Contact structure and population immunity shape the selective advantage of emerging variants

Contact structure and population immunity shape the selective advantage of emerging variants | medRxiv /* */ /* */ <!-- <!-- /*! * yepnope1.5.4 * (c) WTFPL, GPLv2 */ (function(a,b,c){function d(a){return"[object Function]"==o.call(a)}function e(a){return"string"==typeof a}function f(){}function g(a){return!a||"loaded"==a||"complete"==a||"uninitialized"==a}function h(){var a=p.shift();q=1,a?a.t?m(function(){("c"==a.t?B.injectCss:B.injectJs)(a.s,0,a.a,a.x,a.e,1)},0):(a(),h()):q=0}function i(a,c,d,e,f,i,j){function k(b){if(!o&&g(l.readyState)&&(u.r=o=1,!q&&h(),l.onload=l.onreadystatechange=null,b)){"img"!=a&&m(function(){t.removeChild(l)},50);for(var d in y[c])y[c].hasOwnProperty(d)&&y[c][d].onload()}}var j=j||B.errorTimeout,l=b.createElement(a),o=0,r=0,u={t:d,s:c,e:f,a:i,x:j};1===y[c]&&(r=1,y[c]=[]),"object"==a?l.data=c:(l.src=c,l.type=a),l.width=l.height="0",l.onerror=l.onload=l.onreadystatechange=function(){k.call(this,r)},p.splice(e,0,u),"img"!=a&&(r||2===y[c]?(t.insertBefore(l,s?null:n),m(k,j)):y[c].push(l))}function j(a,b,c,d,f){return q=0,b=b||"j",e(a)?i("c"==b?v:u,a,b,this.i++,c,d,f):(p.splice(this.i++,0,a),1==p.length&&h()),this}function k(){var a=B;return a.loader={load:j,i:0},a}var l=b.documentElement,m=a.setTimeout,n=b.getElementsByTagName("script")[0],o={}.toString,p=[],q=0,r="MozAppearance"in l.style,s=r&&!!b.createRange().compareNode,t=s?l:n.parentNode,l=a.opera&&"[object Opera]"==o.call(a.opera),l=!!b.attachEvent&&!l,u=r?"object":l?"script":"img",v=l?"script":u,w=Array.isArray||function(a){return"[object Array]"==o.call(a)},x=[],y={},z={timeout:function(a,b){return b.length&&(a.timeout=b[0]),a}},A,B;B=function(a){function b(a){var a=a.split("!"),b=x.length,c=a.pop(),d=a.length,c={url:c,origUrl:c,prefixes:a},e,f,g;for(f=0;f<d;f++)g=a[f].split("="),(e=z[g.shift()])&&(c=e(c,g));for(f=0;f<b;f++)c=x[f](c);return c}function g(a,e,f,g,h){var i=b(a),j=i.autoCallback;i.url.split(".").pop().split("?").shift(),i.bypass||(e&&(e=d(e)?e:e[a]||e[g]||e[a.split("/").pop().split("?")[0]]),i.instead?i.instead(a,e,f,g,h):(y[i.url]?i.noexec=!0:y[i.url]=1,f.load(i.url,i.forceCSS||!i.forceJS&&"css"==i.url.split(".").pop().split("?").shift()?"c":c,i.noexec,i.attrs,i.timeout),(d(e)||d(j))&&f.load(function(){k(),e&&e(i.origUrl,h,g),j&&j(i.origUrl,h,g),y[i.url]=2})))}function h(a,b){function c(a,c){if(a){if(e(a))c||(j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}),g(a,j,b,0,h);else if(Object(a)===a)for(n in m=function(){var b=0,c;for(c in a)a.hasOwnProperty(c)&&b++;return b}(),a)a.hasOwnProperty(n)&&(!c&&!--m&&(d(j)?j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}:j[n]=function(a){return function(){var b=[].slice.call(arguments);a&&a.apply(this,b),l()}}(k[n])),g(a[n],j,b,n,h))}else!c&&l()}var h=!!a.test,i=a.load||a.both,j=a.callback||f,k=j,l=a.complete||f,m,n;c(h?a.yep:a.nope,!!i),i&&c(i)}var i,j,l=this.yepnope.loader;if(e(a))g(a,0,l,0);else if(w(a))for(i=0;i (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0];var j=d.createElement(s);var dl=l!='dataLayer'?'&l='+l:'';j.src='//www.googletagmanager.com/gtm.js?id='+i+dl;j.type='text/javascript';j.async=true;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-P4HH5NV'); Skip to main content Home About Submit ALERTS / RSS Search for this keyword Advanced Search Contact structure and population immunity shape the selective advantage of emerging variants View ORCID Profile Pourya Toranj Simin , Juliana C. Taube , Elisabeta Vergu , Shweta Bansal , Lulla Opatowski , View ORCID Profile Chiara Poletto doi: https://doi.org/10.1101/2025.11.07.25339691 Pourya Toranj Simin 1 Sorbonne Université, INSERM, Institut Pierre Louis d’Epidémiologie et de Santé Publique , Paris, France Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Pourya Toranj Simin Juliana C. Taube 2 Department of Biology, Georgetown University , Washington, DC, United States Find this author on Google Scholar Find this author on PubMed Search for this author on this site Elisabeta Vergu 3 Université Paris-Saclay, INRAE, MaIAGE , 78350 Jouy-en-Josas, France Find this author on Google Scholar Find this author on PubMed Search for this author on this site Shweta Bansal 2 Department of Biology, Georgetown University , Washington, DC, United States Find this author on Google Scholar Find this author on PubMed Search for this author on this site Lulla Opatowski 4 Epidemiology and Modelling of Antibiotic Evasion Unit, Institute Pasteur , 75475 Paris Cedex 15, France 5 CESP, UMR1018, Université de Versailles Saint Quentin , Inserm, Paris Saclay Find this author on Google Scholar Find this author on PubMed Search for this author on this site Chiara Poletto 6 Department of Molecular Medicine, University of Padova , 35121 Padova, Italy Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Chiara Poletto For correspondence: chiara.poletto{at}unipd.it Abstract Full Text Info/History Metrics Supplementary material Data/Code Preview PDF Abstract Epidemics are shaped by the interplay between host and pathogen population characteristics, which are themselves intertwined. In particular, host behavior and immunity profile shape pathogen population structure by affecting both the likelihood of new variant emergence and its subsequent dynamics. Theoretical studies provided a fragmented description of this complex dynamical dependency, and the empirical evidence is limited. The SARS-CoV-2 pandemic presents an unprecedented opportunity to study the emergence of new variants. The relative growth of emerging variants over the resident ones, i.e., the selection coefficient, showed spatiotemporal variations that could be associated with population immunity and the mean and dispersion of contacts, which varied greatly according to epidemic intensity and human response. We first investigated the impact of these three features on the selection coefficient using a stochastic network-based model of new variant emergence, which incorporates tunable connectivity and heterogeneity. Results systematically chart the parameter space, uncover the boundaries of previously known associations, and quantify their strength. The mean number of contacts was positively associated with the selection coefficient, the effect being more robust for low immune-escape variants. The impact of immunity diminished as immunity increased. Importantly, greater contact dispersion slowed down the spread of variants lacking immune escape, but this effect quickly reversed once immune escape became non-zero. We then analysed the emergence of the SARS-CoV-2 Alpha variant in the United States at the state level, examining the association of the selection coefficient with the three population features under study, reconstructed from serological, vaccination, and contact survey data. Regression analyses revealed a strong effect of population characteristics. Comparing empirical trends with model predictions showed consistency and suggested that the selection coefficient was more affected by contact statistics than by immunity. These results shed light on how human population structure mediates variant dynamics and help interpret the heterogeneity observed in variant emergence. Introduction Epidemics are shaped by the population structure of both hosts and pathogens. Host characteristics, such as population immunity and contact patterns at different scales, drive population-level epidemic propagation by affecting the epidemic growth rate and final size [ 1 , 2 ]. At the same time, pathogen populations often comprise multiple co-circulating variants with diverse epidemiological traits, including transmissibility, time course of infectivity, and antigenicity [ 3 – 6 ]. These differences influence both short-term dynamics and long-term evolutionary trajectories. Host and pathogen population structure are thus essential factors to account for in epidemic response, from the short-term planning of social restrictions to the long-term design of vaccines and antimicrobials [ 7 – 9 ]. Still, the effects of host and pathogen population structure on the epidemic dynamics are intertwined. The structure of host contacts and pathogen transmissibility jointly determine the epidemic transmission potential, with greater average and variability of host contacts both favouring large-scale epidemics [ 2 , 10 – 12 ]. Importantly, these population properties also affect the pathogen characteristics, e.g., by altering the conditions under which a new variant emerges [ 5 , 13 – 22 ]. Indeed, theoretical work revealed that the selective advantage of traits, such as increased transmissibility and immune escape, depends on hosts’ contact structure together with pre-existing immunity [ 5 , 13 – 22 ]. For instance, the growth advantage conferred by increased transmissibility is higher at the onset of the epidemic, when a larger fraction of the population is susceptible and the level of transmission is higher [ 23 ]. Higher mixing favours a more transmissible variant [ 22 ], while contact heterogeneity makes its emergence less likely when pre-existing immunity provides full protection against it [ 5 , 15 , 17 ]. Conversely, an immune-escaping variant is favoured under different conditions: its growth advantage increases as population immunity builds [ 19 ], and the transmission threshold for its emergence is reduced in the presence of contact heterogeneity [ 20 , 21 ]. Taken together, these findings highlight a complex interplay of host and pathogen factors, where the relationships among these different associations and the boundaries separating their respective dynamical regimes remain unclear. In addition, while theoretical findings have been validated by empirical observations in some cases [ 5 , 23 – 26 ], their validation is, in general, difficult due to the limited availability of extensive, high-resolution microbiological records. The emergence and spread of SARS-CoV-2 provide an unprecedented case study for examining the interplay between pathogen diversity and human population structure on outbreak dynamics. New, more fit variants (e.g., Alpha, Beta, Gamma, Delta, and Omicron) have repeatedly emerged, triggering rises in cases, hospitalisations, and deaths, and prompting many countries to reinstate restrictions [ 4 , 27 , 28 ]. Virological and epidemiological investigations have shown that these variants had an advantage in transmission over the resident virus, combined with an immune-escape advantage at different degrees – limited for Alpha, more substantial for Beta, Gamma, and Omicron [ 27 , 28 ]. Unprecedented sequencing efforts have enabled monitoring of variant dynamics across space [ 29 , 30 ], allowing the quantification of their growth rates relative to the resident variant – i.e., their selection coefficients – to provide anticipation and inform control [ 4 , 31 ]. However, the same variant has exhibited different selection coefficients across geographic regions. The emergence of Alpha across US states provides a paradigmatic example. The Alpha variant was first identified in the US in late November 2020 [ 32 – 34 ]. In the majority of states, the growth began in late December 2020 and took around 90 days. Figure 1a shows the fraction of Alpha cases among total cases for each US state, based on data obtained through the CoV-Spectrum project [ 35 ]. Figure 1b highlights that the selection coefficient varied from one state to another. Its value in New York, corresponding to the 10% percentile was as low as 0.033 days −1 , while the 90% percentile was 0.063 days −1 and was registered in Arizona (see Methods for the mathematical definition of the selection coefficient and the methodology used to compute it from data). Download figure Open in new tab Fig 1. Alpha variant frequency and selection coefficients across U.S. states. a) Time series of the frequency of Alpha for each of the 50 US states, extracted from CoV-Spectrum [ 35 ] using genomic data from GISAID [ 36 ]. The frequency of Alpha is defined as the incidence of Alpha divided by the total incidence. Time series are smoothed with 21-day sliding window. Thick lines indicate the frequency in the three states where the selection coefficient is 10% percentile (New York, orange line), median (North Carolina, red line), and 90% percentile (Arizona, black line). b) Map showing the selection coefficient by state. The selection coefficient, measures the growth of the emerging variant frequency relative to the resident variant one (more details are provided in the Methods). The fitting method to compute the selection coefficient from the data is provided in the Methods section. Plots of the time series of the raw frequency for each state, together with the fit for the selection coefficient computation, are reported in Figure S1 in the Supporting Information. Similar to the US example above, other studies highlighted spatial heterogeneities in the variant emergence dynamics [ 26 , 31 , 37 , 38 ]. This could be the result of variable human population characteristics. However, this link was investigated only by a few works [ 26 , 37 , 38 ]. Here, we focus on three key host population features, the population immunity, the average level of human social mixing, and its variability; and two variant traits, transmissibility advantage and immune escape. We analyse their role on the selection coefficient, combining extensive network-based model simulations and the analysis of empirical data for Alpha emergence in the US. Results Modelling variant emergence and growth We first provide theoretical predictions on the spread of an emerging variant on a human-to-human contact network. We focus on the drivers of the variant’s growth upon emergence, quantified by the selection coefficient. The frequency of a variant is computed by its incidence divided by total incidence, i.e. , where V ( t ) and W ( t ) are the incidences of the emerging and the resident variant, respectively. The selection coefficient, s ( t ), is then defined as the growth of the emerging variant frequency relative to the resident variant one, (see Methods section for details). We implemented a stochastic individual-based model to simulate the emergence of a variant and its co-circulation with the resident one. As a compromise between simplicity and realism, we group transmission settings in two: within and outside the household ( figure 2a ). Household contacts play a central role in infection propagation due to their high clustering. On the other hand, non-household contacts are the most affected by interventions and adaptive behavioural response. Households are modelled as cliques of variable size. The number of non-household contacts is distributed as a negative binomial distribution, with mean (MD) and standard deviation (SD) explored during the analysis. On top of the network, we simulated the co-circulation of two variants interacting with cross-immunity ( figure 2b ), i.e. the immunity acquired upon infection with one variant confers full protection to the same variant and a certain degree of protection to the other variant. This is governed by the immune evasion or susceptibility rescale parameter, 0 ≤ σ ≤ 1. The case σ = 0 corresponds to no immune evasion (i.e. full cross immunity), while σ > 0 corresponds to some degree of immune evasion. The emerging variant is introduced at a time when a fraction IM of the population has already acquired natural immunity to the resident variant. The co-circulation dynamics that follow are depicted in figure 2 c-e for a specific set of parameters taken as an example – panel c shows the number of cases for each variant, panel d the frequency of the emerging variant, i.e. its incidence relative to the total incidence, and panel e the distribution of the selection coefficient computed from each stochastic run. Other examples of the dynamics are reported in Figure S2 in the SI. Download figure Open in new tab Fig 2. Schematic representation of the model and of the simulation output. a) Modelled contact network. This consists of two layers: the household layer and the social layer. b) Scheme of the 2-strain SIR model. Compartments represent: S (susceptible), I r (Infected with resident variant), I e (infected with emerging variant), R r (recovered from resident variant), R e (recovered from emerging variant), I ( r ) e (infected with emerging variant after recovering from resident variant), I ( e ) r (infected with resident variant after recovering from emerging variant), R r (recovered from both variants). Mathematical symbols reported near the arrows indicate the transition rate: β r and β e indicates the transmissibility of the resident and emerging variant respectively (β e = r β β r ), γ the recovery rate (assumed to be the same for each variant), and σ the reduction in susceptibility due to cross-immunity. c) The incidence of the resident variant, the emerging variant, and the total incidence as a function of time. d) The frequency of the emerging variant, defined as the proportion of the incidence of the new variant to the total incidence. The vertical dashed lines represent the time window used to compute the selection coefficient. The shaded area shows the 95% confidence interval. e) The distribution of computed selection coefficient of each realisation, representing stocastic fluctuation of the estimated selection coefficient. The vertical orange line represents the mean value of the distribution which is equal to 0.100 (1/day). In the plots of panels c) timing of emergence and d) distribution of selection coefficient, curves corresponding to each realisation are aligned in correspondence with the time of the emerging variant appearance, occurring when the resident variant reaches the desired fraction of immune. Parameters of the model used to generate figures c, d, and e are: mean number of contacts (MD = 6), standard deviation of contacts (SD = 6), immunity level (IM = 15%), immune evasion (σ = 0), transmissibility advantage ( r β = 2), transmissibility of the resident variant β = 0. 0175, recovery rate μ = 0. 143 (1/7) days −1 , and size of the network N = 10 4 .. Statistics is computed over 100 stochastic epidemic realisations for each of the 40 stochastic generations of the network with fixed parameters, for a total of 4000 runs. The complex landscape of viral emergence dynamics We use the simulation set-up described above to systematically quantify the selection coefficient in varying the parameters on human population contacts and immunity – level of immunity (IM), mean (MD) and standard deviation in contacts (SD). In Figures 3 a-c, we plot the selection coefficient when one of the three parameters is varied, while the others are kept fixed. We consider an emerging variant with twice the transmissibility of the resident one, and we compare different values of immune escape. For the range of parameters considered, the selection coefficient is always positive, i.e. the emerging variant spreads more efficiently than the resident one. Panel 3a shows that the trend of the selection coefficient in varying population immunity depends on the degree of immune evasion. For low immune evasion, the selection coefficient decreases with increasing immunity, whereas for immune evasion above a certain threshold, it increases. In both cases, the selection coefficient varies more substantially when immunity is low, while it tends to plateau for high immunity. The simulations highlight that population immunity reduces the advantage of a more transmissible variant if it cannot efficiently reinfect immunised individuals. This finding aligns with predictions from a mean-field model and experimental observations [ 22 , 23 ]. In contrast, as expected, an immune-escaping variant becomes increasingly advantaged over the resident one as population immunity increases, since this expands the pool of individuals who can be infected by the former but not by the latter [ 19 ]. Download figure Open in new tab Fig 3. Impact of population chacareteristics on the selection coefficient of the emerging variant. Panels a, b, and c show the selection coefficient (y-axis) as a function of the immunity level (IM) at the time of emergence, mean (MD) and standard deviation (SD) of the number of contacts, respectively, with the transmissibility advantage of the emerging variant fixed at r β = 2. Colours indicate the assumptions on immune evasion (σ). Panel (a) corresponds to fixed values of MD = 8 and SD = 8; panel (b) to fixed SD = 8 and IM = 10%; and panel (c) to fixed MD = 8 and IM = 10%. Panel (d) shows the selection coefficient (y-axis) as a function of standard deviation SD for a fixed immune evasion (σ = 0.05), mean of the number of contacts (MD = 8) and immunity level (IM=10%), where colours represent different transmissibility rescaling factors of the emerging variant. Points are means over 100 stochastic epidemic realisations for each of the 40 stochastic generations of the network with fixed parameters, for a total of 4000 runs. Dots are the mean and the error bars the standard errors (smaller than the size of the dots for almost all cases). Other parameters are: transmissibility of the resident variant β = 0. 0175, recovery rate μ = 0. 143 (1/7) days −1 , and size of the network N = 10 4 In panel b we look at variations in the selection coefficient with the mean degree of the population network. For low immune evasion, the average selection coefficient increases with MD ( figure 3 b ). It becomes, instead, non-monotonic when immune evasion is high, decreasing for low MD values and then increasing. The increase in population mixing, thus, favours disproportionately the more transmissible variant compared to the resident one [ 22 ]. This is also true for a variant with strong immune escape, but only at a high mixing level. The selection coefficient, as a function of SD, for fixed MD and IM values, is illustrated in Figure 3c . The dispersion in the number of contacts slows down a variant with zero or near-zero immune escape, while accelerating an immune-escaping one. Contacts’ dispersion was found to have a detrimental role on new variant emergence in previous studies that were focusing on the probability of emergence [ 5 , 15 , 17 ]. Here, we found a similar effect on the selection coefficient. This behaviour can be attributed to the effect of hubs. Hubs can act as super-spreaders or super-blockers. Given their high number of connections, hubs are among the first to be infected, further transmit the infection to a large number of individuals, and are then among the first to become immune. When a variant with no immune escape emerges, hubs that are already immunised will act as super-blockers. Conversely, a variant able to reinfect immunised hubs can take advantage of their super-spreading role. Of note, Figure 3c shows that the super-blocker effect of hubs is fragile, and it is predicted to happen only for full or nearly full cross-immunity. The threshold level of immune escape above which this effect disappears is very low, i.e. around 0.1 for r β = 2 ( figure 3 c ), while it is even lower for smaller r β . The case of immune escape 0.05 is analysed in depth in Figure 3d , where different values of r β are compared. For this value of immune escape, the slowdown with increasing contact dispersion occurs for high transmissibility rescaling r β . For r β = 1. 25 or 1.5 the selection coefficient does not show a clear trend across varying SD. We conducted a systematic exploration of the parameters IM, MD, and SD, comparing the case of low immune evasion (σ = 0) with high immune evasion (σ = 0.6) (details in Figure S3). For σ = 0, the increase of the selection coefficient with MD is consistent across all values of immunity and SD tested. The decline of the selection coefficient vs. SD, instead, happens only if immunity is sufficiently high. For low values of immunity (below 10%, with all other parameters as in the figure), the trends become non-monotonic. For σ = 0. 6, the non-monotonic trend of the selection coefficient vs. MD is consistent across all IM and SD tested, with MD corresponding to the minimum shifting with IM and SD. The average growth of the emerging variant incidence shows in most cases supercritical spread (see Figure S4). However, for certain parameter values, the spread remains subcritical, with a negative average growth rate and a low probability of positive growth. We find that a variant can still be advantaged over the resident one (i.e., have a positive selection coefficient) even when its spread is subcritical. Analysis of Alpha emergence across US states We now consider the Alpha spread in the US as a case study and analyse the relationship between the selection coefficient, social contact statistics and population immunity. Data on seroprevalence [ 39 ], vaccination [ 40 ], and contact data statistics [ 41 ] were gathered for all US states during the period of SARS-CoV-2 Alpha emergence and growth, between late December 2020 and late April 2021 (see Methods for a detailed description of the data). Figure 4a and b show the evolution in time of the contacts’ mean and standard deviation for each state during the period of Alpha spread [ 41 ]. Both the mean and standard deviation of contacts increased during this time and varied significantly between states. We define the set of covariates MD and SD as the averages over time of the mean and standard deviation, respectively, for each state. Their values ranged between 4.1 and 10.9 (mean value 8.0) for MD and between 8.5 and 15.2 (mean value 13.0) for SD. By combining seroprevalence, vaccination data, and vaccine effectiveness estimates we estimated population immunity at the time of Alpha’s emergence for each state, which defines the IM covariate. The mean value across states was 18.6%. Immunity estimated for Vermont was extraordinarily low (1.1%), as the state was rural, with limited connections with urban hubs, and had a strong public health response. On the other hand, the highest immunity was estimated for Ohio and was as high as 30.7%. In Figure 4 c and d, we show the distribution of MD, SD and IM, as well as their correlation. MD was strongly correlated with SD. This may be explained by the fact that the adaptive behavioural response and the public health interventions, which caused spatiotemporal contact variations during COVID-19, affected different settings (e.g. workplace, social events) and thus they likely altered both the average and the dispersion in the number of contacts. MD was also correlated with IM, likely due to the relatively stable ranking of states according to contact activity: states with higher mixing likely experienced more transmission, leading to more cumulative infections and, thus, greater built-in immunity by the time Alpha was introduced. Download figure Open in new tab Fig 4. Data from the Alpha variant emergence in the US and statistical analysis. a) Time series of the mean number of contacts or mean degree, and b) time series of the standard deviation of degree. Black curves represent the value of each state over time, and the blue curve is the smoothed curve of the average value. c) The scatter plot of the time-averaged mean degree (MD) against the time-averaged standard deviation (SD) during the period of Alpha emergence, together with their histograms. d) The scatter plot of the time-averaged mean degree (MD) during the period of Alpha emergence and the estimated effective immunity (IM) at the time of the Alpha variant’s emergence, together with their histograms. e) Color map of MD for each US state. Figures f-h show the scatter plots of the fitted selection coefficients depending on the state characteristics: f) selection coefficient and MD. g) selection coefficient and SD. h) selection coefficient and IM. The selection coefficient was fitted over the same period used for defining the time-averaged mean and standard deviation of contacts (MD and SD, respectively). The map of figure 4e allows for a visual comparison of the average number of contacts by state with the state’s selection coefficient of Figure 1 b, highlighting that the two quantities have a similar geographical pattern. For instance, all states on the West Coast and many states on the East Coast (especially the Northeast) had both selection coefficient and average number of contacts below the average. On the other hand, Oklahoma, Nebraska, South Dakota, Arkansas, Louisiana, and Mississippi were among the states with both selection coefficient and average number of contacts above the average. Beyond the visual comparison, the scatter plots in Figure 4f–h show that the selection coefficient was positively correlated with all three covariates. Its correlation with MD and SD is substantially higher than the one with IM . This suggests that contact statistics could affect the selection coefficient. We used ridge multivariate regression to disentangle the relative role of the three covariates while accounting for collinearity (see Methods). The coefficients obtained identified a stronger effect of MD (5.26×10 −2 ), followed by SD (4.96×10 −2 ), and IM (2.87×10 −2 ). Consistency between empirical evidence and theoretical findings We now discuss whether these results are compatible with the predictions of the theoretical model. The transmissibility advantage of Alpha is estimated at around 50% ( r β = 1.5) [ 33 , 34 , 42 ]. Virological studies reported some, though limited, immune evasion for Alpha [ 28 ]. A small reduction in vaccine efficacy was observed [ 43 ]. Protection against Alpha from prior infection by the historical variant was estimated at 90% [ 44 ]. Another study, however, found no evidence that reinfections were more frequent with Alpha than with the pre-existing variant [ 45 ]. These findings are compatible with zero or very low values of immune evasion in our model. Within this parameter range, the theoretical model predicts a positive association between the selection coefficient and MD, which is consistent with the empirical trend. Model predictions regarding the selection coefficient’s association with SD could also be compatible with observations if we assume low but non-zero immune escape for which the negative effect of SD is absent. To test the consistency between the model and the data, in this parameter regime, we ran simulations with our model with as input the triplets of MD, SD, and IM values estimated for each US state, along with the average household size of each state. We then took transmissibility advantage r β = 1.5, and immune evasion σ = 0. 1. The simulated and empirical values of the selection coefficient across states were positively correlated (Pearson correlation coefficient equal to 0.55 p < 10 −4 , after removing Vermont, which was an outlier, Figure S5). In line with the data, the simulated selection coefficient showed positive trends with all covariates, MD, SD, and IM – Pearson correlations 0.84, p = 10 −4 , and 0.81, p < 10 −4 , and 0.4, p < 0. 02, respectively. Simulations with lower values of immune escape, i.e. σ = 0.05, were also correlated with empirical trends. When we set to σ = 0, instead, the model failed to reproduce the empirical trends, and its outputs were not correlated with the data. Discussion The COVID-19 pandemic has brought to the forefront the consequences of viral evolution for epidemic preparedness. The emergence of new variants with distinct epidemiological traits has created a rapidly shifting landscape, challenging both anticipation and the planning of interventions. Understanding the drivers of variant emergence and selection dynamics is thus crucial, not only for anticipating long-term viral evolution, but also for the short-term outbreak response. Previous work has shown that the interaction between host and pathogen population structure has complex and multifaceted consequences on viral emergence dynamics. Our contribution to this research effort was twofold. First, we deepened our understanding of these consequences through an extensive numerical reconstruction of the possible dynamical trends in varying host characteristics and pathogen traits. Second, we analysed real-world data to show that empirical patterns are consistent with theoretical expectations. We focused on two variant traits – the transmissibility advantage and the immune escape – and three host features – the mean and dispersion of social contacts, and the population immunity. We designed a spreading model to capture the effect of these ingredients, maintaining a parsimonious and general description of the disease dynamics to provide an understanding of the basic ecological behaviour. This allowed us to systematically explore the regime shifts driven by these parameters and to better delimit the boundaries of previously identified dynamical regimes. We found that the monotonic increase of the selection coefficient with the mean number of contacts is a robust feature of variants – provided that the immune escape is not high – observed across a wide range of parameter values. By contrast, contact heterogeneity hindered variant emergence only in a narrow parameter region, i.e. near-zero immune escape, high immunity, and strong transmissibility advantage; otherwise, it was found to favor emergence. The choice of variant traits explored in the study was motivated by SARS-CoV-2 Alpha. In particular, we did not explore differences in the generation time between variants. Previous studies have examined this factor, showing a complex interplay with the timescale of host contacts: high mixing levels favour pathogens with shorter generation times, while lower mixing longer generation times [ 16 , 46 , 47 ]. Given the strong changes in individual mixing caused by social restrictions, it has been suggested that such effects could play a role in the case of SARS-CoV-2 variants with altered infectivity profiles. Still, the majority of studies agree that the generation time of Alpha was in line with the one of the historical variant [ 47 , 48 ]. Human face-to-face contacts are important drivers of respiratory infection dynamics. These are, however, difficult to reconstruct from data. Over the last two decades, extensive efforts have produced either detailed, high-resolution records of contact dynamics in specific settings, for small populations, and over short time periods [ 49 – 53 ], or coarse contact statistics of daily interactions obtained from large representative samples of the population at national scales [ 41 , 54 – 56 ]. These data sources provide a fragmented picture of human contact patterns, which complicates epidemic assessment—particularly in situations such as the COVID-19 pandemic, where adaptive behavioural responses and public health interventions caused abrupt changes in these patterns. The COVID-19 pandemic has intensified efforts to reconstruct contact patterns and, more broadly, human behaviour, and has opened a reflection on essential information that needs to be collected for modelling [ 41 , 56 , 57 ]. Here, we show that, in addition to mean contact numbers, their dispersion also plays an important role in determining the dynamics of variant emergence, yet this indicator is rarely investigated in survey studies. It is important to note that to integrate this information into models, a challenge arises from the nature of surveys, which typically record the number of contacts on a single day and thus cannot capture the temporal dynamics of contacts on timescales relevant for infection spread. Univariate analyses of Alpha emergence across U.S. states showed that the selection coefficient correlated with the mean and variability of contacts as well as with population immunity, supporting the role of host population structure in shaping variant dynamics. To account for strong correlations among predictors, particularly between the mean and standard deviation of contacts, we applied multivariate ridge regression, which confirmed positive associations with all three factors, though weaker for immunity. Model simulations, parameterized with Alpha-specific traits and state-level characteristics, reproduced these empirical patterns and showed agreement with observed trends, under the hypothesis of limite dalbeit non-zero immune evasion, which could be compatible with virological findings. Overall, results show an important role of contact statistics. A previous study reported a negative association between the stringency of interventions and the selection coefficient across regions in the UK [ 26 ]. Considering that interventions reduce contact rates, our results are therefore consistent with that work. Here, however, we directly investigated the role of contacts, which, while influenced by interventions, are not solely determined by them, as factors such as adherence and adaptive behavioural response also play a role [ 41 , 58 – 60 ]. The effect of immunity appeared weaker than that of contact patterns. Differently from Alpha, Omicron was marked by strong immune evasion, and a stronger impact of immunity would therefore be expected. Future work could extend the present analysis to Omicron emergence, leveraging model-based extrapolations of human contact statistics that have been validated in previous studies [ 41 ]. However, the complexity of the immunity landscape and the greater change in epidemiology compared to Alpha, would challenge this analysis. It is important to stress that our aim was to qualitatively reproduce trends in the selection coefficient, not to realistically simulate incidence or achieve a precise quantitative match with the data. The stylised model developed here was not designed to capture the full complexity of the COVID-19 epidemic across US states. Furthermore, using survey-based statistics as input to a static network model may introduce bias. Variations in contact numbers observed across individuals on a given day may reflect short-term fluctuations rather than stable behavioural differences. Daily survey data may therefore overestimate heterogeneities that would average out over longer timescales. This was previously found to bias projections on the herd immunity threshold [ 61 – 65 ]. The same effect would limit here the ability of the model to simultaneously capture ecological dynamics and incidence trends, which is beyond the scope of the work. The fact that the modelled trends of the selection coefficient nonetheless correlate with empirical observations, under reasonable assumptions, indicates that the qualitative behaviour described here is robust to many dynamic complexities not explicitly represented in the model. This study is subject to limitations. First, we modeled immunity in a simplified, polarized way and neglected homologous reinfection. While this assumption was reasonable during the early phase of COVID-19, reinfections – often milder – became increasingly common within a year due to immune waning and partial cross-immunity. Second, we did not account for behavioral factors other than contacts, such as use of mask and adoption of hygiene measures. Also we could not distinguish between indoor and outdoor contacts. Third, while we investigated contact heterogeneities, the model does not incorporate the socioeconomic and demographic drivers of such heterogeneities—e.g., age, education, income—which have been shown to shape patterns of COVID-19 spread [ 41 , 66 , 67 ]. Explicitly accounting for their role in variant dynamics is an important future direction. Fourth, our analysis of Alpha spread across the US was conducted at the state level. Differences in contact statistics across states were largely driven by marked differences in urbanisation [ 41 ]. Analyses at higher spatial resolution could better account for this factor and reduce noise; however, sequencing coverage was insufficient to support analyses at finer scales. Finally, the model does not account for spatial structure or the dynamics of introduction of cases from outside the US, factors that could potentially affect the advantage of a variant [ 68 ]. This study demonstrates that social contact patterns and immunity levels significantly shape the competitive advantage of emerging pathogen variants, with contact heterogeneity playing opposing roles depending on a variant’s immune escape capacity. From a public health perspective, these findings suggest that surveillance and intervention strategies should be tailored to local population characteristics, with contact reduction measures potentially having differential effects on variant competition. Understanding these dynamics could inform targeted vaccination strategies and enhance variant emergence prediction when integrated with real-time behavioral surveillance systems that capture evolving contact patterns. Methods Selection coefficient The ecological dynamics of the emerging variant can be reconstructed by looking at its frequency, p ( t ), i.e. the percentage of incidence cases attributed to it: , where V ( t ) and W ( t ) are the incidences of the emerging and the resident variant, respectively. The selection coefficient, s ( t ), measures the growth of the emerging variant frequency relative to the resident variant one, When s ( t ) is constant in time, the emerging variant frequency p ( t ) follows a logistic function, where p ( t ) increases if s > 0 and decreases when s < 0. We estimate the selection coefficient by fitting the logistic function p ( t ) to the frequency curve, for both the simulated and real frequency - see below for details on the definition of the time window for the fit. We regress a binomial Generalized Linear Model (GLM) with a logit link function in the R programming language to estimate the parameters of the logistic function and their confidence intervals. Computational model of two-variant infection dynamics Modelling human-to-human contact network We model a two-layer contact network with N nodes. The first layer accounts for household contacts, while the second layer accounts for non-household contacts (e.g. occurring in the workplace, school or community). To construct the household layer, we distribute the nodes in households of different sizes and consider a clique, i.e. a fully connected graph, inside each household. We assume that the household size follows a zero-truncated Poisson distribution [ 69 ], . This leads to a Poisson distribution for the nodes’ degree in the household layer, , where the subscript h indicates the household layer. For the non-houseshold layer, that we label o. h ., we assume that the probability of an individual having a degree k follows a negative binomial distribution shifted by 1 (to avoid agents with no contacts in this layer), , with k ≥ 1 and free parameters n and p , that are related to the mean degree and the standard deviation of the degree as follow: . Once assigned a degree to each node based on this distribution, we connect the stubs using the configuration model to avoid degree-degree correlations. We parametrised the networks as follows. The network size was N = 10 4 , a compromise between being reasonably large and maintaining computational times to feasible levels. We set the household-size distribution parameter to λ = 2, yielding an average household size of 2.3 – similar to values observed in European countries and the US [ 70 , 71 ]. We then varied the parameters of the non-household layer to explore MD and SD within ranges estimated from US (see Figure 1 ) and compatible with European surveys as well – namely, MD in [ 4 , 16 ] and SD in [ 3 , 16 ]. Modelling transmission We use a susceptible-infected-recovered model to simulate the spread of two variants in the network. In this model, the population is distributed into the compartments, S (Susceptible), I (Infected), and R (Recovered), with respect to each variant. We assume that co-infection is not possible, i.e. no individuals can be infected with more than one strain at the same time. Therefore, Individuals are divided into eight compartments: fully susceptible ( S ), infected by the resident variant ( I r ), infected by the emerging variant ( I e ), recovered from the resident variant and still susceptible to the emerging variant ( R ( r ) ), recovered from the resident variant and infected by the emerging variant ( I ( r ) e ), recovered from the emerging variant and still susceptible to the resident variant ( R ( e ) ), recovered from the emerging variant and infected by the resident variant ( I ( e ) r ), and recovered from all variants ( R ). Susceptible agents can be infected by their infectious neighbors with transmission rate per contact β r and β e for resident and emerging variant, respectively. We quantify the advantage in transmission of the emerging variant with the parameter . The cross-immunity parameter σ j,i represents the reduction in susceptibility to the variant i due to the acquired immunity to j . Namely, individuals previously infected by the variant j can contract the infection by i with the per-contact rate σ j,i β i , where σ i,j = 0 and σ i,j = 1 represent full cross-immunity and no cross-immunity, respectively. We assume the symmetry condition applies, i.e. σ i,j = σ j,i = σ, as depicted in Figure 2 b . Infectious individuals recover with a rate γ, assumed to be the same for the two variants. The transmission model was parametrised to consider a scenario similar to the emergence of a SARS-CoV-2 variant. The recovery rate was set to γ = 1/(7 days ). The per-contact transmission rate of the resident variant β r was set to 0. 0175 per day. The basic reproductive ratio of the resident variant corresponding to the input MD and SD can be computed from the equation and falls within the range [0.98 - 4.82]. Variant’s advantages in transmission and immune escape were varied, considering values of r β between 1.25 and 2 and σ between 0 and 0.6. These values were in the ballpark of estimates for SARS-CoV-2 variants. Details of the numerical simulations Simulations are stochastic and discrete-time. At each simulation time-step, corresponding to one day, we simulate the two processes: - Infection: Every agent infected with the variant i can transmit the infection to neighbouring nodes that are susceptible to i with probability β i and σ i,j β i , according to whether they have no infection history or a history of j infection, respectively. - Recovery: Every infectious individual recovers from the infection at a probability γ. To simulate the spreading of the variants, we initially infect 50 random individuals with the resident variant and set the state of the rest of the population to S . The value of 50 was chosen to avoid the initial epidemic extinction from occurring too often. Then we let the simulation continue until the time when the proportion of recovered by the resident variant reached IM. At this point, the emerging variant is seeded by randomly infecting 50 susceptible individuals. The simulation continues until no infectious agents remain. For each set of parameters of the contact network, we generate an ensemble of 40 random networks. For each generated network, we then run 100 stochastic realizations of the epidemic. For each stochastic run, we computed the selection coefficient by fitting the logistic curve to the emerging variant frequency (see the dedicated section at the beginning of the Methods section). The time window used for the fit was defined from the emergence of the variant until either the time of the incidence peak or 7 days after the emergence, whichever occurs later. In the Supplementary Information we also analyzed the growth rate of the emerging variant’s incidence. The growth rate at time step t is defined as G ( t ) = ln( V ( t + 1)) − ln( V ( t )), where V ( t ) is the incidence of the emerging variant. We computed the time-averaged growth rate as the average G ( t ) over seven days from the emergence day. Both the growth rate and the selection coefficient were computed for each stochastic run. Figure 2 , 3 and other figures in the SI show statistics over all runs. Empirical analysis of Alpha spread across US states SARS-CoV-2 Alpha’s frequency and selection coefficient The data used to estimate Alpha frequency are provided by the CoV-Spectrum project [ 35 ], based on genomic sequences and metadata retrieved from GISAID [ 36 ]. The Alpha variant took about 90 days to reach its peak incidence in each state. Therefore, we defined the time window of Alpha growth as the period with a length of 12 weeks (84 days) centred around the midpoint date when the Alpha frequency reaches half of its maximum value. In the cases in which the Alpha was peaking and starting to decline before the midpoint plus six weeks, the date of the peak frequency was taken as the end of the window (Figure S1). Face-to-Face contact statistics For all US states, the number of non-household contacts was obtained from the U.S. COVID-19 Trends and Impact Survey (Delphi Research Group at Carnegie Mellon University in partnership with Facebook). The dataset is described and analysed in [ 41 ]. Briefly, in the survey, contact was defined as a face-to-face conversation or physical contact. Contacts are aggregated on a weekly basis for each U.S. state, and the mean and standard deviation over the number of contacts are computed after removing outlier responses with the number of contacts above the 95th percentile. The data spans the period from April 27, 2020, to April 26, 2021. Weekly mean and standard deviation in the number of contacts are then averaged during the time window of Alpha growth defined above to obtain the covariates, MD and SD. Population’s immunity For each US state, we estimated the effectivve population immunity against SARS-CoV-2 resident variant using seroprevalence [ 39 ] and vaccination [ 40 ] data provided by the Centers for Disease Control and Prevention (CDC). Assuming z ( t ) is the proportion of the population infected before the time of Alpha emergence, v p ( t ) is partial vaccination coverage (i.e. one dose of a two-dose vaccine schedule) and v c ( t ) is complete vaccination coverage, then effective immunity can be estimated as [ 72 ]: Where ϵ z , ϵ p and ϵ c are natural immunity efficacy, partial vaccination efficacy, and complete vaccination efficacy, respectively. We used the following values informed by the literature [ 72 ]: ϵ z = 1, ϵ p = 0. 6 and ϵ c = 0. 9. The covariate IM of each state used in the regression analysis below was defined as IM ( t ), where t is the starting date of the window of Alpha growth. Regression analysis To disentangle the association of MD, SD, and IM with the selection coefficient (S), we performed ridge regression. Ordinary least squares regression can be unstable in the presence of multicollinearity (as observed between MD, SD, and IM) and ridge regression addresses this issue by imposing an L2-norm penalty on all model coefficients, shrinking them towards zero while keeping them in the model. We used a Gaussian family and z-normalized all covariates. We performed k-fold cross-validation using the R package glmnet to choose a penalty tuning parameter, λ, which optimizes the bias-variance tradeoff. We select the maximum value of λ so that the mean-squared error is within one standard deviation of the minimum [ 73 ]. Our results are qualitatively robust to an alternative choice of λ that minimizes the mean squared error. North Dakota is omitted from this analysis due to missing immunity level (IM) data. Standard errors are not reported as confidence intervals do not have usual frequentist properties in ridge regression. Data availability The data used to estimate Alpha variant frequencies were obtained from the CoV-Spectrum project [ 35 ] (publicly available from https://cov-spectrum.org/ ), which is based on genomic sequences and metadata retrieved from GISAID ( https://gisaid.org ) [ 36 ]. Seroprevalence [ 39 ] and vaccination [ 40 ] data were obtained from the Centers for Disease Control and Prevention (CDC). Data on the average household size for each U.S. state were retrieved from Statista ( https://www.statista.com/statistics/242265/average-size-of-us-households-by-state/ ) [ 74 ]. The number of non-household contacts was obtained from the U.S. COVID-19 Trends and Impact Survey (Delphi Research Group at Carnegie Mellon University in partnership with Facebook). The weekly, state-level means and variances of the number of contacts used in this study, along with scripts for running simulations, generating raw simulation data, and estimating selection coefficients, are publicly available on GitHub at https://github.com/PouryaTS/MultiStrain_Network . Supporting information S1 Appendix. Supplementary text and figures. Competing interests The authors have declared that no competing interests exist. Author contributions Conceptualization: EV, SB, LO, CP. Data curation: PTS, JCT, SB, LO, CP. Formal analysis: PTS, JCT, EV, SB, LO, CP. Investigation: PTS, JCT, EV, SB, LO, CP. Methodology: PTS, JCT, EV, SB, LO, CP. Software: PTS. Supervision: SB, CP. Validation: PTS, JCT, SB, LO, CP. Visualization: PTS. Writing – original draft: PTS, CP. Writing – review & editing: PTS, JCT, SB, LO, CP. Acknowledgments We thank Pierre-Yves Boëlle and Mircea Sofonea for usefull discussion. We acknowledge financial support by the Municipality of Paris through the programme Emergence(s) to C.P. and P.T.S.; Cariparo Foundation through the program Starting Package to C.P.; Department of Molecular Medicine through the program SID from BIRD funding to C.P. We gratefully acknowledge support from the ANRS MIE Network on Modelling Infectious Diseases, which provided a travel grant to P.T.S. for a research visit. The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. Footnotes ↵ † Dr. Elisabeta Vergu has passed away during the completion of this work. Dr. Vergu has contributed to conceiving and designing the study, and developing the network-based multi-variant computational model. References 1. ↵ Matt J. Keeling PR. Modeling Infectious Diseases in Humans and Animals . Princeton University Press ; 2008 . 2. ↵ Pastor-Satorras R , Castellano C , Van Mieghem P , Vespignani A. Epidemic processes in complex networks . Rev Mod Phys . 2015 ; 87 : 925 – 979 . doi: 10.1103/RevModPhys.87.925 OpenUrl CrossRef 3. ↵ Bedford T , Riley S , Barr IG , Broor S , Chadha M , Cox NJ , et al. Global circulation patterns of seasonal influenza viruses vary with antigenic drift . Nature . 2015 ; 523 : 217 – 220 . doi: 10.1038/nature14460 OpenUrl CrossRef PubMed 4. ↵ Volz E. Fitness, growth and transmissibility of SARS-CoV-2 genetic variants . Nat Rev Genet . 2023 ; 1 – 11 . doi: 10.1038/s41576-023-00610-z OpenUrl CrossRef 5. ↵ Pinotti F , Fleury É, Guillemot D , Böelle P-Y , Poletto C. Host contact dynamics shapes richness and dominance of pathogen strains . PLOS Comput Biol . 2019 ; 15 : e1006530 . doi: 10.1371/journal.pcbi.1006530 OpenUrl CrossRef PubMed 6. ↵ Shrestha S , King AA , Rohani P. Statistical Inference for Multi-Pathogen Systems . PLOS Comput Biol . 2011 ; 7 : e1002135 . doi: 10.1371/journal.pcbi.1002135 OpenUrl CrossRef PubMed 7. ↵ Atkins KE , Lafferty EI , Deeny SR , Davies NG , Robotham JV , Jit M. Use of mathematical modelling to assess the impact of vaccines on antibiotic resistance . Lancet Infect Dis . 2018 ; 18 : e204 – e213 . doi: 10.1016/S1473-3099(17)30478-4 OpenUrl CrossRef 8. Wu JT , Leung GM , Lipsitch M , Cooper BS , Riley S. Hedging against Antiviral Resistance during the Next Influenza Pandemic Using Small Stockpiles of an Alternative Chemotherapy . PLOS Med . 2009 ; 6 : e1000085 . doi: 10.1371/journal.pmed.1000085 OpenUrl CrossRef PubMed 9. ↵ Day T , Kennedy DA , Read AF , Gandon S. Pathogen evolution during vaccination campaigns . PLOS Biol . 2022 ; 20 : e3001804 . doi: 10.1371/journal.pbio.3001804 OpenUrl CrossRef PubMed 10. ↵ Bansal S , Grenfell BT , Meyers LA . When individual behaviour matters: homogeneous and network models in epidemiology . J R Soc Interface . 2007 ; 4 : 879 – 891 . doi: 10.1098/rsif.2007.1100 OpenUrl CrossRef PubMed Web of Science 11. Valdano E , Colombi D , Poletto C , Colizza V. Epidemic graph diagrams as analytics for epidemic control in the data-rich era . Nat Commun . 2023 ; 14 : 8472 . doi: 10.1038/s41467-023-43856-1 OpenUrl CrossRef PubMed 12. ↵ Kiss IZ , Miller J , Simon PL . Mathematics of Epidemics on Networks: From Exact to Approximate Models . Springer International Publishing ; 2017 . doi: 10.1007/978-3-319-50806-1 OpenUrl CrossRef 13. ↵ Park SW , Cobey S , Metcalf CJE , Levine JM , Grenfell BT . Predicting pathogen mutual invasibility and co-circulation . Science . 2024 ; 386 : 175 – 179 . doi: 10.1126/science.adq0072 OpenUrl CrossRef PubMed 14. Lion S , Boots M. Are parasites ‘“prudent”’ in space? Ecol Lett . 2010 ; 13 : 1245 – 1255 . doi: 10.1111/j.1461-0248.2010.01516.x OpenUrl CrossRef PubMed Web of Science 15. ↵ Leventhal GE , Hill AL , Nowak MA , Bonhoeffer S. Evolution and emergence of infectious diseases in theoretical and real-world networks . Nat Commun . 2015 ; 6 : 6101 . doi: 10.1038/ncomms7101 OpenUrl CrossRef PubMed 16. ↵ Poletto C , Meloni S , Colizza V , Moreno Y , Vespignani A. Host Mobility Drives Pathogen Competition in Spatially Structured Populations . PLoS Comput Biol . 2013 ; 9 : e1003169 . doi: 10.1371/journal.pcbi.1003169 OpenUrl CrossRef PubMed 17. ↵ Elie B , Selinger C , Alizon S. The source of individual heterogeneity shapes infectious disease outbreaks . Proc R Soc B Biol Sci . 2022 ; 289 : 20220232 . doi: 10.1098/rspb.2022.0232 OpenUrl CrossRef 18. Reyné B , Djidjou-Demasse R , Sofonea MT , Alizon S. Mutant emergence timing and population immunisation status impact epidemiological dynamics . J Theor Biol . 2025 ; 608 : 112140 . doi: 10.1016/j.jtbi.2025.112140 OpenUrl CrossRef PubMed 19. ↵ Bushman M , Kahn R , Taylor BP , Lipsitch M , Hanage WP . Population impact of SARS-CoV-2 variants with enhanced transmissibility and/or partial immune escape . Cell . 2021 ; 184 : 6229 - 6242.e18 . doi: 10.1016/j.cell.2021.11.026 OpenUrl CrossRef 20. ↵ Bansal S , Meyers LA . The impact of past epidemics on future disease dynamics . J Theor Biol . 2012 ; 309 : 176 – 184 . doi: 10.1016/j.jtbi.2012.06.012 OpenUrl CrossRef PubMed 21. ↵ Fox SJ , Miller JC , Meyers LA . Seasonality in risk of pandemic influenza emergence . PLOS Comput Biol . 2017 ; 13 : e1005749 . doi: 10.1371/journal.pcbi.1005749 OpenUrl CrossRef PubMed 22. ↵ Day T , Gandon S , Lion S , Otto SP . On the evolutionary epidemiology of SARS-CoV-2 . Curr Biol . 2020 ; 30 : R849 – R857 . doi: 10.1016/j.cub.2020.06.031 OpenUrl CrossRef PubMed 23. ↵ Berngruber TW , Froissart R , Choisy M , Gandon S. Evolution of Virulence in Emerging Epidemics . PLOS Pathog . 2013 ; 9 : e1003209 . doi: 10.1371/journal.ppat.1003209 OpenUrl CrossRef PubMed 24. Boots M , Mealor M. Local Interactions Select for Lower Pathogen Infectivity . Science . 2007 ; 315 : 1284 – 1286 . doi: 10.1126/science.1137126 OpenUrl Abstract / FREE Full Text 25. Benhamou W , Blanquart F , Choisy M , Berngruber TW , Choquet R , Gandon S. Evolution of virulence in emerging epidemics: from theory to experimental evolution and back . Virus Evol . 2024 ; 10 : veae069 . doi: 10.1093/ve/veae069 OpenUrl CrossRef PubMed 26. ↵ Otto SP , Day T , Arino J , Colijn C , Dushoff J , Li M , et al. The origins and potential future of SARS-CoV-2 variants of concern in the evolving COVID-19 pandemic . Curr Biol . 2021 ; 31 : R918 – R929 . doi: 10.1016/j.cub.2021.06.049 OpenUrl CrossRef PubMed 27. ↵ Alizon S , Sofonea MT . SARS-CoV-2 virulence evolution: Avirulence theory, immunity and trade-offs . J Evol Biol . 2021 ; 34 : 1867 – 1877 . doi: 10.1111/jeb.13896 OpenUrl CrossRef 28. ↵ Tao K , Tzou PL , Nouhin J , Gupta RK , de Oliveira T , Kosakovsky Pond SL , et al. The biological and clinical significance of emerging SARS-CoV-2 variants . Nat Rev Genet . 2021 ; 22 : 757 – 773 . doi: 10.1038/s41576-021-00408-x OpenUrl CrossRef PubMed 29. ↵ Brito AF , Semenova E , Dudas G , Hassler GW , Kalinich CC , Kraemer MUG , et al. Global disparities in SARS-CoV-2 genomic surveillance . Nat Commun . 2022 ; 13 : 7003 . doi: 10.1038/s41467-022-33713-y OpenUrl CrossRef PubMed 30. ↵ Subissi L , von Gottberg A , Thukral L , Worp N , Oude Munnink BB , Rathore S , et al. An early warning system for emerging SARS-CoV-2 variants . Nat Med . 2022 ; 28 : 1110 – 1115 . doi: 10.1038/s41591-022-01836-w OpenUrl CrossRef 31. ↵ Campbell F , Archer B , Laurenson-Schafer H , Jinnai Y , Konings F , Batra N , et al. Increased transmissibility and global spread of SARS-CoV-2 variants of concern as at June 2021 . Eurosurveillance . 2021 ; 26 : 2100509 . doi: 10.2807/1560-7917.ES.2021.26.24.2100509 OpenUrl CrossRef PubMed 32. ↵ Alpert T , Brito AF , Lasek-Nesselquist E , Rothman J , Valesano AL , MacKay MJ , et al. Early introductions and transmission of SARS-CoV-2 variant B.1.1.7 in the United States . Cell . 2021 ; 184 : 2595 - 2604.e13 . doi: 10.1016/j.cell.2021.03.061 OpenUrl CrossRef PubMed 33. ↵ Washington NL , Gangavarapu K , Zeller M , Bolze A , Cirulli ET , Schiabor Barrett KM , et al. Emergence and rapid transmission of SARS-CoV-2 B.1.1.7 in the United States . Cell . 2021 ; 184 : 2587 - 2594.e7 . doi: 10.1016/j.cell.2021.03.052 OpenUrl CrossRef PubMed 34. ↵ Faucher B , Sabbatini CE , Czuppon P , Kraemer MUG , Lemey P , Colizza V , et al. Drivers and impact of the early silent invasion of SARS-CoV-2 Alpha . Nat Commun . 2024 ; 15 : 2152 . doi: 10.1038/s41467-024-46345-1 OpenUrl CrossRef PubMed 35. ↵ Chen C , Nadeau S , Yared M , Voinov P , Xie N , Roemer C , et al. CoV-Spectrum: analysis of globally shared SARS-CoV-2 data to identify and characterize new variants . Bioinformatics . 2022 ; 38 : 1735 – 1737 . doi: 10.1093/bioinformatics/btab856 OpenUrl CrossRef PubMed 36. ↵ Elbe S , Buckland-Merrett G. Data, disease and diplomacy: GISAID’s innovative contribution to global health . Glob Chall . 2017 ; 1 : 33 – 46 . doi: 10.1002/gch2.1018 OpenUrl CrossRef PubMed 37. ↵ Hinch R , Panovska-Griffiths J , Probert WJM , Ferretti L , Wymant C , Di Lauro F , et al. Estimating SARS-CoV-2 variant fitness and the impact of interventions in England using statistical and geo-spatial agent-based models . Philos Transact A Math Phys Eng Sci . 380 : 20210304 . doi: 10.1098/rsta.2021.0304 OpenUrl CrossRef PubMed 38. ↵ Roquebert B , Trombert-Paolantoni S , Haim-Boukobza S , Lecorche E , Verdurme L , Foulongne V , et al. The SARS-CoV-2 B.1.351 lineage (VOC β) is outgrowing the B.1.1.7 lineage (VOC α) in some French regions in April 2021 . Eurosurveillance . 2021 ; 26 : 2100447 . doi: 10.2807/1560-7917.ES.2021.26.23.2100447 OpenUrl CrossRef PubMed 39. ↵ Nationwide Commercial Laboratory Seroprevalence Survey | Data | Centers for Disease Control and Prevention . [cited 19 Aug 2025 ]. Available: https://data.cdc.gov/Laboratory-Surveillance/Nationwide-Commercial-Laboratory-Seroprevalence-Su/d2tw-32xv/about_data 40. ↵ COVID-19 Vaccinations in the United States,Jurisdiction | Data | Centers for Disease Control and Prevention . [cited 19 Aug 2025 ]. Available: https://data.cdc.gov/Vaccinations/COVID-19-Vaccinations-in-the-United-States-Jurisdi/unsk-b7fc/about_data 41. ↵ Taube JC , Susswein Z , Colizza V , Bansal S. Characterising non-household contact patterns relevant to respiratory transmission in the USA: analysis of a cross-sectional survey . Lancet Digit Health . 2025 ; 7 . doi: 10.1016/j.landig.2025.100888 OpenUrl CrossRef 42. ↵ Volz E , Mishra S , Chand M , Barrett JC , Johnson R , Geidelberg L , et al. Assessing transmissibility of SARS-CoV-2 lineage B.1.1.7 in England . Nature . 2021 ; 593 : 266 – 269 . doi: 10.1038/s41586-021-03470-x OpenUrl CrossRef PubMed 43. ↵ Supasa P , Zhou D , Dejnirattisai W , Liu C , Mentzer AJ , Ginn HM , et al. Reduced neutralization of SARS-CoV-2 B.1.1.7 variant by convalescent and vaccine sera . Cell . 2021 ; 184 : 2201 - 2211.e7 . doi: 10.1016/j.cell.2021.02.033 OpenUrl CrossRef PubMed 44. ↵ Altarawneh HN , Chemaitelly H , Hasan MR , Ayoub HH , Qassim S , AlMukdad S , et al. Protection against the Omicron Variant from Previous SARS-CoV-2 Infection . N Engl J Med . 2022 ; 386 : 1288 – 1290 . doi: 10.1056/NEJMc2200133 OpenUrl CrossRef PubMed 45. ↵ Graham MS , Sudre CH , May A , Antonelli M , Murray B , Varsavsky T , et al. Changes in symptomatology, reinfection, and transmissibility associated with the SARS-CoV-2 variant B.1.1.7: an ecological study . Lancet Public Health . 2021 ; 6 : e335 – e345 . doi: 10.1016/S2468-2667(21)00055-4 OpenUrl CrossRef 46. ↵ Poletto C , Meloni S , Van Metre A , Colizza V , Moreno Y , Vespignani A. Characterising two-pathogen competition in spatially structured environments . Sci Rep . 2015 ; 5 : 7895 . doi: 10.1038/srep07895 OpenUrl CrossRef PubMed 47. ↵ Blanquart F , Hozé N , Cowling BJ , Débarre F , Cauchemez S. Selection for infectivity profiles in slow and fast epidemics, and the rise of SARS-CoV-2 variants . Cooper BS, Davenport MP, editors. eLife . 2022 ; 11 : e75791 . doi: 10.7554/eLife.75791 OpenUrl CrossRef 48. ↵ Manica M , Litvinova M , Bellis AD , Guzzetta G , Mancuso P , Vicentini M , et al. Estimation of the incubation period and generation time of SARS-CoV-2 Alpha and Delta variants from contact tracing data . Epidemiol Infect . 2023 ; 151 : e5 . doi: 10.1017/S0950268822001947 OpenUrl CrossRef 49. ↵ Mastrandrea R , Fournet J , Barrat A. Contact Patterns in a High School: A Comparison between Data Collected Using Wearable Sensors, Contact Diaries and Friendship Surveys . PLOS ONE . 2015 ; 10 : e0136497 . doi: 10.1371/journal.pone.0136497 OpenUrl CrossRef PubMed 50. Salathé M , Kazandjieva M , Lee JW , Levis P , Feldman MW , Jones JH . A high-resolution human contact network for infectious disease transmission . Proc Natl Acad Sci . 2010 ; 107 : 22020 – 22025 . doi: 10.1073/pnas.1009094108 OpenUrl Abstract / FREE Full Text 51. Génois M , Vestergaard CL , Fournet J , Panisson A , Bonmarin I , Barrat A. Data on face-to-face contacts in an office building suggest a low-cost vaccination strategy based on community linkers . Netw Sci . 2015 ; 3 : 326 – 347 . doi: 10.1017/nws.2015.10 OpenUrl CrossRef 52. Obadia T , Silhol R , Opatowski L , Temime L , Legrand J , Thiébaut ACM , et al. Detailed contact data and the dissemination of Staphylococcus aureus in hospitals . PLoS Comput Biol . 2015 ; 11 : e1004170 . doi: 10.1371/journal.pcbi.1004170 OpenUrl CrossRef PubMed 53. ↵ Vanhems P , Barrat A , Cattuto C , Pinton J-F , Khanafer N , Régis C , et al. Estimating Potential Infection Transmission Routes in Hospital Wards Using Wearable Proximity Sensors . PLOS ONE . 2013 ; 8 : e73970 . doi: 10.1371/journal.pone.0073970 OpenUrl CrossRef PubMed 54. ↵ Mossong J , Hens N , Jit M , Beutels P , Auranen K , Mikolajczyk R , et al. Social Contacts and Mixing Patterns Relevant to the Spread of Infectious Diseases . PLoS Med . 2008 ; 5 : e74 . doi: 10.1371/journal.pmed.0050074 OpenUrl CrossRef PubMed 55. Béraud G , Kazmercziak S , Beutels P , Levy-Bruhl D , Lenne X , Mielcarek N , et al. The French Connection: The First Large Population-Based Contact Survey in France Relevant for the Spread of Infectious Diseases . PLOS ONE . 2015 ; 10 : e0133203 . doi: 10.1371/journal.pone.0133203 OpenUrl CrossRef PubMed 56. ↵ Verelst F , Hermans L , Vercruysse S , Gimma A , Coletti P , Backer JA , et al. SOCRATES-CoMix: a platform for timely and open-source contact mixing data during and in between COVID-19 surges and interventions in over 20 European countries . BMC Med . 2021 ; 19 : 254 . doi: 10.1186/s12916-021-02133-y OpenUrl CrossRef 57. ↵ Wong KLM , Gimma A , Coletti P , Paolotti D , Tizzani M , Cattuto C , et al. Social contact patterns during the COVID-19 pandemic in 21 European countries – evidence from a two-year study . BMC Infect Dis . 2023 ; 23 : 268 . doi: 10.1186/s12879-023-08214-y OpenUrl CrossRef PubMed 58. ↵ Di Domenico L , Sabbatini CE , Boëlle P-Y , Poletto C , Crépey P , Paireau J , et al. Adherence and sustainability of interventions informing optimal control against the COVID-19 pandemic . Commun Med . 2021 ; 1 : 1 – 13 . doi: 10.1038/s43856-021-00057-5 OpenUrl CrossRef 59. Weitz JS , Park SW , Eksin C , Dushoff J. Awareness-driven behavior changes can shift the shape of epidemics away from peaks and toward plateaus, shoulders, and oscillations . Proc Natl Acad Sci . 2020 ; 117 : 32764 – 32771 . doi: 10.1073/pnas.2009911117 OpenUrl Abstract / FREE Full Text 60. ↵ Pangallo M , Aleta A , del Rio-Chanona RM , Pichler A , Martín-Corral D , Chinazzi M , et al. The unequal effects of the health–economy trade-off during the COVID-19 pandemic . Nat Hum Behav . 2024 ; 8 : 264 – 275 . doi: 10.1038/s41562-023-01747-x OpenUrl CrossRef PubMed 61. ↵ Britton T , Ball F , Trapman P. A mathematical model reveals the influence of population heterogeneity on herd immunity to SARS-CoV-2 . Science . 2020 ; 369 : 846 – 849 . doi: 10.1126/science.abc6810 OpenUrl Abstract / FREE Full Text 62. Gomes MGM , Ferreira MU , Corder RM , King JG , Souto-Maior C , Penha-Gonçalves C , et al. Individual variation in susceptibility or exposure to SARS-CoV-2 lowers the herd immunity threshold . J Theor Biol . 2022 ; 540 : 111063 . doi: 10.1016/j.jtbi.2022.111063 OpenUrl CrossRef PubMed 63. Loedy N , Wallinga J , Hens N , Torneri A. Repetition in social contacts: implications in modelling the transmission of respiratory infectious diseases in pre-pandemic and pandemic settings . Proc R Soc B Biol Sci . 2024 ; 291 : 20241296 . doi: 10.1098/rspb.2024.1296 OpenUrl CrossRef PubMed 64. Pung R , Firth JA , Russell TW , Rogers T , Lee VJ , Kucharski AJ . Temporal contact patterns and the implications for predicting superspreaders and planning of targeted outbreak control . J R Soc Interface . 2024 ; 21 : 20240358 . doi: 10.1098/rsif.2024.0358 OpenUrl CrossRef PubMed 65. ↵ Tkachenko AV , Maslov S , Elbanna A , Wong GN , Weiner ZJ , Goldenfeld N. Time-dependent heterogeneity leads to transient suppression of the COVID-19 epidemic, not herd immunity . Proc Natl Acad Sci . 2021 ; 118 : e2015972118 . doi: 10.1073/pnas.2015972118 OpenUrl Abstract / FREE Full Text 66. ↵ Manna A , Dall’Amico L , Tizzoni M , Karsai M , Perra N. Generalized contact matrices allow integrating socioeconomic variables into epidemic models . Sci Adv . 2024 ; 10 : eadk4606 . doi: 10.1126/sciadv.adk4606 OpenUrl CrossRef PubMed 67. ↵ Tizzoni M , Nsoesie EO , Gauvin L , Karsai M , Perra N , Bansal S. Addressing the socioeconomic divide in computational modeling for infectious diseases . Nat Commun . 2022 ; 13 : 2897 . doi: 10.1038/s41467-022-30688-8 OpenUrl CrossRef PubMed 68. ↵ Benhamou W , Choquet R , Gandon S. The interplay between migration and selection on the dynamics of pathogen variants . medRxiv ; 2025 . p. 2025.04.28.25326566 . doi: 10.1101/2025.04.28.25326566 OpenUrl Abstract / FREE Full Text 69. ↵ Jennings V , Lloyd-Smith B , Ironmonger D. Household size and the poisson distribution . J Aust Popul Assoc . 1999 ; 16 : 65 – 84 . doi: 10.1007/BF03029455 OpenUrl CrossRef 70. ↵ Eurostat. Average household size . In: Eurostat [Internet] . [cited 15 July 2025 ]. Available: https://ec.europa.eu/eurostat/databrowser/product/page/ILC_LVPH01 71. ↵ U.S. Census Bureau . Households and Families (Table S1101) , 2023 American Community Survey 1-Year Estimates . [cited 15 July 2025 ]. Available: https://data.census.gov/table/ACSST1Y2023.S1101?q=household+size&g=010XX00US$0400000 72. ↵ Susswein Z , Valdano E , Brett T , Rohani P , Colizza V , Bansal S. Ignoring spatial heterogeneity in drivers of SARS-CoV-2 transmission in the US will impede sustained elimination . medRxiv . 2021 ; 2021.08.09.21261807 . doi: 10.1101/2021.08.09.21261807 OpenUrl Abstract / FREE Full Text 73. ↵ Hastie T , Tibshirani R , Friedman J. The Elements of Statistical Learning . New York, NY : Springer ; 2009 . doi: 10.1007/978-0-387-84858-7 OpenUrl CrossRef 74. ↵ Average size of U.S. households, by state 2021 . In: Statista [Internet] . [cited 15 July 2025 ]. Available: https://www.statista.com/statistics/242265/average-size-of-us-households-by-state/ View the discussion thread. Back to top Previous Next Posted November 09, 2025. Download PDF Supplementary Material Data/Code Email Thank you for your interest in spreading the word about medRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following Contact structure and population immunity shape the selective advantage of emerging variants Message Subject (Your Name) has forwarded a page to you from medRxiv Message Body (Your Name) thought you would like to see this page from the medRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share Contact structure and population immunity shape the selective advantage of emerging variants Pourya Toranj Simin , Juliana C. Taube , Elisabeta Vergu , Shweta Bansal , Lulla Opatowski , Chiara Poletto medRxiv 2025.11.07.25339691; doi: https://doi.org/10.1101/2025.11.07.25339691 Share This Article: Copy Citation Tools Contact structure and population immunity shape the selective advantage of emerging variants Pourya Toranj Simin , Juliana C. Taube , Elisabeta Vergu , Shweta Bansal , Lulla Opatowski , Chiara Poletto medRxiv 2025.11.07.25339691; doi: https://doi.org/10.1101/2025.11.07.25339691 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Area Epidemiology Subject Areas All Articles Addiction Medicine (569) Allergy and Immunology (863) Anesthesia (300) Cardiovascular Medicine (4441) Dentistry and Oral Medicine (444) Dermatology (383) Emergency Medicine (609) Endocrinology (including Diabetes Mellitus and Metabolic Disease) (1510) Epidemiology (15229) Forensic Medicine (30) Gastroenterology (1126) Genetic and Genomic Medicine (6607) Geriatric Medicine (668) Health Economics (998) Health Informatics (4541) Health Policy (1369) Health Systems and Quality Improvement (1613) Hematology (543) HIV/AIDS (1266) Infectious Diseases (except HIV/AIDS) (15923) Intensive Care and Critical Care Medicine (1103) Medical Education (623) Medical Ethics (147) Nephrology (668) Neurology (6605) Nursing (346) Nutrition (998) Obstetrics and Gynecology (1146) Occupational and Environmental Health (957) Oncology (3334) Ophthalmology (974) Orthopedics (369) Otolaryngology (420) Pain Medicine (436) Palliative Medicine (130) Pathology (663) Pediatrics (1693) Pharmacology and Therapeutics (692) Primary Care Research (711) Psychiatry and Clinical Psychology (5448) Public and Global Health (9236) Radiology and Imaging (2200) Rehabilitation Medicine and Physical Therapy (1370) Respiratory Medicine (1196) Rheumatology (594) Sexual and Reproductive Health (713) Sports Medicine (530) Surgery (712) Toxicology (99) Transplantation (289) Urology (265) (function(){function c(){var b=a.contentDocument||a.contentWindow.document;if(b){var d=b.createElement('script');d.innerHTML="window.__CF$cv$params={r:'a017ee48ee628f53',t:'MTc3OTc0NzY2Mg=='};var a=document.createElement('script');a.src='/cdn-cgi/challenge-platform/scripts/jsd/main.js';document.getElementsByTagName('head')[0].appendChild(a);";b.getElementsByTagName('head')[0].appendChild(d)}}if(document.body){var a=document.createElement('iframe');a.height=1;a.width=1;a.style.position='absolute';a.style.top=0;a.style.left=0;a.style.border='none';a.style.visibility='hidden';document.body.appendChild(a);if('loading'!==document.readyState)c();else if(window.addEventListener)document.addEventListener('DOMContentLoaded',c);else{var e=document.onreadystatechange||function(){};document.onreadystatechange=function(b){e(b);'loading'!==document.readyState&&(document.onreadystatechange=e,c())}}}})();