Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
CEC Research | https://doi.org/10.21973/N3766T Fall 2017 1/7
Habitatpreferenceandspeciesinteractionsofthedesertwoodrat(Neotomalepida)
intheMojaveDesert
BeccaCosmero1,EmilyFieberling2,KatherineMarlin3,RicardoRuiz4,RizzieVermont4,andLaraVolski5
1UniversityofCalifornia,Merced;2UniversityofCalifornia,LosAngeles;
3UniversityofCalifornia,SantaBarbara;4UniversityofCalifornia,SantaCruz;5UniversityofCalifornia,Berkeley
ABSTRACT
Ecosystemengineersoftenhavecascading impactsontheecosystems inwhichtheyoccur.Understandingthefactorsthatcontributetotheirhabitatselectioncan provide valuable information on how ecosystems function. Our studyinvestigatesthehabitatdistributionofdesertwoodrats(Neotomalepida)intheEasternMojave Desert.We show that there is a difference in desertwoodratabundance among habitats, and found that specific aspects of theseenvironmentsareassociatedwithwoodrats.Theseaspects includegrasscover,shrubrichness,andespeciallysubstratearchitecture.Our findings indicatethatwoodratsmayfacilitateplantproductivitywithinthisaridenvironment.
INTRODUCTION
Preferentialhabitatselectionisoneofthedrivingfactorsthatallowspeciestocoexist (Rosenzweig 1981). Habitatquality, determined by resourceavailability,isaprimarydriverofhabitatselection (Orians and Wittenberger1991).Agreaterunderstandingofwhatleads an organism to choose a specifichabitatcanuncoverecologicaldynamicsand interactions between species. Thisis especially the case for ecosystemengineers that modify their environment
and in turn support biodiversity. Forexample, American pikas (Ochotonaprinceps)areecosystemengineersinalpineenvironments because their burrowsincreasethelevelsofnutrientsavailableto plants (Aho et al. 1998). Thus, theplaces that pikas prefer to live wouldtherefore be associated with thedistribution of alpine vegetation. Inother words, understanding whatdetermines the habitat preference of asingle organism can reveal what driveshabitat selection of other species. This
CEC Research | https://doi.org/10.21973/B8R93X 2/7
in turn can lend greater understandingofthespatialdistributionofspeciesandresourceavailabilitywithinthathabitat.Abioticallyextremeenvironmentsthat
contain limited resources, such asdeserts, can provide insight into thegreater ecological impacts associatedwithhabitatselectionofkeyorganisms.Ecosystemengineersplaymajorrolesinthe distribution of patchy and limitedresources. Rodents that create soildisturbancesinaridregions,suchasthedesert woodrat (Neotoma lepida), areoften classified as ecosystemengineersbecausecollectedorganicmaterialsandfecal detritus surrounding desertwoodrat middens promote nitrogenmineralization that in turn promoteslocal biodiversity (Whitford andSteinberger 2010). Reciprocally, desertwoodrats are also heavily reliant ontheir habitat. For example, desertwoodrat survivorship was found to beheavily correlated with the success orfailure of desert annual and perennialplant production (Smith 1995). WhenMojave yucca (Yucca schidigera) failedtosetseedduringdroughtyears,desertwoodrats also failed to reproduce(Smith 1995). Thus, vegetation caninfluencewoodratsurvivorship,andthefact that woodrats have been cited asecosystem engineers indicates thatwoodrats may also be responsible forthe composition of their surroundingvegetation. Nevertheless, the specificcompositionof thesesurroundingplantcommunitieshasyettobedocumented.In this study, we investigated howwoodrat habitat preference correlateswithdifferencesintheircommunities.
In order to examine the reciprocalrelationship between woodrat habitatpreference and the associatedsurroundingvegetation,weinvestigatedwhether (1) woodrats prefer specifichabitats andmicrohabitats, (2) middensubstratevegetationdifferedfromnon-midden vegetation, and (3) woodratpresencedriveschanges in surroundingvegetation. Together, our study helpscontributetoabetterunderstandingofthe underlying mechanisms drivinghabitat preference of desert woodrats,andinturn,thepotentiallylargeimpactof desert woodrats on desertvegetation.Furtheringourknowledgeofthis relationshipwill allow us to betterunderstand the dynamics of desertecosystems at large, and what enablesdesert natives to persist in this harsh,resourcelimitedenvironments.
METHODS
Studysite
We conducted our study in threeseparate sites in the University ofCalifornia Sweeney Granite MountainsDesertResearchCenter(UCGMR)intheEastern Mojave Desert (24°47' N,115°43' W). This area is a transitionzone and contains flora and faunacharacteristic of the Great Basin,Sonoran,andMojaveDeserts.Thereare14 species of rodents in the area and502speciesofplants.Atroughly1200melevation,ourstudysitesdidnotbreachdusky-footed woodrat (Neotomafuscipes) territory and instead featuredonly desert woodrat middens. Amongthree sites, we characterized eachaccording to their predominant
CEC Research | https://doi.org/10.21973/B8R93X 3/7
vegetation.“MixedCholla/YuccaScrub”was abundant in Mojave yucca (Yuccaschidigera) and buckhorn cholla(Cylindropuntia acanthocarpa). “MixedYucca Scrub,” was abundant in yuccabuthad fewerbuckhorn cholla. “MixedAcaciaScrub”had few tonoyuccaandcholla, but was abundant in cat-clawacacia(Senegaliagreggii).Allthreesiteshadavaryingnumberofcreosotebush(Larreatridentata).
Shrubrichness
To test whether desert woodratdensities correlated with richness ofsurrounding shrubsweused five50m2macro-plotswithineachsite.Toreducebias, we randomly determined thedistance between macro-plots. Eachmacro-plot was divided into fourtransectstogivefive10mx50mbandsandwithinthreebandswecountedthetotalnumberofyucca,cholla,creosote,and acacia individuals. We used thisdatatoestimatetheabundanceofeachplant species. Additionally, a belttransectwasperformedalongthex-axisofeachmacro-plot inordertomeasurespeciesrichnessofallplantsinadditiontothepreviouslysurveyedyucca,cholla,creosote,andacaciaindividuals.
Middendata
Toassesswoodrathabitatpreference,we surveyed every woodrat middenwithin each macro-plot. For eachmiddenwe encounteredwithinmacro-plots,werecordeditsposition,theplantspecies that hosted the midden(hereafterreferredtoassubstrate),andthe midden construction material. Inorder to test if woodrat presence was
associated with shrub richness, wemeasured the following variables forevery substrate: surrounding plantspecies richness, percent grass coverundershrubs,andplantarchitecture(asnumber of clones). We defined anindividual yucca clone as a newrhizomatous growth and an individualcholla clone as a main branch at thebase. Then, we chose the next closestnon-midden substrate of the samespeciesofsimilarheight,andmeasuredthesamevariables.
Statisticalanalyses
We used a one-way ANOVA to testdifferences in midden abundanceamong sites, followed by Tukey’s HSDpost-hoc test to distinguish specificdifferences among sites. To determinewhat habitat characteristics wereassociatedwithwoodratabundance,weused linear regressions to examinerelationships between middenabundance and shrub species richness,shrubdensity,proportionofcholla,andproportionofcreosote.We conducted twowayANOVAs to test
the relationships of midden presenceand substrate species on surroundingshrubrichness,percentgrasscover,andshrub architecture. To determinespecificdifferenceamongtheeffectsofsubstrate specieson surrounding shrubrichness, understory percent grasscover, and substrate architecture weused Tukey HSD post-hoc tests. Acaciawasexcluded from theANOVAanalysisof number of clones because acaciadoes not exhibit the same growthpatternsasyuccaandcholla.
CEC Research | https://doi.org/10.21973/B8R93X 4/7
Lastly we used ANCOVA to comparenumber of yucca clones andsurroundingspeciesrichnessbymiddenpresence. All statistical analyses wereconductedusingJMP13.0.0.
RESULTS
Mixed cholla scrub predominantlyconsisted of Mojave yucca (30%),buckhorn cholla (60%), and desertcreosotebush(10%).Mixedyuccascrubwas composed ofMojave yucca (41%),buckhorn cholla (6%), and desertcreosotebush(53%).Mixedacaciascrubconsistedofdesertcreosotebush(57%)andcatclawacacia(43%).There were significantly more
middens inthemixedchollascrubthananyotherhabitat (p<0.001; Figure1).Mixed cholla scrub also had highershrub richness and total shrub densitythan the other two habitats. Middendensity was positively correlated withshrubrichness(p<0.01;R2=0.8;Figure2A) and shrub density (p < 0.001; R2 =0.85; Figure 2B). Midden abundanceincreased as proportion of chollaincreased(p<0.0001;R2=0.8;Figure2C)butdecreasedasproportionof
Figure1.Meannumberofmiddens±1S.E.perhabitat. Difference Tukey HSD: Mixed ChollaScrub (A),MixedCholla/Yucca Scrub (B),MixedAcaciaScrub(B).
creosoteincreased(p<0.001,R2=0.99;Figure 2D). Also, 76% of middenscontainedchollaasmiddenmaterial. Midden presence was positively
correlated with shrub richness ofsurrounding vegetation (p < 0.002;Figure 3A). Shrub richness wassignificantly higher in yucca than incholla,butacaciashrubrichnessdidnotsignificantly differ from either yucca orcholla (p < 0.04; Figure 3B). Middensubstrates also had greater percentgrasscoverthannon-middensubstrates(p < 0.03; Figure 3C). Grass percentcover was highest under acacia,followed by cholla, then yucca. Acaciawas significantly different from chollaand yucca, and cholla and yucca werenot significantly different from eachother (p < 0.0001; Figure 3D). Numberof clones had the strongest correlationwith midden presence (p < 0.0001;Figure 3E). The number of clones wassignificantly higher in yucca than incholla(p<0.0001;Figure3F).Woodrat presence is more strongly
associatedwithahighnumberofclonesinyuccathanincholla(p< .001;Figure4). Yuccas with woodrat middens hadalmost double the number of clonesthanyuccaswithoutwoodratmiddens.In yuccawithmiddens, as number of
clonesincreased,shrubrichnessslightlyincreased.Inyuccawithoutmiddens,asnumber of clones increased, shrubrichness slightly decreased. However,these results were not significant(P(midden presence) = 0.09, P(middenpresence*no.ofclones)=0.15,P(no.ofclones) = 0.76. R2 = 0.045. LinearregressionR2pres=0.02,R2abs=0.02;Figure5.)
P"<"0.0001"A"
B"
C"
CEC Research | https://doi.org/10.21973/B8R93X 5/7
Figure 2. The relationship between habitat characteristics and midden density. Graphs demonstrateresponse of midden density to shrub richness (A), shrub density (B), proportion of cholla (C), andproportionofcreosote(D).
Figure3.Meanvegetationcharacteristics±1S.Eby midden presence (A, C, E) and middensubstrate species (B,D, F). Barsnot connectedby the same letterswere significantly differentafterTukey’sHSDpost-hoctests.
Figure4.Relationshipofmiddenpresenceandsubstrate species with substrate architecture.Dark and lightbars representmiddenpresenceand absence, respectively. Bars representmeans±1S.E.Barsnotconnectedbythesameletterswere significantlydifferentafterTukey’sHSDpost-hoctests.
CEC Research | https://doi.org/10.21973/B8R93X 6/7
Figure 5. Relationships between yuccaarchitecture and shrub richness for middensand non-middens. Yucca architecture wasmeasured by number of clones. Middenpresence is represented by blue and middenabsenceisrepresentedbyred.
DISCUSSION
Our research suggests that woodrathabitat preference influences thedistribution of desert vegetation bypromotinggrowtharoundmiddens.Wefoundhigherwoodratmiddendensitiesin areaswith the highest proportion ofcholla, demonstrating that woodratsexhibit habitat preference. This findingcomplements the fact that woodratsurvivorshipisgreaterinchollamiddensbecause it confers more protectionagainst predators (Smith 1995). Thischolla-woodrat relationship is furtherunderscored by the high proportion ofcholla used as midden constructionmaterial. The fact that the speciesrichnessandthepercentcoverofgrassaround middens were higher than inareasthatlackedmiddensindicatesthatthere is clear relationship betweendesertwoodratsanddesertvegetation.We suggest a manipulation study toconfirmacause-and-effectrelationship,butthesefindingsneverthelessendorse
the desert woodrat’s position as anecosystemengineer.There were also habitats that desert
woodratsselectedagainst.Forexample,midden density was negativelycorrelatedwith proportion of creosote.Habitats with a greater proportion ofcreosotewere also less dense and hadlowerdiversity.Thisrelationshipmaybedue to the fact that creosote hasallelopathic effects on surroundingvegetation and avoids growing aroundcompeting root systems (Brisson andReynolds 1994). Since woodrats preferdenseareasthatprovideamplecover,itis appropriate that a woodrat wouldavoid this habitat (Bonaccorso andBrown1972).There were strong differences in the
biotic communities around middenswhen compared to non-middensubstrates. Midden presence wasstrongly linkedwithgreaterdiversityofsurrounding shrubs, grass cover, andarchitectural complexity, most notablyin yucca. Since woodrats promotenitrogen mineralization, it is quitepossible that midden presence isresponsible for the correspondinglyhigher diversity and density ofsurrounding vegetation. Nevertheless,the association of woodrat middenswithgreatershrubdiversityandpercentgrass cover could be confounded byinherent differences in the threehabitatswesurveyed. It isalsounclearwhetherwoodratpresenceallowsyuccato live longer and thus generate moreclones, or whether woodrats simplyprefer to build theirmiddens in yuccasthat are more architecturally complex.Yet the relationship betweenwoodrats
CEC Research | https://doi.org/10.21973/B8R93X 7/7
and yucca shows promise becausewoodrats pass down their middens forseveral generations (Dial1988)and thenumber of clones in yucca increaseswithage(LaPre1979).Desert woodrats exhibited clear
habitat preference and there weredifferences between the vegetationaround middens versus non-middens.This demonstrates that desertwoodrathabitat preference is biotically drivenand there are strong interspecificrelationships between woodrats anddesert vegetation. Desert woodrathabitat preferencewas associatedwithshrub density and diversity, but thispreference was driven by theproportion cholla. Our findings furtherillustrate the complexity of desertwoodrathabitatselectioninaresource-deficient environment. There are likelyotherspecies interactionsthatwehavenot accounted for, but this studyportraysabaselineunderstandingofthedynamic relationship woodrats havewith desert vegetation. Our resultsindicate that shrub diversity andabundancemaydrivemiddenpresencewhich in turncanhaveecosystem levelconsequences. These findings furtherclarifywhatdriveshabitatselection,andwhatenablestheseorganismstosurviveinaridenvironments.
ACKNOWLEDGEMENTS
This work was performed at theUniversity of California’s SweeneyGranite Mountains Desert ResearchCenter,doi:10.21973/N3S942.
REFERENCES
Aho,K.,N.Huntly,J.Moen,andT.Oksanen.1998Pikas(Ochotonaprinceps:Lagomorpha)asallogenicengineersinanalpineecosystem.Oecologia,114:405–409.
Bonaccorso,F.J.andJ.H.Brown.1972.Houseconstructionofthedesertwoodrat,Neotomalepidalepida.JournalofMammalogy53(2):283–288.
Brisson,J.andJ.F.Reynolds.1994.Theeffectofneighborsonrootdistributioninacreosotebush(Larreatridentata)population.Ecology75:1693–1702.
Dial,K.P.1988.ThreesympatricspeciesofNeotoma:dietaryspecializationandcoexistence.Oecologia76(4):531–537.
LaPre,L.F.1979.PhysiologicalecologyofYuccaschidigera.Riverside,CA:UniversityofCalifornia.117p.Dissertation.
Orians,G.H.andJ.F.Wittenberger.1991.Spatialandtemporalscalesinhabitatselection.TheAmericanNaturalist137:29–49.
Rosenzweig,M.L.1981.Atheoryofhabitatselection.Ecology62(2):327–335.
Smith,F.A.1995.Dencharacteristicsandsurvivorshipofwoodrats(Neotomalepida)intheEasternMojaveDesert.TheSouthwesternNaturalist40(4):366–372.
Whitford,W.G.andY.Steinberger.2010.Packrats(Neotomaspp.):Keystoneecologicalengineers?JournalofAridEnvironments74:1450–1455.