|
|
Proc Natl Acad Sci U S A. 2001 January 30; 98(3): 1095–1100. | PMCID: PMC14714 |
Copyright © 2001, The National Academy of Sciences Ecology Patterns of spread in biological invasions dominated by
long-distance jump dispersal: Insights from Argentine ants Andrew V. Suarez,* David A. Holway, and Ted J. Case Department of Biology 0116, University of California at San Diego, 9500 Gilman Drive, La Jolla, CA 92093-0116 Received June 14, 2000; Accepted December 4, 2000. Invading organisms may spread through local movements (giving
rise to a diffusion-like process) and by long-distance jumps, which are
often human-mediated. The local spread of invading organisms has been
fit with varying success to models that couple local population growth
with diffusive spread, but to date no quantitative estimates exist for
the relative importance of local dispersal relative to human-mediated
long-distance jumps. Using a combination of literature review, museum
records, and personal surveys, we reconstruct the invasion history of
the Argentine ant (Linepithema humile), a widespread
invasive species, at three spatial scales. Although the inherent
dispersal abilities of Argentine ants are limited, in the last century,
human-mediated dispersal has resulted in the establishment of this
species on six continents and on many oceanic islands. Human-mediated
jump dispersal has also been the primary mode of spread at a
continental scale within the United States. The spread of the Argentine
ant involves two discrete modes. Maximum distances spread by colonies
undergoing budding reproduction averaged 150 m/year, whereas annual
jump-dispersal distances averaged three orders of magnitude higher.
Invasions that involve multiple dispersal processes, such as those
documented here, are undoubtedly common. Detailed data on invasion
dynamics are necessary to improve the predictive power of future
modeling efforts. Keywords: stratified diffusion, invasive ants, Linepithema
humile, rate of invasion One of the triumphs of
invasion biology is the well-developed mathematical theory that
describes the spread of invasive species ( 1). Skellam ( 2) advanced
models based on reaction-diffusion equations that have been used
commonly by ecologists to predict asymptotic rate of invasion.
Predicted rates of spread are often in agreement with observed rates,
demonstrating the value of this approach ( 1, 3– 6). These models are
often based on two parameters: the intrinsic rate of growth and the
diffusion coefficient ( 2, 7). An important assumption concerning the
diffusion coefficient is that the distances individuals move over a
given length of time are drawn from a normal distribution ( 8, 9). Although theoretical models have reasonably predicted invasion
rates in many cases, violations of their assumptions may limit their
usefulness. In particular, the distributions of dispersal distances for
many taxa are leptokurtic rather than normally distributed ( 7– 12).
Such non-normal distributions may result from a variety of processes.
First, rare long-distance jump-dispersal events can skew distributions
of dispersal distances. Second, deviations from normality may arise
from stratified diffusion ( 3) where an invading species spreads by two
or more modes (e.g., diffusion and jump-dispersal) simultaneously. Both
long-distance dispersal and stratified diffusion can greatly increase
invasion rates and result in a lack of agreement between models and
empirical data ( 1, 3, 9, 13, 14). The frequency and distances of
jump-dispersal events are thought to be stochastic, difficult to
determine, and therefore have rarely been quantified (refs. 12 and 13,
but see ref. 10). However, estimates of the rate and distance of
long-distance dispersal events are essential for accurate model
construction, a limitation that is widely recognized ( 1, 9, 12– 14). Despite their value in guiding modeling efforts, there are few data
documenting large-scale patterns of invasion for species that spread
primarily via jump-dispersal or that spread by multiple processes. To
address this issue, we quantified invasion dynamics for a highly
invasive species, the Argentine ant (Linepithema humile).
Despite the widespread distribution of this species, no recent attempt
has been made to synthesize the growing amount of information regarding
its distribution and invasion history. By analyzing museum records,
personal collections, and the literature, we reconstructed invasion
dynamics of Argentine ants at three spatial scales. First, we
determined the current worldwide distribution of the Argentine ant.
Second, we constructed a chronological history of invasion for this
species at a continental scale in the United States after its
introduction into New Orleans around 1891. Finally, we examined
patterns of spread at a local scale by providing new data and reviewing
the literature. Examining invasion dynamics of the Argentine ant at
these three spatial scales allowed us to gauge the relative importance
of alternate modes of spread at each scale. Biology of Linepithema humile. Native to South America, the Argentine ant causes a variety of economic
and ecological problems throughout its introduced range. Perhaps most
notably, the Argentine ant competitively displaces native ant species
wherever it is introduced ( 15– 21). The loss of native ants has led to
a number of indirect effects, including reduced recruitment of
myrmecochorous shrubs in South Africa ( 16) and declines in populations
of coastal horned lizards in California ( 22). In addition, Argentine
ants have been implicated in the decline of endemic arthropods in
Hawaii ( 23) and in the disruption of arthropod communities in
California ( 24, 25). Argentine ants are most successful in Mediterranean and some
subtropical climates but appear unable to survive in cold-temperate,
tropical, or extremely arid environments ( 26, 27). However, through
their close association with humans, Argentine ants may persist locally
in areas with unfavorable climates in the vicinity of human
habitations. While Argentine ants are associated with disturbed
habitats throughout their introduced range, in some locations L.
humile penetrates natural areas that have experienced little
anthropogenic disturbance. Examples include matorral in Chile ( 28),
fynbos in South Africa ( 16), coastal sage scrub in southern California
( 20), riparian woodlands in California ( 17, 21, 29), subalpine
shrubland in Hawaii ( 23), and oak and pine woodland in Portugal ( 18). The dispersal of Argentine ants involves at least two
discrete processes: diffusion and jump dispersal. Once
established, Argentine ant colonies reproduce by budding; this pattern
of spread resembles that of diffusion. When new colonies are formed by
budding, inseminated queens leave established nests on foot along with
a group of workers and form new nests nearby. This is in contrast to
the prevailing mode of colony reproduction in ants where queens undergo
mating flights, founding colonies independently of and often well away
from their natal nest ( 26). Argentine ants queens are not known to
undergo mating flights in their introduced range ( 21, 30, 31). A second
form of dispersal involves human-mediated transport of colonies. Such
jump dispersal is probably common for Argentine ants because they often
associate closely with humans. For example, early this century it was
noted that nearly every one of over 100 steamships landing between New
Orleans and Baton Rouge, Louisiana, was heavily invested with Argentine
ants ( 32, 33). Their commensal habits result in part from opportunistic
nesting requirements and a general diet ( 32). Because Argentine ants
lack mating flights, it easy to distinguish between these two distinct
modes of spread. Argentine ants are also known to spread locally by
rafting downstream ( 33). We feel that this mode of spread is of
relatively minor importance compared with human-mediated jump dispersal
given that patterns of expansion are predominately upstream, across
watersheds, and overland (see below). Worldwide Distribution. We used the following methods to determine the current global
distribution of the Argentine ant. First, in a thorough review of the
scientific literature on ants, we examined both regional surveys as
well as publications pertaining specifically to Argentine ants. Second,
we contacted 140 public and personal entomological collections for
information regarding the presence or absence of Argentine ants. Last,
we conducted visual surveys for Argentine ants along the west coast of
North America (from Guerrero Negro, Baja California, Mexico to
Vancouver, Canada) and in northern Argentina (from Buenos Aires north,
primarily along the Rio Parana and the Rio Uruguay). The taxonomy of Linepithema, a genus confined to the
Neotropical region, is poorly studied and unresolved. In particular,
the species boundaries of L. humile in its native range are
largely unknown. While introduced populations are all likely the same
species, some records for Central and South America (where other native
Linepithema occur) may pertain to species other than
L. humile. When possible, we examined material from these
regions to determine whether the specimens resembled the invasive form.
In our surveys of museum records, an attempt was made to determine
whether specimens were morphologically similar to those of introduced
populations. Regional Patterns of Invasion. We used four different sources of information to construct a
chronological history of invasion for the Argentine ant in the United
States. These sources included published accounts, museum surveys,
personal surveys, and unpublished personal communications with
academics, pest company entomologists, and state agriculture extension
personnel. We then constructed a GIS database ( arcview for
MS Windows NT) that consisted of dates Argentine ants were first
detected (if ever) for counties in the United States. We used counties
as our unit of measure because information in many museum records and
publications was limited to the county level. This approach allowed a
detailed reconstruction of invasion for the southeastern United States
where counties are relatively small. In the western United States,
however, where counties are much larger, county-level analysis
exaggerated the area occupied. Therefore, rather than use area as our
overall measure of occupied territory, we used the number of counties.
To depict invasion history chronologically, we partitioned records into
four periods (1891–1910, 1911–1930, 1931–1950, and 1951–1999)
starting with the year of first detection (1891) in New Orleans, LA
( 32). Although our sources provide information on whether Argentine
ants have been recorded in a county, they do not always reveal whether
they were able to persist, or if they still occur. In some locations,
Argentine ants cannot survive outside of human-modified landscapes
because of their inability to tolerate arid (e.g., Arizona) or
cold-temperate (e.g., Minnesota and Illinois) climates. In other areas,
Argentine ants have been locally eradicated through control measures
(e.g., ref. 34) or displaced by the red imported fire ant,
Solenopsis invicta ( 35). Because the purpose of our study is
to examine the frequency and scale at which long-distance
jump-dispersal events occur, we do not distinguish between counties
that still have Argentine ants versus those that presently do not. We used the above reconstruction of the Argentine ant's invasion of
the United States along with published accounts that monitored invasion
fronts for at least 1 year (see below) to assemble a distribution of
rates of spread. Yearly jump-dispersal distances were estimated from
the invasion history of the Argentine ant by using all new foci (newly
occupied counties) through the year 1930. Early in the invasion, most
jump-dispersal events likely originated from New Orleans, the site of
the original introduction ( 32). However, later foci may have originated
either from New Orleans or from a closer infested county. Because
sources of introduction are not known, we estimated jump-dispersal
distances in two ways. First, we determined the distribution of
distances assuming New Orleans was the source for all new introductions
(out of New Orleans model). This method provides maximum estimates of
human-mediated jump-dispersal distances. For counties in California,
only the first record was used; subsequent spread was assumed to occur
from other counties in California. Second, we assumed that the source
of new foci came from the nearest county that had been occupied for at
least 1 year (nearest occupied county model); this method provides a
distribution composed of minimum estimates. While the contrasting models above provide a realistic range of
estimates for long-distance jump-dispersal at a regional scale, two
aspects of this approach may lead to overestimates of jump-dispersal
distances. First, jump-dispersal events occurring within counties would
not be detected by our methods. Average county diameter therefore sets
a lower limit to our estimation of jump-dispersal distances. However,
while important for local consolidation of occupied areas,
within-county jumps contribute less to the overall pattern of
colonization at the regional scale than do between-county jumps. For
this reason, we feel that a county-level approach is appropriate for an
analysis of the pattern of invasion at this regional scale. Second,
unequal spatial sampling could bias the reconstruction of the Argentine
ant's invasion history and also lead to overestimates of
jump-dispersal distances. For example, the absence of Argentine ants
from many intervening counties may have resulted from inadequate
surveys there. This would inflate estimates of jump-dispersal events in
our nearest occupied county model as the actual distances between a new
infestation and its putative source would be overestimated if a closer
infestation existed but remained unnoticed. There are several reasons
why we believe that this error is minimal. As a result of its prominent
pest status earlier this century ( 32), extensive surveys were conducted
throughout the southeastern United States, and L. humile was
not detected in many areas. Because information concerning absences is
as important as presence data, collections and published surveys that
did not detect Argentine ants are also included in the supplemental
tables, which are published on the PNAS web site ( www.pnas.org). In
addition, rates of spread from budding are far too slow to account for
the occupation of these distant counties (see Discussion). Local Patterns of Invasion. Rates of spread at invasion fronts through colony budding were collated
from the literature and unpublished surveys. For each account, we noted
the location, duration of study, number of fronts monitored, habitat
type, and the minimum and maximum rates of spread. Subsequent analyses
used the maximum rate of spread from each study. Interannual variation in rates of spread by budding was examined by
following 20 distinct invasion fronts of Argentine ants for 3–4 years
(depending on the site) in riparian woodlands in northern California
(see ref. 29 for complete description of study areas). Variation in
invasion rate across years was examined by using repeated-measures
ANOVA. Only fronts that spread at least 3 years were included in
subsequent analyses ( n = 10). Invasion fronts that did
not advance were dropped from this analysis because the environment at
these sites may not be abiotically suitable for Argentine ants ( 29).
For example, Holway ( 29) found that Argentine ants spread at sites with
permanent stream flow but did not at sites with intermittent stream
flow. Therefore, including sites at which Argentine ants did not spread
would overestimate variation in invasion rates. Worldwide Distribution. Argentine ants now occur throughout the world, with at least 28
separate introductions known from six continents and many oceanic
islands (Fig. ). The earliest reported
introduction occurred in 1882 on Madeira Island. Although new records
continue to accumulate, many populations were established by the 1950s
(Table 1). Although Fig. undoubtedly
underestimates the Argentine ant's current distribution, it
illustrates that this species has become a worldwide invader in areas
with Mediterranean-type climates and in some temperate and subtropical
areas. Given the innate limitations in the Argentine ant's natural
dispersal capabilities, it is clear that the colonization ability of
this species hinges on human commerce. Although the introduced range of
this species is now fairly well known, its native range in South
America deserves further study. Morphological and genetic similarity
exists between many introduced populations and populations along the
Rios Parana and Uruguay in northern Argentina (unpublished data).
However, at this point it is unclear whether populations from Brazil
belong to L. humile (ref. 36 and P. Ward, personal
communication).
| Table 1Dates Argentine ants were first detected in countries and
islands throughout the
world |
Regional Patterns of Invasion. In addition to published accounts, 66 collections provided
information regarding the presence or absence of Argentine ants in
counties in the United States (see supplemental Tables 3 and 4).
Linepithema humile has been recorded in 335 counties in 21
states (Fig. ). Their initial spread
throughout the United States was characterized by new populations
established well away from the original site of introduction (Fig.
A and B). The two models of jump dispersal
provide a range of estimates of annual jump-dispersal distances based
on the pattern of spread of Argentine ants across the United States
through 1930: out of New Orleans model = 361.72 ± 416.82 km
(mean ± 1 SD); nearest occupied county model = 160.58
± 323.83 km (see Fig. ). Although these distributions are different
(paired t test, P < 0.05), they are of the
same order of magnitude. Moreover, these distances are an order of
magnitude greater than the estimated lower limit of the distribution
set by average county diameter (28.79 ± 17.69 km,
n = 3,140 counties) (see Fig. ), suggesting that, on
average, jumps are occurring across more than one county. Apparent from
the later stages of this invasion, Argentine ants filled in their range
as new populations were discovered between originally separate
colonized counties forming continuous regions of occupation (Fig.
C and D). When plotted against time, the pattern
of occupation appears sigmoidal (Fig. ).
In the first 20 years of this invasion, few new counties were
colonized. This was followed by a period of rapid expansion through the
1930s and then a period (that continues through the present) during
which few new counties were colonized (Fig. ). Although historical
records for Argentine ants are fairly detailed, the current
distribution of this species in the southeastern United States is
unknown given the widespread presence of the red imported fire ant
which may have displaced L. humile from portions of its
former range ( 35).
Local Patterns of Invasion. Sixteen studies followed Argentine ant invasion fronts for at
least 1 year (Table 2). Although examined
in different habitats throughout the world, with few exceptions, the
maximum yearly rates of spread reported were largely consistent across
sites (0.154 ± 0.021 km) (mean ± SE). Jump-dispersal
distances, in contrast, averaged three orders of magnitude higher. When
both distributions are plotted together, the disparity in dispersal
distances between these two processes is clearly evident (Fig.
). Despite the consistency among studies
in their maximum rate of invasion (Table 2), within locations, rates of
spread show much temporal variation. For example, at 10 sites in
northern California, average rates of spread varied by a factor of 3
over a 3- to 4-year period ( F = 3.14, df = 3,18,
P = 0.05) (Fig. ).
| Table 2Reported rates of spread by budding reproduction for
Argentine ants |
The results of this study demonstrate the importance of
human-mediated jump-dispersal in determining invasion dynamics
subsequent to establishment. This is evident from both the worldwide
and regional reconstruction of Argentine ant invasion history. Fig.
illustrates that virtually no country or island is too isolated for
potential establishment. The patchy and somewhat piecemeal pattern of
invasion at the regional scale (Fig. ) is in contrast to the
well-defined and continuous invasion fronts that characterize some
invasions (refs. 1 and 3, but see ref. 37). To account for this
observed pattern through a diffusion-like processes, the rates of
spread through colony budding would have to be three orders of
magnitude higher (Fig. ). Taken together, the results of this study
provide an unusually thorough analysis of a worldwide invasion. In
addition, unlike previous reconstructions of invasion history, this is
the first comprehensive assessment for a species that spreads primarily
by human-mediated jump dispersal. Long-distance jump-dispersal events are believed to be rare and
difficult to measure. Even when infrequent, long-distance dispersal
events may greatly influence overall invasion rate. For example, using
simulations, Higgins and Richardson ( 12) demonstrated that
long-distance dispersal by as little as 0.001% could increase
predicted rate of spread by an order of magnitude. Our study
illustrates that for species that associate closely with humans these
events are common and can drive the overall pattern of invasion.
Specifically, by determining the distance and rate at which
long-distance jump-dispersal events occur, we can estimate parameters
for use in modeling the spread of invasive organisms that rely heavily
on these events. The spread of the house finch ( Carpodacus
mexicanus) throughout the northeastern United States provides
another good example of how historical data sets can enhance the
predictive quality of modeling efforts. Because of long-term census
data provided by Christmas Bird Counts, Veit and Lewis ( 10) were able
to measure long-distance dispersal events and to estimate their
relative contribution to overall invasion rate ( 9, 10). When jump dispersal is common, simple diffusion fails to describe the
pattern of spread. Such cases are better modeled by stratified
diffusion, where more than one process is involved in the spread of an
invading species. This is particularly clear in Argentine ant invasions
given the huge disparity between rates of spread for alternate modes of
dispersal (Fig. ). Stratified diffusion “may be the rule rather
than the exception” for invading organisms ( 3) and quantification of
alternate modes of dispersal is key to the development of more
realistic models to predict patterns of invasion dynamics in such
cases. In particular, for species that spread through stratified
diffusion, the distance and rate at which new foci are created may be
more important than the rate of spread through diffusion from
established foci. The extent to which species spread by stratified diffusion may
influence the implementation of control strategies. For example, the
effectiveness of control measures can be greatly increased by
preventing the establishment of new foci or by eliminating new foci
rather than focusing efforts on established invasion fronts ( 37). As
demonstrated in this study, the establishment of new foci through
human-mediated jump dispersal is of paramount importance in the spread
of Argentine ants and control efforts should focus on preventing their
spread through these means. However, Argentine ants establish new
populations easily. For example, laboratory experiments demonstrate
that queens with as few as 10 workers exhibit high rates of colony
growth, suggesting that such small propagules can easily establish
beachheads ( 38). Another interesting result of this study was that, at different spatial
scales, rates of invasion vary through time. First, at a regional
scale, there was a clear lag time in the spread of Argentine ants
throughout the United States (Fig. ). Such lag times are common
features of invasions but their underlying causes often remain unclear
( 14, 39). Explanations include the inherent features of population
growth, environmental changes that benefit invasive species such as
increased urbanization, and genetic changes occurring subsequent to
introduction that result in fitness increases ( 39). The relatively
rapid spread of Argentine ants following the lag period may also
reflect coincident increases in human transportation and commerce.
Although unequal sampling effort could give rise to the pattern seen in
Fig. , this is unlikely because of the intense sampling for Argentine
ants early this century ( 32, 33). Second, at a local scale, rates of
diffusion through budding reproduction varied temporally across sites
(Fig. ), possibly because of differences in abiotic conditions (e.g.,
precipitation) among years. A better appreciation of the causes
underlying such variation will be useful in guiding theoretical work on
the spread of invading organisms. While we describe a case of stratified diffusion involving two
dispersal processes, other invasions are more complex. For example, red
imported fire ants ( Solenopsis invicta) and zebra mussels
( Dreissena polymorpha) spread via three primary modes of
dispersal. In red imported fire ants, budding, mating flights and human
introductions all contribute to overall spread ( 40). In zebra mussels,
spread results from diffusive (within watershed), advective (within
watershed), and jump-dispersal (across watershed) events ( 41, 42).
Quantifying the rates and distances of these dispersal events is
therefore difficult. For the spread of zebra mussels, an attempt has
been made to examine the potential for over-ground dispersal between
watersheds by examining the rates and distances which recreational
boaters travel in Wisconsin ( 11). Approaches such as these offer great
promise in the quantification of jump-dispersal events. A major challenge in the study of biological invasions lies in
determining factors that contribute to or limit the spread of exotic
species. This can be difficult because detailed chronological histories
of invasions rarely exist. In addition, despite the obvious value in
making the study of biological invasions a more predictive science,
estimating the rate and pattern of invasions remains a difficult task.
Given the unpredictable nature of long-distance jump-dispersal events,
accurately determining the range at which they occur can greatly
enhance future modeling efforts. A careful reconstruction of invasion
dynamics at contrasting spatial scales will also aid in the development
management or eradication strategies. This manuscript would not have been possible without the help
of all of the museum curators mentioned in supplemental Table 4. In
addition, we would like to especially thank A. Bachman, B. Brown, S.
Cover, M. Deyrup, L. Loope, B. Norden, N. Reimer, T. Schultz, R.
Snelling, D. Summers, P. Ward, and A. Wild for sharing unpublished
data and for their hospitality. Work in Argentina was made possible by
P. Cichero, F. Menvielle, and L. Raffo from the Administracion de
Parques Nacionales Argentinas, R. (Mima) Venguet for logistical
support, and I. C. Quilmes for many refreshing insights. The
manuscript benefited from comments by L. Levin, T. Price, and P. Ward.
This work was supported by the Canon National Parks Science Scholars
Program (A.V.S.), U.S. Department of Agriculture National Research
Initiative Competitive Grants Program Grants 99-35302-8675 (D.A.H.),
and 00-35302-9417 (A.V.S.); and National Science Foundation Grant
DEB-9610306 (T.J.C.). 1. Shigesada N, Kawasaki K. Biological Invasion: Theory and Practice. New York: Oxford Univ. Press; 1997. 2. Skellam J G. Biometrika. 1951;38:196–218. [PubMed] 3. Hengeveld R. Dynamics of Biological Invasions. London: Chapman and Hall; 1989. 4. Okubo A, Maini P K, Williamson M H, Murray J D. Proc R Soc London Ser B. 1989;238:113–125. [PubMed] 5. Andow D A, Karieva P M, Levin S A, Okubo A. Landscape Ecol. 1990;4:177–188. 6. Allen L J S, Allen E J, Kunst C R G. J Ecol. 1991;79:1123–1135. 7. Okubo A. Diffusion and Ecological Problems: Mathematical Models. New York: Springer; 1980. 8. Kot M, Lewis M A, van der Driessche P. Ecology. 1996;77:2027–2042. 9. Lewis M A. In: Spatial Ecology: The Role of Space in Population Dynamics and Interspecific Interactions. Tilman D, Kareiva P, editors. Princeton, NJ: Princeton Univ. Press; 1997. pp. 46–69. 10. Veit R R, Lewis M A. Am Nat. 1996;148:255–274. 11. Buchan L A J, Padilla D K. Ecol Appl. 1999;9:254–265. 12. Higgins S L, Richardson D M. Am Nat. 1999;153:464–475. 13. Hengeveld R. Trends Ecol Evol. 1994;9:339–342. 14. Hastings A. Ecology. 1996;77:1675–1679. 15. Erickson J M. Psyche. 1971;78:257–266. 16. Bond W, Slingsby P. Ecology. 1984;65:1031–1037. 17. Ward P S. Hilgardia. 1987;55(2):1–16. 18. Cammell M E, Way M J, Paiva M R. Insectes Soc. 1996;43:37–46. 19. Human K G, Gordon D M. Oecologia. 1996;105:405–412. 20. Suarez A V, Bolger D T, Case T J. Ecology. 1998;79:2041–2056. 21. Holway D A. Oecologia. 1998;116:252–258. 22. Suarez A V, Richmond J Q, Case T J. Ecol Appl. 2000;10:711–725. 23. Cole F R, Medeiros A C, Loope L L, Zuehlke W W. Ecology. 1992;73:1313–1322. 24. Human K G, Gordon D M. Cons Biol. 1997;11:1242–1248. 25. Bolger D T, Suarez A V, Crooks K R, Morrison S A, Case T J. Ecol Appl. 2000;10:1230–1248. 26. Holldobler B, Wilson E O. The Ants. Cambridge, MA: Harvard Univ. Press; 1990. 27. Passera L. In: Exotic Ants: Biology, Impact, and Control of Introduced Species. Williams D F, editor. Boulder: Westview Press; 1994. pp. 23–43. 28. Fuentes E R. In: Biogeography of Mediterranean Invasions. Grives R H, Di Castri F, editors. Cambridge, MA: Cambridge Univ. Press; 1991. pp. 43–49. 29. Holway D A. Oecologia. 1998;115:206–212. 30. Markin G P. Ann Entomol Soc Am. 1970;63:1238–1242. 31. Fluker S S, Beardsley J W. Ann Entomol Soc Am. 1970;63:1290–1296. 32. Newell W, Barber T C. US Dep Agric Bur Entomol Bull. 1913;122:1–98. 33. Barber T C. US Dep Agric Bur Entomol Bull. 1916;377:1–23. 34. Harned R W, Smith M R. J Econ Entomol. 1922;15:261–264. 35. Wilson E O. Evolution. 1951;5:68–79. 36. Orr M R, Seike S H. Oecologia. 1998;117:420–425. 37. Moody M E, Mack R N. J Appl Ecol. 1988;25:1009–1021. 38. Hee J J, Holway D A, Suarez A V, Case T J. Cons Biol. 2000;14:559–563. 39. Crooks J, Soulé M E. In: in Proceedings of the Norway/UN Conference on Alien Species. Sandlund O, Schei P, Viken A, editors. Norway: Trondheim; 1996. pp. 39–46. 40. Porter S D, Van Eimeren B, Gilbert L E. Ann Entomol Soc Am. 1988;81:913–918. 41. Johnson L E, Carlton J T. Ecology. 1996;77:1686–1690. 42. Johnson L E, Padilla D K. Biol Cons. 1996;78:23–33. 43. Zimmerman E C. Proc Hawaiian Entomol Soc. 1941;11:108. 44. Crowell K L. Ecology. 1968;49:551–555. 45. Tremper B D. Ph.D. thesis. Berkeley: Univ. of California; 1976. 46. Dale W E. Revista Peruana Entomol. 1974;17:126–127. 47. Morrison L W. Acta Oecologica. 1997;18:685–695. 48. Donisthorpe H. Entomol Record J Var. 1930;42:13–16. 49. King J. M.S. thesis. San Diego: Univ. of California; 1998. 50. Way M J, Cammell M E, Paiva M R, Collingwood C A. Insectes Soc. 1997;44:415–433. 51. Chopard L. Ann Epiphyties. 1921;7:237–266. 52. Kutter H. Mitt Schweiz Entomol Ges. 1981;54:171–172. 53. Casevitz-Weulersse J. Ann Soc Entomol Fr. 1974;10:611–621. 54. Martinez M D, Ornosa C, Gamarra P. Boln Asoc Esp Ent. 1997;21:275–276. 55. Stoll O. Mittheil Schweize Entomol Gesellschaft. 1898;10:120–126. 56. Prins A J, Robertson H G, Prins A. In: Applied Myrmecology, A World Perspective. Vandermeer R K, Jaffe K, Cedeno A, editors. Boulder: Westview Press; 1990. pp. 25–33. 57. Collingwood C A, Tigas B J, Agnosti D. J Arid Environ. 1997;37:505–512. 58. Sugiyama T. Jpn J Appl Entomol Zool. 2000;44:125–126. 59. Pasfield G. Australian Nat Hist. 1968;16:12–15. 60. Green O R. Weta. 1990;13:14–15. 61. Della Lucia T M C, Loureira M C, Chandler L, Freire J A H, Galvno J D, Fernandes B. Experientiae. 1982;18:67–94. 62. Woodworth C W. Univ Calif Agric Exp Stn Bull. 1910;207:51–82. 63. Fullaway D T. Proc Hawaiian Entomol Soc. 1944;12:26.
|
This article has been 
