Pith. sign in

REVIEW 2 major objections 5 minor 64 references

Information and fitness in two-state systems: self-replicating individuals in a fluctuating environment

T0 review · 2 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read With phenotype and environment marginals fixed, mutual information and normalized population fitness are one-to-one increasing functions of the phenotype-environment correlation, independent of the growth rates.

desk verdict A sound, honestly scoped extension of information-fitness equivalence to a population-level two-state model; the fixed-marginals caveat is in the paper itself, so referee it. read the letter →

arxiv 2508.00150 v1 pith:VEW4UPYF submitted 2025-07-31 q-bio.PE

Poulami Chatterjee , Cesar Nieto , Juan Manuel Pedraza , Abhyudai Singh This is my paper · ORCID
classification q-bio.PE MSC 92D2594A17
keywords bet-hedgingmutualinformationphenotypicswitchingfluctuatingenvironmentnormalizedfitnesstwo-statephenotypebacterialpersistencevalue
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper asks when population-level bet-hedging pays in a fluctuating two-state environment, and proposes that the payoff is fully captured by the correlation between phenotype and environment. For a population of self-replicating individuals with two phenotypes, it defines a normalized fitness $\Gamma$ as the time-averaged growth rate relative to the hypothetical independent population, divided by the maximum possible relative gain. The main result is that, when the marginal frequencies of the two environments and the two phenotypes are held fixed, $\Gamma$ and the mutual information $I$ between environment and phenotype are one-to-one increasing functions of the same quantity $P_{x_1y_1}$, the probability that the fittest phenotype appears in its matching environment. The relation does not depend on the numerical values of the growth or switching rates. If true, it means that in this whole class of models the fitness benefit of bet-hedging is governed by a single correlation measure, and that extra information about the environment is worth progressively less as the population already has more of it.

What carries the argument

The load-bearing object is the stationary joint probability $P_{x_i y_j}$, obtained from the time-averaged phenotype fraction $\langle f_{y_j}\rangle$ and the environmental indicator function. This turns the dynamical system into a time-independent information channel whose input is the environment and whose output is the phenotype. The key identity is Eq. (18): with fixed $P_{x_1}$ and $P_{y_1}$, $\Gamma = (P_{x_1y_1} - P_{x_1}P_{y_1})/(P_{x_1}(1-P_{x_1}))$, which is independent of the growth rates. The mutual information is likewise parameterized by the single correlation $P_{x_1y_1}$ through the four entries of the joint distribution. The argument that carries the paper is that both $\Gamma$ and $I$ are monotone functions of this one parameter, so a one-to-one curve between them exists for each fixed pair of marginals; the normalization by the maximum relative growth rate is what removes the growth rates from the formula.

What would settle it

Run the dormant-proliferating model while actively clamping $P_{y_1}$ constant as $\mu_1$ varies; if $\Gamma$ and $I$ still separate, with one falling while the other rises, the fixed-marginal claim fails in exactly the regime intended. Alternatively, scan the full parameter space of the general model and search for any pair of parameter sets with identical $P_{x_1}$ and $P_{y_1}$ but with a larger $P_{x_1y_1}$ and a smaller $\Gamma$; none should exist if Eq. (18) is correct.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is a parameter-free equivalence between information and fitness. Starting from the deterministic fraction dynamics for two phenotypes in a two-state telegraph environment, the authors construct the stationary joint distribution $P_{x_i y_j}$ by time-averaging the phenotype fraction inside each environmental state, and define the normalized fitness $\Gamma = (\langle\gamma\rangle - \langle\gamma\rangle_{\mathrm{ind}})/(\langle\gamma\rangle_{\max} - \langle\gamma\rangle_{\mathrm{ind}})$. For fixed marginals $P_{x_1}$ and $P_{y_1}$, this reduces to $\Gamma = (P_{x_1y_1} - P_{x_1}P_{y_1})/(P_{x_1}(1-P_{x_1}))$, and the mutual information $I=\sum P_{x_i y_j}\log_2(P_{x_i y_j}/(P_{x_i}P_{y_j}))$ also depends on $P_{x_1y_1}$ alone; consequently both quantities increase together and each one determines the other. Because the growth rates cancel out of $\Gamma$, the relation is universal across all parameter choices. The maximum achievable $\Gamma$ and $I$ occur when the phenotype marginal equals the environmental marginal, $P_{y_1}=P_{x_1}$, and the maximum information is then the environmental entropy. The paper stresses that this master-curve behavior holds only while the marginals are fixed; when the marginals vary, as they do in a dormant-proliferating (persister) population under increasing antibiotic stress, $I$ can rise while $\Gamma$ peaks and falls, so the two are not equivalent as raw numbers in that broader setting.

Load-bearing premise

The equivalence rests on defining fitness as the normalized, time-averaged growth advantage over the independence baseline, and on keeping the marginal phenotype and environment frequencies fixed; if fitness means the actual long-term growth rate, or if the marginals move, the one-to-one relation can fail.

Editorial extensions

If this is right

  • In the symmetric bet-hedging model with equally likely environments, all growth and switching parameters collapse onto a single master curve $\Gamma_{\mathrm{sym}}=4P_{x_1y_1}-1$ against $I_{\mathrm{sym}}$, so measuring one quantity determines the other.
  • Any strategy that raises the phenotype-environment correlation, such as responsive switching that favors the fittest phenotype or a slower environment that lets the fittest phenotype dominate, raises both mutual information and normalized fitness for fixed marginals.
  • The marginal value of information, $\partial\Gamma/\partial I$, is always positive but decreases with $I$; near zero information it diverges, meaning the first bits of correlation yield the largest normalized-fitness gain per bit.
  • To maximize both information and normalized fitness, a population should match its phenotype marginal to the environmental marginal ($P_{y_1}=P_{x_1}$), reaching an information level equal to the environmental entropy.
  • When the marginals are free to vary, as in the dormant-proliferating model under rising antibiotic death rate, normalized fitness and mutual information can move in opposite directions, so the equivalence is specifically a fixed-marginal property rather than a universal proportionality.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Because the mutual information here is computed from the stationary joint distribution rather than from trajectories, the paper's equivalence does not by itself capture how quickly the phenotype population tracks the environment; a natural extension would be to define a rate-dependent mutual information and test whether the one-to-one relation survives.
  • A testable prediction follows: in a synthetic two-state bet-hedging system with the marginal phenotype fraction clamped by a controller, any perturbation that increases phenotype-environment correlation should increase the normalized growth rate, even if the absolute growth rate falls, paralleling persister cells under stress.
  • The persister example suggests that the quantity evolutionary biology should equate with information is $\Gamma$, the relative advantage over independence, not the absolute long-term growth rate; if absolute growth is the fitness that matters, the paper's 'information implies fitness' statement applies to a different objective.
  • The decreasing marginal value curve suggests an optimization problem: if acquiring and storing information has a per-bit cost, an organism's optimal information level would sit below the achievable maximum, which the paper notes but does not model.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

2 major / 5 minor

Summary. The paper analyzes a two-state population model in a two-state randomly switching environment. It defines time-averaged joint probabilities for environment and phenotype, the mutual information I, and a normalized fitness Gamma defined as the growth rate relative to the independent case, normalized by the maximum possible relative growth rate. With fixed marginal probabilities Px1 and Py1, the paper shows that Gamma and I are both strictly increasing functions of the joint probability Px1y1, and it derives a growth-rate-independent expression for Gamma, the maximum achievable Gamma and I for fixed marginals, and a marginal information value dGamma/dI. The formalism is illustrated with a symmetric bet-hedging model and a dormant-proliferating phenotype model. The central algebraic derivations are sound, but the biological scope of the claimed information-fitness equivalence is narrower than the abstract suggests.

Significance. As a mathematical characterization of a two-by-two probability table, the paper is correct and provides closed-form results: Eq. (18) shows that Gamma is independent of the growth rates, and Eq. (21) correctly evaluates the marginal information value, which I verified to be positive and divergent at I=0. The paper also ships reproducible simulation code and includes a detailed algorithmic description of its numerics. The biological significance, however, is limited by the fact that the equivalence is a property of the 2x2 parametrization rather than a demonstrated property of the population dynamics: changing biological rates generically changes the marginals as well as Px1y1, and the dormant-proliferating example in Fig. 6E shows that Gamma and I can move in opposite directions when marginals are free. The contribution is therefore a moderate unification of information and normalized fitness for two-state bet-hedging, not a general proof that information acquisition increases population fitness.

major comments (2)
  1. [Section V, Eqs. (16)-(18), and abstract] The headline claim 'with fixed marginal distributions, an increase in information implies an increase in population fitness' is a theorem about a 2x2 probability table, not about the dynamics of Eqs. (1)-(2). For fixed Px1 and Py1, both Gamma and I are strictly increasing functions of Px1y1, and the algebra is correct. However, the paper does not exhibit any control strategy in the general asymmetric model that varies Px1y1 while keeping both Px1 and Py1 fixed. The only mechanism explicitly shown is the symmetric model, where Py1 = Px1 = 1/2 is fixed by symmetry. In the general model, changing growth or switching rates generically moves Py1 and Px1y1 together, as the dormancy example in Fig. 6E demonstrates: there I and Gamma are non-monotonically related precisely because the marginals are not fixed. Since the abstract promises 'strategies' with fixed marginal distributions, the authors should either construct such a mechanism or explicitly restrict the biological claim to the conditional statement and to the symmetric special case.
  2. [Section IV, Eq. (13)] The normalized fitness Gamma is defined as (⟨γ⟩−⟨γ⟩ind)/(⟨γ⟩max−⟨γ⟩ind), which is a normalized linear function of Px1y1−Px1Py1. For fixed marginals, every such function is monotone in Px1y1, and the mutual information I is also monotone in Px1y1 on the admissible interval of Eq. (16). Thus the 'equivalence' between information and fitness is largely installed by the choice of Gamma rather than derived from the population dynamics. I verified Eq. (18) and Eq. (21), but the paper's language—'we propose that information and fitness can be treated as equivalent measures' and 'we prove... any increase in population fitness through an arbitrary strategy corresponds to a simultaneous increase in information'—overstates the novelty. The authors should state explicitly that the equivalence is a property of the 2x2 parametrization, not a dynamical law.
minor comments (5)
  1. [Section II.B, Eq. (2)] In the x=x2 branch of Eq. (2), the selection term should be (gx2_y1 − gx2_y2) fy1 (1 − fy1), not (gx1_y2 − gx2_y2) fy1 (1 − fy1); as written, the expression mixes growth rates from the two different environments.
  2. [Section II.D] The text states 'gx1_y2 = gx2_y2 = 0, gx1_y1 = −µ1 and gx2_y1 = µ2', which is inconsistent with Table V and with Algorithm 1; those sources use the opposite assignment, with y1 dormant (zero growth) and y2 proliferating (−µ1 in stress, µ2 in normal conditions). Please reconcile the text, the table, and the algorithm.
  3. [Section V, Eq. (20)] The second branch of the piecewise expression for Imax is incomplete: it lacks the condition 'if Py1 < Px1' and the third logarithmic term analogous to the first branch. Please supply the full expression.
  4. [Section V.B] The sentence 'This maximum occurs when Py1 ≈ 0.5' is inconsistent with the Fig. 6C caption ('reaching a maximal value when Py1 = Px1') and with Eq. (19), which gives a global maximum at Py1 = Px1. Please correct the text.
  5. [Throughout] There are several typographical errors, including 'bet-heging' in the Introduction, 'a a proliferation rate' in Section II.A, and an index mismatch in Eq. (5) where Pyi is defined with fyj. A careful proofread is needed.

Circularity Check

1 steps flagged · score 5.0 of 10

The central equivalence is installed by the definition of normalized fitness; it restates a standard 2x2-table monotonicity as a biological discovery, without exhibiting a general dynamic mechanism that realizes fixed-marginal sweeps.

  1. renaming known result [Section V, summary paragraph after Eq. (20); Eqs. (13) and (18)]
    "In summary, we prove how for given marginal distributions Px1 and Py1 of environment and phenotype, respectively, any increase in population fitness through an arbitrary strategy corresponds to a simultaneous increase in information, and vice versa."

    Eq. (13) defines normalized fitness Gamma as a linear combination of Px1y1 minus the independent-case term, divided by a constant. With the 2x2 constraints of Eq. (17), this reduces in Eq. (18) to Gamma = (Px1y1 - Px1 Py1)/(Px1(1-Px1)), i.e. a standardized covariance. Mutual information I (Eq. 9) is, for fixed Px1 and Py1, a strictly increasing function of the same parameter Px1y1 over the admissible interval of Eq. (16). Therefore the claim that increasing information implies increasing fitness is a property of the chosen definitions of Gamma and I, not a consequence of the population dynamics in Eq. (1).

full rationale

The paper contains no load-bearing self-citation chain and no fitted parameter renamed as a prediction; the algebra from Eq. (13) to Eq. (18) and the monotonicity of MI in Px1y1 are correct and self-contained. The circularity is conceptual: normalized fitness Gamma is deliberately defined as a standardized phenotype-environment covariance, and mutual information is another monotone function of the same covariance when marginals are fixed. Hence the headline statement 'an increase in information implies an increase in population fitness' is a consequence of the definitions, not an independent result of the stochastic population model. The dormant-proliferating example (Section V.B) explicitly acknowledges that when Py1 is not fixed, Gamma and I can move oppositely, which is consistent with the interpretation that the equivalence is an artifact of holding the marginals fixed. The paper is honest about proposing 'normalized fitness' as a metric, but the presentation in the abstract and Section V as a general proof overstates the extent to which the dynamics realize the fixed-marginal condition. For these reasons, the central equivalence is substantially definitional, though not statistically fitted or self-citation-driven; a moderate circularity score is appropriate.

Assumptions & free parameters 3 free parameters · 6 assumptions · 1 invented entities

No parameters are fitted to data. The central Gamma-I relation is parameter-free given the joint distribution. The illustrative kinetic rates (lambda, k, g) only select where the system sits on the curve; one of the paper's results is that Gamma does not depend on them. The load-bearing inputs are the choice of Gamma as the fitness measure and the treatment of the stationary time-averaged joint distribution as the information channel. No physical entities are postulated; the only new ledger entry is the metric Gamma itself.

free parameters (3)
  • Environmental switching rates lambda12 and lambda21 = Illustrative values, 0.1 to 10 in Fig. 3; lambda21 = 0.5 in Fig. 6.
    Hand-chosen to illustrate slow and fast fluctuation regimes. Not fitted to data; each value only places the system at a point on the derived Gamma-I curve.
  • Phenotype switching rates ka and kb, or scalar k = ka = 0.1, kb = 1.2 in Fig. 3; k = 0.01 in Fig. 6.
    Hand-chosen values distinguishing responsive from non-responsive switching. Not fitted to data.
  • Growth rates mu, rho, mu1 and mu2 = mu - rho = 3 in Fig. 3; mu2 = 1 with mu1 varied in Fig. 6.
    Illustrative. A central result is that Gamma is independent of these values, so they affect the simulated locations but not the equivalence.
assumptions (6)
  • domain assumption Large-population deterministic fraction dynamics, Eq. (1) and SI S1.
    Neglects demographic fluctuations in population growth; fractions evolve by ODEs with continuous boundary conditions at environment switches.
  • domain assumption Environment is a two-state telegraph process with exponentially distributed dwell times.
    Section II.A; yields Px1 = lambda21/(lambda12 + lambda21) and justifies the time-average limits.
  • standard math The time averages in Eqs. (4)-(6) exist and define a proper joint probability.
    Section III states the probabilities are assumed to exist; used to define the channel and the mutual information.
  • domain assumption The fitness matrix is diagonal-favored (g1, g2 > 0) and the diagonal correlation is non-negative (Pxiyi >= Pxi Pyi).
    Eq. (11) and the restriction Px1y1 in [Px1 Py1, min(Px1, Py1)]; selects the branch where Gamma and I increase together.
  • standard math Any 2x2 joint distribution with fixed marginals is parameterized by Px1y1 in the Frechet bounds, Eqs. (16)-(17).
    Section V; the basis for expressing Gamma and I as functions of a single correlation parameter.
  • domain assumption The stationary time-averaged joint distribution is the information channel (time-independent reduction).
    Section III explicitly restricts the analysis to time-independent information transmission, discarding temporal correlations; this acknowledged limitation bounds the equivalence.
invented entities (1)
  • Normalized fitness Gamma independent evidence
    purpose: Growth-rate-independent measure of bet-hedging benefit, defined as the fitness gain over the environment-independent baseline divided by the maximum possible gain (Eq. 13).
    A new metric rather than a physical entity. It is falsifiable in principle: any two-state system with diagonal-favored growth must place its joint distribution on the derived Gamma-I curve. Its biological relevance depends on accepting Gamma as the operative fitness measure, which the paper proposes but does not derive from selection dynamics.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Information and fitness in two-state systems: self-replicating individuals in a fluctuating environment." pith.science (2026). https://pith.science/paper/VEW4UPYF

@misc{pith2026250800150,
  author       = {Pith},
  title        = {Pith review of: Information and fitness in two-state systems: self-replicating individuals in a fluctuating environment},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VEW4UPYF}},
  note         = {Machine review of arXiv:2508.00150}
}
read the original abstract

A population of individuals with the same genes can present heterogeneous traits (phenotypes). The prevalence of this heterogeneity can be explained as a bet-hedging strategy that improves the population proliferation rate (fitness) in fluctuating environments. The phenotype distribution is influenced by factors such as competition between phenotypes, the duration of environmental states, and the rate of phenotype-switching. We illustrate these effects in a system where both the environment and the phenotype can adopt two states. This system includes scenarios such as symmetric bet-hedging and dormant-proliferating phenotypes. We examine how environmental and phenotypic states share mutual information, measured in bits, and explore the relationship between this information and population fitness. We propose that when fitness is measured relative to the case where phenotype and environment are independent, information and fitness can be treated as equivalent measures. We investigate strategies that individuals can use to improve this information, such as adjusting the rates of proliferation and phenotype-switching relative to the environmental fluctuation rate. Through these strategies, with fixed marginal distributions, an increase in information implies an increase in population fitness. We also identify limits to the maximum achievable fitness and information and discuss the value of the information in terms of this new normalized fitness. Our framework offers new insights into how organisms adapt to fluctuating environmental conditions.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

64 extracted references · 64 canonical work pages

  1. [1]

    Dekel and U

    E. Dekel and U. Alon, Nature 436, 588 (2005)

    work page 2005

  2. [2]

    Aub´ e and C

    S. Aub´ e and C. R. Landry, Nature Ecology & Evolution , 1 (2024)

    work page 2024

  3. [3]

    B. Xue, P. Sartori, and S. Leibler, Proceedings of the National Academy of Sciences 116, 13847 (2019)

    work page 2019

  4. [4]

    De Martino, T

    A. De Martino, T. Gueudr´ e, and M. Miotto, Physical Review E 99, 012417 (2019)

    work page 2019

  5. [5]

    Cvijovi´ c, B

    I. Cvijovi´ c, B. H. Good, E. R. Jerison, and M. M. Desai, Proceedings of the National Academy of Sciences 112, E5021 (2015)

    work page 2015

  6. [6]

    Levien, J

    E. Levien, J. Min, J. Kondev, and A. Amir, Reports on Progress in Physics 84, 116601 (2021)

    work page 2021

  7. [7]

    D. H. de Groot, A. J. Tjalma, F. J. Bruggeman, and E. van Nimwegen, Proceedings of the National Academy of Sciences 120, e2211091120 (2023)

    work page 2023

  8. [8]

    Kussell and S

    E. Kussell and S. Leibler, Science 309, 2075 (2005)

    work page 2005

Show all 64 references
  1. [9]

    Thattai and A

    M. Thattai and A. Van Oudenaarden, Genetics 167, 523 (2004)

    2004

  2. [10]

    Fern´ andez-Fern´ andez, D

    R. Fern´ andez-Fern´ andez, D. R. Olivenza, E. Weyer, A. Singh, J. Casades´ us, and M. Antonia S´ anchez-Romero, Proceedings of the National Academy of Sciences 121, e2322371121 (2024)

    2024

  3. [11]

    L. P. Morawska, J. A. Hernandez-Valdes, and O. P. Kuipers, WIREs Mechanisms of Disease 14, e1544 (2022)

    2022

  4. [12]

    K. M. Hawkins and C. D. Smolke, Journal of Biological Chemistry 281, 13485 (2006)

    2006

  5. [13]

    Solopova, J

    A. Solopova, J. Van Gestel, F. J. Weissing, H. Bachmann, B. Teusink, J. Kok, and O. P. Kuipers, Proceedings of the National Academy of Sciences 111, 7427 (2014)

    2014

  6. [14]

    Maslov and K

    S. Maslov and K. Sneppen, Scientific Reports 5, 10523 (2015)

    2015

  7. [15]

    H. R. Bonifield and K. T. Hughes, Journal of Bacteriology 185, 3567 (2003)

    2003

  8. [16]

    C. D. Kenkel and M. V. Matz, Nature Ecology & Evolution 1, 0014 (2016)

    2016

  9. [17]

    Grissom, Z

    S. Grissom, Z. Dixon, A. Singh, and M. Blenner, bioRxiv , 2024 (2024)

    2024

  10. [18]

    N. Q. Balaban, J. Merrin, R. Chait, L. Kowalik, and S. Leibler, Science 305, 1622 (2004)

    2004

  11. [19]

    Patra and S

    P. Patra and S. Klumpp, PLoS One 8, e62814 (2013)

    2013

  12. [20]

    J. C. R. Hernandez-Beltran, J. Rodr ´ ıguez-Beltr´ an, O. B. Aguilar-Luviano, J. Velez-Santiago, O. Mondrag´ on-Palomino, R. C. MacLean, A. Fuentes-Hern´ andez, A. San Mill´ an, and R. Pe˜ na-Miller, Nature Communications 15, 2610 (2024)

    2024

  13. [21]

    S. E. Diamond and R. A. Martin, in Phenotypic plasticity & evolution (CRC Press, 2021) pp. 185–209

    2021

  14. [22]

    K. T. Rahman, R. Amaratunga, X. Y. Butzin, A. Singh, T. Hossain, and N. C. Butzin, International Journal of Antimicrobial Agents 65, 107386 (2025)

    2025

  15. [23]

    S. M. Shaffer, M. C. Dunagin, S. R. Torborg, E. A. Torre, B. Emert, C. Krepler, M. Beqiri, K. Sproesser, P. A. Brafford, M. Xiao, et al., Nature 546, 431 (2017)

    2017

  16. [24]

    C. A. Chang, J. Jen, S. Jiang, A. Sayad, A. S. Mer, K. R. Brown, A. M. Nixon, A. Dhabaria, K. H. Tang, D. Venet, et al., Cancer Discovery 12, 1022 (2022)

    2022

  17. [25]

    Harmange, R

    G. Harmange, R. A. R. Hueros, D. L. Schaff, B. Emert, M. Saint-Antoine, L. C. Kim, Z. Niu, S. Nellore, M. E. Fane, G. M. Alicea, et al., Nature Communications 14, 7130 (2023)

    2023

  18. [26]

    Henrion, J

    L. Henrion, J. A. Martinez, V. Vandenbroucke, M. Delvenne, S. Telek, A. Zicler, A. Gr¨ unberger, and F. Delvigne, Nature Communications 14, 6128 (2023)

    2023

  19. [27]

    E. B. Josephs, New Phytologist 219, 31 (2018)

    2018

  20. [28]

    Vahdat, S

    Z. Vahdat, S. Rezaee, and A. Singh, IF AC-PapersOnLine 58, 276 (2024). xii

    2024

  21. [29]

    Zhang, I

    Z. Zhang, I. Zabaikina, C. Nieto, Z. Vahdat, P. Bokes, and A. Singh, PLOS Computational Biology 21, e1013014 (2025)

    2025

  22. [30]

    M. A. Siddiq, F. Duveau, and P. J. Wittkopp, Nature Ecology & Evolution , 1 (2024)

    2024

  23. [31]

    Singh and M

    A. Singh and M. Saint-Antoine, Frontiers in Microbiology 13, 1050516 (2023)

    2023

  24. [32]

    H. H. Mattingly and T. Emonet, Proceedings of the National Academy of Sciences 119, e2117377119 (2022)

    2022

  25. [33]

    S. M. Scheiner, Annual Review of Ecology and Systematics 24, 35 (1993)

    1993

  26. [34]

    J. R. Bernhardt, M. I. O’Connor, J. M. Sunday, and A. Gonzalez, Philosophical Transactions of the Royal Society B 375, 20190454 (2020)

    2020

  27. [35]

    Libby, T

    E. Libby, T. J. Perkins, and P. S. Swain, Proceedings of the National Academy of Sciences 104, 7151 (2007)

    2007

  28. [36]

    E. V. Koonin, Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences 374, 20150065 (2016)

    2016

  29. [37]

    C. E. Shannon, Comput. Commun. Rev 5, 3 (2001)

    2001

  30. [38]

    Bermek, Foundations of Science , 1 (2024)

    E. Bermek, Foundations of Science , 1 (2024)

    2024

  31. [39]

    M. C. Donaldson-Matasci, C. T. Bergstrom, and M. Lachmann, Oikos 119, 219 (2010)

    2010

  32. [40]

    Hidalgo, J

    J. Hidalgo, J. Grilli, S. Suweis, M. A. Munoz, J. R. Banavar, and A. Maritan, Proceedings of the National Academy of Sciences 111, 10095 (2014)

    2014

  33. [41]

    J. M. Pedraza, D. A. Garcia, and M. F. P´ erez-Ortiz, Frontiers in Physics 6, 83 (2018)

    2018

  34. [42]

    A. S. Moffett and A. W. Eckford, Physical Review E105, 014403 (2022)

    2022

  35. [43]

    Rivoire and S

    O. Rivoire and S. Leibler, Journal of Statistical Physics 142, 1124 (2011)

    2011

  36. [44]

    S. F. Taylor, N. Tishby, and W. Bialek, arXiv preprint arXiv:0712.4382 (2007)

    2007 arXiv

  37. [45]

    M. A. Slein, J. R. Bernhardt, M. I. O’Connor, and S. B. Fey, Proceedings of the Royal Society B 290, 20222225 (2023)

    2023

  38. [46]

    Vermeersch, L

    L. Vermeersch, L. Cool, A. Gorkovskiy, K. Voordeckers, T. Wenseleers, and K. J. Verstrepen, Frontiers in Microbiology 13, 1004488 (2022)

    2022

  39. [47]

    Skanata and E

    A. Skanata and E. Kussell, Physical review letters 117, 038104 (2016)

    2016

  40. [48]

    D´ avila-Romero, F

    M. D´ avila-Romero, F. J. Cao-Garc ´ ıa, and L. Dinis, arXiv preprint arXiv:2406.11482 (2024)

    2024 arXiv

  41. [49]

    Shin and L

    Y.-J. Shin and L. Bleris, PloS one 5, e12785 (2010)

    2010

  42. [50]

    J. R. Van Der Meer and S. Belkin, Nature Reviews Microbiology 8, 511 (2010)

    2010

  43. [51]

    Taheri-Araghi, S

    S. Taheri-Araghi, S. Bradde, J. T. Sauls, N. S. Hill, P. A. Levin, J. Paulsson, M. Vergassola, and S. Jun, Current Biology 25, 385 (2015)

    2015

  44. [52]

    D. C. Dur´ an, C. A. Hern´ andez, E. Suesca, R. Acevedo, I. M. Acosta, D. A. Forero, F. E. Rozo, and J. M. Pedraza, Micromachines 12, 4 (2020)

    2020

  45. [53]

    M. Guha, A. Singh, and N. C. Butzin, Communications Biology 8, 206 (2025)

    2025

  46. [54]

    Hossain, A

    T. Hossain, A. Singh, and N. C. Butzin, Microbiology Spectrum 11, e01219 (2023)

    2023

  47. [55]

    Bokes and A

    P. Bokes and A. Singh, Journal of Theoretical Biology 598, 111996 (2025)

    2025

  48. [56]

    S. Tang, Y. Liu, J. Zhu, X. Cheng, L. Liu, K. Hammerschmidt, J. Zhou, and Z. Cai, Nature Communications 15, 2063 (2024)

    2024

  49. [57]

    Gardiner, Stochastic Methods, Vol

    C. Gardiner, Stochastic Methods, Vol. 4 (Springer Berlin, 2009)

    2009

  50. [58]

    Biswas, S

    S. Biswas, S. Manicka, E. Hoel, and M. Levin, Iscience 24 (2021)

    2021

  51. [59]

    Lynch and G

    M. Lynch and G. K. Marinov, Proceedings of the National Academy of Sciences 112, 15690 (2015)

    2015

  52. [60]

    Mehta, A

    P. Mehta, A. H. Lang, and D. J. Schwab, Journal of Statistical Physics 162, 1153 (2016)

    2016

  53. [61]

    Cohen, Journal of Theoretical Biology 12, 119 (1966)

    D. Cohen, Journal of Theoretical Biology 12, 119 (1966)

    1966

  54. [62]

    L. F. Seoane and R. V. Sol´ e, Royal Society Open Science 5, 172221 (2018)

    2018

  55. [63]

    Cheong, A

    R. Cheong, A. Rhee, C. J. Wang, I. Nemenman, and A. Levchenko, Science 334, 354 (2011)

    2011

  56. [64]

    L. Hahn, A. M. Walczak, and T. Mora, Physical Review Letters 131, 128401 (2023). xiii Supplementary Information S1. DERIV A TION OF FRACTION RA TE EQUA TION Let Nyj be the number of individuals of phenotype yj in environment xi, ∈ R+; i, j∈ {1, 2}. Nyj is growing exponentially...

    2023

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.