The positive correlation between the abundance and prevalence of a parasite (Krasnov et al., 2002) is a manifestation of the general abundance–occupancy relationship of free-living organisms (Gaston, 1999, 2003). There are three common hypotheses to explain this relationship. Firstly, the sampling artefact hypothesis assumes that a positive correlation between abundance and distribution will emerge from a random process in which abundant species have a higher probability of being sampled in occupied sites. Secondly, the core–satellite hypothesis predicts that core species (which are locally abundant and have a wide regional distribution) have a high probability of producing emigrants, whereas satellite species (which have a low abundance and a narrow regional distribution) disperse less successfully. Finally, the ecological specialization hypothesis (host specificity for parasites) predicts that species that exploit a wider resource range are more widespread and more abundant (Hanski et al., 1993). Our results for fleas and sucking lice tend to support the ecological specialization hypothesis. Fleas exploit a wider host range and are more widespread and prevalent than lice. However, the mean abundance of fleas is lower than that of lice. Parasites may possess different adaptations to exploit host species and these may involve some degree of trade-off between parasitized host species and host individuals (Krasnov et al., 2004). A parasite might evade its host immune response because the relevant genes in the parasite (such as in the sucking louse) might evolve two to three times faster than those in the host (Woolhouse et al., 2001; Light & Hafener, 2007; Light & Reed, 2009). Moreover, Khokhlova (2004) found the host immune response to differ for various parasites: for example, phagocytes of Gerbillus dasyurus (which is parasitized by several flea species) are significantly more active than those of Gerbillus andersoni allenbyi (parasitized by a single flea species).
Although both fleas and sucking lice demonstrated an aggregative pattern of distribution, different slope values for Taylor's relationship in the two taxa of insects indicate different degrees of aggregation. Parasites must compromise between being too aggregated and being too randomly distributed (Shaw & Dobson, 1996; Krasnov et al., 2004; Krasnov, 2008). Consequently, parasite distribution may be optimized in relation to the life history characteristics and ecology of the parasite. However, sucking lice showed a higher degree of aggregation than fleas. Fleas and lice share some adaptations in terms of how they anchor themselves to the skin of a host and in their haematophagy, which may play a role in the similarity of patterns observed amongst parasites (i.e. aggregated distribution), but the cause of the differences in aggregation between fleas and sucking lice is still not understood. Sucking lice (which reproduce on the host) demonstrated a higher degree of aggregation than fleas (which reproduce off the host). Hence, the direct reproduction of a parasite on or in a host individual is one of the main parasite-associated reasons for aggregation (Poulin, 2007). Furthermore, most species of sucking louse usually have narrow host ranges and choose a few mammalian species as their main hosts [e.g. Sathrax durus (Phthiraptera: Polyplacidae) parasitizes a single host, Tupaia belangeri]. Different sucking louse species show dramatic niche divergence in host selection, as evidenced by the analysis of ecological niches, niche overlap, numerical classification of community and correspondence analysis, which suggests a high rate of co-evolution between sucking lice and their mammalian hosts (Meng et al., 2007, 2008; Zuo et al., 2010). In particular, the mitochondrial genome of the human body louse, P. humanus, is fragmented into 18 minichromosomes, suggesting that multiple minichromosomes may have evolved along with blood-feeding (Rand, 2009; Shao et al., 2009). Khokhlova (2004, 2008) measured the feeding and reproduction of fleas in host-specific Parapulex chephrenis (Siphonaptera: Pulicidae) (typical host: Cairo spiny mouse, Acomys cahirinus) and host-opportunistic Xenopsylla ramesis (Siphonaptera: Pulicidae) (typical host: Wagner's gerbil, Dipodillus dasyurus). Higher feeding success, egg production and latency of oviposition were found in fleas feeding on a typical host than in those feeding on an atypical host, which suggests that the response of a parasite to the acquired immunity of a host may depend on the host species and host specificity. In addition, the slope of Taylor's power relationship decreased significantly as the number of hosts exploited by a flea species rose, which indicates that highly host-specific parasites tend to have higher degrees of aggregation (Krasnov et al., 2005b, 2006). However, the distribution of a host-generalist flea parasite of the host genus Rattus, such as L. segnis, which reproduces off the host, was also highly aggregated, suggesting that aggregation may occur for not only mechanistic reasons and that some other factors may also be responsible for species-specific limits of aggregation. Our study proposes the existence of potential co-evolutionary processes which contribute towards determining the distribution of ectoparasites on rodents.
A simple epidemiological model incorporating aggregation with mean abundance and variance of abundance can be accurately applied to a wide range of parasite–host relationships. Such information may be critical to formulating recommendations for the prevention and control of flea-borne and louse-borne diseases in this plague-endemic region.