Sensitivity of evolutionary entropy in Lefkovitch matrices
Henrique M. Oliveira · 2026
Read Sensitivity of evolutionary entropy in Lefkovitch matrices by Henrique M. Oliveira (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-26170.txt.
Introduction
Henrique Oliveira studies how a population measure reacts to small changes in birth and survival rates. The measure is evolutionary entropy, coined by Demetrius. It captures how spread out a life cycle is across time. Age-based models got the first formulas. Many real populations are grouped by stage, not age. Stage-based Lefkovitch matrices describe them, and Oliveira covers the well-behaved case in full.
Plain formulas are the payoff. The stable stage mix, the generation time, and the entropy itself all come out in closed form. So do the rates at which entropy moves when fertility, growth, or retention rates shift. The entropy splits into two parts, one from stage changes and one from staying put. That split shows two distinct ways spread arises. Open-group age models drop out as a special case. Worked examples compare species and life histories.
Catalog record
| CREATOR | Henrique M. Oliveira |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 46 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2606.26170
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Sensitivity of evolutionary entropy in Lefkovitch matrices Henrique M. Oliveira Abstract Evolutionary entropy, introduced by Demetrius, is a demographic invariant that quantifies the temporal organization of structured populations. Explicit sensitivity expressions for this quantity were derived by Demetrius, Gundlach and Ziehe for age-structured Leslie matrices, establishing the foundations of entropy-based perturbation theory. In this paper we develop a complete sensitivity theory for evolutionary entropy in irreducible Lefkovitch matrices. Using the Perron–Frobenius representation of the associated Markov chain, we derive explicit closed-form expressions for the stationary distribution, generation time, evolutionary entropy and its partial derivatives with respect to fertility, transition and retention parameters. The resulting identities are expressed directly in terms of demographic coefficients, Perron eigenvectors, the dominant eigenvalue and the reproductive potential. The entropy representation obtained here gives a natural decomposition into transition and retention components and clarifies the distinct mechanisms through which demographic uncertainty is generated in stage-structured populations. We further show that the theory specializes immediately to open-group Leslie matrices, a class that has been shown to comprise a large fraction of empirical demographic models. The results extend the entropy sensitivity theory of Demetrius–Gundlach–Ziehe from age-structured to general stage-structured populations and provide practical tools for comparative demographic analysis, perturbation studies, demographic robustness, and the investigation of life-history strategies. Several biological examples are presented, illustrating how entropy decomposition and sensitivity analysis reveal complementary aspects of population organization. 1 Departamento de Matemática, Instituto Superior Técnico, Universidade de Lisboa, 1049-001 Lisboa, Portugal 2 Center for Mathematical Analysis, Geometry and Dynamical Systems (CAMGSD), Instituto Superior Técnico, Universidade de Lisboa, 1049-001 Lisboa, Portugal E-mail: [email protected] Keywords: evolutionary entropy; Lefkovitch matrices; entropy sensitivity; open-group Leslie matrices; stage-structured populations; demographic robustness. MSC 2020: 92D25, 15B48, 60J10, 94A17. 1 Introduction Structured population models constitute one of the most successful applications of matrix theory to biology. Since the pioneering works of Leslie [ 13 , 14 ] , matrix population models have provided a mathematical setting for describing population growth, stable demographic distributions, reproductive value and long-term population dynamics. To overcome the limitations of purely age-structured descriptions, Lefkovitch [ 12 ] introduced stage-structured matrices allowing individuals to remain within the same developmental stage for variable periods of time. These models have become standard tools in ecology, conservation biology, forestry, plant demography and evolutionary biology [ 3 , 21 , 15 , 20 , 17 , 16 , 19 , 11 ] . Today, matrix population models form the mathematical core of large demographic repositories such as COMADRE and COMPADRE and are routinely used to analyse structured populations across a broad range of ecological contexts. A complementary line of research was initiated by Demetrius [ 8 , 9 , 10 ] , who introduced evolutionary entropy as a quantitative measure of demographic organisation. Whereas the dominant eigenvalue describes asymptotic population growth, evolutionary entropy characterises the temporal organisation of demographic trajectories generated by the underlying life cycle. Through its interpretation in terms of Markov chains, it connects demographic structure, robustness, adaptation and organisation, thereby providing information complementary to that contained in the population growth rate. An important aspect of this theory concerns sensitivity analysis. Demetrius [ 7 ] derived sensitivity expressions for the dominant eigenvalue, relating population growth to reproductive value and age-specific vital rates. This theory was subsequently developed by Caswell and collaborators [ 3 , 4 ] . Subsequently, Demetrius, Gundlach and Ziehe [ 6 ] obtained explicit sensitivity formulae for evolutionary entropy in age-structured Leslie matrices. Their results established evolutionary entropy as a natural object for perturbation analysis and provided formulae directly expressed in terms of biologically meaningful demographic parameters. The theory of [ 6 ] , however, is restricted to Leslie matrices. This limitation is significant because many contemporary applications involve stage-, size- or state-structured populations for which Lefkovitch matrices provide a more realistic description. Although sensitivity and elasticity theory for population growth is well developed in the Lefkovitch setting, explicit sensitivity formulae for evolutionary entropy do not appear to have been obtained. A key observation motivating the present work is the similarity theorem established in [ 1 ] , which shows that every irreducible Lefkovitch matrix is similar to a suitable pseudo-Leslie matrix. This result establishes a bridge between age-structured and stage-structured population models and suggests that an entropy sensitivity theory for Lefkovitch matrices should exist. The purpose of this paper is to derive explicit closed-form expressions for evolutionary entropy, generation time, the associated Markov representation and the corresponding sensitivity coefficients in irreducible Lefkovitch matrices. The resulting formulae are expressed directly in terms of fertility, transition and retention parameters, the dominant eigenvalue and the associated reproductive potential. The extension from Leslie to Lefkovitch matrices is not obtained by a simple replacement of variables. Stage-retention effects introduce new demographic quantities, additional sources of uncertainty and new sensitivity directions. Table 1 summarizes the correspondence between the principal quantities appearing in the Demetrius–Gundlach–Ziehe theory and their Lefkovitch counterparts. Table 1: Correspondence between the entropy sensitivity theory of Demetrius–Gundlach–Ziehe and its Lefkovitch extension. Concept DGZ (Leslie) Present paper (Lefkovitch) Population structure Age classes Stage classes with retention Survivorship \ell_{j}=b_{1}\cdots b_{j-1} R_{j}=\prod_{h=1}^{j-1}\frac{b_{h}}{\lambda-c_{h+1}} Net reproductive contribution V_{j}=\ell_{j}m_{j} No direct analogue; depends on m_{j} , R_{j} and c_{j} Transition distribution p_{j}=\frac{\ell_{j}m_{j}}{\lambda^{j}} q_{j}=\frac{m_{j}R_{j}}{\lambda} Tail probabilities \sum_{k=j+1}^{d}p_{k} Q_{j}=\sum_{k=j}^{d}q_{k} Generation time T=\sum_{j}jp_{j} T=1+\sum_{j=2}^{d}\frac{Q_{j}}{\eta_{j}} Evolutionary entropy H=-\frac{1}{T}\sum_{j}p_{j}\log p_{j} H=H_{\rm tr}+H_{\rm ret} Retention uncertainty Absent -\eta_{j}\log\eta_{j}-\rho_{j}\log\rho_{j} Sensitivity variables m_{j},\ell_{j},b_{j},V_{j} m_{j},b_{j},c_{j} Main sensitivities \partial H/\partial m_{j} \partial H/\partial b_{j} \partial H/\partial\ell_{j} \partial H/\partial V_{j} \partial H/\partial m_{j} \partial H/\partial b_{j} \partial H/\partial c_{j} Limiting case — c_{j}\to 0 recovers DGZ theory The paper is organised as follows. Section 2 develops an explicit Markov representation for irreducible Lefkovitch matrices and derives closed-form expressions for the stationary distribution, generation time and evolutionary entropy. Particular attention is given to the important class of open-group Leslie matrices, which arise frequently in demographic applications. Section 3 introduces a natural decomposition of evolutionary entropy into transition and retention components. Section 4 contains the main sensitivity formulae. Section 5 presents illustrative biological examples. Finally, Section 6 discusses the implications of the results and directions for future research. 2 Entropy representation for Lefkovitch matrices Let L=\begin{pmatrix}m_{1}&m_{2}&\cdots&m_{d}\\ b_{1}&c_{2}&\cdots&0\\ 0&b_{2}&\ddots&\vdots\\ \vdots&\ddots&\ddots&c_{d}\end{pmatrix} (1) be an irreducible pure Lefkovitch matrix, where m_{j}\geq 0,\qquad b_{j}>0,\qquad c_{j}\geq 0. Denote by \lambda the Perron eigenvalue of L , and let u=(u_{1},\ldots,u_{d})^{T} be the corresponding positive right Perron eigenvector. Remark 2.1 . Here and throughout the paper, empty sums are interpreted as zero and empty products are interpreted as 1 . Furthermore, following the standard convention in information theory and entropy theory, terms of the form 0\log 0 are interpreted as zero, since \lim_{x\to 0^{+}}x\log x=0. This convention is classical in the definition of Shannon entropy and related entropy functionals [ 18 , 5 ] . Our first objective is to derive explicit expressions for the probability distribution associated with L , the stationary distribution of the corresponding Markov chain, and the resulting formula for evolutionary entropy. Theorem 2.2 (Generalized Euler–Lotka representation) . Define R_{1}=1,\qquad R_{j}=\prod_{h=1}^{j-1}\frac{b_{h}}{\lambda-c_{h+1}},\qquad j=2,\ldots,d. (2) Then u_{j}=u_{1}R_{j},\qquad j=1,\ldots,d, (3) and the quantities q_{j}=\frac{m_{j}R_{j}}{\lambda},\qquad j=1,\ldots,d, (4) satisfy \sum_{j=1}^{d}q_{j}=1. (5) Equivalently, \sum_{j=1}^{d}\frac{m_{j}R_{j}}{\lambda}=1, (6) which may be regarded as a generalized Euler–Lotka equation for irreducible Lefkovitch matrices. The proof is given in A.2 . The quantities q_{j} define a probability distribution associated with