Exploring associations between international trade

14 Pages·2011·0.45 MB·English

BiolInvasions
DOI10.1007/s10530-011-0039-2
ORIGINAL PAPER
Exploring associations between international trade
and environmental factors with establishment patterns
of exotic Scolytinae
Lorenzo Marini • Robert A. Haack •
Robert J. Rabaglia • Edoardo Petrucco Toffolo •
Andrea Battisti • Massimo Faccoli
Received:16December2010/Accepted:5June2011
(cid:2)SpringerScience+BusinessMediaB.V.2011
Abstract Althoughinvasion of exoticambrosiabee- evaluatedtherelativeimportanceofthepotentialdrivers
tles(fungusfeeders)andbarkbeetles(phloemfeeders) ofspeciesrichnessofnativeandexoticScolytinaein20
(Coleoptera:Curculionidae:Scolytinae)isconsidereda European countries and the 48 contiguous continental
majorthreattoforesthealthworldwide,nostudieshave US states. Analyses were conducted separately for
quantitativelyinvestigatedtheanthropogenicandenvi- ambrosiaandbarkbeetlespecies.Valueofimportswasa
ronmental factors shaping the biogeographical patterns strong predictor of the number of exotic Scolytinae
of invasion by these insects across large spatial scales. speciesinbothregions.Inaddition,intheUSA,warmer
Theprimaryaimofthisstudywastoassesstherelative and wetter climate was positively linked to increased
importance of international trade and several environ- numbers of both native and exotic ambrosia beetles.
mental variables of the recipient region on species Forest heterogeneity and climatic heterogeneity and
richnessofestablishedexoticScolytinae.Asareference, secondarily forest area were key drivers in explaining
we also evaluated the relationships between the same patterns of species richness for native bark beetles but
environmental variables and species richness of native not for exotic species in both regions. Our findings
Scolytinae. Using an information-theoretic framework suggest that if current infestation levels continue on
for model selection and hierarchical partitioning, we importedplantsandwoodpackagingmaterial,increas-
inginternationaltradewilllikelyleadtomoreestablish-
ments of exotic Scolytinae with concomitant negative
effects on forest health in both Europe and the USA.
ComparedtoEuropetheriskofinvasionappearshigher
in the USA, especially for ambrosia beetles in the
L.Marini(&)(cid:2)E.PetruccoToffolo(cid:2) southeastern USA where the climate appears highly
A.Battisti(cid:2)M.Faccoli
suitableforexoticestablishment.
DepartmentofEnvironmentalAgronomyandCrop
Production,UniversityofPadova,Vialedell’Universita`
16,35020Legnaro,Padova,Italy Keywords Alien insect (cid:2) Ambrosia beetle (cid:2)
e-mail:[email protected]
Bark beetle (cid:2) Forest pest (cid:2) Import (cid:2) Scolytidae
R.A.Haack
NorthernResearchStation,USDAForestService,
1407SHarrisonRoad,EastLansing,MI48823,USA Introduction
R.J.Rabaglia
Invasive exotic forest insects are considered a major
ForestHealthProtection,USDAForestService,
1601NKentStreet,RPC–7,Arlington,VA22209,USA threat to forest health (Gandhi and Herms 2010). In
123
L.Marinietal.
particular, ambrosia beetles and bark beetles (Cole- between extinction and colonization rates, with
optera: Curculionidae: Scolytinae) are amongst the higher extinction rates in smaller versus larger areas
most important forest pests causing significant eco- (MacArthur and Wilson 1967). The habitat diversity
nomic damage to trees worldwide. In the past few hypothesis predicts that larger areas are more likely
decades, establishment of exotic species has steadily to contain greater environmental heterogeneity and
increased in both Europe (Roques et al. 2009; therefore, more species as predicted by niche theory
Kirkendall and Faccoli 2010; Sauvard et al. 2010) (Huston 1979). A large body of biogeographical
and North America (Haack 2001, 2006; Rabaglia research also supports the view that, in regions of
et al. 2006, 2008), posing severe threats to native roughly equal habitat area, temperature is the prime
forestecosystems(Holmesetal.2009;Aukemaetal. determinant of species richness in temperate areas as
2010; Haack and Rabaglia 2011). the amount of available energy sets limits to the
Exotic insect invasions can be described in terms species richness of the system (Currie 1991). All of
of a series of ecological barriers or filters, from the processes described above are expected to be
arrivaltoestablishmenttospread,whichmayprevent important for both native and established exotic
an organism from proceeding to the next stage species. Whereas these environmental factors dom-
(Liebhold and Tobin 2008). The first stage corre- inate the larger scale domains, biotic interactions are
sponds to the human-mediated arrival of a pool of expected to be important at local and smaller spatial
exotic species that constitute the potential colonizers scales (Milbau et al. 2009). However, potential
ofadefinedrecipient region.Thearrivalofambrosia effects of native species on the invasiveness of
and bark beetles is expected to be enhanced by exotic Scolytinae are complex and difficult to
increases in global trade (e.g. Haack 2006; Haack predict as both competition and facilitation can be
et al. 2010; Koch et al. 2010). For instance, these expected as important forces in the community
organisms can be easily transported in live plants, ecology of Scolytinae that often breed in multiple-
plant products, wood packaging material, and cargo species associations within the same host plant (Lu
containers where they are often sheltered from et al. 2007).
detection (Work et al. 2005; Brockerhoff et al. Althoughmoststudiesoffactorsthatcontributeto
2006a, b; Haack 2006; McCullough et al. 2006; successful establishment of exotic species have
Brockerhoff 2009; Colunga-Garcia et al. 2009). largely focused on species traits and local habitat
Thenextstageofthebiologicalinvasionprocessis characteristics (both biotic and abiotic) that deter-
establishmentofasubsetofthespeciesthatarrivedat mine vulnerability to invasion, increasing empirical
any given point of entry. Once an exotic insect evidencehasdemonstrated thatpropagulepressureis
species arrives in a new ecosystem, establishment cruciallyimportantinbothofthesestagesofinvasion
will depend on various abiotic and biotic factors (Lockwood et al. 2009; Simberloff 2009). The
(Holmes et al. 2009; Roura-Pascual et al. 2011). At linkage of vulnerability to invasion with propagule
this stage, several factors are expected to determine pressure may be especially strong for exotic Scolyt-
thenumberofScolytinaespeciesthatcanestablishin inae due to their high facility to be transported
a region: habitat area (Stevens 1986), resource and accidentally in various commodities through interna-
environmental heterogeneity (Huston 1979), climate tionaltradeandtravel(Worketal.2005;McCullough
(Coulson 1979), and biotic interactions with native et al. 2006). Although recent works indicate that
communities (Lu et al. 2007), while availability of ongoingbiologicalinvasionsmightbewellpredicted
novelnichescreatedbyhumansisnotexpectedtobe with variables related to human activity (Pysˇek et al.
important.Apositivespecies–arearelationshipisone 2010; Essl et al. 2011; but see Roura-Pascual et al.
of the most consistent biogeographical patterns 2011), no studies have quantitatively assessed the
(Rosenzweig 1995). Two suggested mechanisms relativeimportanceofbothnaturalandanthropogenic
underpinning this positive relationship are the area processeson thepatternsofinvasionbyexoticforest
per se and the habitat diversity hypotheses. The area insects across large spatial scales (but see Langor
per se hypothesis derives from the island biogeogra- et al. 2009). However, the need for improved
phytheoryandassumesthatthenumberofspeciesin understanding of the interactions between species
a region represents an area-mediated equilibrium invasionsandotherglobalchangeprocesseshasbeen
123
Exploringassociations
recently highlighted as a research priority for reduc- Data on Scolytinae distribution
ing the threat of invasive forests pests (Gandhi and
Herms 2010). Information on the distribution of native and exotic
Hence, the primary aim of our study was to Scolytinae was gathered in both regions from
evaluatetherelativeimportanceofinternationaltrade published and unpublished sources. We included in
and various abiotic and biotic environmental vari- our analyses only those exotic species that have
ablesinshapingpatternsofestablishedexoticspecies established self-sustaining populations, regardless of
richnessforambrosiaandbarkbeetlesinbothEurope whether they cause ecological or economic impacts.
and the USA. In particular, we evaluated whether The list of native Scolytinae species of each
propagule pressure was more important than the European country was obtained from the catalogue
quality of the recipient regions in explaining species ofthePalaearcticscolytidspecies(Kn´ızˇek2011).The
richness of the exotic Scolytinae. In addition, as a list of exotic Scolytinae naturalized in Europe was
means to identify the unique contribution of propa- recently compiled by Kirkendall and Faccoli (2010)
gule pressure and environmental factors to exotic and Sauvard et al. (2010). The list by country was
species richness, we studied the relationships of the createdusingtheEuropeandatabaseforexoticorgan-
same environmental variables with native species of isms DAISIE (Delivering Alien Invasive Species
ambrosia and bark beetles in Europe and the USA. InventoriesforEurope)(DAISIE2009),theEuropean
NetworkonInvasiveAlienSpecies(NOBANIS2010),
andtheFaunaEuropaea(Kn´ızˇek2004).
Methods ThelistofnativeandexoticScolytinaenaturalized
in the USA was initially based on Wood (1977,
Study regions 1982), Wood and Bright (1992), and Bright and
Skidmore(1997,2002).Updatedlistsbasedprimarily
We had two study regions: Europe and the on scientific literature were published by Haack
continental USA. The European region consisted (2001, 2006), Rabaglia et al. (2006), Cognato et al.
of 20 countries, while US region consisted of the (2009), and Haack and Rabaglia (2011). The lists
continental 48 contiguous US states. For each of used in the present paper were current through 2010
these countries and states, we had detailed and and integrated with unpublished records of bark
comparable information about the distribution of beetles in the USDA Forest Service, Early Detection
both native and exotic ambrosia and bark beetles and Rapid Response project (Rabaglia et al. 2008;
(Scolytinae). Rabaglia et al. unpublished data).
Despitethedata’scoarsespatialresolution,webelieve
thatthisisthelargestavailabledatasetonthedistribution
of Scolytinae species in temperate countries and no
information at a finer spatial resolution is currently
available.Weacknowledgethatthespeciescountsmight
be partiallyincompleteand thatthenumbers ofknown
native and exotic Scolytinae would likely increase if
more thorough surveys were conducted; however, we
feel that the relative rankings between countries and
stateswillremainratherstable.Wethereforeconsiderthe
data collated here provide a sound basis for gaining
insights into the potential drivers of Scolytinae species
richnessacrossthetwoselectedregions.
Fig.1 Frequency distribution of seed-feeding, phloem-feed- Scolytinae feeding habits
ing (bark beetles), and fungus-feeding (ambrosia beetles)
native and exotic Scolytinae species pool within Europe and
The Scolytinae exhibit several forms of feeding
theUSA.Therelativeproportion(%)ofeachfeedingguildis
alsoshown.Twigfeederswerepooledwithbarkbeetles specialization (Wood 1982), and these different
123
L.Marinietal.
feeding habits likely affect the probability that for bark- and wood-boring insects. Total imports are
individual species will be transported in imported therefore expected to be more related to the intro-
goods and associated wood packaging material. duction of exotic species than simply imports of
Given the above, and that different feeding habits wood products. Although exotic species richness of
mayalsoinfluenceestablishmentsuccess,wesplitthe other taxonomic groups such as plants or fungi may
Scolytinae into three groups: better reflect historical rather than contemporary
(i) Phloem-feeding species which feed in the human activities, species richness of exotic insects
phloem layer, and include species that develop in has been demonstrated to be more closely related to
twigs (hereafter referred to as ‘bark beetles’). The currentratherthanhistoricalsocioeconomicvariables
trophic niches of these species are separated by host (Essl et al. 2011). We therefore used the value of
tree species as well as by host requirements within recent import data to explain the number of exotic
the same species such as phloem thickness, tree Scolytinae species in both regions.
defense chemistry, nutritional quality and water Indicatorsofcumulativeeconomicprosperitysuch
content (Byers 1989). aswealthareoftenconsideredbetterpredictorsofthe
(ii) Fungus-feeding species, which bore into the distribution of established exotic species than vari-
woodandfeedprimarilyonsymbioticfungi(‘ambro- ables measuring flow of capital and goods within
sia’) that grow along the tunnel walls (hereafter shorter time spans (Pysˇek et al. 2010). However, we
called‘ambrosiabeetles’).Thesespeciescandevelop did not include any cumulative measures because
in both living and dead wood and therefore usually exotic Scolitynae do not require further human
exhibit low host tree specificity. intervention or novel niches created by humans to
(iii) Seed-feeding species, which develop in hard establish. The expected main limiting factors for
seeds and fruit. We excluded this group from the establishment of exotic Scolytinae are suitable cli-
analyses because their numbers were too low to maticconditions,availabilityofhosttrees,andbiotic
include as a separate analysis. interactionswithnativecommunitieswithinthesame
or higher trophic levels.
For Europe, we used the average annual value of
Explanatory variables
goods imported during the period 1999-2009 based
onofficialeconomicstatisticsoftheEuropeanUnion
We evaluated the role of several variables indicative
(www.ec.europa.eu/eurostat). For the USA, similar
of international trade and the local environment for
data were available for the period 2008–2010 (US
eachcountry/statethatwereexpectedtobeimportant
Economic Census Bureau, US International Trade in
drivers of Scolytinae species richness at the spatial
Goods and Services FT900). For both regions, we
scale of our analyses (Table 1). We selected a priori
used data for the final destination of the imports,
the most likely predictors based on current knowl-
rather than the first port of arrival because the vast
edge of Scolytinae ecology and biogeography as
majorityofimports arrive inshipping containersthat
explained below.
arenotopeneduntiltheyreachtheirfinaldestination.
IMPORT was log-transformed in all analyses to
International trade improve linearity and uniformity. Although import
data were available only for rather short periods,
By analyzing exotic Scolytinae interceptions in the longer time series were available only for other
USA between 1985 and 2000, Haack (2001) found indicators of flow of capital such as gross domestic
that 73% of the interceptions were found in associ- product (GDP). For the period for which both vari-
ation with wood packing material, only 22% in food ables were available the correlation between imports
or plants, and 5% in unspecified material (no similar and GDP was always very high (r = -0.90). Pre-
information is available for Europe). On the basis of liminary analyses using GDP with longer time series
these data we used the total value of imports instead of IMPORT yielded very similar results,
(IMPORT) as proxy for propagule pressure because indicating that relative differences in flow of capital
most products are now transported in crating or on between countries or states have remained almost
palletsandtherebycouldserveasapotentialpathway constant over the last decades.
123
Exploringassociations
Table1 DescriptivestatisticsoftheexplanatoryvariablesforbothEurope(20countries)andtheUSA(48continentalcontiguous
states)includedinthisstudy
Variablenameanddescription Unit Mean(SD)
Europe(n=20) USA(n=48)
IMPORTa Cumulativevalueofimports(allgoods) 109$year-1 181.81(189.66) 36.80(59.47)
AREA Areacoveredbyforest km2 68,359(81,015) 61,607(74,877)
PRECIPb Meanannualprecipitation mm 799(786) 909(352)
TEMPc Minimumtemperatureofthecoldestmonth (cid:3)C -0.45(0.47) -0.06(0.60)
PRECIP-H SDofmeanannualprecipitation mm 174(169) 137(150)
TEMP-H SDofmeanannualtemperatured (cid:3)C 1.99(1.93) 1.76(1.62)
FOREST-H Foresthabitatheterogeneity ShannonH’ 1.33(1.35) 1.41(0.36)
Refertomethodsfordetailsaboutthedatasources
a Euro–dollarexchangerate:1€=1.30$
b BioclimaticvariableBIO12fromWorldClim(http://www.worldclim.org/bioclim)
c BioclimaticvariableBIO6fromWorldClim(http://www.worldclim.org/bioclim)
d BioclimaticvariableBIO1fromWorldClim(http://www.worldclim.org/bioclim)
Environmental factors (both ambrosia and bark beetles) would decrease
when moving from warmer low-latitude to colder
Given that the vast majority of native and exotic high-latitude areas according to the species–energy
Scolytinae in Europe and the USA utilize trees to hypothesis (Currie 1991). Warmer climes are also
completetheirlifecycle,weusedtheareacoveredby expected to generally favor the establishment of
forest (AREA) instead of administrative area to exoticspecies(Waltheretal.2009).Precipitationwas
account for the species–habitat area relationship included because it is a key climatic variable
(Stevens1986).Asforestareawasstronglycorrelated affecting tree susceptibility and insect physiology
toadministrativearea(r = 0.87and0.67forEurope (Coulson1979).Forambrosiabeetles,precipitationis
s
and the USA, respectively) the simultaneous inclu- also a key factor in the development of their
sion of the two variables was not possible. AREA associated fungi (Wood 1982).
was based on data from CORINE for Europe (www. Given that broad geographical gradients might be
eea.europa.eu), and the National Forest Inventory associated with large variations in environmental
andAnalysisdatabasefortheUSA(USDA-FS2010). gradients,particularlyasthesizeofthesamplingunit
We expected that increasing the area covered by increases, we quantified three measures of environ-
forests would likely increase the number of both mentalheterogeneity.Weusedthestandarddeviation
native and exotic Scolytinae species by increasing of mean annual temperature (bioclimatic variables
habitatavailability,irrespectiveoftheirfeedinghabit. BIO1 in WorldClim) (TEMP-H) and precipitation
AREA was log-transformed in all analyses to (PRECIP-H) within each country and state as a
improve linearity and uniformity. measure of climatic heterogeneity. We expected that
Two climatic variables were used based on the number of both native and exotic Scolytinae
monthly data: minimum temperature of the coldest should increase with increasing number of climatic
months (TEMP) and mean annual precipitation niches available. Finally, we quantified forest habitat
(PRECIP). Both were derived from the WorldClim diversity with the Shannon–Wiener diversity index
database and corresponded to the bioclimatic vari- (FOREST-H) of the forested area covered by differ-
ables BIO6 and BIO12, respectively (Hijmans et al. ent forest categories within each country or state as
2005; 1 km resolution; 1960-1990 period). Both explained below. We expected this variable to be
variableswereaveragedwithineachcountryorstate. positivelyrelatedtothediversityofhosttreespecies,
As our two regions cover wide latitudinal gradients, and therefore to the diversity of trophic niches. As
we expected that the number of native Scolytinae ambrosiabeetleshavelowerhostspecificitythanbark
123
L.Marinietal.
beetle species (Wood 1982), we expected the bark between native and exotic Scolytinae would indicate
beetles to be more related to forest habitat diversity stronginfluenceofthesameenvironmentaldriverson
than ambrosia beetles. For Europe the only informa- both species groups, i.e. large-scale environmental
tion available on forests was the proportion of ICP factors enhancing native species richness would also
(International Co-operative Programme on Assess- increase exotic species richness.
ment and Monitoring of Air Pollution Effects on
Forests) plots per forest type within each country. Multi-model inference and hierarchical partitioning
ThissystemclassifiedEuropeanforestsinto14broad
categories (EEA 2007). Although the latter measure Due to the relatively low number of replicates
was not optimal to estimate the cover of different (n = 20 in Europe and n = 48 in the USA) and the
forest categories, it was the only variable available relatively high number of potential predictors, we
for the whole of Europe. For the USA, we retrieved used multi-model inference within an information-
up-to-date information on the area covered by 31 theoretic framework to evaluate the role of the
forest categories by state from the USDA Forest selectedvariablesinexplainingpatternsofScolytinae
Service, Forest Inventory Data Online (FIDO) species richness (Burnham and Anderson 2002). As
(USDA-FS 2010). Although the European and US explained above, all selected variables in Table 1
classification systems are not fully comparable the corresponded to ecological meaningful predictors of
criteriato define the forest categories inbothregions exoticScolytinaespeciesrichnesspatterns.Fornative
were similar. Each category is defined as a physiog- Scolytinae, we excluded IMPORT from the analyses
nomically uniform group of plant associations shar- because the native species were not expected to be
ing one or more dominant tree species. The higher directly related tothis variable. Moreover, given that
number of categories in the USA than in Europe wehadnoapriorihypothesessupportinginteractions
reflects the larger climatic gradient and greater betweentheconsideredexplanatoryvariables,wedid
diversity of tree species in the USA. not include any interaction terms in our model
selection procedure. We used ordinary least square
Data analysis multiple linear regression to estimate model param-
eters as model esiduals approximated a normal
Relationship between native and exotic Scolytinae distribution. The number of native or exotic species
species richness reported for each country or state was used as the
response variable in these models. We ran separate
Considering that at our large spatial scale there was analysesforambrosiaandbarkbeetles.Similarly,the
no clear hypothesis on the direct effects of native low number of exotic bark beetles in Europe did not
species richness on exotic species richness, we did allow for analysis. Given that severe collinearity
notincludenativespeciesrichnessinourmultivariate among explanatory variables can hamper identifica-
analyses. For ambrosia beetles, we expected weak tion of the most causal variables also within an
interactions between natives and exotics due to their information theoretic framework, we calculated a
low host specificity and broad trophic niches. For variance inflation factor (VIF) within our set of
bark beetles, no prediction was made on the interac- predictors relative to all our response variables in
tion between native and exotic species, as both both regions. If apredictorhas avalueofVIF[10, it
competition and facilitation can be expected as is considered an indicator of strong collinearity
important processes in the community ecology of associated with that particular predictor. As none of
Scolytinae that breed in tree phloem and often in our predictors had a VIF[10 (maximum VIF was
multiple-species associations (Lu et al. 2007). How- 6.26 for AREA in Europe), we retained all the
ever, we performed a separate univariate analysis variables in our multi-model inference procedure.
where we tested the co-variation between native and Our information-theoretic approach compared the
exotic species richness separately for ambrosia and fit of all the possible candidate models obtained by
bark beetle species in the two regions. According to thecombinationofourpredictorsusingsecond-order
the ‘rich get richer’ hypothesis (e.g. Stohlgren et al. Akaike’sinformationcriterion(AICc).TheAICcisa
2003),weexpectedthatastrongpositiveco-variation measure of relative model fit, proportional to the
123
Exploringassociations
likelihoodofthemodelandthenumberofparameters phloem-feedersinbothEurope(93.7%)andtheUSA
used to generate it. The best fitting model is the one (80%), with ambrosia beetles and seed-feeders being
with the lowest AICc. In a set of n models each lessfrequent(Fig. 1).Exoticspeciesweremoreoften
model i can be ranked using its difference in AICc ambrosia beetles than bark beetles compared with
score with the best-fitting model (DAICc = AICc– native proportions in both Europe and the USA
i i
AICc minimum). The difference in AICc values (Fig. 1). In Europe, native ambrosia beetle species
indicatestherelativesupportforthedifferentmodels. richness was highest in the central European coun-
A model is usually considered plausible if its DAICc tries (Fig. 2a), while species richness of native bark
is below 2 (Burnham and Anderson 2002). For each beetles was largest in the southern countries such as
model i we also calculated an Akaike’s weight (w), Italy and France (Fig. 2b). The European country
i
which is the probability that model i would be with the highest number of exotic ambrosia beetles
selected as the best fitting model if the data were was Italy followed by the Czech Republic (Fig. 2c).
collected again under identical circumstances (Burn- In the USA, native ambrosia beetle species were
hamandAnderson2002).Akaike’sweightshouldbe concentrated in the southern and eastern states
interpreted as a measure of model selection uncer- (Fig. 2d), while the states with the largest species
tainty. The multi-model inference analyses were richness of native bark beetles were located on the
performed using the ‘MuMIn’ package (Barton west coast (Fig. 2e). Species richness distribution of
2010) implemented in R version 2.12 (R Develop- exotic ambrosia beetles reflected that of native
ment Core Team 2011). ambrosia beetle species (Fig. 2f), while species
Although it is possible to use Akaike’s model richness hotspots of exotic bark beetles were distrib-
weights to evaluate the relative importance of vari- uted on both the east and the west coast (Fig. 2g).
ables, this approach has been recently criticized in
cases of collinearity between the predictors included Relationship between native and exotic
in model selection (Murray and Conner 2009). We Scolytinae species richness
thereforeusedhierarchicalpartitioning(HP)(Chevan
and Sutherland 1991) to rank the importance of each On the one hand, we found a positive relationship
variable most likely to affect variation in species between native and exotic ambrosia beetle species
richness. HP allows the joint consideration of all richness in both Europe (Fig. 3a) and the USA
possible models in a multiple regression attempting (Fig. 3b). The relationship was much stronger in the
to identify the most likely causal factors. This USAthaninEurope.Ontheotherhand,wefoundno
analysis splits the variation explained by each association between species richness of native and
explanatoryvariableintoa joint effect with the other exotic bark beetles in the USA (Fig. 3c). In Europe,
explanatory variables and an independent effect not thenumberofexoticbarkbeetlespecieswastoolow
shared with any other variable. HP was conducted to perform the analysis.
using the ‘Hier.Part’ package (version 0.5-1, Mac
Nally and Walsh 2004) implemented in R version Multi-model inference and hierarchical
2.12 (R Development Core Team 2011). We used a partitioning
normal error distribution and R2 as a measure of
goodness-of-fit. For both regions, model selection indicated that no
single best model was supported and that several
candidate models were plausible in explaining pat-
Results terns of Scolytinae species richness for both native
and exotic species. There was therefore considerable
General results model selection uncertainty, as reflected by the low
Akaike’s weights (w). However, we found contrast-
i
Thetotal number ofScolytinae species analyzed was ing results concerning the relative importance of the
241forthe20Europeancountries(221nativeand20 variables included in the sets of plausible models for
exotic) and 537 for the 48 US states (469 native and each species groups, i.e. native versus exotic species
58 exotic). The vast majority of native species were and ambrosia versus bark beetles.
123
L.Marinietal.
Fig.2 Nativeandexotic
speciesrichnessfor
ambrosiabeetlesandbark
beetlesinEurope(20
countries)andthe48
contiguousUSstates.The
numberofexoticbark
beetlespeciesinEurope
wastoolowandthereforeis
notshown
Fig.3 Relationship between exotic species richness and Afittedlineindicatesasignificantlinearregression(P\0.05).
native species richness for a ambrosia beetles in Europe, The number of exotic bark beetle species in Europe was not
bambrosiabeetlesintheUSA,andcbarkbeetlesintheUSA. sufficienttoperformtheanalysis
Europe heterogeneity (Table 2a). The proportion of
explained variation by these models was relatively
For native European ambrosia beetles there was low (22–31%). For native European bark beetles, we
support for four models including mostly forest found three plausible models (i.e. DAICc\2) which
123
Exploringassociations
includedmostlyforestheterogeneityandtemperature USA
heterogeneity variables. These models explained a
larger proportion of the total variation in species For native ambrosia beetles in the USA, we found
richness (60–69%) than did the models for native two plausible models that included mainly tempera-
ambrosia beetles (Table 2b). Concerning the exotic ture,precipitation,precipitationvariability,andforest
ambrosia beetles, we found three plausible models area explaining 64 and 67% of the total variation in
which always supported the hypothesis that exotic species richness, respectively (Table 3a). For native
species richness was positively associated with the bark beetles, we found six plausible models mostly
value of imports and secondarily to temperature including minimum temperature and standard devia-
heterogeneity (Table 2c). The low number of exotic tion of mean temperature, forest heterogeneity, and
bark beetle species in Europe did not allow for forest area explaining more than 80% of the total
analysis. variation in species richness (Table 3b). For exotic
Generally, results of the HP analyses reflected ambrosia beetles, we found three plausible models
those ascertained by model selection, but revealed mainly including positive effects of minimum tem-
slightly different results concerning the relative perature, value of imports and precipitation explain-
importance of some variables. The variable ranking ing 71–73% of the total variation (Table 3c). For
indicated that forest heterogeneity was always the exotic bark beetles, we found four plausible models,
best predictor with the highest independent effect on which included mostly the value of imports explain-
the number of native species for both ambrosia ing alone c. 40% of the total variation in species
beetles and bark beetles (Fig. 4a, b). For exotic richness (Table 3d).
ambrosia beetle species the value of imports was the The results of the HP analyses indicated that
only predictor with a large independent effect native ambrosia beetles were mostly associated
(Fig. 4c). with climatic variables (minimum temperature and
Table2 Plausiblecandidatemodels(within2DAICoftopmodel)explainingScolytinaespeciesrichnessinEurope,separatelyfor
(a)nativeambrosiabeetles,(b)nativebarkbeetles,and(c)exoticambrosiabeetles
Intercepta FOREST- AREA TEMP PRECIP PRECIP- TEMP- IMPORTb k R2 Adj. AICc DAICc w
i
H H H R2
(a)Nativeambrosiabeetles(EU)
11.25 0.968 – – – – – n.a. 3 0.22 0.18 86.87 0.00 0.18
11.25 1.027 – – – 0.607 – n.a. 4 0.31 0.23 87.66 0.79 0.12
11.25 0.921 0.468 – – – – n.a. 4 0.27 0.19 88.66 1.79 0.07
11.25 0.898 – – – – 0.437 n.a. 4 0.27 0.18 88.86 1.99 0.06
(b)Nativebarkbeetles(EU)
89.7 14.90 – – -7.949 – 11.83 n.a. 5 0.69 0.63 174.12 0.00 0.25
89.7 15.93 8.66 – – – – n.a. 4 0.61 0.56 175.10 0.98 0.15
89.7 15.45 – – – – 8.439 n.a. 4 0.60 0.55 175.51 1.40 0.12
(c)Exoticambrosiabeetles(EU)
3.250 – – – – –1.125 1.537 1.720 5 0.51 0.41 94.00 0.00 0.16
3.250 – – – – – 0.853 1.474 4 0.39 0.32 94.57 0.57 0.12
3.250 – – – – – – 1.384 3 0.28 0.25 94.64 0.64 0.11
ThenumberofexoticbarkbeetlesinEuropewasnotsufficienttoperformtheanalysis.Modelsarerankedaccordingtotheirsecond-
orderAkaike’sinformationcriterion(AICc).Parameterestimates,numberofparameters(k),adjustedR2andmodelweight(w)are
i
reported.Allexplanatoryvariableswerestandardized(mean=0,SD=1).VariablesnamesarepresentedinTable1
a Foreachresponsevariable,theplausiblecandidatemodelshavethesameinterceptvaluebecausetheexplanatoryvariableswere
standardized(mean=0,SD=1)
b n.a.Notapplicable.ThevariableIMPORTwasnotincludedintheanalysesfornativespecies
123
L.Marinietal.
Discussion
Our cross-continental study supports the view that
both climate and imports are strong predictors of
species richness of exotic Scolytinae, despite some
differences between regions and feeding guilds. In
Europe the only predictor of the number of estab-
lished ambrosia beetles was the value of imports
while in the USA, the feeding guild of an exotic
species appeared to modify the probability of its
establishment. Here, along with the strong influence
of imports, a warmer and wetter climate appeared
linked to an increase in the number of established
exotic ambrosia beetles, while climate was not
important for bark beetles. By contrast, measures of
environmentalheterogeneityplayednoimportantrole
inexplainingspecies richnessofexotic Scolytinaein
either region, while they were key drivers in
explaining patterns of native species richness. In
both regions, exotic Scolytinae were more often
ambrosiabeetlesthanbarkbeetlescomparedwiththe
relative proportions in the native species pool. There
are several life history traits of exotic ambrosia
beetles that can favor their arrival and establishment
in a new environment: (1) low host specificity and
thustheabilitytobreedinbothlivinganddeadwood
of many tree species, (2) sib-mating, i.e. brother-
sister mating prior to emergence from the host tree,
and (3) symbiotic trophic specialization with fungi
that obviates the need to overcome many host
Fig.4 The independent and shared contributions estimated defensesandreducescompetitionwithnativespecies
fromhierarchicalpartitioningofeachexplanatoryvariablefor
(Kirkendall1983).Alongwiththeselife-historytraits
the number of native and exotic Scolytinae in both Europe
(EU)a–candtheUSAd–g,presentedseparatelyforambrosia theincreasingshareofimportsfromAsiancountries,
and bark beetles. Variable ranking is conducted according to which have a rich ambrosia beetle fauna, may also
the size of the independent effect, i.e. variable importance helpexplainwhytheambrosiabeetlesrepresentsuch
declinesfromlefttoright.VariablenamesasgiveninTable2.
a large proportion of the exotic Scolytinae fauna in
AREA and IMPORT were log-transformed. The number of
exoticbarkbeetlesinEuropewasnotsufficienttoperformthe both Europe and the USA (Rabaglia et al. 2008;
analysis Kirkendall and Faccoli 2010).
We found a positive co-variation between native
and exotic ambrosia beetles species richness in both
precipitation) (Fig. 4d), while native bark beetles regions whilenosuch correlation was found forbark
were more associated with temperature heterogene- beetles (no data were available for Europe). The
ity and forest area (Fig. 4e). Amongst the exotic positive co-variation for ambrosia beetles was prob-
species, we found that annual precipitation was the ably related to the fact that both native and exotic
most important variable for exotic ambrosia beetles, species responded strongly to the same climatic
followed by temperature and the value of imports factors, while at the scale of our analysis potential
(Fig. 4f), while for exotic bark beetles the only interspecific interactions within this feeding guild
variable with a large independent effect was the could not be elucidated (Wood 1982). The lack of
value of imports (Fig. 4g). correlationamongstbarkbeetlesindicatesthatnative
123