POPULATION ECOLOGY

Effects of Abiotic Factors on the Distribution and Activity of theInvasive Argentine Ant (Hymenoptera Formicidae)

KATHLEEN G HUMAN STUART WEISS ANDREW WEISS BENET SANDLER AND

DEBORAH M GORDON]

Department of Biological Sciences Stanford University Stanford CA 94305-5020

Environ Entomol 27(4) 822-833 (1998)ABSTRACT The Argentine ant Linepithema humile (Mayr) has spread worldwide often deci-mating native ant populations and other arthropod species in invaded areas It is not known whatabiotic and biotic conditions limit its distribution We investigated the distribution of the Argentineant in the Jasper Ridge Biological Preserve in northern California a nature preserve that has beenpartially invaded by these ants Canonical correlation analysis showed that the Argentine ant is mostlikely to occur near the edges of the preserve which are next to disturbed areas and in low-elevationareas Native ant species are associated with higher-elevation areas farther from the preserve edgeDistance to surface water and insolation were less important in predicting Argentine ant distributionThis suggests that dispersal from disturbed areas strongly determines the spread of the invasion Weexamined how the daily activity patterns of Argentine ants and several native ant species dependon soil temperature air temperature and relative humidity There was considerable overlap inactivity times and there were no values of any of the abiotic factors measured in which native antspecies were more likely to be active than Argentine ants Because the Argentine ant is usually activeat the same times as any native species most native species in invaded areas are likely to encounterthe invaders frequently Diet overlap was high with most native ant species except for seed-eatingspecies Overlap in activity times may intensify both exploitative and interference competitionbetween the Argentine and native ant species

KEY WORDS Linepithema humile colony activity foraging competition invasion

STUDIES OF COMMUNITlES in flux such as those subject toa biological invasion can elucidate the roles of abioticand biotic factors in structuring communities TheArgentine ant Linepithema humile (Mayr) has in-vaded many areas of the world displacing native antspecies and other invertebrates where it invades (Fos-ter 1908 Haskins and Haskins 1965 1988 Crowell1968 Erickson 1971 Tremper 1976 Ward 1987 DeKock 1990 Cole et al1992) It has been suggested thatextremely successful biological invaders such as theArgentine ant generally have broad niches (Elton1958 Baker and Stebbins 1965 Ehrlich 1986 Holdgate1986 Porter and Savignano 1990) If broad environ-mental tolerance allows an invasive species to be ac-tive in the same range of abiotic conditions as a nativeone competitive or predatory interactions may lead tothe displacement of native species during an invasion(Elton 1958 Fausch and White 1981 Ward 1987Banks and Williams 1989 Porter and Savignano 1990DAntonio and Vitousek 1992 Human and Gordon1996 1997)

Abiotic factors may affect the distribution of theArgentine ant In the Central Valley of California therange of the Argentine ant is probably limited byproximity to water (Ward 1987 Holway 1995) An antspecies activity patterns and range may depend in

I To whom all correspondence should be addressed

part on its size and body color which affect rates ofheating and desiccation (Casey 1976 Tremper 1976)Behavioral strategies may affect tolerance of abioticconditions (Marsh 1988) and the physical structure ofnests may modulate temperature and humidity insidethe nest (Andrews 1927 Brian 1973) Argentine antworkers are small relative to those of many species ofnative ant and may therefore be more vulnerable todesiccation (Tremper 1976)

The spread of the Argentine ant has been monitoredsince 1993 in the Jasper Ridge Biological Preserve inCalifornia a reserve of 482 ha surrounded by residen-tial and agricultural development The invasion is pro-ceeding at rates up to 300 myr accompanied by adrastic decline in populations of native species andchanges in the distributions of many other arthropods(Human and Gordon 1996 1997) Both interferenceand exploitative competition seem to be important inthe displacement of native ant species from invadedareas

Here we relate the local distribution of the Argentineant in the Jasper Ridge Biological Preserve to severalabiotic factors Ant species ranges may be limited byelevation proximity to water or minimum annual tem-perature (Davidson 1977 Bernstein 1979 Cushman1993) We consider the effects of distance to the nearestsource of water distance to the nearest edge of thepreserve and thus to disturbed areas elevation and in-

0046-225X980822-0833$02000 copy 1998 Entomological Society of America

August 1998 HUMAN ET AL DISTRffiUTION AND ACTIVITY OF ARGENTINE ANT 823

solation on Argentine ant distribution We then comparethe daily activity patterns and diets of the Argentine antand several native ant species The daily activity patternsof ant colonies may depend on abiotic factors such as soiltemperature and relative humidity (Fielde 1904Talbot1943 Brian 1964 Marsh 1988 VepsaIainen and Savol-ainen 1990 Dean 1992) and also on interactions withcompetitors prey or predators (Brian 1964Lynch et al1980 Briese and Macauley 1980 Gordon 1988 Savol-ainen and VepsaIainen 1988Andersen 1992Dean 1992Andersen and Patel 1994)We relate activity patterns to4 abiotic factors (time of day soil temperature air tem-perature and relative humidity) and calculate the over-lap in diet of Argentine ant and native ant species Fi-nally we investigate whether abiotic conditions maylimit the invasion of the Argentine ant and consider howtemporal and resource overlap may lead to competitionbetween the invasive and native ant species

Materials and Methods

Study Site This study was conducted at the 482-haJasper Ridge Biological Preserve in northern Califor-nia (San Mateo County 1220 13 Wand 3 24 N60-120 m in elevation) Vegetation types found at thepreserve include serpentine grassland annual grass-land redwood forests chaparral evergreen forest oakwoodland wetland and riparian systems (specieslisted in Sawyer and Keeler-Wolf 1995) Argentineants have invaded =30 of the area of Jasper Ridgeprimarily along the edge of the preserve Jasper Ridgeis surrounded by low-density residential areas andagricuturalland most of which has already been in-vaded by Argentine ants

Ant Species Distributions Between May 1993 andJanuary 1996we surveyed all ofJasper Ridge Biolog-ical Preserve for ants 3 times yearly Surveys wereconducted in May (following the end of the rainyseason) September (at the end of the dry summer)and January (during the rainy season) For ant sur-veys a map of the entire preserve was overlaid with a100-m grid By using survey points at the center ofeach hectare we recorded the presence or absence ofnative ants and Argentine ants For 5 min we searchedvisually for ants within a 20-m-radius circle of thesurvey point If no ants were found we left honeytraps consisting of 40-ml vials filled with 10 ml ofhoney and collected them 24 h later Details of thesurvey method and a test of its accuracy are describedelsewhere (Human and Gordon 1996) In the Sep-tember 1993 January 1994 and May 1994 surveysnative ant species were identified only as not Argen-tine ants but thereafter they were identified to genusIn each time period 296-318 ha were surveyed A fewsurvey points could not be reached during particulartimes of year because of flooding or poison oak (Tox-icodendron diversilobum Jepson) Ant distributiondata were entered into a geographical informationsystem (ArcView GIS 1996)

For analysis ant distribution data for each hectarewere converted into a measure of incidence the pro-portion of all surveys in which ants of a particular

genus were found For example ifArgentine ants weredetected in 7 of 9 surveys Argentine ant incidence inthat hectare was 078 Incidence was calculated for allant species sampled in each hectare Data from the 3earliest surveys in which native species were notidentified were not included in this calculation Thismeasure of incidence does not take into account tem-poral changes in the distribution of ants but the dis-tribution of the Argentine ant at Jasper Ridge changedvery slowly its front advancing at a rate of =300 my(Human and Gordon 1996)

By using spatial databases of the Jasper Ridge Pre-serve developed at the Center for Conservation Biol-ogy Stanford University we calculated the followingfor each survey point distance to the nearest sourceof surface water (m) distance to the nearest edge ofthe preserve (m) mean elevation (m) mean summersolstice insolation (kJm2d) and slope (deg) Manystreams in the preserve are present only briefly duringthe rainy season of especially wet years to excludethese streams only streams containing water for atleast 4 mo out of the year were included in the cal-culation of distance to water Physiographic variableswere calculated from a digital elevation model with ahOlizontal resolution of5 m developed using the ArcInfo TOPOGRID function (Hutchinson 1989 1993)from US Geological Survey 124000digital hypsog-raphy (62-m contour intervals) spot elevations andhydrology (Palo Alto Quad USGS 1990) We calcu-lated slope the rate of change in elevation by usingthe average maximum technique (Burrough 1986) ap-plied in a 3 by 3 cell neighborhood over each cell inthe raster data set Potential clear sky insolation forJune 21 was calculated using the Arc Marco Languageprogram SOLARFLUX (Hetrick et al 1993) with anatmospheric transmittance of 065 Mean values foreach abiotic factor were calculated within a 20-m di-ameter circle around the survey point the same areawithin which we searched for ants

To examine the relationship of incidence of antgenera to abiotic factors we used canonical corre-spondence analysis Canonical correspondence anal-ysis is an ordination technique that relates unimodalresponses of distribution data including abundance orincidence to environmental variables (ter Braak 19871988 Whittaker 1989 Palmer 1993) The techniquetakes into account relationships among independentvariables In canonical correspondence analysis thelinear combination of independent variables that bestexplains the differences in dependent variables is the1st canonical axisand additional axes are calculated toaccount for the remaining variation (ter Braak andPrentice 1988 Palmer 1993) Eigenvalues whichrange from 0 to 1 are calculated for each axis todescribe the fraction of variance that is accounted forby that axis

Monte Carlo permutations were used to test statis-tically whether the incidence of ant genera was re-lated to the abiotic variables by using CANOCO (terBraak 1987) In each of 99 unrestricted Monte Carlopermutations values of the abiotic variables and gen-era incidence were randomized and new canonical

824 ENVIRONMENTAL ENTOMOLOGY Vo 27 no 4

axes and eigenvalues were calculated If 95 of thenew eigenvalues calculated for axis I were lower thanthat eigenvalue in the fitted canonical correspon-dence analysis model then axis I of the fitted modelwas considered to account for a significant proportionof the variance in dependent variables (ter Braak1987) For a test of significance of the entire canonicalcorrespondence analysis model the eigenvalues of allaxes were summed after each permutation and thosesums were compared with the sum of the originaleigenvalues Similar Monte Carlo permutations wereused to test the significance of the contribution of eachabiotic factor to the variance in ant distribution

Distance to the preserve edge and elevation ap-peared to account for most of the differences in thedistribution of all native ant genera (considered as 1group) and the Argentine ant and previous worksuggested that the Argentine ant is limited to areasnear permanently flowing water (Ward 1987Holway1995) To examine further the relationship of Argen-tine ant incidence and these 3 factors hectares wereclassified by degree of Argentine ant establishment asfollows (1) firmly established (found in 75-100of allsurveys in 3 yr) (2) transitional (found in 14-74 ofall surveys) and (3) absent (never found) To com-pare areas where Argentine ants were firmly estab-lished transitional and absent we plotted the valuesof the abiotic variables with box plots We performedno statistical analyses on these data because they wereused in statistical analysis of the canonical correspon-dence analysis model and because the 3 abiotic factorsmay not be independent of each other

Activity Patterns To compare the activity patternsof native ant species and Argentine ants we observedcolonies of Argentine ants and 4 native ant speciesCamponotus semitestaceus Snelling Formica subpolitaMayr Messor andrei Mayr and Pheidole californicaMayr During each observation day 6 colonies of eachspecies of native ant and or 12 Argentine ant nestswere observed at 3-h intervals for 21 consecutive hDuring each activity observation we recorded thenumber of ants entering and exiting the colony within2 min The activity of all colonies was observed over6 d native ant colonies for 3 d in June 1994and 3 d inJuly 1994 and Argentine ant colonies for 2 d in June19943 d in July 1994 and 1 d in August 1994 For 1 din June only colonies of native ants were observedand for 1 d in August only colonies of Argentine antswere observed Data were collected in the summermonths when activity is highest and most cOnstant forall species (KGH unpublished data) When activitycounts were made we also recorded the soil surfacetemperature with an infrared thermometer (Omega-scope model 0582) and relative humidity and airtem-perature 3 cm above the soil surface with an Omegatemperature and relative humidity meter (model RH-21C Omega Engineering Inc Stamford CT USA)

Soil temperature air temperature relative humid-ity and time of day are all related To examine speciesdifferences in the relation of ant colony activity andthe abiotic factors we used canonical correspondenceanalysis which accounts for relationships among in-

dependent variables In this canonical correspon-dence analysis colony activity (numbers of ants en-tering and exiting colonies) was considered thedependent variable and soil temperature (0C) airtemperature (0C) and relative humidity () wereindependent variables Time of day was not includedin the canonical correspondence analysis because theanalysis cannot handle circular data Monte Carlo sim-ulations were used to test how well the canonicalcorrespondence analysis model describes the relationof ant colony activity to the abiotic factors

Diet Overlap and Forage Composition To comparethe diets of 4 native ant species and the Argentine antwe intercepted foragers of each species returning tothe colonies and their forage wa~ identified In July1994 and April 1995 we collected forage from 9 col-onies of Argentine ants 12 colonies of M andrei 10colonies of F subpolita 11 colonies of P californicaand 9 colonies of C semitestaceus in a chaparral-oakwoodland at Jasper Ridge The native ant species over-lapped in range with each other but not with theArgentine ant the area with the colonies of nativespecies was 100-200 m from the invasion front of theArgentine ant The forage of 100ants returning to eachcolony during peak foraging hours was identified onsite or collected and later identified in the laboratoryForage was classified as follows seeds meat (dead orlive insects) liquid food (indicated by distention of anants gaster) or other (twigs bits ofleaves soil rockor unidentifiable matter) For ants of 1 native speciesF subpolita we could not determine whether thegasters of workers were distended

To compare the diet of the Argentine ant and eachnative ant species we calculated overlap 0 ij as

where Pia and Pja are the proportions of food categorya in the diets of species i and j (Pianka 1973) Thevalue approaches 1 as overlap increases

Results

Ant Species Distributions Argentine ants appear tohave invaded Jasper Ridge Biological Preserve fromthe edge and they appear to be most strongly estab-lished in areas near water (Fig 1) Distance to edgeelevation insolation and distance to water togetherexplained a significant proportion of the variation inthe distribution of ant genera In the canonical cor-respondence analysis eigenvalues of Axis I and Axis IIwere 046 and 002 respectively indicating that thelinear combination of environmental variables in AxisI described most of the variation in ant distributionexplained by the mode The linear combinations wereas follows

Axis I = 075 e + 042 v - 004 s - 021 i

+ 018 w [1]

August 1998 HUMAN ET AL DISTRIBUTION AND AcrNITY OF ARGENTINE ANT

1 kIn

Incidence of Argentine ant

bull 075

Ea 014-074

Do

825

Fig 1 Argentineant incidence at JasperRidgeBiologicalPreservein 9 surveysover 3 yrEach squarerepresents 1 haand incidencewascalculated for each ha as the proportion of 9 surveysin which the Argentineant was foundThe blacklinearoundthe perimeterof the figureis the preserveboundaryblacklinesin the interiorof the preserverepresent streams[mda lake and a wetland area are shown in gray

Axis 11= -035 e + 09 v - 114 s - 075 i

- 065 W [2]

where e = distance to edge v = elevation s = slopei == insolation and w = distance to water The coef-ficient of each describes the magnitude of its contri-bution to the axis Monte Carlo simulations deter-mined that both the 1st canonical axis alone and allaxes combined contdbute significantly to variationamong genera in incidence (P lt 001 for axis 1 P lt001 for the whole model) Axis II was not significantso further axes were discounted

In the ordination diagram (Fig 2) abiotic variablesare represented by arrows Arrows point toward gen-era that tend to occur in areas with high values of theabiotic variable represented The projection of eacharrow on an axis indicates the magnitude of its con-tribution to that axis and arrows are numbered by thevalue of the Axis I canonical coefficients of the abioticfactors they represent For example although arrow 5representing the variable slope is relatively long in thedirection of Axis II its projection on Axis I is small

reflecting a small canonical coefficient in that axis(-004 equation 1)

Distance to the preserve edge weighted mostheavily in Axis I accounted for most of the varHltionamong genera in distribution (canonical coefficient =075 in axis I P = 001 Monte Carlo permutations)The arrow for distance to the preserve edge (arrow 1)points generally to the right indicating that ant generathat tend to occur far from the edge of the preserve fallon the right side of the ordination diagram (Fig 2)Crematogaster spp Formica spp and M andrei aregenerally found in areas far from the preserve edgewhereas Argentine ants and Liometopum occidentaleEmery are generally found much closer to the edgePrenolepis imparis Mayr and Leptothorax spp werefound in areas at intermediate distances from the edgeA dotted line drawn perpendicular to the arrow andthrough the origin (1 in Fig 2) separates the Argen-tine ant (LH) from most native ant genera

Elevation contributed significantly to differencesamong ant genera in distribution (P = 001 MonteCarlo permutations) Ants found at highest elevations

TS CSPC

826

LH

La

(Insolation) -i 4

l

ENVIRONMENTAL ENTOMOLOGY

Axis II

2 (Elevation)

ST

3 (Distance to water)

SM

MA

1 (Distance to edge)

Vol 27 no 4

Axis

Fig 2 Canonical correspondence ordination of ant incidence data during 9 surveys Abiotic factors are represented byvectors as follows 1Distance to edge 2 elevation 3 distance to water 4 insolation 5 slope To aid in the visual interpretationof the diagram dotted line l is drawn perpendicular to vector 1 Ant genera that fall on opposite sides of this line occuron average in areas with different values of the abiotic factor represented distance to edge Axis IT is plotted for conventiononly its contribution to the variance in ant distribution is minimal CC Crematogaster spp CS C semitestaceus FM Formicaspp LH L humile LO L occidentale LP Leptothorax spp MA Missor andrei PC Pheidole californica PI Prenolepis imparisSM SnlOlesta and ST Stenamma spp LO PC ST and SM (small letters) were rarely discovered in ant surveys CC CS FMLH MA PI and TS (large letters) were discovered frequently

included Stenamma spp and Camponotus spp andthose found in lower elevations included Solenopsismolesta Mayr L occidentale and the Argentine ant

Insolation and distance to water explained small butsignificant amounts of variation among genera in dis-tribution (P = 001 for insolation P = 002 fordistance to water) S molesta and the Argentine antwere found in areas with the highest insolation and Pimparis Leptothomx spp and the Argentine ant werefound closest to water Slope did not explain any vari-ation in distribution among ant genera (P = 023)

Of the 7 native ant species sampled frequentlyshown in large letters in Fig 2 P imparis is mostsimilar to the Argentine ant P imparis is the 1 speciesof native ant often seen foraging in the vicinity ofArgentine ant nests P imparis colonies are active fromNovember to June during times when most native antcolonies are not active and when the Argentine antsare frequently inactive L occidentale sampled rarely

is also close to the Argentine ant on the ordinationdiagranl Like the Argentine ant L occidentale is adoichoderine ant that makes long trails as it foragesfor the liquid exudate of insects (Wheeler andWheeler 1986)

When the effects of distance to edge elevation anddistance to water were each plotted separately theresults supported the patterns suggested by the ca-nonical correspondence analysis On average the Ar-gentine ant was most strongly established closer to thepreserve edge closer to water and in areas at lowerelevation (Fig 3)

Activity Patterns Soil temperature relative humid-ity and air temperature explained a significant pro-portion of the variation in the distribution of ant gen-era (Fig 4) Eigenvalues of Axis I and II were 019 and003 respectively indicating that the linear combina-tion of abiotic factors in Axis I accounted for most of

Fig 3 Boxplots of the relationshipbetween degree ofestablishment by the Argentine ant and 3 environmentalvariables (a-c) Each dot represents 1 ha and the box sur-rounds50ofallobservationsHorizontallinesinthe centersof the boxes are medians and diamonds show the meanvalues (center of the diamond) and standard error of themean (height of the diamond) The right column in eachgraph represents hectares where Argentine ants are wellestablished(found in 75-100of all surveysover 3 yr) Themiddle column represents transitionalhectares (Argentineants found in 14-74of all surveys) and the left columnrepresents hectares in which Argentine ants were neverobserved

827

axis I = 101 Ts + 094 Rh + 003 Ta [3]

axis II = 081 Ts - 072 Rh - 049 Ta [4]

where Ts is soil temperature in degrees centigrade Rhis relative humidity in percent and Ta is air temper-ature in degrees centigrade Monte Carlo simulationsdetermined that both the 1st canonical axis alone andall axes combined contributed significantly to varia-tion between species in activity (P lt 001 for axis IP lt 001 for the whole model 99 permutations) Thelow eigenvalue of axis I in this canonical correspon-dence analysis reveals that there is substantial overlapamong ant species in abiotic conditions at which theyare active Axis II was not significant so further axeswere discounted

All 3 of the abiotic factors contributed significantlyto variation in colony activity (P = 001 for soiltemperature P = 001 for relative humidity P = 001for air temperature Monte Carlo permutations) Soiltemperature and relative humidity represented moststrongly in the 1st canonical axis account for most ofthe differences among colonies in activity patternsThe outlying position of F subpalita in the ordinationdiagram (Fig 4) is probably partly a result of lowoverall activity of F subpolita colonies relative to col-onies of the other species tested

The Argentine ant was somewhat more likely thanthe native ant species to be active in higher soil tem-peratures up to 70degCColonies of P calif arnica and Csemitestaceus both of which are crepuscular and noc-turnal did not forage in temperatures gt40degC F sub-palita and Mandrei remained active up to =60degC (FigSa) Air temperature measured with a thermocoupleappeared to be highly correlated with infrared-de-rived soil temperature (Fig 6c) so it is not surprisingthat for all species the relations of colony activity toair and soil temperature were similar (Fig 5 a and b)Colonies of the Argentine ant were active in air tem-peratures up to 45degC as were colonies of M andreiwhereas other native ant species were more limited byair temperature (Fig 5b)

Relative humidity may limit slightly the activity ofall species (Fig 5c) all ant species were more likelyto be inactive when relative humidity was lt40 (Fig5c) The Argentine ant was slightly more likely thannative species to be active when relative humidity washigh (Fig 5c)

Time of day influenced colony activity C semi-testaceus M andrei and P californica were usuallyinactive for some time in the middle of the day thoughM andrei only for a short period (Fig 5c) The Ar-gentine ants were often active throughout the day

Overall during the summer Argentine ants are ac-tive in a wider range of temperatures and relativehumidities than colonies of C semitestaceus and Fsubpalita and in a slightly wider range of tempera-tures relative humidities and times than P californica(Fig 5 a-d) Colonies of M andrei are active undermost of the same conditions as the Argentine ants

the variance in colony activity explained by the modelThe linear combinations were as follows

HUMAN ET AL DISTRIBUTION AND ACTIVITY OF ARGENTINE ANT

noLh transitional Lh well-established

Degree of L humile establishment

a

Au~ust 1998

S--QJ 800beclQJQJ

~ 600QJltfJQJbull0-0 400bullQJU

~ 200~Cl

0

~-g 180sect 160tll1l

j 140QJ

~ 120QJ

E100

80

60500-g 40Jbull

QJbulll1l~ 3008Q)

~ 200bull~Cl 100

828 ENVIRONMENTAL ENTOMOLOGY

Axis IIFS

Vol 27 no 4

MA PC

CS

1 (Soil temperature)

2 (Relative humidity)

Axis I

Fig 4 CANOCO ordination of ant colony activity data Abiotic factors are represented by vectors as follows 1 soiltemperature 2 relative humidity and 3 air temperature AxisII is plotted for convention only its contribution to the variancein ant distribution is minimal CS Camponotus semitestaceus FS F subpolita LH Linepithema hwnile MAMessor andrei PCPheidole califomico

with the exception of a short period of time in themiddle of the day characterized by high soil temper-atures when Mandrei colonies are usually inactive fora few hours but the Argentine ants are not (Fig 5 a andd)

There are many sets of abiotic conditions in whichcolonies of native and Argentine ants are both active(Fig 5) For example whenever Pcalif arnica coloniesare active during the day so are the Argentine ants(Fig 5d) There are no values of any of the abioticfactors measured in which the native ant species weremore likely to be active than the Argentine ants

Abiotic factors all depend upon each other Relativehumidity was high at low air temperatures (Fig 6a)soil temperature increases from sunrise to middaybefore decreasing again (Fig 6b) and air and soiltemperatures are positively correlated (Fig 6c) Ca-nonical correspondence analysis takes the interactionof variables into account

Diet Overlap The 2 seed-eating ants tested Mandrei and P calif arnica overlapped only slightly indiet with the Argentine ant (0 = 0011 and 0015

respectively Table 1) Most of the forage of M andreiand P califarnica was seeds although like the Argen-tine ants both the seed-eating species also broughtsome dead insects back to the colony (Table 1) Fsubpalita and the Argentine ant overlap only in thecollection of insects (Table 1) but insects made up afar larger fraction of the forage of F subpolita (0631)than of the Argentine ant (0055) The diet of Csemitestaceus was similar to that of the Argentine antboth species primarily collected liquid food

Discussion

Distribution of Ants at Jasper Ridge At JasperRidge the Argentine ant ismost likely to occur in areasthat are close to the preserve edge and areas at lowelevation whereas native ant species are more abun-dant in the center of the preserve and at higher ele-vation

The association of the Argentine ant with areas nearthe preserve edge may be due to its response to dis-turbance and to its mechanism of dispersal Many

August 1998 HUMAN ET AL DISTRIBUTION AND ACTIVlTI OF ARGENTINE ANT 829

C semitestaceus F subpolita L humile M andrei P califarnica

I

0 I bull I r-~~720 1440o

bull pobullbull 1II bull

I bullbullbull - ~ bullbull~bull bull It

bull I304050607080

I~r ~ I

L70 10 30 50 70

bullbull

bull bull It ~-

bull

-r--

304050607080

-- bull

ItS ~

r bull -T

bull

~----

bull

720 1440

~ t bullbullbull I

I bullbull~ ~-- - ~

o

70 10 30 50 70 10 30 50

Soil temperature (0C)

bull10 bullbullbull

bull

bullbull

bull bull bullbull - -- bull bullbullbullbullbullbull bull bullbullbull bullbullbull- ~II bullbullbull - middot bull

bullbullbullbull middot bullbull bull bullbull bull i - bullbull bull 1bullbullbull bull 1bull ~ bull bull bullbull

30 40 50 60 70 80 30 40 50 60 70 80

Relative humidity ()

1440midnight

bull i~ ~ Imiddot

bullbull

20 30 40 50 10 20 30 40 50 10 20 30 40 50 10 20 30 40 50 10 20 30 40 50

Air Temperature (0C)

bull

720noon

4middot

bull

middot

bull-1bull

30 40 50 60 70 8e

bull

d

c200

50

a200

1150-

100 bull -jI~

50 -Ibull

10 30 50 70

b200

150 bull- bull100 middot

100

150

100

150

200

omidnight

Time (min from previous midnight)

Fig 5 The influence of the abiotic factors (a) soil temperature (b) air temperature (c) relative humidity and (d) timeon ant colony activity for 5 ant species Ant colony activity was measured as the numbers of ants entering and exiting a colonywithin 2 min Each dot on the figures represents 1 observation of 1 colony

exotic species including the Argentine ant more eas-ily invade areas that are disturbed by human activity(Ward 1987Elton 1958Simberloff1981) This may berelated both to increased opportunity for introduction

in areas with a great deal of human activity and lackof competitors or predators if native species are sen-sitive to disturbance and retreat It is likely that theArgentine ant moved into the reserve from the de-

830 ENVIRONMENTAL ENTOMOLOGY Vol 27 no 4

Time (min from midnight)

c80

E 60QJbull2bull 40QJ0Ell0

20Ul

00 10 20 30 40

Air temperature (0C)

1440

b80

~ 60QJ

2~ 40QJ0-Ell0

20Ul

00 720

a

C 80

c~ 60Elc 40QJ

iO 20Oip 00

0 0 0

0 10 20 30 40

Air temperature (0C)

Fig6 Examplesof the relationshipsamongabioticfactors (a-c) Eachdot represents 1 measurementof colonyactivityand the indicated abioticfactor

veloped surrounding areas that have already beeninvaded

The dispersal strategy of the Argentine ant mayfurther explain why it tends to occur near the edges ofthe preserve Mating flights have been observed onlyrarely in this species (Newell and Barber 1913 Skaife1961) Occasional spot introductions of Argentine antshave been described in Californias Central Valley(Holway 1995) This may have occurred through theaccidental introduction of queens by humans throughqueens or fragments of colonies rafting downstream infloods (Holway 1995) or through the founding of newcolonies by winged queens Over the last 3 yr we haveobserved no spot introductions In the Jasper RidgePreserve most areas along creek corridors are alreadyinvaded by Argentine ants and there is little oppor-tunity for the species to be introduced accidentallyinside the preserve because human activities within itare limited Our results suggest that the Argentine antoccurs mostly at the preserve edges because it is ex-panding its range primarily by spreading along theground through colony fission

Table 1 Composition of forage and diet overlap between Ar-gentine ants and 4 native ant species

Ant species

LH CS FS MA PC

Proportion of food itemsLiquid food 0916 0825 0000 0000 0000Seeds 0004 0004 0036 0765 0661Insects 0055 0102 0631 0093 0120Other 0025 0069 0333 0141 0219

Total number of food items1313 484 279 686 183

Proportion overlap with Argentine antLow 0997 0060 0011 0015High 0998 0066 0016 0023

In (C) values closest to 1 indicate the greatest degree of overlapThe low overlap value assumed that forage in the unidentified cate-gory (other) differed between species and the high value assumedthat forage in this category was identical CS C semitestacetls FS Fsltbpolita LH L hmile MA M andrei PC P califomica

It seems unlikely that previous disturbance of theareas surrounding the preserve had already caused thenative species to disappear from inside the preserveprior to invasion by Argentine ants Within the pre-serve native ant species and Argentine ants competeon the margin of the invasion and native ant speciesdisappear from invaded areas but persist elsewhere(Human and Gordon 1996 1997) All of the 9 hainvaded by the Argentine ant between September1993 and September 1995 were occupied in 1993 bynative ant species

Argentine ant incidence was high in low-elevationareas However elevation per se is unlikely to be thelimiting factor because the Argentine ant can befound in a few high-elevation areas of Jasper Ridgeand the species has been found up to 2000 m in theHawaiian islands (Zimmerman 1940 Cole et al 1992)Elevation may be related to another environmentalvariable such as soil moisture that affects the Argen-tine ant directly There may be a historical effect if theArgentine ant were introduced more frequently tolow-elevation areas

Studies in the Central Valley of California havesuggested that Argentine ant encroachment into nat-ural habitat is almost exclUSivelylimited to areas withpermanent sources of water and that the Argentine antis more limited by proximity to water than are nativeant species (Ward 1987) We found that several nativetaxa including Prenoepis imparis Stenamma spp andLeptothorax spp were as close to water on average aswere Argentine ants Many other taxa including Cam-ponotus spp P californica and Tapinoma sessile Smithwere found only marginally farther from water thanthe Argentine ant (Fig 2) Although Argentine antstend to be more strongly established in areas closer towater (Fig 1) there are many areas near water not yetinvaded and many areas far from water where Ar-gentine ants are firmly established The differencebetween the results of this study and previous onesmay be due to climatic differences the Central Valleyis much dryer and hotter especially in the summerthan Jasper Ridge which lies in the coastal foothills

August 1998 HUMAN ET AL DISTRIBUTION AND ACTIVITY OF ARGENTINE ANT 831

Colony Activity and Abiotic Conditions It may bethat the Argentine ant avoids temperature extremesby some behavioral means A laboratory study thatcompared Argentine ants and other ant species nativeto central coastal California suggested that the Argen-tine ant has a low thermal tolerance dying at tem-peratures close to 46degC (Tremper 1976) In the fieldArgentine ant nests frequently experience soil tem-peratures gt46degC and they are active up to 70degC It ispossible that the Argentine ant copes behaviorallywith heat more effectively than native species In thelaboratory M andrei workers died when exposed totemperatures gt485degC (Tremper 1976) We foundthat M andrei colonies were rarely active when soiltemperatures exceeded 50degC M andrei are large blackants that are often found on trails and open areaswhere they may be unable to avoid temperature ex-tremes

Compared with most native ant species Argentineants are active in a wide range of abiotic conditions(Fig 5) M andrei and P californica were active in arange of conditions similar to that of Argentine antsBy contrast the activities of C semitestaceus and Fsubpolita were confined to a narrower range of timesand temperatures Because the Argentine ant is usu-ally active whenever any native species is most nativespecies in invaded areas are likely to encounter theinvaders frequently

The diet of the Argentine ant overlaps with that ofseveral species (Table 1) Argentine ant nests areactive in a wider range of abiotic conditions than areC semitestaceus or F subpolita the species with whichit competes most for food Diet overlap with C semi-testaceus is almost complete Overlap with F subpolitamay be higher than measured in our study becauseprevious studies suggest that many Formica speciesincluding F subpalita collect insect and plant exudate(Cook 1953 Dumpert 1981 Wheeler and Wheeler1986 Cherix 1987) which makes up a significant pro-portion of the diet of Argentine ants Diet overlap withP calif arnica and M andrei was minimal We measuredthe diets of native ant species and Argentine ant spe-cies in different areas where resource availability mayhave varied affecting food choice Further work isneeded at an invasion front to investigate effects offood availability on diet overlap

The strong association of the Argentine ant with theedges of Jasper Ridge suggests that the most significantlimits to its range expansion may be opportunity forintroduction and its own mechanism of dispersal Dis-tance to water was less important than we expectedbut a study that spans a wider range of conditions thanours might discover a stronger effect of moistureLimiting human activity within nature preserves andcreating reserves with large areaperimeter ratios mayprevent invasion of interior areas by Argentine antsand other invasive species that tend to track humandisturbance Elevation or some parameter related toit is also important to Argentine ant distribution andfurther investigation of this might contribute to othermanagement practices The possibility of spatial au-tocorrelation should be kept in mind sites near to each

other are likely to have similar environmental condi-tions and similar histories of Argentine ant establish-ment

Abiotic factors explain much of the difference indistribution of Argentine and native ant species whichdo not overlap on a scale of hectares but little of thedifference in activity patterns At the edge of theinvasion on a scale of meters Argentine and native antspecies forage in the same areas with substantial over-lap in temporal patterns of foraging activity Theseresults suggest that whenever native ant colonies onthe edge of the invasion are active they are likely tointeract with Argentine ants Interactions between theinvasive and native ant species may promote displace-ment of the native species Frequent encounters maylead native species to retreat from food sources andpossibly to emigrate Where Argentine ants and nativeants overlap substantially in diet exploitation compe-tition could reduce the foraging success of native antcolonies

Acknowledgments

We thank Eric Bjorkstedt (Stanford University) MarkBrown (Stanford University) Diane Wagner (Stanford Uni-versity) and 2 anonymous reviewers for their comments onthe manuscript Kristen Duin (Stanford University) and LisaStadlen (Stanford University) were terrific field assistantsRobert Swierk (Center for Conservation Biology StanfordUniversity) provided technical assistance with GeographicalInformation Systems The work was supported by disserta-tion fellowships to KGH from National Science Foundationand the American Association of University Women grantsto KGH from the Jasper Ridge Biological Preserve TheConservation and Research Foundation and the Phi BetaKappa Northern California Chapter and USDA grant No95-37302-1885 to DMG

References Cited

Adams E S and IFA Trainello 1980 Chemical interfer-ence competition by Monomorium minimum Oecologia(Ber) 51 265-270

Andersen A N and A D Patel 1994 Meat ants as domi-nant members of Australian ant communities an exper-imental test of their influence on the foraging success andforager abundance of other species Oecologia (Ber) 9815-24

Anderson A N 1992 Regulation of momentary diversityby dominant species in exceptionally rich ant communi-ties of the Australian seasonal tropics Am Nat 140 401-420

Andrews E A 1927 Ant-mounds as to temperature andsunshine J of Morpho Physio 44 1-20

ArcView GIS 1996 ArcInfo 701 ArcView 21 Environ-mental Systems Research Institute Redlands CA

Baker H G and G L Stebbins 1965 The genetics ofcolonizing species Proceedings International Union ofBiological Sciences Symposia on General Biology Aca-demic New York

Banks W A and D F Williams 1989 Competitive dis-placement of Paratrechina longicomis (Latreille) (Hyme-noptera Formicidae) from baits by fire ants in MattoGrosso Brazi J Entomo Sci 24 381-391

August 1998 HUMAN ET AL DISTRffiUTION AND ACTIVITY OF ARGENTINE ANT 823

solation on Argentine ant distribution We then comparethe daily activity patterns and diets of the Argentine antand several native ant species The daily activity patternsof ant colonies may depend on abiotic factors such as soiltemperature and relative humidity (Fielde 1904Talbot1943 Brian 1964 Marsh 1988 VepsaIainen and Savol-ainen 1990 Dean 1992) and also on interactions withcompetitors prey or predators (Brian 1964Lynch et al1980 Briese and Macauley 1980 Gordon 1988 Savol-ainen and VepsaIainen 1988Andersen 1992Dean 1992Andersen and Patel 1994)We relate activity patterns to4 abiotic factors (time of day soil temperature air tem-perature and relative humidity) and calculate the over-lap in diet of Argentine ant and native ant species Fi-nally we investigate whether abiotic conditions maylimit the invasion of the Argentine ant and consider howtemporal and resource overlap may lead to competitionbetween the invasive and native ant species

Materials and Methods

Study Site This study was conducted at the 482-haJasper Ridge Biological Preserve in northern Califor-nia (San Mateo County 1220 13 Wand 3 24 N60-120 m in elevation) Vegetation types found at thepreserve include serpentine grassland annual grass-land redwood forests chaparral evergreen forest oakwoodland wetland and riparian systems (specieslisted in Sawyer and Keeler-Wolf 1995) Argentineants have invaded =30 of the area of Jasper Ridgeprimarily along the edge of the preserve Jasper Ridgeis surrounded by low-density residential areas andagricuturalland most of which has already been in-vaded by Argentine ants

Ant Species Distributions Between May 1993 andJanuary 1996we surveyed all ofJasper Ridge Biolog-ical Preserve for ants 3 times yearly Surveys wereconducted in May (following the end of the rainyseason) September (at the end of the dry summer)and January (during the rainy season) For ant sur-veys a map of the entire preserve was overlaid with a100-m grid By using survey points at the center ofeach hectare we recorded the presence or absence ofnative ants and Argentine ants For 5 min we searchedvisually for ants within a 20-m-radius circle of thesurvey point If no ants were found we left honeytraps consisting of 40-ml vials filled with 10 ml ofhoney and collected them 24 h later Details of thesurvey method and a test of its accuracy are describedelsewhere (Human and Gordon 1996) In the Sep-tember 1993 January 1994 and May 1994 surveysnative ant species were identified only as not Argen-tine ants but thereafter they were identified to genusIn each time period 296-318 ha were surveyed A fewsurvey points could not be reached during particulartimes of year because of flooding or poison oak (Tox-icodendron diversilobum Jepson) Ant distributiondata were entered into a geographical informationsystem (ArcView GIS 1996)

For analysis ant distribution data for each hectarewere converted into a measure of incidence the pro-portion of all surveys in which ants of a particular

genus were found For example ifArgentine ants weredetected in 7 of 9 surveys Argentine ant incidence inthat hectare was 078 Incidence was calculated for allant species sampled in each hectare Data from the 3earliest surveys in which native species were notidentified were not included in this calculation Thismeasure of incidence does not take into account tem-poral changes in the distribution of ants but the dis-tribution of the Argentine ant at Jasper Ridge changedvery slowly its front advancing at a rate of =300 my(Human and Gordon 1996)

By using spatial databases of the Jasper Ridge Pre-serve developed at the Center for Conservation Biol-ogy Stanford University we calculated the followingfor each survey point distance to the nearest sourceof surface water (m) distance to the nearest edge ofthe preserve (m) mean elevation (m) mean summersolstice insolation (kJm2d) and slope (deg) Manystreams in the preserve are present only briefly duringthe rainy season of especially wet years to excludethese streams only streams containing water for atleast 4 mo out of the year were included in the cal-culation of distance to water Physiographic variableswere calculated from a digital elevation model with ahOlizontal resolution of5 m developed using the ArcInfo TOPOGRID function (Hutchinson 1989 1993)from US Geological Survey 124000digital hypsog-raphy (62-m contour intervals) spot elevations andhydrology (Palo Alto Quad USGS 1990) We calcu-lated slope the rate of change in elevation by usingthe average maximum technique (Burrough 1986) ap-plied in a 3 by 3 cell neighborhood over each cell inthe raster data set Potential clear sky insolation forJune 21 was calculated using the Arc Marco Languageprogram SOLARFLUX (Hetrick et al 1993) with anatmospheric transmittance of 065 Mean values foreach abiotic factor were calculated within a 20-m di-ameter circle around the survey point the same areawithin which we searched for ants

To examine the relationship of incidence of antgenera to abiotic factors we used canonical corre-spondence analysis Canonical correspondence anal-ysis is an ordination technique that relates unimodalresponses of distribution data including abundance orincidence to environmental variables (ter Braak 19871988 Whittaker 1989 Palmer 1993) The techniquetakes into account relationships among independentvariables In canonical correspondence analysis thelinear combination of independent variables that bestexplains the differences in dependent variables is the1st canonical axisand additional axes are calculated toaccount for the remaining variation (ter Braak andPrentice 1988 Palmer 1993) Eigenvalues whichrange from 0 to 1 are calculated for each axis todescribe the fraction of variance that is accounted forby that axis

Monte Carlo permutations were used to test statis-tically whether the incidence of ant genera was re-lated to the abiotic variables by using CANOCO (terBraak 1987) In each of 99 unrestricted Monte Carlopermutations values of the abiotic variables and gen-era incidence were randomized and new canonical

824 ENVIRONMENTAL ENTOMOLOGY Vo 27 no 4

axes and eigenvalues were calculated If 95 of thenew eigenvalues calculated for axis I were lower thanthat eigenvalue in the fitted canonical correspon-dence analysis model then axis I of the fitted modelwas considered to account for a significant proportionof the variance in dependent variables (ter Braak1987) For a test of significance of the entire canonicalcorrespondence analysis model the eigenvalues of allaxes were summed after each permutation and thosesums were compared with the sum of the originaleigenvalues Similar Monte Carlo permutations wereused to test the significance of the contribution of eachabiotic factor to the variance in ant distribution

Distance to the preserve edge and elevation ap-peared to account for most of the differences in thedistribution of all native ant genera (considered as 1group) and the Argentine ant and previous worksuggested that the Argentine ant is limited to areasnear permanently flowing water (Ward 1987Holway1995) To examine further the relationship of Argen-tine ant incidence and these 3 factors hectares wereclassified by degree of Argentine ant establishment asfollows (1) firmly established (found in 75-100of allsurveys in 3 yr) (2) transitional (found in 14-74 ofall surveys) and (3) absent (never found) To com-pare areas where Argentine ants were firmly estab-lished transitional and absent we plotted the valuesof the abiotic variables with box plots We performedno statistical analyses on these data because they wereused in statistical analysis of the canonical correspon-dence analysis model and because the 3 abiotic factorsmay not be independent of each other

Activity Patterns To compare the activity patternsof native ant species and Argentine ants we observedcolonies of Argentine ants and 4 native ant speciesCamponotus semitestaceus Snelling Formica subpolitaMayr Messor andrei Mayr and Pheidole californicaMayr During each observation day 6 colonies of eachspecies of native ant and or 12 Argentine ant nestswere observed at 3-h intervals for 21 consecutive hDuring each activity observation we recorded thenumber of ants entering and exiting the colony within2 min The activity of all colonies was observed over6 d native ant colonies for 3 d in June 1994and 3 d inJuly 1994 and Argentine ant colonies for 2 d in June19943 d in July 1994 and 1 d in August 1994 For 1 din June only colonies of native ants were observedand for 1 d in August only colonies of Argentine antswere observed Data were collected in the summermonths when activity is highest and most cOnstant forall species (KGH unpublished data) When activitycounts were made we also recorded the soil surfacetemperature with an infrared thermometer (Omega-scope model 0582) and relative humidity and airtem-perature 3 cm above the soil surface with an Omegatemperature and relative humidity meter (model RH-21C Omega Engineering Inc Stamford CT USA)

Soil temperature air temperature relative humid-ity and time of day are all related To examine speciesdifferences in the relation of ant colony activity andthe abiotic factors we used canonical correspondenceanalysis which accounts for relationships among in-

dependent variables In this canonical correspon-dence analysis colony activity (numbers of ants en-tering and exiting colonies) was considered thedependent variable and soil temperature (0C) airtemperature (0C) and relative humidity () wereindependent variables Time of day was not includedin the canonical correspondence analysis because theanalysis cannot handle circular data Monte Carlo sim-ulations were used to test how well the canonicalcorrespondence analysis model describes the relationof ant colony activity to the abiotic factors

Diet Overlap and Forage Composition To comparethe diets of 4 native ant species and the Argentine antwe intercepted foragers of each species returning tothe colonies and their forage wa~ identified In July1994 and April 1995 we collected forage from 9 col-onies of Argentine ants 12 colonies of M andrei 10colonies of F subpolita 11 colonies of P californicaand 9 colonies of C semitestaceus in a chaparral-oakwoodland at Jasper Ridge The native ant species over-lapped in range with each other but not with theArgentine ant the area with the colonies of nativespecies was 100-200 m from the invasion front of theArgentine ant The forage of 100ants returning to eachcolony during peak foraging hours was identified onsite or collected and later identified in the laboratoryForage was classified as follows seeds meat (dead orlive insects) liquid food (indicated by distention of anants gaster) or other (twigs bits ofleaves soil rockor unidentifiable matter) For ants of 1 native speciesF subpolita we could not determine whether thegasters of workers were distended

To compare the diet of the Argentine ant and eachnative ant species we calculated overlap 0 ij as

where Pia and Pja are the proportions of food categorya in the diets of species i and j (Pianka 1973) Thevalue approaches 1 as overlap increases

Results

Ant Species Distributions Argentine ants appear tohave invaded Jasper Ridge Biological Preserve fromthe edge and they appear to be most strongly estab-lished in areas near water (Fig 1) Distance to edgeelevation insolation and distance to water togetherexplained a significant proportion of the variation inthe distribution of ant genera In the canonical cor-respondence analysis eigenvalues of Axis I and Axis IIwere 046 and 002 respectively indicating that thelinear combination of environmental variables in AxisI described most of the variation in ant distributionexplained by the mode The linear combinations wereas follows

Axis I = 075 e + 042 v - 004 s - 021 i

+ 018 w [1]

August 1998 HUMAN ET AL DISTRIBUTION AND AcrNITY OF ARGENTINE ANT

1 kIn

Incidence of Argentine ant

bull 075

Ea 014-074

Do

825

Fig 1 Argentineant incidence at JasperRidgeBiologicalPreservein 9 surveysover 3 yrEach squarerepresents 1 haand incidencewascalculated for each ha as the proportion of 9 surveysin which the Argentineant was foundThe blacklinearoundthe perimeterof the figureis the preserveboundaryblacklinesin the interiorof the preserverepresent streams[mda lake and a wetland area are shown in gray

Axis 11= -035 e + 09 v - 114 s - 075 i

- 065 W [2]

where e = distance to edge v = elevation s = slopei == insolation and w = distance to water The coef-ficient of each describes the magnitude of its contri-bution to the axis Monte Carlo simulations deter-mined that both the 1st canonical axis alone and allaxes combined contdbute significantly to variationamong genera in incidence (P lt 001 for axis 1 P lt001 for the whole model) Axis II was not significantso further axes were discounted

In the ordination diagram (Fig 2) abiotic variablesare represented by arrows Arrows point toward gen-era that tend to occur in areas with high values of theabiotic variable represented The projection of eacharrow on an axis indicates the magnitude of its con-tribution to that axis and arrows are numbered by thevalue of the Axis I canonical coefficients of the abioticfactors they represent For example although arrow 5representing the variable slope is relatively long in thedirection of Axis II its projection on Axis I is small

reflecting a small canonical coefficient in that axis(-004 equation 1)

Distance to the preserve edge weighted mostheavily in Axis I accounted for most of the varHltionamong genera in distribution (canonical coefficient =075 in axis I P = 001 Monte Carlo permutations)The arrow for distance to the preserve edge (arrow 1)points generally to the right indicating that ant generathat tend to occur far from the edge of the preserve fallon the right side of the ordination diagram (Fig 2)Crematogaster spp Formica spp and M andrei aregenerally found in areas far from the preserve edgewhereas Argentine ants and Liometopum occidentaleEmery are generally found much closer to the edgePrenolepis imparis Mayr and Leptothorax spp werefound in areas at intermediate distances from the edgeA dotted line drawn perpendicular to the arrow andthrough the origin (1 in Fig 2) separates the Argen-tine ant (LH) from most native ant genera

Elevation contributed significantly to differencesamong ant genera in distribution (P = 001 MonteCarlo permutations) Ants found at highest elevations

TS CSPC

826

LH

La

(Insolation) -i 4

l

ENVIRONMENTAL ENTOMOLOGY

Axis II

2 (Elevation)

ST

3 (Distance to water)

SM

MA

1 (Distance to edge)

Vol 27 no 4

Axis

Fig 2 Canonical correspondence ordination of ant incidence data during 9 surveys Abiotic factors are represented byvectors as follows 1Distance to edge 2 elevation 3 distance to water 4 insolation 5 slope To aid in the visual interpretationof the diagram dotted line l is drawn perpendicular to vector 1 Ant genera that fall on opposite sides of this line occuron average in areas with different values of the abiotic factor represented distance to edge Axis IT is plotted for conventiononly its contribution to the variance in ant distribution is minimal CC Crematogaster spp CS C semitestaceus FM Formicaspp LH L humile LO L occidentale LP Leptothorax spp MA Missor andrei PC Pheidole californica PI Prenolepis imparisSM SnlOlesta and ST Stenamma spp LO PC ST and SM (small letters) were rarely discovered in ant surveys CC CS FMLH MA PI and TS (large letters) were discovered frequently

included Stenamma spp and Camponotus spp andthose found in lower elevations included Solenopsismolesta Mayr L occidentale and the Argentine ant

Insolation and distance to water explained small butsignificant amounts of variation among genera in dis-tribution (P = 001 for insolation P = 002 fordistance to water) S molesta and the Argentine antwere found in areas with the highest insolation and Pimparis Leptothomx spp and the Argentine ant werefound closest to water Slope did not explain any vari-ation in distribution among ant genera (P = 023)

Of the 7 native ant species sampled frequentlyshown in large letters in Fig 2 P imparis is mostsimilar to the Argentine ant P imparis is the 1 speciesof native ant often seen foraging in the vicinity ofArgentine ant nests P imparis colonies are active fromNovember to June during times when most native antcolonies are not active and when the Argentine antsare frequently inactive L occidentale sampled rarely

is also close to the Argentine ant on the ordinationdiagranl Like the Argentine ant L occidentale is adoichoderine ant that makes long trails as it foragesfor the liquid exudate of insects (Wheeler andWheeler 1986)

When the effects of distance to edge elevation anddistance to water were each plotted separately theresults supported the patterns suggested by the ca-nonical correspondence analysis On average the Ar-gentine ant was most strongly established closer to thepreserve edge closer to water and in areas at lowerelevation (Fig 3)

Activity Patterns Soil temperature relative humid-ity and air temperature explained a significant pro-portion of the variation in the distribution of ant gen-era (Fig 4) Eigenvalues of Axis I and II were 019 and003 respectively indicating that the linear combina-tion of abiotic factors in Axis I accounted for most of

Fig 3 Boxplots of the relationshipbetween degree ofestablishment by the Argentine ant and 3 environmentalvariables (a-c) Each dot represents 1 ha and the box sur-rounds50ofallobservationsHorizontallinesinthe centersof the boxes are medians and diamonds show the meanvalues (center of the diamond) and standard error of themean (height of the diamond) The right column in eachgraph represents hectares where Argentine ants are wellestablished(found in 75-100of all surveysover 3 yr) Themiddle column represents transitionalhectares (Argentineants found in 14-74of all surveys) and the left columnrepresents hectares in which Argentine ants were neverobserved

827

axis I = 101 Ts + 094 Rh + 003 Ta [3]

axis II = 081 Ts - 072 Rh - 049 Ta [4]

where Ts is soil temperature in degrees centigrade Rhis relative humidity in percent and Ta is air temper-ature in degrees centigrade Monte Carlo simulationsdetermined that both the 1st canonical axis alone andall axes combined contributed significantly to varia-tion between species in activity (P lt 001 for axis IP lt 001 for the whole model 99 permutations) Thelow eigenvalue of axis I in this canonical correspon-dence analysis reveals that there is substantial overlapamong ant species in abiotic conditions at which theyare active Axis II was not significant so further axeswere discounted

All 3 of the abiotic factors contributed significantlyto variation in colony activity (P = 001 for soiltemperature P = 001 for relative humidity P = 001for air temperature Monte Carlo permutations) Soiltemperature and relative humidity represented moststrongly in the 1st canonical axis account for most ofthe differences among colonies in activity patternsThe outlying position of F subpalita in the ordinationdiagram (Fig 4) is probably partly a result of lowoverall activity of F subpolita colonies relative to col-onies of the other species tested

The Argentine ant was somewhat more likely thanthe native ant species to be active in higher soil tem-peratures up to 70degCColonies of P calif arnica and Csemitestaceus both of which are crepuscular and noc-turnal did not forage in temperatures gt40degC F sub-palita and Mandrei remained active up to =60degC (FigSa) Air temperature measured with a thermocoupleappeared to be highly correlated with infrared-de-rived soil temperature (Fig 6c) so it is not surprisingthat for all species the relations of colony activity toair and soil temperature were similar (Fig 5 a and b)Colonies of the Argentine ant were active in air tem-peratures up to 45degC as were colonies of M andreiwhereas other native ant species were more limited byair temperature (Fig 5b)

Relative humidity may limit slightly the activity ofall species (Fig 5c) all ant species were more likelyto be inactive when relative humidity was lt40 (Fig5c) The Argentine ant was slightly more likely thannative species to be active when relative humidity washigh (Fig 5c)

Time of day influenced colony activity C semi-testaceus M andrei and P californica were usuallyinactive for some time in the middle of the day thoughM andrei only for a short period (Fig 5c) The Ar-gentine ants were often active throughout the day

Overall during the summer Argentine ants are ac-tive in a wider range of temperatures and relativehumidities than colonies of C semitestaceus and Fsubpalita and in a slightly wider range of tempera-tures relative humidities and times than P californica(Fig 5 a-d) Colonies of M andrei are active undermost of the same conditions as the Argentine ants

the variance in colony activity explained by the modelThe linear combinations were as follows

HUMAN ET AL DISTRIBUTION AND ACTIVITY OF ARGENTINE ANT

noLh transitional Lh well-established

Degree of L humile establishment

a

Au~ust 1998

S--QJ 800beclQJQJ

~ 600QJltfJQJbull0-0 400bullQJU

~ 200~Cl

0

~-g 180sect 160tll1l

j 140QJ

~ 120QJ

E100

80

60500-g 40Jbull

QJbulll1l~ 3008Q)

~ 200bull~Cl 100

828 ENVIRONMENTAL ENTOMOLOGY

Axis IIFS

Vol 27 no 4

MA PC

CS

1 (Soil temperature)

2 (Relative humidity)

Axis I

Fig 4 CANOCO ordination of ant colony activity data Abiotic factors are represented by vectors as follows 1 soiltemperature 2 relative humidity and 3 air temperature AxisII is plotted for convention only its contribution to the variancein ant distribution is minimal CS Camponotus semitestaceus FS F subpolita LH Linepithema hwnile MAMessor andrei PCPheidole califomico

with the exception of a short period of time in themiddle of the day characterized by high soil temper-atures when Mandrei colonies are usually inactive fora few hours but the Argentine ants are not (Fig 5 a andd)

There are many sets of abiotic conditions in whichcolonies of native and Argentine ants are both active(Fig 5) For example whenever Pcalif arnica coloniesare active during the day so are the Argentine ants(Fig 5d) There are no values of any of the abioticfactors measured in which the native ant species weremore likely to be active than the Argentine ants

Abiotic factors all depend upon each other Relativehumidity was high at low air temperatures (Fig 6a)soil temperature increases from sunrise to middaybefore decreasing again (Fig 6b) and air and soiltemperatures are positively correlated (Fig 6c) Ca-nonical correspondence analysis takes the interactionof variables into account

Diet Overlap The 2 seed-eating ants tested Mandrei and P calif arnica overlapped only slightly indiet with the Argentine ant (0 = 0011 and 0015

respectively Table 1) Most of the forage of M andreiand P califarnica was seeds although like the Argen-tine ants both the seed-eating species also broughtsome dead insects back to the colony (Table 1) Fsubpalita and the Argentine ant overlap only in thecollection of insects (Table 1) but insects made up afar larger fraction of the forage of F subpolita (0631)than of the Argentine ant (0055) The diet of Csemitestaceus was similar to that of the Argentine antboth species primarily collected liquid food

Discussion

Distribution of Ants at Jasper Ridge At JasperRidge the Argentine ant ismost likely to occur in areasthat are close to the preserve edge and areas at lowelevation whereas native ant species are more abun-dant in the center of the preserve and at higher ele-vation

The association of the Argentine ant with areas nearthe preserve edge may be due to its response to dis-turbance and to its mechanism of dispersal Many

August 1998 HUMAN ET AL DISTRIBUTION AND ACTIVlTI OF ARGENTINE ANT 829

C semitestaceus F subpolita L humile M andrei P califarnica

I

0 I bull I r-~~720 1440o

bull pobullbull 1II bull

I bullbullbull - ~ bullbull~bull bull It

bull I304050607080

I~r ~ I

L70 10 30 50 70

bullbull

bull bull It ~-

bull

-r--

304050607080

-- bull

ItS ~

r bull -T

bull

~----

bull

720 1440

~ t bullbullbull I

I bullbull~ ~-- - ~

o

70 10 30 50 70 10 30 50

Soil temperature (0C)

bull10 bullbullbull

bull

bullbull

bull bull bullbull - -- bull bullbullbullbullbullbull bull bullbullbull bullbullbull- ~II bullbullbull - middot bull

bullbullbullbull middot bullbull bull bullbull bull i - bullbull bull 1bullbullbull bull 1bull ~ bull bull bullbull

30 40 50 60 70 80 30 40 50 60 70 80

Relative humidity ()

1440midnight

bull i~ ~ Imiddot

bullbull

20 30 40 50 10 20 30 40 50 10 20 30 40 50 10 20 30 40 50 10 20 30 40 50

Air Temperature (0C)

bull

720noon

4middot

bull

middot

bull-1bull

30 40 50 60 70 8e

bull

d

c200

50

a200

1150-

100 bull -jI~

50 -Ibull

10 30 50 70

b200

150 bull- bull100 middot

100

150

100

150

200

omidnight

Time (min from previous midnight)

Fig 5 The influence of the abiotic factors (a) soil temperature (b) air temperature (c) relative humidity and (d) timeon ant colony activity for 5 ant species Ant colony activity was measured as the numbers of ants entering and exiting a colonywithin 2 min Each dot on the figures represents 1 observation of 1 colony

exotic species including the Argentine ant more eas-ily invade areas that are disturbed by human activity(Ward 1987Elton 1958Simberloff1981) This may berelated both to increased opportunity for introduction

in areas with a great deal of human activity and lackof competitors or predators if native species are sen-sitive to disturbance and retreat It is likely that theArgentine ant moved into the reserve from the de-

830 ENVIRONMENTAL ENTOMOLOGY Vol 27 no 4

Time (min from midnight)

c80

E 60QJbull2bull 40QJ0Ell0

20Ul

00 10 20 30 40

Air temperature (0C)

1440

b80

~ 60QJ

2~ 40QJ0-Ell0

20Ul

00 720

a

C 80

c~ 60Elc 40QJ

iO 20Oip 00

0 0 0

0 10 20 30 40

Air temperature (0C)

Fig6 Examplesof the relationshipsamongabioticfactors (a-c) Eachdot represents 1 measurementof colonyactivityand the indicated abioticfactor

veloped surrounding areas that have already beeninvaded

The dispersal strategy of the Argentine ant mayfurther explain why it tends to occur near the edges ofthe preserve Mating flights have been observed onlyrarely in this species (Newell and Barber 1913 Skaife1961) Occasional spot introductions of Argentine antshave been described in Californias Central Valley(Holway 1995) This may have occurred through theaccidental introduction of queens by humans throughqueens or fragments of colonies rafting downstream infloods (Holway 1995) or through the founding of newcolonies by winged queens Over the last 3 yr we haveobserved no spot introductions In the Jasper RidgePreserve most areas along creek corridors are alreadyinvaded by Argentine ants and there is little oppor-tunity for the species to be introduced accidentallyinside the preserve because human activities within itare limited Our results suggest that the Argentine antoccurs mostly at the preserve edges because it is ex-panding its range primarily by spreading along theground through colony fission

Table 1 Composition of forage and diet overlap between Ar-gentine ants and 4 native ant species

Ant species

LH CS FS MA PC

Proportion of food itemsLiquid food 0916 0825 0000 0000 0000Seeds 0004 0004 0036 0765 0661Insects 0055 0102 0631 0093 0120Other 0025 0069 0333 0141 0219

Total number of food items1313 484 279 686 183

Proportion overlap with Argentine antLow 0997 0060 0011 0015High 0998 0066 0016 0023

In (C) values closest to 1 indicate the greatest degree of overlapThe low overlap value assumed that forage in the unidentified cate-gory (other) differed between species and the high value assumedthat forage in this category was identical CS C semitestacetls FS Fsltbpolita LH L hmile MA M andrei PC P califomica

It seems unlikely that previous disturbance of theareas surrounding the preserve had already caused thenative species to disappear from inside the preserveprior to invasion by Argentine ants Within the pre-serve native ant species and Argentine ants competeon the margin of the invasion and native ant speciesdisappear from invaded areas but persist elsewhere(Human and Gordon 1996 1997) All of the 9 hainvaded by the Argentine ant between September1993 and September 1995 were occupied in 1993 bynative ant species

Argentine ant incidence was high in low-elevationareas However elevation per se is unlikely to be thelimiting factor because the Argentine ant can befound in a few high-elevation areas of Jasper Ridgeand the species has been found up to 2000 m in theHawaiian islands (Zimmerman 1940 Cole et al 1992)Elevation may be related to another environmentalvariable such as soil moisture that affects the Argen-tine ant directly There may be a historical effect if theArgentine ant were introduced more frequently tolow-elevation areas

Studies in the Central Valley of California havesuggested that Argentine ant encroachment into nat-ural habitat is almost exclUSivelylimited to areas withpermanent sources of water and that the Argentine antis more limited by proximity to water than are nativeant species (Ward 1987) We found that several nativetaxa including Prenoepis imparis Stenamma spp andLeptothorax spp were as close to water on average aswere Argentine ants Many other taxa including Cam-ponotus spp P californica and Tapinoma sessile Smithwere found only marginally farther from water thanthe Argentine ant (Fig 2) Although Argentine antstend to be more strongly established in areas closer towater (Fig 1) there are many areas near water not yetinvaded and many areas far from water where Ar-gentine ants are firmly established The differencebetween the results of this study and previous onesmay be due to climatic differences the Central Valleyis much dryer and hotter especially in the summerthan Jasper Ridge which lies in the coastal foothills

August 1998 HUMAN ET AL DISTRIBUTION AND ACTIVITY OF ARGENTINE ANT 831

Colony Activity and Abiotic Conditions It may bethat the Argentine ant avoids temperature extremesby some behavioral means A laboratory study thatcompared Argentine ants and other ant species nativeto central coastal California suggested that the Argen-tine ant has a low thermal tolerance dying at tem-peratures close to 46degC (Tremper 1976) In the fieldArgentine ant nests frequently experience soil tem-peratures gt46degC and they are active up to 70degC It ispossible that the Argentine ant copes behaviorallywith heat more effectively than native species In thelaboratory M andrei workers died when exposed totemperatures gt485degC (Tremper 1976) We foundthat M andrei colonies were rarely active when soiltemperatures exceeded 50degC M andrei are large blackants that are often found on trails and open areaswhere they may be unable to avoid temperature ex-tremes

Compared with most native ant species Argentineants are active in a wide range of abiotic conditions(Fig 5) M andrei and P californica were active in arange of conditions similar to that of Argentine antsBy contrast the activities of C semitestaceus and Fsubpolita were confined to a narrower range of timesand temperatures Because the Argentine ant is usu-ally active whenever any native species is most nativespecies in invaded areas are likely to encounter theinvaders frequently

The diet of the Argentine ant overlaps with that ofseveral species (Table 1) Argentine ant nests areactive in a wider range of abiotic conditions than areC semitestaceus or F subpolita the species with whichit competes most for food Diet overlap with C semi-testaceus is almost complete Overlap with F subpolitamay be higher than measured in our study becauseprevious studies suggest that many Formica speciesincluding F subpalita collect insect and plant exudate(Cook 1953 Dumpert 1981 Wheeler and Wheeler1986 Cherix 1987) which makes up a significant pro-portion of the diet of Argentine ants Diet overlap withP calif arnica and M andrei was minimal We measuredthe diets of native ant species and Argentine ant spe-cies in different areas where resource availability mayhave varied affecting food choice Further work isneeded at an invasion front to investigate effects offood availability on diet overlap

The strong association of the Argentine ant with theedges of Jasper Ridge suggests that the most significantlimits to its range expansion may be opportunity forintroduction and its own mechanism of dispersal Dis-tance to water was less important than we expectedbut a study that spans a wider range of conditions thanours might discover a stronger effect of moistureLimiting human activity within nature preserves andcreating reserves with large areaperimeter ratios mayprevent invasion of interior areas by Argentine antsand other invasive species that tend to track humandisturbance Elevation or some parameter related toit is also important to Argentine ant distribution andfurther investigation of this might contribute to othermanagement practices The possibility of spatial au-tocorrelation should be kept in mind sites near to each

other are likely to have similar environmental condi-tions and similar histories of Argentine ant establish-ment

Abiotic factors explain much of the difference indistribution of Argentine and native ant species whichdo not overlap on a scale of hectares but little of thedifference in activity patterns At the edge of theinvasion on a scale of meters Argentine and native antspecies forage in the same areas with substantial over-lap in temporal patterns of foraging activity Theseresults suggest that whenever native ant colonies onthe edge of the invasion are active they are likely tointeract with Argentine ants Interactions between theinvasive and native ant species may promote displace-ment of the native species Frequent encounters maylead native species to retreat from food sources andpossibly to emigrate Where Argentine ants and nativeants overlap substantially in diet exploitation compe-tition could reduce the foraging success of native antcolonies

Acknowledgments

We thank Eric Bjorkstedt (Stanford University) MarkBrown (Stanford University) Diane Wagner (Stanford Uni-versity) and 2 anonymous reviewers for their comments onthe manuscript Kristen Duin (Stanford University) and LisaStadlen (Stanford University) were terrific field assistantsRobert Swierk (Center for Conservation Biology StanfordUniversity) provided technical assistance with GeographicalInformation Systems The work was supported by disserta-tion fellowships to KGH from National Science Foundationand the American Association of University Women grantsto KGH from the Jasper Ridge Biological Preserve TheConservation and Research Foundation and the Phi BetaKappa Northern California Chapter and USDA grant No95-37302-1885 to DMG

References Cited

Adams E S and IFA Trainello 1980 Chemical interfer-ence competition by Monomorium minimum Oecologia(Ber) 51 265-270

Andersen A N and A D Patel 1994 Meat ants as domi-nant members of Australian ant communities an exper-imental test of their influence on the foraging success andforager abundance of other species Oecologia (Ber) 9815-24

Anderson A N 1992 Regulation of momentary diversityby dominant species in exceptionally rich ant communi-ties of the Australian seasonal tropics Am Nat 140 401-420

Andrews E A 1927 Ant-mounds as to temperature andsunshine J of Morpho Physio 44 1-20

ArcView GIS 1996 ArcInfo 701 ArcView 21 Environ-mental Systems Research Institute Redlands CA

Baker H G and G L Stebbins 1965 The genetics ofcolonizing species Proceedings International Union ofBiological Sciences Symposia on General Biology Aca-demic New York

Banks W A and D F Williams 1989 Competitive dis-placement of Paratrechina longicomis (Latreille) (Hyme-noptera Formicidae) from baits by fire ants in MattoGrosso Brazi J Entomo Sci 24 381-391

824 ENVIRONMENTAL ENTOMOLOGY Vo 27 no 4

axes and eigenvalues were calculated If 95 of thenew eigenvalues calculated for axis I were lower thanthat eigenvalue in the fitted canonical correspon-dence analysis model then axis I of the fitted modelwas considered to account for a significant proportionof the variance in dependent variables (ter Braak1987) For a test of significance of the entire canonicalcorrespondence analysis model the eigenvalues of allaxes were summed after each permutation and thosesums were compared with the sum of the originaleigenvalues Similar Monte Carlo permutations wereused to test the significance of the contribution of eachabiotic factor to the variance in ant distribution

Distance to the preserve edge and elevation ap-peared to account for most of the differences in thedistribution of all native ant genera (considered as 1group) and the Argentine ant and previous worksuggested that the Argentine ant is limited to areasnear permanently flowing water (Ward 1987Holway1995) To examine further the relationship of Argen-tine ant incidence and these 3 factors hectares wereclassified by degree of Argentine ant establishment asfollows (1) firmly established (found in 75-100of allsurveys in 3 yr) (2) transitional (found in 14-74 ofall surveys) and (3) absent (never found) To com-pare areas where Argentine ants were firmly estab-lished transitional and absent we plotted the valuesof the abiotic variables with box plots We performedno statistical analyses on these data because they wereused in statistical analysis of the canonical correspon-dence analysis model and because the 3 abiotic factorsmay not be independent of each other

Activity Patterns To compare the activity patternsof native ant species and Argentine ants we observedcolonies of Argentine ants and 4 native ant speciesCamponotus semitestaceus Snelling Formica subpolitaMayr Messor andrei Mayr and Pheidole californicaMayr During each observation day 6 colonies of eachspecies of native ant and or 12 Argentine ant nestswere observed at 3-h intervals for 21 consecutive hDuring each activity observation we recorded thenumber of ants entering and exiting the colony within2 min The activity of all colonies was observed over6 d native ant colonies for 3 d in June 1994and 3 d inJuly 1994 and Argentine ant colonies for 2 d in June19943 d in July 1994 and 1 d in August 1994 For 1 din June only colonies of native ants were observedand for 1 d in August only colonies of Argentine antswere observed Data were collected in the summermonths when activity is highest and most cOnstant forall species (KGH unpublished data) When activitycounts were made we also recorded the soil surfacetemperature with an infrared thermometer (Omega-scope model 0582) and relative humidity and airtem-perature 3 cm above the soil surface with an Omegatemperature and relative humidity meter (model RH-21C Omega Engineering Inc Stamford CT USA)

Soil temperature air temperature relative humid-ity and time of day are all related To examine speciesdifferences in the relation of ant colony activity andthe abiotic factors we used canonical correspondenceanalysis which accounts for relationships among in-

dependent variables In this canonical correspon-dence analysis colony activity (numbers of ants en-tering and exiting colonies) was considered thedependent variable and soil temperature (0C) airtemperature (0C) and relative humidity () wereindependent variables Time of day was not includedin the canonical correspondence analysis because theanalysis cannot handle circular data Monte Carlo sim-ulations were used to test how well the canonicalcorrespondence analysis model describes the relationof ant colony activity to the abiotic factors

Diet Overlap and Forage Composition To comparethe diets of 4 native ant species and the Argentine antwe intercepted foragers of each species returning tothe colonies and their forage wa~ identified In July1994 and April 1995 we collected forage from 9 col-onies of Argentine ants 12 colonies of M andrei 10colonies of F subpolita 11 colonies of P californicaand 9 colonies of C semitestaceus in a chaparral-oakwoodland at Jasper Ridge The native ant species over-lapped in range with each other but not with theArgentine ant the area with the colonies of nativespecies was 100-200 m from the invasion front of theArgentine ant The forage of 100ants returning to eachcolony during peak foraging hours was identified onsite or collected and later identified in the laboratoryForage was classified as follows seeds meat (dead orlive insects) liquid food (indicated by distention of anants gaster) or other (twigs bits ofleaves soil rockor unidentifiable matter) For ants of 1 native speciesF subpolita we could not determine whether thegasters of workers were distended

To compare the diet of the Argentine ant and eachnative ant species we calculated overlap 0 ij as

where Pia and Pja are the proportions of food categorya in the diets of species i and j (Pianka 1973) Thevalue approaches 1 as overlap increases

Results

Ant Species Distributions Argentine ants appear tohave invaded Jasper Ridge Biological Preserve fromthe edge and they appear to be most strongly estab-lished in areas near water (Fig 1) Distance to edgeelevation insolation and distance to water togetherexplained a significant proportion of the variation inthe distribution of ant genera In the canonical cor-respondence analysis eigenvalues of Axis I and Axis IIwere 046 and 002 respectively indicating that thelinear combination of environmental variables in AxisI described most of the variation in ant distributionexplained by the mode The linear combinations wereas follows

Axis I = 075 e + 042 v - 004 s - 021 i

+ 018 w [1]

August 1998 HUMAN ET AL DISTRIBUTION AND AcrNITY OF ARGENTINE ANT

1 kIn

Incidence of Argentine ant

bull 075

Ea 014-074

Do

825

Fig 1 Argentineant incidence at JasperRidgeBiologicalPreservein 9 surveysover 3 yrEach squarerepresents 1 haand incidencewascalculated for each ha as the proportion of 9 surveysin which the Argentineant was foundThe blacklinearoundthe perimeterof the figureis the preserveboundaryblacklinesin the interiorof the preserverepresent streams[mda lake and a wetland area are shown in gray

Axis 11= -035 e + 09 v - 114 s - 075 i

- 065 W [2]

where e = distance to edge v = elevation s = slopei == insolation and w = distance to water The coef-ficient of each describes the magnitude of its contri-bution to the axis Monte Carlo simulations deter-mined that both the 1st canonical axis alone and allaxes combined contdbute significantly to variationamong genera in incidence (P lt 001 for axis 1 P lt001 for the whole model) Axis II was not significantso further axes were discounted

In the ordination diagram (Fig 2) abiotic variablesare represented by arrows Arrows point toward gen-era that tend to occur in areas with high values of theabiotic variable represented The projection of eacharrow on an axis indicates the magnitude of its con-tribution to that axis and arrows are numbered by thevalue of the Axis I canonical coefficients of the abioticfactors they represent For example although arrow 5representing the variable slope is relatively long in thedirection of Axis II its projection on Axis I is small

reflecting a small canonical coefficient in that axis(-004 equation 1)

Distance to the preserve edge weighted mostheavily in Axis I accounted for most of the varHltionamong genera in distribution (canonical coefficient =075 in axis I P = 001 Monte Carlo permutations)The arrow for distance to the preserve edge (arrow 1)points generally to the right indicating that ant generathat tend to occur far from the edge of the preserve fallon the right side of the ordination diagram (Fig 2)Crematogaster spp Formica spp and M andrei aregenerally found in areas far from the preserve edgewhereas Argentine ants and Liometopum occidentaleEmery are generally found much closer to the edgePrenolepis imparis Mayr and Leptothorax spp werefound in areas at intermediate distances from the edgeA dotted line drawn perpendicular to the arrow andthrough the origin (1 in Fig 2) separates the Argen-tine ant (LH) from most native ant genera

Elevation contributed significantly to differencesamong ant genera in distribution (P = 001 MonteCarlo permutations) Ants found at highest elevations

TS CSPC

826

LH

La

(Insolation) -i 4

l

ENVIRONMENTAL ENTOMOLOGY

Axis II

2 (Elevation)

ST

3 (Distance to water)

SM

MA

1 (Distance to edge)

Vol 27 no 4

Axis

Fig 2 Canonical correspondence ordination of ant incidence data during 9 surveys Abiotic factors are represented byvectors as follows 1Distance to edge 2 elevation 3 distance to water 4 insolation 5 slope To aid in the visual interpretationof the diagram dotted line l is drawn perpendicular to vector 1 Ant genera that fall on opposite sides of this line occuron average in areas with different values of the abiotic factor represented distance to edge Axis IT is plotted for conventiononly its contribution to the variance in ant distribution is minimal CC Crematogaster spp CS C semitestaceus FM Formicaspp LH L humile LO L occidentale LP Leptothorax spp MA Missor andrei PC Pheidole californica PI Prenolepis imparisSM SnlOlesta and ST Stenamma spp LO PC ST and SM (small letters) were rarely discovered in ant surveys CC CS FMLH MA PI and TS (large letters) were discovered frequently

included Stenamma spp and Camponotus spp andthose found in lower elevations included Solenopsismolesta Mayr L occidentale and the Argentine ant

Insolation and distance to water explained small butsignificant amounts of variation among genera in dis-tribution (P = 001 for insolation P = 002 fordistance to water) S molesta and the Argentine antwere found in areas with the highest insolation and Pimparis Leptothomx spp and the Argentine ant werefound closest to water Slope did not explain any vari-ation in distribution among ant genera (P = 023)

Of the 7 native ant species sampled frequentlyshown in large letters in Fig 2 P imparis is mostsimilar to the Argentine ant P imparis is the 1 speciesof native ant often seen foraging in the vicinity ofArgentine ant nests P imparis colonies are active fromNovember to June during times when most native antcolonies are not active and when the Argentine antsare frequently inactive L occidentale sampled rarely

is also close to the Argentine ant on the ordinationdiagranl Like the Argentine ant L occidentale is adoichoderine ant that makes long trails as it foragesfor the liquid exudate of insects (Wheeler andWheeler 1986)

When the effects of distance to edge elevation anddistance to water were each plotted separately theresults supported the patterns suggested by the ca-nonical correspondence analysis On average the Ar-gentine ant was most strongly established closer to thepreserve edge closer to water and in areas at lowerelevation (Fig 3)

Activity Patterns Soil temperature relative humid-ity and air temperature explained a significant pro-portion of the variation in the distribution of ant gen-era (Fig 4) Eigenvalues of Axis I and II were 019 and003 respectively indicating that the linear combina-tion of abiotic factors in Axis I accounted for most of

Fig 3 Boxplots of the relationshipbetween degree ofestablishment by the Argentine ant and 3 environmentalvariables (a-c) Each dot represents 1 ha and the box sur-rounds50ofallobservationsHorizontallinesinthe centersof the boxes are medians and diamonds show the meanvalues (center of the diamond) and standard error of themean (height of the diamond) The right column in eachgraph represents hectares where Argentine ants are wellestablished(found in 75-100of all surveysover 3 yr) Themiddle column represents transitionalhectares (Argentineants found in 14-74of all surveys) and the left columnrepresents hectares in which Argentine ants were neverobserved

827

axis I = 101 Ts + 094 Rh + 003 Ta [3]

axis II = 081 Ts - 072 Rh - 049 Ta [4]

where Ts is soil temperature in degrees centigrade Rhis relative humidity in percent and Ta is air temper-ature in degrees centigrade Monte Carlo simulationsdetermined that both the 1st canonical axis alone andall axes combined contributed significantly to varia-tion between species in activity (P lt 001 for axis IP lt 001 for the whole model 99 permutations) Thelow eigenvalue of axis I in this canonical correspon-dence analysis reveals that there is substantial overlapamong ant species in abiotic conditions at which theyare active Axis II was not significant so further axeswere discounted

All 3 of the abiotic factors contributed significantlyto variation in colony activity (P = 001 for soiltemperature P = 001 for relative humidity P = 001for air temperature Monte Carlo permutations) Soiltemperature and relative humidity represented moststrongly in the 1st canonical axis account for most ofthe differences among colonies in activity patternsThe outlying position of F subpalita in the ordinationdiagram (Fig 4) is probably partly a result of lowoverall activity of F subpolita colonies relative to col-onies of the other species tested

The Argentine ant was somewhat more likely thanthe native ant species to be active in higher soil tem-peratures up to 70degCColonies of P calif arnica and Csemitestaceus both of which are crepuscular and noc-turnal did not forage in temperatures gt40degC F sub-palita and Mandrei remained active up to =60degC (FigSa) Air temperature measured with a thermocoupleappeared to be highly correlated with infrared-de-rived soil temperature (Fig 6c) so it is not surprisingthat for all species the relations of colony activity toair and soil temperature were similar (Fig 5 a and b)Colonies of the Argentine ant were active in air tem-peratures up to 45degC as were colonies of M andreiwhereas other native ant species were more limited byair temperature (Fig 5b)

Relative humidity may limit slightly the activity ofall species (Fig 5c) all ant species were more likelyto be inactive when relative humidity was lt40 (Fig5c) The Argentine ant was slightly more likely thannative species to be active when relative humidity washigh (Fig 5c)

Time of day influenced colony activity C semi-testaceus M andrei and P californica were usuallyinactive for some time in the middle of the day thoughM andrei only for a short period (Fig 5c) The Ar-gentine ants were often active throughout the day

Overall during the summer Argentine ants are ac-tive in a wider range of temperatures and relativehumidities than colonies of C semitestaceus and Fsubpalita and in a slightly wider range of tempera-tures relative humidities and times than P californica(Fig 5 a-d) Colonies of M andrei are active undermost of the same conditions as the Argentine ants

the variance in colony activity explained by the modelThe linear combinations were as follows

HUMAN ET AL DISTRIBUTION AND ACTIVITY OF ARGENTINE ANT

noLh transitional Lh well-established

Degree of L humile establishment

a

Au~ust 1998

S--QJ 800beclQJQJ

~ 600QJltfJQJbull0-0 400bullQJU

~ 200~Cl

0

~-g 180sect 160tll1l

j 140QJ

~ 120QJ

E100

80

60500-g 40Jbull

QJbulll1l~ 3008Q)

~ 200bull~Cl 100

828 ENVIRONMENTAL ENTOMOLOGY

Axis IIFS

Vol 27 no 4

MA PC

CS

1 (Soil temperature)

2 (Relative humidity)

Axis I

Fig 4 CANOCO ordination of ant colony activity data Abiotic factors are represented by vectors as follows 1 soiltemperature 2 relative humidity and 3 air temperature AxisII is plotted for convention only its contribution to the variancein ant distribution is minimal CS Camponotus semitestaceus FS F subpolita LH Linepithema hwnile MAMessor andrei PCPheidole califomico

with the exception of a short period of time in themiddle of the day characterized by high soil temper-atures when Mandrei colonies are usually inactive fora few hours but the Argentine ants are not (Fig 5 a andd)

There are many sets of abiotic conditions in whichcolonies of native and Argentine ants are both active(Fig 5) For example whenever Pcalif arnica coloniesare active during the day so are the Argentine ants(Fig 5d) There are no values of any of the abioticfactors measured in which the native ant species weremore likely to be active than the Argentine ants

Abiotic factors all depend upon each other Relativehumidity was high at low air temperatures (Fig 6a)soil temperature increases from sunrise to middaybefore decreasing again (Fig 6b) and air and soiltemperatures are positively correlated (Fig 6c) Ca-nonical correspondence analysis takes the interactionof variables into account

Diet Overlap The 2 seed-eating ants tested Mandrei and P calif arnica overlapped only slightly indiet with the Argentine ant (0 = 0011 and 0015

respectively Table 1) Most of the forage of M andreiand P califarnica was seeds although like the Argen-tine ants both the seed-eating species also broughtsome dead insects back to the colony (Table 1) Fsubpalita and the Argentine ant overlap only in thecollection of insects (Table 1) but insects made up afar larger fraction of the forage of F subpolita (0631)than of the Argentine ant (0055) The diet of Csemitestaceus was similar to that of the Argentine antboth species primarily collected liquid food

Discussion

Distribution of Ants at Jasper Ridge At JasperRidge the Argentine ant ismost likely to occur in areasthat are close to the preserve edge and areas at lowelevation whereas native ant species are more abun-dant in the center of the preserve and at higher ele-vation

The association of the Argentine ant with areas nearthe preserve edge may be due to its response to dis-turbance and to its mechanism of dispersal Many

August 1998 HUMAN ET AL DISTRIBUTION AND ACTIVlTI OF ARGENTINE ANT 829

C semitestaceus F subpolita L humile M andrei P califarnica

I

0 I bull I r-~~720 1440o

bull pobullbull 1II bull

I bullbullbull - ~ bullbull~bull bull It

bull I304050607080

I~r ~ I

L70 10 30 50 70

bullbull

bull bull It ~-

bull

-r--

304050607080

-- bull

ItS ~

r bull -T

bull

~----

bull

720 1440

~ t bullbullbull I

I bullbull~ ~-- - ~

o

70 10 30 50 70 10 30 50

Soil temperature (0C)

bull10 bullbullbull

bull

bullbull

bull bull bullbull - -- bull bullbullbullbullbullbull bull bullbullbull bullbullbull- ~II bullbullbull - middot bull

bullbullbullbull middot bullbull bull bullbull bull i - bullbull bull 1bullbullbull bull 1bull ~ bull bull bullbull

30 40 50 60 70 80 30 40 50 60 70 80

Relative humidity ()

1440midnight

bull i~ ~ Imiddot

bullbull

20 30 40 50 10 20 30 40 50 10 20 30 40 50 10 20 30 40 50 10 20 30 40 50

Air Temperature (0C)

bull

720noon

4middot

bull

middot

bull-1bull

30 40 50 60 70 8e

bull

d

c200

50

a200

1150-

100 bull -jI~

50 -Ibull

10 30 50 70

b200

150 bull- bull100 middot

100

150

100

150

200

omidnight

Time (min from previous midnight)

Fig 5 The influence of the abiotic factors (a) soil temperature (b) air temperature (c) relative humidity and (d) timeon ant colony activity for 5 ant species Ant colony activity was measured as the numbers of ants entering and exiting a colonywithin 2 min Each dot on the figures represents 1 observation of 1 colony

exotic species including the Argentine ant more eas-ily invade areas that are disturbed by human activity(Ward 1987Elton 1958Simberloff1981) This may berelated both to increased opportunity for introduction

in areas with a great deal of human activity and lackof competitors or predators if native species are sen-sitive to disturbance and retreat It is likely that theArgentine ant moved into the reserve from the de-

830 ENVIRONMENTAL ENTOMOLOGY Vol 27 no 4

Time (min from midnight)

c80

E 60QJbull2bull 40QJ0Ell0

20Ul

00 10 20 30 40

Air temperature (0C)

1440

b80

~ 60QJ

2~ 40QJ0-Ell0

20Ul

00 720

a

C 80

c~ 60Elc 40QJ

iO 20Oip 00

0 0 0

0 10 20 30 40

Air temperature (0C)

Fig6 Examplesof the relationshipsamongabioticfactors (a-c) Eachdot represents 1 measurementof colonyactivityand the indicated abioticfactor

veloped surrounding areas that have already beeninvaded

The dispersal strategy of the Argentine ant mayfurther explain why it tends to occur near the edges ofthe preserve Mating flights have been observed onlyrarely in this species (Newell and Barber 1913 Skaife1961) Occasional spot introductions of Argentine antshave been described in Californias Central Valley(Holway 1995) This may have occurred through theaccidental introduction of queens by humans throughqueens or fragments of colonies rafting downstream infloods (Holway 1995) or through the founding of newcolonies by winged queens Over the last 3 yr we haveobserved no spot introductions In the Jasper RidgePreserve most areas along creek corridors are alreadyinvaded by Argentine ants and there is little oppor-tunity for the species to be introduced accidentallyinside the preserve because human activities within itare limited Our results suggest that the Argentine antoccurs mostly at the preserve edges because it is ex-panding its range primarily by spreading along theground through colony fission

Table 1 Composition of forage and diet overlap between Ar-gentine ants and 4 native ant species

Ant species

LH CS FS MA PC

Proportion of food itemsLiquid food 0916 0825 0000 0000 0000Seeds 0004 0004 0036 0765 0661Insects 0055 0102 0631 0093 0120Other 0025 0069 0333 0141 0219

Total number of food items1313 484 279 686 183

Proportion overlap with Argentine antLow 0997 0060 0011 0015High 0998 0066 0016 0023

In (C) values closest to 1 indicate the greatest degree of overlapThe low overlap value assumed that forage in the unidentified cate-gory (other) differed between species and the high value assumedthat forage in this category was identical CS C semitestacetls FS Fsltbpolita LH L hmile MA M andrei PC P califomica

It seems unlikely that previous disturbance of theareas surrounding the preserve had already caused thenative species to disappear from inside the preserveprior to invasion by Argentine ants Within the pre-serve native ant species and Argentine ants competeon the margin of the invasion and native ant speciesdisappear from invaded areas but persist elsewhere(Human and Gordon 1996 1997) All of the 9 hainvaded by the Argentine ant between September1993 and September 1995 were occupied in 1993 bynative ant species

Argentine ant incidence was high in low-elevationareas However elevation per se is unlikely to be thelimiting factor because the Argentine ant can befound in a few high-elevation areas of Jasper Ridgeand the species has been found up to 2000 m in theHawaiian islands (Zimmerman 1940 Cole et al 1992)Elevation may be related to another environmentalvariable such as soil moisture that affects the Argen-tine ant directly There may be a historical effect if theArgentine ant were introduced more frequently tolow-elevation areas

Studies in the Central Valley of California havesuggested that Argentine ant encroachment into nat-ural habitat is almost exclUSivelylimited to areas withpermanent sources of water and that the Argentine antis more limited by proximity to water than are nativeant species (Ward 1987) We found that several nativetaxa including Prenoepis imparis Stenamma spp andLeptothorax spp were as close to water on average aswere Argentine ants Many other taxa including Cam-ponotus spp P californica and Tapinoma sessile Smithwere found only marginally farther from water thanthe Argentine ant (Fig 2) Although Argentine antstend to be more strongly established in areas closer towater (Fig 1) there are many areas near water not yetinvaded and many areas far from water where Ar-gentine ants are firmly established The differencebetween the results of this study and previous onesmay be due to climatic differences the Central Valleyis much dryer and hotter especially in the summerthan Jasper Ridge which lies in the coastal foothills

August 1998 HUMAN ET AL DISTRIBUTION AND ACTIVITY OF ARGENTINE ANT 831

Colony Activity and Abiotic Conditions It may bethat the Argentine ant avoids temperature extremesby some behavioral means A laboratory study thatcompared Argentine ants and other ant species nativeto central coastal California suggested that the Argen-tine ant has a low thermal tolerance dying at tem-peratures close to 46degC (Tremper 1976) In the fieldArgentine ant nests frequently experience soil tem-peratures gt46degC and they are active up to 70degC It ispossible that the Argentine ant copes behaviorallywith heat more effectively than native species In thelaboratory M andrei workers died when exposed totemperatures gt485degC (Tremper 1976) We foundthat M andrei colonies were rarely active when soiltemperatures exceeded 50degC M andrei are large blackants that are often found on trails and open areaswhere they may be unable to avoid temperature ex-tremes

Compared with most native ant species Argentineants are active in a wide range of abiotic conditions(Fig 5) M andrei and P californica were active in arange of conditions similar to that of Argentine antsBy contrast the activities of C semitestaceus and Fsubpolita were confined to a narrower range of timesand temperatures Because the Argentine ant is usu-ally active whenever any native species is most nativespecies in invaded areas are likely to encounter theinvaders frequently

The diet of the Argentine ant overlaps with that ofseveral species (Table 1) Argentine ant nests areactive in a wider range of abiotic conditions than areC semitestaceus or F subpolita the species with whichit competes most for food Diet overlap with C semi-testaceus is almost complete Overlap with F subpolitamay be higher than measured in our study becauseprevious studies suggest that many Formica speciesincluding F subpalita collect insect and plant exudate(Cook 1953 Dumpert 1981 Wheeler and Wheeler1986 Cherix 1987) which makes up a significant pro-portion of the diet of Argentine ants Diet overlap withP calif arnica and M andrei was minimal We measuredthe diets of native ant species and Argentine ant spe-cies in different areas where resource availability mayhave varied affecting food choice Further work isneeded at an invasion front to investigate effects offood availability on diet overlap

The strong association of the Argentine ant with theedges of Jasper Ridge suggests that the most significantlimits to its range expansion may be opportunity forintroduction and its own mechanism of dispersal Dis-tance to water was less important than we expectedbut a study that spans a wider range of conditions thanours might discover a stronger effect of moistureLimiting human activity within nature preserves andcreating reserves with large areaperimeter ratios mayprevent invasion of interior areas by Argentine antsand other invasive species that tend to track humandisturbance Elevation or some parameter related toit is also important to Argentine ant distribution andfurther investigation of this might contribute to othermanagement practices The possibility of spatial au-tocorrelation should be kept in mind sites near to each

other are likely to have similar environmental condi-tions and similar histories of Argentine ant establish-ment

Abiotic factors explain much of the difference indistribution of Argentine and native ant species whichdo not overlap on a scale of hectares but little of thedifference in activity patterns At the edge of theinvasion on a scale of meters Argentine and native antspecies forage in the same areas with substantial over-lap in temporal patterns of foraging activity Theseresults suggest that whenever native ant colonies onthe edge of the invasion are active they are likely tointeract with Argentine ants Interactions between theinvasive and native ant species may promote displace-ment of the native species Frequent encounters maylead native species to retreat from food sources andpossibly to emigrate Where Argentine ants and nativeants overlap substantially in diet exploitation compe-tition could reduce the foraging success of native antcolonies

Acknowledgments

We thank Eric Bjorkstedt (Stanford University) MarkBrown (Stanford University) Diane Wagner (Stanford Uni-versity) and 2 anonymous reviewers for their comments onthe manuscript Kristen Duin (Stanford University) and LisaStadlen (Stanford University) were terrific field assistantsRobert Swierk (Center for Conservation Biology StanfordUniversity) provided technical assistance with GeographicalInformation Systems The work was supported by disserta-tion fellowships to KGH from National Science Foundationand the American Association of University Women grantsto KGH from the Jasper Ridge Biological Preserve TheConservation and Research Foundation and the Phi BetaKappa Northern California Chapter and USDA grant No95-37302-1885 to DMG

References Cited

Adams E S and IFA Trainello 1980 Chemical interfer-ence competition by Monomorium minimum Oecologia(Ber) 51 265-270

Andersen A N and A D Patel 1994 Meat ants as domi-nant members of Australian ant communities an exper-imental test of their influence on the foraging success andforager abundance of other species Oecologia (Ber) 9815-24

Anderson A N 1992 Regulation of momentary diversityby dominant species in exceptionally rich ant communi-ties of the Australian seasonal tropics Am Nat 140 401-420

Andrews E A 1927 Ant-mounds as to temperature andsunshine J of Morpho Physio 44 1-20

ArcView GIS 1996 ArcInfo 701 ArcView 21 Environ-mental Systems Research Institute Redlands CA

Baker H G and G L Stebbins 1965 The genetics ofcolonizing species Proceedings International Union ofBiological Sciences Symposia on General Biology Aca-demic New York

Banks W A and D F Williams 1989 Competitive dis-placement of Paratrechina longicomis (Latreille) (Hyme-noptera Formicidae) from baits by fire ants in MattoGrosso Brazi J Entomo Sci 24 381-391

August 1998 HUMAN ET AL DISTRIBUTION AND AcrNITY OF ARGENTINE ANT

1 kIn

Incidence of Argentine ant

bull 075

Ea 014-074

Do

825

Fig 1 Argentineant incidence at JasperRidgeBiologicalPreservein 9 surveysover 3 yrEach squarerepresents 1 haand incidencewascalculated for each ha as the proportion of 9 surveysin which the Argentineant was foundThe blacklinearoundthe perimeterof the figureis the preserveboundaryblacklinesin the interiorof the preserverepresent streams[mda lake and a wetland area are shown in gray

Axis 11= -035 e + 09 v - 114 s - 075 i

- 065 W [2]

where e = distance to edge v = elevation s = slopei == insolation and w = distance to water The coef-ficient of each describes the magnitude of its contri-bution to the axis Monte Carlo simulations deter-mined that both the 1st canonical axis alone and allaxes combined contdbute significantly to variationamong genera in incidence (P lt 001 for axis 1 P lt001 for the whole model) Axis II was not significantso further axes were discounted

In the ordination diagram (Fig 2) abiotic variablesare represented by arrows Arrows point toward gen-era that tend to occur in areas with high values of theabiotic variable represented The projection of eacharrow on an axis indicates the magnitude of its con-tribution to that axis and arrows are numbered by thevalue of the Axis I canonical coefficients of the abioticfactors they represent For example although arrow 5representing the variable slope is relatively long in thedirection of Axis II its projection on Axis I is small

reflecting a small canonical coefficient in that axis(-004 equation 1)

Distance to the preserve edge weighted mostheavily in Axis I accounted for most of the varHltionamong genera in distribution (canonical coefficient =075 in axis I P = 001 Monte Carlo permutations)The arrow for distance to the preserve edge (arrow 1)points generally to the right indicating that ant generathat tend to occur far from the edge of the preserve fallon the right side of the ordination diagram (Fig 2)Crematogaster spp Formica spp and M andrei aregenerally found in areas far from the preserve edgewhereas Argentine ants and Liometopum occidentaleEmery are generally found much closer to the edgePrenolepis imparis Mayr and Leptothorax spp werefound in areas at intermediate distances from the edgeA dotted line drawn perpendicular to the arrow andthrough the origin (1 in Fig 2) separates the Argen-tine ant (LH) from most native ant genera

Elevation contributed significantly to differencesamong ant genera in distribution (P = 001 MonteCarlo permutations) Ants found at highest elevations

TS CSPC

826

LH

La

(Insolation) -i 4

l

ENVIRONMENTAL ENTOMOLOGY

Axis II

2 (Elevation)

ST

3 (Distance to water)

SM

MA

1 (Distance to edge)

Vol 27 no 4

Axis

Fig 2 Canonical correspondence ordination of ant incidence data during 9 surveys Abiotic factors are represented byvectors as follows 1Distance to edge 2 elevation 3 distance to water 4 insolation 5 slope To aid in the visual interpretationof the diagram dotted line l is drawn perpendicular to vector 1 Ant genera that fall on opposite sides of this line occuron average in areas with different values of the abiotic factor represented distance to edge Axis IT is plotted for conventiononly its contribution to the variance in ant distribution is minimal CC Crematogaster spp CS C semitestaceus FM Formicaspp LH L humile LO L occidentale LP Leptothorax spp MA Missor andrei PC Pheidole californica PI Prenolepis imparisSM SnlOlesta and ST Stenamma spp LO PC ST and SM (small letters) were rarely discovered in ant surveys CC CS FMLH MA PI and TS (large letters) were discovered frequently

included Stenamma spp and Camponotus spp andthose found in lower elevations included Solenopsismolesta Mayr L occidentale and the Argentine ant

Insolation and distance to water explained small butsignificant amounts of variation among genera in dis-tribution (P = 001 for insolation P = 002 fordistance to water) S molesta and the Argentine antwere found in areas with the highest insolation and Pimparis Leptothomx spp and the Argentine ant werefound closest to water Slope did not explain any vari-ation in distribution among ant genera (P = 023)

Of the 7 native ant species sampled frequentlyshown in large letters in Fig 2 P imparis is mostsimilar to the Argentine ant P imparis is the 1 speciesof native ant often seen foraging in the vicinity ofArgentine ant nests P imparis colonies are active fromNovember to June during times when most native antcolonies are not active and when the Argentine antsare frequently inactive L occidentale sampled rarely

is also close to the Argentine ant on the ordinationdiagranl Like the Argentine ant L occidentale is adoichoderine ant that makes long trails as it foragesfor the liquid exudate of insects (Wheeler andWheeler 1986)

When the effects of distance to edge elevation anddistance to water were each plotted separately theresults supported the patterns suggested by the ca-nonical correspondence analysis On average the Ar-gentine ant was most strongly established closer to thepreserve edge closer to water and in areas at lowerelevation (Fig 3)

Activity Patterns Soil temperature relative humid-ity and air temperature explained a significant pro-portion of the variation in the distribution of ant gen-era (Fig 4) Eigenvalues of Axis I and II were 019 and003 respectively indicating that the linear combina-tion of abiotic factors in Axis I accounted for most of

Fig 3 Boxplots of the relationshipbetween degree ofestablishment by the Argentine ant and 3 environmentalvariables (a-c) Each dot represents 1 ha and the box sur-rounds50ofallobservationsHorizontallinesinthe centersof the boxes are medians and diamonds show the meanvalues (center of the diamond) and standard error of themean (height of the diamond) The right column in eachgraph represents hectares where Argentine ants are wellestablished(found in 75-100of all surveysover 3 yr) Themiddle column represents transitionalhectares (Argentineants found in 14-74of all surveys) and the left columnrepresents hectares in which Argentine ants were neverobserved

827

axis I = 101 Ts + 094 Rh + 003 Ta [3]

axis II = 081 Ts - 072 Rh - 049 Ta [4]

where Ts is soil temperature in degrees centigrade Rhis relative humidity in percent and Ta is air temper-ature in degrees centigrade Monte Carlo simulationsdetermined that both the 1st canonical axis alone andall axes combined contributed significantly to varia-tion between species in activity (P lt 001 for axis IP lt 001 for the whole model 99 permutations) Thelow eigenvalue of axis I in this canonical correspon-dence analysis reveals that there is substantial overlapamong ant species in abiotic conditions at which theyare active Axis II was not significant so further axeswere discounted

All 3 of the abiotic factors contributed significantlyto variation in colony activity (P = 001 for soiltemperature P = 001 for relative humidity P = 001for air temperature Monte Carlo permutations) Soiltemperature and relative humidity represented moststrongly in the 1st canonical axis account for most ofthe differences among colonies in activity patternsThe outlying position of F subpalita in the ordinationdiagram (Fig 4) is probably partly a result of lowoverall activity of F subpolita colonies relative to col-onies of the other species tested

The Argentine ant was somewhat more likely thanthe native ant species to be active in higher soil tem-peratures up to 70degCColonies of P calif arnica and Csemitestaceus both of which are crepuscular and noc-turnal did not forage in temperatures gt40degC F sub-palita and Mandrei remained active up to =60degC (FigSa) Air temperature measured with a thermocoupleappeared to be highly correlated with infrared-de-rived soil temperature (Fig 6c) so it is not surprisingthat for all species the relations of colony activity toair and soil temperature were similar (Fig 5 a and b)Colonies of the Argentine ant were active in air tem-peratures up to 45degC as were colonies of M andreiwhereas other native ant species were more limited byair temperature (Fig 5b)

Relative humidity may limit slightly the activity ofall species (Fig 5c) all ant species were more likelyto be inactive when relative humidity was lt40 (Fig5c) The Argentine ant was slightly more likely thannative species to be active when relative humidity washigh (Fig 5c)

Time of day influenced colony activity C semi-testaceus M andrei and P californica were usuallyinactive for some time in the middle of the day thoughM andrei only for a short period (Fig 5c) The Ar-gentine ants were often active throughout the day

Overall during the summer Argentine ants are ac-tive in a wider range of temperatures and relativehumidities than colonies of C semitestaceus and Fsubpalita and in a slightly wider range of tempera-tures relative humidities and times than P californica(Fig 5 a-d) Colonies of M andrei are active undermost of the same conditions as the Argentine ants

the variance in colony activity explained by the modelThe linear combinations were as follows

HUMAN ET AL DISTRIBUTION AND ACTIVITY OF ARGENTINE ANT

noLh transitional Lh well-established

Degree of L humile establishment

a

Au~ust 1998

S--QJ 800beclQJQJ

~ 600QJltfJQJbull0-0 400bullQJU

~ 200~Cl

0

~-g 180sect 160tll1l

j 140QJ

~ 120QJ

E100

80

60500-g 40Jbull

QJbulll1l~ 3008Q)

~ 200bull~Cl 100

828 ENVIRONMENTAL ENTOMOLOGY

Axis IIFS

Vol 27 no 4

MA PC

CS

1 (Soil temperature)

2 (Relative humidity)

Axis I

Fig 4 CANOCO ordination of ant colony activity data Abiotic factors are represented by vectors as follows 1 soiltemperature 2 relative humidity and 3 air temperature AxisII is plotted for convention only its contribution to the variancein ant distribution is minimal CS Camponotus semitestaceus FS F subpolita LH Linepithema hwnile MAMessor andrei PCPheidole califomico

with the exception of a short period of time in themiddle of the day characterized by high soil temper-atures when Mandrei colonies are usually inactive fora few hours but the Argentine ants are not (Fig 5 a andd)

There are many sets of abiotic conditions in whichcolonies of native and Argentine ants are both active(Fig 5) For example whenever Pcalif arnica coloniesare active during the day so are the Argentine ants(Fig 5d) There are no values of any of the abioticfactors measured in which the native ant species weremore likely to be active than the Argentine ants

Abiotic factors all depend upon each other Relativehumidity was high at low air temperatures (Fig 6a)soil temperature increases from sunrise to middaybefore decreasing again (Fig 6b) and air and soiltemperatures are positively correlated (Fig 6c) Ca-nonical correspondence analysis takes the interactionof variables into account

Diet Overlap The 2 seed-eating ants tested Mandrei and P calif arnica overlapped only slightly indiet with the Argentine ant (0 = 0011 and 0015

respectively Table 1) Most of the forage of M andreiand P califarnica was seeds although like the Argen-tine ants both the seed-eating species also broughtsome dead insects back to the colony (Table 1) Fsubpalita and the Argentine ant overlap only in thecollection of insects (Table 1) but insects made up afar larger fraction of the forage of F subpolita (0631)than of the Argentine ant (0055) The diet of Csemitestaceus was similar to that of the Argentine antboth species primarily collected liquid food

Discussion

Distribution of Ants at Jasper Ridge At JasperRidge the Argentine ant ismost likely to occur in areasthat are close to the preserve edge and areas at lowelevation whereas native ant species are more abun-dant in the center of the preserve and at higher ele-vation

The association of the Argentine ant with areas nearthe preserve edge may be due to its response to dis-turbance and to its mechanism of dispersal Many

August 1998 HUMAN ET AL DISTRIBUTION AND ACTIVlTI OF ARGENTINE ANT 829

C semitestaceus F subpolita L humile M andrei P califarnica

I

0 I bull I r-~~720 1440o

bull pobullbull 1II bull

I bullbullbull - ~ bullbull~bull bull It

bull I304050607080

I~r ~ I

L70 10 30 50 70

bullbull

bull bull It ~-

bull

-r--

304050607080

-- bull

ItS ~

r bull -T

bull

~----

bull

720 1440

~ t bullbullbull I

I bullbull~ ~-- - ~

o

70 10 30 50 70 10 30 50

Soil temperature (0C)

bull10 bullbullbull

bull

bullbull

bull bull bullbull - -- bull bullbullbullbullbullbull bull bullbullbull bullbullbull- ~II bullbullbull - middot bull

bullbullbullbull middot bullbull bull bullbull bull i - bullbull bull 1bullbullbull bull 1bull ~ bull bull bullbull

30 40 50 60 70 80 30 40 50 60 70 80

Relative humidity ()

1440midnight

bull i~ ~ Imiddot

bullbull

20 30 40 50 10 20 30 40 50 10 20 30 40 50 10 20 30 40 50 10 20 30 40 50

Air Temperature (0C)

bull

720noon

4middot

bull

middot

bull-1bull

30 40 50 60 70 8e

bull

d

c200

50

a200

1150-

100 bull -jI~

50 -Ibull

10 30 50 70

b200

150 bull- bull100 middot

100

150

100

150

200

omidnight

Time (min from previous midnight)

Fig 5 The influence of the abiotic factors (a) soil temperature (b) air temperature (c) relative humidity and (d) timeon ant colony activity for 5 ant species Ant colony activity was measured as the numbers of ants entering and exiting a colonywithin 2 min Each dot on the figures represents 1 observation of 1 colony

exotic species including the Argentine ant more eas-ily invade areas that are disturbed by human activity(Ward 1987Elton 1958Simberloff1981) This may berelated both to increased opportunity for introduction

in areas with a great deal of human activity and lackof competitors or predators if native species are sen-sitive to disturbance and retreat It is likely that theArgentine ant moved into the reserve from the de-

830 ENVIRONMENTAL ENTOMOLOGY Vol 27 no 4

Time (min from midnight)

c80

E 60QJbull2bull 40QJ0Ell0

20Ul

00 10 20 30 40

Air temperature (0C)

1440

b80

~ 60QJ

2~ 40QJ0-Ell0

20Ul

00 720

a

C 80

c~ 60Elc 40QJ

iO 20Oip 00

0 0 0

0 10 20 30 40

Air temperature (0C)

Fig6 Examplesof the relationshipsamongabioticfactors (a-c) Eachdot represents 1 measurementof colonyactivityand the indicated abioticfactor

veloped surrounding areas that have already beeninvaded

The dispersal strategy of the Argentine ant mayfurther explain why it tends to occur near the edges ofthe preserve Mating flights have been observed onlyrarely in this species (Newell and Barber 1913 Skaife1961) Occasional spot introductions of Argentine antshave been described in Californias Central Valley(Holway 1995) This may have occurred through theaccidental introduction of queens by humans throughqueens or fragments of colonies rafting downstream infloods (Holway 1995) or through the founding of newcolonies by winged queens Over the last 3 yr we haveobserved no spot introductions In the Jasper RidgePreserve most areas along creek corridors are alreadyinvaded by Argentine ants and there is little oppor-tunity for the species to be introduced accidentallyinside the preserve because human activities within itare limited Our results suggest that the Argentine antoccurs mostly at the preserve edges because it is ex-panding its range primarily by spreading along theground through colony fission

Table 1 Composition of forage and diet overlap between Ar-gentine ants and 4 native ant species

Ant species

LH CS FS MA PC

Proportion of food itemsLiquid food 0916 0825 0000 0000 0000Seeds 0004 0004 0036 0765 0661Insects 0055 0102 0631 0093 0120Other 0025 0069 0333 0141 0219

Total number of food items1313 484 279 686 183

Proportion overlap with Argentine antLow 0997 0060 0011 0015High 0998 0066 0016 0023

In (C) values closest to 1 indicate the greatest degree of overlapThe low overlap value assumed that forage in the unidentified cate-gory (other) differed between species and the high value assumedthat forage in this category was identical CS C semitestacetls FS Fsltbpolita LH L hmile MA M andrei PC P califomica

It seems unlikely that previous disturbance of theareas surrounding the preserve had already caused thenative species to disappear from inside the preserveprior to invasion by Argentine ants Within the pre-serve native ant species and Argentine ants competeon the margin of the invasion and native ant speciesdisappear from invaded areas but persist elsewhere(Human and Gordon 1996 1997) All of the 9 hainvaded by the Argentine ant between September1993 and September 1995 were occupied in 1993 bynative ant species

Argentine ant incidence was high in low-elevationareas However elevation per se is unlikely to be thelimiting factor because the Argentine ant can befound in a few high-elevation areas of Jasper Ridgeand the species has been found up to 2000 m in theHawaiian islands (Zimmerman 1940 Cole et al 1992)Elevation may be related to another environmentalvariable such as soil moisture that affects the Argen-tine ant directly There may be a historical effect if theArgentine ant were introduced more frequently tolow-elevation areas

Studies in the Central Valley of California havesuggested that Argentine ant encroachment into nat-ural habitat is almost exclUSivelylimited to areas withpermanent sources of water and that the Argentine antis more limited by proximity to water than are nativeant species (Ward 1987) We found that several nativetaxa including Prenoepis imparis Stenamma spp andLeptothorax spp were as close to water on average aswere Argentine ants Many other taxa including Cam-ponotus spp P californica and Tapinoma sessile Smithwere found only marginally farther from water thanthe Argentine ant (Fig 2) Although Argentine antstend to be more strongly established in areas closer towater (Fig 1) there are many areas near water not yetinvaded and many areas far from water where Ar-gentine ants are firmly established The differencebetween the results of this study and previous onesmay be due to climatic differences the Central Valleyis much dryer and hotter especially in the summerthan Jasper Ridge which lies in the coastal foothills

August 1998 HUMAN ET AL DISTRIBUTION AND ACTIVITY OF ARGENTINE ANT 831

Colony Activity and Abiotic Conditions It may bethat the Argentine ant avoids temperature extremesby some behavioral means A laboratory study thatcompared Argentine ants and other ant species nativeto central coastal California suggested that the Argen-tine ant has a low thermal tolerance dying at tem-peratures close to 46degC (Tremper 1976) In the fieldArgentine ant nests frequently experience soil tem-peratures gt46degC and they are active up to 70degC It ispossible that the Argentine ant copes behaviorallywith heat more effectively than native species In thelaboratory M andrei workers died when exposed totemperatures gt485degC (Tremper 1976) We foundthat M andrei colonies were rarely active when soiltemperatures exceeded 50degC M andrei are large blackants that are often found on trails and open areaswhere they may be unable to avoid temperature ex-tremes

Compared with most native ant species Argentineants are active in a wide range of abiotic conditions(Fig 5) M andrei and P californica were active in arange of conditions similar to that of Argentine antsBy contrast the activities of C semitestaceus and Fsubpolita were confined to a narrower range of timesand temperatures Because the Argentine ant is usu-ally active whenever any native species is most nativespecies in invaded areas are likely to encounter theinvaders frequently

The diet of the Argentine ant overlaps with that ofseveral species (Table 1) Argentine ant nests areactive in a wider range of abiotic conditions than areC semitestaceus or F subpolita the species with whichit competes most for food Diet overlap with C semi-testaceus is almost complete Overlap with F subpolitamay be higher than measured in our study becauseprevious studies suggest that many Formica speciesincluding F subpalita collect insect and plant exudate(Cook 1953 Dumpert 1981 Wheeler and Wheeler1986 Cherix 1987) which makes up a significant pro-portion of the diet of Argentine ants Diet overlap withP calif arnica and M andrei was minimal We measuredthe diets of native ant species and Argentine ant spe-cies in different areas where resource availability mayhave varied affecting food choice Further work isneeded at an invasion front to investigate effects offood availability on diet overlap

The strong association of the Argentine ant with theedges of Jasper Ridge suggests that the most significantlimits to its range expansion may be opportunity forintroduction and its own mechanism of dispersal Dis-tance to water was less important than we expectedbut a study that spans a wider range of conditions thanours might discover a stronger effect of moistureLimiting human activity within nature preserves andcreating reserves with large areaperimeter ratios mayprevent invasion of interior areas by Argentine antsand other invasive species that tend to track humandisturbance Elevation or some parameter related toit is also important to Argentine ant distribution andfurther investigation of this might contribute to othermanagement practices The possibility of spatial au-tocorrelation should be kept in mind sites near to each

other are likely to have similar environmental condi-tions and similar histories of Argentine ant establish-ment

Abiotic factors explain much of the difference indistribution of Argentine and native ant species whichdo not overlap on a scale of hectares but little of thedifference in activity patterns At the edge of theinvasion on a scale of meters Argentine and native antspecies forage in the same areas with substantial over-lap in temporal patterns of foraging activity Theseresults suggest that whenever native ant colonies onthe edge of the invasion are active they are likely tointeract with Argentine ants Interactions between theinvasive and native ant species may promote displace-ment of the native species Frequent encounters maylead native species to retreat from food sources andpossibly to emigrate Where Argentine ants and nativeants overlap substantially in diet exploitation compe-tition could reduce the foraging success of native antcolonies

Acknowledgments

We thank Eric Bjorkstedt (Stanford University) MarkBrown (Stanford University) Diane Wagner (Stanford Uni-versity) and 2 anonymous reviewers for their comments onthe manuscript Kristen Duin (Stanford University) and LisaStadlen (Stanford University) were terrific field assistantsRobert Swierk (Center for Conservation Biology StanfordUniversity) provided technical assistance with GeographicalInformation Systems The work was supported by disserta-tion fellowships to KGH from National Science Foundationand the American Association of University Women grantsto KGH from the Jasper Ridge Biological Preserve TheConservation and Research Foundation and the Phi BetaKappa Northern California Chapter and USDA grant No95-37302-1885 to DMG

References Cited

Adams E S and IFA Trainello 1980 Chemical interfer-ence competition by Monomorium minimum Oecologia(Ber) 51 265-270

Andersen A N and A D Patel 1994 Meat ants as domi-nant members of Australian ant communities an exper-imental test of their influence on the foraging success andforager abundance of other species Oecologia (Ber) 9815-24

Anderson A N 1992 Regulation of momentary diversityby dominant species in exceptionally rich ant communi-ties of the Australian seasonal tropics Am Nat 140 401-420

Andrews E A 1927 Ant-mounds as to temperature andsunshine J of Morpho Physio 44 1-20

ArcView GIS 1996 ArcInfo 701 ArcView 21 Environ-mental Systems Research Institute Redlands CA

Baker H G and G L Stebbins 1965 The genetics ofcolonizing species Proceedings International Union ofBiological Sciences Symposia on General Biology Aca-demic New York

Banks W A and D F Williams 1989 Competitive dis-placement of Paratrechina longicomis (Latreille) (Hyme-noptera Formicidae) from baits by fire ants in MattoGrosso Brazi J Entomo Sci 24 381-391

TS CSPC

826

LH

La

(Insolation) -i 4

l

ENVIRONMENTAL ENTOMOLOGY

Axis II

2 (Elevation)

ST

3 (Distance to water)

SM

MA

1 (Distance to edge)

Vol 27 no 4

Axis

Fig 2 Canonical correspondence ordination of ant incidence data during 9 surveys Abiotic factors are represented byvectors as follows 1Distance to edge 2 elevation 3 distance to water 4 insolation 5 slope To aid in the visual interpretationof the diagram dotted line l is drawn perpendicular to vector 1 Ant genera that fall on opposite sides of this line occuron average in areas with different values of the abiotic factor represented distance to edge Axis IT is plotted for conventiononly its contribution to the variance in ant distribution is minimal CC Crematogaster spp CS C semitestaceus FM Formicaspp LH L humile LO L occidentale LP Leptothorax spp MA Missor andrei PC Pheidole californica PI Prenolepis imparisSM SnlOlesta and ST Stenamma spp LO PC ST and SM (small letters) were rarely discovered in ant surveys CC CS FMLH MA PI and TS (large letters) were discovered frequently

included Stenamma spp and Camponotus spp andthose found in lower elevations included Solenopsismolesta Mayr L occidentale and the Argentine ant

Insolation and distance to water explained small butsignificant amounts of variation among genera in dis-tribution (P = 001 for insolation P = 002 fordistance to water) S molesta and the Argentine antwere found in areas with the highest insolation and Pimparis Leptothomx spp and the Argentine ant werefound closest to water Slope did not explain any vari-ation in distribution among ant genera (P = 023)

Of the 7 native ant species sampled frequentlyshown in large letters in Fig 2 P imparis is mostsimilar to the Argentine ant P imparis is the 1 speciesof native ant often seen foraging in the vicinity ofArgentine ant nests P imparis colonies are active fromNovember to June during times when most native antcolonies are not active and when the Argentine antsare frequently inactive L occidentale sampled rarely

is also close to the Argentine ant on the ordinationdiagranl Like the Argentine ant L occidentale is adoichoderine ant that makes long trails as it foragesfor the liquid exudate of insects (Wheeler andWheeler 1986)

When the effects of distance to edge elevation anddistance to water were each plotted separately theresults supported the patterns suggested by the ca-nonical correspondence analysis On average the Ar-gentine ant was most strongly established closer to thepreserve edge closer to water and in areas at lowerelevation (Fig 3)

Activity Patterns Soil temperature relative humid-ity and air temperature explained a significant pro-portion of the variation in the distribution of ant gen-era (Fig 4) Eigenvalues of Axis I and II were 019 and003 respectively indicating that the linear combina-tion of abiotic factors in Axis I accounted for most of

Fig 3 Boxplots of the relationshipbetween degree ofestablishment by the Argentine ant and 3 environmentalvariables (a-c) Each dot represents 1 ha and the box sur-rounds50ofallobservationsHorizontallinesinthe centersof the boxes are medians and diamonds show the meanvalues (center of the diamond) and standard error of themean (height of the diamond) The right column in eachgraph represents hectares where Argentine ants are wellestablished(found in 75-100of all surveysover 3 yr) Themiddle column represents transitionalhectares (Argentineants found in 14-74of all surveys) and the left columnrepresents hectares in which Argentine ants were neverobserved

827

axis I = 101 Ts + 094 Rh + 003 Ta [3]

axis II = 081 Ts - 072 Rh - 049 Ta [4]

where Ts is soil temperature in degrees centigrade Rhis relative humidity in percent and Ta is air temper-ature in degrees centigrade Monte Carlo simulationsdetermined that both the 1st canonical axis alone andall axes combined contributed significantly to varia-tion between species in activity (P lt 001 for axis IP lt 001 for the whole model 99 permutations) Thelow eigenvalue of axis I in this canonical correspon-dence analysis reveals that there is substantial overlapamong ant species in abiotic conditions at which theyare active Axis II was not significant so further axeswere discounted

All 3 of the abiotic factors contributed significantlyto variation in colony activity (P = 001 for soiltemperature P = 001 for relative humidity P = 001for air temperature Monte Carlo permutations) Soiltemperature and relative humidity represented moststrongly in the 1st canonical axis account for most ofthe differences among colonies in activity patternsThe outlying position of F subpalita in the ordinationdiagram (Fig 4) is probably partly a result of lowoverall activity of F subpolita colonies relative to col-onies of the other species tested

The Argentine ant was somewhat more likely thanthe native ant species to be active in higher soil tem-peratures up to 70degCColonies of P calif arnica and Csemitestaceus both of which are crepuscular and noc-turnal did not forage in temperatures gt40degC F sub-palita and Mandrei remained active up to =60degC (FigSa) Air temperature measured with a thermocoupleappeared to be highly correlated with infrared-de-rived soil temperature (Fig 6c) so it is not surprisingthat for all species the relations of colony activity toair and soil temperature were similar (Fig 5 a and b)Colonies of the Argentine ant were active in air tem-peratures up to 45degC as were colonies of M andreiwhereas other native ant species were more limited byair temperature (Fig 5b)

Relative humidity may limit slightly the activity ofall species (Fig 5c) all ant species were more likelyto be inactive when relative humidity was lt40 (Fig5c) The Argentine ant was slightly more likely thannative species to be active when relative humidity washigh (Fig 5c)

Time of day influenced colony activity C semi-testaceus M andrei and P californica were usuallyinactive for some time in the middle of the day thoughM andrei only for a short period (Fig 5c) The Ar-gentine ants were often active throughout the day

Overall during the summer Argentine ants are ac-tive in a wider range of temperatures and relativehumidities than colonies of C semitestaceus and Fsubpalita and in a slightly wider range of tempera-tures relative humidities and times than P californica(Fig 5 a-d) Colonies of M andrei are active undermost of the same conditions as the Argentine ants

the variance in colony activity explained by the modelThe linear combinations were as follows

HUMAN ET AL DISTRIBUTION AND ACTIVITY OF ARGENTINE ANT

noLh transitional Lh well-established

Degree of L humile establishment

a

Au~ust 1998

S--QJ 800beclQJQJ

~ 600QJltfJQJbull0-0 400bullQJU

~ 200~Cl

0

~-g 180sect 160tll1l

j 140QJ

~ 120QJ

E100

80

60500-g 40Jbull

QJbulll1l~ 3008Q)

~ 200bull~Cl 100

828 ENVIRONMENTAL ENTOMOLOGY

Axis IIFS

Vol 27 no 4

MA PC

CS

1 (Soil temperature)

2 (Relative humidity)

Axis I

Fig 4 CANOCO ordination of ant colony activity data Abiotic factors are represented by vectors as follows 1 soiltemperature 2 relative humidity and 3 air temperature AxisII is plotted for convention only its contribution to the variancein ant distribution is minimal CS Camponotus semitestaceus FS F subpolita LH Linepithema hwnile MAMessor andrei PCPheidole califomico

with the exception of a short period of time in themiddle of the day characterized by high soil temper-atures when Mandrei colonies are usually inactive fora few hours but the Argentine ants are not (Fig 5 a andd)

There are many sets of abiotic conditions in whichcolonies of native and Argentine ants are both active(Fig 5) For example whenever Pcalif arnica coloniesare active during the day so are the Argentine ants(Fig 5d) There are no values of any of the abioticfactors measured in which the native ant species weremore likely to be active than the Argentine ants

Abiotic factors all depend upon each other Relativehumidity was high at low air temperatures (Fig 6a)soil temperature increases from sunrise to middaybefore decreasing again (Fig 6b) and air and soiltemperatures are positively correlated (Fig 6c) Ca-nonical correspondence analysis takes the interactionof variables into account

Diet Overlap The 2 seed-eating ants tested Mandrei and P calif arnica overlapped only slightly indiet with the Argentine ant (0 = 0011 and 0015

respectively Table 1) Most of the forage of M andreiand P califarnica was seeds although like the Argen-tine ants both the seed-eating species also broughtsome dead insects back to the colony (Table 1) Fsubpalita and the Argentine ant overlap only in thecollection of insects (Table 1) but insects made up afar larger fraction of the forage of F subpolita (0631)than of the Argentine ant (0055) The diet of Csemitestaceus was similar to that of the Argentine antboth species primarily collected liquid food

Discussion

Distribution of Ants at Jasper Ridge At JasperRidge the Argentine ant ismost likely to occur in areasthat are close to the preserve edge and areas at lowelevation whereas native ant species are more abun-dant in the center of the preserve and at higher ele-vation

The association of the Argentine ant with areas nearthe preserve edge may be due to its response to dis-turbance and to its mechanism of dispersal Many

August 1998 HUMAN ET AL DISTRIBUTION AND ACTIVlTI OF ARGENTINE ANT 829

C semitestaceus F subpolita L humile M andrei P califarnica

I

0 I bull I r-~~720 1440o

bull pobullbull 1II bull

I bullbullbull - ~ bullbull~bull bull It

bull I304050607080

I~r ~ I

L70 10 30 50 70

bullbull

bull bull It ~-

bull

-r--

304050607080

-- bull

ItS ~

r bull -T

bull

~----

bull

720 1440

~ t bullbullbull I

I bullbull~ ~-- - ~

o

70 10 30 50 70 10 30 50

Soil temperature (0C)

bull10 bullbullbull

bull

bullbull

bull bull bullbull - -- bull bullbullbullbullbullbull bull bullbullbull bullbullbull- ~II bullbullbull - middot bull

bullbullbullbull middot bullbull bull bullbull bull i - bullbull bull 1bullbullbull bull 1bull ~ bull bull bullbull

30 40 50 60 70 80 30 40 50 60 70 80

Relative humidity ()

1440midnight

bull i~ ~ Imiddot

bullbull

20 30 40 50 10 20 30 40 50 10 20 30 40 50 10 20 30 40 50 10 20 30 40 50

Air Temperature (0C)

bull

720noon

4middot

bull

middot

bull-1bull

30 40 50 60 70 8e

bull

d

c200

50

a200

1150-

100 bull -jI~

50 -Ibull

10 30 50 70

b200

150 bull- bull100 middot

100

150

100

150

200

omidnight

Time (min from previous midnight)

Fig 5 The influence of the abiotic factors (a) soil temperature (b) air temperature (c) relative humidity and (d) timeon ant colony activity for 5 ant species Ant colony activity was measured as the numbers of ants entering and exiting a colonywithin 2 min Each dot on the figures represents 1 observation of 1 colony

exotic species including the Argentine ant more eas-ily invade areas that are disturbed by human activity(Ward 1987Elton 1958Simberloff1981) This may berelated both to increased opportunity for introduction

in areas with a great deal of human activity and lackof competitors or predators if native species are sen-sitive to disturbance and retreat It is likely that theArgentine ant moved into the reserve from the de-

830 ENVIRONMENTAL ENTOMOLOGY Vol 27 no 4

Time (min from midnight)

c80

E 60QJbull2bull 40QJ0Ell0

20Ul

00 10 20 30 40

Air temperature (0C)

1440

b80

~ 60QJ

2~ 40QJ0-Ell0

20Ul

00 720

a

C 80

c~ 60Elc 40QJ

iO 20Oip 00

0 0 0

0 10 20 30 40

Air temperature (0C)

Fig6 Examplesof the relationshipsamongabioticfactors (a-c) Eachdot represents 1 measurementof colonyactivityand the indicated abioticfactor

veloped surrounding areas that have already beeninvaded

The dispersal strategy of the Argentine ant mayfurther explain why it tends to occur near the edges ofthe preserve Mating flights have been observed onlyrarely in this species (Newell and Barber 1913 Skaife1961) Occasional spot introductions of Argentine antshave been described in Californias Central Valley(Holway 1995) This may have occurred through theaccidental introduction of queens by humans throughqueens or fragments of colonies rafting downstream infloods (Holway 1995) or through the founding of newcolonies by winged queens Over the last 3 yr we haveobserved no spot introductions In the Jasper RidgePreserve most areas along creek corridors are alreadyinvaded by Argentine ants and there is little oppor-tunity for the species to be introduced accidentallyinside the preserve because human activities within itare limited Our results suggest that the Argentine antoccurs mostly at the preserve edges because it is ex-panding its range primarily by spreading along theground through colony fission

Table 1 Composition of forage and diet overlap between Ar-gentine ants and 4 native ant species

Ant species

LH CS FS MA PC

Proportion of food itemsLiquid food 0916 0825 0000 0000 0000Seeds 0004 0004 0036 0765 0661Insects 0055 0102 0631 0093 0120Other 0025 0069 0333 0141 0219

Total number of food items1313 484 279 686 183

Proportion overlap with Argentine antLow 0997 0060 0011 0015High 0998 0066 0016 0023

In (C) values closest to 1 indicate the greatest degree of overlapThe low overlap value assumed that forage in the unidentified cate-gory (other) differed between species and the high value assumedthat forage in this category was identical CS C semitestacetls FS Fsltbpolita LH L hmile MA M andrei PC P califomica

It seems unlikely that previous disturbance of theareas surrounding the preserve had already caused thenative species to disappear from inside the preserveprior to invasion by Argentine ants Within the pre-serve native ant species and Argentine ants competeon the margin of the invasion and native ant speciesdisappear from invaded areas but persist elsewhere(Human and Gordon 1996 1997) All of the 9 hainvaded by the Argentine ant between September1993 and September 1995 were occupied in 1993 bynative ant species

Argentine ant incidence was high in low-elevationareas However elevation per se is unlikely to be thelimiting factor because the Argentine ant can befound in a few high-elevation areas of Jasper Ridgeand the species has been found up to 2000 m in theHawaiian islands (Zimmerman 1940 Cole et al 1992)Elevation may be related to another environmentalvariable such as soil moisture that affects the Argen-tine ant directly There may be a historical effect if theArgentine ant were introduced more frequently tolow-elevation areas

Studies in the Central Valley of California havesuggested that Argentine ant encroachment into nat-ural habitat is almost exclUSivelylimited to areas withpermanent sources of water and that the Argentine antis more limited by proximity to water than are nativeant species (Ward 1987) We found that several nativetaxa including Prenoepis imparis Stenamma spp andLeptothorax spp were as close to water on average aswere Argentine ants Many other taxa including Cam-ponotus spp P californica and Tapinoma sessile Smithwere found only marginally farther from water thanthe Argentine ant (Fig 2) Although Argentine antstend to be more strongly established in areas closer towater (Fig 1) there are many areas near water not yetinvaded and many areas far from water where Ar-gentine ants are firmly established The differencebetween the results of this study and previous onesmay be due to climatic differences the Central Valleyis much dryer and hotter especially in the summerthan Jasper Ridge which lies in the coastal foothills

August 1998 HUMAN ET AL DISTRIBUTION AND ACTIVITY OF ARGENTINE ANT 831

Colony Activity and Abiotic Conditions It may bethat the Argentine ant avoids temperature extremesby some behavioral means A laboratory study thatcompared Argentine ants and other ant species nativeto central coastal California suggested that the Argen-tine ant has a low thermal tolerance dying at tem-peratures close to 46degC (Tremper 1976) In the fieldArgentine ant nests frequently experience soil tem-peratures gt46degC and they are active up to 70degC It ispossible that the Argentine ant copes behaviorallywith heat more effectively than native species In thelaboratory M andrei workers died when exposed totemperatures gt485degC (Tremper 1976) We foundthat M andrei colonies were rarely active when soiltemperatures exceeded 50degC M andrei are large blackants that are often found on trails and open areaswhere they may be unable to avoid temperature ex-tremes

Compared with most native ant species Argentineants are active in a wide range of abiotic conditions(Fig 5) M andrei and P californica were active in arange of conditions similar to that of Argentine antsBy contrast the activities of C semitestaceus and Fsubpolita were confined to a narrower range of timesand temperatures Because the Argentine ant is usu-ally active whenever any native species is most nativespecies in invaded areas are likely to encounter theinvaders frequently

The diet of the Argentine ant overlaps with that ofseveral species (Table 1) Argentine ant nests areactive in a wider range of abiotic conditions than areC semitestaceus or F subpolita the species with whichit competes most for food Diet overlap with C semi-testaceus is almost complete Overlap with F subpolitamay be higher than measured in our study becauseprevious studies suggest that many Formica speciesincluding F subpalita collect insect and plant exudate(Cook 1953 Dumpert 1981 Wheeler and Wheeler1986 Cherix 1987) which makes up a significant pro-portion of the diet of Argentine ants Diet overlap withP calif arnica and M andrei was minimal We measuredthe diets of native ant species and Argentine ant spe-cies in different areas where resource availability mayhave varied affecting food choice Further work isneeded at an invasion front to investigate effects offood availability on diet overlap

The strong association of the Argentine ant with theedges of Jasper Ridge suggests that the most significantlimits to its range expansion may be opportunity forintroduction and its own mechanism of dispersal Dis-tance to water was less important than we expectedbut a study that spans a wider range of conditions thanours might discover a stronger effect of moistureLimiting human activity within nature preserves andcreating reserves with large areaperimeter ratios mayprevent invasion of interior areas by Argentine antsand other invasive species that tend to track humandisturbance Elevation or some parameter related toit is also important to Argentine ant distribution andfurther investigation of this might contribute to othermanagement practices The possibility of spatial au-tocorrelation should be kept in mind sites near to each

other are likely to have similar environmental condi-tions and similar histories of Argentine ant establish-ment

Abiotic factors explain much of the difference indistribution of Argentine and native ant species whichdo not overlap on a scale of hectares but little of thedifference in activity patterns At the edge of theinvasion on a scale of meters Argentine and native antspecies forage in the same areas with substantial over-lap in temporal patterns of foraging activity Theseresults suggest that whenever native ant colonies onthe edge of the invasion are active they are likely tointeract with Argentine ants Interactions between theinvasive and native ant species may promote displace-ment of the native species Frequent encounters maylead native species to retreat from food sources andpossibly to emigrate Where Argentine ants and nativeants overlap substantially in diet exploitation compe-tition could reduce the foraging success of native antcolonies

Acknowledgments

We thank Eric Bjorkstedt (Stanford University) MarkBrown (Stanford University) Diane Wagner (Stanford Uni-versity) and 2 anonymous reviewers for their comments onthe manuscript Kristen Duin (Stanford University) and LisaStadlen (Stanford University) were terrific field assistantsRobert Swierk (Center for Conservation Biology StanfordUniversity) provided technical assistance with GeographicalInformation Systems The work was supported by disserta-tion fellowships to KGH from National Science Foundationand the American Association of University Women grantsto KGH from the Jasper Ridge Biological Preserve TheConservation and Research Foundation and the Phi BetaKappa Northern California Chapter and USDA grant No95-37302-1885 to DMG

References Cited

Adams E S and IFA Trainello 1980 Chemical interfer-ence competition by Monomorium minimum Oecologia(Ber) 51 265-270

Andersen A N and A D Patel 1994 Meat ants as domi-nant members of Australian ant communities an exper-imental test of their influence on the foraging success andforager abundance of other species Oecologia (Ber) 9815-24

Anderson A N 1992 Regulation of momentary diversityby dominant species in exceptionally rich ant communi-ties of the Australian seasonal tropics Am Nat 140 401-420

Andrews E A 1927 Ant-mounds as to temperature andsunshine J of Morpho Physio 44 1-20

ArcView GIS 1996 ArcInfo 701 ArcView 21 Environ-mental Systems Research Institute Redlands CA

Baker H G and G L Stebbins 1965 The genetics ofcolonizing species Proceedings International Union ofBiological Sciences Symposia on General Biology Aca-demic New York

Banks W A and D F Williams 1989 Competitive dis-placement of Paratrechina longicomis (Latreille) (Hyme-noptera Formicidae) from baits by fire ants in MattoGrosso Brazi J Entomo Sci 24 381-391

Fig 3 Boxplots of the relationshipbetween degree ofestablishment by the Argentine ant and 3 environmentalvariables (a-c) Each dot represents 1 ha and the box sur-rounds50ofallobservationsHorizontallinesinthe centersof the boxes are medians and diamonds show the meanvalues (center of the diamond) and standard error of themean (height of the diamond) The right column in eachgraph represents hectares where Argentine ants are wellestablished(found in 75-100of all surveysover 3 yr) Themiddle column represents transitionalhectares (Argentineants found in 14-74of all surveys) and the left columnrepresents hectares in which Argentine ants were neverobserved

827

axis I = 101 Ts + 094 Rh + 003 Ta [3]

axis II = 081 Ts - 072 Rh - 049 Ta [4]

where Ts is soil temperature in degrees centigrade Rhis relative humidity in percent and Ta is air temper-ature in degrees centigrade Monte Carlo simulationsdetermined that both the 1st canonical axis alone andall axes combined contributed significantly to varia-tion between species in activity (P lt 001 for axis IP lt 001 for the whole model 99 permutations) Thelow eigenvalue of axis I in this canonical correspon-dence analysis reveals that there is substantial overlapamong ant species in abiotic conditions at which theyare active Axis II was not significant so further axeswere discounted

All 3 of the abiotic factors contributed significantlyto variation in colony activity (P = 001 for soiltemperature P = 001 for relative humidity P = 001for air temperature Monte Carlo permutations) Soiltemperature and relative humidity represented moststrongly in the 1st canonical axis account for most ofthe differences among colonies in activity patternsThe outlying position of F subpalita in the ordinationdiagram (Fig 4) is probably partly a result of lowoverall activity of F subpolita colonies relative to col-onies of the other species tested

The Argentine ant was somewhat more likely thanthe native ant species to be active in higher soil tem-peratures up to 70degCColonies of P calif arnica and Csemitestaceus both of which are crepuscular and noc-turnal did not forage in temperatures gt40degC F sub-palita and Mandrei remained active up to =60degC (FigSa) Air temperature measured with a thermocoupleappeared to be highly correlated with infrared-de-rived soil temperature (Fig 6c) so it is not surprisingthat for all species the relations of colony activity toair and soil temperature were similar (Fig 5 a and b)Colonies of the Argentine ant were active in air tem-peratures up to 45degC as were colonies of M andreiwhereas other native ant species were more limited byair temperature (Fig 5b)

Relative humidity may limit slightly the activity ofall species (Fig 5c) all ant species were more likelyto be inactive when relative humidity was lt40 (Fig5c) The Argentine ant was slightly more likely thannative species to be active when relative humidity washigh (Fig 5c)

Time of day influenced colony activity C semi-testaceus M andrei and P californica were usuallyinactive for some time in the middle of the day thoughM andrei only for a short period (Fig 5c) The Ar-gentine ants were often active throughout the day

Overall during the summer Argentine ants are ac-tive in a wider range of temperatures and relativehumidities than colonies of C semitestaceus and Fsubpalita and in a slightly wider range of tempera-tures relative humidities and times than P californica(Fig 5 a-d) Colonies of M andrei are active undermost of the same conditions as the Argentine ants

the variance in colony activity explained by the modelThe linear combinations were as follows

HUMAN ET AL DISTRIBUTION AND ACTIVITY OF ARGENTINE ANT

noLh transitional Lh well-established

Degree of L humile establishment

a

Au~ust 1998

S--QJ 800beclQJQJ

~ 600QJltfJQJbull0-0 400bullQJU

~ 200~Cl

0

~-g 180sect 160tll1l

j 140QJ

~ 120QJ

E100

80

60500-g 40Jbull

QJbulll1l~ 3008Q)

~ 200bull~Cl 100

828 ENVIRONMENTAL ENTOMOLOGY

Axis IIFS

Vol 27 no 4

MA PC

CS

1 (Soil temperature)

2 (Relative humidity)

Axis I

Fig 4 CANOCO ordination of ant colony activity data Abiotic factors are represented by vectors as follows 1 soiltemperature 2 relative humidity and 3 air temperature AxisII is plotted for convention only its contribution to the variancein ant distribution is minimal CS Camponotus semitestaceus FS F subpolita LH Linepithema hwnile MAMessor andrei PCPheidole califomico

with the exception of a short period of time in themiddle of the day characterized by high soil temper-atures when Mandrei colonies are usually inactive fora few hours but the Argentine ants are not (Fig 5 a andd)

There are many sets of abiotic conditions in whichcolonies of native and Argentine ants are both active(Fig 5) For example whenever Pcalif arnica coloniesare active during the day so are the Argentine ants(Fig 5d) There are no values of any of the abioticfactors measured in which the native ant species weremore likely to be active than the Argentine ants

Abiotic factors all depend upon each other Relativehumidity was high at low air temperatures (Fig 6a)soil temperature increases from sunrise to middaybefore decreasing again (Fig 6b) and air and soiltemperatures are positively correlated (Fig 6c) Ca-nonical correspondence analysis takes the interactionof variables into account

Diet Overlap The 2 seed-eating ants tested Mandrei and P calif arnica overlapped only slightly indiet with the Argentine ant (0 = 0011 and 0015

respectively Table 1) Most of the forage of M andreiand P califarnica was seeds although like the Argen-tine ants both the seed-eating species also broughtsome dead insects back to the colony (Table 1) Fsubpalita and the Argentine ant overlap only in thecollection of insects (Table 1) but insects made up afar larger fraction of the forage of F subpolita (0631)than of the Argentine ant (0055) The diet of Csemitestaceus was similar to that of the Argentine antboth species primarily collected liquid food

Discussion

Distribution of Ants at Jasper Ridge At JasperRidge the Argentine ant ismost likely to occur in areasthat are close to the preserve edge and areas at lowelevation whereas native ant species are more abun-dant in the center of the preserve and at higher ele-vation

The association of the Argentine ant with areas nearthe preserve edge may be due to its response to dis-turbance and to its mechanism of dispersal Many

August 1998 HUMAN ET AL DISTRIBUTION AND ACTIVlTI OF ARGENTINE ANT 829

C semitestaceus F subpolita L humile M andrei P califarnica

I

0 I bull I r-~~720 1440o

bull pobullbull 1II bull

I bullbullbull - ~ bullbull~bull bull It

bull I304050607080

I~r ~ I

L70 10 30 50 70

bullbull

bull bull It ~-

bull

-r--

304050607080

-- bull

ItS ~

r bull -T

bull

~----

bull

720 1440

~ t bullbullbull I

I bullbull~ ~-- - ~

o

70 10 30 50 70 10 30 50

Soil temperature (0C)

bull10 bullbullbull

bull

bullbull

bull bull bullbull - -- bull bullbullbullbullbullbull bull bullbullbull bullbullbull- ~II bullbullbull - middot bull

bullbullbullbull middot bullbull bull bullbull bull i - bullbull bull 1bullbullbull bull 1bull ~ bull bull bullbull

30 40 50 60 70 80 30 40 50 60 70 80

Relative humidity ()

1440midnight

bull i~ ~ Imiddot

bullbull

20 30 40 50 10 20 30 40 50 10 20 30 40 50 10 20 30 40 50 10 20 30 40 50

Air Temperature (0C)

bull

720noon

4middot

bull

middot

bull-1bull

30 40 50 60 70 8e

bull

d

c200

50

a200

1150-

100 bull -jI~

50 -Ibull

10 30 50 70

b200

150 bull- bull100 middot

100

150

100

150

200

omidnight

Time (min from previous midnight)

Fig 5 The influence of the abiotic factors (a) soil temperature (b) air temperature (c) relative humidity and (d) timeon ant colony activity for 5 ant species Ant colony activity was measured as the numbers of ants entering and exiting a colonywithin 2 min Each dot on the figures represents 1 observation of 1 colony

exotic species including the Argentine ant more eas-ily invade areas that are disturbed by human activity(Ward 1987Elton 1958Simberloff1981) This may berelated both to increased opportunity for introduction

in areas with a great deal of human activity and lackof competitors or predators if native species are sen-sitive to disturbance and retreat It is likely that theArgentine ant moved into the reserve from the de-

830 ENVIRONMENTAL ENTOMOLOGY Vol 27 no 4

Time (min from midnight)

c80

E 60QJbull2bull 40QJ0Ell0

20Ul

00 10 20 30 40

Air temperature (0C)

1440

b80

~ 60QJ

2~ 40QJ0-Ell0

20Ul

00 720

a

C 80

c~ 60Elc 40QJ

iO 20Oip 00

0 0 0

0 10 20 30 40

Air temperature (0C)

Fig6 Examplesof the relationshipsamongabioticfactors (a-c) Eachdot represents 1 measurementof colonyactivityand the indicated abioticfactor

veloped surrounding areas that have already beeninvaded

The dispersal strategy of the Argentine ant mayfurther explain why it tends to occur near the edges ofthe preserve Mating flights have been observed onlyrarely in this species (Newell and Barber 1913 Skaife1961) Occasional spot introductions of Argentine antshave been described in Californias Central Valley(Holway 1995) This may have occurred through theaccidental introduction of queens by humans throughqueens or fragments of colonies rafting downstream infloods (Holway 1995) or through the founding of newcolonies by winged queens Over the last 3 yr we haveobserved no spot introductions In the Jasper RidgePreserve most areas along creek corridors are alreadyinvaded by Argentine ants and there is little oppor-tunity for the species to be introduced accidentallyinside the preserve because human activities within itare limited Our results suggest that the Argentine antoccurs mostly at the preserve edges because it is ex-panding its range primarily by spreading along theground through colony fission

Table 1 Composition of forage and diet overlap between Ar-gentine ants and 4 native ant species

Ant species

LH CS FS MA PC

Proportion of food itemsLiquid food 0916 0825 0000 0000 0000Seeds 0004 0004 0036 0765 0661Insects 0055 0102 0631 0093 0120Other 0025 0069 0333 0141 0219

Total number of food items1313 484 279 686 183

Proportion overlap with Argentine antLow 0997 0060 0011 0015High 0998 0066 0016 0023

In (C) values closest to 1 indicate the greatest degree of overlapThe low overlap value assumed that forage in the unidentified cate-gory (other) differed between species and the high value assumedthat forage in this category was identical CS C semitestacetls FS Fsltbpolita LH L hmile MA M andrei PC P califomica

It seems unlikely that previous disturbance of theareas surrounding the preserve had already caused thenative species to disappear from inside the preserveprior to invasion by Argentine ants Within the pre-serve native ant species and Argentine ants competeon the margin of the invasion and native ant speciesdisappear from invaded areas but persist elsewhere(Human and Gordon 1996 1997) All of the 9 hainvaded by the Argentine ant between September1993 and September 1995 were occupied in 1993 bynative ant species

Argentine ant incidence was high in low-elevationareas However elevation per se is unlikely to be thelimiting factor because the Argentine ant can befound in a few high-elevation areas of Jasper Ridgeand the species has been found up to 2000 m in theHawaiian islands (Zimmerman 1940 Cole et al 1992)Elevation may be related to another environmentalvariable such as soil moisture that affects the Argen-tine ant directly There may be a historical effect if theArgentine ant were introduced more frequently tolow-elevation areas

Studies in the Central Valley of California havesuggested that Argentine ant encroachment into nat-ural habitat is almost exclUSivelylimited to areas withpermanent sources of water and that the Argentine antis more limited by proximity to water than are nativeant species (Ward 1987) We found that several nativetaxa including Prenoepis imparis Stenamma spp andLeptothorax spp were as close to water on average aswere Argentine ants Many other taxa including Cam-ponotus spp P californica and Tapinoma sessile Smithwere found only marginally farther from water thanthe Argentine ant (Fig 2) Although Argentine antstend to be more strongly established in areas closer towater (Fig 1) there are many areas near water not yetinvaded and many areas far from water where Ar-gentine ants are firmly established The differencebetween the results of this study and previous onesmay be due to climatic differences the Central Valleyis much dryer and hotter especially in the summerthan Jasper Ridge which lies in the coastal foothills

August 1998 HUMAN ET AL DISTRIBUTION AND ACTIVITY OF ARGENTINE ANT 831

Colony Activity and Abiotic Conditions It may bethat the Argentine ant avoids temperature extremesby some behavioral means A laboratory study thatcompared Argentine ants and other ant species nativeto central coastal California suggested that the Argen-tine ant has a low thermal tolerance dying at tem-peratures close to 46degC (Tremper 1976) In the fieldArgentine ant nests frequently experience soil tem-peratures gt46degC and they are active up to 70degC It ispossible that the Argentine ant copes behaviorallywith heat more effectively than native species In thelaboratory M andrei workers died when exposed totemperatures gt485degC (Tremper 1976) We foundthat M andrei colonies were rarely active when soiltemperatures exceeded 50degC M andrei are large blackants that are often found on trails and open areaswhere they may be unable to avoid temperature ex-tremes

Compared with most native ant species Argentineants are active in a wide range of abiotic conditions(Fig 5) M andrei and P californica were active in arange of conditions similar to that of Argentine antsBy contrast the activities of C semitestaceus and Fsubpolita were confined to a narrower range of timesand temperatures Because the Argentine ant is usu-ally active whenever any native species is most nativespecies in invaded areas are likely to encounter theinvaders frequently

The diet of the Argentine ant overlaps with that ofseveral species (Table 1) Argentine ant nests areactive in a wider range of abiotic conditions than areC semitestaceus or F subpolita the species with whichit competes most for food Diet overlap with C semi-testaceus is almost complete Overlap with F subpolitamay be higher than measured in our study becauseprevious studies suggest that many Formica speciesincluding F subpalita collect insect and plant exudate(Cook 1953 Dumpert 1981 Wheeler and Wheeler1986 Cherix 1987) which makes up a significant pro-portion of the diet of Argentine ants Diet overlap withP calif arnica and M andrei was minimal We measuredthe diets of native ant species and Argentine ant spe-cies in different areas where resource availability mayhave varied affecting food choice Further work isneeded at an invasion front to investigate effects offood availability on diet overlap

The strong association of the Argentine ant with theedges of Jasper Ridge suggests that the most significantlimits to its range expansion may be opportunity forintroduction and its own mechanism of dispersal Dis-tance to water was less important than we expectedbut a study that spans a wider range of conditions thanours might discover a stronger effect of moistureLimiting human activity within nature preserves andcreating reserves with large areaperimeter ratios mayprevent invasion of interior areas by Argentine antsand other invasive species that tend to track humandisturbance Elevation or some parameter related toit is also important to Argentine ant distribution andfurther investigation of this might contribute to othermanagement practices The possibility of spatial au-tocorrelation should be kept in mind sites near to each

other are likely to have similar environmental condi-tions and similar histories of Argentine ant establish-ment

Abiotic factors explain much of the difference indistribution of Argentine and native ant species whichdo not overlap on a scale of hectares but little of thedifference in activity patterns At the edge of theinvasion on a scale of meters Argentine and native antspecies forage in the same areas with substantial over-lap in temporal patterns of foraging activity Theseresults suggest that whenever native ant colonies onthe edge of the invasion are active they are likely tointeract with Argentine ants Interactions between theinvasive and native ant species may promote displace-ment of the native species Frequent encounters maylead native species to retreat from food sources andpossibly to emigrate Where Argentine ants and nativeants overlap substantially in diet exploitation compe-tition could reduce the foraging success of native antcolonies

Acknowledgments

We thank Eric Bjorkstedt (Stanford University) MarkBrown (Stanford University) Diane Wagner (Stanford Uni-versity) and 2 anonymous reviewers for their comments onthe manuscript Kristen Duin (Stanford University) and LisaStadlen (Stanford University) were terrific field assistantsRobert Swierk (Center for Conservation Biology StanfordUniversity) provided technical assistance with GeographicalInformation Systems The work was supported by disserta-tion fellowships to KGH from National Science Foundationand the American Association of University Women grantsto KGH from the Jasper Ridge Biological Preserve TheConservation and Research Foundation and the Phi BetaKappa Northern California Chapter and USDA grant No95-37302-1885 to DMG

References Cited

Adams E S and IFA Trainello 1980 Chemical interfer-ence competition by Monomorium minimum Oecologia(Ber) 51 265-270

Andersen A N and A D Patel 1994 Meat ants as domi-nant members of Australian ant communities an exper-imental test of their influence on the foraging success andforager abundance of other species Oecologia (Ber) 9815-24

Anderson A N 1992 Regulation of momentary diversityby dominant species in exceptionally rich ant communi-ties of the Australian seasonal tropics Am Nat 140 401-420

Andrews E A 1927 Ant-mounds as to temperature andsunshine J of Morpho Physio 44 1-20

ArcView GIS 1996 ArcInfo 701 ArcView 21 Environ-mental Systems Research Institute Redlands CA

Baker H G and G L Stebbins 1965 The genetics ofcolonizing species Proceedings International Union ofBiological Sciences Symposia on General Biology Aca-demic New York

Banks W A and D F Williams 1989 Competitive dis-placement of Paratrechina longicomis (Latreille) (Hyme-noptera Formicidae) from baits by fire ants in MattoGrosso Brazi J Entomo Sci 24 381-391

828 ENVIRONMENTAL ENTOMOLOGY

Axis IIFS

Vol 27 no 4

MA PC

CS

1 (Soil temperature)

2 (Relative humidity)

Axis I

Fig 4 CANOCO ordination of ant colony activity data Abiotic factors are represented by vectors as follows 1 soiltemperature 2 relative humidity and 3 air temperature AxisII is plotted for convention only its contribution to the variancein ant distribution is minimal CS Camponotus semitestaceus FS F subpolita LH Linepithema hwnile MAMessor andrei PCPheidole califomico

with the exception of a short period of time in themiddle of the day characterized by high soil temper-atures when Mandrei colonies are usually inactive fora few hours but the Argentine ants are not (Fig 5 a andd)

There are many sets of abiotic conditions in whichcolonies of native and Argentine ants are both active(Fig 5) For example whenever Pcalif arnica coloniesare active during the day so are the Argentine ants(Fig 5d) There are no values of any of the abioticfactors measured in which the native ant species weremore likely to be active than the Argentine ants

Abiotic factors all depend upon each other Relativehumidity was high at low air temperatures (Fig 6a)soil temperature increases from sunrise to middaybefore decreasing again (Fig 6b) and air and soiltemperatures are positively correlated (Fig 6c) Ca-nonical correspondence analysis takes the interactionof variables into account

Diet Overlap The 2 seed-eating ants tested Mandrei and P calif arnica overlapped only slightly indiet with the Argentine ant (0 = 0011 and 0015

respectively Table 1) Most of the forage of M andreiand P califarnica was seeds although like the Argen-tine ants both the seed-eating species also broughtsome dead insects back to the colony (Table 1) Fsubpalita and the Argentine ant overlap only in thecollection of insects (Table 1) but insects made up afar larger fraction of the forage of F subpolita (0631)than of the Argentine ant (0055) The diet of Csemitestaceus was similar to that of the Argentine antboth species primarily collected liquid food

Discussion

Distribution of Ants at Jasper Ridge At JasperRidge the Argentine ant ismost likely to occur in areasthat are close to the preserve edge and areas at lowelevation whereas native ant species are more abun-dant in the center of the preserve and at higher ele-vation

The association of the Argentine ant with areas nearthe preserve edge may be due to its response to dis-turbance and to its mechanism of dispersal Many

August 1998 HUMAN ET AL DISTRIBUTION AND ACTIVlTI OF ARGENTINE ANT 829

C semitestaceus F subpolita L humile M andrei P califarnica

I

0 I bull I r-~~720 1440o

bull pobullbull 1II bull

I bullbullbull - ~ bullbull~bull bull It

bull I304050607080

I~r ~ I

L70 10 30 50 70

bullbull

bull bull It ~-

bull

-r--

304050607080

-- bull

ItS ~

r bull -T

bull

~----

bull

720 1440

~ t bullbullbull I

I bullbull~ ~-- - ~

o

70 10 30 50 70 10 30 50

Soil temperature (0C)

bull10 bullbullbull

bull

bullbull

bull bull bullbull - -- bull bullbullbullbullbullbull bull bullbullbull bullbullbull- ~II bullbullbull - middot bull

bullbullbullbull middot bullbull bull bullbull bull i - bullbull bull 1bullbullbull bull 1bull ~ bull bull bullbull

30 40 50 60 70 80 30 40 50 60 70 80

Relative humidity ()

1440midnight

bull i~ ~ Imiddot

bullbull

20 30 40 50 10 20 30 40 50 10 20 30 40 50 10 20 30 40 50 10 20 30 40 50

Air Temperature (0C)

bull

720noon

4middot

bull

middot

bull-1bull

30 40 50 60 70 8e

bull

d

c200

50

a200

1150-

100 bull -jI~

50 -Ibull

10 30 50 70

b200

150 bull- bull100 middot

100

150

100

150

200

omidnight

Time (min from previous midnight)

Fig 5 The influence of the abiotic factors (a) soil temperature (b) air temperature (c) relative humidity and (d) timeon ant colony activity for 5 ant species Ant colony activity was measured as the numbers of ants entering and exiting a colonywithin 2 min Each dot on the figures represents 1 observation of 1 colony

exotic species including the Argentine ant more eas-ily invade areas that are disturbed by human activity(Ward 1987Elton 1958Simberloff1981) This may berelated both to increased opportunity for introduction

in areas with a great deal of human activity and lackof competitors or predators if native species are sen-sitive to disturbance and retreat It is likely that theArgentine ant moved into the reserve from the de-

830 ENVIRONMENTAL ENTOMOLOGY Vol 27 no 4

Time (min from midnight)

c80

E 60QJbull2bull 40QJ0Ell0

20Ul

00 10 20 30 40

Air temperature (0C)

1440

b80

~ 60QJ

2~ 40QJ0-Ell0

20Ul

00 720

a

C 80

c~ 60Elc 40QJ

iO 20Oip 00

0 0 0

0 10 20 30 40

Air temperature (0C)

Fig6 Examplesof the relationshipsamongabioticfactors (a-c) Eachdot represents 1 measurementof colonyactivityand the indicated abioticfactor

veloped surrounding areas that have already beeninvaded

The dispersal strategy of the Argentine ant mayfurther explain why it tends to occur near the edges ofthe preserve Mating flights have been observed onlyrarely in this species (Newell and Barber 1913 Skaife1961) Occasional spot introductions of Argentine antshave been described in Californias Central Valley(Holway 1995) This may have occurred through theaccidental introduction of queens by humans throughqueens or fragments of colonies rafting downstream infloods (Holway 1995) or through the founding of newcolonies by winged queens Over the last 3 yr we haveobserved no spot introductions In the Jasper RidgePreserve most areas along creek corridors are alreadyinvaded by Argentine ants and there is little oppor-tunity for the species to be introduced accidentallyinside the preserve because human activities within itare limited Our results suggest that the Argentine antoccurs mostly at the preserve edges because it is ex-panding its range primarily by spreading along theground through colony fission

Table 1 Composition of forage and diet overlap between Ar-gentine ants and 4 native ant species

Ant species

LH CS FS MA PC

Proportion of food itemsLiquid food 0916 0825 0000 0000 0000Seeds 0004 0004 0036 0765 0661Insects 0055 0102 0631 0093 0120Other 0025 0069 0333 0141 0219

Total number of food items1313 484 279 686 183

Proportion overlap with Argentine antLow 0997 0060 0011 0015High 0998 0066 0016 0023

In (C) values closest to 1 indicate the greatest degree of overlapThe low overlap value assumed that forage in the unidentified cate-gory (other) differed between species and the high value assumedthat forage in this category was identical CS C semitestacetls FS Fsltbpolita LH L hmile MA M andrei PC P califomica

It seems unlikely that previous disturbance of theareas surrounding the preserve had already caused thenative species to disappear from inside the preserveprior to invasion by Argentine ants Within the pre-serve native ant species and Argentine ants competeon the margin of the invasion and native ant speciesdisappear from invaded areas but persist elsewhere(Human and Gordon 1996 1997) All of the 9 hainvaded by the Argentine ant between September1993 and September 1995 were occupied in 1993 bynative ant species

Argentine ant incidence was high in low-elevationareas However elevation per se is unlikely to be thelimiting factor because the Argentine ant can befound in a few high-elevation areas of Jasper Ridgeand the species has been found up to 2000 m in theHawaiian islands (Zimmerman 1940 Cole et al 1992)Elevation may be related to another environmentalvariable such as soil moisture that affects the Argen-tine ant directly There may be a historical effect if theArgentine ant were introduced more frequently tolow-elevation areas

Studies in the Central Valley of California havesuggested that Argentine ant encroachment into nat-ural habitat is almost exclUSivelylimited to areas withpermanent sources of water and that the Argentine antis more limited by proximity to water than are nativeant species (Ward 1987) We found that several nativetaxa including Prenoepis imparis Stenamma spp andLeptothorax spp were as close to water on average aswere Argentine ants Many other taxa including Cam-ponotus spp P californica and Tapinoma sessile Smithwere found only marginally farther from water thanthe Argentine ant (Fig 2) Although Argentine antstend to be more strongly established in areas closer towater (Fig 1) there are many areas near water not yetinvaded and many areas far from water where Ar-gentine ants are firmly established The differencebetween the results of this study and previous onesmay be due to climatic differences the Central Valleyis much dryer and hotter especially in the summerthan Jasper Ridge which lies in the coastal foothills

August 1998 HUMAN ET AL DISTRIBUTION AND ACTIVITY OF ARGENTINE ANT 831

Colony Activity and Abiotic Conditions It may bethat the Argentine ant avoids temperature extremesby some behavioral means A laboratory study thatcompared Argentine ants and other ant species nativeto central coastal California suggested that the Argen-tine ant has a low thermal tolerance dying at tem-peratures close to 46degC (Tremper 1976) In the fieldArgentine ant nests frequently experience soil tem-peratures gt46degC and they are active up to 70degC It ispossible that the Argentine ant copes behaviorallywith heat more effectively than native species In thelaboratory M andrei workers died when exposed totemperatures gt485degC (Tremper 1976) We foundthat M andrei colonies were rarely active when soiltemperatures exceeded 50degC M andrei are large blackants that are often found on trails and open areaswhere they may be unable to avoid temperature ex-tremes

Compared with most native ant species Argentineants are active in a wide range of abiotic conditions(Fig 5) M andrei and P californica were active in arange of conditions similar to that of Argentine antsBy contrast the activities of C semitestaceus and Fsubpolita were confined to a narrower range of timesand temperatures Because the Argentine ant is usu-ally active whenever any native species is most nativespecies in invaded areas are likely to encounter theinvaders frequently

The diet of the Argentine ant overlaps with that ofseveral species (Table 1) Argentine ant nests areactive in a wider range of abiotic conditions than areC semitestaceus or F subpolita the species with whichit competes most for food Diet overlap with C semi-testaceus is almost complete Overlap with F subpolitamay be higher than measured in our study becauseprevious studies suggest that many Formica speciesincluding F subpalita collect insect and plant exudate(Cook 1953 Dumpert 1981 Wheeler and Wheeler1986 Cherix 1987) which makes up a significant pro-portion of the diet of Argentine ants Diet overlap withP calif arnica and M andrei was minimal We measuredthe diets of native ant species and Argentine ant spe-cies in different areas where resource availability mayhave varied affecting food choice Further work isneeded at an invasion front to investigate effects offood availability on diet overlap

The strong association of the Argentine ant with theedges of Jasper Ridge suggests that the most significantlimits to its range expansion may be opportunity forintroduction and its own mechanism of dispersal Dis-tance to water was less important than we expectedbut a study that spans a wider range of conditions thanours might discover a stronger effect of moistureLimiting human activity within nature preserves andcreating reserves with large areaperimeter ratios mayprevent invasion of interior areas by Argentine antsand other invasive species that tend to track humandisturbance Elevation or some parameter related toit is also important to Argentine ant distribution andfurther investigation of this might contribute to othermanagement practices The possibility of spatial au-tocorrelation should be kept in mind sites near to each

other are likely to have similar environmental condi-tions and similar histories of Argentine ant establish-ment

Abiotic factors explain much of the difference indistribution of Argentine and native ant species whichdo not overlap on a scale of hectares but little of thedifference in activity patterns At the edge of theinvasion on a scale of meters Argentine and native antspecies forage in the same areas with substantial over-lap in temporal patterns of foraging activity Theseresults suggest that whenever native ant colonies onthe edge of the invasion are active they are likely tointeract with Argentine ants Interactions between theinvasive and native ant species may promote displace-ment of the native species Frequent encounters maylead native species to retreat from food sources andpossibly to emigrate Where Argentine ants and nativeants overlap substantially in diet exploitation compe-tition could reduce the foraging success of native antcolonies

Acknowledgments

We thank Eric Bjorkstedt (Stanford University) MarkBrown (Stanford University) Diane Wagner (Stanford Uni-versity) and 2 anonymous reviewers for their comments onthe manuscript Kristen Duin (Stanford University) and LisaStadlen (Stanford University) were terrific field assistantsRobert Swierk (Center for Conservation Biology StanfordUniversity) provided technical assistance with GeographicalInformation Systems The work was supported by disserta-tion fellowships to KGH from National Science Foundationand the American Association of University Women grantsto KGH from the Jasper Ridge Biological Preserve TheConservation and Research Foundation and the Phi BetaKappa Northern California Chapter and USDA grant No95-37302-1885 to DMG

References Cited

Adams E S and IFA Trainello 1980 Chemical interfer-ence competition by Monomorium minimum Oecologia(Ber) 51 265-270

Andersen A N and A D Patel 1994 Meat ants as domi-nant members of Australian ant communities an exper-imental test of their influence on the foraging success andforager abundance of other species Oecologia (Ber) 9815-24

Anderson A N 1992 Regulation of momentary diversityby dominant species in exceptionally rich ant communi-ties of the Australian seasonal tropics Am Nat 140 401-420

Andrews E A 1927 Ant-mounds as to temperature andsunshine J of Morpho Physio 44 1-20

ArcView GIS 1996 ArcInfo 701 ArcView 21 Environ-mental Systems Research Institute Redlands CA

Baker H G and G L Stebbins 1965 The genetics ofcolonizing species Proceedings International Union ofBiological Sciences Symposia on General Biology Aca-demic New York

Banks W A and D F Williams 1989 Competitive dis-placement of Paratrechina longicomis (Latreille) (Hyme-noptera Formicidae) from baits by fire ants in MattoGrosso Brazi J Entomo Sci 24 381-391

August 1998 HUMAN ET AL DISTRIBUTION AND ACTIVlTI OF ARGENTINE ANT 829

C semitestaceus F subpolita L humile M andrei P califarnica

I

0 I bull I r-~~720 1440o

bull pobullbull 1II bull

I bullbullbull - ~ bullbull~bull bull It

bull I304050607080

I~r ~ I

L70 10 30 50 70

bullbull

bull bull It ~-

bull

-r--

304050607080

-- bull

ItS ~

r bull -T

bull

~----

bull

720 1440

~ t bullbullbull I

I bullbull~ ~-- - ~

o

70 10 30 50 70 10 30 50

Soil temperature (0C)

bull10 bullbullbull

bull

bullbull

bull bull bullbull - -- bull bullbullbullbullbullbull bull bullbullbull bullbullbull- ~II bullbullbull - middot bull

bullbullbullbull middot bullbull bull bullbull bull i - bullbull bull 1bullbullbull bull 1bull ~ bull bull bullbull

30 40 50 60 70 80 30 40 50 60 70 80

Relative humidity ()

1440midnight

bull i~ ~ Imiddot

bullbull

20 30 40 50 10 20 30 40 50 10 20 30 40 50 10 20 30 40 50 10 20 30 40 50

Air Temperature (0C)

bull

720noon

4middot

bull

middot

bull-1bull

30 40 50 60 70 8e

bull

d

c200

50

a200

1150-

100 bull -jI~

50 -Ibull

10 30 50 70

b200

150 bull- bull100 middot

100

150

100

150

200

omidnight

Time (min from previous midnight)

Fig 5 The influence of the abiotic factors (a) soil temperature (b) air temperature (c) relative humidity and (d) timeon ant colony activity for 5 ant species Ant colony activity was measured as the numbers of ants entering and exiting a colonywithin 2 min Each dot on the figures represents 1 observation of 1 colony

exotic species including the Argentine ant more eas-ily invade areas that are disturbed by human activity(Ward 1987Elton 1958Simberloff1981) This may berelated both to increased opportunity for introduction

in areas with a great deal of human activity and lackof competitors or predators if native species are sen-sitive to disturbance and retreat It is likely that theArgentine ant moved into the reserve from the de-

830 ENVIRONMENTAL ENTOMOLOGY Vol 27 no 4

Time (min from midnight)

c80

E 60QJbull2bull 40QJ0Ell0

20Ul

00 10 20 30 40

Air temperature (0C)

1440

b80

~ 60QJ

2~ 40QJ0-Ell0

20Ul

00 720

a

C 80

c~ 60Elc 40QJ

iO 20Oip 00

0 0 0

0 10 20 30 40

Air temperature (0C)

Fig6 Examplesof the relationshipsamongabioticfactors (a-c) Eachdot represents 1 measurementof colonyactivityand the indicated abioticfactor

veloped surrounding areas that have already beeninvaded

The dispersal strategy of the Argentine ant mayfurther explain why it tends to occur near the edges ofthe preserve Mating flights have been observed onlyrarely in this species (Newell and Barber 1913 Skaife1961) Occasional spot introductions of Argentine antshave been described in Californias Central Valley(Holway 1995) This may have occurred through theaccidental introduction of queens by humans throughqueens or fragments of colonies rafting downstream infloods (Holway 1995) or through the founding of newcolonies by winged queens Over the last 3 yr we haveobserved no spot introductions In the Jasper RidgePreserve most areas along creek corridors are alreadyinvaded by Argentine ants and there is little oppor-tunity for the species to be introduced accidentallyinside the preserve because human activities within itare limited Our results suggest that the Argentine antoccurs mostly at the preserve edges because it is ex-panding its range primarily by spreading along theground through colony fission

Table 1 Composition of forage and diet overlap between Ar-gentine ants and 4 native ant species

Ant species

LH CS FS MA PC

Proportion of food itemsLiquid food 0916 0825 0000 0000 0000Seeds 0004 0004 0036 0765 0661Insects 0055 0102 0631 0093 0120Other 0025 0069 0333 0141 0219

Total number of food items1313 484 279 686 183

Proportion overlap with Argentine antLow 0997 0060 0011 0015High 0998 0066 0016 0023

In (C) values closest to 1 indicate the greatest degree of overlapThe low overlap value assumed that forage in the unidentified cate-gory (other) differed between species and the high value assumedthat forage in this category was identical CS C semitestacetls FS Fsltbpolita LH L hmile MA M andrei PC P califomica

It seems unlikely that previous disturbance of theareas surrounding the preserve had already caused thenative species to disappear from inside the preserveprior to invasion by Argentine ants Within the pre-serve native ant species and Argentine ants competeon the margin of the invasion and native ant speciesdisappear from invaded areas but persist elsewhere(Human and Gordon 1996 1997) All of the 9 hainvaded by the Argentine ant between September1993 and September 1995 were occupied in 1993 bynative ant species

Argentine ant incidence was high in low-elevationareas However elevation per se is unlikely to be thelimiting factor because the Argentine ant can befound in a few high-elevation areas of Jasper Ridgeand the species has been found up to 2000 m in theHawaiian islands (Zimmerman 1940 Cole et al 1992)Elevation may be related to another environmentalvariable such as soil moisture that affects the Argen-tine ant directly There may be a historical effect if theArgentine ant were introduced more frequently tolow-elevation areas

Studies in the Central Valley of California havesuggested that Argentine ant encroachment into nat-ural habitat is almost exclUSivelylimited to areas withpermanent sources of water and that the Argentine antis more limited by proximity to water than are nativeant species (Ward 1987) We found that several nativetaxa including Prenoepis imparis Stenamma spp andLeptothorax spp were as close to water on average aswere Argentine ants Many other taxa including Cam-ponotus spp P californica and Tapinoma sessile Smithwere found only marginally farther from water thanthe Argentine ant (Fig 2) Although Argentine antstend to be more strongly established in areas closer towater (Fig 1) there are many areas near water not yetinvaded and many areas far from water where Ar-gentine ants are firmly established The differencebetween the results of this study and previous onesmay be due to climatic differences the Central Valleyis much dryer and hotter especially in the summerthan Jasper Ridge which lies in the coastal foothills

August 1998 HUMAN ET AL DISTRIBUTION AND ACTIVITY OF ARGENTINE ANT 831

Colony Activity and Abiotic Conditions It may bethat the Argentine ant avoids temperature extremesby some behavioral means A laboratory study thatcompared Argentine ants and other ant species nativeto central coastal California suggested that the Argen-tine ant has a low thermal tolerance dying at tem-peratures close to 46degC (Tremper 1976) In the fieldArgentine ant nests frequently experience soil tem-peratures gt46degC and they are active up to 70degC It ispossible that the Argentine ant copes behaviorallywith heat more effectively than native species In thelaboratory M andrei workers died when exposed totemperatures gt485degC (Tremper 1976) We foundthat M andrei colonies were rarely active when soiltemperatures exceeded 50degC M andrei are large blackants that are often found on trails and open areaswhere they may be unable to avoid temperature ex-tremes

Compared with most native ant species Argentineants are active in a wide range of abiotic conditions(Fig 5) M andrei and P californica were active in arange of conditions similar to that of Argentine antsBy contrast the activities of C semitestaceus and Fsubpolita were confined to a narrower range of timesand temperatures Because the Argentine ant is usu-ally active whenever any native species is most nativespecies in invaded areas are likely to encounter theinvaders frequently

The diet of the Argentine ant overlaps with that ofseveral species (Table 1) Argentine ant nests areactive in a wider range of abiotic conditions than areC semitestaceus or F subpolita the species with whichit competes most for food Diet overlap with C semi-testaceus is almost complete Overlap with F subpolitamay be higher than measured in our study becauseprevious studies suggest that many Formica speciesincluding F subpalita collect insect and plant exudate(Cook 1953 Dumpert 1981 Wheeler and Wheeler1986 Cherix 1987) which makes up a significant pro-portion of the diet of Argentine ants Diet overlap withP calif arnica and M andrei was minimal We measuredthe diets of native ant species and Argentine ant spe-cies in different areas where resource availability mayhave varied affecting food choice Further work isneeded at an invasion front to investigate effects offood availability on diet overlap

The strong association of the Argentine ant with theedges of Jasper Ridge suggests that the most significantlimits to its range expansion may be opportunity forintroduction and its own mechanism of dispersal Dis-tance to water was less important than we expectedbut a study that spans a wider range of conditions thanours might discover a stronger effect of moistureLimiting human activity within nature preserves andcreating reserves with large areaperimeter ratios mayprevent invasion of interior areas by Argentine antsand other invasive species that tend to track humandisturbance Elevation or some parameter related toit is also important to Argentine ant distribution andfurther investigation of this might contribute to othermanagement practices The possibility of spatial au-tocorrelation should be kept in mind sites near to each

other are likely to have similar environmental condi-tions and similar histories of Argentine ant establish-ment

Abiotic factors explain much of the difference indistribution of Argentine and native ant species whichdo not overlap on a scale of hectares but little of thedifference in activity patterns At the edge of theinvasion on a scale of meters Argentine and native antspecies forage in the same areas with substantial over-lap in temporal patterns of foraging activity Theseresults suggest that whenever native ant colonies onthe edge of the invasion are active they are likely tointeract with Argentine ants Interactions between theinvasive and native ant species may promote displace-ment of the native species Frequent encounters maylead native species to retreat from food sources andpossibly to emigrate Where Argentine ants and nativeants overlap substantially in diet exploitation compe-tition could reduce the foraging success of native antcolonies

Acknowledgments

We thank Eric Bjorkstedt (Stanford University) MarkBrown (Stanford University) Diane Wagner (Stanford Uni-versity) and 2 anonymous reviewers for their comments onthe manuscript Kristen Duin (Stanford University) and LisaStadlen (Stanford University) were terrific field assistantsRobert Swierk (Center for Conservation Biology StanfordUniversity) provided technical assistance with GeographicalInformation Systems The work was supported by disserta-tion fellowships to KGH from National Science Foundationand the American Association of University Women grantsto KGH from the Jasper Ridge Biological Preserve TheConservation and Research Foundation and the Phi BetaKappa Northern California Chapter and USDA grant No95-37302-1885 to DMG

References Cited

Adams E S and IFA Trainello 1980 Chemical interfer-ence competition by Monomorium minimum Oecologia(Ber) 51 265-270

Andersen A N and A D Patel 1994 Meat ants as domi-nant members of Australian ant communities an exper-imental test of their influence on the foraging success andforager abundance of other species Oecologia (Ber) 9815-24

Anderson A N 1992 Regulation of momentary diversityby dominant species in exceptionally rich ant communi-ties of the Australian seasonal tropics Am Nat 140 401-420

Andrews E A 1927 Ant-mounds as to temperature andsunshine J of Morpho Physio 44 1-20

ArcView GIS 1996 ArcInfo 701 ArcView 21 Environ-mental Systems Research Institute Redlands CA

Baker H G and G L Stebbins 1965 The genetics ofcolonizing species Proceedings International Union ofBiological Sciences Symposia on General Biology Aca-demic New York

Banks W A and D F Williams 1989 Competitive dis-placement of Paratrechina longicomis (Latreille) (Hyme-noptera Formicidae) from baits by fire ants in MattoGrosso Brazi J Entomo Sci 24 381-391

830 ENVIRONMENTAL ENTOMOLOGY Vol 27 no 4

Time (min from midnight)

c80

E 60QJbull2bull 40QJ0Ell0

20Ul

00 10 20 30 40

Air temperature (0C)

1440

b80

~ 60QJ

2~ 40QJ0-Ell0

20Ul

00 720

a

C 80

c~ 60Elc 40QJ

iO 20Oip 00

0 0 0

0 10 20 30 40

Air temperature (0C)

Fig6 Examplesof the relationshipsamongabioticfactors (a-c) Eachdot represents 1 measurementof colonyactivityand the indicated abioticfactor

veloped surrounding areas that have already beeninvaded

The dispersal strategy of the Argentine ant mayfurther explain why it tends to occur near the edges ofthe preserve Mating flights have been observed onlyrarely in this species (Newell and Barber 1913 Skaife1961) Occasional spot introductions of Argentine antshave been described in Californias Central Valley(Holway 1995) This may have occurred through theaccidental introduction of queens by humans throughqueens or fragments of colonies rafting downstream infloods (Holway 1995) or through the founding of newcolonies by winged queens Over the last 3 yr we haveobserved no spot introductions In the Jasper RidgePreserve most areas along creek corridors are alreadyinvaded by Argentine ants and there is little oppor-tunity for the species to be introduced accidentallyinside the preserve because human activities within itare limited Our results suggest that the Argentine antoccurs mostly at the preserve edges because it is ex-panding its range primarily by spreading along theground through colony fission

Table 1 Composition of forage and diet overlap between Ar-gentine ants and 4 native ant species

Ant species

LH CS FS MA PC

Proportion of food itemsLiquid food 0916 0825 0000 0000 0000Seeds 0004 0004 0036 0765 0661Insects 0055 0102 0631 0093 0120Other 0025 0069 0333 0141 0219

Total number of food items1313 484 279 686 183

Proportion overlap with Argentine antLow 0997 0060 0011 0015High 0998 0066 0016 0023

In (C) values closest to 1 indicate the greatest degree of overlapThe low overlap value assumed that forage in the unidentified cate-gory (other) differed between species and the high value assumedthat forage in this category was identical CS C semitestacetls FS Fsltbpolita LH L hmile MA M andrei PC P califomica

It seems unlikely that previous disturbance of theareas surrounding the preserve had already caused thenative species to disappear from inside the preserveprior to invasion by Argentine ants Within the pre-serve native ant species and Argentine ants competeon the margin of the invasion and native ant speciesdisappear from invaded areas but persist elsewhere(Human and Gordon 1996 1997) All of the 9 hainvaded by the Argentine ant between September1993 and September 1995 were occupied in 1993 bynative ant species

Argentine ant incidence was high in low-elevationareas However elevation per se is unlikely to be thelimiting factor because the Argentine ant can befound in a few high-elevation areas of Jasper Ridgeand the species has been found up to 2000 m in theHawaiian islands (Zimmerman 1940 Cole et al 1992)Elevation may be related to another environmentalvariable such as soil moisture that affects the Argen-tine ant directly There may be a historical effect if theArgentine ant were introduced more frequently tolow-elevation areas

Studies in the Central Valley of California havesuggested that Argentine ant encroachment into nat-ural habitat is almost exclUSivelylimited to areas withpermanent sources of water and that the Argentine antis more limited by proximity to water than are nativeant species (Ward 1987) We found that several nativetaxa including Prenoepis imparis Stenamma spp andLeptothorax spp were as close to water on average aswere Argentine ants Many other taxa including Cam-ponotus spp P californica and Tapinoma sessile Smithwere found only marginally farther from water thanthe Argentine ant (Fig 2) Although Argentine antstend to be more strongly established in areas closer towater (Fig 1) there are many areas near water not yetinvaded and many areas far from water where Ar-gentine ants are firmly established The differencebetween the results of this study and previous onesmay be due to climatic differences the Central Valleyis much dryer and hotter especially in the summerthan Jasper Ridge which lies in the coastal foothills

August 1998 HUMAN ET AL DISTRIBUTION AND ACTIVITY OF ARGENTINE ANT 831

Colony Activity and Abiotic Conditions It may bethat the Argentine ant avoids temperature extremesby some behavioral means A laboratory study thatcompared Argentine ants and other ant species nativeto central coastal California suggested that the Argen-tine ant has a low thermal tolerance dying at tem-peratures close to 46degC (Tremper 1976) In the fieldArgentine ant nests frequently experience soil tem-peratures gt46degC and they are active up to 70degC It ispossible that the Argentine ant copes behaviorallywith heat more effectively than native species In thelaboratory M andrei workers died when exposed totemperatures gt485degC (Tremper 1976) We foundthat M andrei colonies were rarely active when soiltemperatures exceeded 50degC M andrei are large blackants that are often found on trails and open areaswhere they may be unable to avoid temperature ex-tremes

Compared with most native ant species Argentineants are active in a wide range of abiotic conditions(Fig 5) M andrei and P californica were active in arange of conditions similar to that of Argentine antsBy contrast the activities of C semitestaceus and Fsubpolita were confined to a narrower range of timesand temperatures Because the Argentine ant is usu-ally active whenever any native species is most nativespecies in invaded areas are likely to encounter theinvaders frequently

The diet of the Argentine ant overlaps with that ofseveral species (Table 1) Argentine ant nests areactive in a wider range of abiotic conditions than areC semitestaceus or F subpolita the species with whichit competes most for food Diet overlap with C semi-testaceus is almost complete Overlap with F subpolitamay be higher than measured in our study becauseprevious studies suggest that many Formica speciesincluding F subpalita collect insect and plant exudate(Cook 1953 Dumpert 1981 Wheeler and Wheeler1986 Cherix 1987) which makes up a significant pro-portion of the diet of Argentine ants Diet overlap withP calif arnica and M andrei was minimal We measuredthe diets of native ant species and Argentine ant spe-cies in different areas where resource availability mayhave varied affecting food choice Further work isneeded at an invasion front to investigate effects offood availability on diet overlap

The strong association of the Argentine ant with theedges of Jasper Ridge suggests that the most significantlimits to its range expansion may be opportunity forintroduction and its own mechanism of dispersal Dis-tance to water was less important than we expectedbut a study that spans a wider range of conditions thanours might discover a stronger effect of moistureLimiting human activity within nature preserves andcreating reserves with large areaperimeter ratios mayprevent invasion of interior areas by Argentine antsand other invasive species that tend to track humandisturbance Elevation or some parameter related toit is also important to Argentine ant distribution andfurther investigation of this might contribute to othermanagement practices The possibility of spatial au-tocorrelation should be kept in mind sites near to each

other are likely to have similar environmental condi-tions and similar histories of Argentine ant establish-ment

Abiotic factors explain much of the difference indistribution of Argentine and native ant species whichdo not overlap on a scale of hectares but little of thedifference in activity patterns At the edge of theinvasion on a scale of meters Argentine and native antspecies forage in the same areas with substantial over-lap in temporal patterns of foraging activity Theseresults suggest that whenever native ant colonies onthe edge of the invasion are active they are likely tointeract with Argentine ants Interactions between theinvasive and native ant species may promote displace-ment of the native species Frequent encounters maylead native species to retreat from food sources andpossibly to emigrate Where Argentine ants and nativeants overlap substantially in diet exploitation compe-tition could reduce the foraging success of native antcolonies

Acknowledgments

We thank Eric Bjorkstedt (Stanford University) MarkBrown (Stanford University) Diane Wagner (Stanford Uni-versity) and 2 anonymous reviewers for their comments onthe manuscript Kristen Duin (Stanford University) and LisaStadlen (Stanford University) were terrific field assistantsRobert Swierk (Center for Conservation Biology StanfordUniversity) provided technical assistance with GeographicalInformation Systems The work was supported by disserta-tion fellowships to KGH from National Science Foundationand the American Association of University Women grantsto KGH from the Jasper Ridge Biological Preserve TheConservation and Research Foundation and the Phi BetaKappa Northern California Chapter and USDA grant No95-37302-1885 to DMG

References Cited

Adams E S and IFA Trainello 1980 Chemical interfer-ence competition by Monomorium minimum Oecologia(Ber) 51 265-270

Andersen A N and A D Patel 1994 Meat ants as domi-nant members of Australian ant communities an exper-imental test of their influence on the foraging success andforager abundance of other species Oecologia (Ber) 9815-24

Anderson A N 1992 Regulation of momentary diversityby dominant species in exceptionally rich ant communi-ties of the Australian seasonal tropics Am Nat 140 401-420

Andrews E A 1927 Ant-mounds as to temperature andsunshine J of Morpho Physio 44 1-20

ArcView GIS 1996 ArcInfo 701 ArcView 21 Environ-mental Systems Research Institute Redlands CA

Baker H G and G L Stebbins 1965 The genetics ofcolonizing species Proceedings International Union ofBiological Sciences Symposia on General Biology Aca-demic New York

Banks W A and D F Williams 1989 Competitive dis-placement of Paratrechina longicomis (Latreille) (Hyme-noptera Formicidae) from baits by fire ants in MattoGrosso Brazi J Entomo Sci 24 381-391

August 1998 HUMAN ET AL DISTRIBUTION AND ACTIVITY OF ARGENTINE ANT 831

Colony Activity and Abiotic Conditions It may bethat the Argentine ant avoids temperature extremesby some behavioral means A laboratory study thatcompared Argentine ants and other ant species nativeto central coastal California suggested that the Argen-tine ant has a low thermal tolerance dying at tem-peratures close to 46degC (Tremper 1976) In the fieldArgentine ant nests frequently experience soil tem-peratures gt46degC and they are active up to 70degC It ispossible that the Argentine ant copes behaviorallywith heat more effectively than native species In thelaboratory M andrei workers died when exposed totemperatures gt485degC (Tremper 1976) We foundthat M andrei colonies were rarely active when soiltemperatures exceeded 50degC M andrei are large blackants that are often found on trails and open areaswhere they may be unable to avoid temperature ex-tremes

Compared with most native ant species Argentineants are active in a wide range of abiotic conditions(Fig 5) M andrei and P californica were active in arange of conditions similar to that of Argentine antsBy contrast the activities of C semitestaceus and Fsubpolita were confined to a narrower range of timesand temperatures Because the Argentine ant is usu-ally active whenever any native species is most nativespecies in invaded areas are likely to encounter theinvaders frequently

The diet of the Argentine ant overlaps with that ofseveral species (Table 1) Argentine ant nests areactive in a wider range of abiotic conditions than areC semitestaceus or F subpolita the species with whichit competes most for food Diet overlap with C semi-testaceus is almost complete Overlap with F subpolitamay be higher than measured in our study becauseprevious studies suggest that many Formica speciesincluding F subpalita collect insect and plant exudate(Cook 1953 Dumpert 1981 Wheeler and Wheeler1986 Cherix 1987) which makes up a significant pro-portion of the diet of Argentine ants Diet overlap withP calif arnica and M andrei was minimal We measuredthe diets of native ant species and Argentine ant spe-cies in different areas where resource availability mayhave varied affecting food choice Further work isneeded at an invasion front to investigate effects offood availability on diet overlap

The strong association of the Argentine ant with theedges of Jasper Ridge suggests that the most significantlimits to its range expansion may be opportunity forintroduction and its own mechanism of dispersal Dis-tance to water was less important than we expectedbut a study that spans a wider range of conditions thanours might discover a stronger effect of moistureLimiting human activity within nature preserves andcreating reserves with large areaperimeter ratios mayprevent invasion of interior areas by Argentine antsand other invasive species that tend to track humandisturbance Elevation or some parameter related toit is also important to Argentine ant distribution andfurther investigation of this might contribute to othermanagement practices The possibility of spatial au-tocorrelation should be kept in mind sites near to each

other are likely to have similar environmental condi-tions and similar histories of Argentine ant establish-ment

Abiotic factors explain much of the difference indistribution of Argentine and native ant species whichdo not overlap on a scale of hectares but little of thedifference in activity patterns At the edge of theinvasion on a scale of meters Argentine and native antspecies forage in the same areas with substantial over-lap in temporal patterns of foraging activity Theseresults suggest that whenever native ant colonies onthe edge of the invasion are active they are likely tointeract with Argentine ants Interactions between theinvasive and native ant species may promote displace-ment of the native species Frequent encounters maylead native species to retreat from food sources andpossibly to emigrate Where Argentine ants and nativeants overlap substantially in diet exploitation compe-tition could reduce the foraging success of native antcolonies

Acknowledgments

We thank Eric Bjorkstedt (Stanford University) MarkBrown (Stanford University) Diane Wagner (Stanford Uni-versity) and 2 anonymous reviewers for their comments onthe manuscript Kristen Duin (Stanford University) and LisaStadlen (Stanford University) were terrific field assistantsRobert Swierk (Center for Conservation Biology StanfordUniversity) provided technical assistance with GeographicalInformation Systems The work was supported by disserta-tion fellowships to KGH from National Science Foundationand the American Association of University Women grantsto KGH from the Jasper Ridge Biological Preserve TheConservation and Research Foundation and the Phi BetaKappa Northern California Chapter and USDA grant No95-37302-1885 to DMG

References Cited

Adams E S and IFA Trainello 1980 Chemical interfer-ence competition by Monomorium minimum Oecologia(Ber) 51 265-270

Andersen A N and A D Patel 1994 Meat ants as domi-nant members of Australian ant communities an exper-imental test of their influence on the foraging success andforager abundance of other species Oecologia (Ber) 9815-24

Anderson A N 1992 Regulation of momentary diversityby dominant species in exceptionally rich ant communi-ties of the Australian seasonal tropics Am Nat 140 401-420

Andrews E A 1927 Ant-mounds as to temperature andsunshine J of Morpho Physio 44 1-20

ArcView GIS 1996 ArcInfo 701 ArcView 21 Environ-mental Systems Research Institute Redlands CA

Baker H G and G L Stebbins 1965 The genetics ofcolonizing species Proceedings International Union ofBiological Sciences Symposia on General Biology Aca-demic New York

Banks W A and D F Williams 1989 Competitive dis-placement of Paratrechina longicomis (Latreille) (Hyme-noptera Formicidae) from baits by fire ants in MattoGrosso Brazi J Entomo Sci 24 381-391

832 ENVIRONMENTAL ENTOMOLOGY Vo 27 no 4

Bernstein R A 1979 Schedules of foraging activity in spe-cies of ants J Anim Eco 48 921-930

Brian M V 1964 Ant distribution in a southern Englishheath J Anim Eco 33 451-461

1973 Temperature choise and its relevance to brood sur-vival and caste determination in the ant Myrmica TlIbra LPhysio Zoo 46 254-252

Briese D T andJ B Macauley 1980 Temporal structureof an ant community in semi-arid Australia Aust J ofEco 5 121-134

Burrough P A 1986 Principles of geographic informationsystems for land resources assessment Oxford UniversityPress New York

Cheri x D 1987 Relation between diet and polyethism inFomlica colonies pp 93-117 In J M Pasteels and J LDeneubourg [eds] From individual to collective behav-ior in social insects Les Treilles Workshop BirkhauserBath

Cole F R A Cl1edeiros L L Loope and W W Zuehlke1992 Effects of the Argentine ant on arthropod fauna ofHawaiian high-elevation shrubhmd Ecology 73 1313-1322

Cook T W 1953 The ants of California Pacific Books PaloAlto CA

Crowell K 1968 Rates of competitive exclusion by the Ar-gentine ant in Bermuda Ecology 49 551-555

Cushman J H 1993 Latitudinal patterns in European antassemblages variation in species richness and body sizeOecologia (Ber) 95 30-37

DAntonio C M and P M Vitousek 1992 Biological in-vasions by exotic grasses the grass fire cycle and globalchange Annu Rev Eco Syst 23 63-87

Davidson D W 1977 Foraging ecology and communityorganization in desert seed-eating ants Ecology 58 725-737

De Kock A E 1990 Interactions between the introducedArgentine ant Indomyrmex hllmiis MayI and two in-digenous fynbos ant species J Entomo Soc S Afr 53107-111

Dean WRJ 1992 Temperatures determining activity pat-terns of some ant species in the southern Karoo SouthAfrica J Entomo Soc of S Afr 55 149-156

Dumpert K 1981 The social bioloty of ants Pitman BathUK

Ehrlich P R 1986 Which animal will invade In H AMooney and J A Drake [eds] Ecology of biologicalinvasions of North America and Hawaii Springer NewYork

Elton C S 1958 The ecology of invasions by animals andplants Methuen London

Erickson J M 1971 The displacement of native ant speciesby the introduced Argentine ant Indomyrmex humiisMayI Psyche 78 257-266

Fausch K D and R J White 1981 Competition betweenbrook trout (Salvelinlls fontinalis) and brown trout(Salmo trutta) for positions in a Michigan stream Can JFish Aquat Sci 38 1220-1227

Fielde A M 1904 Observations on ants in their relation totemperature and to submergence Bio Bull 7 170-174

Foster E 1908 The introduction of Indomyrmex humiis(MayI) into New Orleans J Econ Entomo 1 289-293

Gordon D M 1988 Nest-plugging interference competi-tion in desert ants (Novomessor cockerelli and Pogono-myrmex barbatlls) Oecologia (Ber) 75 114-118

Haskins C and E Haskins 1965 Pheidole megacephala andIndomyrmex 11lImilis in Bermuda Equilibrium or slowreplacement Ecology 46 736-740

1988 Final observations on Pheidole megaceplUlla and In-domymlex humilis in Bermuda Psyche 95 177-184

Hetrick W AP M Rich FJ Barnes and SB Weiss 1993GIS-based solar radiation poundlux model American Societyfor Photogrammetry and Remote Sensing Technical Pa-pers GIS PhotogI Mode 3 132-143

Holdgate M W 1986 Summary and conclusions charac-teristics and consequences of biological invasions PhilosTrans R Soc Lond B 314 733-742

Holldobler B and E O Wilson 1990 The ants HarvardUniversity Press Cambridge MA

Holway D A 1995 Distribution of the Argentine ant (Line-pithema hwnile) in northern California Conserv Bio 91634-1637

Human K G and D M Gordon 1996 Exploitation andinterference competition between the invasive Argen-tine ant Linepithema hwnile and native ant species Oe-cologia (Ber) 105 405-412

1997 Effects of Argentine ants on invertebrate diversity innorthern California Conserv Bio 11 1242-1248

Hutchinson M F 1989 A new procedure for gridding el-evation and stream line data with automatic removal ofspurious pits J Hydro 106 211-232

1993 Development of a continent-wide DEM with appli-cations to terrain and climate analysis pp 393-399 InM F Goodchild [ed] Environmental modeling withGIS Oxford University Press New York

Lynch J F E C Balinsky and S G Vail 1980 Foragingpatterns in three sympatric forest ant species Prenolepisimparis Paratrechina melanderi and Aphaenogaster mdis(Hymenoptera Formicidae) Eco Entomo 5 353-371

Macdonald IAW D L Clark and H C Taylor 1989 Thehistory and effects of alien plant control in the Cape ofGood Hope Nature Reserve S Afr J Bot 55 56-75

Marsh A C 1985 Microclimatic factors influencing forag-ing patterns and success of the thermophilic desert antOcymyrmex barbiger Insectes Soc 32 286-296

1988 Activity patterns of some Namib Desert ants J AridEnviron 14 61-73

Medeiros A C L L Loope and F R Cole 1986 Distri-bution of ants and their effects on endemic biota ofHaelakala and Hawaii Volcanoes National Parks a pre-liminary assessment pp 39-51 In C P Stone and C WSmith [eds] Proceedings of the 6th conference in nat-ural sciences Hawaii Volcanoes National Park Universityof Hawaii Cooperative National Park Resources StudiesUnit Honolulu

Newell W and T C Barber 1913 The Argentine ant USDep Agric Bull 122 1-98

Palmer M W 1993 Putting things in even better order theadvantages of canonical correspondence analysis Ecol-ogy 74 2215-2230

Petersen K and T J Case 1996 An experimental demon-stration of exploitation competition in an ongoing inva-sion Ecology 77 118-132

Pianka E R 1973 The structure of lizard communitiesAnnu Rev Eco Syst 4 53-74

Porter S D and D A Savignano 1990 Invasion of poly-gyne fire ants decimates native ants and disrupts arthro-pod community Ecology 71 2095-2106

Savolainen R and K Vepsaainen 1988 A competition hi-erarchy among boreal ants Impact on resource partition-ing and community structure Oikos 51 135-155

Sawyer J 0 and T Keeler-Wolf 1995 A manual of Cali-fornia vegetation California Native Plant Society

Simberloff D 1981 Community effects of introduced spe-cies pp 53-81 In M H Nitecki [ed] Biotic crises inecological and evolutionary time Academic New York

Au~st 1998 HUMAN ET AL DISTRIBUTION AND ACTIVlTY OF ARGENTINE ANT 833

Skaife SH 1961 The study of ants Longmans Green Lon-don

ter Braak CJF 1987 CANOCO-a FORTRAN programfor canonical community ordination MicrocomputerPower Ithaca NY

ter Braak C]F and I C Prentice 1988 A theory of gra-dient analysis Adv Ecol Res 18 271-313

Talbot M 1943 Response of the ant Prenolepis imparis Sayto temperature and humidity changes Ecology 24 345-352

Tremper B S 1976 Distribution of the Argentine ant lri-domyrmex humilis Mayr in relation to certain native antsof Califomia ecological physiological and behavioralaspects PhD dissertation University of CaliforniaBerkeley

epsiiliHnen K and R Savolainen 1990 The effect of in-terference by formicine ants on the foraging of MynnicaJ Anim Ecol 59 643-654

Ward P S 1987 Distribution of the introduced Argentineant (Iridomyrmex humitis) in natural habitats of theLower Sacramento Valley and its effects on the indige-nous ant fauna Hilgardia 55 1-16

Wehner R 1987 Spatial organization of foraging behaviorin individually searching desert ants Cataglyphis (Saharadesert) and Ocymynnex (Namib desert) pp 15-42 In

J M Pasteels and J L Deneuboulg reds J From indi-vidual to collective behavior in the social insects Expe-rientia supplement 54 Birkhauser Basel

Wheeler G C and J N Wheeler 1986 The ants of Ne-vada Natural History Museum of Los Angeles CountyLos Angeles CA

Whitford W G D J Depree p Hamilton and G Etter-shank 1981 Foraging ecology of seed-harvesting antsPheidole spp in a Chihuahuan desert ecosystem AmMid Nat 105 159-167

Whitford W G and G Ettershank 1975 Factors affectingforaging activity in Chihuahuan Desert harvester antsEnviron Entomol 4 689-696

Whittaker R J 1989 The vegetation of the StorbreenGletschervorfeld Jotunheimen Norway III Vegetation-environment relationships J Biogreogr 16 413- 433

Wilson E O 1955 A monographic revision of the ant genusLasius Bull Mus Compo Zool Harv 113 1-99

Zimmerman E C 1941 Adaptive radiation in Hawaii withspecial reference to insects Biotropica 2 32-38

Receivedforpublication 11 April 1997 accepted 18 February1998