See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/51522750

A framework for studying transient dynamics of population projection matrix

models

Article in Ecology Letters · September 2011

DOI: 10.1111/j.1461-0248.2011.01659.x · Source: PubMed

CITATIONS READS

147 850

3 authors:

Iain Stott Stuart Townley

University of Southern Denmark University of Exeter
43 PUBLICATIONS 1,823 CITATIONS 152 PUBLICATIONS 3,164 CITATIONS

SEE PROFILE SEE PROFILE

David James Hodgson

University of Exeter
221 PUBLICATIONS 9,968 CITATIONS

SEE PROFILE

All content following this page was uploaded by David James Hodgson on 24 August 2015.

The user has requested enhancement of the downloaded file.

 Ecology Letters, (2011) 14: 959–970 doi: 10.1111/j.1461-0248.2011.01659.x

REVIEW AND
SYNTHESIS A framework for studying transient dynamics of population
projection matrix models

Abstract
Iain Stott,1 Stuart Townley2 and Empirical models are central to effective conservation and population management, and should be predictive
David James Hodgson1* of real-world dynamics. Available modelling methods are diverse, but analysis usually focuses on long-term
1
Centre for Ecology and dynamics that are unable to describe the complicated short-term time series that can arise even from simple
Conservation, Biosciences, College models following ecological disturbances or perturbations. Recent interest in such transient dynamics has led to
of Life and Environmental Sciences, diverse methodologies for their quantification in density-independent, time-invariant population projection
University of Exeter Cornwall matrix (PPM) models, but the fragmented nature of this literature has stifled the widespread analysis of
Campus, Tremough, Treliever Road, transients. We review the literature on transient analyses of linear PPM models and synthesise a coherent
Penryn, Cornwall, TR10 9EZ, UK framework. We promote the use of standardised indices, and categorise indices according to their focus on
2
College of Engineering, either convergence times or transient population density, and on either transient bounds or case-specific
Mathematics and Physical Sciences, transient dynamics. We use a large database of empirical PPM models to explore relationships between indices
University of Exeter, Harrison of transient dynamics. This analysis promotes the use of population inertia as a simple, versatile and informative
Building, Streatham Campus, Exeter,
predictor of transient population density, but criticises the utility of established indices of convergence times.
Devon, EX4 4QF, UK
Our findings should guide further development of analyses of transient population dynamics using PPMs or
*Correspondence: E-mail:
other empirical modelling techniques.
[email protected]

Keywords
Amplification, attenuation, damping ratio, demography, distance measures, population inertia, population
projection matrix, reactivity, short-term dynamics, transient dynamics.

Ecology Letters (2011) 14: 959–970

and interpretation. Although density-dependent and stochastic forms

INTRODUCTION
exist (Costantino et al. 1997; Fieberg & Ellner 2001), many research
Our aim is to promote a framework for the analysis of transients projects are unable to collect the data required to parameterise such
emerging from empirical models of population dynamics. Models of models, especially for populations of conservation concern. Thus, it
prospective population dynamics are integral to conservation and is linear, time-invariant PPM models that dominate the applied
management (Menges 1990; Fujiwara & Caswell 2001; Wilson 2003; demographic literature (e.g. Pavlik & Barbour 1988; Crooks et al.
Baxter et al. 2006) and should ideally be predictive of real-world 1998; Grenier et al. 2007), despite criticisms that they have
population dynamics. Numerous population modelling methods now limitations with respect to describing observed dynamics of natural
exist, including simple linear or logistic equations (e.g. Verhulst 1838; populations (Bierzychudek 1999; Stephens et al. 2002). Historically,
Volterra 1926; Lotka 1932), difference equations (e.g. May 1975), analysis of these models (and indeed of many other classes of
matrix models (e.g. Leslie 1945; Lefkovitch 1965), integral projection population model) has focused on long-term population dynamics.
models (e.g. Childs et al. 2004; Ellner & Rees 2006, 2007) and Simple models tend to concentrate on analysing properties of stable
individual-based models (e.g. van Winkle et al. 1993). Most models asymptotic equilibria such as long-term growth rate and population
have simple density-independent, time-invariant forms, but many structure (e.g. Pinard 1993; Shea & Kelly 1998; Otway et al. 2004).
models may include drivers of more complicated population Density-dependent models largely focus on analysing attractors
dynamics: nonlinear density-dependent forms incorporate intrinsic including stable equilibria such as carrying capacity or unstable
drivers, such as density dependence in vital rates or Allee effects (e.g. equilibria such as limit cycles and chaotic dynamics (e.g. Turchin
Costantino et al. 1997; Dennis et al. 2001; Fowler & Ruxton 2002), 1993; Dennis et al. 2001). Stochastic models aim to simulate or
whereas time-varying stochastic forms incorporate extrinsic drivers, provide analytical solutions to expected values of and variation in
such as variation in climate or resource abundance (e.g. Fieberg & long-term stochastic growth rate and population size (e.g. Tuljapur-
Ellner 2001; Tuljapurkar et al. 2002). Nonlinear, time-varying forms kar 1990; Morris & Doak 2005; Nantel et al. 1996). The tendency to
combine both these density dependent and stochastic approaches (e.g. focus on equilibria and long-term dynamics is perhaps a hangover
Bjørnstad & Grenfell 2001; Costantino et al. 2005). from early theoretical paradigms that perceived ecological systems as
One particular class of model, the population projection matrix ÔbalancedÕ (Cuddington 2001). In addition, because such long-term
(PPM), has held enduring popularity in population modelling. PPM dynamics are largely independent of population attributes such as
models benefit from being intuitive and tractable, and from the initial population size or structure, they lend themselves to relatively
availability of a mature literature on their parameterisation, analysis easy calculation.

 2011 Blackwell Publishing Ltd/CNRS

960 I. Stott, S. Townley and D. J. Hodgson Review and Synthesis

However, there has been a recent surge of interest in short-term, non-equilibrium dynamics without recourse to the computationally
TRANSIENT DYNAMICS of populations (see Glossary for defini- expensive study of fully stochastic systems. In a predictive sense, they
tion of capitalised terms). Density-independent, time-invariant PPM are perhaps most useful for populations that usually experience
models predict that if a population experiences a constant environ- relatively stable conditions, but which are occasionally subject to
ment with unlimited resources, then it settles to a long-term stable rate significant disturbance or perturbation.
of ASYMPTOTIC GROWTH (or decline) and a theoretical STABLE Transient dynamics have already been shown to have important
DEMOGRAPHIC DISTRIBUTION. However, in reality, popula- consequences for population management. Anthropogenic fires cause
tions experience a changeable and heterogeneous world, where increased transient growth of the reproductive proportion of popu-
extrinsic disruptions to population structure of biotic origin (e.g. lations of endangered golden mountain heather (Hudsonia Montana) in
disease, predation pressure or competition), abiotic origin (e.g. North Carolina (Gross et al. 1998). Following hunting cessation, a
changing climate, extreme weather events or fire) and anthropogenic managed population of red deer (Cervus elaphus) on the Isle of Rum
origin (e.g. harvesting or management intervention) lead to a realised showed prolonged transient fluctuations in size (Coulson et al. 2004).
INITIAL DEMOGRAPHIC DISTRIBUTION that is different from Analysis of transients also helps to improve our understanding of
the populationÕs stable demographic distribution. This can be a result population invasion. In invasive plant populations, early life-stage vital
of either DISTURBANCES to population structure that are asym- rates strongly influence the transient dynamics that promote estab-
metric across the life cycle (changing the initial demographic lishment and population growth (McMahon & Metcalf 2008; Ezard
distribution) and ⁄ or PERTURBATIONS to vital rates of the et al. 2010): this is intuitive considering seeds are the dispersive unit for
population (changing the stable demographic distribution). Discrep- plants. Conversely, in many animal populations, adults are more likely
ancies between initial and stable distributions will result in short-term to disperse, and pea aphid (Acyrthosiphon pisum) populations may exhibit
increases in population density (population AMPLIFICATION), immediate population growth that is greater than predicted asymptotic
decreases in population density (population ATTENUATION), growth, following experimentally induced adult invasion scenarios
and ⁄ or fluctuations in density that are very different to those that (Tenhumberg et al. 2009). Comparative studies of transient dynamics
might be expected in an immutable environment (Hastings 2004; provide information that should help advise population management
Townley et al. 2007; Tenhumberg et al. 2009). In the absence of in the absence of sufficient data. Monocarpic herbs and trees have been
further disturbance or perturbation, transient dynamics would dampen shown to exhibit greater potential amplitudes of transient amplification
and the population would settle back to stable state. The time taken and attenuation than perennial herbs and shrubs (Stott et al. 2010a).
for stable state to be reached following disturbance or perturbation is Long-lived animal species with slow reproduction have been shown to
termed the TRANSIENT PERIOD (Fig. 1). The ability to understand have greater variability in potential transient growth than short-lived
and quantify such transient dynamics would enable managers to both species with fast reproduction (Koons et al. 2005). The importance of
ameliorate adverse transient response to natural disturbance and transients in shaping natural population dynamics is likely to become
perturbation, and deliberately manipulate transient response through more apparent, as transient analyses become a more established part of
anthropogenic disturbance and perturbation. A thorough understand- the demographerÕs toolbox.
ing of transients may improve the predictive power of the simple Unfortunately, the emerging literature on transient analysis is
density-independent, time-invariant PPM models that dominate the disparate and rather fragmented, lacking the coherence of approaches
literature and, combined with other approaches, may aid in our to the analysis of long-term dynamics in population models. A large
comprehension of how complicated population dynamics are shaped. number of methodologies have been developed in recent years for
The transient dynamics that we discuss here form a middle-ground of quantifying transient dynamics and their relationship to vital rates, and
empirical modelling: they describe deterministic responses to (possibly these have almost exclusively focused on density-independent, time-
stochastic) disturbances or perturbations, allowing a consideration of invariant PPM models: this class of model lends itself readily to
development of transient analyses. A coherent approach to the analysis
Transient period of transient dynamics in such models is warranted. In this article, we
Density in current stage distribution

1 Initial provide a critical review of existing methodologies that focuses on

1 indices that have been developed, how they are calculated, their
Population density

Amplification
interpretation and potential applications. We highlight differences,
0 similarities and even mathematical equivalences among such published
indices. Our primary aim is to consolidate this information into a
0.5 1 Stable common framework for transient analysis and identify important
Attenutation
Initial indices that capture much of the variation in transient dynamics of a
Stable population. Although our focus is on linear, time-invariant models, we
0 anticipate that this emerging framework will inform the development of
λmax = 0.958
0 1 8 transient analyses for density-dependent and stochastic PPM models
0 10 20 30 Life stage and wider classes of population model. Throughout, we provide
Projection intervals necessary mathematical methodology and clarify terminology surround-
ing the study of transient dynamics and wider population ecology.
Figure 1 Illustration of transient population dynamics arising from a density-
independent, time-invariant PPM model. A population with stable demographic
distribution (dashed line; hatched bar plot) declines at a rate equal to kmax each time TRANSIENT DYNAMICS IN PPM MODELS
interval. However, real-world populations have initial demographic distributions
that differ from the stable demographic distribution (solid line; grey bar plot) and Transient dynamics in density-independent, time-invariant PPM
exhibit transient dynamics. models do not benefit from the simple analytical solutions available

 2011 Blackwell Publishing Ltd/CNRS

Review and Synthesis A synthesis of transient demography 961

for asymptotic dynamics. For the model nt = Atn0 (where A is the initial distribution of the population, and so cannot provide
PPM, nt is the demographic distribution vector of the population at information on convergence and the transient period given a
time t and n0 is the initial demographic distribution vector of the particular disturbance scenario. As such, it is likely to be of limited
population), the rate of asymptotic growth (or decline) is simply equal use for population managers and researchers interested in population
to kmax (the dominant eigenvalue of A). The dominant right response to specific natural or anthropogenic disturbance. Indeed, we
eigenvector w represents the relative proportions of life stages in the use a large set of empirical PPM models to show below that the
stable demographic distribution, and the dominant left eigenvector v damping ratio correlates relatively weakly with convergence times of
represents the relative reproductive value of each life stage (see Caswell realistic population projections.
2001). Once the population settles to stable state, asymptotic properties Another class of convergence measures makes it possible to
have sole influence over population density, growth and structure. incorporate specific disturbance scenarios. Distance measures calcu-
They are independent of the initial demographic distribution of the late the ÔdistanceÕ to stable state, incorporating explicitly the degree of
population, and they are insensitive to time. However, transient difference between the initial and stable demographic distributions in
dynamics (and the indices used to describe them) vary in two their calculation. Each distance measure has a different interpretation
dimensions: first, along the ÔtimeÕ axis of the population projection (i.e. and corresponding strengths and weaknesses. The simplest distance
in rate of convergence to stable state) and second, along the Ôpopulation measures suffer from the opposite problem to the damping ratio in
densityÕ axis of the population projection (i.e. in whether they amplify, that they do not consider intrinsic resilience of the population to
attenuate and ⁄ or oscillate and in the magnitude of these fluctuations). disturbance. For example, KeyfitzÕs D (Keyfitz 1968) measures the
Variation along both axes depends strongly on the initial demographic proportional difference between the initial and stable demographic
distribution. However, variation along both axes is also limited: distribution vectors (Table S1), ignoring other PPM parameters in its
transient dynamics are bounded both in rate of convergence and in calculation. Some distance measures incorporate further PPM
possible amplification and attenuation. Therefore, indices of transient parameters: for example, projection distance (Haridas & Tuljapurkar
dynamics focus either on convergence time or transient population 2007) measures the difference in the reproductive value of the initial
density, and can differ in their measurement of either TRANSIENT and the stable demographic distributions (Table S1), utilising the
BOUNDS or CASE-SPECIFIC indices of transient dynamics. reproductive value vector v in its calculation. Other distance measures
are even more informative: CohenÕs cumulative distance metric (Cohen
1979) incorporates the intrinsic resilience of the population to
Convergence rates
disturbance, with a formula that incorporates the ÔpathÕ taken by the
The earliest widely used transient index focuses on model conver- initial demographic distribution, as it converges towards stable
gence rate. The damping ratio (Caswell 2001 p. 95) of a PPM model is demographic distribution through time.
calculated as the ratio of the dominant eigenvalue to the magnitude With each distance metric having a different interpretation, the
of the first subdominant eigenvalue (see Table S1). It can be utility of each will depend on the nature of the study under
considered as a measure of the intrinsic resilience of the population, consideration. One drawback common to all distance measures is that
describing how quickly transient dynamics decay following distur- they cannot give an objective estimate of convergence time given a
bance or perturbation, regardless of population structure (the larger particular disturbance scenario. As such, they are most useful in
the damping ratio, the quicker the population converges). It is comparative analyses, comparing case-specific response to disturbance
certainly a useful measure in comparing relative resilience across or perturbation. Despite this, distance measures have not seen the
populations or species, and has been employed as such in many same popularity as the damping ratio in comparative analyses. There is
comparative analyses. For example, slower-growing corals may have a a clear need for an index that can provide an estimate of time to
higher damping ratio than faster-growing corals (Hughes & Tanner convergence given a particular disturbance scenario. Quasi-conver-
2000), indicating that slower-growing corals may be more susceptible gence given a specific population structure can be simulated by
to disturbance or perturbation. Similarly, late-successional plant projecting the model and calculating the time taken to achieve
species such as trees or shrubs tend to have a lower damping ratio asymptotic growth to a specified accuracy (e.g. for the model to
than early-successional species such as perennial herbs (Franco & exhibit growth within 1% of kmax); however, this is a relatively
Silvertown 2004). Many other comparative analyses have also used the computationally intensive solution to the problem.
damping ratio to infer population resilience (e.g. OÕConnor 1993;
Mollet & Cailliet 2002; Salguero-Gómez & Casper 2010).
Population density & growth
The damping ratio itself is a dimensionless measure and does not
provide a direct quantification of the transient period. However, it can There are a number of benefits to studying transient population size,
be implemented in equations that will provide this information. For density and growth. Population managers will often aim either to
example, the time for the influence of kmax to become x times as great boost the density and growth of populations (e.g. for conservation or
as k2 is equal to the logarithm of x divided by the logarithm of the harvesting) or to curb them (e.g. for control of pests and invasive
damping ratio (Caswell 2001 p. 96). In a strict mathematical sense, the species). Understanding transient response to natural or induced
population never reaches stable state, but merely continues to disturbance and perturbation can help to achieve these aims; indeed an
approach it indefinitely. Therefore, considering quasi-convergence of ignorance of transient responses could hinder progress when
the model in this way is perhaps the only way of prescribing a finite employing management strategies developed using asymptotic analysis
timeframe to the transient period. However, the length of the (Koons et al. 2007a). When studying resistance and resilience to
transient period depends not only on the inherent resilience of disturbance and perturbation, measuring immediate changes in
the population but also on its structure: the damping ratio ignores the population density following disturbance or perturbation is an

 2011 Blackwell Publishing Ltd/CNRS

 962 I. Stott, S. Townley and D. J. Hodgson Review and Synthesis

alternative to evaluating convergence rate (Neubert & Caswell 1997). Disentangling transient and asymptotic effects
Populations may never reach stable state in the natural environment There are two approaches to the study of transient amplification or
(Townley et al. 2007; Buckley et al. 2010; Stott et al. 2010a) and in any attenuation in population density. First, absolute measures of transient
case, most ecological studies are conducted on timescales that are population density describe how big or small a population can become
shorter than the time it would take to do so (Hastings 2004). Hence, in the short term: such measures describe the combined influence of
there is a conceptual flaw in analysing asymptotic population density transient growth rates and asymptotic dynamics (Fig. 2a). Second,
and growth whilst ignoring transients, and studying convergence rate relative measures of transient population density describe how big or
alone will be of limited use. Consequently, recent literature has been small a population can become in the short term, relative to
concerned with quantifying transient population size, structure, asymptotic dynamics (Fig. 2b). Relative measures of transient dynam-
density and growth. ics hold several advantages over absolute measures of transient
However, the quantification of transient population density and population density. First, they clarify the concept of attenuation in
growth requires a number of decisions to be made. The first decision asymptotically declining populations (attenuation is short-term decline
is simply how to define ÔtransientÕ. It is difficult to objectively delimit in density at a faster rate than asymptotic decline) and amplification in
the transient period (Maron et al. 2010), and the length of the transient asymptotically increasing populations (amplification is short-term
period is highly variable among different models. Defining ÔtransientÕ increase in density at a faster rate than asymptotic increase). Figure 2a
in terms of population density and structure is equally difficult: illustrates how, for an asymptotically declining population of the
asymptotic model properties always exert some influence over the desert tortoise Gopherus agassizii, it is difficult to capture attenuation as
population projection, albeit a decreasing influence, the nearer-term the population approaches extinction. Conversely, Fig. 2b demon-
the analysis. Furthermore, by their very nature, transient dynamics are strates that attenuation of the population is much easier to understand
highly sensitive to the demographic population structure used in the when the influence of asymptotic growth is discounted. Second,
model. Considering these issues, we have identified three main relative measures enable the fair comparison of both amplified and
problems to overcome in quantifying transients: (1) disentangling attenuated dynamics of a single population, through projection of
transient and asymptotic effects, (2) choosing an initial demographic different initial demographic distributions through the same PPM
distribution and (3) choosing the timeframe for analysis. We discuss (Fig. 2b). Third, relative measures enable fair comparative study of
existing methodologies with respect to how they deal with these three transient dynamics among populations or species with widely varying
problems. asymptotic growth (Stott et al. 2010a). Therefore, relative measures

6 (a)
Population density

1 Upper
4 bound

0
2

1 Initial i

0
0 10 20 30
Density in current stage distribution

Projection intervals 0
(b)
10 1 Initial ii

Upper bound
Population density relative to λmax(A)

0
Initial i
1 Stable

Intial ii
1 0 Figure 2 Transient dynamics of the desert tortoise Gopherus
Stable agassizii with medium fecundity (Doak et al. 1994). (a) Absolute
1 Initial iii
population dynamics, including both transient and asymptotic
influences; (b) standardised transient dynamics, excluding the
Initial iii influence of asymptotic growth. All demographic distributions are
0 scaled, so that overall initial population density equals 1. Bold lines
and black barplots indicate transient bounds (note that sometimes,
1 Lower unlike in this case, different stage-biased projections define the
Lower bound bound
amplification envelope at different times in the projection); thin
lines and grey barplots indicate case-specific initial demographic
0.1 distributions; dashed lines and hatched barplots indicate the stable
0
1 8 demographic distribution. Areas shaded in light grey indicate the
0 10 20 30
transient envelope, which is the range of values in which all case-
Projection intervals Life stage specific projections will lie.

 2011 Blackwell Publishing Ltd/CNRS

Review and Synthesis A synthesis of transient demography 963

add to the list of standardised, emergent properties of a PPM model BIASED VECTORS of demographic distribution (Townley &
that are both qualitatively and quantitatively comparable within and Hodgson 2008), where all individuals in the population are grouped
across studies and models. Relative measures do not lack an intuitive in a single stage. In the G. agassizii model, maximal possible
biological interpretation: discounting asymptotic growth yields values amplification is achieved from a population of just adults, whereas
for population density, relative to those of a population with the same maximal possible attenuation is achieved from a population of just
initial density that grows at its predicted asymptotic rate (Koons et al. yearlings (Fig. 2b). Although there do exist situations where all
2007b; Townley et al. 2007). In addition, relative measures of density individuals in a population might be in the same life stage (for
can easily be converted back to absolute measures of density if example, in a biological invasion or reintroduction programme), it is
required, simply by multiplying relative density by ktmax . unlikely that the dynamics of real populations will follow such extreme
The process of removing the influence of asymptotic dynamics can trajectories. The transient envelope is therefore useful in describing
be done very easily in practice. An intuitive solution may be to project the range of possible transient densities, but fails to inform on the
two models, one with the initial and one with the stable demographic probability of certain population sizes being achieved. For example, a
distribution, and compare the two. However, this is unnecessarily model might show an increased propensity for amplification over
computationally intensive. A second solution is to incorporate the attenuation (indeed this is the case for the G. agassizii model in Fig. 2),
correction into formulae for transient indices, for example, by scaling but the transient envelope does not provide this vital information.
the eigenvectors of the matrix (e.g. Koons et al. 2007b). This solution, Simulation of population dynamics over the entire range of possible
although adequate for individual indices, is lacking in generality: it demographic distributions (e.g. Koons et al. 2005) can provide this
does not allow for more detailed study, such as of other transient detailed information. Imagine the shaded areas in Fig. 2, but shaded
timeframes or of lifestage-specific dynamics. A third solution used by darker for more likely transient population densities and lighter for
Townley & Hodgson (2008) in calculating their transient indices less likely transient densities. This may be useful for population
(Table S1) is to use the ÔstandardisedÕ PPM Â, which is equal to the management, but is computationally demanding and an unwieldy
PPM A divided by kmax. This is a computationally simple solution, and output to use in comparative research.
enables complete projection of population dynamics in the absence of Case-specific indices of transient dynamics and transient bounds are
asymptotic dynamical effects. Hence, any analysis that can be applied both valuable measures of transient dynamics, but it is important that
to the PPM can also be applied to the standardised PPM, including both are comparable across models. Standardising the initial stage
analysis at any timeframe of the projection or analysis of lifestage- distribution so that overall population density is equal to 1 (e.g. Maron
specific dynamics. We promote this as the simplest and most et al. 2010; Koons et al. 2005; Townley et al. 2007; Townley &
comprehensive solution to the problem of disentangling transient Hodgson 2008; Stott et al. 2010a) achieves this. Initialising a
from asymptotic effects. population projection with an overall density of 1 holds several
major advantages: first, it enables fair comparison of transient
Choosing an initial demographic distribution dynamics of different initial stage distributions projected through
Transient dynamics are very sensitive to the initial demographic the same model. In Fig. 2b, for example, it is easy to see how different
distribution used in the population projection (Koons et al. 2005; initial demographic distributions result in different transient responses
Caswell 2007; Townley & Hodgson 2008), and there are two when each starts with at an initial overall density of 1. Second, it
established approaches available when it comes to choosing an initial enables fair comparative study of transient dynamics among models
demographic distribution to work with. First, case-specific measures for different populations or species (Maron et al. 2010; Stott et al.
of transient density describe the predicted dynamics of the population, 2010a). Third, coupling this approach with the study of relative
given a known initial demographic distribution. Second, bounds on measures of transient dynamics yields values for projected population
transient population density represent the most extreme possible density that are not only relative to asymptotic dynamics but also
values of amplification and attenuation, and require no prior relative to initial density. For example, in Fig. 2b, a projected density
knowledge of the populationÕs demographic distribution. For effective of 2 means that the population doubles in size relative to asymptotic
population management, it will most often be better to study case- growth, whereas a projected density of 0.5 means that the population
specific transient dynamics where possible (e.g. Zúñiga-Vega et al. halves in size relative to asymptotic growth.
2007). However, estimating the demographic distribution of a
population can prove to be costly and labour-intensive and so is Choosing the timeframe for analysis
often infeasible. Transient bounds provide an alternative approach to Transient dynamics can be measured at any point along the population
population management when the demographic distribution of the projection. The first, and perhaps most intuitive, approach is to
population is unknown, and provide best- and worst-case scenarios of measure near-term, time-dependent transient dynamics: in this case,
transient change in population density (Townley & Hodgson 2008). the timeframe chosen is very important. Very near-term measure-
The realised dynamics of the population will lie anywhere between the ments risk overlooking the full extent of transient dynamics if the
transient bounds on population density, and we call this range of population continues to amplify or attenuate beyond the timeframe
possible values the TRANSIENT ENVELOPE (Fig. 2b). Transient studied. Conversely, measurements taken further along the projection
bounds can also prove useful for comparative studies, both because will risk increased dilution by asymptotic dynamics as the population
they require no knowledge of population structure and because they converges to stable state. Choosing arbitrary time points along the
invoke like responses among models (Stott et al. 2010a). projection at which to analyse transient dynamics has been the
Calculation of case-specific transient dynamics merely requires solution of many comparative studies (e.g. Koons et al. 2005;
projection of the known initial demographic distribution. Transient McMahon & Metcalf 2008; Maron et al. 2010), but often such
bounds on population density result from projection of STAGE- measurements are not comparable as they are not necessarily

 2011 Blackwell Publishing Ltd/CNRS

964 I. Stott, S. Townley and D. J. Hodgson Review and Synthesis

indicative of preceding or subsequent dynamics. A second approach is Townley & Hodgson 2008; Table S1) and the lower Kreiss bound is an
to measure the asymptotic effects of transient dynamics: although this upper bound on maximal possible attenuation (Townley & Hodgson
may seem paradoxical, this method provides easily calculable and 2008; Table S1). These measures are time-independent and, thanks to
amenable indices that can supply very useful information. A third the existence of analytical formulae for their calculation, provide a
approach is to do away with time altogether in calculations, with time- gateway to perturbation analysis for transient bounds on maximal
independent measures of transient dynamics. The benefits and amplification and attenuation. However, the Kreiss bounds also suffer
drawbacks of each method depend on the context in which they are from drawbacks: first, they have no precise biological interpretation.
used. Second, they are currently defined only as transient bounds and not
For near-term transient analysis, an intuitive solution is to study for case-specific dynamics (although the necessary algebraic adjust-
immediate transient response, i.e. population density and growth in ment is not complicated). Last, they are somewhat redundant when
the first time interval. Reactivity and first-timestep attenuation are the considering their relationship to other indices, as we show below.
maximal possible amplification and attenuation in the first timestep
(Neubert & Caswell 1997; Townley et al. 2007; Townley & Hodgson
A FRAMEWORK FOR TRANSIENT ANALYSIS
2008; Table S1). This instantaneous transient response will always be
in the transient period, but limiting evaluation to the first time interval The transient analysis of density-independent, time-invariant PPM
risks missing the full extent of transient dynamics. A complementary models would benefit from a coherent framework within which to
solution to this problem is to measure maximal transient response work. When measuring transient population density and growth, the
alongside first-timestep response. Maximal amplification and maximal discounting of asymptotic model properties has a number of benefits,
attenuation are the largest possible amplification and attenuation that and this is most easily done by using the standardised PPM Â, where
may be achieved at any time point of population projection (Neubert  = A ⁄ kmax. In addition, standardising the initial demographic
& Caswell 1997; Townley et al. 2007; Townley & Hodgson 2008; distribution n0 to give b n0 , where jjb
n0 jj1 ¼ 1 means that transient
Table S1). Measuring this ÔbiggestÕ transient response is effective for dynamics can be studied relative to both initial density as well as to
capturing the full extent of transient dynamics: population density asymptotic growth. In combination, these standardisations make
should always be smaller than maximal amplification and larger than transient indices more meaningful both for population management
maximal attenuation. These four near-term indices were all originally and comparative analysis, and allow fair comparison of results both
defined as transient bounds, but they also exist for case-specific within and among models.
projections and collectively capture the majority of variation in near- Deciding the timeframe for analysis of transient density and growth
term transient response, as we show below. However, they are not is a harder problem to solve: different transient indices have different
particularly amenable: maximal amplification and attenuation must be interpretations, but often show highly correlated relationships with
calculated numerically, and are therefore not readily disposed to one another (Stott et al. 2010a). We exploit a database of 563
PERTURBATION ANALYSIS. Analytical perturbation analyses of published, irreducible (Stott et al. 2010b) PPM models for 202
reactivity and first-timestep attenuation are more feasible, but will do species of animals and plants (Appendix S1) to further explore the
little to inform on transient response beyond the first timestep. relationships between transient indices measured using different
Alternatively, studying the asymptotic effects of transient dynamics timeframes. All mathematical and statistical modelling was conducted
can prove to be very informative. Early demographic studies of human using R version 2.12.1 (R Development Core Team 2011). First of all,
populations considered population momentum (Keyfitz 1971; Table S1), we note a special relationship between the Kreiss bound and transient
the latent increase in human population size following an immediate bounds on population inertia. The derivation of the Kreiss bounds
decrease in birth rates to the level of replacement. Population inertia involves maximising or minimising over a scaling factor r for r > 1
(Koons et al. 2007b; Table S1) is a logical extension of this: a population (Table S1). However, where the maximum ⁄ minimum occurs at
with any given demographic distribution, following any disturbance or r fi 1, the upper and lower Kreiss bounds are identical to the upper
perturbation, will settle asymptotically to a fixed ratio above or below and lower bounds on population inertia respectively. We calculated
what is expected from a similar population growing at asymptotic rate these measurements for each PPM in our database and found that in
(Fig. 2b). Therefore, population momentum is a special case of 76.1% of cases, the upper Kreiss bound was identical to the upper
population inertia (Koons et al. 2007b; Table S1). Population inertia bound on population inertia, whereas in 99.8% of cases, the lower
was originally defined for case-specific projections, but bounds on Kreiss bound was identical to the lower bound on population inertia.
population inertia also exist. Amplified asymptotic multiplication and In both cases, SpearmanÕs rank correlations between the Kreiss
attenuated asymptotic multiplication (Townley & Hodgson 2008; Table S1) bounds and their respective bound on inertia yielded coefficients of
are bounds on population inertia, despite being calculated differently. greater than 0.99. So, the Kreiss bounds are somewhat redundant in
Although more a consequence of transient dynamics than a measure of comparison to population inertia, which is easier to interpret, simpler
transient dynamics per se, population inertia can be thought of as a to calculate and more flexible.
holistic measure that results from the combined effect of dynamics over Given this information, we would consider first-timestep (reactivity,
the whole transient period and we show below that population inertia first-timestep attenuation), maximal (maximal amplification, maximal attenu-
correlates tightly with the other indices of transient dynamics. A final ation) and asymptotic (amplified inertia, attenuated inertia) indices of
strength of inertia is that it is very amenable to perturbation analysis, as it transient dynamics to be the most useful measurements of transient
is a simple function of the initial stage distribution and the dominant population density and growth (Table 1). These indices and their
eigenvectors of the PPM. bounds together describe the majority of variation in transient
A final method for overcoming the problem of time is to do away population density, are comparable within and across models and all
with it in calculations altogether. The upper Kreiss bound is a lower have a definite biological interpretation. Relationships between these
bound on maximal possible amplification (Townley et al. 2007; measurements are presented in Fig. 3. To study relationships between

 2011 Blackwell Publishing Ltd/CNRS

Review and Synthesis A synthesis of transient demography 965

bounds on transient dynamics, we calculated the six bounds for each We repeated this process 1000 times to obtain distributions of
PPM and correlated them against one another, the results of which are correlation coefficients for the pairwise comparisons. Although the
presented in the upper-right triangle of Fig. 3. To study relationships use of a uniform distribution is somewhat artificial given that relative
between case-specific indices, we generated random demographic densities of life stages are likely to co-vary in natural populations, it
distribution vectors for each PPM by drawing numbers from uniform allowed exploration of a wide range of potential stage structures. The
distributions, standardised these vectors to sum to 1, calculated the six results are presented in the lower-left triangle of Fig. 3. It appears that
case-specific transient indices for each model and performed for both transient bounds and case-specific indices, amplified
SpearmanÕs rank correlations on each pairwise combination of indices. measures show very tight positive correlations with one another

Table 1 Important indices of transient population dynamics. Matrices are presented as capitalised, and in bold. Vectors are in small type and in bold. Numbers and scalars are
in normal font

INDEX FORMULA BIOLOGICAL MEANING

 
REACTIVITY  1 ¼ A
TRANSIENT BOUND q ^
1
TRANSIENT BOUND – maximum population growth in a
single timestep, relative to stable growth rate.
   
 1 ¼ A
CASE-SPECIFIC P ^0 1 when A
^n ^0 1 >1
^n CASE-SPECIFIC – population growth achieved by a given
population structure in a single timestep, relative to stable
growth rate. Assumes the population will amplify immediately.
FIRST-TIMESTEP ^Þ
TRANSIENT BOUND q1 ¼ minCSðA TRANSIENT BOUND – minimum population growth in a
ATTENUATION single timestep, relative to stable growth rate.
   
CASE-SPECIFIC P1 ¼ A^
^ n0  when A^
1
^ n0  <1
1
CASE-SPECIFIC – population growth achieved by a given
population in a single timestep, relative to stable growth rate.
Assumes the population will attenuate immediately.
 t 
MAXIMUM TRANSIENT BOUND q ^ k1
max ¼ max kA TRANSIENT BOUND – the largest possible future population
t>0
AMPLIFICATION density achievable, relative to a population with stable growth
rate and same initial density.
 t 
^n
CASE-SPECIFIC Pmax ¼ max kA ^ tn
^0 k1 whenkA ^0 k1 >1 for some t CASE-SPECIFIC – the largest possible future density of a given
t>0
population structure, relative to a population with stable growth
rate and same initial density. Assumes the population amplifies
at some point in the future.
 
MAXIMUM ^ tÞ
TRANSIENT BOUND qmin ¼ min minCSðA TRANSIENT BOUND – the smallest possible future population
t>0
ATTENUATION density achievable, relative to a population with stable growth
rate and same initial density.
 t   t 
CASE-SPECIFIC Pmin ¼ min A
^n^0 1 when A
^n^0 1 <1 for some t CASE-SPECIFIC – the smallest possible future density of a
t>0
given population structure, relative to a population with
stable growth rate and same initial density. Assumes the
population attenuates at some point in the future.
vmax kwk1
AMPLIFIED INERTIA 1 ¼
TRANSIENT BOUND q TRANSIENT BOUND – the largest possible long-term
vT w
population density, relative to a population with stable growth
rate and same initial density.
vT n
^0 kwk1 vT n
^0 kwk1
CASE-SPECIFIC P1 ¼ T when >1 CASE-SPECIFIC – the long-term population density of a
v w vT w
given population structure, relative to a population with stable
growth and same initial density. Assumes the population
amplifies in the long term.
vmin kwk1
ATTENUATED TRANSIENT BOUND q1 ¼ TRANSIENT BOUND – the smallest possible long-term
vT w
INERTIA population density, relative to a population with stable growth
rate and same initial density.
vT n
^0 kwk1 vT n
^0 kwk1
CASE-SPECIFIC P1 ¼ when <1 CASE-SPECIFIC – the long-term population density of a
vT w vT w
given population structure, relative to a population with
stable growth and same initial density. Assumes the population
attenuates in the long term.

 represents the standardised population projection matrix and is equal to A ⁄ kmax (where A is the PPM and kmax is the dominant eigenvalue of A); w represents the dominant
right eigenvector of A (the stable demographic distribution vector); v represents the dominant left eigenvector of A (the reproductive value vector); n^0 represents the initial
demographic distribution, standardised to sum to 1. minCS denotes the minimum column sum of a matrix and jjmjj1 is the one-norm of a vector m (equal to the sum of its
entries). mmin and mmax are the smallest and largest entries of a vector m respectively. We have chosen to use the Greek q to represent transient bounds (in accordance with
Townley & Hodgson 2008), whereas the Latin P represents case-specific indices of transient dynamics. An overbar indicates a bound or index of amplification, whereas an
underbar represents a bound or index of attenuation (Townley & Hodgson 2008). A subscript provides information on the timeframe of study: 1 for first-timestep indices; max
or min for maximal amplification or attenuation, respectively, and ¥ for asymptotic indices.

 2011 Blackwell Publishing Ltd/CNRS

966 I. Stott, S. Townley and D. J. Hodgson Review and Synthesis

log(transient bound)

log(transient bound)
ρmax ρ∞ ρ1 ρmin ρ∞
15
0.94 0.92
10
ρ1
5
−0.45 −0.55 −0.41
0
0 5 10 15
15
0.3
0.99
0.2 10
Pmax ρmax
0.1 5
−0.32 −0.55 −0.43
0
0
0.92 0.94 0.96 0 5 10 15
15
0.3

0.2 10
P∞ ρ∞
0.1 5
−0.29 −0.53 −0.41
0
0
0.88 0.91 0.94 0.980 0.986 0.992 –15 −10 −5 0
0
0.3
−5
0.2
P1 ρ1
0.1 −10
0.77 0.68
0 −15
−0.8 −0.1 0.6 −0.85 −0.15 0.55 −15 −10 −5 0
0
0.3
−5
0.2
Pmin ρmin
0.1 −10
0.95
0 −15
−0.75 0 0.75 −0.60 0.05 0.50 −0.9 −0.3 0.3 0.62 0.75 0.88 −15 −10 −5 0
0.3

0.2
P∞
0.1
Frequency

0
−0.8 0 0.8 −0.8 0 0.8 0.50 0.65 0.80 0.87 0.93 0.99

P1 Pmax P∞ P1 Pmin
Spearman’s correlation coefficient

Figure 3 Correlations between transient bounds and case-specific indices for 563 published population projection matrix models. The upper-right triangle shows pairwise
correlations between six transient bounds and their associated SpearmanÕs rho values; both axes are on a log scale. The lower-left triangle consists of histograms that show
distributions of SpearmanÕs rho values for 1000 pairwise correlations of case-specific indices for randomly generated case-specific demographic distributions.

(upper-left quadrant of Fig. 3), whereas attenuated measures show values. Most correlations between bounds and median convergence
tight positive correlations, but not to the same degree as among time were significant, but all were relatively weak with absolute values
amplified measures (lower-right quadrant of Fig. 3). However, of SpearmanÕs rho ranging between 0.21 and 0.7. Distributions of
amplified measures are not good predictors of attenuated measures correlation coefficients between case-specific transient dynamics and
and vice versa (upper-right and lower-left quadrants of Fig. 3). The time to convergence were similarly weak and widely varying (see
correlation coefficients of transient bounds lie comfortably within the Appendix S2 for more detailed methodology and results from these
modal range for those of case-specific transient indices, which analyses). This indicates that larger transient departures from
indicates that in relation to one another, bounds on transient dynamics asymptotic growth are not necessarily linked to a longer time to
behave similar to case-specific indices of transient dynamics. Of the convergence, either for transient bounds or for case-specific transient
three pairs of indices, population inertia correlates best with its indices. Studying the relationship between the damping ratio and the
amplified or attenuated partner indices. median simulated time to convergence also indicated that the damping
We also explored the relationships between the six indices and ratio is a relatively poor predictor of convergence, with a SpearmanÕs
measures of convergence. For each randomly generated initial correlation coefficient of )0.66 and high variability in the distribution
demographic distribution, we simulated the time taken for population of points (Appendix S2). The relationship between approximate time
growth to settle within 1% of kmax. We correlated transient bounds to convergence calculated using the damping ratio and simulated
with the median simulated convergence time of the 1000 iterations for median time to convergence is similarly weak, although with median
each PPM. For case-specific indices, we correlated the six transient time to convergence rarely exceeding approximated time to conver-
indices with time to convergence at each iteration of the model, and gence, there is a potential utility of the damping ratio in providing a
1000 iterations of this model provided distributions of SpearmanÕs rho weak bound on convergence rate. Nonetheless, there is a clear need

 2011 Blackwell Publishing Ltd/CNRS

Review and Synthesis A synthesis of transient demography 967

for an index that measures finite convergence time, which would fit population is expected to decrease by 10% compared with asymptotic
into the framework for analysis we describe here. Such a measure growth in one year, whereas the value for attenuated inertia indicates
would be qualitatively and quantitatively informative for population that given environmental stability the population is expected to
management, be calculable as both a bound and case-specific index become up to 70% smaller than asymptotic growth predicts. This is
and be a standardised index that is comparable among PPM models. important to consider: kmax for the population is 1.02, indicating that
Simulating quasi-convergence of the model goes some way to it will grow over time, when in fact the transient dynamics of the
achieving these goals, but is computationally intensive and rather ill- population may cause it to decline. Eventually (assuming no change in
defined. vital rates or canopy cover), population growth will settle to
kmax = 1.02, but at much reduced density. The ÔafterÕ projection
simulates disturbance caused by a hurricane opening up 50% of the
CASE STUDIES
existing canopy. Far from being damaging to the population, this
The transient indices and bounds identified here provide standardised, would cause amplification, even with no change in the vital rates of
comparable, qualitative and quantitative measures that can be used in the population. It would grow at a rate faster than kmax in the short
conjunction with other model parameters to inform on population term: reactivity indicates that within just 1 year, it is predicted to
state and potential management (Fig. 4). Figure 4a shows two become 5% larger than asymptotic growth predicts, whereas amplified
projections for the subcanopy tree Styrax obassia in central Japan inertia predicts the density will eventually settle to be up to 30% larger
(Abe et al. 1998). The PPM distinguishes between those individuals than expected of a population initiated at stable stage structure (again,
found in the shade and individuals found in canopy gaps: vital rates of assuming environmental stability following the disturbance event).
individuals in each environment differ considerably. The ÔbeforeÕ Figure 4b illustrates transient bounds for the Amsterdam albatross
projection uses the recorded demographic distribution, with only Diomedea amsterdamensis on Amsterdam Island in the South-Eastern
6.2% of the canopy open. This population attenuates increasingly over Indian Ocean (Inchausti & Weimerskirch 2001). This is one of the
time: the value for first-timestep attenuation indicates that the worldÕs rarest species of bird, with the Amsterdam Island colony being
the known population. Although the demographic structure of the
species has previously been recorded in 1997, the exact demographic
Population density relative to λmax(A)

1 After
Density in current stage distribution

(a) structure of the current population is unknown. In this case, transient

1.5
Pmax = P∞ bounds can provide vital information. kmax for the population is equal
After to 1.06, which is encouraging, and the values of bounds on reactivity
P1 0
and first-timestep attenuation indicate that in 1 year, the population
1 Stable 1 Stable should not decrease or increase within anything more than 10% of its
P1 size as predicted by asymptotic growth. However, despite having a
positive kmax, the worst-case scenario is that the population almost
Before
0 halves relative to asymptotic growth within the next 10 years as
Pmin = P∞
1 Before indicated by the lower bound on maximal attenuation. So, despite
having a positive predicted asymptotic growth, there is a real danger
0.5 that the population may in fact decline in size. This worst-case
0 50 100 scenario is unlikely as it results from a population composed entirely
0
Projection intervals 1 10 of chicks, but sets the goalposts for population management. Were a
Life stage reintroduction programme is to take place, the transient bounds
indicate that age 11 individuals would be the best to translocate, as it is
Population density relative to λmax(A)

1 Upper this age class that provides the largest overall population amplification
Density in current stage distribution

1.5 (b) bound

ρmax both immediately and asymptotically.
ρ∞
Upper bound
0
ρ1 11 DISCUSSION
1 Stable 0.2 Stable
ρ1 Transient population dynamics are important to consider in any
management scenario: short-term fluctuations in population density
Lower bound 0 and growth can swamp predicted asymptotic trends. Recent interest in
ρ∞ transients has given rise to diverse methodologies for calculating
1 Lower
ρmin bound transients for density-independent, time-invariant PPM models.
0.5 However, the fragmented nature of this literature has left the field
0 50 of transient analysis relatively inaccessible to non-specialists. We have
0
Projection intervals 1 24 identified an emerging framework, with a number of decisions to
Life stage make when evaluating transient population density, growth and
structure. We strongly recommend that transient dynamics be studied
Figure 4 Case studies of transient dynamics in natural populations. (a) Case-specific
relative to both asymptotic growth and initial population density. The
dynamics of the subcanopy tree Styrax obassia; (b) transient bounds of the
Amsterdam albatross Diomedea amsterdamensis. Bold lines and black barplots indicate
easiest ways to achieve this are to standardise the PPM by dividing it
transient bounds; thin lines and grey barplots indicate case-specific initial by kmax and standardise the initial demographic distribution by scaling
demographic distributions; dashed lines and hatched barplots indicate the stable it, so that overall population density is equal to 1. Analysing transient
demographic distribution. dynamics at arbitrary points along the population projection is

 2011 Blackwell Publishing Ltd/CNRS

968 I. Stott, S. Townley and D. J. Hodgson Review and Synthesis

unsatisfactory, and so the three pairs of indices identified here can be asymptotic growth and changes in vital rates of populations are often
used as comparable measures to capture the majority of variation in markedly nonlinear (Hodgson & Townley 2004) and there is evidence
transient dynamics of a population. Our analysis of correlations to show that this may also be the case for transient dynamics (Townley
between these indices reveals population inertia (Koons et al. 2007b) et al. 2007). Sensitivity analysis, as a linear approximation, is not
to be a simple yet versatile index that correlates strongly with other sufficient to describe population dynamic responses to non-negligible
indices, both when measured for case-specific transient dynamics and perturbations (Carslake et al. 2008, 2009). Therefore, there is a need
bounds on transient dynamics. There are many available methods for for a transfer-function style approach to transient perturbation
analysing convergence of models, but the biological interpretation of analysis that can model the nonlinear response of transient density
most of these indices is questionable and we have identified a need for to perturbation.
a more robust index of convergence in density-independent, time- The majority of interest in transient dynamics has so far centred
invariant PPM models. on their use in population management: far fewer studies have
Although the study of transient dynamics is relatively well considered their potential impact in other areas of ecology and
developed for density-independent, time-invariant PPM models, evolution. A better understanding of transients could provide
transients have received relatively little attention in density-dependent opportunities to understand life-history evolution from a new
and stochastic models. We anticipate that the framework we identify perspective. A population that experiences unpredictable distur-
here (i.e. study of relative transient dynamics, standardised initial bance may need to evolve to be resistant to disturbance through
conditions and first-timestep, maximal and asymptotic response) may having smaller transients, as an insurance against population
be useful in informing development of transient indices in these areas. decline. Conversely, a population that experiences regular distur-
However, the nonlinear and ⁄ or time-varying nature of such models bance of a particular type may evolve to have a larger transient
presents further obstacles to analytical solutions for transient dynamical response to that disturbance, maximising amplification to
dynamics. Nonlinear projection matrix models vary greatly in the exploit opportunity to outcompete other genotypes. It is likely that
form and influence of their attractors, therefore there is unlikely to be the simultaneous optimisation of asymptotic and transient growth
a single predictor of transient density or convergence rate that suits all rates will combine to maximise long-term stochastic growth rate,
situations. Nonlinear models demonstrating stable equilibrium density but the mechanisms mapping life-history variation onto transient
will prevent the phenomenon of asymptotic inertia in future dynamics and potential trade-offs between selection on short- and
population size as defined here for linear models, but timestep- long-term growth have not yet been explored. Transient dynamics
specific amplification and attenuation (and bounds on these) will also have relevance in ecological systems other than populations:
remain interesting and measurable. Stochastic models incorporate indeed the application of transient theory could extend to any
disturbance and perturbation as a result of small-scale fluctuations in stage-structured biological system. Transients could help to better
the environment, but near-term dynamics following larger, more understand the dynamics of communities, for example in modelling
infrequent disturbances or perturbations may differ from long-term tropic cascades in food webs. They may help to better understand
dynamic trends. It may be useful to have analytical formulae to the spread of infectious diseases in epidemiological models. They
approximate such dynamics, similar to those that exist for long-term may help explain the rapid spread of novel genotypes through
stochastic dynamics (Tuljapurkar 1982), thereby reducing the need for populations. They could be useful in modelling the responses of
full numerical simulation. The extension of transient analysis to classes ecosystems to climate change, tracking the movement of energy or
of population model other than PPMs is another logical next step. matter through different ecosystem compartments. There are some
In particular, integral projection models (Ellner & Rees 2006, 2007) examples of transient dynamics in other ecological systems – the
would benefit from development of methods for transient analysis. flow of matter through a rainforest ecosystem has been shown to
One of the most useful extensions to understanding population exhibit transient dynamics (Neubert & Caswell 1997; Townley &
dynamics per se is to understand the interplay between the vital rates of Hodgson 2008) and models of whooping cough epidemiology in
the population and its dynamics. Perturbation analyses such as humans have shown significant transient response to perturbation,
SENSITIVITY (Caswell 2001 p. 210), ELASTICITY (Caswell 2001 with annual epidemical cycles becoming multiannual cycles follow-
p. 227) and TRANSFER-FUNCTION (Hodgson & Townley 2004) ing an increase in recovery rates (Rohani et al. 2002).
analyses provide this important information for asymptotic population Transient analysis of population dynamics is still a young and
dynamics. A number of methods for transient perturbation analysis emerging field of population ecology, likely to see many advances in
exist, particularly methods for analysing sensitivity of transient coming years. Indeed, transient analysis should prove an essential part
population density to changes in vital rates of the population. These of any study of demography. However, relatively inaccessible literature
methods must make the same decisions regarding standardisation of and a lack of coherency could present a barrier to widespread use of
dynamics, choice of initial population structure and time point along some methodologies. With a common framework within which to
the projection at which to analyse. However, they have an added develop methods, and the formulation of extra tools, transient analysis
decision to make in what form of projection equation to differentiate: has the potential to provide great insight to the fields of conservation,
methods have chosen to use matrix calculus to evaluate state-space population management and evolutionary ecology.
form equations (Caswell 2007), to differentiate the solution to the
projection equation as expressed using model eigenvalues and
ACKNOWLEDGEMENTS
eigenvectors (Fox & Gurevitch 2000; Yearsley 2004) and to evaluate
sensitivity of transient indices that are functions of the PPM (Townley We are grateful to David Carslake, Miguel Franco and Vicky
et al. 2007; Koons et al. 2007b). A review and synthesis of these Warwick-Evans for the provision of PPMs to the database, and for
approaches is certainly needed. However, there is still a need for new intellectual contributions. IS was supported by the European Social
approaches to transient perturbation analysis. Relationships between Fund.

 2011 Blackwell Publishing Ltd/CNRS

Review and Synthesis A synthesis of transient demography 969

AUTHOR CONTRIBUTIONS Franco, M. & Silvertown, J. (2004). A comparative demography of plants based
upon elasticities of vital rates. Ecology, 85, 531–538.
IS wrote the first draft of the manuscript, performed modelling work Fujiwara, M. & Caswell, H. (2001). Demography of the endangered North Atlantic
and statistical analyses and created figures, tables and supporting right whale. Nature, 414, 537–541.
documents. IS, ST and DJH all provided intellectual contributions, Grenier, M.B., McDonald, D.B. & Buskirk, S.W. (2007). Rapid population growth
and were together responsible for editing final versions of the of a critically endangered carnivore. Science, 317, 779.
manuscript, figures, tables and supporting documents. Gross, K., Lockwood, J.R. III, Frost, C.C. & Morris, W.F. (1998). Modelling
controlled burning and trampling reduction for conservation of Hudsonia montana.
Conserv. Biol., 12, 1291–1301.
REFERENCES Haridas, C.V. & Tuljapurkar, S. (2007). Time, transients and elasticity. Ecol. Lett., 10,
1143–1153.
Abe, S., Nakashizuka, T. & Tanaka, H. (1998). Effects of canopy gaps on the Hastings, A. (2004). Transients: the key to long-term ecological understanding?
demography of the subcanopy tree Styrax obassia. J. Veg. Sci., 9, 787–796. Trends Ecol. Evol., 19, 39–45.
Baxter, P.W.J., McCarthy, M.A., Possingham, H.P., Menkhorst, P.W. & McLean, N. Hodgson, D.J. & Townley, S. (2004). Linking management changes to population
(2006). Accounting for management costs and sensitivity analyses of matrix dynamic responses: the transfer function of a projection matrix perturbation.
population models. Conserv. Biol., 20, 893–905. J. Appl. Ecol., 41, 1155–1161.
Bierzychudek, P. (1999). Looking backwards: assessing the projections of a tran- Hughes, T.P. & Tanner, J.E. (2000). Recruitment failure, life histories and long-term
sition matrix model. Ecol. Appl., 9, 1278–1287. decline of Caribbean corals. Ecology, 81, 2250–2263.
Bjørnstad, O.N. & Grenfell, B.T. (2001). Noisy clockwork: time series analysis of Inchausti, P. & Weimerskirch, H. (2001). Risks of decline and extinction of the
population fluctuation in animals. Science, 293, 638–643. endangered Amsterdam albatross and the projected impact of long-line fisheries.
Buckley, Y.M., Ramula, S., Blomberg, S.P., Burns, J.H., Crone, E.E., Ehrlén, J. et al. Biol. Conserv., 100, 337–386.
(2010). Causes and consequences of variation in plant population growth rate: a Keyfitz, N. (1968). Introduction to the Mathematics of Populations. Addison-Wesley,
synthesis of matrix population models in a phylogenetic context. Ecol. Lett., 13, Reading, MA.
1182–1197. Keyfitz, N. (1971). On the momentum of population growth. Demography, 8, 71–80.
Carslake, D., Townley, S. & Hodgdon, D.J. (2008). Nonlinearity in eigenvalue- Koons, D.N., Grand, J.B., Zinner, B. & Rockwell, R.F. (2005). Transient population
perturbation curves of simulated population projection matrices. Theor. Popul. dynamics: relations to life history and initial population state. Ecol. Model., 185,
Biol., 73, 498–505. 283–297.
Carslake, D., Townley, S. & Hodgson, D.J. (2009). Predicting the impact of stage- Koons, D.N., Rockwell, R.F. & Grand, J.B. (2007a). Population momentum:
specific harvesting on population dynamics. J. Anim. Ecol., 78, 1076–1085. implications for wildlife management. J. Wildlife Manage., 70, 19–26.
Caswell, H. (2001). Matrix Population Models: Construction, Analysis and Interpretation. Koons, D.N., Holmes, R.R. & Grand, J.B. (2007b). Population inertia and its sen-
Sinauer, Sunderland, MA. sitivity to changes in vital rates and population structure. Ecology, 88, 2857–2867.
Caswell, H. (2007). Sensitivity analysis of transient population dynamics. Ecol. Lett., Lefkovitch, L.P. (1965). The study of population growth in organisms grouped by
10, 1–15. stages. Biometrics, 21, 1–18.
Childs, D.Z., Rees, M., Rose, K.E., Grubb, P.J. & Ellner, S.P. (2004). Evolution of Leslie, P.H. (1945). On the use of matrices in certain population mathematics.
size-dependent flowering in a variable environment: construction and analysis of Biometrika, 33, 183–212.
a stochastic integral projection model. Proc. R. Soc. Lond. B, 271, 425–434. Lotka, A.J. (1932). The growth of mixed populations: two species competing for a
Cohen, J.E. (1979). The cumulative distance from an observed to a stable age common food supply. J. Wash. Acad. Sci., 22, 461–469.
structure. SIAM J. Appl. Math., 36, 169–175. Maron, J.L., Horvitz, C.C. & Williams, J.L. (2010). Using experiments, demography
Costantino, R.F., Desharnais, R.A. & Cushing, J.M. (1997). Chaotic dynamics in an and population models to estimate interaction strength based on transient and
insect population. Science, 275, 389–391. asymptotic dynamics. J. Ecol., 98, 290–301.
Costantino, R.F., Desharnais, R.A., Cushing, J.M., Dennis, B., Henson, S.M. & May, R.M. (1975). Biological populations obeying difference equations: stable
King, A.A. (2005). Nonlinear stochastic population dynamics: the flour beetle points, stable cycles and chaos. J. Theor. Biol., 51, 511–524.
Tribolium as an effective tool of discovery. Adv. Ecol. Res., 37, 10–141. McMahon, S.M. & Metcalf, C.J.E. (2008). Transient sensitivities of non-indigenous
Coulson, T., Guinness, F., Pemberton, J. & Clutton-Brock, T. (2004). The demo- shrub species indicated complicated invasion dynamics. Biol. Invasions, 10, 833–
graphic consequences of releasing a population of red deer from culling. Ecology, 846.
85, 411–422. Menges, E.S. (1990). Population viability for an endangered plant. Conserv. Biol., 4,
Crooks, K.R., Sanjayan, M.A. & Doak, D.F. (1998). New insights on cheetah 52–62.
conservation through demographic modelling. Conserv. Biol., 12, 889–895. Mollet, H.F. & Cailliet, G.M. (2002). Comparative population demography of
Cuddington, K. (2001). The Ôbalance of natureÕ metaphor and equilibrium in elasmobranchs using life history tables, Leslie matrices and stage-based matrix
population ecology. Biol. Philos., 16, 463–479. models. Mar. Freshwater Res., 53, 503–516.
Dennis, B., Desharnais, R.A., Cushing, J.M., Henson, S.M. & Costantino, R.F. Morris, W.F. & Doak, D.F. (2005). How general are the determinants of the
(2001). Estimating chaos and complex dynamics in an insect population. Ecol. stochastic population growth rate across nearby sites? Ecol. Monogr., 75, 119–137.
Monogr., 71, 277–303. Nantel, P., Gagnon, D. & Nault, A. (1996). Population viability analysis of
Doak, D., Kareiva, P. & Klepta, B. (1994). Modeling population viability for the American ginseng and wild leek harvested in stochastic environments. Conserv.
desert tortoise in the western Mojave desert. Ecol. Appl., 4, 446–460. Biol., 10, 608–621.
Ellner, S.P. & Rees, M. (2006). Integral projection models for species with complex Neubert, M.G. & Caswell, H. (1997). Alternatives to resilience for measuring the
demography. Am. Nat., 167, 410–428. responses of ecological systems to perturbations. Ecology, 78, 653–665.
Ellner, S.P. & Rees, M. (2007). Stochastic stable population growth in integral OÕConnor, T.G. (1993). The influence of rainfall and grazing on the demography of
projection models: theory and application. J. Math. Biol., 54, 227–256. some African savanna grasses: a matrix modelling approach. J. Appl. Ecol., 30,
Ezard, H.G., Bullock, J.M., Dalgleish, H.J., Millon, A., Pelletier, F., Ozgul, A. et al. 119–132.
(2010). Matrix models for a changeable world: the importance of transient Otway, N.M., Bradshaw, J.A. & Harcourt, R.G. (2004). Estimating the rate of quasi-
dynamics in population management. J. Appl. Ecol., 47, 515–523. extinction of the Australian grey nurse shark (Carcharias taurus) population using
Fieberg, J. & Ellner, S.P. (2001). Stochastic matrix models for conservation and deterministic age- and stage-classified models. Biol. Conserv., 119, 341–350.
management: a comparative review of methods. Ecol. Lett., 4, 244–266. Pavlik, B.M. & Barbour, M.G. (1988). Demographic monitoring of endemic sand
Fowler, M.S. & Ruxton, G.D. (2002). Population dynamic consequences of Allee dune plants, Eureka valley, California. Biol. Conserv., 46, 217–242.
effects. J. Theor. Biol., 215, 39–46. Pinard, M. (1993). Impacts of stem harvesting on populations of Iriartea deltoidea in
Fox, G.A. & Gurevitch, J. (2000). Population numbers count: tools for near-term an extractive reserve in Acre, Brazil. Biotropica, 25, 2–14.
demographic analysis. Am. Nat., 156, 242–256.

 2011 Blackwell Publishing Ltd/CNRS

970 I. Stott, S. Townley and D. J. Hodgson Review and Synthesis

R Development Core Team (2011). R: A Language and Environment for Statistical Disturbance: A change to the demographic structure of the
Computing, reference index version 2.12.1. R Foundation for Statistical Com- population (usually as a result of exogenous forces); mathematically,
puting, Vienna, Austria. Available at: http://cran.r-project.org. Last accessed 16 a change in the ratio of life stages in the initial demographic
December 2010.
distribution vector.
Rohani, P., Keeling, M.T. & Grenfell, B.T. (2002). The interplay between deter-
minism and stochasticity in childhood diseases. Am. Nat., 159, 469–481. Elasticity: The change in population density or growth resulting from
Salguero-Gómez, R. & Casper, B.B. (2010). Keeping plant shrinkage in the a perturbation of one or more vital rates of a population, relative to
demographic loop. J. Ecol., 98, 312–323. the magnitude of the vital rate(s) perturbed.
Shea, K. & Kelly, D. (1998). Estimating biocontrol agent impact with matrix Perturbation: A change to the vital rates of the population (usually as
models: Carduus nutans in New Zealand. Ecol. Appl., 8, 824–832. a result of exogenous forces); mathematically, a change in one or more
Stephens, P.A., Frey-Roos, F., Arnold, W. & Sutherland, W.J. (2002). Model
PPM elements.
complexity and population predictions. The alpine marmot as a case study.
J. Anim. Ecol., 71, 343–361.
Perturbation analysis: Any analysis that considers the change of a
Stott, I., Franco, M., Carslake, D., Townley, S. & Hodgson, D.J. (2010a). Boom or model output with respect to the change of a model input; usually
bust? A comparative analysis of transient population dynamics in plants. J. Ecol., considering the change in population density or growth resulting from
98, 302–311. perturbation of one or more vital rates of the population.
Stott, I., Townley, S., Carslake, D. & Hodgson, D.J. (2010b). On reducibility and Sensitivity: The absolute rate of change in population density or
ergodicity of population projection matrix models. Methods Ecol. Evol., 1, 242–252. growth resulting from infinitesimal perturbations of one or more vital
Tenhumberg, B., Tyre, A.J. & Rebarber, R. (2009). Model complexity affects
rates of a population.
transient population dynamics following a dispersal event: a case study with pea
aphids. Ecology, 90, 1878–1890. Stable demographic distribution: The ratios of life stages in the
Townley, S. & Hodgson, D. (2008). Erratum et addendum: transient amplification and population when it reaches stable state; mathematically, equal to the
attenuation in stage-structured population dynamics. J. Appl. Ecol., 45, 1836–1839. dominant right eigenvector of the PPM, w.
Townley, S., Carslake, D., Kellie-Smith, O., McCarthy, D. & Hodgson, D. (2007). Stage-biased vector: An initial demographic distribution vector that
Predicting transient amplification in perturbed ecological systems. J. Appl. Ecol., has all individuals in a single stage and an overall density of 1;
44, 1243–1251. mathematically, a standard basis vector with zeroes in each entry,
Tuljapurkar, S.D. (1982). Population dynamics in variable environments. III.
except for a single entry that is equal to one.
Evolutionary dynamics of r selection. Theor. Popul. Biol., 21, 141–165.
Tuljapurkar, S.D. (1990). Population dynamics in variable environments. Lect. notes Transfer function: A means of perturbation analysis derived from
Biomath., 85. Springer-Verlag, New York. systems and control theory. Describes the exact nonlinear relationship
Tuljapurkar, S., Horvitz, C.C. & Pascarella, J.B. (2002). The many growth rates and between a perturbation to a vital rate and resultant change in
elasticities of populations in random environments. Am. Nat., 162, 489–502. population density or growth.
Turchin, P. (1993). Chaos and stability in rodent population dynamics: evidence Transient bound: The most extreme values of transient dynamics that
from non-linear time-series analysis. Oikos, 68, 167–172.
may result from a PPM. Any case-specific transient dynamics must lie
Verhulst, P.F. (1838). Notice sur la loi que la population suit dans son accroisee-
ment. Correspondes Mathematiques et Physiques, 10, 113–121.
between the bounds on amplification and attenuation.
Volterra, V. (1926). Variations and Fluctuations of the Numbers of Individuals in Animal Transient dynamics: The short-term dynamics of the population,
Species Living Together. McGraw Hill, New York. (Reprinted in 1931: R.N. Chap- which are dependent on the initial conditions of the model.
man, Animal Ecology) Transient envelope: The area between the transient bounds, in which
Wilson, P.H. (2003). Using population projection matrices to evaluate recovery case-specific transient dynamics must lie.
strategies for Snake River spring and summer Chinook salmon. Conserv. Biol., 17, Transient period: The period of time in which the model exhibits
782–794.
transient dynamics, before it settles to stable state.
van Winkle, W., Rose, K.A. & Chambers, R.C. (1993). Individual-based approach to
fish population dynamics: an overview. T. Am. Fish. Soc., 122, 397–403.
Yearsley, J.M. (2004). Transient population dynamics and short-term sensitivity SUPPORTING INFORMATION
analysis of matrix population models. Ecol. Model., 177, 245–258.
Zúñiga-Vega, J.J., Valverde, T., Rojas-González, R.I. & Lemos-Espinal, J.A. (2007). Additional Supporting Information may be found in the online
Analysis of the population dynamics of an endangered lizard (Xenosaurus grandis) version of this article:
through the use of projection matrices. Copeia, 2007, 324–335.
Table S1 Table of published indices of transient dynamics.
Appendix S1 Database of population projection matrix models and
GLOSSARY associated data used in correlation analyses.
Amplification: Short-term increase in population density relative to Appendix S2 Detailed information on methodology and results of
asymptotic growth. correlation analyses.
Asymptotic growth: The long-term, geometric rate of population As a service to our authors and readers, this journal provides
increase or decline that the model exhibits when it reaches stable state; supporting information supplied by the authors. Such materials are
mathematically, equal to the dominant eigenvalue of the PPM, kmax. peer-reviewed and may be re-organized for online delivery, but are not
Attenuation: Short-term decrease in population density relative to copy-edited or typeset. Technical support issues arising from
asymptotic growth. supporting information (other than missing files) should be addressed
Case-specific transient dynamics: The transient dynamics resulting to the authors.
from a specified initial demographic distribution.
Initial demographic distribution: The actual ratios of life stages in the Editor, Michael Bonsall
population; mathematically, the vector used to project population Manuscript received 29 April 2011
dynamics. Represented here using n0, where ||n0||1 (the one-norm or First decision made 2 June 2011
column sum of n0) is equal to overall population size or density. Manuscript accepted 15 June 2011

 2011 Blackwell Publishing Ltd/CNRS

View publication stats