by
|
by |
|
| This article will appear in a forthcoming issue of Trends in Parasitology. | |
Abstract
To understand the risk of exposure to pathogens has been the major goal of epidemiologists since Snow discovered the Broad Street pump (famously the source of a cholera epidemic in 19th century London) [1]. In recent years, advances in genetics and immunology have thrown the epidemiological spotlight on individual variation in susceptibility to infection. Now advances in molecular biology should open up the study of individual variation in exposure to vector-borne disease. The current model paradigm for vector-borne diseases has remained largely unchanged since the 1950s [2]. This model recognizes that some host species are bitten more than others, but averages the exposure to the susceptible host species by assuming that the vector population is distributed evenly over all individuals. However, to stop research at the population level when investigating exposure to vector-borne disease is akin to abandoning research on the immunology of Plasmodium falciparum, having learned only that humans succumb and not other animals.
Patterns and Preferences
The pattern of contact between blood-sucking insects and their host animals is extremely heterogeneous and non-random. While most hosts are bitten relatively infrequently, a subset will consistently be heavily attacked [3]. Intriguingly, the attraction of blood-sucking insects to humans and other animals has been reported to vary with epidemiologically interesting demographic parameters, including pregnancy [4], age, body size, sex, disease status, and even skin color and blood type [5]. Quantifying the effect of this "patchiness" by using empirical data on the bite distribution of vectors of malaria and leishmaniasis, Woolhouse et al. [6] estimated that, on average, 20% of any host population contributes 80% of the net transmission potential. Put another way, the heterogeneous distribution of vectors tends to create extreme transmission "hot spots" and "cold spots." If the extent and intensity of these hot spots is predictable, then vector control, vaccination, or chemoprophylaxis can be targeted, with resultant cost savings and increased sustainability.
The plethora of host-derived chemical and visual cues proximately responsible for host choice are far too complex and idiosyncratic to be usefully modeled. For example, the tsetse Glossina morsitans morsitans responds to more than ten different host odors or kairomones [7]; Glossinidae also have strong visual senses, hence shape, color, contrast, and patterning are important determinants in attracting G. m. morsitans across the savannah, to hosts and artificial targets. Exhaustive research to describe these attractive cues has yielded baits that were considerably successful in trapping G. m. morsitans from areas of southern Africa. However, other Glossina spp., such as Glossina fuscipes, which live in riverine gallery forests and do not feed habitually on ungulates, respond very poorly to such traps. If we attempted to make quantitative predictions of host choice by G. m. morsitans based on this complex array of attractive cues, the compounding of errors associated with each parameter will make the model output meaningless, and the model is unlikely to be applicable to other tsetse species, let alone other groups of blood-sucking insects. By contrast, the ultimate, evolutionary determinants of host choice could provide the common currency needed to produce a simple, yet general, ecological model of host choice. To do so, we need to understand what it is that insects seek from their hosts that leads them to prefer pregnant women or malarious mice.
Following Feeding Success
Female insects take a bloodmeal primarily to acquire protein for egg production, and the quantity rather than quality of the blood is the major limit to reproductive output. The amino acid composition of a bloodmeal can affect egg production by as much as 20% (8), but the probability of obtaining a bloodmeal and the quantity of blood obtained can vary by an order of magnitude depending on the host animal bitten [9-12] (figure 1a). Perhaps more importantly, the probability of surviving the feeding event varies directly with feeding success: a re-analysis of the data from Webber and Edman [11] indicated that the probability of vector survival falls linearly from 1.0 to 0.5 when feeding success falls to zero (figure 1a).
Such variation in feeding success (and survival) between hosts appears to be driven universally by host defensive behavior. Studies with sandflies [12], mosquitoes [13], horseflies [14], tsetse [15], and reduviid bugs [16] all show that the more defensive the host, the higher the probability that a biting insect will be interrupted before the insect has fed, or fed to repletion. Furthermore, with only one exception [17], feeding success varies as a function of biting intensity: the greater the density of insects biting an individual host per unit of time, the more defensive the host becomes and the lower the per capita feeding success. Again, over the natural range in vector-biting density, feeding success varies by an order of magnitude [10,12] (figure 1b).
In summary, density-dependent host defensiveness appears to be the major host-related determinant of lifetime reproductive success for most blood-sucking insects, affecting the quantity of blood ingested and the probability that the vector lives to feed again; hence there must be intense pressure for vectors to evolve strategies to discriminate and feed on the least defensive hosts.
Ideal Free Distribution
How might insects evolve to minimize their encounters with host defensive behavior? Clearly, insects should evolve strategies to avoid a more defensive host (host X) in favor of a less defensive host (host Y). However, there will be a point where the evolved host choice strategy results in a distribution of vectors between hosts X and Y, such that the defensiveness of host X is equal to that of host Y because the defensiveness of hosts is density-dependent. This is known as the ideal free distribution (IFD) [18], at which point evolution should "cease." The probability of surviving and successfully feeding is the same for all flies, whichever host an individual fly feeds upon, and any new host-seeking strategy which led an individual fly to "move" hosts would cause a decrease in the particular fly's feeding success and could not, therefore, invade.
Preliminary models of the IFD applied to blood-sucking insects have recently been published. These models capture, a priori, from evolutionary theory, two desirable and intuitively plausible behaviors: the distribution of vectors over hosts is heterogeneous both between and within host species; and the intensity of heterogeneity can vary with host and vector abundance [19]. The concept of relative defensiveness, and therefore relative attraction, is an important one. Most host animals encountered by biting flies are in aggregations, such as humans in houses, domestic animals in pens, birds in roosts, or game animals in herds. These feeding opportunities permit vectors to choose readily between hosts, and small differences in host suitability can, therefore, drive large differences in effective host choice (box 1).
The rate of density-dependent defensiveness varies between host species [10,19], and IFD models can be structured accordingly. The evolutionary premise of the IFD predicts that the wide range of demographic groups that are known to be more attractive (e.g. age, sex, disease status) should also be the least defensive. If true, then it should be possible to incorporate these parameters into the IFD models to generate predictions about biting rate on these separate demographic groups as they vary with vector density and the relative defensiveness of the alternative host groups present. Recently, evidence has begun to emerge that variation in the attractiveness of individual hosts is indeed correlated with defensive behavior [17]. To assay individual preference in choice tests accurately, it has been necessary to inject experimental host animals with different trace metal markers [20], with obvious limitations. In the past few years, however, molecular techniques have been developed which allow identification of the individual host fed upon by the vector [21]. The first study applying this technique to tsetse flies on oxen confirms that the less defensive the host, the more the host is bitten, and conveniently, immature cattle are the most defensive and least attractive hosts [22]. Finally, these new molecular tools should permit fascinating insights into how blood-sucking insects assess the defensiveness of their potential hosts (box 2).
Conclusion and Perspectives
Three strands of theory and empirical study contribute to the question, "why are some people bitten more than others." At a proximate level, some people provide more attractive cues either directly, or indirectly, by supporting larger populations of signaling insects. At an evolutionary level, flies preferentially bite particular individuals because the host cues signal lower defensiveness of those individuals. Finally, at a functional level, attractiveness and defensiveness are relative, and the extreme skew in biting rate, for example, towards you and away from your partner, when sharing a leaky hotel room, can come down to very small differences in these attributes. If the new techniques for bloodmeal typing continue to reveal convenient demographic markers that predict even a fraction of the 20% of hosts contributing to 80% of the disease, then targeted reduction in exposure will play an important role in sustainable and cost-effective vector-borne disease control.
Infochemicals in Mosquito Host Selection: Human Skin Microflora and Plasmodium Parasites - introduces the role of human microflora in the process of host selection. From Parasitology Today, 1999, 15:10:409-413.
Olfactory Cues in Mosquito Host Location - reviews research showing that preferential biting does occur.
Sensitivities of Antennal Olfactory Neurons of the Malaria Mosquito, Anopheles gambiae, to Carboxylic Acids - examines the sensitivity of mosquitoes to human body odors. From the Journal of Insect Physiology, 45:365-373.
Insect Olfaction - lecture notes from an Insect Physiology course taught by David Stanley, Department of Entomology, University of Nebraska.
Mosquitoes and Mosquito Repellents: A Clinician's Guide - a review from the June 1, 1998, issue of Annals of Internal Medicine on preventing bites, as well as good background information.
Mosquito Biology - offers general mosquito biology, habitats, life cycle, and anatomy. Includes videos.
Entomology on World-Wide Web - offers an alphabetized list of links.
Insects on the WWW - links to dozens of insect-related sites, listed according to topic.
Insect Physiology Online - links to abstracts, articles, teaching resources, and more.
Related HMS Beagle articles: