Organisms As Ecosystem Engineers Author(S): Clive G

Nordic Society Oikos

Organisms as Ecosystem Engineers Author(s): Clive G. Jones, John H. Lawton and Moshe Shachak Reviewed work(s): Source: Oikos, Vol. 69, No. 3 (Apr., 1994), pp. 373-386 Published by: Wiley on behalf of Nordic Society Oikos Stable URL: http://www.jstor.org/stable/3545850 . Accessed: 11/02/2013 12:21

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].

Wiley and Nordic Society Oikos are collaborating with JSTOR to digitize, preserve and extend access to Oikos.

http://www.jstor.org

This content downloaded on Mon, 11 Feb 2013 12:21:08 PM All use subject to JSTOR Terms and Conditions OIKOS 69: 373-386.Copenhagen 1994

Organismsas ecosystemengineers

Clive G. Jones,John H. Lawton and Moshe Shachak

Jones,C. G.,Lawton, J. H. andShachak, M. 1994.Organisms as ecosystemengineers. - Oikos 69: 373-386.

Ecosystemengineers are organisms that directly or indirectly modulate the availability of resourcesto otherspecies, by causingphysical state changes in bioticor abiotic materials.In so doingthey modify, maintain and create habitats. Autogenic engineers (e.g. corals,or trees)change the environment via theirown physical structures (i.e. theirliving and dead tissues). Allogenic engineers (e.g. woodpeckers, beavers) change theenvironment bytransforming living or non-living materials from one physical state to another,via mechanicalor othermeans. The directprovision of resources to other species,in theform of livingor dead tissuesis notengineering. Organisms act as engineerswhen they modulate the supplyof a resourceor resourcesother than themselves.We recognise and define five types of engineering and provide examples. Humansare allogenicengineers par excellence,and also mimicthe behaviour of autogenicengineers, for example by constructingglasshouses. We explorerelated conceptsincluding the notions of extendedphenotypes and keystone species. Some (but not all) productsof ecosystemengineering are extendedphenotypes. Many (perhapsmost) impacts of keystone species include not only trophic effects, but also engineersand engineering. Engineers differ in theirimpacts. The biggesteffects are attributabletospecies with large per capita impacts, living at high densities, over large areasfor a long time,giving rise to structuresthat persist for millennia and that modulatemany resource flows (e.g. mimamounds created by fossorial rodents). The ephemeralnests constructed by small,passerine birds lie at theopposite end of this continuum.We provide a tentativeresearch agenda for an explorationof the phenom- enonof organismsas ecosystemengineers, and suggestthat all habitatson earth support,and are influenced by, ecosystem engineers.

Jones,C. G., Inst.of Ecosystem Studies (IES), BoxAB, Millbrook,NY 12545, USA. - J. H. Lawton,NERC Centrefor Population Biology,Imperial College, Silwood Park, Ascot,Berks, UK, SL5 7PY,and IES. - M. Shachak,Mitrani Centerfor Desert Ecology, Blaustein Inst. for Desert Research, Ben-Gurion Univ. of the Negev, Sede Boqer 84900, Israel, and IES.

Interactionsbetween organisms are a majordeterminant common. The ecological literature is rich in examplesof of the distributionand abundanceof species.Ecology habitatmodification by organisms,some of whichhave textbooks(e.g. Ricklefs1984, Krebs 1985, Begon et al. beenextensively studied (e.g. Thayer 1979, Naiman et al. 1990) summarisethese important interactions as intra- 1988).However, in general,population and community and inter-specificcompetition for abiotic and bioticre- ecologyhave neitherdefined nor systematically identi- sources,predation, parasitism and mutualism.Conspic- fiedand studied the role of organisms in thecreation and uouslylacking from the list of key processes in most text maintenanceof habitats.There is noteven a word,or booksis therole that many organisms play in thecre- words,in commonuse to describethe process. We will ation,modification and maintenanceof habitats.These call the process EcosystemEngineering and the orga- activitiesdo notinvolve direct trophic interactions be- nismsresponsible Ecosystem Engineers. tweenspecies, but they are neverthelessimportant and Beaver(Castor canadensis) are familiarexamples of

Accepted21 September1993 Copyright(? OIKOS 1994 ISSN 0030-1299 Printedin Denmark- all rightsreserved

OIKOS 69:3 (1994) 373

This content downloaded on Mon, 11 Feb 2013 12:21:08 PM All use subject to JSTOR Terms and Conditions organismsacting as ecosystemengineers. By cutting impactsof engineerson communitiesand ecosystems. treesand using them to construct dams they alter hydrol- We areinterested solely in discoveringproperties that all ogy,creating wetlands that may persistfor centuries. engineershave in common.We addressthe scale and "Theseactivities retain sediments and organic matter in magnitudeof theireffects later. the channel,... modifynutrient cycling and decomposi- tiondynamics, modify the structure and dynamics of the riparianzone, influence the character of water and materials transporteddownstream, and ultimatelyinfluence plantand animal community composition and diversity" Classificationof organisms as engineers (Naimanet al. 1988). Table1 summarisesexamples of organisms as ecosystem However,beaver are by no meansthe only ecosystem engineers.The table is illustrative,not exhaustive. Addi- engineers.As we willshow, a vastarray of specieshave tionalexamples are discussed at greater length in the text. effectsthat are fundamentallysimilar, albeit often on All theexamples of whichwe are awarecan be as- moremodest spatial and temporal scales. Yet there is no signedto one of fivepossible cases (Fig. 1), or to a commonlanguage to describewhat ecosystem engineers combinationof twoor moreof these cases. As in many do, no formalstructure to modeltheir effects, and no otherareas of ecology,the diversity of biologicalpro- generaltheory round which to organiseunderstanding of cesses meansthat precise pigeon-holing is sometimes theprocess. Examples, which are developedmore for- difficult.The boundariesbetween types of engineering mallybelow (for others, see Table 1), includenot only areoccasionally fuzzy and, in thereal world, separating beaverand their dams but also gophers,ants and termites engineeringfrom other ecological processes may also be thatmove soil, woodpeckersthat drill holes, alligators difficult,simply because these non-trophic interactions thatmake wallows, rock-eating snails, trees, corals, sea- alwaysco-occur with trophic interactions. We discuss grassbeds and Sphagnum blanket bogs. some difficultcases as we proceed.The majorityof The purposesof thisarticle are fourfold: (i) to define examplesare, however, easy to classify.The legendto andto giveexamples of ecosystem engineering by orga- Fig. 1 explainsthe conventional notation used to describe nisms;(ii) to developa conceptualframework that ex- them.For clarity, itis easiestto introducethe arguments plainsand classifiesits effects; (iii) to showhow orga- usingbeaver and their dams. nismalengineering differs from related concepts (e.g. Beaverconform to case 4 in Fig. 1. Thatis theyare 'keystonespecies' Paine 1969,Krebs 1985); (iv) andto allogenicengineers, taking materials in theenvironment identifyquestions for further work on organismsas eco- (in thiscase trees,but in themore general case it can be systemengineers. First we definewhat we meanby an anyliving or non-living material), and turning them (en- ecosystemengineer, before providing examples and a gineeringthem) from physical state 1 (livingtrees) into conceptualframework for what is, andis not,ecosystem physicalstate 2 (deadtrees in a beaverdam). This act of engineering. engineeringthen creates a pond,and it is thepond which hasprofound effects on a wholeseries of resource-flows usedby other organisms. The critical step in thisprocess is thetransformation oftrees from state 1 (living)to state 2 (a dam).This transformation modulates the supply of Definitions otherresources, particularly water, but also sediments, Ecosystemengineers are organismsthat directly or in- nutrientsetc. A criticalcharacteristic ofecosystem engi- directlymodulate the availability of resources (other than neeringis thatit mustchange the availability (quality, themselves)to otherspecies, by causingphysical state quantity,distribution) ofresources utilised by other taxa, changesin bioticor abioticmaterials. In so doingthey excludingthe biomass provided directly by thepopula- modify,maintain and/or create habitats. tionof allogenicengineers. Engineering is not the direct The directprovision of resourcesby an organismto provisionof resources in the form of meat, fruits, leaves, otherspecies, in theform of living or dead tissues is not orcorpses. Beaver are not the direct providers of water, in engineering.Rather, it is thestuff of most contemporary the way that prey are a directresource for predators, or ecologicalresearch, for example plant-herbivore orpred- leavesare food for caterpillars. ator-preyinteractions, food web studiesand decomposi- Now considerthe autogenicequivalents of beaver tionprocesses. (Fig. 1, case 3). Simpleexamples are thegrowth of a Autogenicengineers change the environment via their forestor a coralreef. Trees and corals are direct sources ownphysical structures, i.e. theirliving and dead tissues. offood and living space for numerous organisms, but the Allogenicengineers change the environmentby trans- productionof branches, leaves or living coral tissue does formingliving or non-living materials from one physical notconstitute engineering. Rather, it conformsto case 1 stateto another,via mechanicalor othermeans. inFig. 1 (thedirect provision of resources). However, the Armedwith these definitions we can nowproceed to developmentof the forest or thereef results in physical considersome detailedexamples. We are not,at this structureswhich do changethe environment and mod- juncture,concerned with the magnitude or scale of the ulatethe distribution and abundance of otherresources.

374 OIKOS 69:3 (1994)

This content downloaded on Mon, 11 Feb 2013 12:21:08 PM All use subject to JSTOR Terms and Conditions Table1. Examples of organisms acting as ecosystemengineers. Classification according to Fig. 1. Additionalexamples are discussed in the text. Organism Habitat Activity Impact Refs.

Case 2 (allogenic) Americanalligators, Everglades createwallows retainwater in droughts;provide Finlayson& Alligatormississippiensis NationalPark refugesfor fish, fisheating birds, etc. Moser(1991) Rabbits,Oryctolagus Europe digextensive burrows burrowsoccupied by otherspecies, Southern(1964); cuniculus,badgers, Meles (rabbitwarrens, badger e.g fox,Vulpes vulpes, and by many Neal & Roper meles setts) invertebrates (1991) Case 3 (autogenic) Marinephytoplankton Gulfof Maine bloomsof phytoplankton enhancewarming of surfacewaters Townsendet al. particlesscatter and absorb thatmay initiate development of (1992) lightin upperlayers of thermocline watercolumn Microalgaein sea ice Antarctica scatterand absorb light enhancemelting and break up ofice Buynitskiy(1986); withinice andunderlying Arrigoet al. seawater;reduce strength of (1991) ice Freshwaterphytoplankton Lake St. George, interceptlight in upper lightinterception leads to shallower Mazumderet al. Ontario watercolumn; small algal mixingdepth, lower metalimnetic (1990) spp.more effective than temperaturesand lower heat content largespp. of watercolumn Cyanobacteriaand other desertand exudemucilaginous organic gluethe organisms, organic matter West(1990) nonvascularplants semi-desertsoils compounds andsoil particlestogether to forma microphyticcrust; change infiltration, percolation,retention and evaporation of water;reduce soil erosion;affect seedlingemergence Bog moss,Sphagnum spp. Northernand build'blanket' and 'raised' majorchanges in hydrology,pH, and Tansley(1949) westernBritain bogsvia accumulatedpeat topography Submergedmacrophytes freshwaterlakes, growto createweed beds attenuatelight; steepen vertical Carpenter& Lodge pondsand rivers temperaturegradient; retard flow; (1986) enhancesedimentation; oxygenate rhizosphere Foresttrees (broad-leaved HubbardBrook shedbranches and trunks createdebris dams; alter morphology Likens & Bilby andconiferous) Experimental intostreams andstability of streamchannels, (1982); Hedinet Forest,New storageand transport of dissolved al. (1988) Hampshire organicmatter and sediments; differenttree species may create dams whichdiffer in persistence Higherplants ubiquitous deadleaves etc. accumulate altermicroenvironment of soil; Facelli& Pickett as litter changesurface structure, affecting (1991) drainage,and transfer of heatand gasses;act as physicalbarrier for seedsand seedlings; numerous impactson structureand composition of plantcommunities Terrestrialplants in 29 ubiquitous growstructures (modified createsmall aquatic habitats, Fish(1983) families,with >1,500 leaves,leaf axils etc.) that supportinga highly specialised insect species impoundwater fauna

Case 4 (allogenic) Marinemeiofauna ubiquitous biodeposition,bioturbation, change physical, chemical and Reichelt(1991) (protozoaand porewatercirculation, and biologicalproperties of sediments; representativesof many faecalpellet production changedirection and magnitude of invertebratephyla) nutrientfluxes; increase oxygenation of sediments Marineburrowing ubiquitous burrowinto and redistribute create dynamic sediment mosaics; Anderson& macrofauna sediments;bioturbation; activelytransport solutes into Kristensen(1991); burrowventilation burrows;increase oxygenation of de Wilde(1991); sediments;stimulate microflora; Meadows& increasedecomposition rates Meadows(1991b) (cont.) OIKOS 69:3 (1994) 375

This content downloaded on Mon, 11 Feb 2013 12:21:08 PM All use subject to JSTOR Terms and Conditions Tab. 1. (cont.) Organism Habitat Activity Impact Refs.

Marinezooplankton ubiquitous fiterliving, dead organic sinkingfaecal pellets important in Dunbar& Berger andinorganic (e.g. clay) verticaltransport and exchange of (1981); Wallaceet particles,and concentrate elementsand organic compounds in al. (1981); Fowler intofaecal pellets oceans & Knauer(1986) Fiddlercrab, Uca pugnax New England digburrows increasesoil drainageand oxidation- Bertness(1985) saltmarsh reductionpotential; increase decompositionrates; increase primary productionat intermediatetidal heights Europeanperiwinkle, New England bulldozesediments from preventsediment accumulation and Bertness(1984a) Littorinalittorea rockybeach hardsubstrates hencegrowth and establishment of algalcanopy; algae arecase 3 engineersand further increase sedimentationrates; faunal compositionmarkedly different with andwithout snails Snails,Euchondrus spp. Negevdesert eat endolithiclichens and increaserate of nitrogencycling, soil Shachaket al. therock they grow in formationand rock erosion (1987); Jones& Shachak(1990) Bagwormcaterpillars, GoldenGate eat endolithiclichens and smallincrease in erosionrate, nutrient Wessels & Wessels ?Penestoglossasp. Highlands,South constructlarval shelters cyclingand soil formation (1991) Africa ('bags') fromquartz crystals Mound-buildingtermites, widespreadin moundand subterranean changemineral and organic Wood& Sands Isoptera tropicsand galleryconstruction; compositionof soils;alter hydrology (1978); Lal (1991) subtropics redistributionof soil anddrainage particles Ants,Formicidae ubiquitous nestand subterranean changelocal structureand Elmes(1991) galleryconstruction; compositionof soils;alter 'above redistributionof soil nest'vegetation; produce microsite particles enrichment Earthworms,Lumbricidae, ubiquitous burrowing,mixing and changemineral and organic Lal (1991); Megascolecidae casting compositionof soils;affect nutrient Thompsonet al. cycling;alter hydrology and drainage; (1993) affectplant population dynamics and communitycomposition Blindmole rats, Spalax Israel diggingand tunnelling movelarge quantities of soil; increase Heth(1991) ehrenbergi aeration;create distinctive ecosystem Mole rats,Bathyergidae SouthAfrican diggingand tunnelling createimpressive, cratered landscapes, Richardson et al. (severalgenera) lowlandfynboss witheffects on soil formation,plant (in press) productivityand speciescomposition Prairiedogs, Cynomys spp. NorthAmerican continualintense disruption changephysical and chemical Whicker& Detling shortand mixed byburrowing, creating soil propertiesof soil persistingfor 100- (1988) grassprairie mounds 1000s of years Pocketgophers, Geomys NorthAmerican constructtunnels and move alterpatterns and rates of soil Huntly& Inouye bursarius grasslandsand soil to surfacemounds development,nutrient availability and (1988); Moloneyet aridshrublands microtopography;change plant al. (1992) demography,diversity and primary productivity;affect behaviour and abundanceof otherherbivores Indiancrested porcupine, Negevdesert diggingfor food dig up to 2-3 holesm-2; diggings Yair& Rutin Hystrixindica accumulateorganic matter, runoff (1981); Gutterman water;create favourable sites for seed (1982) germination Elephants,Loxodonta EastAfrican physicaldisturbance and widespreadvegetation changes; Naiman(1988) africana woodlandand destructionof treesand alterationof fireregime; effects on savannah shrubs foodsupply and population dynamics of otheranimals; ultimately changes in soil formation,riparian zones, and biogeochemicalcycling (cont.)

376 OIKOS 69:3 ( 1994)

This content downloaded on Mon, 11 Feb 2013 12:21:08 PM All use subject to JSTOR Terms and Conditions Tab. 1. (cont.) Organism Habitat Activity Impact Refs.

Case 5 (autogenic)and case 6 (allogenic)(examples combining elements of both) Crustosecoralline algae, coralreefs overgrowand cement breakforce of waterand protect Anderson(1992) Porolithon,Lithophyllum togetherdetritus on outer coralsagainst major wave action; algalridge of barrier reef effectvia ownbodies (case 5) and secretionof 'cement'(case 6) Ribbedmussels, RhodeIsland secretebyssal threads, and on marshedge, dense beds of mussels Bertness(1984b) Geukensiademissa Spartinasalt formdense mussel beds (case 5) andbyssal threads (case 6) marsh bindand protect sediments and preventphysical erosion and disturbance,e.g. by storms

AUTOGENIC ALLOGENIC

...... ; X,i: ..0.. . . t

.- AISI -ORGANISM [j ] STATE2 - -| STATEI | l i? =REOURCE

-. . :. ,". . . . , ., ...... E . 0.f~. i.'e':.'-: ,:'. ---' '. - ...... '. . . I...... I . . . -l ...... -t...... - NOT ENGINEERINGgo CASE 1 CASE 2

Fig. 1. Conceptualmodels of autogenicand allogenic engineeringby organisms. ERESOURCE FLO Fordefinitions and examples see textand Table 1. The symbolX definespoints of ORGAMSU modulation.For example, STATE 2. STATE STATES - allogenic engineerstransform ...La ______livingor non-livingmaterials fromstate 1 (rawmaterials) to state2 (engineeredobjects ORGOANI andmaterials), via mechanicalor othermeans. The equivalent(state 2) productsof autogenic CASE 3 CASE 4 engineeringare the living anddead tissues of the engineer.These products of bothautogenic and allogenic rESOURCEFLOWS engineeringthen modulate theflow of one or more resourcesto otherspecies OROA= or modulatea STATEM ORGAESM STATE2 (cases 2-4) T.TlI S=AT, L 1] majorabiotic controller (e.g. . fire),which in turn modulatesresource flows A5TIC (cases 5-6). Case 1, the ABNOTIC ORGANIsMu directprovision of resources CONTROL C#BTROL by one speciesto anotheris notengineering, and involvesno modulationof resourceflows. CASE 5 CASE 6

25 OIKOS 69:3 (1994) 377

This content downloaded on Mon, 11 Feb 2013 12:21:08 PM All use subject to JSTOR Terms and Conditions This modulationconstitutes autogenic engineering. Trees It is not universallythe case thatplants are allogenic alterhydrology, nutrient cycles and soil stability,as well and animals are autogenicengineers. Enhanced rates of as humidity,temperature, windspeed and lightlevels (see physicaland chemical weatheringof rocks into soil by Holling 1992); corals modulatecurrent speeds, siltation algae or higherplants (Bloom 1978) constituteallogenic rates and so on. It is obvious, but surprisinglyrarely engineering(case 4). Plantsmay act eitheras autogenic explicitlystated, that numerous inhabitants of thehabitats or as allogenicengineers, and providesome of the most so created are dependentupon the physicalconditions complex cases of ecosystemengineering, to which we modulatedby theautogenic engineers, and uponresource now turn. flows whichthey influence but do not directlyprovide; Autogenic(case 5) and allogenic(case 6) engineering withoutthe engineers,most of these other organisms are at the extremesof continuathat merge with cases 3 would disappear. and 4 respectively.Cases 5 and 6 have the common One furtherexample may help to clarifythe distinction propertythat autogenicor allogenic engineersinteract betweencase 1 and case 3. The growthof seagrassbeds with,and modulatepowerful abiotic forces, for instance modulatesocean currents,which in turnmay altersedi- firesor hurricanes.Examples in cases 5 and 6 are dis- mentationrates and hence food suppliesfor other orga- tinguishedfrom cases 3 and 4 by theextreme magnitude nisms with substantialeffects upon theirperformance of the processesmodulated by the engineers,and by the (e.g growthand survivalin the clam Mercenariamerce- factthat these major abiotic forces are themselvesfunda- naria (Irlandiand Peterson1991)). The directprovision mentalmodulators of the distributionand abundanceof of food or living space by seagrasses (case 1) is not resources. criticalfor Mercenaria but the clam's survivalis nonethe- Fire providesa particularlyinteresting case. It is log- less dependentupon theseplants. ical, albeit unconventional,to regardthe productionof It could be argued that the growthof tree-trunks, combustibleliving and dead biomass as autogenicengi- branches,reefs or similarsubstantial biological structures neering (case 5). Differentspecies of plants produce (case 1, Fig. 1) itselfconstitutes ecosystem engineering. differentqualities and quantitiesof livingand dead fuel, Inclusion or exclusion is a matterof choice. We have modulatingthe magnitude,intensity and durationof fire chosento excludeit (whetherit is theprovision of foodor and, in turn,profoundly altering the supplyof resources of 'architecture'(Southwood et al. 1979 and Lawton formany other species (Christensen1985). High grass- 1983)) when the structuresare consideredsolely and land productivityin Serengeti-Marain the 1960s mark- directlyas resources,because thisdiffers in kindfrom the edly increased the incidence of fire,resulting in con- remainingcases in Fig. 1. To qualifyfor our definitionof versionof savannahwoodland to an alternativestate - an ecosystemengineer, an organismmust modulate the grassland- whichis now maintainedby elephants(Du- supplyof otherresources for other species, rather than be blin et al. 1990). The effectsof elephantsas allogenic thedirect provider of resources.The growthof biological engineersare summarisedin Table 1. structuresis thusa necessarybut not a sufficientrequire- Plants also act as allogenic engineers(case 6). In mentfor autogenic engineering. PuertoRico Dacryodesexcelsa treesare able to withstand We can illustrateour argumentsfurther by considering hurricanesbecause theirextensive roots and root grafts the simplest kind of allogenic engineering(case 2, bind and stabilise bedrock and superficialrocks; this Fig. 1). Variousorganisms make holes in treetrunks and species thereforedominates tropical montain forests branches, some quickly (woodpeckers), others more wherehurricanes are common(Basnet et al. 1992). slowly (rot fungi).They transformwood withoutholes intowood withholes, and indirectlyprovide resources for Difficultcases: Pollinators,gall former and cows othercreatures, nesting and roostingcavities for birds and The richnessof biological processesmeans thata com- bats for instance.The holes are the resource,not the pletely satisfactory,comprehensive yet exclusive de- organismsthat make them.Notice thatif some of the finitionof ecosystemengineering may be impossibleto holes fillup withwater (Kitching 1983), the littleponds achieve,although numerous examples are easy to classify so createdare examples of case 4, and are conceptually (Table 1). Given the diversityof species interactionsin identicalto beaver dams. nature,efforts to classifymany other ecological phenom- Inevitablythere are some greyareas in thisclassifica- ena suffersimilar problems 'at the margin'. tion and at the risk of being pedanticit is worthwhile Pollinators and gall-formerspresent an interesting consideringjust one. The naturalhollows and cavities challenge for our definitionof engineering.Both have formedat branchjunctions and root bases as the tree profoundeffects on the growthof plant tissue; in so grows,and which may subsequentlyfill up withwater doing,pollinators modulate the supplyof resourcesfor ('pans' sensuKitching (1983)) are notthe same as rot-or seed predators,and gall-formerscreate structures that are woodpecker-holesand they do not belong in case 4; used notonly by themselvesfor shelter and food,but also ratherthey conform to case 3 (trees are now autogenic by inquilines(e.g. Askew 1975). Both typesof interac- engineersbecause theirbiomass gives rise to water-filled tionconform broadly to case 2 (Fig. 1). However,we do hollows). Numerous examples of these 'phytotelmata' not findit helpfulto regardpollinators as engineers,not are summarisedby Fish (1983) (Table 1). least since self-pollinationis case 1 and not engineering.

378 OIKOS 69:3 (1994)

This content downloaded on Mon, 11 Feb 2013 12:21:08 PM All use subject to JSTOR Terms and Conditions But it is possible to regardgall-formers as engineersfor inquilines;they physically modify plant-tissues, and cre- Relatedconcepts ate new habitatsand resourcesfor other organisms. The The idea thatsome organismsalter the physical structure distinctionbetween pollinators and gall-formersis, how- of theirenvironment, with impacts on theirown and other ever,a fineone. populationsis notnew. But earlierwork either focuses on Our definitionof engineeringalso embraces other, particularspecies and habitatsand lacks generality,or unexpectedecological phenomena,for example dung takes a more general view but fails clearly to define masses producedby large herbivores.Cows turngrass ecological engineering,or to distinguishit fromother intocow pats,which are thencolonised by a richcommu- processes. For example,in an importantset of reviews nityof invertebrates,dependent upon the pats for food dealing withanimal influenceson ecosystemdynamics, and shelter(Mohr 1943). The physicalstructure and envi- co-ordinatedby Naiman (1988), engineeringand direct ronmentprovided by the droppingsis at least as impor- trophiceffects are interwoven.Within this series of pa- tantto its inhabitantsas the concentrationof food re- pers,Huntly and Inouye(1988) explicitlydescribe pocket sources(Elton 1966). It does not stretchthe definition of gophersGeomys bursarius as "soil engineers"because of engineeringtoo farto regardcows as allogenicengineers, theirrole as earth-movers.Gophers are, indeed, excellent turninggrass into cow pats(case 2). We wouldbe thefirst examples of allogenic (case 4) engineers(Table 1). to admit,however, that it is an unconventionalperspec- Ecologistsin generalhave paid surprisinglylittle atten- tive.Similar remarks apply to faecal pelletproduction by tion to how environmentsare createdand maintained; oceanic zooplankton(Table 1). For theseand otherbor- most appear contentto follow Andrewarthaand Birch derlinecases, thecommon sense way to view theissue is (1954) in recognising"a place to live" and "weather"as to ask whetherunderstanding the ecological interactions two of the four essential featuresof species' environ- is enhancedby recognisingthe engineeringdimension. ments,without formally considering the role of engineering in habitatmodification, creation and maintenance.

Bioturbators In marinebenthic environments the activitiesof large burrowinganimals are knownto play a dominantrole in Human analogues determiningthe physical structure of sediments,altering habitatsuitability for otherspecies (Rhoads and Young The parallels betweenecological and humanengineers 1970, Thayer1979, Lopez and Levinton1987, Meadows are, not surprisingly,very close. Humans are tool-using and Meadows 1991a, see also Table 1). Rhoads and organismsthat specialise in engineering.While human Young (1970) called the process 'trophicamensalism' engineeringoften has intentor purpose,it is probablytrue when large deposit feederscreate unstable sediments, to say that the major reason why humans have such restrictingthe presence of suspensionfeeders and attach- adverse effectson the environmentis because of the mentby sessile epifauna.The termamensalism is reason- unintendedconsequences of our activitiesas engineers. able, because theeffect is asymmetricaland is a formof Indeed people are now the primaryagents of environ- competitiveexclusion (Lawton and Hassell 1981), butthe mentalchange in mostareas of theworld (Naiman 1988, mechanismis nottrophic and clearlydiffers from normal Likens 1992). Manyhuman activities, from dam-building exploitationcompetition for food. 'Trophicamensalism' and skyscraperconstruction to forestclearance and the is actually anothergood example of case 4 allogenic dredgingand canalizationof watercourses, conform ex- engineeringbrought about by bioturbationof sediments. actly to cases in Fig. 1, in whichhumans are allogenic engineers,altering the physical environmentand mod- Patch dynamics ulatingthe flow of resourcesto otherspecies. Numerousstudies recognise the importance of patchesof Constructionof nestingboxes forbirds and hives for bare or differentsubstrates in otherwiseclosed communi- bees are examples of case 2. Ploughingby farmersand ties (e.g. Dayton 1971, Wiens 1976, Paine 1-979,Pickett the constructionof dams and reservoirsby waterengi- and White 1985). Patches may be createdby physical neersprovide examples of case 4. Buildingharbours and disturbance(waves, fire,landslips) or by theactivities of sea walls to reducestorm damage fromwaves are exam- organisms(grazing, predation or engineering),or by in- ples of case 6. Humans mimic autogeniceffects, using teractionsbetween engineering, trophic and physicalpro- tools to constructglasshouses and build air conditioning cesses. For some applicationsof patch dynamictheory plants(mimicking case 3), and by bulldozingfire breaks theway in whichpatches are createdmay be less impor- to counteractfire (mimicking case 5). We classifyhuman tantthan their existence. In general,however, we believe engineeringactivities as heavyor light,construction, civ- thatit is desirable to recogniseengineering as one of il, heating,plumbing, and air-conditioningto name buta several distinctways in which patches are createdand few. Organismsdo all thesejobs, and froma functional maintained,particularly since the factorsthat control perspectivewe see no fundamentaldifference between patch formationby engineersare oftendifferent from humanand non-humanengineering. those controllingpatch formation by abioticforces.

25* OIKOS 69:3 (1994) 379

This content downloaded on Mon, 11 Feb 2013 12:21:08 PM All use subject to JSTOR Terms and Conditions Animaland plant artifacts He also pointsout that not every example of what we In theirTheory of EnvironmentAndrewartha and Birch arenow calling allogenic engineering can be regardedas (1984) definea frameworkfor examining all thepro- an extendedphenotype, because impacts on the envi- cesses thatimpinge upon a singlespecies population. ronmentare of no consequenceto theengineer's fitness This targetspecies occupies the centrumof a web of and henceare notsubject to naturalselection. A good directlyand indirectly acting components, and Andrewar- examplewould be water-filledfootprints made by an thaand Birchpoint out that a linkin theweb maybe ungulate.The distinctionbetween engineering that is anotherliving organism or its artifactor residue.The subjectto naturalselection (because it is an extended Theoryof Environment therefore clearly allows for engi- phenotype)and engineering that is not('accidental' engineering,without explicitly identifying it as a defined neering)appears to be unimportantinterms of its shorter- modifierin the web, or as a processworthy of study in its termecological consequences; all typesof engineering ownright. modifyand modulate resource flows for other organisms. Meadowsand Meadows (1991a) and Meadows (1991) But theremay be interestinglonger-term differences, reviewthe environmental impacts of animal burrows and particularlyin the natureof the feed-backloops that burrowinganimals, and Hansell(1993) the ecological operateon 'extendedphenotype' versus 'accidental' consequencesof animalburrows and nests.Many of formsof engineeredartifacts. We returnto thispoint theseartifacts (e.g. meiofaunal burrows, megapode nests, later. termitemounds and molerat colonies) have landscape level effects,and serveto concentrateand redistributeKeystone species resourcesfor other species - thatis theyare classic Keystonespecies (Paine 1969,Krebs 1985, Daily et al. examplesof ecologicalengineering (Table 1). 1993) play a criticalrole in determiningcommunity Meadows(1991) pointsout that there are "underlying structure. By definition,removal of keystonespecies similaritiesbetween the impact of [burrowing]animals causesmassive changes in speciescomposition and other fromdifferent terrestrial and aquatichabitats on envi- ecosystemattributes. The criticallinks are usuallyre- ronmentalchange and modification." He providesa for- gardedas trophicand therefore within the realm of tradi- mal systemto quantifythe impacts of burrowing,dis- tionalecological thinking For example,removing top tinguishingbetween animals with large per capitabut predatorshas a cascadingeffect throughout the foodweb, geographicallyrestricted effects (e.g. badgers) and those alteringspecies composition and hence physical structure specieswith small per capitaeffects that nevertheless, and nutrient cycling at lower levels (Estes and Palmisano becauseof their abundance and distribution, have impacts 1974,Carpenter et al. 1987). on entirelandscapes (e.g. earthworms).Hansell (1993) Butcritical effects frequently involve engineering, for recognisesthat the "servicesand substancesof the example via disturbance(e.g. bioturbators(Thayer builderscreate a newrange of habitatniches which can (1979),above; case 4). In thefrequently cited example of be exploitedby a widevariety of specialists"and sug- sea-ottersEnhydra lutris, removal of ottersleads to an geststhat "the presence of nestbuilders and burrowers increasein sea urchins(Strongylocentrotus sp.) and can ... significantlycontribute to species diversityin henceto thedisappearance of kelpbeds, which in turn habitats."The examplesprovided by bothauthors all changeswave actionand siltationrates, with profound conformto eithercase 2 (otherspecies use thenests and consequencesfor other inshore flora and fauna (Estes and burrows)or case 4 (speciesrespond to changesin distri- Palmisano1974). Kelp areautogenic engineers (case 3); butionand abundance of resources). Their work therefore removal of kelp by urchinsis, amongstother things, differsfrom ours only in itsmore restricted focus. allogenicengineering (case 4). In otherwords, in this familiarexample, the species traditionally regarded as the Extendedphenotypes keystone(sea otter)has majoreffects because it changes The importanceof animalartifacts is also recognisedby theimpact of one engineer(urchin) on another(kelp), Dawkins(1982), as an exampleof species' extended withknock-on effects on otherspecies in the web of phenotypes.Dawkins writes (p.200): "A beaverdam is interactions.The equally well known impact of kangaroo builtclose to the lodge, but the effect of the dam may be rats(Dipodomys spp.) on desertvegetation occurs be- to floodan area thousandsof squaremeters in extent. causethe rodents not only eat seedsbut also causecon- As to theadvantage of the pond from the beaver's point siderablephysical disturbance. By burrowing and moving of view,the best guess seemsto be thatit increases largequantities of soil they create many shallow pits and thedistance the beavercan travelby water,which is littlemounds, which facilitate decomposition and the saferthan travelling by land,and easier for transporting establishment of annualplants (case 4 engineering) wood.)... If thisinterpretation is right the lake maybe (Brownand Heske 1990). regardedas a hugeextended phenotype." Dawkins recog- Directeffects of keystonespecies via theirrole as nisesother products of animalengineering as extensions engineershave recentlybeen reportedby Daily et al. ofspecies' phenotypes and hence subject to natural selec- (1993). Red-naped sapsuckersSphyrapicus nuchalis (a tion,including caddis-fly cases, termitemounds and typeof woodpecker) act as keystonespecies in two ways birds'nests. in Coloradosubalpine meadows. Their nesting holes

380 OIKOS 69:3 (1994)

This content downloaded on Mon, 11 Feb 2013 12:21:08 PM All use subject to JSTOR Terms and Conditions drilledin aspens, Populus tremuloides, are essential nest- 1992). Ecologicalengineering adds enormouslyto the ingsites for two species of swallows(case 2); theswal- catalogueof important,highly asymmetrical species in- lowsare missing from the community in theabsence of teractions,because engineersimpact upon manytaxa sapsuckers.Feeding holes drilledby thesapsuckers in (positivelyor negatively),but theremay oftenbe no willows,Salix spp.,also makesap flowsavailable to direct,reciprocal effects of the impacted species upon the severalbirds, mammals and insects (directly changing the populationof engineers. distributionand abundanceof thisresource for other One examplewill suffice.Beaver beneficially influ- species- andtherefore again conforming to case 2). encethe abundances of aquatic biota, but not vice versa; It is theoreticallypossible (though we thinkit willbe thatis theyhave massive 'bottom up' effects(o/+) that uncommonin practice) for a keystonespecies to exert its benefitnumerous other aquatic organisms. Their activ- effectsentirely trophically, without also actingas an itiesare also detrimentalto terrestrial species living up- engineer,or withoutchanging the engineering role of stream(and perhaps downstream) from the dam (o/-) just otherspecies in theweb. On theother hand, many engi- as bioturbatorsexclude sessile epifauna requiring stable neersare keystonespecies even though they play rela- substrates(see above). Generallywe expectboth the tivelyminor roles in communityfood webs. positiveand negativeeffects of engineersto be highly Krebsconcludes his textbook review of keystone spe- asymmetrical. ciesas follows:"Keystone species may be relativelyrare Thisis notto saythat there cannot be anyfeedbacks in naturalcommunities, or theymay be commonbut not fromorganisms in theengineered habitat, back to the recognised(our emphasis).At present,few terrestrialengineer. Undoubtedly there are, although feedback path- communitiesare believedto be organisedby keystone waysare probablyoften rather long, indirect, and fre- species,but in aquatic communities keystone species may quentlyslow. They remain virtually unstudied. For some be common".We believe that such views probably reflect engineers,itis difficulttoimagine any reciprocal effects. a consensusamong ecologists. They persist because we Forinstance, the insect inhabitants of abandonedbirds' havefailed to recognizethe role of ecosystem engineers nests(e.g. sometineid moths in thegenus Monopis and as keystonespecies. It is trite,but true, that a forestis a staphylinidbeetles in thegenus Microglotta ( Walsh and forestbecause it has trees,which not only provide food Dibb 1954,Emmet 1979)) probably never encounter the and livingspace butwhich also autogenicallyengineer builder.Gophers, in contrast,engineer soil and change theforest climate, and modulate the flows of many other vegetationcomposition, biomass and productivity (Table resourcesto forestinhabitants, both above- and below- 1); in turn,grasshopper populations become more abun- ground.Many singlespecies of treesin temperateor dantin the vicinity of gopher mounds (Huntly and Inouye borealforests (with low tree-speciesrichness) are both 1988).Two feedbacks may operate on gophers.The first keystonespecies and significantecosystem engineers. is reasonablywell documented,positive and relatively Ourviews are very close to Holling's (1992) Extended direct,via plantsthat are food for gophers; soil disturb- KeystoneHypothesis, in whichhe arguesthat "all ter- ancefavours the plant species that gophers prefer to eat. restrialecosystems are controlledand organisedby a Second,it is at leastconceivable, but untested, (D. Til- smallset of key plant, animal, and abiotic processes that man,pers. comm.) that grasshoppers compete with goph- structurethe landscape at different scales." We would add ersfor food, providing a longer, negative feed-back loop twopoints. First a critical,but not exclusive controlling on gophernumbers. mechanismis someform of engineering; and second, we We predictthat if feed-backs exist at all betweenengi- believethat keystone engineers occur in virtuallyall neersand the organisms they affect, they will character- habitatson earth,not just terrestrial ones. isticallybe indirect,involving several intermediate pro- cessesand species.

'Top down' vs 'bottomup', asymmetrical and indirecteffects Spatial and temporalscales Traditionalpopulation models focus on reciprocallycou- The impactof an ecologicalengineer depends upon the pledpairs of interactions,interspecific competition (-I-), spatialand temporalscale of its actions.Water filled enemy-victim(-/+) and so on (Williamson1972). Highly woodpeckerholes and beaver dams may both be exam- asymmetricalcompetitive interactions (amensalism; 0/-) ples of case 4 engineering,but there is notmuch doubt arecommon, possibly the norm, in some situations (Law- aboutwhich is themore significant ecological phenom- tonand Hassell 1981). Enemy-victiminteractions may enon.Six factorsscale the impact of engineers. They are: also be asymmetrical(donor-controlled; 0/+), that is prey abundancecontrols predator abundance, but not vice (i) Lifetime per capita activity of individual organisms. versa(Lawton 1989, Hawkins 1992), witha growing (ii) Populationdensity. debatein ecologyabout the relative importance of such (iii) The spatialdistribution, both locally and regionally, 'bottomup' vs 'top down' effects(Hunter and Price of thepopulation.

OIKOS 69:3 (1994) 381

This content downloaded on Mon, 11 Feb 2013 12:21:08 PM All use subject to JSTOR Terms and Conditions (iv) The lengthof timethe population has been presentat designedto quantifythe impactof ecosystemengineers a site. by removingor adding species. Studies by Bertness (v) The durabilityof constructs,artifacts and impactsin (1984a,b, 1985) are excellentexamples of manipulative the absence of the originalengineer. experimentson bothallogenic (cases 4 and 6) and auto- (vi) The numberand types of resourceflows that are genic(case 5) engineers(Table 1). A recentstudy by Hall modulatedby the constructsand artifacts,and the et al. (1993) shows the potentialpower of field manip- numberof otherspecies dependentupon these flows. ulationsfor disentangling per capita impactsfrom pop- ulationimpacts (although in the presentcontext it is not Thus,the most obvious ecological engineeringis attribut- ideal because predationand disturbance[case 4 engineer- able to species withlarge per capita effects, living at high ing] effectsare confounded).Edible crabs (Cancer pagu- densities,over large areas fora long time,giving rise to rus) huntingfor prey dig pitsin shallow subtidalareas of structuresthat persist for millennia and which affect the west coast of Scotland. The pits are conspicuous manyresource flows - forinstance mima mounds created featuresof theseabed topography,yet exclusion of crabs by fossorialrodents, including gophers (Cox and Gakahu fromareas of the sea bed for twelve monthsfailed to 1985, 1986, Cox et al. 1987, Naiman 1988). Autogenic reveal any landscape-leveleffects of crab pits,either on engineersmay also have massive effects;as Holling substratestructure and composition(particle sizes, orga- (1992) succinctlystates: "To a degree, ... the boreal nic carbon etc.) or faunal diversity,composition and forest'makes its own weather' and the animals living abundance.Crabs appear not to be abundantenough to thereinare exposed to more moderateand slower var- significantlyalter community structure, either by preda- iationin temperatureand moisturethan they would other- tionor by engineering,despite large and visuallyconspic- wise be." Boreal foresttrees have largeper capitaeffects uous individualimpacts. This example contrastsmark- on hydrologyand climaticregimes, occur at highdensi- edly with the substantialeffects of fiddlercrabs (Uca ties over large areas, and live for decades. But their pugnax) on productivity,decomposition, oxygenation impacts as autogenic engineersmay have a relatively and drainagein a New England salt marsh,revealed by shortmemory if the forestis logged. experimentalremoval of crabs(Bertness 1985) (Table 1). Organismswith small individual impacts can also have Anotherextremely poorly researched problem is the hugeecological effects,providing that they occur at suffi- way in whichthe persistence of theproducts or effectsof cientlyhigh densities over large areas, for sufficient peri- engineeringinfluence population, community and eco- ods of time. Burrowingmeiofauna and bogforming systemprocesses. If engineersmake long-livedartifacts, Sphagnummosses (Table 1) are good examples. Accu- thentheir effects will usually,also, be long lived. But mulatedSphagnum peat maypersist for hundreds to thou- ephemeralproducts can also have long-termimpacts. For sands of years afterthe deathof the livingmoss. example,faecal pelletsproduced by marinezooplankton Ecological engineersmay also enhanceand speed up (Table 1) decompose relativelyquickly, but not before large scale physicalprocesses, including geological ero- theyhave sunk into the deep ocean, removingnutrients sion and weathering(Yair and Rutin1981, Krumbeinand fromsurface waters for millenia. Dyer 1985, Hoskinet al. 1986). Examples of rock-eating A usefulthought experiment is to considertaking the snails and caterpillarsare listed in Table 1. Worldwide, engineers away and imaginingthe consequences. In but especially in the tropics,heavily undercutcoastal many cases theirimpacts are ephemeral,operating on cliffsof sedimentaryrock are apparentlybeing eroded by timescalesshorter than, or similarto, the lifetimeof the tidesand storms.In factthe process is greatlyaccelerated organismitself (e.g. the nests of small passerinebirds). by two groups of engineers,both with low per capita But in othercases, the engineersleave monumentswith effects,but very abundant.Cyanobacteria (Hyella spp.) impactsthat extend many lifetimes beyond their own - bore the rock and are food forchitons which rasp away mima mounds,termite nests, buffalowallows, beaver the rock to reach them,apparently speeding up coastal dams,peat, sedimentary rocks and so on. These persistent erosionby an orderof magnitudeor more(Krumbein and effects must greatly slow down rates of ecological Dyer 1985). Similarly,organisms whose shells, body change,and imposeconsiderable buffering and inertiaon partsand dead tissueshelp formsedimentary rocks, coal manyecological systemsand processes.Rates of decay, and soil (e.g. molluscs,diatoms and manyhigher plants) the 'half-lives'of the productsof ecological engineers, createstructures whose effectson ecosystemspersist for and theircontributions to population,community and eons. ecosystem stability,resistance and resilience (Pimm Engineeringimpacts are oftengreatest when the re- 1984) deservemuch more attention from theoretical and source flows that are modulatedare utilisedby many experimentalecologists. other species, or when the engineermodulates abiotic forcesthat affect many otherspecies. Not surprisingly, engineeringthat effects soils, sediments,rocks, hydrology,fire and hurricanesprovides some of themost strik- ing examples. We know of veryfew fieldmanipulation experiments

382 OIKOS 69:3 (11994)

This content downloaded on Mon, 11 Feb 2013 12:21:08 PM All use subject to JSTOR Terms and Conditions Evolutionaryeffects Questions Earlier, we distinguishedbetween engineeredartifacts We finishwith a haphazardlist of open questions. subjectto naturalselection as extendedphenotypes, and Are thereany ecosystemson earththat have not been 'therest' - by-productsof some otheractivity that are not physicallyengineered by one or more organismsto a themselvesdirectly subject to selection;water-filled un- significantdegree? A cautious,preliminary answer is no, gulate hoof printswere given as an example. Extended thereare not. We initiallythought that it would be diffi- phenotypeengineering, by definition,creates structures cult to identifyevidence of ecosystemengineering by or effectsthat directlyinfluence individual fitness (or biota in the open watersof oceans or large lakes, or in colony fitnessin social insects),for instancea beaver snow fieldsand ice packs,for instance. But theexamples dam, woodpeckerhole or termitemound. But theevolu- in Table 1 show that this predictionwas wrong. We tionaryeffects of extendedphenotype engineering, other currentlycannot identify any habitaton earththat is not 'accidental' engineering,and of organismsin the engi- engineeredin some way by one or more species. neered habitatupon the engineerare far fromstraight- How manyspecies (or whatproportion of species) in forward,and generallyunstudied. For example,the ex- various communitieshave a clearly definedand mea- tendedphenotype may be subjectto selectionfrom the surable impact as engineers?Is it 10%, 1% or 0.1%? physicalenvironment, implying no bioticfeedbacks. Or it What are the relativefrequencies of the five classes of may be subjectto selectionfrom polyphagous predators engineeringidentified in Fig. 1, say in termsof the (e.g. nest-robbingsnakes) thatare in no way dependent numbersof species actingas engineers?Cases 5 and 6 are upon the engineeredhabitat for theirexistence. On the presumablyrather rare; but how muchrarer are theythan otherhand, engineering might generate habitats with spe- the otherstypes of engineering?Is the predominanceof cies populationsthat, ultimately, feed back positivelyor examplesinvolving burrowing animals in case 4 real (we negativelyupon the engineers,via predation,disease, could easily have includedstill more examples) or is it competition,or theinvasion of additionalspecies of engi- because thisform of engineeringis particularlyeasy to neers. see? Some engineeringundoubtedly has had evolutionary Are themost physically structured ecosystems (or sub- effectson otherorganisms. One of the best documented systems,e.g. soil or sediments)the ones in whichengi- examplesin thefossil record is a decline,from the Devo- neers are most important?How much of the structure nian onwards,in the diversityof immobilesuspension have theycreated and modified? feedersliving on softmarine substrata, as mobile taxa How manyother species are impactedby engineersin diversified(Thayer 1979). Thayerattributes these major any ecosystem?What happensto species richnessif we changes in the structureof marinebenthic communities remove or add engineers?How much of the effectsof to theevolution of 'biological bulldozers'- bioturbators keystonespecies are due to engineeringversus trophic or engineers- thatdisturb sediments (see above), fouling, effects?Earlier, we speculatedthat few keystoneeffects overturningand buryingimmobile suspension feeders, are purelytrophic; is this hypothesiscorrect? How do which are now largely confined to hard substrates. engineeringand trophicrelations interact? Thayer also speculatesthat by increasingthe turnover Should conservationistsand naturereserve managers rateof nutrientsin sedimentsof continentalshelves, bull- pay moreattention to therole of ecological engineersin dozersmay have contributedto theMesozoic diversifica- maintainingecosystem integrity, or do managerslargely tion of phytoplankton(coccoliths, diatoms and dinofla- knowwhich the important species are,without having put gellates) and, via trophiclinkage, to diversificationof a name to the idea, or withouthaving recognisingthe zooplankton(radiolaria and foraminifera). commonthemes identified in thispaper? An intriguing,but rarelyconsidered problem is the How should we model engineering?The biological degree to which engineeringby othertaxa mightsimi- detailsin each case will be complicated,and thereare at larly have changed major patternsin the radiationand least fivekinds of engineering;but there are also several extinctionof earth's biota. To the extentthat engineers sortsof interspecificcompetition, various ways of beinga shape and modifymost, possibly all, habitatson earth herbivoreand a richcatalogue of enemy-victiminterac- (see below), and giventhe tritebut trueobservation that tions,none of whichhas stoppedtheoreticians from de- all organismsare adaptedto theirenvironment, engineer- velopingappropriate families of relativelysimple models ing in some formor othermust have driven,or contrib- to understandand to predictthe dynamics of suchinterac- uted to, the evolutionof myriadsof species. But the tions. There is, in principle,no reason why we cannot extentto whichmajor patternsof evolutionmight have writedown the equation: been differentif some types of ecological engineering had not evolved,or had takena differentform, is almost dmayfly/dbeaver= F(xy,z), entirelyunknown. wheremayfly populations respond to changesin beaver numberson long time-scales,and wherethe responseis influencedby variouskey variables,including feedbacks

OIKOS 69:3 (1994) 383

This content downloaded on Mon, 11 Feb 2013 12:21:08 PM All use subject to JSTOR Terms and Conditions to beaver fromother components in theengineered habitat.Interesting theoretical questions centre on thegener- References ation timeof the engineer,the half-lifeof whateverit is Anderson,F 0. andKristensen, E. 1991.Effects of burrowing of non-engi- macrofaunaon organicmatter decomposition in coastal ma- thatis engineered,the rate of restoration rinesediments. - Symp. zool. Soc. Lond.63: 69-88. neeredhabitat, the generation times of impactedspecies, Anderson,R. A. 1992.Diversity of eukaryotic algae. - Biodiv. and theirvarious interactions. There are intriguingprob- Conserv.1: 267-292. lems of nested time-scales,delayed responses,donor- Andrewartha,H. G. andBirch, L. C. 1954.The distribution and and so on, abundanceof animals.- UniversityofChicago Press, Chi- control,long chains of indirectinteractions cago. thatmight usefully be exploredusing relativelysimple - and Birch,L. C. 1984.The ecologicalweb. Moreon the models. distributionand abundanceof Animals.- Universityof Extendingthese arguments, there is no reasonin prin- ChicagoPress, Chicago. A notbe Arrigo,K. R., Sullivan,C. W. and Kremer,J. N. 1991. ciple whyprocesses driven by engineeringshould bio-opticalmodel of Antarctic sea ice.- J.Geophys. Res. 96 coupled to therich diversity of trophiclinkages to create C6: 10581-10592. not simplydescriptions and models of foodwebs (e.g. Askew,R. R. 1975. The organisationof chalcid-dominated Pimm et al. 1991), but of interactionwebs, thatmore parasitoidcommunities centred upon endophytic hosts. - In: Price,P. W.(ed.), Evolutionary strategies of parasitic insects accuratelyreflect interactions in communitiesand eco- andmites. Plenum, New York,pp. 130-153. systems.The conceptualframework brought together by Basnet,K., Likens,G. E., Scatena,F N. andLugo, A. E. 1992. Carpenter(1988) underthe title of 'complexinteractions' HurricaneHugo: damage to a tropicalrain forest in Puerto should clearlyembrace engineering as one component. Rico.- J.Trop. Ecol. 8: 47-55. Begon,M., Harper,J. L. andTownsend, C. R. 1990.Ecology. Intriguingly,once theneed to defineand studyinterac- Individuals,populations and communities,Second ed. - tion webs is recognised,it also becomes apparentthat BlackwellScientific, Boston. ecological engineeringas we have definedit is onlypart Bertness,M. D. 1984a.Habitat and community modification by of thepicture. We have focusedon physicalstate changes an introducedherbivorous snail. - Ecology65: 370-381. - 1984b. Ribbed musselsand Spartinaalternfilora production wroughton biotic and abiotic materialsby organisms. in a New Englandsalt marsh. - Ecology65: 1794-1807. But physicalengineering is not theonly formof ecosys- - 1985. Fiddlercrab regulationof Spartina alterniflorapro- tem engineeringthat organisms carry out. Chemical and ductionon a NewEngland salt marsh. - Ecology66: 1042- transportengineering are twoobvious other forms that we 1055. Bloom,A. L. 1978. Geomorphology.- Prentice Hall, New considerwill conceptuallyfit into the same generalclas- York. sificationscheme. For brevity,we have not examined Brown,J. H. andHeske, E. J. 1990.Control of a desert-grass- themhere. landtransition by a keystonerodent guild. - Science250: Last, butby no meansleast, what new insightswill the 1705-1707. Buynitskiy,V. K. 1968. The influenceof microalgaeon the concept of engineeringbring to ecology? This review structureand strength ofAntarctic sea ice. - Oceanology8: attemptsto define and classify the phenomenon.But 771-776. definitionand classificationare merelya small begin- Carpenter,S. R. (ed.). 1988.Complex interactions in lake com- ning,not an end. Are theremajor patterns in thedistribu- munities.- Springer,New York. - andLodge, D. M. 1986.Effects of submerged macrophytes tionand abundanceof organismsthat might be explained, on ecosystemprocesses. - Aquat.Bot. 26: 341-370. at least in part, by ecosystemengineering? What are -, Kitchell,J. F, Hodgson,J. R., Cochran,P. A., Elser,J. J., they?What predictionsmight we make thatwould not Elser,M. M., Lodge,D. M., Kretchmer,D., He, X. andvon have been made withoutthe conceptualframework pro- Ende,C. N. 1987.Regulation of lakeprimary productivity byfood web structure.- Ecology 68: 1863-1876. videdhere, or somethingakin to it? We do not,currently, Christensen,N. L. 1985.Shrubland fire regimes and their evolu- knowthe answers to thesequestions. But if thenotion of tionaryconsequences. - In: Pickett,S. T. A. andWhite, P. S. organismsas ecosystemengineers results simply in an (eds),The ecology of natural disturbance and patch dynam- accumulationof 'just-so' stories,it will not have been ics. AcademicPress, Orlando, FL, pp. 85-100. Cox, G. W. and Gakahu,C. G. 1985. Mima moundmicro- particularlyuseful. topographyand vegetationpattern in Kenyansavannas. - J.Trop. Ecol. 1: 23-36. Acknowledgements- This workwas supportedby theMary - andGakahu, C. G. 1986.A latitudinaltest of thefossorial FlaglerCary Charitable Trust, the US-Israel Binational Science rodenthypothesis of Mima moundorigin. - Z. Geomor- Foundation,and the NERC Centrefor Population Biology, UK. phologie30: 485-501. Numerouscolleagues have allowed us tosharpen and refine our -, Lovegrove,B. G. and Siegfried,W. R. 1987.The small ideasabout engineering by discussing problems with them, and stonecontent of mima-likemounds in theSouth African havegenerously provided us withexamples and references; to Cape region.- Catena14: 165-176. all of themwe expressour sincere thanks. Contribution to the Daily,G. C., Ehrlich,P. R. and Haddad,N. M. 1993.Double programmeof theInstitute of EcosystemStudies. Publication keystonebird in a keystonespecies complex. - PNAS 90: 171 fromthe Mitrani Center. 592-594. Dawkins,R. 1982. The extendedphenotype. - OxfordUni- versityPress, Oxford. Dayton,P. K. 1971.Competition, disturbance,, and community organisation:the provisionand subsequentutilization of spacein a rockyintertidal community. - Ecol. Monogr.41: 351-389. Dublin,H. T., Sinclair,A. R. E. and McGlade,J. 1990.Ele-

384 OIKOS 69:3 (1994)

This content downloaded on Mon, 11 Feb 2013 12:21:08 PM All use subject to JSTOR Terms and Conditions phantsand fireas causes of multiplestable states in the Krumbein,W. E. andDyer, B. D. 1985.This planet is alive- Serengeti-Marawoodlands. - J. Anim.Ecol. 1990: 1147- weatheringand biology,a multi-facetedproblem. - In: 1164. Drever,J. I. (ed.), The chemistryof weathering.D.Reidel Dunbar,R. B. and Berger,W. H. 1981. Fecal pelletflux to PublishingCo.,, pp. 143-160. modembottom sediment of Santa BarbaraBasin (Cali- Lal, R. 1991.Soil conservationand biodiversity. - In: Hawks- fornia)based on sedimenttrapping. - Geol. Soc. Amer.Bull. worth,D. L. (ed.),The biodiversity ofmicroorganisms and pt.192: 212-218. invertebrates:its rolein sustainableagriculture. CAB In- Elmes,G. W. 1991.Ant colonies and environmentaldisturb- ternational,Wallingford, pp. 89-103. ance.- Symp.zool. Soc. Lond.63: 1-13. Lawton,J. H. 1983. Plantarchitecture and the diversityof Elton,C. S. 1966.The pattern of animal communities. - Meth- phytophagousinsects. - Ann.Rev. Ent.28: 23-39. uen,London. - 1989. Food webs.- In: Cherrett,J. M. (ed.), Ecological Emmet,A. M. (ed.). 1979.A fieldguide to thesmaller British concepts.Blackwell Scientific Publications, Oxford, pp. Lepidoptera.- BritishEntomological and NaturalHistory 43-78. Society,London. - and Hassell,M. P. 1981.Asymmetrical competition in in- Estes,J. A. andPalmisano, J. F 1974.Sea-otters: their role in sects.- Nature289: 793-795. structuringnearshore communities. - Science 185: 1058- Likens,G. E. 1992.The ecosystem approach: its use and abuse. 1060. - EcologyInstitute, Oldendorf/Luhe. Facelli,J. M. andPickett, S. T. A. 1991.Plant litter: its dynamics - andBilby, R. E. 1982.Development, maintenance, and role and effectson plantcommunity structure. - Bot. Rev. 57: oforganic-debris dams in New England streams. - In: Swan- 1-32. son,F. J.,Janda, R. J.,Dunne, T. andSwanston, D. N. (eds), Finlayson,M. and Moser,M. (eds.). 1991. Wetlands.- In- Sedimentbudgets and routingin forestdrainage basins. ternationalWaterfowl and Wetlands Research Bureau, Ox- USDA ForestService General Technical Report PNWl41. ford. USDA ForestService, Pacific Northwest Forest and Range Fish,D. 1983.Phytotelmata: flora and fauna. - In: Frank,J. H. ExperimentalStation, pp. 122-128. andLounibos, L. P. (eds),Phytotelmata: terrestrial plants as Lopez, G. R. and Levinton,J. S. 1987. Ecologyof deposit- hostsfor aquatic insect communities. Plexus, Medford, NJ, feedinganimals in marine sediments. - Quart. Rev. Biol. 62: pp. 1-27. 235-260. Fowler,S. W.and Knauer, G. A. 1986.Role of large particles in Mazumder,A., Taylor,W. D., McQueen,D. J.and Lean, D. R. thetransport of elementsand organiccompounds through S. 1990.Effects of fishand planktonon lake temperature theoceanic water column. - Prog.Oceanog. 16: 147-194. andmixing depth. - Science247: 312-315. Gutterman,Y. 1982.Observations on thefeeding habits of the Meadows,P. S. 1991.The environmental impact of burrows and Indiancrested porcupine (Hystrix indica) and thedistribu- burrowinganimals - conclusionsand a model.- Symp.zool. tionof some hemicryptophytes andgeophytes in theNegev Soc. Lond.63: 327-338. deserthighlands. - J.Arid. Env. 5: 261-268. - andMeadows, A. (eds),1991 a. Theenvironmental impact of Hall,S. J.,Robertson, M. R., Basford,D. J.and Fryer, R. 1993. burrowinganimals and animal burrows. - ClarendonPress, Pitdiggingby the crab Cancer pagurus: a testfor long-term, Oxford. large-scaleeffects on infaunalcommunity structure. - J. - andMeadows, A. 1991b.The geotechnical and geochemical Anim.Ecol. 62: 59-66. implicationsof bioturbation in marine sedimentary ecosys- Hansell,M. H. 1993.The ecological impact of animal nests and tems.- Symp.zool. Soc. Lond.63: 157-181. burrows.- Funct.Ecol. 7: 5-12. Mohr,C. 0. 1943.Cattle droppings as ecologicalunits. - Ecol. Hawkins,B. A. 1992. Parasitoid-hostfood webs and donor Monogr.13: 276-298. control.- Oikos65: 159-162. Moloney,K. A., Levin,S. A., Chiariello,N. R. andButtel, L. Hedin,L. O., Mayer,M. S. andLikens, G. E. 1988.The effect of 1992.Pattern and scale in a serpentinegrassland. - Theoret. deforestationon organic debris dams. - Verh.Internat. Ve- Pop. Biol. 41: 257-276. rein.Limnol. 23: 1135-1141. Naiman,R. J.1988. Animal influences on ecosystemdynamics. Heth,G. 1991.The environmental impact of subterranean mole - BioScience38: 750-752. rats (Spalax ehrenbergi)and theirburrows. - Symp. zool. -, Johnston,C. A. andKelley, J. C. 1988.Alteration of North Soc. Lond.63: 265-280. Americanstreams by beaver. - BioScience38: 753-762. Holling,C. S. 1992. Cross-scalemorphology, geometry, and Neal,E. G. andRoper, T. J.1991. The environmental impact of dynamicsof ecosystems.- Ecol. Monogr.62: 447-502. badgers (Meles meles) and theirsets. - Symp. zool. Soc. Hoskin,C. M., Reed,J. K. andMook, D. H. 1986.Production Lond.63: 89-106. andoff-bank transport of carbonatesediment, Black Rock, Paine,R. T. 1969.A noteon trophic complexity and community southwestLittle Bahama Bank. - Mar.Geol. 73: 125-144. stability.- Am. Nat. 103: 91-93. Hunter,M. D. andPrice, P. W. 1992.Playing chutes and ladders: - 1979.Disaster, catastrophe, and local persistence of the sea heterogeneityand therelative roles of bottom-upand top- palm Postelia palmaeformis.- Science 205: 685-687. downforces in naturalcommunities. - Ecology 73: 724- Pickett,S. T. A. andWhite, P. S. (eds), 1985.The ecologyof 732. naturaldisturbance and patch dynamics. - Academic Press, Huntly,N. andInouye, R. 1988.Pocket gophers in ecosystems: Orlando. patternsand mechanisms. - BioScience 38: 786-793. Pimm,S. L. 1984.The complexity and stability of ecosystems. - Irlandi,E. A. andPeterson, C. H. 1991.Modification of animal Nature307: 321-326. habitatby largeplants: mechanisms by whichseagrasses - , Lawton,J. H. andCohen, J. E. 1991.Food web patterns and influenceclam growth. - Oecologia87: 307-318. theirconsequences. - Nature350: 669-674. Jones,C. G. andShachak, M. 1990.Fertilization of thedesert Reichelt,A. C. 1991.Environmental effects of meiofaunal bur- soil byrock-eating snails. - Nature346: 839-841. rowing.- Symp.zool. Soc. Lond.63: 33-52. Kitching,R. L. 1983.Community structure in water-filled tree- Rhoads,D. C. andYoung, D. G. 1970.The influence of deposit- holes in Europeand Australia- comparisonsand spec- feedingorganisms on sedimentstability and community ulations.- In: Frank,J. K. andLounibos, L. P. (eds),Phyto- trophicstructure. - J. Mar. Res. 28: 150-178. telmata:terrestrial plants as hostsfor aquatic insect commu- Richardson,D. M., Cowling,R. M., Bond,W. J., Stock, W. D. nities.Plexus, Medford, NJ, pp. 205-222. andDavis, G. W. in press.Links between biodiversity and Krebs,C. J.1985. Ecology. The experimental analysis of distri- ecosystemfunction: evidence from the Cape FloristicRe- butionand abundance,Third ed. - Harperand Row,New gion.- In: Davis, G. W. and Richardson,D. M. (eds), York.

OIKOS 69:3 (1994) 385

This content downloaded on Mon, 11 Feb 2013 12:21:08 PM All use subject to JSTOR Terms and Conditions Biodiversityand ecosystem function in Mediterranean-typeWessels, D. C. J.and Wessels, L.-A. 1991.Erosion of biogen- ecosystems.Springer-Verlag, Heidelberg. icallyweathered Clarens sandstone by lichenophagous bag- Ricklefs,R. E. 1984.Ecology, Second ed. - ChironPress, New wormlarvae (Lepidoptera; Pyschidae). - Lichenol. 23: 283- York. 291. Shachak,M., Jones,C. G. and Granot,Y. 1987.Herbivory in West,N. E. 1990.Structure and functionof microphyticsoil rocksand the weathering of a desert.- Science236: 1098- crustsin wildland ecosystems of arid to semi-arid regions. - 1099. Adv.Ecol. Res. 20: 180-223. Southern,H. N. (ed.). 1964.The handbook of British mammals, Whicker,A. D. and Detling,J. K. 1988. Ecologicalconse- Firsted. - BlackwellScientific, Oxford. quencesof prairie dog disturbances. - BioScience 38: 778- Southwood,T. R. E., Brown,V. K. andReader, P. M. 1979.The 785. relationshipof plantand insect diversities in succession.- Wiens,J. A. 1976. Populationresponses to patchyenviron- Biol. J.Linn. Soc. 12: 327-348. ments.- Annu.Rev. Ecol. Syst.7: 81-120. Tansley,A. G. 1949.Britain's green mantle. - GeorgeAllen and de Wilde,A. W.J. 1991. Interactions inburrowing communities Unwin,London. and theireffects on thestructure of marineecosystems. - Thayer,C. W. 1979.Biological bulldozers and the evolution of Symp.zool. Soc. Lond.63: 107-117. marinebenthic communities. - Science 203: 458-461. Williamson,M. 1972.The analysisof biological populations. - Thompson,L., Thomas,C. D., Radley,J. M. A.,Williamson, S. EdwardArnold, London. andLawton, J. H. 1993.The effect of earthworms and snails Wood,T. G. and Sands,W. A. 1978.The roleof termitesin in a simpleplant community. - Oecologia 95: 171-178. ecosystems.- In: Brian,M. V. (ed.),Production ecology of Townsend,D. W.,Keller, M. D., Sieracki,M. E. andAckleson, antsand termites. Cambridge University Press, Cambridge, S. G. 1992.Spring phytoplankton blooms in theabsence of pp. 245-292. verticalwater column stratification. - Nature 360: 59-62. Yair, A. and Rutin,J. 1981. Some aspectsof the regional Wallace,G. T.,Jr., Mahoney, 0. M.,Dulmage, R., Storti,F and variationin theamount of availablesediment produced by Dudek,N. 1981.First-order removal of particulate alumini- isopodsand porcupines,northern Negev, Israel. - Earth umin oceanicsurface water. - Nature293: 729-731. Surf.Proc. Land. 6: 221-234. Walsh,G. B. and Dibb, J. R. (eds.). 1954. A coleopterist's handbook.- The AmateurEntomologists' Society, London.

386 OIKOS 69:3 (1994)

This content downloaded on Mon, 11 Feb 2013 12:21:08 PM All use subject to JSTOR Terms and Conditions