A biogeographical and landscape perspective on within-host infection dynamics

There is a growing appreciation in general population and community ecology of the critical importance of space and spatial phenomena for understanding a wide range of ecological phenomena. At the scale of an invading pathogenic organism, a host organism in some important respects is like a large, complex landscape, on which the infection may play out in space as well as time. Many of the topics of contemporary spatial ecology — invasion dynamics, pattern generation, persistence mechanisms, source-sink dynamics, and spatial subsidies — have analogues in within-host infection processes. Introduction: The emerging discipline of spatial ecology From the perspective of an invading microbe, a vertebrate host individual is in effect a landscape or even continent with considerable internal heterogeneity in resource availability and mortality risks [14]. A valuable and largely unexplored apparatus for looking at withinhost infection dynamics is the conceptual lens of modern spatial ecology. Because this paper is meant to be a thought piece, rather than a thorough review, references to the empirical microbial literature are drawn from a small set of authoritative medical microbiology textbooks. An increasingly central theme in ecological research is the need to consider dynamical processes in a spatial context [16]. This realization has sparked several active areas of inquiry, including: (1) Metapopulation biology, which focuses on species which occupy habitats that are patchy and discontinuously distributed [3], e.g., butterflies occupying discrete meadows separated by forest [2]); (2) Landscape ecology, which examines the consequences and causes of spatial heterogeneity among habitat patches, and in particular is concerned with how the explicit spatial configuration of habitats influences ecological processes [16]; (3) Analytical biogeography, which takes a strongly quantitative approach to issues such as spatial patterns in species richness at very large scales [5,10], patterns in species’ distributions [1] and invasion dynamics [13]. Many processes studied by spatial ecologists have clear analogues in within-host infections. There are of course major differences which may limit the utility of such analogies. For instance, patterns of connectivity among sites in a host are considerably more complex (including diffusion (e.g., on the skin), advective flows on fluid-covered surfaces, and rapid flows in the lymphatic system or bloodstream) and potentially more rapid than in terrestrial biomes,. Moreover, unlike a geographical landscape, the host body The Ecology of Infectious Diseases has been sculpted by selection over aeons of evolutionary history to prevent or keep in check infection (defined here as successful pathogenic invasion). Vertebrate hosts mount an astonishing array of constitutive and induced defenses against invasion [7]. Nonetheless, these complications merely ornament spatial processes, essentially the same as describing the dynamics of any species in a landscape. I propose here that recent advances in spatial ecology can provide fresh perspectives on familiar facts in infectious disease systems. The biogeography of individual hosts A biogeographer grapples with two principal issues regarding the biota of, say, Canada. One is at the level of single species and the other at the community level: (1) Some species have widespread ranges, and others are spatially restricted. The factors that explain such differences among species include differing degrees of ecological specialization, limitations on dispersal, and antagonistic species interactions. (2) Geographical patterns exist in species richness, as well as differences in richness among habitats. Determinants of richness patterns include both intrinsic or local factors (e.g., productivity, severity and frequency of disturbances) and extrinsic factors (e.g., coupling to external source pools) [10]. Typically, habitat area and distance to sources of colonization explain much variation in species richness. There are clear analogues for these broad biogeographical themes in microbial communities within individual hosts. Habitat range and specialization Microbes must contend with considerable within-host spatial and temporal heterogeneity in resource availability, abiotic conditions, mortality factors, permeability, and forces of advection [7,p. 293, 15]. Many microbes exhibit moderate to precise specialization to particular tissues and organ systems and along gradients in resources and abiotic conditions such as temperature. Just to cite an illustrative example, the agent of whooping cough, Bordetella pertusis, infects epithelial cells of bronchioles; this species is said to be “nutritionally fastidious” and very sensitive to fluctuations in physical/chemical factors [11, p. 291]. A large tome remains to be written, taking existing information on tissues and condition specialization by microbes and interpreting this information in terms of habitat specialization, as measured against the templet of available internal heterogeneities and gradients in the host body in resources, abiotic conditions, and defenses. Within-host patterns of species richness in microbial communities There is substantial spatial variation in microbial species richness within hosts. In a healthy body, blood, body fluids and deep tissues (e.g., the lungs, [7, p. 17] are normally sterile [12, p. 140]. Most commensal microbes are found on the skin, or within the lower intestinal tract or outer respiratory tract [11, p. 17]. Likewise, most infectious microbes are either restricted to the respiratory or intestinal tracts [7, p. 9]. Along the alimentary tract, there is a “U-shaped” pattern in species richness, with many species in the oral cavity and colon, and almost none in the stomach [12, p. 142]. As in macroscopic biogeography, patterns of within-host microbial species’ richness should reflect the interplay of local (within-host) properties and processes operating at larger spatial scales. In large-scale biogeography one first attempts to characterize the nature of the species pool available for potential colonization. The properties of the pool The Ecology of Infectious Diseases are determined by processes such as speciation, colonization, and extinction operating at larger spatial and temporal scales than defined by just the local community at hand. What determines the ‘pool’ of microbial species that are potentially available for infection at particular sites in an individual host? Several authors (e.g., [8] have commented on the close parallel between metapopulation models and epidemiological models. We can draw on this parallel to examine the effects of spatial constraints on species richness. Consider a microbe specialized to a single tissue type (=habitat) which causes an infection invoking a defensive host response, but without permanent host immunity. We assume each individual host is a ‘patch’, and total host numbers are fixed. Let p be the fraction of hosts infected. A general expression for disease dynamics in the host population is dp/dt = [net infection rate] [net rate of infection die-out]. Let k be the fraction of hosts potentially available for successful infection out of the entire host population, and e the per host rate of infection dieout. The net infection die-out rate is thus ep. The parameter c will scale the infection rate of the pathogen (a rate of successful colonizations of available, susceptible hosts by pathogens). Consider two distinct colonization scenarios, one for pathogens that are specialists on a focal host, and the other for generalist pathogens sustained by multiple host species which incidentally infect the focal host species. For specialist microbes, persistence depends upon recurrent infection within the focal host population. A simple model for the rate of infection by a specialist microbe is cp( k p) . For a generalist microbe, maintained in an external reservoir with no cross-infection among individuals of the focal host species, an alternative expression for net infection is c( k p). Given a constant per host rate of loss of infection e from infected hosts (akin to local extinction), the equilibrial level of infection in the focal host is predicted to be: p* = k e/c (specialist); p* = kc/(e+c) (generalist). For the specialist pathogen, there are threshold values of the parameters, defining when the microbe will persist: if k < e/c, the specialist pathogen cannot persist in the focal host population. There are no such thresholds for the generalist. These simple metapopulation models suggest several qualitative conclusions. First, pathogens which are habitat specialists should be most common in habitats (e.g., tissue types) that are most common in the host population (high k), or most easily colonized (high c), or experience low rates of die-out due to host defenses (low e). Second, sparse or inaccessible habitats (low k) should be dominated by species that mainly utilize other, more accessible habitats. Third, specialists on rare or inaccessible habitats should have unusually low extinction rates (avoid host defenses), or utilize specialized dispersal modes, relative to the entire ensemble of microbial species on the host. The natural history of infectious diseases and the normal microflora seems consistent with these theoretical expectations. Some tissues (=habitats) have few specialist pathogens. Two quotes provide examples: (i). “Specific infections of skeletal muscle are obscure” [11, p. 786]; (ii), “Infections of the central nervous system are relatively infrequent” [11., p. 716]. Both tissue types are likely to have low c, compared to say epithelial surfaces. Internal tissues in most hosts are unavailable for infection, but are accessible in a few hosts (e.g., due to wounds); such tissues in effect also have a low k. The model predicts that relatively few microbes will be specialized to such tissues; those microbes which do manage to infect there occasionally should also be able to utilize other tissues, or be sustained in reservoirs outside the host population. The Ecology of Infectious Diseases The abstract parameters of c and e determining persistence of an infection in a host population as a whole reflect the details of within-host infection dynamics, including factors such as resource availability and within-host spatial dynamics. The typical stages of infection in a host individual are: 1. Initial entry, 2.Establishment, 3. Spread, 4. Stabilization, 5. Decline, 6. Elimination, or reduction to a persistent chronic level of infection in the host. Parallels with phenomena in spatial ecology arise in each of these stages; here I focus on establishment.