TrendsNear universal uptake of smartphoneshas moved human mobility researchinto big data.

These data describe human mobilityand behaviour with unprecedentedresolution.

Applying ecological theory provides aframework to inform and predicthuman mobility.

Applications include but are not limitedto epidemics, commuting, and crowdbehaviour[389_TD$DIFF].

In turn, big data on human movementcan provide solutions for urgent con-servation problems and managementchallenges.

1Australian Institute of MarineScience, Indian Ocean MarineResearch Centre (IOMRC), Universityof Western Australia (M470), 35Stirling Highway, Crawley, WA 6009,Australia2King Abdullah University of Scienceand Technology (KAUST), Red SeaResearch Center (RSRC), Biologicaland Environmental Sciences andEngineering (BESE), Thuwal 23955-6900, Saudi Arabia3Department of Epidemiology, HarvardT.H. Chan School of Public Health,Boston, MA, USA4IOMRC and UWA Oceans Institute,The University of Western Australia,School of Animal Biology[386_TD$DIFF], M470, 35Stirling Highway, Crawley, WA 6009,Australia5Department of Biological Sciences,Macquarie University, Sydney, NSW2109, Australia

TREE 2198 1–13

ReviewThe Ecology of HumanMobilityMark G. Meekan,1 Carlos M. Duarte,2 Juan Fernández-Gracia,3

Michele Thums,1,* Ana M.M. Sequeira,4 Rob Harcourt,5 andVíctor M. Eguíluz6

Mobile phones and other geolocated devices have produced unprecedentedvolumes of data on human movement. Analysis of pooled individual humantrajectories using [388_TD$DIFF]big data approaches has revealed a wealth of emergentfeatures that have ecological parallels in animals across a diverse array ofphenomena including commuting, epidemics, the spread of innovations andculture, and collective behaviour. Movement ecology, which explores howanimals cope with and optimize variability in resources, has the potential toprovide a theoretical framework to aid an understanding of humanmobility andits impacts on ecosystems. In turn, big data on human movement can beexplored in the context of animal movement ecology to provide solutions forurgent conservation problems and management challenges.

Human Mobility and Big DataMobility is a central component of human activity, receiving significant allocation of resources,including time, commodities, and energy, with transportation now accounting for 20–25% ofglobal energy use [1]. Development of an understanding of the patterns and drivers of humanmobility is, therefore, a key to optimizing this critical element of human lifestyles in a worldincreasingly facing resource limitation and growing human impacts on the biosphere.

Until relatively recently, studies of human mobility have focused at the extreme ends of thespectrum of the scale of movement patterns. At the smallest scales ( [390_TD$DIFF]metres to kilometres), thefield known as human behavioural ecology [2] has applied models of optimal foraging theoryderived from animal ecology to identify the key factors shaping foraging subsistence behaviourin hunter–gatherer communities [2]. At the largest scales (thousands of [391_TD$DIFF]kilometres), the driversof migrations of humans within and between countries have been analysed in terms ofoptimization (see [392_TD$DIFF]Glossary) of ‘push’ and ‘pull’ factors [3,4]. However, these studies focuson only a very limited part of the spectra of human mobility as it occurs in society today,because the majority of these movements occur in the context of the day-to-day existence ofthe 50% of the world’s population that lives in cities [5]. Empirical studies of human mobility atthese scales were galvanized a decade ago in a seminal paper that based inferences on humanmovement patterns on the analysis of circulation of banknotes in the United States [6] (Figure 1).This contrasts with studies of animal behaviour where technological developments starting withthe ring-banding of birds [7] and culminating with state-of-the-art [393_TD$DIFF]animal borne satellite trackingtags [8] (Figure 1) have allowed accurate and detailed tracking of the individuals of many speciesand have been a major driver of the theory and study of animal movement ecology [7].

From the outset of the study of animal movement it has been deemed ethically acceptable toattach or implant tracking devices to animals to generate detailed data sets of movementpatterns. Until very recently, however, this has not been the case for humans, hence the use of

Trends in Ecology & Evolution, Month Year, Vol. xx, No. yy http://dx.doi.org/10.1016/j.tree.2016.12.006 1© 2016 Elsevier Ltd. All rights reserved.

TREE 2198 1–13

6Instituto de Física Interdisciplinar ySistemas Complejos IFISC (CSIC-UIB), E07122 Palma de Mallorca,Spain

*Correspondence:m.thums@aims.gov.au (M. Thums).

proxies for the tracking of human movement patterns such as banknotes [6] or the distributionof family names [8]. However, the arrival of the geolocated smartphone, effectively a trackingdevice now voluntarily carried by billions of people, has generated massive volumes of geo-referenced movements, which now constitute [394_TD$DIFF]big data (see [99]). The data from these devicesdescribe human mobility and behaviour with unprecedented resolution, and for the first time inhuman history, individual patterns of human movement [9–11] and interactions [12–14] can berevealed at almost every spatial scale with a remarkable degree of detail, immediacy, andprecision (Box 1). With the advent of smartphones and other new technologies such as

110110101100001101101010111110101000000 0110010101011111 0100000000000

Bird marking

Radio-tracking developed

Acous�ctelemetry

Argos satellitenetwork

Mobilephones

satelitelaunched

Small radiotransmi�ers

GPS Tags

Camera tags

Big data

Fastloc GPSand Web2.0

Wearables

19001910

19501960

1970

2000

Smartphones

ICARUS

19901980

2010

PSAT

Proximity tags and Tagswith accelerometers

Time-depthrecorder

1st humanmobility paper

Satellitetransmi�er

First GPS

Figure 1. Timeline of Technological Advances in Animal Movement and Human Mobility. Timeline shows the technological advances from animal movement andhuman mobility research since 1900 to present. [367_TD$DIFF]Pop-up satellite archival tags (PSATs) are data loggers with a means to transmit the collected data via satellitedeveloped for gill-breathing animals that spend little time at the surface. The International Cooperation for Animal Research Using Space Initiative (ICARUS) is a newanimal tracking antenna on the International Space Station that would allow smaller tags to send data back through the low-orbit satellite. The dollar bill represents thefirst published paper on human mobility, which tracked records of dollar bills across the United States as a proxy for human movement [6]. [368_TD$DIFF]Abbreviation: GPS, globalpositioning system.

2 Trends in Ecology & Evolution, Month Year, Vol. xx, No. yy

TREE 2198 1–13

GlossaryAnimal movement ecology:describes the patterns, mechanisms,drivers, and effects of the movementof whole organisms from mammalsto microbes. The movement ecologyparadigm, a related concept [76],characterizes animal movement asdependent on four components:internal state (reason to move),motion (ways to move), navigation(timing and direction of movement),and external factors that can affectthe movement.Big data: it is widely agreed that bigdata share the characteristics of the‘three Vs’: ‘volume’ – data sets arevery large and increasing in volume;‘velocity’ – data arrive in acontinuous stream with everincreasing velocity; and ‘variety’ –data are provided by a wide range ofsources (e.g., sensors, social media,transaction records, video) that isboth structured and unstructured.Collective behaviour: refers to thecoordinated and emergentbehaviours that occur whenindividuals aggregate into groups.Classic examples are the behaviourof fishes in schools or the flight offlocks of birds. Emergent propertiesinvolve the transfer of informationamong group members and groupdecision making so that movementsare synchronized.Data mining: a phenomenologicalapproach to data exploration wherethere are no a priori concepts as towhat would be a ‘useful’ outcomefrom the analysis, such as thesupport or rejection of theoreticalpredictions. It can be both predictive,in which case models used topredict system states are developeddirectly from the data, anddescriptive, in which analysesuncover patterns within large datasetOptimization: in the study of animalmovement ecology, optimization isthe theory that animals, throughnatural selection, seek to maximizebenefits and minimize costs ofdecision making in a landscape ofvariable resources. Two criticalelements of this process include‘currency’ (the factor that the animalmust maximize), which in movementcontexts is typically energy intake,and ‘constraints’, which are thelimitations placed on the currency,for example, search or transit time.

wearable fitness trackers, data on mobility at scales relevant to the day-to-day lives of themajority of human populations exceed by many orders of magnitude the combined total of alldata that are available for any other species on the planet. At the same time, the sheer scalecombined with the difficulties of manipulating and analysing these big data has meant that thestudy of patterns in human movements available from these sources has largely been theprovince of complex systems analysts. Not surprisingly, results have been interpreted using theparadigms, such as broad distributions and complex networks, of a research communitydominated by computer scientists and physicists. For the most part, analyses of these datahave taken a [395_TD$DIFF]‘data mining’ approach [15] and little effort has been invested in studies that arehypothesis-led or based in ecological theory. We argue that ecological approaches couldinform a deeper understanding of the drivers of human mobility given that such data are muchwealthier in motivational content than any data available for non-human subjects. Only humanmobility couples movement patterns to both social media and electronic interactions that canexplicitly link movement to individual motivations and outcomes [14,16]. Together, this canprovide unique insights into the relationships between patterns of human mobility and theimpact of humans on the ecosystems they inhabit and in so doing, offer solutions to thepressing conservation andmanagement challenges that now confront society. Here, we reviewthe most salient studies of human mobility and show how these contribute towards aframework for the ecology of human mobility.

Human Mobility: Insights into Populations from Individual-Based DataThe accessibility and volume of geo-referenced data sets of individual human movements havepropelled the search for emergent patterns at population scales. In the following sections, weprovide examples of analyses that have expanded our understanding of patterns in humanmovement, their links with social structureswithin societies and show how these reveal parallelsto animal studies.

Statistics of DisplacementThe analysis of individual human trajectories shows that the net displacement of humansbetween two consecutive positions at time t and t + Dt (Dr = |r(t +Dt)� r(t)|) can be described bya power law distribution, P(Dr)�Dr� [384_TD$DIFF]a, where a is a characteristic scaling exponent. Conse-quently, individual trajectories cannot be easily characterized by a single displacement, sinceboth short and long movements can occur. Such a power law distribution is characteristic of aLévy distribution (i.e., Lévy flight, or random walks) with the scaling exponent a < 2 (e.g.,[6,10]). The underlying scaling of human movement thus conforms to the same patterndescribed for many animals, notably marine predators. For these species, including sharks,bony fishes, reptiles, and seabirds, Lévy flight is a behaviour that optimizes the rate at which ananimal encounters prey in patchy environments [17,18].

Site FidelitySimilar to many animals, the most frequent human movements occur around a small set ofhighly visited locations, typically homes, workplaces, shops etc. [19,20]. A high degree of sitefidelity is thus common to the movement of both many animals (e.g., breeding grounds,foraging areas, resting sites) and humans. The displacement range (measured by the gyrationradius) for humans can be described by a power law distribution with exponent a = 1.65truncated by an exponential with a characteristic distance of 350 km [10]. This outcome is aconsequence of very heterogeneous mobility patterns, with the majority of individuals movingwithin a restricted range, whereas others travel over a region that is orders of magnitude larger(with a gyration radius ranging from 1 to 1000 km). For animals, site fidelity is thought to be anoutcome of predictable resource availability [21] and there are clear parallels for humanmovements given that the urban environments are also very predictable in the distributionof resources (schools, homes, workplaces, etc.).

Trends in Ecology & Evolution, Month Year, Vol. xx, No. yy 3

TREE 2198 1–13

Social networks: typically networksof friends, acquaintances, and workor business colleagues that arelinked by interpersonal relationships.The term can also apply to the onlinesite or service that serves to createand sustain these interpersonallinkages.

Collective Patterns of MovementAggregation occurs in both animals (e.g., fish, birds, mammals) and humans [22]. Thebehaviour imposes constraints on movement patterns and a variety of models derived fromphysics have been used to understand and quantify the collective patterns that emerge fromthese behaviours. Analyses of commuting fluxes (recursive travelling between home and work)have been a common subject of studies of collective patterns in human mobility. The dis-tributions of these fluxes are heavy tailed [23–26], that is, distributions whose tails are notexponentially bounded. Spatial distributions of animal movement show similar patterns ofrecursive travel, in many cases between foraging, resting, and reproductive habitats. Thesedistributions of travel have been modelled using approaches based on Brownian motion suchas Brownian bridges that aid definition of home ranges and quantify patterns of space usewithin the range (e.g., [27])

Relationships between Human Mobility and Social Interactions within Populations andMeta-PopulationsDaily encounters among individuals are the main driver of the formation of social relationships[28]. Individuals tend to spend more time with people who are friends and family, making astrong link between physical location and social relationships. Consequently, the location of anyindividual can be predicted from the locations of friends and family and bonds of friendship canbe predicted from spatial and temporal co-occurrence of multiple individuals that are visitedregularly by any individual. The probability of being ‘socially connected’ (i.e., having friendshipbonds with others) decays with distance (d) as d�a [29,30]. Geolocated data on socialinteractions, derived from telephone communications or interactions through online socialnetworks, have been used to measure the relationship of social bonding and geography[30–32]. Bluetooth [33] and radiofrequency identification tags have been developed that storeproximity data to map patterns of social networks (the SocioPatterns project; http://www.sociopatterns.org/) [34]. This latter approach has been used tomap contact patterns in schools[35,36], a hospital ward [37], and conferences [38]. The concept of movement by humans inphysical spatial networks [12] has now been expanded into virtual, digital space through theanalysis of proximity networks in online virtual worlds [39] and social media platforms [12,39].This use of human movement data to understand social interactions is paralleled by a growinginterest in the use of multi-individual tracking strategies to study the structure and dynamics ofsocial networks in populations of wild animals [40,41].

The Ecology of HumanMobilityThe aforementioned examples suggest that the ecological principles that explain themovementpatterns of marine and terrestrial animals also govern humanmobility. A central paradigm of thestudy of animal ecology is that a detailed understanding of movement offers insights into howvariability in the distribution of resources across landscapes affects the performance ofindividuals and, in turn, population-level demography [42,43]. We argue that this concept isequally applicable to the study of humans and can provide new insights into emergentbehaviours of human populations and the impacts of their movement patterns on the ecologyof ecosystems they inhabit. Furthermore, the analytical approaches developed to link mobilityto social networks offer a powerful tool for gaining insights into the social organization of animalspecies [44] (see Outstanding Questions). We illustrate these points with key examples in thefollowing sections.

EpidemicsCompartmental reaction–diffusion models [45] have long been used to describe the dynamicsof population invasion and range expansion in populations of a broad range of species (e.g.,[46]). These models are now also being applied to human populations, notably in the context ofthe dynamics of epidemics, where individuals (or ‘agents’) are divided into different classes

4 Trends in Ecology & Evolution, Month Year, Vol. xx, No. yy

TREE 2198 1–13

Box 1. Resolving Patterns of Human Mobility

P(Δ

r/<Δ

r>)

104

102

100

10-2

10-4

10-6 10-4 10-2 100 102

102 102

102102

102 102 104104 106106 108108

102

<Δr>

104

104 106

102

102100

100

ΔT = 10s100s100s10000s100000s(Δr/<Δr>)-1

Δr/<Δr>

T(s)

NijD /

Niα

Njβ

NijD /

NijM

NijD /

NijM

NijD /

NijM102

100

10-2 10-2

10-210-2

10-410-4100 100

100

100100

101 101

Popula�on of origin Ni Popula�on of origin Ni

High

Low

High

Low

Av. vectorsuitability

Humanmobility

Epi.risk

Voteshare dist.5

4

3

2

1

0 -0.4 -0.2 0v-<v>

0.2 0.4

Spa�al corr.1

0.8

0.6

0.4

0.20

C(r)

101 102 103

r(km)

P(v)

t=0t=4yt=8yt=12yt=16yt=20yt=24yt=28yt=32y

t=0t=4yt=8yt=12yt=16yt=20yt=24yt=28yt=32y

IIII

1.00.90.80.70.60.50.40.30.20.10.0

1.00.90.80.70.60.50.40.30.20.10.0

2012 Data

Model

IIII

II

III

III

I

I

IV

IV

(B)(A)

(C)

(E)

(D)

(F)

r(km) r(km)

Humanmobility has been studied at different levels of resolution. (A) Individual trajectories of 10 000 people across the United States derived from geolocated Twitterdata from June 2015. When analysed as random walks, these data reveal scaling laws that describe the paths followed by humans, the most basic being thedisplacement that occurs in a certain time window, [Dr = r(t + Dt) � r(t)], whose distribution is heavy-[369_TD$DIFF]tailed and consistent with a power-law P(Dr)�Dr�a of exponenta = 1.59 [6] or 1.75 [10]. (B) To illustrate this point, we show the analysis of the worldwide trajectories of 3 million Twitter users in June 2015, which shows adisplacement distribution consistent with a power law of exponent a = 1.0. The inset shows the superdiffusive scaling of the average displacement with time [<Dr>(T) � Tb, with b = 1.27]. (C) The network formed by fluxes of people commuting between counties in the United States [data from US census 2001 (see http://www.census.gov), only the top 20% of the dominant fluxes shown]. Fluxes are explained by population distributions and distances between locations (encapsulated by thegravity and radiationmodels of movement). (D) The fit of the data in C (boxplots) to the gravity model (red lines)Nij�ANi

aNjb/dij

g, whereNij is the number of people livingin county i and working in county j, Ni (Nj) is the population of county i ( j), and dij is the distance between the centroids of counties i and j, with parameters a = 0.35,

Trends in Ecology & Evolution, Month Year, Vol. xx, No. yy 5

TREE 2198 1–13

b = 0.54, and g = 2.7. (D, i) shows the data NijD[370_TD$DIFF] divided by Ni

aNjb as a function of distance and the other plots show the data Nij

D divided by the model fit NijM as a

function of distance (D, ii), population of origin Ni ( [372_TD$DIFF]D, iii), and population of destination Nj ( [373_TD$DIFF]D, iv).

The knowledge of the laws behind human mobility is used to model processes that are mediated by human interaction (red boxes). For example, spatiotemporalfeatures of election results can be reproduced by simple models that take into account human mobility, particularly the vote-share distributions [374_TD$DIFF](E, iii) and spatialcorrelations (see E [375_TD$DIFF],). (E, i and ii) show maps of real (i) and simulated (ii) electoral results. Reprinted from [24]. (F) The spatial spread of epidemics is modelled using ourknowledge of human mobility. In (F) we illustrate this by showing a calculation of the epidemic risk of dengue fever[376_TD$DIFF](F, iii) in Pakistan using data on human mobility[377_TD$DIFF](F, ii)and habitat suitability for the animal vector (mosquitos[378_TD$DIFF], F, i). Reprinted from [79].

depending on their infectious state. How agents meet and spread the infection follows patternsdictated by human mobility [47] and are thus central to these models. A metapopulationapproach is commonly used in order to cope with the computational challenge of having a vastamount of agents that human populations can provide, particularly on whole country or globalscales [48]. These models operate both at large scales and also within constrained environ-ments such as hospitals, schools, or conference venues and at these smaller scales, have beenused to study the spread of disease and tomodel appropriate health-care responses. Recently,Vanhems et al. [49] showed that in a hospital environment, health-care workers were thesubpopulation of individuals who were more likely to act as ‘super-spreaders’ of contagion andthus should be targeted for vaccination [50]. On larger, global scales, the impact of multiscalenetworks of human mobility has been shown to be crucial in the forecasting of infection routesfor global pandemics [23,50,51]. Notably, these have shown how the advent of air transporta-tion has facilitated epidemic spread. For example, in the Middle Ages the wavefront of infectionof the Black Death spread at around 200 km/year [52], whereas today transport rates ofhumanity have accelerated, so that such fronts no longer exist [23]. Importantly, knowledge ofhuman patterns of mobility and social networks offers the opportunity to model the spread ofcontagion to reduce the impact of disease [23]. Such models show that epidemics spreadquickly to remote locations (through air transportation connections), whereas at local scales,the spread patterns follow highly frequented routes typical of commuting flows. Understandingthe impacts of human transport systems for epidemics is directly relevant to the diffusion ofpathogens and invasive species in ecosystems, as many of these rely on human vehicles,notably airplanes, cars, and ships for long distance transport. These allow pathogens andinvasive species to expand their distributions across natural barriers to movement, broadeningtheir range and impact [53].

Diffusion of Innovations, Opinions, and Cultural TraitsThe diffusion of innovations can bemodelled in a manner similar to that of epidemics. The basicpremise of the model is that adopters can be classified as innovators or as imitators and thespeed and timing of adoption of an innovation depend on both the degree of their innova-tiveness and strength of imitation among adopters. Bothmobility (through face-to-face contact)and social networks contribute to the social contagion of innovations [54].

The diffusion of cultural traits such as words [55], language [56], and opinions [24] has beenshown to be dependent on human mobility in both real and digital space (i.e., face-to-facecontacts mediated through movement and online connections), although the dynamics ofsocial interchange might vary. Recent studies have used [396_TD$DIFF]databases provided by geolocatedtweets to study the stability of culture and the impact of seasonal tourism on the heterogeneityof cultural patterns within society [56]. For animals, the biological and evolutionary signifi-cance of social learning has remained a controversial issue, despite clear evidence of thisprocess in fish, reptiles, birds, and mammals [57]. The analysis of movement patterns toinform the study of social learning in humans might prove a useful direction for futureinvestigation of parallels between social learning in animals and the development of culturein humans.

6 Trends in Ecology & Evolution, Month Year, Vol. xx, No. yy

TREE 2198 1–13

Behaviours during StressIn the case of natural disasters or emergencies in confined environments, understanding thecollective behaviour of crowds might help devise better strategies to cope with such events[58,59]. Studies of human movement during panic situations show that optimal strategies forescape from hazardous situations, such as a smoke-filled room, involve a mixture of individu-alistic behaviour and collective ‘herding’ instinct [60]. At larger scales, mobile phone data havebeen used to track the movements of people after the 2010 earthquake and subsequentcholera outbreak in Haiti [61], Hurricane Sandy in the United States [62], and other disasters[59]. Rapid description of such movement patterns via data from mobile phones and otherdevices could allow aid and disaster relief to be targeted more effectively for communities.Overall, such analyses provide compelling evidence that human escape reactions conform tothe patterns of coordinated collective behaviour described for animals. These result in thesynchronized behaviours exhibited by schooling fish and flocking birds under attack [63]. Thefinding that animal escape responses rely on rapid and efficient transfer of information amongindividuals, and the diffusion of reactions among neighbours [64], might help develop predictiveframeworks for the response of humans to emergencies or stress, a major issue for publicsafety.

Linking Human Mobility and Animal Movement for Conservation andManagement OutcomesThe analysis and conceptualization of the big data produced by geolinked devices and socialnetworks allow patterns of human and animal movement to be superimposed and synthesizedto recognize, understand, and solve urgent conservation and management problems. Threekey examples of this approach are outlined here.

FishingAccurate and reliable data on the behaviour of fishers fundamentally underpin the successfulmanagement of a fishery. The analysis of big data provided by smartphones and socialnetworks is poised to revolutionize this process, replacing conventional surveys that are limitedboth in accuracy and in spatial and temporal extent due to cost [65]. Mobile phone appsproduce synoptic, real-time data that not only reveal patterns of fishing pressure and catch andeffort data [66], but also connectivity in movement patterns of fishers that can also be used totrack the risk of the spread of invasive species. At the scale of stocks, information from socialnetworks can be used to optimize fishing practices across entire fisheries to minimize bycatch,enhance sustainability, and reduce environmental impacts [67]. Globally, the introduction ofautomatic ship identification systems makes it possible to observe the behaviour and move-ment patterns of entire fishing fleets and to map compliance with management strategies. Forthe first time, this offers a cost-effective means to address the issue of illegal fishing on the highseas [68].

The Illegal Trade in WildlifeTrade in wildlife is now acknowledged to be a global crisis. The practice is pushing the mostvalued species to extinction and depriving many developing countries of billions of dollars ofresources. The quantities of animals and plants involved are truly staggering, as are the lossesto the front-line park ranger community tasked with protecting these resources; it is estimatedthat over one thousand rangers have been killed in the line of duty during the last decade (UnitedNations Environment Programme; http://www.unep.org/documents/itw/ITW_fact_sheet.pdf).

Geolocated devices can play a key role in combating this trade and in safeguarding the lives ofrangers. In India, the smartphone app ‘Hejji’ allows rangers to monitor the movement andsighting patterns of tigers in protected reserves, and to also place encounters in context, byadding information on water resources, forest fires, and importantly, suspicious human activity.

Trends in Ecology & Evolution, Month Year, Vol. xx, No. yy 7

TREE 2198 1–13

Similarly, ‘Wildapp’ is used by rangers in Africa to collate social and environmental data andtrackmovements of rangers and encounter rates with dead animals. Wildapp is producing datavital to determining poaching rates (https://ilabafricastrathmore.wordpress.com/2016/09/20/the-wild-app-helping-conserve-our-wild-animals/). In Kenya ‘tenBoma’ collates informationfrom rangers and satellite remote sensing, and uses mobile phones confiscated from poachersto expose the social networks that facilitate the trade (http://www.ifaw.org/sites/default/files/IFAW-tenBoma.pdf). The latter program illustrates an important point: the same big data andsoftware platforms that can aid conservation can also facilitate the illegal trade in wildlife.

Human Mobility As a Threat to AnimalsOur transport systems are responsible for the deaths of many millions of animals and trillions ofinsects each year [69]. Road and train networks fragment habitats, form barriers to migration,act as conduits for the spread of invasive species, and are a source of light, noise, and

Box 2. Ship Strikes and Marine Megafauna

103

104

102

101

100

10-1

[379_TD$DIFF]Maritime traffic volumes and vessel speeds have grown rapidly over the last few decades. Shipping is responsible for thetransportation of 90% of global trade [80] (http://www.imo.org/) and in 2014, there were 89 464 registered vesselsundertaking commercial passage [81]. [380_TD$DIFF]Figure shows the average monthly density of all vessels, calculated on alogarithmic scale per 1 km2

[371_TD$DIFF] grid cell for 2014. There has also been an emergence of new trade routes in areas that wereonce relatively untrafficked, such as the northern sea route through the high Arctic, where there is projected to beaverage annual increase in shipping of 20% per year over the next 25 years [82]. The global increase in shipping has ledto a corresponding increase in the number of ship strikes onmarinemegafauna, particularly on those species that spendlong periods of time near the surface [83] such as whales, sirenids, turtles, and whale (Rhincodon typus) and basking(Cetorhinusmaximus) sharks. There are numerous documented cases of both serious injuries andmortalities caused byship strikes on this fauna [84]. Ship strikes are exacerbated by the size and the speed of vessels and the likelihood ofcollision increases in situations where high volumes of maritime traffic overlap with areas important tomarinemegafaunasuch asmigratory corridors, feeding, or breeding grounds [84,85]. This threat is of most concern for small populations innear-shore habitats with high levels of maritime traffic. For example, collisions with vessels are a causative factor thathave prevented population recovery in the North Atlantic right whale (Eubalaena glacialis) despite the cessation ofcommercial whaling [86]. Similarly, vessels are implicated in the decline in abundance of some whale shark populations[87] and of some marine turtles (e.g., Chelonia mydas), which display strong site and migratory route fidelities to areasthat overlap with high densities of shipping [88].

Mitigation through Mobility

The combination of fine-scale information on ship movements through the AIS system [68] and advances in ourunderstanding of animal movement behaviour through extensive telemetry [7] has enabled real opportunity for effectivemitigation. Since US NOAA regulations lowered ship speed limits in the vicinity of right whale habitats in 2008, deathsfrom vessel strikes have declined [89] and devices such as Whale Alert: A Tool for Reducing Collisions Between Shipsand Endangered Whales [90] are now being developed to initiate real-time mitigation. Big data approaches combinedwith insights from refinements of animal movement models provide the potential for real-time dynamic and cost-effective mitigation of this otherwise growing threat.

8 Trends in Ecology & Evolution, Month Year, Vol. xx, No. yy

TREE 2198 1–13

particulate pollution that alter animal behaviour and survivorship. Dead and injured animals andinsects on roads and verges can attract scavengers that are then also at greater risk of mortality[70].

Until recently, mitigation of these effects has been difficult because the quantification of roadkillis time consuming and effort intensive, limiting the spatial and temporal scope of data collec-tion. However, there is a recent proliferation of smartphone apps that apply citizen scienceapproaches to register roadkills. These apps provide data that allow ranking of speciesvulnerability, seasonal patterns, and monitoring and detection of rare species in roadkills.Importantly, the data provided by the apps allow identification of roadkill hot spots andlandscape connectivity issues across broad areas, so that managers can take evidence-basedactions to mitigate effects. The data can also be used to increase vehicle safety by incorpo-rating warnings of collision hot spots into global positioning system navigation systems for cars[71]. These issues andmitigation approaches are paralleled inmarine systemswhere ship strikeposes a major threat to marine megafauna (Box 2).

Future Insights on Human Mobility � Lessons from Animal Ecology andSolutions for Management and ConservationThe aforementioned examples show how studies of emergent patterns in human trajectories atdifferent spatial scales are consistent with theory and observations derived for animal move-ment. Thus, ecological movement theory can inform our understanding of biological and socialprocesses in human societies beyond the phenomenological approach prevalent to date. Howdo we facilitate this shift? One obvious route forward utilizes the approaches of earlier studiesthat examined the movement patterns of hunter–gatherer societies and large-scale migrationpatterns [2]. Many of these applied an optimization framework equivalent to optimal foragingtheory in animal ecology [72,73] to describe and understand the drivers of movement patterns.Such frameworks are clearly applicable to analyses of big data because they are robust in theface of differences in patterns of mobility between humans and animals (Box 3) and can beapplied in contexts well beyond foraging. In some limited circumstances these ideas are alreadywell developed, such as cost and benefit studies of commuting and urban sprawl [74,75].However, the new generation of wearable devices and the linking of digital social networks tomobility patterns offer a means to bring a far great degree of complexity and detail tooptimization studies.

In turn, the flexibility, detail, and immediacy of data gathered by smartphones and geolocateddevices offer new opportunities to improve management and conservation outcomes for someof the most important ecological issues faced by human society. Minimizing adverse impacts

Box 3. Similarities and Differences in Human and Animal Mobility

[381_TD$DIFF]The mobility patterns of humans are most similar to those of many marine animals because, through the use oftechnology, we have escaped the constraints of body size that apply to the movements of most terrestrial species thattransport themselves on land [91]. Some of the other species that have also escaped the constraints of terrestriallocomotion are those that are either deliberately or inadvertently included by humans in their travels, such as microbes,domestic animals, and invasive species. Clear examples of convergence in the studies of animal movement and humanmobility are the identification of scaling laws that characterize movement patterns such as Lévy flight, the character-ization of spatial patterns in residency such as fidelity, recursive movements, and home ranges, and in the spread ofinvasive species and epidemics. In the case of human epidemics the convergence is complete, since the microbe bothinfects and is transported by the human host. However, these comparisons also reveal a fundamental difference inhuman and animal search behaviour because today, many human ‘foraging’ movements are directed; humans oftenknowwhere the item is before they embark on a journey to find it. For example, humans are likely to use prior knowledgeor to do a search online to find a restaurant, bar, or specific good for sale and then go directly to that place where theservice or item is available. Analysis of the most extensive data sets of human movements within cities show that this isthe case � the majority of movements within a city by people occur between a home and a workplace [10,20,92,93].Thus, unlike animals, most human movements might not necessarily involve (at least extrinsic) searching behaviour.

Trends in Ecology & Evolution, Month Year, Vol. xx, No. yy 9

TREE 2198 1–13

Box 4. Biologging and the Internet of Things

(A)(A) (B)(B)

Predicted growth in the number and capacity of electronic devices will not only increase the amount of data they provideon human mobility, but also the velocity at which they are generated and the complexity of information these datacontain. Wearable devices, such as the ‘Fitbit’ shown in (A) provide a new data stream that opens a window intomovement and body physiology [94]. This development mirrors the field of [382_TD$DIFF]telemetry and biologging in animals [7] wheretracking tags (B – southern elephant seals, Mirounga leonina, note tag on head of seal) [383_TD$DIFF]not only transmit location databut also store data from a variety of sensors that can record body position in three dimensions and internal (e.g., heartrate) and environmental (e.g., temperature, depth, altitude) variables. As they have done for animals, wearable deviceswill allow the development of theories connecting human behaviour, body condition, and health to movement andactivity [95,96]. In turn, this will be coupled with innovations that connect objects within human environments to theInternet. Tracking devices are now commonplace not just on people, but also on all forms of transport (e.g., cars, ships,planes) and the environments in which theymove are being linked by technologies such as smart grids, smart cities, andsmart homes. By 2020 it is predicted that human society will have 50 billion Internet-connected devices, with averageownership of 6.58 devices per person [97]. As a result, humans will move within an environment described as the‘Internet of Things’ [98]. Data onmovement patterns, the internal state of the personmoving, the environment traversed,and the mode of transport used will be available at all times and in all environments on the planet, including those thatwere once both data-poor and costly to monitor. This has implications for many aspects of social organization andbehaviour. It will speed the diffusion of innovations, culture, and opinions as well as allow much closer regulation ofbehaviour. For example, the remoteness and expense of monitoring the open ocean have meant that these environ-ments have long been a haven for illegal fishing. Today, vessels are being equipped with automatic identificationsystems that report positions to satellites to ensure compliance with fisheries regulations [68]. This type of monitoringcan be done remotely and at a fraction of the cost of most other types of compliance enforcement.

[397_TD$DIFF]Outstanding Questions� Movement studies of animals, par-ticularly those involving telemetry andbiologging (Box 4) are now generatingbig data. To what extent can the ana-lytical techniques that have beenapplied to human mobility now informand improve the exploration andunderstanding of animal mobility?� Big data analyses of human move-ment have been driven by complexsystem analyses. Animal movementstudies provide a framework withinwhich to [398_TD$DIFF]pose hypotheses abouthuman movement, but to what degreedo external drivers produced by thecomplexity of human sociality skewthe potential outcomes?� The analysis of data sets of con-straints and motivations for humanmovement poses ethical issues, evenwhen these use existing public datasets, because the more detail that isadded to the profile of the person whois making a movement, the higher thelikelihood that he or she could be indi-vidually identified from results. This isan important issue for research thatuses big data and will be particularlyrelevant in the context of optimizationstudies.� Big data can be used to understandand optimize the management of legit-imate activities for resource extractionsuch as fishing, but can also empowerand facilitate illicit and unregulatedbehaviours that now threaten thefuture of species and even entire habi-tats. How do we curb such behaviourswhen connectivity within the samesoftware platforms can be critical toboth the legal and illegal use ofresources?

and ensuring sustainable harvests of resources are central to humanity’s future. The analysis ofbig data provided by geolocated devices offers a multitude of opportunities because it providesa cheap and simple means to gather comprehensive and synoptic data sets that allowvisualization of patterns and interactions within human movement and social networks atspatial scales that have never been available in the past.

Concluding RemarksThe infiltration of mobile phones into human societies is now almost complete, with 6.8 billionsubscribers (i.e., 96% of the world population in 2013) [9]. The rapidly developing study of thepatterns of mobility of human populations derived from these big data provided by mobilephones and other digital devices has many clear parallels with studies of animal movement.However, a key difference between the two fields is that the emergent patterns foundby the combined analysis of individual animal trajectories form the evidential basis ofthe theory of animal movement ecology [76], whereas human studies do not yet have anequivalent theoretical framework that operates across all spatial scales. Some argue thatsuch a framework is unnecessary, because ‘with enough data, the numbers speak forthemselves’ [77] (http://www.edge.org/3rd_culture/anderson08/anderson08_index.html). In

10 Trends in Ecology & Evolution, Month Year, Vol. xx, No. yy

TREE 2198 1–13

reality, all explorations of big data sift through an array of variables and include only subsets ofthese in analyses, so that preconceptions of researchers must always influence results [78].Optimization theory provides a relevant, simple, and transparent framework for this process.Because humanmobility data can be linked to social networks, economic transactions, and thephysiological state of an individual, the currency and constraints both motivating and limitingmovement patterns can be explored in very fine detail. The volume, scope, and immediacy ofthese big data offer robust and novel insights into some of the major problems surrounding thesustainable use of ecological resources and the minimization of human impacts on terrestrialand aquatic ecosystems that now confront society.

[399_TD$DIFF]AcknowledgmentsThe research reported in this publication was supported by King Abdullah University of Science and Technology (KAUST)

through the baseline fund and the Award No. 63150414 from the Office of Sponsored Research (OSR) and the Australian

Institute of Marine Science. A.M.M.S. was supported by an IOMRC (AIMS, CSIRO, and UWA) Collaborative Post-doctoral

Fellowship, J.F.-G. by NIH grant U54GM088558-06 (Lipsitch) [400_TD$DIFF]and V.M.E. by Agencia Estatal de Investigación (AEI) and

Fondo Europeo de Desarrollo Regional (FEDER) through project SPASIMM (FIS2016-80067-P (AEI/FEDER, UE)). We

thank Ivan D. Gromicho for the artwork in Figure 1.

References

1. Rodrigue, J.-P. and Comtois, C. (2013) Transportation and

energy. In The Geography of Transport Systems (Rodrigue, J.-P., ed.), Routledge, pp. 1–41

2. Mulder, M.B. and Schacht, R. (2012) Human behavioural ecol-ogy. In Encyclopedia of Life Sciences. Wiley http://dx.doi.org/10.1002/9780470015902.a0003671.pub2

3. Black, R. et al. (2011) The effect of environmental change onhuman migration. Glob. Environ. Change 21, S3–S11

4. Kline, D.S. (2003) Push and pull factors in international nursemigration. J. Nurs. Sch. 35, 107–111

5. United Nations Department of Economic and Social Affairs, Pop-ulation Division (2014) World Urbanization Prospects: The 2014Revision, United Nations

6. Brockmann, D. et al. (2006) The scaling laws of human travel.Nature 439, 462–465

7. Hussey, N.E. et al. (2015) Aquatic animal telemetry: a panoramicwindow into the underwater world. Science 348, 1255642

8. Piazza, A. et al. (1987) Migration rates of human populations fromsurname distributions. Nature 329, 714–716

9. Blondel, V.D. et al. (2015) A survey of results on mobile phonedatasets analysis. EPJ Data Sci. 4, 10

10. González, M.C. et al. (2008) Understanding individual humanmobility patterns. Nature 453, 779–782

11. Saramäki, J. and Moro, E. (2015) From seconds to months: anoverview of multi-scale dynamics of mobile telephone calls. Eur.Phys. J. B 88, B882015164

12. Eagle, N. et al. (2009) Inferring friendship network structure byusing mobile phone data. Proc. Natl. Acad. Sci. U. S. A. 106,15274–15278

13. Stopczynski, A. et al. (2014) Measuring large-scale social net-works with high resolution. PLoS One 9, e95978

14. Wang, D. et al. (2011) Human mobility, social ties, and linkprediction. In Proceedings of the 17th ACM SIGKDD InternationalConference on Knowledge Discovery and DataMining, pp. 1100–1108, KDD ’11

15. Wu, X. et al. (2014) Data mining with big data. IEEE Trans. Knowl.Data Eng. 26, 97–107

16. Lazer, D. et al. (2009) Life in the network: the coming age ofcomputational social science. Science 323, 723

17. Humphries, N.E. et al. (2010) Environmental context explainsLévy and Brownian movement patterns of marine predators.Nature 465, 1066–1069

18. Sims, D.W. et al. (2008) Scaling laws of marine predator searchbehaviour. Nature 451, 1098–1102

19. Krumme, C. et al. (2013) The predictability of consumer visitationpatterns. Sci. Rep. 3, 1645

20. Song, C. et al. (2010) Limits of predictability in human mobility.Science 327, 1018–1021

21. Switzer, P.V. (1993) Site fidelity in predictable and unpredictablehabitats. Evol. Ecol. 7, 533–555

22. Parrish, J.K. and Hamner, W.M. (1997) Animal Groups in ThreeDimensions: How Species Aggregate, Cambridge UniversityPress

23. Balcan, D. et al. (2009) Multiscale mobility networks and thespatial spreading of infectious diseases. Proc. Natl. Acad. Sci.U. S. A. 106, 21484–21489

24. Fernández-Gracia, J. et al. (2014) Is the voter model a model forvoters? Phys. Rev. Lett. 112, 158701

25. Masucci, A.P. et al. (2013) Gravity versus radiation models: on theimportance of scale and heterogeneity in commuting flows. Phys.Rev. E 88, 022812

26. Simini, F. et al. (2012) A universal model for mobility and migrationpatterns. Nature 484, 96–100

27. Horne, J.S. et al. (2007) Analyzing animal movements usingBrownian bridges. Ecology 88, 2354–2363

28. Toole, J.L. et al. (2015) Coupling humanmobility and social ties. J.R. Soc. Interface 12, 20141128

29. Grabowicz, P.A. et al. (2014) Entangling mobility and interactionsin social media. PLoS One 9, e92196

30. Lambiotte, R. et al. (2008) Geographical dispersal of mobilecommunication networks. Physica A 387, 5317–5325

31. Lengyel, B. et al. (2015) Geographies of an online social network.PLoS One 10, e0137248

32. Onnela, J.-P. et al. (2011) Geographic constraints on socialnetwork groups. PLoS One 6, e16939

33. Eagle, N. and Pentland, A. (2006) Reality mining: sensingcomplex social systems. Pers. Ubiquitous Comput. 10, 255–268

34. Cattuto, C. et al. (2010) Dynamics of person-to-person interac-tions from distributed RFID sensor networks. PLoS One 5,e11596

35. Fournet, J. and Barrat, A. (2014) Contact patterns among highschool students. PLoS One 9, e107878

36. Stehlé, J. et al. (2011) High-resolution measurements of face-to-face contact patterns in a primary school. PLoS One 6, e23176

37. Isella, L. et al. (2011) Close encounters in a pediatric ward:measuring face-to-face proximity and mixing patterns with wear-able sensors. PLoS One 6, e17144

38. Barrat, A. et al. (2010) Social dynamics in conferences: analysesof data from the Live Social Semantics application. In TheSemantic Web–ISWC, pp. 17–33, Springer

Trends in Ecology & Evolution, Month Year, Vol. xx, No. yy 11

TREE 2198 1–13

39. Szell, M. (2010) Multirelational organization of large-scale socialnetworks in an online world. Proc. Natl. Acad. Sci. U. S. A. 107,13636–13641

40. Armansin, N.C. et al. (2016) Integrating social network analysisand fine-scale positioning to characterise the associations of abenthic shark. Anim. Behav. 115, 245–258

41. Jacoby, D.M.P. and Freeman, R. (2016) Emerging network-based tools in movement ecology. Trends Ecol. Evol. 31,301–314

42. Gaillard, J.-M. et al. (2010) Habitat–performance relationships:finding the right metric at a given spatial scale. Philos. Trans. R.Soc. Lond. B Biol. Sci. 365, 2255–2265

43. Morales, J.M. et al. (2010) Building the bridge between animalmovement and population dynamics.Philos. Trans. R. Soc. Lond.B Biol. Sci. 365, 2289–2301

44. Kurvers, R.H.J.M. et al. (2014) The evolutionary and ecologicalconsequences of animal social networks: emerging issues.Trends Ecol. Evol. 29, 326–335

45. Kermack, W.O. and McKendrick, A.G. (1927) A contribution tothe mathematical theory of epidemics. Philos. R. Soc. Lond. AMath. Phys. Eng. Sci. 115, 700–721

46. Andow, D.A. et al. (1990) Spread of invading organisms. Landsc.Ecol. 4, 177–188

47. Funk, S. (2010) Modelling the influence of human behaviour onthe spread of infectious diseases: a review. J. R. Soc. Interface 7,1247–1256

48. Mata, A.S. and Ferreira, S.C. (2013) Pair quenched mean-fieldtheory for the susceptible-infected-susceptible model on com-plex networks. Europhys. Lett. 103, 48003

49. Vanhems, P. et al. (2013) Estimating potential infection transmis-sion routes in hospital wards using wearable proximity sensors.PLoS One 8, e73970

50. Salathe, M. et al. (2012) Digital epidemiology. PLoS Comput. Biol.8, e1002616

51. Tizzoni, M. et al. (2014) On the use of human mobility proxies formodeling epidemics. PLoS Comput. Biol. 10, e1003716

52. Murray, J. (2003) In Mathematical Biology: Spatial Models andBiomedical Applications (Vol. II), Springer-Verlag

53. Hulme, P.E. (2009) Trade, transport and trouble: managing inva-sive species pathways in an era of globalization. J. Appl. Ecol. 46,10–18

54. Toole, J.L. et al. (2012) Modeling the adoption of innovations inthe presence of geographic and media influences. PLoS One 7,e29528

55. Lizana, L. et al. (2011) Modeling the spatial dynamics of culturespreading in the presence of cultural strongholds. Phys. Rev. E83, 066116

56. Mocanu, D. et al. (2013) The Twitter of Babel: mapping worldlanguages through microblogging platforms. PLoS One 8, e61981

57. Laland, K.N. and Janik, V.M. (2006) The animal cultures debate.Trends Ecol. Evol. 21, 542–547

58. Kryvasheyeu, Y. et al. (2016) Rapid assessment of disaster dam-age using social media activity. Sci. Adv. 2, e1500779

59. Wang, Q. and Taylor, J.E. (2016) Patterns and limitations of urbanhuman mobility resilience under the influence of multiple types ofnatural disaster. PLoS One 11, e0147299

60. Helbing, D. et al. (2000) Simulating dynamical features of escapepanic. Nature 407, 487–490

61. Bengtsson, L. et al. (2011) Improved response to disasters andoutbreaks by tracking population movements with mobile phonenetwork data: a post-earthquake geospatial study in Haiti. PLoSMed. 8, e1001083

62. Wang, Q. and Taylor, J.E. (2014) Quantifying human mobilityperturbation and resilience in Hurricane Sandy. PLoS One 9,e112608

63. Procaccini, A. et al. (2011) Propagating waves in starling,Sturnus vulgaris, flocks under predation. Anim. Behav. 82,759–765

64. Rosenthal, S.B. et al. (2015) Revealing the hidden networks ofinteraction in mobile animal groups allows prediction of complex

12 Trends in Ecology & Evolution, Month Year, Vol. xx, No. yy

behavioral contagion. Proc. Natl. Acad. Sci. U. S. A. 112,4690–4695

65. Gutowsky, L.F.G. et al. (2013) Smartphones and digitaltablets: emerging tools for fisheries professionals. Fisheries 38,455–461

66. Papenfuss, J.T. et al. (2015) Smartphones reveal angler behavior:a case study of a popular mobile fishing application in Alberta,Canada. Fisheries 40, 318–327

67. Barnes, M. et al. (2016) Social networks and environmental out-comes. Proc. Natl. Acad. Sci. U. S. A. 113, 6466–6471

68. McCauley, D.J. et al. (2016) Ending hide and seek at sea. Science351, 1148–1150

69. Muñoz, P.T. et al. (2015) Effects of roads on insects: a review.Biodivers. Conserv. 24, 659–682

70. Trombulak, S.C. and Frissell, C.A. (2000) Review of ecologicaleffects of roads on terrestrial and aquatic communities [Revisiónde los Efectos de Carreteras en Comunidades Terrestres yAcuáticas]. Conserv. Biol. 14, 18–30

71. Vercayie, D. and Herremans, M. (2015) Citizen science andsmartphones take roadkill monitoring to the next level. Nat.Conserv. 11, 29–40

72. Charnov, E.L. (1976) Optimal foraging, the marginal valuetheorem. Theor. Popul. Biol. 9, 129–136

73. Perry, G. and Pianka, E.R. (1997) Animal foraging: past, presentand future. Trends Ecol. Evol. 12, 360–364

74. Brueckner, J.K. (2001) Urban sprawl: lessons from urbaneconomics. In Brookings-Wharton Papers on Urban Affairs,pp. 65–97, Brookings Institution Press

75. Nechyba, T.J. and Walsh, R.P. (2004) Urban sprawl. J. Econ.Perspect. 18, 177–200

76. Nathan, R. et al. (2008) A movement ecology paradigm forunifying organismal movement research. Proc. Natl. Acad. Sci.U. S. A. 105, 19052–19059

77. Anderson, C. (2008) The End of Theory: The Data Deluge Makesthe Scientific Method Obsolete, Wired Magazine

78. Boyd, D. and Crawford, K. (2012) Critical questions for big data:provocations for a cultural, technological, and scholarly phenom-enon. Inf. Commun. Soc. 15, 662–679

79. Wesolowski, A. et al. (2015) Impact of human mobility on theemergence of dengue epidemics in Pakistan. Proc. Natl. Acad.Sci. U. S. A. 112, 11887–11892

80. International Maritime Organization (2016) http://www.imo.org(2016).

81. United Nations Organization (2015) United Nations Conferenceon Trade and Development Review of Maritime Transport 2015(UNCTAD/RMT/2015), pp. 1–108, United Nations Organization

82. Miller, A.W. and Ruiz, G.M. (2014) Arctic shipping and marineinvaders. Nature Clim. Change 4, 413–416

83. Laist, D.W. et al. (2001) Collisions between ships and whales.Mar. Mamm. Sci. 17, 35–75

84. Vanderlaan, A.S.M. and Taggart, C.T. (2007) Vessel collisionswith whales: the probability of lethal injury based on vessel speed.Mar. Mamm. Sci. 23, 144–156

85. Conn, P.B. and Silber, G.K. (2013) Vessel speed restrictionsreduce risk of collision-related mortality for North Atlantic rightwhales. Ecosphere 4, 1–16

86. Kraus, S.D. et al. (2005) North Atlantic right whales in crisis.Science 309, 561

87. Speed, C.W. et al. (2008) Scarring patterns and relativemortality rates of Indian Ocean whale sharks. J. Fish Biol. 72,1488–1503

88. Hazel, J. et al. (2007) Vessel speed increases collision risk forthe green turtle Chelonia mydas. Endangered Species Res. 3,103–113

89. van der Hoop, J.M. et al. (2015) Vessel strikes to largewhales before and after the 2008 ship strike rule. Conserv. Lett.8, 24–32

90. Wiley, D. et al. (2015) Whale alert: a tool for reducing collisionsbetween ships and endangered whales. In AAAS 2015 AnnualMeeting. AAAS

TREE 2198 1–13

91. Tamburello, N. (2015) Energy and the scaling of animal spaceuse. Am. Nat. 186, 196–211

92. Bagrow, J.P. and Lin, Y.-R. (2012) Mesoscopic structure andsocial aspects of human mobility. PLoS One 7, e37676

93. Csáji, B.C. et al. (2013) Exploring the mobility of mobile phoneusers. Physica A 392, 1459–1473

94. Gibney, E. (2015) The inside story on wearable electronics.Nature 528, 26–28

95. Beiwinkel, T. et al. (2016) Using smartphones to monitor bipolardisorder symptoms: a pilot study. JMIR Ment. Health 3, e2

96. Onnela, J.-P. and Rauch, S.L. (2016) Harnessing smartphone-based digital phenotyping to enhance behavioral and mentalhealth. Neuropsychopharmacology 41, 1691–1696

97. Evans, D. (2011) The Internet of things: How the next evolutionof the Internet is changing everything. In CISCO White Paper(Vol. 1), pp. 1–11, CISCO

98. Ashton, K. (2009) That ‘Internet of things’ thing. RFID J. 22,97–114

99. Ward, J.S. and Barker, A. (2013) Undefined by data: a survey ofbig data definitions. arXiv, 1309.5821

Trends in Ecology & Evolution, Month Year, Vol. xx, No. yy 13