The Effects of Density Dependence and Immigration on Local Adaptation and Niche Evolution in a Black-Hole Sink Environment

We examine the effects of density dependence and immigration on local adaptation in a "black-hole sink" habitat, i.e., a habitat in which isolated populations of a species would tend to extinction but where a population is demographically maintained by recurrent one-way migration from a separate source habitat in which the species persists. Using a diploid, one-locus model of a discrete-generation sink population maintained by immigration from a fixed source population, we show that a locally favored allele will spread when rare in the sink if the absolute fitness (or, in some cases, the geometric-mean absolute fitness) of heterozygotes with the favored allele is above one in the sink habitat. With density dependence, the criterion for spread can depend on the rate of immigration, because immigration affects local densities and, hence, absolute fitness. Given the successful establishment of a locally favored allele, it will be maintained by a migration-selection balance and the resulting polymorphic population will be sustained deterministically with either stable or unstable dynamics. The densities of stable polymorphic populations tend to exceed densities that would be maintained in the absence of the favored allele. With strong density regulation, spread of the favored allele may destabilize population dynamics. Our analyses show that polymorphic populations which form subsequent to the establishment of favorable alleles have the capacity to persist deterministically without immigration. Finally, we examined the probabilistic rate at which new favored alleles arise and become established in a sink population. Our results suggest that favored alleles are established most readily at intermediate levels of immigration.     

Allee effects, immigration, and the evolution of species' niches

Theoretical studies of adaptation to sink environments (with conditions outside the niche requirements of a species) have shown that immigration from source habitats can either facilitate or inhibit local adaptation. Here, we examine the influence of immigration on the evolution of local adaptation, given an Allee effect (i.e., at low densities, absolute fitness increases with population density). We consider a deterministic model for evolution at a haploid locus, and a stochastic individual-based model for evolution of a quantitative trait, and several kinds of Allee effects. We demonstrate that increased immigration can greatly facilitate adaptive evolution in the sink; with greater immigration, local population sizes rise, and because of the Allee effect, there is a positive indirect effect of immigration on local fitness. This makes it easier for alleles of modest effect to be captured by natural selection, transforming the sink into a locally adapted population that can persist without immigration.     

Temporal variation can facilitate niche evolution in harsh sink environments

We examine the impact of temporal variation on adaptive evolution in "sink" environments, where a species encounters conditions outside its niche. Sink populations persist because of recurrent immigration from sources. Prior studies have highlighted the importance of demographic constraints on adaptive evolution in sinks and revealed that adaptation is less likely in harsher sinks. We examine two complementary models of population and evolutionary dynamics in sinks: a continuous-state quantitative-genetics model and an individual-based model. In the former, genetic variance is fixed; in the latter, genetic variance varies because of mutation, drift, and sampling. In both models, a population in a constant harsh sink environment can exist in alternative states: local maladaptation (phenotype comparable to immigrants from the source) or adaptation (phenotype near the local optimum). Temporal variation permits transitions between these states. We show that moderate amounts of temporal variation can facilitate adaptive evolution in sinks, permitting niche evolution, particularly for slow or autocorrelated variation. Such patterns of temporal variation may particularly pertain to sinks caused by biotic interactions (e.g., predation). Our results are relevant to the evolutionary dynamics of species' ranges, the fate of exotic invasive species, and the evolutionary emergence of infectious diseases into novel hosts.     

On the evolutionary stability of sink populations

The evolution of adaptive behaviours can influence population dynamics. Conversely, population dynamics can affect both the rate and direction of adaptive evolution. This paper examines reasons why sink populations – populations maintained by immigration, preventing local extinction – might persist in the habitat repertoire of a species over evolutionary time-scales. Two such reasons correspond to standard explanations for deviations from an ideal free habitat distribution: organisms may not be free to settle in whichever habitat has the highest potential fitness, and may be constrained by costs, perceptual limitations, or mode of dispersal in the acuity of their habitat selectivity. Here, I argue that a third general reason for persistent sink populations is provided by unstable population dynamics in source habitats. I present a simple model illustrating how use of a sink habitat may be selectively advantageous, when a source population has unstable dynamics (which necessarily reflects temporal variation in local fitnesses). Species with unstable local dynamics in high-quality habitats should be selected to utilize a broader range of habitats than species with stable local dynamics, and in particular in some circumstances should utilize sink habitats. This observation has implications for the direction of niche evolution, and the likelihood of niche conservatism.     

The phenomenology of niche evolution via quantitative traits in a 'black-hole' sink

Previous studies of adaptive evolution in sink habitats (in which isolated populations of a species cannot persist deterministically) have highlighted the importance of demographic constraints in slowing such evolution, and of immigration in facilitating adaptation. These studies have relied upon either single-locus models or deterministic quantitative genetic formulations. We use individual-based simulations to examine adaptive evolution in a 'black-hole' sink environment where fitness is governed by a polygenic character. The simulations track both the number of individuals and their multi-locus genotypes, and incorporate, in a natural manner, both demographic and genetic stochastic processes. In agreement with previous studies, our findings reveal the central parts played by demographic constraints and immigration in adaptation within a sink (adaptation is more difficult in environments with low absolute fitness, and higher immigration can accelerate adaptation). A novel finding is that there is a 'punctuational' pattern in adaptive evolution in sink environments. Populations typically stay maladapted for a long time, and then rapidly shift into a relatively adapted state, in which persistence no longer depends upon recurrent immigration.     

Canalization breakdown and evolution in a source-sink system

Understanding the process of adaptation to novel environments may help to elucidate several ecological phenomena, from the stability of species range margins to host-pathogen specificity and persistence in degraded habitats. We study evolution in one type of novel environment: a sink habitat where populations cannot persist without recurrent immigration from a source population. Previous studies on source-sink evolution have focused on how extrinsic environmental factors influence adaptation to a sink, but few studies have examined how intrinsic genetic factors influence adaptation. We use an individual-based model to explore how genetic canalization that evolves in gene regulation networks influences the adaptation of a population to a sink. We find that as canalization in the regulation network increases, the probability of adaptation to the novel habitat decreases. When adaptation to the habitat does occur, it is usually preceded by a breakdown of canalization. As evolution continues in the novel habitat, canalization reemerges, but a legacy of the breakdown may remain, even after several generations. We also find that environmental noise tends to increase the probability of adaptation to the novel habitat. Our results suggest that the details of genetic architecture can significantly influence the likelihood of niche evolution in novel environments.     

Integrated population model reveals that kestrels breeding in nest boxes operate as a source population

The identification of the source–sink status of a population is critical for the establishment of conservation plans and enacting smart management decisions. We developed an integrated population model to formally assess the source status of a kestrel Falco tinnunculus population breeding in nest boxes in Switzerland. We estimated juvenile and adult survival, reproduction and net dispersal (emigration/immigration) by jointly analyzing capture–recapture, dead recovery, breeding monitoring and population survey data. We also investigated the role of nest boxes on kestrel demography and assessed the contributions of vital rates to realized population growth rates. The results indicate that the kestrel population breeding in nest boxes has acted as a source over the 15 years of the study duration. A quantitative approach suggests that a substantial number of individuals have emigrated annually from this population likely affecting the population dynamics outside the management area. Variation in fecundity explained 34% of the temporal variability of the population growth rate. Moreover, a literature review suggests that kestrel pairs produce on average 1.4 chicks more per breeding attempt in nest boxes compared to natural open nests. Together, these findings suggest that fecundity was an important driver for the dynamics of this population and that nest boxes have contributed to its raise. Nest boxes are regularly used as an efficient tool for conservation management. We suggest that such a conservation action can result in the establishment of a source population being beneficial for populations both inside and outside the managed area.     

Predicting patterns of long‐term adaptation and extinction with population genetics

Population genetics struggles to model extinction; standard models track the relative rather than absolute fitness of genotypes, while the exceptions describe only the short‐term transition from imminent doom to evolutionary rescue. But extinction can result from failure to adapt not only to catastrophes, but also to a backlog of environmental challenges. We model long‐term adaptation to long series of small challenges, where fitter populations reach higher population sizes. The population's long‐term fitness dynamic is well approximated by a simple stochastic Markov chain model. Long‐term persistence occurs when the rate of adaptation exceeds the rate of environmental deterioration for some genotypes. Long‐term persistence times are consistent with typical fossil species persistence times of several million years. Immediately preceding extinction, fitness declines rapidly, appearing as though a catastrophe disrupted a stably established population, even though gradual evolutionary processes are responsible. New populations go through an establishment phase where, despite being demographically viable, their extinction risk is elevated. Should the population survive long enough, extinction risk later becomes constant over time.     

Quantifying the contribution of immigration to population dynamics: a review of methods,evidence and perspectives in birds and mammals

The demography of a population is often reduced to the apparent (or local) survival of individuals and their realised fecundity within a study area defined according to logistical constraints rather than landscape features. Such demographics are then used to infer whether a local population contributes positively to population dynamics across a wider landscape context. Such a simplistic approach ignores a fundamental process underpinning population dynamics: dispersal. Indeed, it has long been accepted that immigration contributed by dispersers that emigrated from neighbouring populations may strongly influence the net growth of a local population. To date however, we lack a clear picture of how widely immigration rate varies both among and within populations, in relation to extrinsic and intrinsic ecological conditions, even for the best‐studied avian and mammalian populations. This empirical knowledge gap precludes the emergence of a sound conceptual framework that ought to inform conservation and population ecology. This review, conducted on both birds and mammals, has thus three complementary objectives. First, we describe and evaluate the relative merits of methods used to quantify immigration and how they relate to widely applicable metrics. We identify two simple and unifying metrics to measure immigration: the immigration rate i_t defined as the ratio of the number of immigrants present in the population at time t + 1 and the total breeding population in year t, and π_t , the proportion of immigrants among new recruits (i.e. new breeders). Two recently developed methods are likely to provide the most valuable data on immigration in the near future: individual parentage (rather than population) assignments based on genetic sampling, and spatially explicit integrated population models combining multiple sources of demographic data (survival, fecundity and population counts). Second, we report on a systematic literature review of studies providing a quantitative measure of immigration. Although the diversity of methods employed precludes detailed analyses, it appears that the number of immigrants exceeds locally born individuals in recruitment for most avian populations (median π_t  = 0.57, N = 45 estimates from 37 studies), a figure twofold higher than estimated for mammalian populations (median π_t  = 0.26, N = 33 estimates from 11 studies). Third, recent quantitative studies reveal that immigration can be the main driver of temporal variation in population growth rates, across a wide array of demographic and spatial contexts. To what extent immigration acts as a regulatory process has however been considered only rarely to date and deserves more attention. Overall, it is likely that most populations benefit from immigrants without necessarily being sink populations. Furthermore, we suggest that quantitative estimates of immigration should be core to future demographic studies and plead for more empirical evidence about the ways in which immigration interacts with local demographic processes to shape population dynamics. Finally, we discuss how to tackle spatial population dynamics by exploring, beyond the classical source–sink framework, the extent to which populations exchange individuals according to spatial scale and type of population distribution throughout the landscape.     

Impact of dispersal status on estimates of local population growth rates in a Temminck's stint Calidris temminckii population

Due to different costs and benefits associated with dispersal and philopatry, life history traits of immigrants and philopatric individuals may differ. Despite of the apparent effects, dispersal status is only rarely considered in analyses of population dynamics. We analysed whether dispersal status explains variation in life history traits of an endangered Temminck's stint Calidris temminckii population breeding at the Baltic Sea. We also estimated the impact of immigration and dispersal status on the population growth rate (λ) with a population matrix model, in which immigrants and philopatric individuals are separated to their own stages. We found that philopatric individuals had a higher apparent survival than immigrants in both sexes. In reproductive parameters, variation due to dispersal status was not clear. Nests incubated by philopatric individuals survived better than those of immigrants, but this did not translate in hatchling production per breeding attempt. Models described a sink population in which the inclusion of both immigration rate to the population and the dispersal status of individuals into the model increased estimates of λ. When the better success of philopatric individuals was considered, the population growth appeared more stable (λ=0.972). If this was not taken into account, λ implied a strong decline (λ=0.911). The results support the hypothesis that immigrants exhibit lower components of lifetime reproductive success and therefore contribute less to population growth and the gene pool than local recruits. While we cannot distinguish whether this difference reflects higher mortality or permanent emigration, the latter explanation seems more plausible. Our results highlight the importance of considering immigration and dispersal status in population modelling. In the case of the endangered study population, the results implied that management options directed to improve local recruitment would be a profitable option.     

Demography in relation to population density in two herbivorous marsupials: testing for source–sink dynamics versus independent regulation of population size

We compared demography of populations along gradients of population density in two medium-sized herbivorous marsupials, the common brushtail possum Trichosurus vulpecula and the rufous bettong Aepyprymnus rufescens, to test for net dispersal from high density populations (acting as sources) to low density populations (sinks). In both species, population density was positively related to soil fertility, and variation in soil fertility produced large differences in population density of contiguous populations. We predicted that if source–sink dynamics were operating over this density gradient, we should find higher immigration rates in low-density populations, and positive relationships of measures of individual fitness—body condition, reproductive output, juvenile growth rates and survivorship—to population density. This was predicted because under source–sink dynamics, immigration from high-density sites would hold population density above carrying capacity in low-density sites. The study included 13 populations of these two species, representing a more than 50-fold range of density for each species, but we found that individual fitness, immigration rates and population turnover were similar in all populations. We conclude that net dispersal from high to low density populations had little influence on population dynamics in these species; rather, all populations appeared to be independently regulated at carrying capacity, with a balanced exchange of dispersers among populations. These two species have suffered recent reductions in range, and they are ecologically similar to other species that have declined to extinction in inland Australia. It has been argued that part of the cause of the vulnerability of species like these is that they exhibit source–sink dynamics, and disturbance to source habitats can therefore cause large-scale population collapses. The results of our study argue against this interpretation.     

Finite metapopulation models with density-dependent migration and stochastic local dynamics

The effects of small density-dependent migration on the dynamics of a metapopulation are studied in a model with stochastic local dynamics. We use a diffusion approximation to study how changes in the migration rate and habitat occupancy affect the rates of local colonization and extinction. If the emigration rate increases or if the immigration rate decreases with local population size, a positive expected rate of change in habitat occupancy is found for a greater range of habitat occupancies than when the migration is density-independent. In contrast, the reverse patterns of density dependence in respective emigration and immigration reduce the range of habitat occupancies where the metapopulation will be viable. This occurs because density-dependent migration strongly influences both the establishment and rescue effects in the local dynamics of metapopulations.     

ROLE OF SEX AND MIGRATION IN ADAPTATION TO SINK ENVIRONMENTS

Understanding the effects of sex and migration on adaptation to novel environments remains a key problem in evolutionary biology. Using a single‐cell alga Chlamydomonas reinhardtii, we investigated how sex and migration affected rates of evolutionary rescue in a sink environment, and subsequent changes in fitness following evolutionary rescue. We show that sex and migration affect both the rate of evolutionary rescue and subsequent adaptation. However, their combined effects change as the populations adapt to a sink habitat. Both sex and migration independently increased rates of evolutionary rescue, but the effect of sex on subsequent fitness improvements, following initial rescue, changed with migration, as sex was beneficial in the absence of migration but constraining adaptation when combined with migration. These results suggest that sex and migration are beneficial during the initial stages of adaptation, but can become detrimental as the population adapts to its environment.     

Self-sustaining populations, population sinks or aggregates of strays: chum (Oncorhynchus keta) and Chinook salmon (Oncorhynchus tshawytscha) in the Wood River system, Alaska

Small populations can provide insights into ecological and evolutionary aspects of species distributions over space and time. In the Wood River system in Alaska, USA, small aggregates of Chinook (Oncorhynchus tshawytscha) and chum salmon (O. keta) spawn in an area dominated by sockeye salmon (O. nerka). Our objective was to determine whether these Chinook and chum salmon are reproductively isolated, self-sustaining populations, population sinks that produce returning adults but receive immigration, or strays from other systems that do not produce returning adults. DNA samples collected from adult chum salmon from 16 streams and Chinook salmon from four streams in the Wood River system over 3 years were compared to samples from large populations in the nearby Nushagak River system, a likely source of strays. For both species, microsatellite markers indicated no significant genetic differentiation between the two systems. Simulations of microsatellite data in a large source and a smaller sink population suggested that considerable immigration would be required to counteract the diverging effects of genetic drift and produce genetic distances as small as those observed, considering the small census sizes of the two species in the Wood River system. Thus, the Wood River system likely receives substantial immigration from neighbouring watersheds, such as the Nushagak River system, which supports highly productive runs. Although no data on population productivity in the Wood River system exist, our results suggest source-sink dynamics for the two species, a finding relevant to other systems where salmonid population sizes are limited by habitat factors.     

Modeling the consequence of increased host tolerance toward avian brood parasitism

Avian brood parasites greatly reduce the reproductive success of their hosts. Empirical studies have demonstrated that some hosts have evolved defenses against parasitism like an ability to recognize and reject parasitic eggs that are dissimilar to their own eggs. Detailed mechanisms of how hosts recognize parasitism still remain unknown, but recent studies have shown that the host’s recognition, in many cases, is based on discordance of the eggs in a clutch, and that hosts are more error-prone when the nest is multiply parasitized, i.e., hosts tend to accept more multiple parasitism than single parasitism. In an area in Hungary, the great reed warbler Acrocephalus arundinaceus, one of the main hosts of the common cuckoo Cuculus canorus, is heavily parasitized and the parasitism rate has been kept at quite a high level for decades. Previous mathematical models suggest that such a high parasitism rate can be maintained because the focal host population behaves as a sink where few hosts can reproduce but immigration from outside replenishes the loss of host reproduction in the sink population. Here, we explore the consequences of the increased host tolerance towards multiple parasitism which has been overlooked in the previous studies using a simple model. Our model analysis shows that the increased host tolerance can dramatically contribute to both the parasite abundance and the parasitism rate being kept at a high level. We suggest that such a host behavior, combined with host immigration, can be an important factor responsible for the observed severe parasitism.     

Augmenting superpopulation capture-recapture models with population assignment data

Summary Ecologists applying capture–recapture models to animal populations sometimes have access to additional information about individuals' populations of origin (e.g., information about genetics, stable isotopes, etc.). Tests that assign an individual's genotype to its most likely source population are increasingly used. Here we show how to augment a superpopulation capture–recapture model with such information. We consider a single superpopulation model without age structure, and split each entry probability into separate components due to births in situ and immigration. We show that it is possible to estimate these two probabilities separately. We first consider the case of perfect information about population of origin, where we can distinguish individuals born in situ from immigrants with certainty. Then we consider the more realistic case of imperfect information, where we use genetic or other information to assign probabilities to each individual's origin as in situ or outside the population. We use a resampling approach to impute the true population of origin from imperfect assignment information. The integration of data on population of origin with capture–recapture data allows us to determine the contributions of immigration and in situ reproduction to the growth of the population, an issue of importance to ecologists. We illustrate our new models with capture–recapture and genetic assignment data from a population of banner‐tailed kangaroo rats Dipodomys spectabilis in Arizona.     

Empirical evidence for source–sink populations: a review on occurrence,assessments and implications

Assessing the role of local populations in a landscape context has become increasingly important in the fields of conservation biology and ecology. A growing number of studies attempt to determine the source–sink status of local populations. As the source–sink concept is commonly used for management decisions in nature conservation, accurate assessment approaches are crucial. Based on a systematic literature review of studies published between 2002 and 2013, we evaluated a priori predictions on methodological and biological factors that may influence the occurrence of source or sink populations. The review yielded 90 assessments from 73 publications that included qualitative and quantitative evidence for either source or sink population(s) for one or multiple species. Overall, sink populations tended to occur more often than source populations. Moreover, the occurrence of source or sink populations differed among taxonomic classes. Sinks were more often found than sources in mammals, while there was a non‐significant trend for the opposite to be true for amphibians. Univariate and multivariate analyses showed that the occurrence of sources was positively related to connectivity of local populations. Our review furthermore highlights that more than 25 years after Pulliam's widely cited publication on ‘sources, sinks, and population regulation’, in‐depth assessments of the source–sink status of populations based on combined consideration of demographic parameters such as fecundity, survival, emigration and immigration are still scarce. To increase our understanding of source–sink systems from ecological, evolutionary and conservation‐related perspectives, we recommend that forthcoming studies on source–sink dynamics should pay more attention to the study design (i.e. connectivity of study populations) and that the assessment of the source–sink status of local populations is based on λ values calculated from demographic rates.     

Impacts of environmental variability in open populations and communities: "inflation" in sink environments

Ecological communities are typically open to the immigration and emigration of individuals, and also variable through time. In this paper we argue that interesting and potentially important effects arise when one splices together spatial fluxes and temporal variability. The particular system we examine is a sink habitat, where a species faces deterministic extinction but is rescued by recurrent immigration. We have shown, using a simple extension of the canonical exponential growth model in a time-varying environment, that variation "inflates" the average abundance of sink populations. We can analytically quantify the magnitude of this effect in several special cases (square-wave temporal variation and Gaussian stochastic variation). The inflationary effect can be large in "intermittent" sinks (where there are periods with positive growth), and when temporal variation is strongly autocorrelated. The effect appears to be robust to incorporation of demographic stochasticity (due to discrete birth-death-immigration processes), and to direct density dependence. With discrete generations, however, one can observe a wide range of effects of temporal variation, including depression as well as inflation. We argue that the inflationary effect of temporal variation in sink habitats can have important implications for community structure, because it can increase the average abundance (and hence local impacts) of species that on average are being excluded from a local community. We illustrate the latter effect using a familiar model of exploitative competition for a single limiting resource. We demonstrate that temporal variation can reverse local competitive dominance, even to the extent of allowing an inferior competitor maintained by immigration to exclude a competing species that would be locally superior in a constant environment.     

The demographic drivers of local population dynamics in two rare migratory birds

The exchange of individuals among populations can have strong effects on the dynamics and persistence of a given population. Yet, estimation of immigration rates remains one of the greatest challenges for animal demographers. Little empirical knowledge exists about the effects of immigration on population dynamics. New integrated population models fitted using Bayesian methods enable simultaneous estimation of fecundity, survival and immigration, as well as the growth rate of a population of interest. We applied this novel analytical framework to the demography of two populations of long-distance migratory birds, hoopoe Upupa epops and wryneck Jynx torquilla, in a study area in south-western Switzerland. During 2002–2010, the hoopoe population increased annually by 11%, while the wryneck population remained fairly stable. Apparent juvenile and adult survival probability was nearly identical in both species, but fecundity and immigration were slightly higher in the hoopoe. Hoopoe population growth rate was strongly correlated with juvenile survival, fecundity and immigration, while that of wrynecks strongly correlated only with immigration. This indicates that demographic components impacting the arrival of new individuals into the populations were more important for their dynamics than demographic components affecting the loss of individuals. The finding that immigration plays a crucial role in the population growth rates of these two rare species emphasizes the need for a broad rather than local perspective for population studies, and the development of wide-scale conservation actions.     

Energy and spatial order in niche and community

The concept of a varying evolutionary tempo that is regulated by energy was first suggested in the 1950s. It was based on the observation that energy‐rich habitats appear to be the centres of evolutionary change, producing novel characters more frequently and having greater speciation rates. Subsequently, a number of studies have found positive relationships between evolutionary rate and energy. Gradients of energy occur across a range of scales and these have been invoked to explain, for example, higher tropical species richness. Precipitation has also been found to influence evolutionary rate, suggesting that biologically available energy and productivity are the important variables in the relationship rather than solar energy alone. Here, we take the theoretical step of investigating at smaller scales the implications of faster evolutionary tempo where productivity is greater: first at the level of the species population and, subsequently, for the same‐guild co‐habitants in a community. To facilitate, this we begin by applying the concept of gradients of available energy, from more productive to less productive sectors within the species niche, in the context of source‐sink theory. We then propose that a species population will have its highest rate of evolution, for changes that confer a positive selection coefficient throughout the niche, in that sector where it has best adaptive fit and greatest per capita energy flux. We also posit that, where it conferred a shift in fitness, an evolutionary change at niche optimum could subsequently affect conspecific populations occupying lower energy niche sectors via the selection‐mediated dispersal of the apomorphy into the more marginal components of the niche. We then infer that this type of change in a species with a generally higher per capita energy flux might negatively affect adjacent more slowly evolving species (living in less productive peripheral niches) in situations where the increased fitness conferred by a particular apomorphy was relevant to conditions occurring beyond the limits of the progenitor niche hypervolume. We therefore suggest a directional component to evolution at the compressed scales of niche and community level microevolution whereby populations and species occupying more productive conditions have greater tempos of change with attendant enhancements of both their competitive influence and their evolutionary potential. In this manner, previously recorded macro‐evolutionary patterns indicating faster evolution with increased energy at larger scales are interpreted in the context of proposed micro‐evolutionary relationships at intraspecific and locally interspecific scales. © 2013 The Linnean Society of London, Biological Journal of the Linnean Society, 2013, 110 , 696–714.