Avilés & Maddison, 1991; Rowell & Main, 1992; Avilés et al., 2000) and larger colonies. Such detailed categorizations can therefore not only facilitate the comparison of species, but also suggest testable hypotheses about the evolution of sociality. It should be noted that this is intended as a first attempt to atomize social behaviour into discrete characters. Owing to the preliminary nature of these categories and limited data, the inclusion of these characteristics in the phylogenetic data matrix seems premature; rather, this table may serve as a guideline on possible ways of identifying potentially homologous behavioural units among social theridiids.

The behaviour of species outside the New World is poorly known, but some are also social, including several recently discovered subsocial species from Madagascar ( Agnarsson & Kuntner, 2005), and Malaysia ( Agnarsson & Zhang, 2006). Positive evidence for solitary lifestyles is rarely reported, and even the two European Anelosimus species (no reports exist of social behaviour of these) cannot be presumed to be solitary [maternal care or subsociality seems likely given their phylogenetic position, see Agnarsson (2004), and Fig. 61 ]. The many solitary species placed in Anelosimus by Levi (1956, 1963, 1967) do not belong to this genus and have been transferred (see Agnarsson, 2004, and below).

Figures

Figure 61. The evolution of sociality. Three categories, solitary (thin lines), subsocial (intermediate lines) and social (thick lines) are mapped on the preferred phylogeny. Question marks indicate unknown behaviour; broken lines indicate ambiguous optimization, ‘1,2’ indicate polymorphism. Subsociality evolved at the node leading to Anelosimus; sociality has arisen repeatedly; 7–8 origins are inferred within the genus (depending on interpretation of A. studiosus, which is polymorphic for this character). Sociality is always preceeded by subsociality (subsociality evolves at a more inclusive node), and subsociality in turn is preceeded by maternal care, which optimizes to the node leading to the lost colulus clade (see Agnarsson, 2004).