niche construction theory: a practical guide for ecologists

8/12/2019 Niche Construction Theory: A Practical Guide for Ecologists

1/27

Niche Construction Theory: A Practical Guide for EcologistsAuthor(s): John Odling-Smee, Douglas H. Erwin, Eric P. Palkovacs, Marcus W. Feldman, KevinN. LalandSource: The Quarterly Review of Biology, Vol. 88, No. 1 (March 2013), pp. 3-28Published by: The University of Chicago PressStable URL: http://www.jstor.org/stable/10.1086/669266.

Accessed: 01/04/2013 08:08

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at.http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of

content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms

of scholarship. For more information about JSTOR, please contact [email protected].

.

The University of Chicago Pressis collaborating with JSTOR to digitize, preserve and extend access to The

Quarterly Review of Biology.

http://www.jstor.org

This content downloaded from 138.251.144.206 on Mon, 1 Apr 2013 08:08:43 AMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/action/showPublisher?publisherCode=ucpresshttp://www.jstor.org/stable/10.1086/669266?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/10.1086/669266?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=ucpress


2/27

THE QUARTERLY REVIEW

of Biology

NICHE CONSTRUCTION THEORY: A PRACTICAL GUIDEFOR ECOLOGISTS

John Odling-Smee

Mansfield College, University of Oxford

Oxford OX1 3TF United Kingdom

e-mail: [email protected]

Douglas H. Erwin

Department of Paleobiology, Smithsonian Institution

Washington, DC 20560 and Santa Fe Institute

Santa Fe, New Mexico 87501 USA


Eric P. Palkovacs

Department of Ecology and Evolutionary Biology, University of California

Santa Cruz, California 95064 USA


Marcus W. Feldman

Department of Biology, Stanford UniversityStanford, California 94305-5020 USA


Kevin N. Laland*

School of Biology, University of St. Andrews

St. Andrews, Fife KY16 9TS United Kingdom


*Corresponding Author

The Quarterly Review of Biology, March 2013, Vol. 88, No. 1

Copyright 2013 by The University of Chicago Press. All rights reserved.

0033-5770/2013/8801-0001$15.00

Volume88,No.1 March2013

3

http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp


3/27

keywordsniche construction, ecological inheritance, ecosystem engineering, eco-

evolutionary feedbacks, ecosystem ecology

abstractNiche construction theory (NCT) explicitly recognizes environmental modification by organisms (niche

construction) and their legacy over time (ecological inheritance) to be evolutionary processes in their own

right. Here we illustrate how niche construction theory provides useful conceptual tools and theoretical

insights for integrating ecosystem ecology and evolutionary theory. We begin by briefly describing NCT, and

illustrating how it differs from conventional evolutionary approaches. We then distinguish between two

aspects of niche constructionenvironment alteration and subsequent evolution in response to constructed

environmentsequating the first of these with ecosystem engineering. We describe some of the ecological

and evolutionary impacts on ecosystems of niche construction, ecosystem engineering, and ecological

inheritance, and illustrate how these processes trigger ecological and evolutionary feedbacks and leave

detectable ecological signatures that are open to investigation. Finally, we provide a practical guide to how

NCT could be deployed by ecologists and evolutionary biologists to explore eco-evolutionary dynamics. Wesuggest that, by highlighting the ecological and evolutionary ramifications of changes that organisms bring

about in ecosystems, NCT helps link ecosystem ecology to evolutionary biology, potentially leading to a deeper

understanding of how ecosystems change over time.

Introduction

ECOLOGY has long been portrayed as adiscipline separated by the different waysin which its two principal subfields, population-community and ecosystem ecology, relate to

evolution (Ehrlich 1986; ONeill et al. 1986;Jones and Lawton 1995; Likens 1995; Loreau2010; Schoener 2011). Why?

A primary concern of many ecologists isto understand how energy and matter flowthrough organisms and their environments. Incontrast, evolutionary biologists are principallyconcerned with information: that is, with theacquisition and inheritance across generationsof algorithmic information by organisms(Chaitin 1987), primarily in the form of the

heritable DNA sequences that underpin theiradaptations. Standard evolutionary theory(henceforth SET) permits evolutionary ecolo-gists to integrate population-community ecol-ogy and evolutionary biology (Whitham et al.2006; Rowntree et al. 2011), but at the price ofrestricting population-community ecologylargely to biota (ONeill et al. 1986). SET rec-ognizes abiota as sources of natural selection,but rarely considers the converse relationship,

where organisms modify abiotic componentsin environments, to be ecologically consequen-tial or evolutionarily generative. As ecosystemsnecessarily include abiota, evolutionary ecolo-gists frequently edit out abiota from commu-

nities, and focus on the interactions of pheno-types in different species (for instance, in food

webs). This means that many forms of environ-mental modification by organisms are left outof coevolutionary analyses, and partly accountsfor the prevailing division between population-community ecology and ecosystem ecology. Asa result: The disciplinary links between ecosys-tem science and evolutionary biology areamong the weakest in the biological sciences(Matthews et al. 2011:690).

The need to overcome this weakness is nowbecoming urgent, particularly with increasingrecognition that evolution and ecology canhappen at the same pace (Kingsolver et al.2001; Hairston et al. 2005; Ellner et al. 2011)and must shape each other (Palkovacs andHendry 2010). Schoener (2011) describesThe Newest Synthesis as the emerging fieldof eco-evolutionary dynamics, whose majorprecept is that both directions of effectecology to evolution and evolution to ecolo-gyare substantial (Schoener 2011:426). Aprimary goal of this new field is to elucidate theconsequences of bidirectional eco-evo interac-tions and eco-evolutionary feedbacks in eco-systems (Post and Palkovacs 2009).

The aim of this article is to illustrate howniche construction theory provides usefultheoretical insights and practical tools thatcontribute to the integration of ecosystem ec-ology and evolutionary theory. The niche-

4 Volume88THE QUARTERLY REVIEW OF BIOLOGY



4/27

construction perspective explicitly recognizesenvironmental modification by organisms(niche construction), and its legacy over time(ecological inheritance), to be evolutionaryprocesses: that is, they cause evolutionary changeby acting as sources of modified selection, as

well as of modified phenotypes (Lewontin1983; Odling-Smee et al. 2003). This stancecan be contrasted with the more tacit recogni-tion of organisms environmental impacts instandard accounts. The extension has pro-duced a body of conceptual and formal theory,known as niche construction theory (hence-forth NCT), which explores the ecological andevolutionary ramifications of niche construc-tion. NCT has begun to be used as a vehiclefor integrating ecosystem ecology and evolu-tion (Erwin 2008; Kylafis and Loreau 2008;Krakauer et al. 2009; Post and Palkovacs2009; Loreau 2010; Van Dyken and Wade2012). We begin by summarizing thesefindings.

NCT is derived from insights that werefirst introduced to evolutionary biology inthe 1980s by Richard Lewontin (1982, 1983,2000). Niche construction refers to the modi-fication of both biotic and abiotic componentsin environments via trophic interactions andthe informed (i.e., based on genetic or ac-quired information) physical work of organ-isms. It includes the metabolic, physiological,and behavioral activities of organisms, as well astheir choices. For example, many species ofanimals manufacture nests, burrows, holes,

webs, and pupal cases; algae and plants changelevels of atmospheric redox states and influ-ence energy and matter flows by modifyingnutrient cycles; fungi and bacteria decomposeorganic matter; and bacteria also fix nutrientsand excrete compounds that alter environ-ments. Niche-constructing species includeseveral categories of species recognized in theecological literature, including ecosystem engi-neers (i.e., species that modify their environ-ments via nontrophic interactions), keystonespecies (i.e., rare species with large effects oncommunities and ecosystems disproportionateto their abundance, often via predation), dom-inant species (common species with large ef-fects on communities and ecosystems, often

via competition), and foundation and facul-tative species (i.e., habitat-creating species;

Wright et al. 2002). For simplicity, we refer toall such activities as ecosystem engineering,although some of these activities are ex-cluded from more strict definitions of thisterm. Ecological inheritance refers to lega-cies of change, in both biota and abiota,bequeathed by niche-constructing organismsto subsequent populations, which modify se-lection pressures on descendant organisms(Odling-Smee et al. 2003); this can be re-garded as a second general inheritance sys-tem in evolution.

NCT has generated a body of conceptualand formal theory that explores the ramifica-tions of niche construction for evolutionary bi-ology (Odling-Smee 1988; Laland et al. 1996,1999; Odling-Smee et al. 2003; Laland andSterelny 2006; Silver and DiPaolo 2006; Leh-mann 2008; Van Dyken and Wade 2012), andfor related disciplines (Boni and Feldman2005; Laland et al. 2010; Kendal et al. 2011),including ecology (Erwin 2008; Kylafis andLoreau 2008; Krakauer et al. 2009; Post andPalkovacs 2009; Loreau 2010), and the Earthsciences (Corenblit et al. 2009, 2011). Insightsfrom mathematical evolutionary theory, sum-marized in Table 1, provide unambiguousevidence that niche construction is of consid-erable ecological and evolutionary importance.

The significance of niche constructionfor evolution is threefold. First, niche con-struction can influence spatial and tempo-ral patterns in the strength and directionof selection acting on the constructors them-selves (Odling-Smee et al. 1996, 2003). Bymodifying their own environments, and bylegating modified selection pressures totheir descendents via ecological inheri-tance, populations can influence the direc-tion and the rate of their evolution; forinstance, generating time-lagged responsesto selection (Laland et al. 1999). Second,niche construction can increase the abun-dance of individuals within species by in-creasing fecundity and/or extending thelongevity of individuals. By influencing pop-ulation structure, it may thus decrease thesignificance of drift and potentially increasethe longevity of species, independent of anydirect selection. Third, through ecologicalspillovers that occur in the process of modi-fying their own niches, organisms can also

March2013 5NCT: A PRACTICAL GUIDE FOR ECOLOGISTS



5/27

change the niches of other species in anecosystem. Where these spillovers are ef-fectively coupled to other species they canlead to coevolution. Thus, niche construc-tion has the potential to percolate throughecosystems and precipitate multiple evolu-tionary and coevolutionary events. In NCT,it is possible for one:many, many:one, andmany:manyrelationships to occur betweenniche-constructing populations and otherpopulations that coevolve as a result of theniche construction. Beavers provide a one:many example because beaver dams alterthe environments of many populations(Naiman et al. 1988), while the soil envi-ronment coconstructed by earthworms,arthropods, plants, and bacteria, amongothers, is a many:many case. By drawingattention to the many important conse-quences that flow from time-lagged effects,

byproducts, acquired characters, and collectiveactivity in ecosystems, the broader conceptual-ization offered by NCT compared to SET po-tentially allows it to make useful contributions

to ecosystem ecology. The principal differ-ences between SET and NCT are summarizedin Table 2.

In the following sections, we first distinguishbetween the environment-altering and subse-quent evolutionary aspects of niche construc-tion; we then consider how these effects canbe detected by ecologists and paleoecologistsand used to comprehend eco-evolutionarydynamics.

Niche Construction Theory andEcosystem-Level Ecology

Post and Palkovacs (2009) distinguish be-tween two aspects of niche construction:

TABLE 1Twelve insights from niche construction theory

Finding References

Niche construction can:1. Fix genes or phenotypes that would, under standard evolutionary theory,

be deleterious; support stable polymorphisms where none are expected

and eliminate polymorphisms that without niche construction would be

stable.

Laland et al. 1996, 1999, 2001; Kerr et al.

1999; Creanza et al. 2012

2. Affect evolutionary rates, both speeding up and slowing down responses

to selection under different conditions.

Laland et al. 1996, 1999, 2001; Silver and

Di Paolo 2006

3. Cause evolutionary time lags, generate momentum, inertia, and

autocatalytic effects. Interactions with evolving environments can produce

catastrophic responses to selection, as well as cyclical dynamics.

Laland et al. 1996, 1999, 2001; Kerr et al.

1999

4. Drive niche-constructing traits to fixation by creating statistical

associations with recipient traits.

Silver and Di Paolo 2006; Rendell et al.

2011

5. Influence the dynamics, competition, and diversity of meta-populations. Hui et al. 2004; Borenstein et al. 20066. Be favored, even when currently costly, because of the benefits that will

accrue to distant descendants.

Lehmann 2007, 2008

7. Allow the persistence of organisms in currently inhospitable

environmental conditions that would otherwise lead to their extinction;

facilitate range expansion.

Kylafis and Loreau 2008

8. Regulate environmental states, keeping essential parameters within

tolerable ranges.

Laland et al. 1996, 1999; Kylafis and

Loreau 2008

9. Facilitate the evolution of cooperative behavior. Lehmann 2007, 2008; Van Dyken and

Wade 2012

10. Drive coevolutionary events, both exacerbate and ameliorate

competition, and affect the likelihood of coexistence.

Krakauer et al. 2009; Kylafis and Loreau

2011

11. Affect carrying capacities, species diversity and robustness, and

macroevolutionary trends.

Krakauer et al. 2009

12. Affect long-term fitness (not just the number of offspring or grand-

offspring) by contributing to the long-term legacy of alleles, genotypes,

or phenotypes within a population.

McNamara and Houston 2006; Lehmann

2007; Palmer and Feldman 2012




6/27

environment alteration and the subsequentevolution of population(s) in an ecosystem inresponse to natural selection pressures previ-ously altered by the niche constructor(s).

Although environmental alteration may com-monly lead to subsequent evolution, some actsof niche construction are dissipated, swamped,or counteracted by other processes, such thattheir ecological and/or evolutionary ramifica-tions are trivial and, in principle, ecologicallyand evolutionarily negligible cases of nicheconstruction need not coincide (Post andPalkovacs 2009). This means that the eco-logical consequences of niche constructionmay sometimes usefully be studied withoutconsidering evolutionary ramifications and

vice versa.

the environment-altering aspect ofniche construction

The environment-altering aspect of nicheconstruction is similar toecosystem engineeringin ecology (Jones et al. 1994, 1997; Wrightand Jones 2006; Cuddington et al. 2007; see

Glossary). Jones et al. (1997) describe allo-genic engineers that modify environments bymechanically changing materials from oneform to another (e.g., nest-building wasps)andautogenic engineers that modify environ-ments by changing themselves (e.g., treesthat provide living space for insects, birds, andmammals, or windbreaks). Researchers haveproposed various classes of ecosystem engi-neering, including structural engineers, bio-

turbators and bioconsolidators, chemicalengineers, light engineers, and wind attenua-tors (Berke 2010; Jones et al. 2010). Althoughto some ecologists the term ecosystem engi-neering is restricted to nontrophic effects, it isimportant to recognize that important envi-ronmental alterations can also occur throughtrophic interactions, which is why we considerthem here.

Jones et al. (1994, 1997) and Cuddington etal. (2007) drew attention to the paucity of eco-logical research dedicated to studying organ-isms that modulate the availability of resourcesand habitats in ecosystems. Many species influ-ence energy flows, mass flows, and trophic

TABLE 2Comparing standard evolutionary theory and niche construction theory

Standard Evolutionary Theory Niche Construction Theory

Focus:Organismic evolution in response to environments. Focus:The coevolution of organisms and environments.Causation: Primarily unidirectional, with autonomous selective

environments shaping organisms. Reciprocal causation is

recognized in some special cases where the source of

selection is biotic (e.g., sexual selection, predator-prey

coevolution).

Causation: Primarily reciprocal, with selective

environments shaping organisms, and organisms

shaping selective environments, either relative to

themselves or other organisms.

Niche construction:Organisms acknowledged to change

environmental states, but this is treated as the product of

natural selection and rarely as an evolutionary process in

its own right. Focus is restricted to adaptations expressed

outside the bodies of the organisms (e.g., extended

phenotypes).

Niche construction:Treated as an evolutionary process in

its own right. Focus is not exclusively on adaptations,

but includes changes in environments caused by the

byproducts of organisms (e.g., detritus), acquired

characters (e.g., learned), or the collective

metabolism or behaviors of multiple individuals/

species.Inheritance: Primarily genetic, although maternal, epigenetic,

cytoplasmic, and cultural inheritances recognized as

special cases.

Inheritance:Genetic and ecological inheritance (i.e.,

legacies of selection pressures previously modified by

niche construction). Genetic and ecological

inheritance interact to form niche inheritance.

Maternal, epigenetic, cytoplasmic, and cultural

inheritances can be examples.

Organism-environment complementarity (adaptation):The

product of natural selection.

Organism-environment complementarity (adaptation):The

match between organism and environment results

from dynamic interactions between niche

construction and natural selection.




7/27

Glossary

Algorithmic information:Structural and functional know-how carried by organismstypically, but not exclusively, in their genomes (Chaitin 1987).

Byproducts:Phenotypic effects that evolve as a consequence of selection on some othercharacter, rather than being an adaptation directly favored by selection for its current role.

Eco-evolutionary dynamics:Interactions between ecology and evolution occurring onoverlapping time scales (Pelletier et al. 2009). The recognition that evolution canhappen at the pace of ecology means that ecological changes can directly shapeevolution and vice versa.

Ecological inheritance:The inheritance, via an external environment, of one or morenatural selection pressures previously modified by niche-constructing organisms

(Odling-Smee et al. 2003). Ecological inheritance typically depends on organismsbequeathing altered selective environments to their descendants, but other organ-isms, including unrelated conspecifics and members of other species that share thesame ecosystem may also be affected by this legacy. Where an act of niche con-struction leads to a change in the species composition of the local ecologicalcommunity, this too is regarded as an aspect of the ecological inheritance.

Ecosystem engineering:The creation, destruction, or modification of habitats and/ormodulation of the availability of resources to other species by organisms (Jones etal. 1994). Ecosystem engineering can be equated with the environment-alteringcomponent of niche construction, although some definitions of ecosystem engi-

neering are more restrictive.EMGAs (environmentally mediated genotypic associations):Indirect but specific connec-tions between distinct genotypes mediated either by biotic or abiotic environmentalcomponents in the external environment (Odling-Smee et al. 2003). EMGAs mayeither associate different genes in a single population or they may associate differ-ent genes in different populations.

Engineering webs: Webs of connectance in ecosystems, caused by species influencingenergy and mass flows, and creating habitat and other resources for other species (Jones etal. 1994, 1997). Engineering webs contribute to the stability and dynamics of ecosystems,alongside webs of trophic interactions. Ecosystem engineers may influence and control

energy and matter flows without themselves being part of those flows.Extended phenotypes:Phenotypic characters expressed outside the body of the or-ganism (Dawkins 1982). Although byproducts are sometimes characterized as ex-tended phenotypes, Dawkins is explicit in stating that this is not profitable, andrestricts use of the term to adaptations. Thus, extended phenotypes correspond tothat subset of niche-constructing activities that are biological adaptations.

Niche construction:The process whereby organisms, through their metabolism, theiractivities, and their choices, modify their own and/or each others niches (Odling-Smee et al. 2003). Niche construction may result in changes in one or more naturalselection pressures in the external environment of populations. Niche-constructing

organisms may either alter the natural selection pressures of their own population, ofother populations, or of both. Niche construction may occur through physical pertur-bation of the environment or through relocation to a new environment. Niche con-struction may have both positive and negative effects on the constructors fitness.




8/27

patterns in ecosystems nontrophically by gen-erating engineering webs based on mosaics

of connectivity among different species. Al-though the ecological consequences ofniche construction should be the same asthose of ecosystem engineering, NCT fo-cuses primarily on those environment-altering activities of organisms that modifynatural selection in ways that affect the evo-lution of population(s) (e.g., Loreau 2010).

As a discipline, evolutionary biology has fo-cused overwhelmingly on the passage of ge-netic information via DNA, although there isincreasing recognition that a broader notionof heredity may be required (Bondurianskyand Day 2009; Danchin et al. 2011; Bonduri-ansky 2012). In contrast, ecosystem ecologistsare concerned with the flux and flow of energyand materials. If researchers are to bridgethese disciplines, they require a framework ca-pable of reconciling these differences. Withthis in mind, we define an ecological variable,R, representing any resource or condition thatis both a potential source of selection for arecipient population, and is modifiable byat least one niche-constructing population(Laland et al. 1996, 1999). An increasein R canhave either positive (e.g., R prey) or negative(e.g., R predator) consequences for the fit-ness of organisms in the recipient population.Here we differentiate between three broadcategories of resource/condition that havedifferent properties, and respond to niche con-struction in different ways (Figure 1). R couldrepresent an abiotic component of the envi-ronment (e.g., water, sediment), or a bioticcomponent (e.g., another organism), or an ar-tifact or other construct built by an organism(e.g., a spiders web).

We further subdivide R into physical

energy and matter resources, henceforthlabeled Rp, and semantic informational re-

sourcese.g., know-how or algorithmicinformation, such as is encoded by genes,or is acquired through learning (Chaitin1987), labeled Ri (Figure 1). Abiota areusually physical energy and matter re-sources only (exclusively Rp), while biotaalways carry both physical Rpand algorith-mic informational Ri resources. All organ-isms carry Ri (i.e., genetic information) intheir genomes in addition to the Rp intheir bodies, while animals also carry Ri

(i.e., acquired knowledge) in their brains.Artifacts primarily carry Rp, but it is possiblefor them to carry Ri too as, for instance, inhuman artifacts such as computers or books.These distinctions are significant since the dif-ferences between different kinds of R poten-tially lead to different ecological responses toniche construction.

signatures of niche construction/ecosystem engineering

Let us assume that R is modified byniche construction, such that R becomesR R, where R could be positive ornegative. Then R is a potential ecologi-cal signature of niche construction (eco-system engineering), and offers a useful

way to measure the ecological changecaused by organisms. Detection and quan-tification of the impact of niche construc-tion on an ecosystem therefore requiresthat R must be differentiated from allother potential sources of change in R, in-cluding those resulting from abiotic (physi-cal or chemical) processes or from otherbiota not of focal interest. This is the usual

Niche inheritance:Combinations of genetic inheritance and ecological inheritancethat descendant organisms inherit from ancestral organisms.

Time lags:Delayed evolutionary response of characters exposed to selection pres-sures modified by niche construction. These time lags are generated by ecologicalinheritance that affects availability of a resource for multiple descendant genera-tions, and can be orders of magnitude larger than the number of generations ofniche constructing affecting that resource. Time lags may generate unusual evolu-tionary dynamics, including momentum, inertia, and autocatalytic effects, as well asopposite and catastrophic responses to selection.




9/27

strategy of researchers who study NCT, eco-system engineering, and eco-evolutionaryfeedbacks (Jones et al. 1994, 1997; Odling-Smee et al. 2003; Cuddington et al. 2007;Post and Palkovacs 2009), with a variety ofobservational, correlational, and experimen-tal methods (for details see Jones and Lawton1995; Odling-Smee et al. 2003; Hairston et al.2005; Ellner et al. 2011). These include exper-imental manipulation of the presence,abundance, or potency of niche construction,common gardening experiments, selection ex-periments and experimental evolution, and of-ten involve breaking a complex system downinto component parts, which involve single-species or species-pair effects.

As the niche-constructing activities ofmany animals are shaped by the knowledge

that they carry in their genomes or brains,the R ecological signatures left by niche con-

struction should reflect this information to theextent that an ecological signature of nicheconstruction might stand out relative tochanges in R resulting from other external pro-cesses. Below we spell out how abiota, biota,and artifacts potentially offer different signa-tures of prior niche construction.

Abiota

As abiota carry only physical resources(Rp), niche-constructing organisms are

only able to change abiota physically orchemically. However, there are often clearsignatures of prior niche construction(Rp) left in abiota; for example, in sedi-

Figure 1. Types of Resource Subject to Niche Construction

$ If the Ri in other organisms can be modified by the separate Ri in niche-constructing organisms (e.g.,genetic manipulation by parasites), then the Ri in other organisms becomes an ecological resource for theniche constructors. Some organisms superimpose signals, such as scent marks, on abiota. If they do that,abiota may carry Ri as well as Rp. However, in that case, abiota are probably better described as Ri-carryingquasi-artifacts rather than as abiota.




10/27

ments, soils, oceans, and rocks (e.g., Erwin2008). Unlike other causes of change in abiota,changes due to niche-construction must bedue to the information (Ri) in those organ-isms. Abiotic resources that have been affectedby multiple generations of niche constructionmay have been driven far from thermody-namic equilibrium by the prior nonrandom

work of organisms, and may therefore occupystates that could not occur, or would be highlyunlikely to occur, on a dead planet. Exam-ples include soil states changed by niche-constructing invertebrates, ocean stateschanged by Ediacaran organisms, and sedi-ment or rock states changed by burrowing or-ganisms (Odling-Smee et al. 2003; Erwin 2008;Erwin and Tweedt 2012). Even when it is dueto byproducts, for instance, by their metabo-lism or their detritus, any alteration to an abi-otic ecosystem component caused by nicheconstruction will ultimately still be shaped bygenetic information. It may therefore be non-random and should frequently stand out rela-tive to any change caused by nonliving agentsby virtue of its improbability (for instance, ele-

vated rates of nutrient flux or improbablechemical composition). Such signatures may

well be produced by multiple individuals, per-haps emanating from multiple species, andpossibly detectable over periods of time sub-stantially longer than the lifetimes of the con-structors. The temporal and spatial patternsassociated with the signature often covary withthe density and duration of the constructorpopulation(s) and the magnitude of their in-dividual effects (Jones et al. 1994, 1997), andmay be characterized by long-term trends in

accumulation, depletion, or transformation ofresources over time. Such patterns are alsolikely to represent an extremely broad range ofscales.

Biota

The biotic components of ecosystems dif-fer from abiota in potentially being able torespond to prior niche construction by recip-rocating with further niche construction oftheir own. Niche-constructing organisms canchange biota in two ways: either by modify-ing the physical (Rp) or informational (Ri)states of other organisms (Figure 1). Niche-constructing organisms modify the physical

states of other organisms in ways long-studied by ecologists, for example, by provi-sioning or protecting other organisms (e.g.,offspring), supplying them with nutrients(e.g., mutualism), eating them (e.g., prey),or competing with them (e.g., for light). Atthe population level, a niche-constructingpopulation should therefore leave an ecolog-ical signature of its prior activities in biota(Rp). It might modify the demographics ofanother population (e.g., the impact of gup-pies on algal growth), or its population struc-ture (e.g., the impact of guppy predators onguppy size distributions), or its distributionacross an environment (e.g., the impact of ale-

wife populations on zooplankton communitiesin different lakes), or one of its engineeringfunctions in an ecosystem (Post and Palkovacs2009).

An ecological signature in a recipient popu-lation could also show signs of feedback to theoriginal niche-constructing population in theform of a modified biotic resource, generatinganother (Rp and probablyRi) ecological sig-nature in the original niche constructors. Thiscould result in demographic signatures amongpopulations in ecosystems, potentially medi-ated by changes in intermediate abiotic ecosys-tem components, even in the absence of anygenetic coevolution. Such interactions contrib-ute to many of the phenomena that areroutinely studied by ecologists (for exam-ple, trophic relations, community assem-blies, energy/matter flows in ecosystems, andengineering webs). NCT predicts that thesephenomena should depend on the niche-constructing activities of organisms far moreoften than has previously been recognized (ex-cept by those ecologists who study ecosystemengineering); it is therefore important thatniche-constructing effects be quantified.

Niche-constructing organisms may alsochange the information (Ri) carried by otherorganisms. For instance, horizontal gene trans-fer can occur in bacteria while a parasite, suchasa virus or aninsect, mayaffectthe genome ofits host by inserting a hostile message into itshosts genome. In animals, including inverte-brates, useful acquired knowledge is oftentransmitted through social learning from con-specifics and heterospecifics (Galef and La-land 2005; Leadbeater and Chittka 2007).




11/27

Flack et al. (2006) describe how aggressive po-licing by dominant pigtail macaques creates asocial niche, stabilizing and integrating ma-caque societies. Similarly, Oh and Badyaev(2010) describe how male house finches ac-tively seek or create social environments withproperties that enhance their attractiveness orconspicuousness to females (see also Cornwal-lis and Uller 2010). One of the strongest Riimpacts of human niche construction is theselective modification of other species, bothintentional (e.g., domestication) and uninten-tional (e.g., evolution in response to humanactivities such as urbanization; Palkovacs et al.2012). In such cases, transmitting organismsleave a Risignature in the receiving organ-isms.

Artifacts

Niche-constructing organisms may alsomodify selection by building artifacts, includ-ing nests, burrows, webs, mounds, and dams,all the way up to the houses, cars, factories, andcomputers of contemporary humans. We in-clude in this category constructed resources

such as crop fields, hedgerows, terraces, canals,artificial lakes, forest clearings, and so forth(Kendal et al. 2011). Artifacts are usually phys-ical abiotic resources, but can be biotic re-sources (e.g., crop fields, ant gardens) and areoften underpinned by information stored inthe genome or brains of the constructor, asthough they were intelligently designed.

Although the most familiar artifacts (nests,mounds) are extended phenotypes (Dawk-ins 1982), this concept is of limited utility here

as it is restricted to niche-constructingadapta-tions, yet artifacts can sometimes be built by thebyproductsof organisms. Examples include thedead coral substrate and sand of coralreefs, which are constructed by interac-tions among multiple species over long pe-riods of time, and the paths and routescreated by habitual animals. Moreover,particularly in the case of those manufac-tured by humans, artifacts built throughthe expression of acquired knowledge may

still be informed and structured, even ifthey are not themselves biological adapta-tions. For instance, human buildings are notbiological adaptations and, accordingly, a sim-

ple adaptation-byproduct dichotomy wouldimply that they are byproducts. However, hu-mans and other animalspossess some very gen-eral, knowledge-gaining adaptations, such asthe ability to learn, which allow them to ac-quire and store information in their brains, tocommunicate this information to others, andin humans to build on that shared knowledgebase cumulatively (for instance, through tech-nological advance). This means that some ar-tifacts may possess a designed property in spiteof the fact that they are not adaptations. Thisproperty should shape the (Rp) ecologicalsignatures found in artifacts, and potentially

facilitate their recognition. Finally, artifactsmay sometimes include (Ri) signs or signalssuch as the advertising signals in a bower-birds bower (Madden et al. 2012).

correlates of signatures of nicheconstruction

If a new niche-constructing activity evolves, itshould start to generate a signature immedi-ately, although probably only locally at first.Conversely, if a niche-constructing activity

changes, or if the niche-constructing pop-ulation goes extinct, then the signature itpreviously generated should dissipate fromthat time on. However, niche-constructedeffects might be expected to accumulateover time, before they become detectable,either because the per-capita impact of a singleact of niche construction is very small or be-cause low-impact effects are dissipated, coun-teracted, or swamped by other processes.However, in searching for the covariation be-

tween putative niche-constructing acts andputative ecological effects, we note that thisrelationship may be time-lagged (Laland et al.1996, 1999; Jones et al. 2010). Analyses of suchtime lags have been based on genetic mod-els, but similar time lags occur in demo-graphic and ecological models (Ihara andFeldman 2004; Jones et al. 2010), partly be-cause structure formation and decay and bioticresponses to structural and abiotic change arerarely instantaneous (Jones et al. 2010),although the dynamics of gene-frequencychange may also contribute to time lags(Laland et al. 1996).




12/27

Abiota

It should be possible to detect associa-tions between specific niche-constructing

populations and specific ecological signa-tures in abiota. It may even also be possibleto detect traces of such associations in re-mains of extinct populations, the ghostsof ancient niche constructors, still capableof affecting contemporary ecological andevolutionary processes (Odling-Smee et al.2003). For instance, bog-forming Sphag-nummosses produce peat that can persistfor hundreds or thousands of years after thedeath of the living moss (van Breemen 1995),

while the plants of the Carboniferousproduced the fossil fuels that with hu-man help are constructing the modernclimate of Earth. These kinds of tracesare particularly likely to be found in abi-ota, which commonly persist for a longertime than the populations that gave rise tothem. Even if the information (Ri) expressedby an extinct niche-constructing populationis now gone, it may still be possible to retrievea proxy for it from a functionally equivalent

contemporary population, thereby allowingits original biological functionality to be in-ferred. For example, the individual organ-isms responsible for the oxygenation of theatmosphere are extinct, but many contem-porary organisms exhibit photosynthesis, sug-gesting parallel characteristics in the originaloxygenators. Features related to theprevalenceor density of extinct or ancestral organisms, thesignatures of their niche-constructing activity,may still be recoverable in contemporary

oceans, sediments, or in the atmosphere (forexamples see Erwin 2008; Corenblit et al.2011).

Biota

Ecological associations between niche-constructing and recipient biota typically in-

volve ongoing interactions between currentlyliving organisms, corresponding to familiarecological relationships, such as predator-prey,host-parasite, mutualisms, or competition for ashared resource. However, NCT draws atten-tion to the possibility that such interactionsmay also be mediated by any number of abioticor biotic intermediate ecosystem components.

Moreover, the density of a recipient popula-tion may covary with the density of aconstructor population ntime steps ago,if their interaction occurs via a resource,manufactured by the constructor populationthat takes time to accumulate, or is graduallydepleted through niche construction (La-land et al. 1999). Indirect associations increasethe likelihood that the effects are time-lagged.

Although these observations might make spe-cies associations appear complicated, in prac-tice, it need not make the job of the ecologistor evolutionary biologist any more difficult,and may often make it easier since established

procedures and methods can address theproblem and render systems more predictable(Jones and Lawton 1995; Odling-Smee et al.2003). The key to progress is to break downcomplicated pathways into tractable compo-nent pieces (e.g., Post and Palkovacs 2009;see the section entitled Methods of Imple-mentation).

Ecological associations should typically oc-cur in the presence of the species expressingthe genetic or acquired knowledge responsi-ble for the niche construction, althoughthis association may be time-lagged and mayfade or disappear when the relevant niche-constructing population disappears. It followsthat ecological signatures of the novel appear-ance, the ongoing presence, or the subsequentdisappearance of niche constructors may af-fect inference about the structure, func-tion, and regulation of ecosystems. HereNCT can offer some useful predictionsfor

instance, concerning when invasions occurand how they should change ecosystem func-tioning when they do occur (e.g., see Odling-Smee et al. 2003; Loreau 2010).

Reciprocal acts of niche constructionby recipient biota may be repeated manytimes, so the knock-on consequences of asingle type of niche construction couldcascade through many compartments ofan ecosystem. Such causal chains, or cas-cades, resulting from niche construction

may include evolutionary as well as ecolog-ical responses and trigger eco-evolutionarydynamics (Pelletier et al. 2009; Post andPalkovacs 2009; Loreau 2010).




13/27

Artifacts

Ecological signatures left in artifacts byniche construction may be very different

from those left in other kinds of abiota,since they often comprise the entire arti-fact and may possess a designed quality.This should be reflected in particularlystrong associations between the expressedinformation (Ri) and the signature (Rp),particularly in extended phenotypes such asnests, webs, and mounds. They should alsoclearly demonstrate their roles in contributingto the fitness of their constructors. Artifacts arealso often highly organized, one consequence

of which is that they may spontaneously dissi-pate or be deconstructed by other organisms,leaving only transitory ecological signaturesunless they are constantly repaired and/or re-constructed. Moreover, since artifacts are oftendifficult and costly to construct and maintain,they, or their component parts, will frequentlybe valuable to other organisms and may re-quire defending. Artifacts can also feed backto the constructor population to modifythe latters developmental environment. A

good example is provided by Leca et al.(2010) who describe how piles of stonetools left by macaques shape the learningof tool-using traditions in these monkeys.

The Evolutionary Consequences ofPrior Niche Construction

We now turn to the second major aspectof niche construction, namely, the subse-quent evolution of populations due to naturalselection modified by organisms. Note the

niche-constructing population that causes theecological change does not necessarily have tobe the same as the recipient population(s)

whose subsequent evolution is affected by theselection that has been modified by that par-ticular ecological change (Odling-Smee et al.2003; Post and Palkovacs 2009). This meansthat niche construction can potentially triggerboth diffuse coevolution (Strauss and Irwin2004), and coupled coevolutionary dynamics.Below, we consider what it takes to convert an(R) ecological signature into an evolu-tionarily significant modified natural selec-tion pressure for one or more populations,and then review the principal alternative

eco-evolutionary paths in ecosystems de-scribed by NCT.

from ecological signatures to

modified selection

To affect evolution, any change in anyecological variable caused by niche con-struction (R) must translate into at leastone modified natural selection pressurefor at least one population in an ecosystem.It cannot do that unless the ecological sig-nature caused by the niche constructionbecomes a component in the ecologicalinheritance of at least one population.

There would be no evolutionary conse-quences if the ecological changes causedby NCT dissipated too rapidly, or if they

were swamped by other processes or byother agents, or if there were a constrainton the rate of evolution such as the ab-sence of genetic variation. In such cases,niche construction would have only eco-logical consequences. Niche constructionmay have evolutionary consequences thatare not easily detectable. For example, theevolutionary consequences of counterac-tive niche construction may be stabilizing,and they may be difficult to detect until theconstructor disappears from an ecosys-tem.

In practice, the criteria necessary for anecological signature of niche constructionto be evolutionarily consequential are eas-ily satisfied. Jones et al. (1994, 1997) de-scribe the principal ecological factors thatscale up the impact of engineering spe-cies in ecosystems: (1) lifetime per capitaactivity of individual engineering organ-isms; (2) population density; (3) the localand regional spatial distribution of thepopulation; (4) the length of time the pop-ulation has been at a site; (5) the type andformation of the constructs, artifacts, orimpacts, and their durability in the ab-sence of the engineers; and (6) the num-ber and types of resources that are directlyor indirectly controlled, the ways these re-sources are controlled, and the number ofother organisms that depend on these re-sources (Jones et al. 1997:1952). All ofthese ecological factors should also makeany (R) change caused by the niche con-




14/27

struction more likely to translate into anevolutionarily significant, modified naturalselection pressure for at least one popula-tion in an ecosystem. The higher the valueof each of these six variables, the more ecolog-ically potent the niche construction should beand the more likely it is that this will subse-quently translate into evolutionarily significantecological inheritance for at least one popula-tion in an ecosystem. It is not necessary for allsix of the above variables to be high simultane-ously. Some organisms, such as beavers, arepotent niche constructors because they pro-duce large, durable effects (Moore 2006).Others, such as photosynthesizing cyanobac-teria whose individual impacts on their en-

vironments are tiny, can have a vast impactif they are present in sufficiently high den-sities over sufficiently large areas for suffi-cient time and may affect vast numbers ofspecies (Odling-Smee 2010).

An additional factor that facilitates the con-version of an ecological signature of niche con-struction into a modified natural selectionpressure is genetic inheritance. Insofar as mostorganisms in a niche-constructing populationshare the same genes, and insofar as thesegenes predispose them to niche construct inthe same way, generation after generation,the collective consequences of their nicheconstruction are likely to accumulate overtime to the point where the (R) ecologi-cal signatures they generate do becomemodified natural selection pressures onthe population, which may cause furtherevolution of that or other populations inan ecosystem.

alternative eco-evolutionaryfeedback paths

Figure 2 describes the principal alter-native eco-evolutionary feedback pathsconnecting the environment-altering andselection-pressure-modifying aspects of nicheconstruction.

The simplest path (1) corresponds tothat emphasized in Dawkinss extendedphenotype. A genotype in a population ex-presses an environment-altering adaptation,for instance, the houses built by caddisfly lar-

vae (Dawkins 1982). This extended phenotypethen modifies a natural selection pressure that

feeds back exclusively to that genotype in thebearer responsible for building the extendedphenotype and subsequently affects its fitness.Path (1) is described by SET and NCT in thesame way. Path (2) is the path modeled in theearliest niche construction models (Lalandet al. 1996, 1999) as well as some other mod-eling frameworks (Bailey 2012). Genotype(s)in a niche-constructing population expressan environment-altering trait, which modi-fies an ecological variable. The (R) signa-ture of change becomes a modified naturalselection pressure for the same population.Unlike path (1), however, the modified selec-tion in this case doesnot solely feed back to theallele(s) or genotype(s) responsible for theenvironment-altering trait(s); rather it (also)affects the fitness of different allele(s) or geno-type(s) in the same population. For example,the beaver dam modifies selection on genesexpressed in beaver browsing or exposureto parasites. Path (2) can be supported byenvironment-altering byproducts as well asadaptations.

Path (3) is similar to path (2), exceptthat the organisms affected by modifiednatural selection are no longer in the samepopulation as the organisms responsiblefor modifying the natural selection. Forinstance, beaver dam building and brows-ing modify selection acting on genes ex-pressed in the production of defensivechemicals byPopulustrees (Martinsen et al.1998). Note that path (3) can also repre-sent environment-altering traits that are ei-ther adaptations (e.g., dam building) orbyproducts (e.g., excretion). This diffusecoevolution is more accurately described asfeed-forward than feedback, and may bedirect or indirect. The latter would happen ifan environment-altering population causes achange in the availability of an abiotic re-source, such as water or light, that indirectlyaffects a different recipient population. Path(3) might also work via intermediate biota thatcould serve as no more than a catalyst for asubsequent evolutionary change in a recipientpopulation. For instance, an intermedi-ate population might respond to a niche-constructing population with a demographicchange, without itself evolving further. Thismight, nevertheless, translate into a modified




15/27

natural selection pressure for a third popula-tion that subsequently evolves further. Post andPalkovacs (2009) give contemporary examples,

while Erwin (2008) discusses several cases ofancient niche construction, including sedi-ment perturbation and the persistence of shellbeds in marine ecosystems throughout thePhanerozoic, as well as their possible macroev-olutionary consequences in deep time extend-ing as far back as the Cambrian. Erwin andTweedt (2012) discuss the role of such ecolog-

ical spillovers in influencing redox gradientsand other aspects of the Cambrian Explosion.

Path (4) resembles path (3) but with theadditional property that it promotes thecoevolution of two or more populations viaan additional feedback path or paths. If arecipient population responds to modifiedselection induced by niche construction,by itself becoming an environment-changingniche-constructing population that subse-quently modifies selection on the original

Figure 2. Eco-Evolutionary Feedback Paths

Alternative eco-evolutionary feedback paths that connect the environment-altering and selection pressure-modifying aspects of niche construction. Here shaded circles represent biota, white squares represent abiotic

compartments in ecosystems, and shaded sections in white squares represent the

R signature of nicheconstruction that has modified the state of the abiota. The grey lines represent chromosomes, with the smallwhite sections specific genes. Continuous arrows represent niche construction, and dashed lines naturalselection. Path (1): A genotype in a population expresses an environment-altering trait that builds an extendedphenotype, which modifies a natural selection pressure that feeds back exclusively to the genotype responsiblefor building the extended phenotype. Path (2): A genotype in a niche-constructing population expresses anenvironment-altering trait that modifies an ecological variable, thereby generating a modified natural selectionpressure for the same niche-constructing population, but here affecting the fitness of different genotypes fromthose responsible for the niche construction. Path (3): Similar to path (2), except that the organisms affectedby modified selection are no longer in the same population as the organisms responsible for modifyingselection. Here the niche construction of the first population drives ecological change and coevolutionaryepisodes in the second population. Path (4): Resembles path (3), but with the additional property that thecoevolution of two or more populations is facilitated via an additional feedback path. Here the secondpopulation modifies another abiotic resource through its own niche construction, which modifies selectionacting back on the original population.




16/27

niche-constructing population, then the twopopulations may coevolve. Once again, thefeedback paths that connect coevolving niche-constructing populations to each other couldbe either direct or indirect. If they are indirect(path 4), particularly if the coevolution in-

volves modified intermediate abiota such as amineral nutrient or soil pH, then the differentphysical dynamics of the abiotic change in-duced by niche construction may need to betaken into account. A possible example is pro-

vided by Goddard (2008) who shows exper-imentally how niche construction by the yeastSaccharomyces cerevisiaeallows it to outcompeteother species. This species eventually domi-nates in fruitniches where it is naturally initiallyrare, by modifying the environment throughfermentation (the Crabtree effect) in ways thatextend beyond ethanol production. NCT alsodescribes more complicated patterns of con-nection between environmental alteration andselection-pressure modification. There can beone:many,many:one, ormany:manyrelationshipsbetween environment-altering populationsand subsequently evolving recipient popula-tions via paths 3 and 4. All of these more com-plicated consequences of niche constructionmay also occur on multiple different scales inspace and time (Loreau 2010).

the role of byproducts

A primary benefit of our broad concep-tualization of niche construction is that ithighlights the important roles that niche-constructing byproducts may play in eco-systems. This is significant because suchroles are often unintuitive and easy to dis-miss. It is far more apparent that the bea-

ver dam may drive coevolutionary episodesthan beaver dung may, yet the latter is a

very real possibility. Numerous exampleshave been documented of seemingly in-consequential and inadvertent acts by or-ganisms whose aggregate activity generatesan important ecological signature. For ex-ample, consider Euchondrus snails whoseconsumption of endolithic lichens inadver-tently generates tonnes of soil, therebyplaying a vital role in fertilizing desert eco-systems (Jones and Shachak 1990).

Typically, evolutionary biologists assumethat if a niche-constructing activity gener-

ates evolutionary feedback to the construc-tor, then it must be an adaptation. In fact,recent theory suggests that this need not be thecase and niche-constructed byproducts can beconsequential for the constructors evolution.This occurs when byproducts precipitate boutsof selection in their own population by induc-ing selection on other traits in the same popu-lation and hitchhiking to fixation on the backof this selection they generate (Silver and DiPaolo 2006; Rendell et al. 2011). In this in-stance, spatial structure (local dispersal andmating) may give rise to statistical associa-tions between niche-constructing traits andany genotypes favored in the constructedenvironments. Drawing on an importantdistinction made by philosopher Elliot So-ber (1984), here there is selection of theniche-constructing trait, but notselection forit, and only the latter meets the defini-tion of an adaptation (Williams 1966; So-ber 1984). Nonetheless, in such hitchhikingcases, there remains evolutionarily conse-quential feedback to a niche-constructingpopulation stemming directly from its niche-constructing activities.

the role of acquired characters

All of the aforementioned paths throughwhich niche construction acts to trigger evolu-tionary episodes remain relevant even if theniche construction is the expression of learnedknowledge. This point is of particular rele-

vance for understanding human anthropo-genic change within ecosystems; clearly it isacquired knowledge that underpins urbaniza-tion, deforestation, agricultural practices, andthe majority of major human impacts on theenvironment. Such processes undoubtedlyprecipitate evolutionary episodes in humansand other species. It would be unwise forresearchers to assume that human-modified se-lective environments can be treated as equiva-lent to such independent sources of selectionas arise from geological or climatic changesince there may be selective feedback to theconstructing population of a form that influ-ences its constructing behaviorfor instance, adairy-farming niche created the conditionsthat favored the spread of alleles facilitatingadult lactase persistence, but the consumptionof dairy products is more likely in individuals




17/27

with the lactose tolerant genotype (Gerbaultet al. 2011). Similarly, deforestation for cropplanting has apparently inadvertently pro-moted insect-propagated diseases (such asmalaria) in many human populations, gener-ating selection for resistant alleles (Durham1991; Laland et al. 2010) and triggering fur-ther human niche construction, such as the

widespread use of chemical insecticides.Hence, anthropogenic change leads tothe type of eco-evolutionary dynamics recog-nized by ecologists, but underpinned by cul-tural learning.

spatial structure

The significance of niche constructionfor the evolution of recipient populations isfurther enhanced by spatial structure. Spatiallyexplicit niche-constructing models have re-

vealed hitherto unrecognized forms of dynam-ical feedback in ecosystems. For instance,niche-constructing traits, even if costly, candrive themselves to fixation through spatiallyenhanced statistical association with genotypesthat they inadvertently favor (Silver and DiPaolo 2006). Rather than passively hitchhiking,these traits may actively create the conditionsthat promote their own spread by association

with recipient traits. Such dynamics not onlyincrease the amount of niche-constructingactivity across a population, driving increasesin the magnitude of the niche construc-tion, but have also been shown, undermore restricted conditions, to trigger a sec-ondary hitchhiking process in which ge-netic variation at other loci that promotethe potency of the niche-constructing be-havior can also be favored (Rendell et al.2011). Local spatial effects are also likely tobe important in cases in which a popula-tion promotes its own range expansionthrough niche construction (Kylafis andLoreau 2008). An important aspect ofspatial structure is its ability to promotesimultaneous genetic and ecological di-

versification if the strength (or nature)of feedbacks differs from patch to patch(Habets et al. 2006). This perspective isimportant because it has the potential tolink eco-evolutionary feedbacks to the pro-cess of adaptive radiation, which is usually at-tributed to an environmental template and the

importance of preexisting empty niches(Schluter 2000; Gavrilets and Losos 2009;Tebbich et al. 2010). Although the recentliterature on adaptive radiations has em-phasized lineage diversification withoutany necessary ecological changes, trophicnovelty is a critical aspect of adaptive radi-ation (Martin and Wainwright 2011).Rather than lineages simply diversifying tofill available niches, niches themselvesmay be diversifying (Erwin 2005), a processthat Losos (2010) termed self-propagatingadaptive radiation. This process of differ-ential niche construction and subsequentevolutionary diversification has been dem-onstrated in several microbial laboratorysystems (Rainey and Travisano 1998; Ha-bets et al. 2006) and appears to underliethe divergence between landlocked andanadromous alewife populations (Palko-

vacs and Post 2008; Schielke et al. 2012).What remains to be explored both empir-ically and through theoretical studies is theextent to which niche construction can drivethe acquisition of new resources sufficientlyto enable adaptive radiation.

temporal dynamics

Temporal dynamics are likely to beequally important here, since engineeringoccurs over a broad range of temporal scales(Hastings et al. 2007) and many niche-constructing activities accumulate over time,sometimes even deep time (Erwin 2008).Formal theory suggests that time-lagged ef-fects stemming from ecological inheritancecan generate rich evolutionary dynamics(e.g., momentum effects, inertia, or cata-strophic responses; Laland et al. 1996, 1999).They also raise some fundamental questionsfor biologists, such as how fitness is measuredin natural populations of organisms. If niche-constructing organisms can influence pat-terns of selection not just on the current andoffspring generations, but for many descen-dent populations, then an accurate estimateof fitness would require consideration of thecosts and benefits of niche construction overmultiple generations (e.g., McNamara andHouston 2006; Lehmann 2008). Under suchcircumstances, it may be better to replaceconventional fitness proxies (such as repro-




18/27


19/27

The recognition of networks of EMGAshas potentially important implications for un-derstanding ecosystem dynamics. In practice,the feedback and feed-forward ramifications ofniche-constructing activities are likely to go

well beyond isolated eco-evo links and pro-duce interwoven causal chains that thread eco-systems. The immediate drivers of evolutionary

changes in recipient populations are the eco-logical changes caused by environment-altering populations, but the preceding causeof evolutionary changes in recipient popula-tions may well be the prior evolution of niche-constructing populationsand, therefore, of theniche-constructing traits that cause the ecolog-ical changes in the first place. Thus, EMGAs

Figure 3




20/27

connect the prior evolution (evo) of geno-types in niche-constructing populations to thesubsequent evolution of genotypes in recipientpopulations via the modification of either bi-otic (e.g., their phenotypes) or abiotic (e.g.,soil chemistry) intermediate ecosystem (eco)components. In other words, EMGAs are not

just eco-evo links in ecosystems, but ratherevo-eco-evo links.

This argument can be extended furtherand, in some instances, the cause of selectionfor the initial evolutionary event in an evo-eco-evo link is the prior niche construction of anearlier population, while there may be furtherdownstream ecological or evolutionary conse-quences of a given act of niche construction. Inthis manner, causal chains underpinning eco-system dynamics can potentially be traced, asillustrated by Post and Palkovacs (2009). Itis clear that such interactions will influenceand to some degree control (Jones et al.1997) ecosystem dynamics. However, it willalso be apparent that ecosystem evolu-tion need not resemble the simulated dy-namics of the same system reduced to itsbiotic components, as in classical modelecosystems (e.g., May 1973), since such mod-els omit EMGAs mediated by abiota andhence many important drivers of coevolu-tionary episodes in ecosystems. The inclusionof these additional connections could poten-tially greatly affect the stability of model eco-

systems. Thus, a full understanding of ecosys-tem dynamics will require incorporatingfood webs and engineering webs into dynam-ical systems that simultaneously considertrophic and competitive interactions along-side ecosystem engineering and niche con-struction.

While, in theory, any such evo-eco-evocausal chain might be traced back ad infi-nitum, in practice it is useful to assumethat some event initiated a particular eco-evolutionary cascade. This might start witha change in the environment, which pre-cipitates an evolutionary response in afocal population, but this is not the onlypossibility; the initiating event may best becharacterized as an act of niche construc-tion. For example, behavioral plasticity in apopulation of animal niche constructorsmay generate an innovation that propa-gates through the population via learning(for instance, birds learning to peck openmilk bottles or monkeys discovering newfood sources). In such cases, where pre-sumably there is comparatively little advan-tage to considering the prior evolution ofthe population responsible for the nicheconstruction, it makes sense to view thecausal chain as triggered by an ecologicalrather than an evolutionary episode (inthis case, a learned behavior).

This is particularly apparent in the case

Figure 3. Feed-forward and Feedback Processes in Ecosystems

(a) Two populations connected indirectly by an EMGA. A genotype in a worm population expresses anenvironment-altering phenotype that modifies the soil. The (R) ecological change in the soil modifies a

selection pressure for a plant population. The recipient plants respond with genotypic evolutionary changes.The biota, worms, and plants are symbolized by shaded circles, and the intermediate abiotic environmentalcomponent, the soil, is symbolized by the square in the middle. The square is partitioned into shaded and whitecomponents, the former representing the (R) ecological change in the soil caused by the earthworms nicheconstruction, while the latter represents the extent to which the soils state is due to other agents. In principle,these components are measurable. (b) This figure symbolizes the guppy system described by Palkovacs et al.(2009). Differential predation in (i) and (ii) by two predator populations (the first shaded circle) lead todifferent size distributions in two guppy populations (the second circle), leading to differences in rates ofexcretion in the two populations (niche construction) leaving different (R) signatures (shaded section ofwhite square) affecting nitrogen cycling in the local environments (white square), affecting algal growth (thethird circle), and feeding back to affect selection on the guppy populations. (c) An EMGA based eco-evolutionary network in a miniature model of an ecosystem. (d) The same ecosystem reduced to a food web,

in a conventional population-community approach. Again, shaded circles represent biota, white squaresrepresent abiotic compartments in ecosystems, and shaded sections in white squares represents the Rsignature of niche construction that has modified the state of the abiota. Ecosystem dynamics may be affectedby engineering processes and their knock-on consequences, such that sometimes (c) cannot safely be reducedto (d).




21/27

of anthropogenic change where, for in-stance, deforestation or agricultural practicesare obviously not triggered by evolution oftree-chopping or crop-planting genes, butnonetheless can initiate eco-evolutionaryevents in an ecosystem (Laland et al. 2010).

Although we see utility in considering genenetworks operating across multiple species inecosystems (Whitham et al. 2006), such exam-ples serve to emphasize that the interplay be-tween ecological and evolutionary processescannot always be reduced to the genetic leveland, sometimes, such as where acquired char-acteristics underpin niche construction, amore appropriate focus may be on envi-ronmentally mediatedphenotypicassociations.One advantage of a niche-construction per-spective is that it emphasizes the value of think-ing of phenotypes as being reconstructed ineach generationby different developmental re-sources (genetic and nongenetic), rather thanas expression of genetically encoded informa-tion only. The focus of some methods (e.g.,community genetics) on heritable phenotypic

variation to the exclusion of nonheritable vari-ation (e.g., plasticity, epigenetics, populationstructure) may miss many interesting eco-evolutionary processes (Uller 2008; Ellner et al.2011; Palkovacs et al. 2012).

Methods of Implementation

The sheer complexity of engineeringwebs and the challenge of tracing evo-eco-evo causal chains through ecosystemsappears to be a daunting ordeal for theecologist. In reality, the practical study ofecology and evolution is not changed bythis perspective, as standard methods reg-ularly and successfully deployed by ecolo-gists and evolutionary biologists can becombined to good effect. What is differentis the focus of investigation, which movesfrom the study of the ecological impact orevolutionary response in a single taxon tothe investigation of eco-evolutionary sys-tems, pathways, or networks. This requiresthat researchers go beyond the normalpractice of evolutionary biology and ask,What causes the selection pressures lead-ing to a specific evolutionary response?rather than treating those selection pres-sures as a starting point. It also requires

researchers to go beyond the normal prac-tice of ecosystem ecology and ask, Whatevolutionary ramifications follow from spe-cies ecological impacts on biota and abi-ota?

Any eco-evolutionary system underpinnedby niche construction can be viewed as achain of component links that either com-prise the environment-altering (ecosystemengineering) aspect of niche construction(or its downstream consequences) or a sub-sequent evolutionary response to prior nicheconstruction (Odling-Smee et al. 2003). Thestudy of eco-evolutionary dynamics will requireresearchers to utilize the methods of both ecol-ogy and evolutionary biology. The key to prog-ress is to break down complicated pathways ineco-evolutionary networks into tractable com-ponent pieces (e.g., Post and Palkovacs 2009),subject each to analysis and then reconstructthe network, including the strength of in-teractions and how they vary over time togain a systems-level understanding. So-phisticated techniques to study networkstructure and dynamics have been devel-oped via statistical mechanics and appliedto a wide variety of problems, from socialdynamics revealed by internet traffic to thedynamics of power grids and food webs(Newman 2010).

The guppy system studied by Palkovacset al. (2009) serves as a useful example(Figure 3b). The first link in the chain isdifferential predation on two guppy popu-lations. This environment-altering aspectof niche construction by the guppy preda-tors leaves an ecological signature in theform of different size distributions in twoguppy populations. Characterizing thislink is, at least in principle, straightfor-

ward: it requires that the predators beidentified, the intensity of predation mea-sured, and the guppy size distributionssampled and quantified. Ideally, statisticalanalyses would link the intensity of preda-tion to the resulting prey distributions. Itis, of course, possible that there has beenan evolutionary response to this predationin the guppies with, for instance, more in-tense predation favoring smaller adult guppysizes, alternative trophic morphology, or di-etary preferences. This is highly plausible




22/27

given the evidence for rapid evolution inguppies (Reznik et al. 1997) and in manyother species in which the response to se-lection has been measured (Endler 1986;Kingsolver et al. 2001; Ellner et al. 2011).

Again, this can be investigated using stan-dard procedures (Endler 1986).

The second link in the chain is the differen-tial excretion by the two populations and itsimpact on nitrogen (N) and phosphorus (P)levels in the water. This second environment-altering component of niche constructionstems from the guppies and leaves an ecologi-cal signature in two abiotic environmentalcomponents: the local aquatic N and P levels.Palkovacs et al. (2009) were able to demon-strate this link experimentally with a mesocosmexperiment. The results showed that high-predation guppy populations contributedapproximately double the amount of Nand P to the total nutrient pool via excre-tion compared to low-predation popula-tions. In this way, differential excretion wasidentified as a correlate of the signature ofniche construction (modified N and P).Under conditions of equal biomass, a pop-ulation dominated by smaller individuals(the high-predation population) is expectedto drive higher nutrient fluxes than a pop-ulation dominated by larger individuals(low predation; Hall et al. 2007).

The third link in the chain connects theexcretion-mediated nutrient fluxes to algalgrowth in the streams. This link could bedemonstrated in the field by correlatingmeasures of guppy excretion rates (or, morerealistically, estimates of these based on as-says of population densities as a function ofsize) with rates of algal growth. It could alsobe demonstrated experimentally, either un-der controlled conditions in the laboratoryor micro-/mesocosms or in the field, by can-celing or enhancing the environment-altering behavior (excretion) or mimicking itsniche-constructed effects (Odling-Smee et al.2003). Relevant experimental manipulationsinclude: canceling or enhancing the nicheconstructors current activity by removing indi-

viduals and/or supplementing their numberswith introduced members of the same species(e.g., decreasing/increasing guppy num-bers or manipulating size distributions);

canceling or enhancing niche constructorscurrent activity by removal and/or providingthem with the resources necessary for popula-tion growth (e.g., removing prey or feeding theguppies); canceling or enhancing niche con-structors current activity by removing from orsupplementing the ecosystem with introducedmembers of the same guild (i.e., removing oradding a different species that engineers in thesame manner, such as another small fish withsimilar impact on nutrient flux); artificially re-moving or manufacturing and introducing theproducts of the niche constructors (e.g., chem-ically extracting or adding N and P to mimicthe effects of excretion); and counteractingnegative effects and facilitating positive influ-ences of both abiotic and biotic factors thatmay affect niche constructors through trophicor nontrophic links (e.g., interfering with thepredation regimes). In all cases, the effects ofthese manipulations on algal growth could bemonitored.

The final link in this chain could be to testa hypothesized feedback to trait evolutionby connecting changes in algal biomass toguppy color patterns by demonstrating a re-sponse to (natural or sexual) selection in thischaracter through sensitivity to the amountof algae-derived carotenoids available in theenvironment, again through standard proce-dures for detecting selection (Endler 1986).

As this example demonstrates, testing linksin eco-evolutionary chains of causality re-quires working at a variety of scalesfrom the

whole-ecosystem scale to the small-scale exper-iment. Small-scale experiments are needed toisolate trait change as the driver of ecologicalchange and can be useful for running select-ion experiments under controlled conditions.However, to show that trait-driven ecosystemdynamics and feedback effects are occurringalso requires simultaneous studies at the whole-ecosystem scale, where such dynamics play outin nature.

In practice, the above procedures arelikely to be complicated by many additionalchallenges, including the aforementionedspatial effects, time-lagged responses andother convoluted temporal dynamics, slowresponses to selection in long-lived species,and so forth. Nonetheless, this example illus-trates the tractability of tracing at least some




23/27

causal pathways via eco-evolutionary feed-backs and feed-forwards. Researchers canpotentially go beyond merely identifyingcausal influences to quantifying their mag-nitude and duration. Other procedures thatcan be deployed include experimental manip-ulation of the potency of niche construction,common gardening experiments, selection ex-periments, and experimental evolution. Whererelevant data are available, statistical ap-proaches such as structural equation modelingand causal graphs can help to isolate or con-firm putative causal influences and/or rejectcausal hypotheses that areinconsistent with thedata (Pearl 2000; Shipley 2000). Once com-peting causal hypotheses are translated intostatistical models, researchers can assess theevidence for each by standard statistical anal-

yses or comparing competing hypotheses byusing information theoretic or Bayesian cri-teria. The same methods allow the magni-tude of causal influences to be quantifiedand niche-constructing effects on variables tobe distinguished from other processes. Withtime-series data, the duration of niche-constructing effects and the rate of response toselection can also be quantified.

Ecosystems are threaded by a bewilderingarray of EMGAs, so a primary concern forresearchers is how to constrain the system toa manageable size. This amounts to deciding

which acts of niche construction and whichresponses to selection are too weak, tran-sient, or trivial to merit inclusion and, con-

versely, which are too important to ignore. Inmany instances, likely or plausible key niche-constructing processes can be identified apriori by taking account of the factors de-scribed in the section entitled From Ecolog-ical Signatures to Modified Selection thatscale up the impact of niche construction,supplemented by a knowledge of the basicbiology of the species (Jones et al. 1994,1997). Ideally, however, researchers would gobeyond identifying a plausible key niche-constructing process to actually demonstratingits importance through data collection and sta-tistical analysis or experimental manipulationsand estimating its magnitude and duration.Likewise, in principle, putatively unimportantprocesses can be confirmed to be so throughthe same procedures.

The starting point for an investigationneed not be the first link in the chain, asresearchers can potentially work backward as

well as forward in identifying causal influ-ences (Odling-Smee et al. 2003). We envis-age that the signatures of niche constructionhighlighted in the sections Niche Construc-tion Theory and Ecosystem-Level Ecologyand The Evolutionary Consequences ofPrior Niche Construction might frequentlybe the starting points for analysis, triggeringinvestigations of their downstream conse-quences for other ecosystem compartmentsor, alternatively, exploration of the causes ofthe niche construction.

The boundary of the system, in terms ofwhere the eco-evolutionary causal chain orweb should start or stop, is a decision forresearchers, given the nature of their re-search interests and the resources availableto them. In principle, systems could range inscale from short-term responses in isolatedeco-evolutionary links to analysis of long-term patterns of change in entire ecosys-tems. In the latter case, we note increasingreference to ecosystem evolution withinthe ecological literature (e.g., Loreau et al.2004), and point out that, to the extent thatecosystems comprise not just biota but alsoabiota, the integrated changes in entire eco-systems over evolutionary scales must be me-diated by niche construction. Related to thispoint, given that much niche construction isstabilizing in effect, by counteracting priorchanges in the environment (Odling-Smeeet al. 2003), it follows that the absence ofchanges in entire ecosystems over evolution-ary scales in spite of evolutionary changes incomposite species must be mediated byniche construction.

Thus far we have said little about genes, yetour definition of EMGAs places emphasis onindirect associations between potentially dis-ta

niche construction theory: a practical guide for ecologists

Documents