Data accessibility statement
This paper contains no data.
Abstract
Environment drives the host movements that shape pathogen transmission
through three mediating processes: host density, host mobility, and
contact. These processes combine with pathogen life-history to give rise
to an “epidemiological landscape” that determines spatial patterns of
pathogen transmission. Yet despite its central role in transmission,
strategies for predicting the epidemiological landscape from real-world
data remain limited. Here, we develop the epidemiological landscape as
an interface between movement ecology and spatial epidemiology. We
propose a movement-pathogen pace-of-life heuristic for prioritizing the
landscape’s central processes by positing that spatial dynamics for fast
pace-of-life pathogens are best-approximated by the spatial ecology of
host contacts; spatial dynamics for slower pace-of-life pathogens are
best approximated by host densities; and spatial dynamics for pathogens
with environmental reservoirs reflect a convolution of those densities
with the spatial configuration of environmental reservoir sites. We then
identify mechanisms that underpin the epidemiological landscape and
match each mechanism to emerging tools from movement ecology. Finally,
we outline workflows for describing the epidemiological landscape and
using it to predict subsequent patterns of pathogen transmission. Our
framework links transmission to environmental context, providing a
scaffold for mechanistically understanding how the environment can
generate and shift existing patterns in spatial epidemiology.
Introduction
Environment shapes the spatial, temporal, and behavioral structure of
host movements and pathogen persistence, leading transmission to occur
in some locations but not in others. The particular role the environment
plays depends on the movement ecology of the host and pathogen: wildlife
hosts can shed pathogens as they track resources across the landscape;
livestock pathogens can jump abruptly between habitat patches as hosts
are transported from farm to feedlot; and transmission of human
pathogens often shows spatial signatures associated with work,
caregiving, and social engagement. Environment also structures disease
management efforts targeting specific locations that could give rise to
super-spreading events or transmission hotspots, curtail the spread of
invading epidemic “waves”, or capitalize on predictable fluctuations
in local transmission patterns.
However, despite these strong implicit connections, our understanding of
the environment-movement-transmission interface remains in its infancy.
Although movement ecology offers an array of tools that could help
integrate environment, host movement, and pathogen deposition, the
tools’ outputs are somewhat detached from critical aspects of spatial
transmission and some important components of spatial transmission are
largely overlooked. Moreover, it is unclear how to best integrate the
various processes each tool infers for optimal epidemiological
predictions. Quantifying the environment’s role in spatial dynamics of
transmission, appropriately combining constituent processes, and
identifying relevant mechanisms will require a stronger integration of
movement ecology into conventional epidemiological frameworks.
There is a reason why movement and disease ecology developed separately
to this point: they are interested in fundamentally different
relationships. Movement ecology usually explores how individuals
interact with their environments (Nathan et al. 2008), whereas disease
ecology focuses on how individuals interact with one another to allow
for pathogen transmission. Scaling individual-environment interactions
up to predict individual-individual interactions is a long-standing
challenge, and one that potentially distracts from other
transmission-related insights that movement ecology has to offer. That
lacuna is symptomatic of a broader interdisciplinary divide: the spatial
modeling community lacks a systematic way to incorporate mechanistic and
empirical insights from movement ecology into spatial models of
transmission while retaining tractability and transportability beyond
the specific system or landscape at hand. This gap impedes our ability
to identify general principles about how the environment – both the
local habitat and its broader scale landscape structure — shapes
spatial patterns of pathogen transmission in real-world environments.
Spatial transmission dynamics depend on three central processes: host
density processes that describe where hosts are located across the
landscape, mobility processes that describe when and how hosts move
among sites, and contact processes that describe the frequency, duration
and form of host-host interactions and how those interactions relate to
rates of pathogen transmission. Together, the host density, mobility,
and contact processes combine with pathogen life-history to form an
epidemiological landscape that interacts with the local volume of
infected hosts to determine epidemiological dynamics through space and
time (Fig. 1). Because it does not incorporate the volume of infected
individuals directly, the epidemiological landscape exists apart from
any specific disease event. Instead, it summarizes the potential for
particular spatial transmission patterns to arise in a given environment
on the basis of local host density, mobility, and contact. Identifying
how density, mobility, and contact relate to the environment, however,
is an open challenge that movement ecology can help to address.
Here, we propose a strategy to strengthen the interface between
environment and transmission by focusing explicitly on the mechanisms
and traits surrounding the epidemiological landscape. First, we use the
epidemiological landscape to formalize the mechanistic relations that
underpin transportable models. We then propose an heuristic that we
refer to as the movement-pathogen pace-of-life hypothesis that suggests
how to prioritize the density, mobility, and contact processes according
to important pathogen life history attributes. Next, we identify a set
of critical mechanisms determining the epidemiological landscape’s
structure, and match both the epidemiological landscape’s central
processes and our proposed mechanisms to metrics and inferential tools
from movement ecology in sections organized around two research
priorities (Fig. 2). The first priority is to infer specific aspects of
pathogen transmission using existing movement ecology tools; the second
is to determine how social context contributes to patterns of density,
mobility, and contact. Finally, we outline three workflows that blend
mechanisms with movement and transmission data to generate
epidemiological forecasts that are transportable across landscapes and
populations (Fig. 3), and end by identifying existing challenges and
open questions.