|

Received: 11 September 2019    Accepted: 19 November 2019

DOI: 10.1111/1365-2745.13332

RESEARCH ARTICLE

Alien plants and flower visitors disrupt the seasonal dynamics

of mutualistic networks

Blanca Arroyo-Correa1,2  | Laura A. Burkle3  | Carine Emer4

1
Integrative Ecology Group, Estación
Biológica de Doñana, CSIC, Sevilla, Spain Abstract
2
School of Biosciences, Autonomous 1. Alien species can drastically disrupt ecological processes such as those involving
University of Barcelona, Barcelona, Spain
plant–pollinator interactions, performing central roles that may affect the struc-
3
Department of Ecology, Montana State
University, Bozeman, MT, USA
ture of native pollination networks. However, most studies to date have focused
4
Conservation Biology Lab, Department on a single trophic level of alien species, evaluating either the impacts of an alien
of Ecology, São Paulo State University, Rio pollinator or an alien plant species, and have neglected their joint effects on the
Claro, SP, Brazil
seasonal dynamics of mutualistic networks.
Correspondence 2. Here, we aim to fill this gap by investigating how alien plant and flower visitor (here
Blanca Arroyo-Correa
Email: [email protected] used as a proxy for pollinator) species structure temporal networks, and how these
species affect the beta-diversity of interactions across a flowering season. Our study
Funding information
Coordination for the Improvement of Higher system is located in the South Island of New Zealand, where 68% of the sampled
Education Personnel (CAPES, Brazil) plant species and 14% of the flower visitor species that interact with them are alien.
Handling Editor: Jane Catford 3. Alien flower visitor species exhibited higher interaction degree, specialization and
strength than their native counterparts, while alien plant species showed the op-
posite pattern. We found that invader complexes (in which alien species inter-
act significantly more with each other than with native species) were established
across the season, and interactions involving alien plant species were the main
connectors of the temporal networks. Both alien plant and flower visitor species
increased total interaction turnover through the flowering season by promoting
interaction rewiring in the case of alien plants and by increasing species turnover
in the case of alien flower visitors.
4. Synthesis. This study provides one of the first empirical reports of alien species shap-
ing the seasonal dynamics of plant–flower visitor networks. We demonstrate that the
presence of alien species may simultaneously lead to a homogenization of plant spe-
cies composition and increase the diversity of plant–flower visitor interactions through
a flowering season. Additionally, we highlight the importance of considering the role
of different trophic levels when analysing the impact of alien species in plant–flower
visitor communities. Overall, our findings suggest that management strategies should
pay particular attention to the timescale at which interactions with alien species
dissolve or form, and to the consequences and drivers of such seasonal dynamics.

KEYWORDS

beta-diversity, ecological networks, interaction turnover, invader complex, invasion ecology,

mutualism, plant–pollinator interactions

Journal of Ecology. 2019;00:1–12. wileyonlinelibrary.com/journal/jec© 2019 British Ecological Society     1 |

 |
2      
Journal of Ecology ARROYO-CORREA et al.

1 |  I NTRO D U C TI O N over extensive temporal spans can provide important information
about overall community patterns and the complexity of biological
In a rapidly changing world, where many species have been intro- systems (Bascompte, Jordano, Melian, & Olesen, 2003; Tylianakis &
duced outside of their native range due to human activities, under- Morris, 2017). However, their long-time frames fail to capture the
standing how alien species affect ecological processes is crucial if we dynamics of interaction build-up, and their structure can be signifi-
aim to reduce biodiversity loss (Gurevitch & Padilla, 2004; Traveset cantly different from temporally resolved networks (Rasmussen,
& Richardson, 2014). The impacts of alien species have long been Dupont, Mosbacher, Trøjelsgaard, & Olesen, 2013). By decompos-
studied in plant–pollinator interactions (e.g. Olesen, Eskildsen, & ing networks into more refined temporal frames, one can make
Venkatasamy, 2002; Padrón et al., 2009). Yet, their effects on the pol- inferences about how communities assemble over time (Olesen,
lination of native species are not consistent across species or regions, Bascompte, Elberling, & Jordano, 2008) and how non-native, alien
varying from negative (Bjerknes, Totland, Hegland, & Nielsen, 2007), species influence the establishment of interactions.
to neutral (Bartomeus, Vilà, & Santamaría, 2008) and even to posi- In this context, temporal variation in network structure over
tive effects (Moragues & Traveset, 2005). To date, most studies focus the course of a flowering season may be caused by phenological
on a single trophic level of the invasion, investigating either the ef- constraints, translating into changes in the identity of the inter-
fects of alien plants (e.g. Emer, Vaughan, Hiscock, & Memmott, 2015; acting partners (Petanidou, Kallimanis, Tzanopoulos, Sgardelis,
Kaiser-Bunbury, Valentin, Mougal, Matatiken, & Ghazoul, 2011) or & Pantis, 2008). Thus, the number of interactions a given species
alien pollinator species (e.g. Goulson, 2003; Trøjelsgaard, Heleno, & establishes within a community can vary within (Basilio, Medan,
Traveset, 2019). In addition, fluctuations in resource availability over Torretta, & Bartolini, 2006) and among years (Alarcón, Waser, &
a flowering season, due to, for example, phenological constraints, Ollerton, 2008), which in turn can influence their role as connec-
may influence how alien species interact among themselves and tors of temporal networks (Olesen, Stefanescu, & Traveset, 2011).
integrate into the native communities (Dupont, Padrón, Olesen, & Hence, the temporal effects of alien species on flower-visiting net-
Petanidou, 2009). Yet, we lack a clear understanding of whether, how works could be assessed by considering the extent to which each
and when the presence of both alien plant and alien animal species species is participating in interactions through a flowering season.
alter community-level patterns of mutualistic interactions. That said, beta-diversity indexes, which are usually applied to inves-
A network approach to animal–plant mutualisms has proven to tigate patterns of species composition and turnover across space,
be an efficient tool to evaluate the integration of alien species into can be useful to detect patterns of temporal changes in mutualistic
plant–pollinator communities (Memmott, 2009; Tylianakis & Morris, interactions (e.g. CaraDonna et al., 2017; Poisot, Canard, Mouillot,
2017). Alien species often exhibit high degree (i.e. number of inter- Mouquet, & Gravel, 2012; Simanonok & Burkle, 2014) in the pres-
actions performed) and high generalism (i.e. the capacity to interact ence of alien species as well. The effects of alien species on the
with a high diversity of partner species) in the networks in which temporal beta-diversity of interactions (i.e. interaction turnover)
they occur (Aizen, Morales, & Morales, 2008; Memmott & Waser, may be caused by (a) species turnover and/or (b) interaction rewir-
2002). Such characteristics can facilitate the establishment of alien ing. Interaction turnover caused by species turnover is defined as
species in communities because they can promptly interact with the changes in pairwise animal–plant interactions as animal or plant
local assemblage of partner species (Traveset & Richardson, 2014). species become present or absent over time. Meanwhile, interac-
Hence, they can function as central nodes in the networks they tion turnover caused by interaction rewiring is defined as the reas-
have invaded, performing many interactions with both native and sembling of pairwise interactions when both partners occur but do
alien partners and increasing the dependence of other species on not interact (Burkle, Myers, & Belote, 2016). Thus, a combination
them, that is, increasing their species strength (Stouffer, Cirtwill, & of network tools and beta-diversity indexes (Trøjelsgaard, Jordano,
Bascompte, 2014). Such concentration of interactions by a few alien Carstensen, & Olesen, 2015) is promising to investigate the role
species can potentially decrease mutualism strength among native of alien species in driving the temporal assembly of flower-visiting
co-occurring species (Aizen et al., 2008). Additionally, preferential networks.
interactions among alien species might create ‘invader complexes’, Here, we aimed to understand the role of alien plant and alien
defined as ‘groups of alien species interacting more with each other flower visitor species on the structure of invaded flower-visiting
than expected by chance’ (D’Antonio & Dudley, 1993). Invader com- networks throughout a flowering season. We collected data over
plexes can promote ‘invasional meltdowns’, in which alien species 2.5 months during a flowering season in the subalpine vegetation of
facilitate one another, thereby improving the conditions for their the South Island of New Zealand, where native and alien plant and
survival and spread (Simberloff & Von Holle, 1999). These alien–alien insect flower visitor species co-occur. Specifically, we investigated (a)
interactions can disrupt native plant–pollinator mutualisms by mod- the extent to which the network roles of alien plant and flower visi-
ifying the structure of the native community (Padrón et al., 2009). tor species (i.e. degree, specialization and species strength) differed
Yet, much of our understanding of the effects of alien species on from the roles performed by native species, and whether the pat-
the structure of flower-visiting networks is derived from data aggre- terns depended on the trophic level analysed; (b) the extent to which
gated across the entire flowering season (but see e.g. Dupont et al., interactions involving alien species represented temporal connec-
2009; Kaiser-Bunbury et al., 2011). Network studies pooling data tions between sequential time frames of the invaded flower-visiting
ARROYO-CORREA et al. Journal of Ecology |
      3

networks; (c) how alien plant and flower visitor species shaped the study sites was considered low. Therefore, each site was treated as
temporal beta-diversity of interactions; and (d) whether alien species an independent assemblage of species. Four consecutive surveys
formed temporal invader complexes. Due to their typically generalist per site were conducted from late December 2012 to early March
behaviour (Traveset & Richardson, 2014), we expected alien species 2013, therefore including the flowering season in our study area
to be involved in a higher number and more diverse interactions, ex- (Figure S1b). With each of these surveys, we built the correspond-
hibiting higher degree, lower specialization and higher strength than ing temporal network, totalling 12 temporal networks (four per site).
their native counterparts. Therefore, we predicted that alien species Hence, the four temporal networks constructed per site represent
would be active for longer periods during the season (Memmott & four subsections of a full-season network. This methodological ap-
Waser, 2002), playing important roles in connecting sequential time proach that accounts for the temporal component of mutualistic
frames of the flower-visiting networks. networks has been widely used in pollination network studies (e.g.
Further, the diversity of interactions performed by alien species Kaiser-Bunbury, Muff, Memmott, Müller, & Caflisch, 2010; Souza et
and their longer phenologies may shuffle interactions among native al., 2018).
species, potentially affecting seasonal beta-diversity patterns by The adequacy of splitting interaction data across the flowering
promoting interaction rewiring while homogenizing the community season into four time slices according to flowering phenology was
composition. Finally, because alien species in our study system were confirmed with linear discriminant analyses (LDA) using the ‘mass’ r

from the same geographic origin (Allen, Duncan, & Lee, 2006), they package (Venables & Ripley, 2002). We constructed a date-by-plant
potentially shared close evolutionary trajectories favouring morpho- species data matrix, in which the matrix entries indicated the abun-
logical trait matching between them (Anderson, Terblanche, & Ellis, dance of flowering species on a particular date of the flowering sea-
2010). Thus, we expected alien plant and alien flower visitor species son. In this matrix, each date was associated with one of the four
to interact more frequently with each other in comparison with the time slices previously established across the season. For each site,
number of interactions established between alien and native spe- we performed a LDA with time slice as the dependent variable and
cies, thereby forming temporal invader complexes throughout the abundance of each flowering plant species as independent variables.
flowering season. To analyse differences among the four time slices in composition
and abundance of flowering species, we used pairwise t test com-
parisons with the first and second discriminant functions’ scores as
2 | M ATE R I A L S A N D M E TH O DS dependent variables and the time slices as independent variables.
LDA results confirmed that flowering phenology was effective in
2.1 | Study site distributing our data into four distinct groups (i.e. time slices; Figure
S2; Table S1).
The study was carried out at Cass Mountain Research Field Station, Within each site, we established a 500 × 5 m plot, subdivided
located in the Cass River basin near the Craigieburn Range, in into 20 subplots (25 × 5 m each; Figure S1a). To avoid temporal and
the middle of the South Island of New Zealand (43°02′09.2″S, spatial pseudoreplication within and among sites, we sampled five
171°45′22.0″E, 580 m a.s.l.). The field station is located in a moist subplots per site per day in the field, randomizing the order of the
and temperate climate zone with subalpine vegetation that includes sites and subplots to be sampled in the morning and afternoon each
open fields composed of grasses and shrubs (Wardle, 1991). The day. We also avoided bad weather conditions that could influence
current biota of New Zealand has been influenced by human-driven flower visitor behaviour, such as rain and strong winds. Interaction
introductions of plant and animal species since European settle- data were collected by the same person slowly walking an estab-
ment in 1,840 (Allen et al., 2006). It has been suggested that plant lished transect in each subplot. All insects observed visiting flowers
introductions in New Zealand have averaged 11 species per year and touching their reproductive structures were collected, and the
since European settlement (Atkinson & Cameron, 1993), represent- interaction was recorded. The amount of time spent in each sub-
ing around 40% of the total plant community richness (Vitousek, plot varied depending on insect and floral abundance (and the time
D'Antonio, Loope, Rejmanek, & Westbrooks, 1997). Likewise, na- needed to process the collected insects). With this methodological
tive European bumblebees (Bombus spp.), the cosmopolitan hoverfly approach, we standardized the sampling effort per flower over a con-
Eristalix tenax, the invasive cabbage butterfly Pieris rapae, and several sistent area. These individual-level observations were then pooled
other bee species including Apis mellifera became established on the at the species-level to build flower-visiting networks. Therefore, the
South Island of New Zealand following a series of introductions in number of hours and days spent to complete a survey (i.e. to sample
the late nineteenth and early twentieth century (Donovan, 1980; all 20 subplots per site) varied according to flowering phenology, the
Goulson & Hanley, 2004). local assemblage of flower visitors and weather conditions (number
We surveyed plant–flower visitor interactions at three sites at of days in each survey: min nine days; max 18 days; Figure S1b).
the base of Cass Mountain range, c. 1.5 km apart. Due to the lim- Specimens of plants were collected for species confirmation
ited average flying capacity of the flower visitor species present at or further identification by local taxonomists when needed. All in-
the study sites and the montane characteristics of the local terrain, sects were collected with an entomological net, transferred to killing
the likelihood of movement of flower visitor individuals among the tubes containing acetyl acetate, frozen and pinned for identification.
 |
4      
Journal of Ecology ARROYO-CORREA et al.

Insects were identified by taxonomists at the National Museum of package (Kuznetsova, Brockhoff, & Christensen, 2016). Prior to
Cardiff, UK. Our sampling removed relatively few of the insects vi- model fitting, we log-transformed normalized degree and species
sually seen in the overall communities. Furthermore, lethal sampling strength to meet model assumptions. Individual models were fit-
of bee specimens in a similar temperate meadow system did not af- ted for each trophic level separately, with each of the three metrics
fect abundance, richness, evenness or functional group composition as the dependent variable tested against species ‘status’ (native vs.
even after sampling the same site every 2 weeks for 5 years (Gezon, alien). Time was nested within site as a random factor to account for
Wyman, Ascher, Inouye, & Irwin, 2015). Therefore, we assumed that differences in species composition among sites and over time.
the removal of insect individuals in our study sites did not affect the
temporal variation in interaction composition. However, because all
sites and subplots were treated equally, even if insect removal had 2.4 | Temporal network connectors
slight temporary effects on the behaviour of remaining insects, it
would be similar across sites and subplots. We aimed to identify which type of species interaction (i.e. native
plant species–native flower visitor species, native plant–alien flower
visitor, alien plant–alien flower visitor and alien plant–native flower
2.2 | Flower-visiting networks visitor) was predominant through the flowering season and therefore
was most important in connecting the temporal networks. To do so,
We built each network as an Amn adjacency matrix in which m repre- we built a binary Amn matrix for each site in which m corresponds to
sents the plant species (rows), and n the flower visitor insect species each of the temporal networks (1 to 4) and n to the plant–flower visi-
(columns), while the mn elements represent the visitation frequency tor interactions recorded through the flowering season in that site;
of insect i1…n in plant j1…m. We constructed temporal networks rep- cells received the value of 1 when an interaction n was recorded in
resenting the data recorded in each of the four surveys, therefore a given time m, and 0 otherwise. Then, we calculated betweenness
creating four matrices per site (N = 12 temporal networks) that rep- centrality (BC) for each plant–flower visitor interaction (Freeman,
resented the dynamics of plant–flower visitor interactions during the 1978; González, Dalsgaard, & Olesen, 2010) as a proxy for the im-
flowering season. portance of alien species associations in connecting the temporal
networks. Interactions with BC > 0 play a significant role as temporal
network connectors, therefore contributing to the cohesiveness of
2.3 | Species role the plant–flower visitor community through the flowering season.
The significance of differences in BC among interaction types was
To determine whether alien and native species played different roles tested with nonparametric pairwise comparisons using Dunn's tests
in structuring flower-visiting networks over the flowering season, (Dinno, 2017) following a Kruskal–Wallis test.
we calculated the following metrics for each species in each tem-
poral network: (a) normalized degree—the number of observed links
per species weighted by the total number of links in the network, a 2.5 | Contribution of alien species to temporal beta-
proxy for the diversity of interactions a given species can perform; diversity
(b) specialization level (d′)—which measures how strongly a species
deviates from an expected random choice of available interaction To investigate the contribution of alien species to interaction beta-
partners; a species that exclusively interacts with an abundant spe- diversity (i.e. turnover of interactions) over time, we estimated the
cies has a lower d′ value (i.e. less specialized) than a species that ex- turnover of plant–flower visitor interactions (βCC) across the flower-
clusively interacts with a rare species (Blüthgen, Menzel, & Blüthgen, ing season using binary temporal networks to calculate the Jaccard
2006); and (c) species strength—the sum of the dependencies of all similarity index among them. Because overall interaction turnover
interaction partners on that species, quantifying the importance (βCC) is driven by several ecological processes, we first partitioned it
of a species for the community with which it interacts (Barrat, as: βCC = βST + β0, where βST denotes interaction turnover due to spe-
Barthelemy, Pastor-Satorras, & Vespignani, 2004). Across species, cies turnover, that is changes in interactions as plant and/or flower
there were no significant relationships between the metrics, except visitor species become present or absent over time, and β0 denotes
for a positive correlation between species strength and normalized interaction turnover due to interaction rewiring, that is changes in
degree (r = .68; Table S2). We retained all metrics in our analyses, and interactions due to changes in interacting partners (e.g. Dupont et
we interpreted the results accordingly, given this positive correla- al., 2009). Then, we divided the total contribution of species turno-
tion. In doing so, we aimed to assess whether higher degree species ver to interaction turnover (βST ) into three components, following
also showed higher strength, probably as a result of being connected the approach developed by Novotny (2009): βST = βP + βA + βPA . That
to low degree species. Then, we investigated whether the role of is, the contribution of species turnover to interaction turnover (βST )
each trophic level (i.e. plant or flower visitor species) was related to can be driven by plant species turnover (βP), flower visitor species
their status as alien or native species by fitting generalized linear turnover (βA), and simultaneous turnover of both plant a flower visi-
mixed models (GLMMs) with Gaussian errors and using the ‘lmertest’ tor species (βPA). We calculated overall interaction turnover (βCC) and
ARROYO-CORREA et al. Journal of Ecology |
      5

its components for sequential pairwise combinations of interac- temporal networks (i.e. 12 contingency tables). Each contingency
tion networks using modified ‘betalink’ R Script (Poisot et al., 2012; table contained two descriptors (flower visitor and plant species)
Simanonok & Burkle, 2014). with two levels (native and alien species). Therefore, each of the
To further analyse the effects of alien species on the temporal four cells of the tables described one type of interaction: alien
dynamics of mutualistic networks, we assessed how alien plant and plant–alien flower visitor, alien plant–native flower visitor, native
flower visitor species contributed to the interaction beta-diversity plant–native flower visitor or native plant–alien flower visitor. For
over time. To do so, we built two simulated networks for each of each temporal network from each site, cells were filled with the fre-
the 12 temporal networks, identical to the observed data except for quency with which the corresponding interaction type was repre-
the removal of (a) alien plant species or (b) alien flower visitor spe- sented in the modules. The significance of the associations between
cies and the interactions they performed. Hence, we obtained 12 plant and flower visitor descriptors was tested with Fisher's exact
‘alien plant removal’ networks and 12 ‘alien flower visitor removal’ tests of independence.
networks. Given that species turnover and interaction rewiring are Network metrics (i.e. species-level metrics for species role,
expected to depend on species richness and/or network size (e.g. betweenness centrality for temporal connectors and modularity
Burkle et al., 2016; Kraft et al., 2011) independently of their status for invader complexes) were estimated using ‘bipartite’ r package
as alien or native, we compared the results from our alien-removal (Dormann, Gruber, & Fründ, 2008). All analyses were performed in R
simulations against the removal of random species using null models. version 3.4.1 (R Core Team, 2017).
Because our aim was to quantify whether the contribution of alien
species was different than expected by chance and not to predict
how the community would change after the removal of alien spe- 3 | R E S U LT S
cies, in this analysis we did not considered the potential interaction
rewiring among the remaining species in the networks (e.g. Kaiser- We recorded a total of 1,405 individuals of insect flower visitors
Bunbury et al., 2010). distributed among 100 species (Table S3) and participating in 412
Our null models involved the creation of new simulated net- plant–flower visitor interactions with 34 plant species (Table S4). Of
works in which we randomly removed the same number of species those, 334 interactions (81%) included at least one alien partner, ei-
as the number of species removed in the ‘alien removal’ networks, ther an alien plant or an alien flower visitor species. Alien plant spe-
considering each trophic level separately. The random removal of cies were the most represented group (23 spp, 68% of plant species).
species was iterated 1,000 times across each of the 12 observed Out of the nine alien plant families recorded, Asteraceae accounted
temporal networks. Therefore, we obtained 12,000 ‘random plant for almost 50% of the alien plant species. The set of native plants
removal’ networks and 12,000 ‘random flower visitor removal’ net- comprised species from more families, and Asteraceae was again the
works. Finally, we obtained the beta-diversity indices described most represented group (36% of all native plant species). Regarding
above for the alien-removal and the random-removal simulated flower visitors, alien species constituted a lower proportion of all
networks. To test the significance of the differences between ‘alien flower visitor species (14 spp, 14%). Approximately 30% of the alien
plant removal’ and ‘random plant removal’, and between ‘alien flower flower visitor species, including Apis mellifera and Bombus spp., were
visitor removal’ and ‘random flower visitor removal’ networks, we found within the Apidae family. The native flower visitor families
used generalized linear mixed models (GLMMs) with a gamma distri- with the greatest species richness were Syrphidae (21%), Tachinidae
bution. Each beta-diversity index was tested as a dependent variable (12%) and Colletidae (11%).
against the type of network (alien-removal or random-removal), and
pairwise temporal comparisons (i.e. temporal network 1 vs. temporal
network 2, etc.) were nested within site as a random factor to control 3.1 | The role of alien versus native species in
for potential spatial and temporal pseudoreplication. structuring flower-visiting networks

The role of alien and native species in the structure of the invaded
2.6 | Invader complexes plant–flower visitor interaction networks depended on the trophic
level considered. Alien plant species had lower normalized degree
To test for the existence of invader complexes, we first estimated (estimate ± SE = 0.39 ± 0.16, t = 2.39, p = .020; Figure 1a), speciali-
the modularity of each temporal network to further determine zation (estimate ± SE = 0.12 ± 0.04, t = 2.64, p = .010, Figure 1b) and
whether species interacted significantly more in the same module species strength (estimate ± SE = −0.41 ± 0.09, t = −4.68, p < .010,
depending on their native or alien status. Modularity was estimated Figure 1c) when compared to native plant species. On the other
by the ‘computeModules’ function, using the Beckett algorithm hand, alien flower visitor species had higher normalized degree (es-
(Beckett, 2016). Module composition stabilized after 50 iterations, timate ± SE = −0.16 ± 0.09, t = −1.92, p = .050; Figure 1a), speciali-
and therefore, this was the final network configuration used to as- zation (estimate ± SE = −0.17 ± 0.03, t = −5.01, p < .010; Figure 1b)
sign species to modules. We used the species composition in each and strength (estimate ± SE  =  −0.99  ±  0.22, t  =  −4.46, p < .010;
module to construct a two-way contingency table for each of the 12 Figure 1c) than the native assemblage of flower visitor species.
 |
6      
Journal of Ecology ARROYO-CORREA et al.

(a) (b) (c)

F I G U R E 1   The roles of native and alien plant and flower visitor species in structuring invaded flower-visiting networks (a, degree; b,
specialization and c, strength). Boxes indicate the 25th to 75th percentiles, the middle line is the median, and the maximum length of the
whiskers is 1.5 times the interquartile range. (*) Significant differences between alien and native species (p < .050)

TA B L E 1   Results from the betweenness centrality (BC) 3.3 | Contribution of alien species to temporal beta-
analyses for the different types of plant–flower visitor interactions diversity
composing the temporal networks

Number of We found significant differences in the interaction composition of

interactions flower-visiting networks across the season (Figure 2; Table S7). In
Interaction type with BC > 0 (%) BC (M ± SD) general, interaction turnover was high between each consecutive
Alien plant–alien flower visitor 26 0.029 ± 0.024 temporal network (βCC = 0.82 ± 0.05 [M ± SD]). The largest contributor
Alien plant–native flower visitor 34 0.036 ± 0.026 to interaction turnover was the turnover of species (67.27% ± 0.06).
Native plant–alien flower visitor 3 0.024 ± 0.017 Total species turnover was slightly lower during the middle (between
Native plant–native flower visitor 37 0.024 ± 0.024 time 2 and 3, βST  =  0.49  ±  0.07) than earlier (between time 1 and
2, βST = 0.59 ± 0.01) or later in the season (between time 3 and 4,
Note: Interactions with BC > 0 are significant temporal network
connectors. The proportion of interactions of the different types is based βST = 0.58 ± 0.06).
on the total number of interactions with BC > 0. Mean and standard Both alien plant and flower visitor species had a significant effect
deviation (SD) correspond to the values among the three study sites. on the temporal beta-diversity of interactions (Figure 3; Table S8).
Total interaction turnover (βCC) was lower when alien plant species
and alien flower visitor species were removed in comparison with
3.2 | Temporal network connectors the random removal of plant (alien removal = 0.84 ± 0.10; random
removal = 0.94 ± 0.06, p < .001) and the random removal of flower
Approximately one fourth of all plant–flower visitor interactions visitor (alien removal = 0.84 ± 0.06; random removal = 0.91 ± 0.04,
played a role in connecting the networks through time in our commu- p < .001) species. However, the difference in interaction turnover be-
nities (Table 1; Table S5). We found differences in betweenness cen- tween alien plant and alien flower visitor species removal was driven
trality (BC) between the pairwise combinations of alien and native by different processes. The removal of alien plant species reduced
plant–flower visitor interactions (Kruskal–Wallis chi-squared = 7.98, the contribution of interaction rewiring to interaction turnover
p < .050; Table S6). Interactions between native plants and native (β0  = 0.06 ± 0.07) compared to the removal of random plant species
flower visitors had lower BC compared with alien plant–alien flower (0.14 ± 0.12, p < .001). Meanwhile, the removal of alien flower visitor
visitor interactions (Dunn z = 1.39, p < .050) and alien plant–native species decreased the contribution of species turnover to interaction
flower visitor interactions (Dunn z = 2.63, p < .050). Overall, inter- turnover (βST = 0.60 ± 0.09) in comparison with the removal of ran-
actions with native plant species had significantly lower potential dom flower visitor species (0.66 ± 0.10, p = .015). Additionally, flower
to connect the temporal networks (lower BC), while interactions visitor species turnover (βA) was lower when alien flower visitor spe-
involving alien plant species, independently of the status of the cies were removed (alien flower visitor species removal = 0.35 ± 0.11,
flower visitor species, contributed the most to temporal connectiv- random flower visitor species removal = 0.44 ± 0.12, p < .001) and
ity (Table 1). higher when alien plant species were removed (alien plant species
ARROYO-CORREA et al. Journal of Ecology |
      7

(a) (b) (c)

1 1 1
2 4 2
4
2 4

3 3
3

F I G U R E 2   Temporal networks of plant–flower visitor interactions from the South Island of New Zealand. Graphs correspond to the
networks from (a) site 1, (b) site 2 and (c) site 3. Each node (circle) represents a unique pairwise interaction between a plant species and a
flower visitor species. Squares represent the four time slices in which interactions were sampled during the flowering season. Grey lines
connect the interactions to the time slice in which they were recorded. Therefore, nodes connected to the square number 1 correspond
to those plant–flower visitor interactions that occur earlier in the flowering season. In the other end of the gradient, nodes connected to
square number 4 are those interactions that appeared only at the end of the flowering season. Black nodes shared among squares mean that
the given interaction was present across time, that is, shared nodes are those interactions connecting the temporal networks, which were
persisting throughout the flowering season. Network graphs were built with ‘Network3D’ r package (Allaire et al., 2017)

F I G U R E 3   The distribution of values for species turnover (βST ), interaction rewiring (β0) and total interaction turnover (βCC) of alien-
removal and random-removal simulated plant–flower visitor networks. The results for the removal of plant species are shown in the top
panel and flower visitors in the bottom ones. Density was calculated using a kernel density estimate to visualize the continuous distribution
of the data
 |
8      
Journal of Ecology ARROYO-CORREA et al.

between alien species were dominant, forming invader complexes

that might have facilitated the establishment of other alien species.
Furthermore, interactions involving alien plant species were the
main connectors of the temporal networks as they persisted longer
throughout the season. In addition, simulated alien species removal
significantly reduced the turnover of interactions, a pattern driven
by distinct underlying mechanisms for alien flower visitor and alien
plant species. First, simulated alien plant species removal reduced
interaction rewiring, suggesting that alien plants, despite homogeniz-
ing plant species composition, may increase interaction rewiring by
continuously attracting new flower visitor species over the season.
Second, alien flower visitor species removal reduced species turnover
even though they were participating in invader complexes through
the season, suggesting that the identity of those alien flower visitors
involved in invader complexes can change over time.
Alien plant species were visited by a lower number of flower vis-

F I G U R E 4   The presence of invader complexes and the itor species and were less specialized in comparison with the native
preferential interactions between alien species in an invaded plant– plant species which, in turn, showed higher degree and higher spe-
flower visitor network from the South Island of New Zealand. Black cialization. Hence, even though alien plants in our study sites were
boxes represent modules and invader complexes are identified visited by fewer flower visitor species (i.e. lower degree), the fre-
when interactions within a given module are predominantly
quency of visits by different flower visitor species was more evenly
performed by a pair of alien species. Darker colours correspond to
distributed (i.e. lower d′ specialization). These findings partially
higher frequency of interactions. This graph is one of the temporal
networks sampled during a flowering season in our study sites in contradict our hypothesis that alien species would perform more
the South Island of New Zealand (fourth temporal network in site 3, generalist interactions but expand our understanding of the poten-
see Section 2 for details) tial mechanisms behind such generalization found in the literature
(e.g. Aizen et al., 2008; Bartomeus et al., 2008; Lopezaraiza-Mikel,
removal = 0.48 ± 0.20, random plant species removal = 0.33 ± 0.21, Hayes, Whalley, & Memmott, 2007). Furthermore, alien plants may
p  =  .028). In contrast, plant species turnover (βP)  did not change create ‘floral markets’ (sensu Chittka & Schürkens, 2001) by at-
when alien plant species were removed (alien plant species re- tracting pollinators for the whole plant community, including native
moval = 0.09 ± 0.11, random plant species removal = 0.16 ± 0.12, and other alien species. In our study sites in the South Island of
p  =  .062) or when alien flower visitor species were removed (alien New Zealand, species within the Asteraceae (e.g. Cirsium spp) and
flower visitor species removal = 0.16 ± 0.10, random flower visitor Fabaceae (e.g. Lotus pedunculatus) families formed dense patches
species removal = 0.11 ± 0.07, p = .063). of floral resources during the flowering season, likely functioning
as an attractor of flower visitors to the local assemblage of plants
(Lopezaraiza-Mikel et al., 2007). Interestingly though, as suggested
3.4 | Invader complexes in a previous study (Chrobock, Winiger, Fischer, & Kleunen, 2013),
the lower strength of the alien plant species also revealed that the
The evidence for the presence of invader complexes was strong in overall flower visitor community depended more on the native flora
eight temporal networks, weak in three networks, and we found no to obtain floral resources. These findings stand in contrast with
evidence in one network (Table S9). The probability of an alien plant some studies (Bartomeus et al., 2008; Maruyama et al., 2016) sug-
species interacting in the same module with an alien flower visitor gesting that flower visitors have a high dependency on alien plant
species was, on average, 11.95 times greater than the probability of species. This discrepancy in outcomes among studies may be re-
interacting in the same module with a native flower visitor species, lated to the use of a multispecies approach that considers all of the
supporting the invader complex hypothesis (Figure 4). alien species in the community (e.g. this study; Timóteo et al., 2018;
Vizentin-Bugoni et al., 2019), instead of focusing on a single or few
widespread and abundant alien species. The higher degree and
4 |  D I S CU S S I O N higher strength of the native plant species indicate that they are
providing food resources to a higher number of flower visitor spe-
Our results suggest that the effects of alien species in structuring the cies, both alien and natives. Still, the combination of higher richness
seasonal dynamics of plant–flower visitor networks largely depend on of flower visitors and uneven frequency of visitations makes native
the trophic level being considered. While alien plant species were less plant species more specialized than the alien ones, corroborating
specialized and performed fewer interactions than native plant spe- the finding of previous studies as well (Norfolk, Gilbert, & Eichhorn,
cies, alien flower visitors showed the opposite pattern. Interactions 2018; Waser, Chittka, Price, Williams, & Ollerton, 1996).
ARROYO-CORREA et al. Journal of Ecology |
      9

Alien flower visitor species participated in a higher number of with CaraDonna et al. (2017), which may strengthen the effects of
interactions (higher degree) than the native ones and still showed the turnover of species. Because it is more likely for two species
higher specialization and higher strength. This result suggests that to continue to co-occur across shorter timespans compared to lon-
the local flora may become more dependent on alien flower visitors ger ones, if we had collected data at a finer temporal resolution,
for pollination than on their sympatric native species. That is exactly we would expect interaction rewiring to contribute more strongly
the opposite pattern we found for alien versus native plants; that is, to the total beta-diversity of interactions. Moreover, the constant
flower visitors may remain more dependent on native flora. In our presence of alien species in our system likely changed the seasonal
communities, the assemblage of alien flower visitors included diet dynamics of the local communities and may explain the different
and habitat generalist species which formed distinct alien-based outcomes between studies.
functional groups such as the bees (Bombus spp. and Apis mellifera), As expected with the directional turnover of species (Anderson
the hoverflies Eristalix tenax and the butterflies Pieris rapae (Goulson, et al., 2011), and thus interactions, over a flowering season, the two
2003; Lewis, 1989). Additionally, alien pollinator species are often temporal networks in the middle of the season shared a higher num-
highly abundant in the areas they invade (Valido, Rodríguez- ber of interactions than those composing the networks at the start
Rodríguez, & Jordano, 2019). In our study sites, we can infer the and at the end of the season. Such a pattern is partially caused by
abundance of alien flower visitor species from the frequency of in- phenology given that the first and the last temporal networks lacked
teractions they established with flowering plants. Therefore, they many flowering plant species, which is not surprising since flow-
potentially act as important pollinators for the local flora, visiting a er-visiting networks are phenologically driven systems (Encinas-Viso,
range of species with distinct flower morphologies, blooming strat- Revilla, & Etienne, 2012; Rafferty & Ives, 2011). Temporal networks
egies and resource quality. A previous study (Trøjelsgaard et al., were connected by a relatively low number of interactions, and
2019) showed a similar pattern in five Galápagos islands, where alien most plant–flower visitor associations occurred in a short window
flower visitors had more interaction partners and higher strength of time. The betweenness centrality results revealed that alien spe-
(i.e. plant species were more dependent on those flower visitors) cies, especially plants, played an important role in connecting the
in comparison with their native counterparts. However, when the temporal networks by performing interactions that persisted longer
frequency of interactions was considered in our study, alien flower in the flowering season. These findings are also supported by the
visitors seem to prefer to visit a small subset of plant species (higher persistence of invader complexes across the season, which may facil-
specialization). itate the temporally prolonged presence of alien flower visitors in the
We found that alien plants and alien flower visitors interact more study area (Morales & Aizen, 2002; Olesen et al., 2002). As pheno-
with each other than expected by chance, supporting the presence logical overlap is one of the major factors influencing pollinator-me-
of invader complexes. Our findings are contrary to previous work diated interactions between plant species (McKinney & Goodell,
in two different oceanic islands whose data did not support the ex- 2011), the longer persistence of alien–alien interactions may have
istence of invader complexes (Olesen et al., 2002), but our results critical consequences on the direction of the impacts of alien spe-
are consistent with the findings of Morales and Aizen (2006) and cies on the pollination of native plant species. However, in order to
Traveset et al. (2013), which report the occurrence of such com- explore the overall ecological and evolutionary implications of such
plexes. The combination of generalist characteristics and the pres- pattern, it would be necessary to analyse the pollen transport fluxes
ence of invader complexes may have facilitated the spread of alien between native and alien plants (Alarcón, 2010), which can assist in
species in New Zealand (D’Antonio & Dudley, 1993). For example, investigating the facilitative and competitive interactions between
bumblebees and honeybees that were first introduced in New those species (Emer et al., 2015; Lopezaraiza-Mikel et al., 2007).
Zealand to enhance crop pollination (Donovan, 1980) are now widely Finally, the simulated removal of alien species reduced in-
established in both the South and North Island, possibly because of teraction turnover between temporal networks; yet, the drivers
the interactions they have been able to establish with both native of the temporal turnover of interactions were also trophic-level
and introduced plant species (Olesen et al., 2002). dependent. This approach does not include the mechanisms by
The temporal turnover of interactions across the season was which alien species integrate and affect native communities, such
extremely high, confirming the importance of temporal scale in as competitive exclusion and preferential attachment (Revilla &
the structure of mutualistic networks (see also Burkle & Alarcón, Křivan, 2018), and does not consider the potential for interaction
2011; Olesen et al., 2008; Petanidou et al., 2008). Changes in rewiring to occur between the native species after aliens have
species composition over the season were the main drivers of been removed. That said and considering a lack of pre-invasion
the seasonal turnover of plant–flower visitor interactions. Our data for this system, these simulations (and comparisons to the
findings contradict a previous study also carried out in subalpine null models) can be seen as a first step towards understanding how
plant–flower visitor communities in which the within-season in- alien species influence the plant–flower visitor interaction com-
teraction turnover was primarily driven by interaction rewiring position through a flowering season. Alien plant species signifi-
(CaraDonna et al., 2017). This discrepancy between studies may cantly increased interaction turnover due to interaction rewiring,
be due to differences in phenophases between the two systems suggesting that alien plants may be outcompeting native plants by
or to the coarser temporal resolution in our study in comparison attracting pollinators that would otherwise interact with the local
 |
10      
Journal of Ecology ARROYO-CORREA et al.

flora, as demonstrated in previous studies (Bjerknes et al., 2007; lead to a homogenization of species composition and increase the
Chittka & Schürkens, 2001). diversity of plant–flower visitor interactions through the flowering
Together with the betweenness analysis, our results demon- season. Such changes may affect the persistence of interactions
strated that alien plant species have a long persistence through performed exclusively by native species and may have conse-
the season – leading to a temporal homogenization of interacting quences for the maintenance of the ecosystem functioning.
species composition – but they simultaneously have an important
role increasing the relative occurrence of interaction rewiring over AC K N OW L E D G E M E N T S
time. Towards the end of the season, many native species have We thank Pedro Jordano, Ignasi Bartomeus and two anonymous
already passed their phenological optimum for reproduction (i.e. reviewers for their helpful comments on earlier drafts of the man-
blooming or flying activities), and the remaining species may start uscript. We thank Jane Memmott, Jason Tylianakis, Laura Young,
to interact with each other even if they are not each other's pref- Michele Lambert, Sarah Kruis, Jane Ladley and the University of
erential partners (Olesen et al., 2008; CaraDonna et al., 2017). For Canterbury for friendship, intellectual, logistical and field-assistance
example, Bombus spp. started to interact with native plants at the support. C.E. was funded by the Coordination for the Improvement
end of the season after the flowers of their alien plant partners had of Higher Education Personnel (CAPES, Brazil).
stopped blooming. Likewise, late-season alien plant species might
offer additional resources for those native species of flower visi- AU T H O R S ’ C O N T R I B U T I O N S
tors whose preferential partner is not flowering anymore (Shea & C.E. conceived the idea and collected the data. B.A-.C. and C.E.
Chesson, 2002). Moreover, the simulated removal of alien plants analysed the data and wrote the first draft of the manuscript.
also promoted an increase in the turnover of flower visitor species, B.A-.C., L.A.B. and C.E. contributed to subsequent drafts and gave
suggesting that alien plants may homogenize the temporal assem- final approval for publication.
blage of flower visitors either by ‘rescuing’ native flower visitor
species that have lost their preferential flower resource (Chittka & DATA AVA I L A B I L I T Y S TAT E M E N T
Schürkens, 2001) or by facilitating the persistence of alien flower Plant–flower visitor interaction data used in this study are depos-
visitor species in the system through the maintenance of invader ited in the Zenodo Digital Repository: https​://doi.org/10.5281/
complexes (D’Antonio & Dudley, 1993). Finally, total species turn- zenodo.3011651 (Arroyo-Correa, Burkle, & Emer, 2019).
over was lower in the networks in which alien flower visitor species
were removed, a pattern driven by the decrease in flower visitor ORCID
species turnover, compared to random species removal simula- Blanca Arroyo-Correa  https://orcid.org/0000-0002-9402-3013
tions. That is, although alien flower visitor species participated in Laura A. Burkle  https://orcid.org/0000-0002-8413-1627
invader complexes over the entire flowering season, the identity Carine Emer  https://orcid.org/0000-0002-1258-2816
of those flower visitor species can change over time and therefore
they may increase the interaction turnover due to species turnover. REFERENCES
Aizen, M., Morales, C., & Morales, J. (2008). Invasive mutualists
erode native pollination webs. PLoS Biology, 6(2), e31. https​://doi.
org/10.1371/journ​al.pbio.0060031
5 |  CO N C LU S I O N S
Alarcón, R. (2010). Congruence between visitation and pollen-transport
networks in a California plant–pollinator community. Oikos, 119(1),
This study provides one of the first empirical reports showing 35–44. https​://doi.org/10.1111/j.1600-0706.2009.17694.x
that alien species have a crucial effect on the seasonal dynam- Alarcón, R., Waser, N., & Ollerton, J. (2008). Year-to-year variation in the
topology of a plant-pollinator interaction network. Oikos, 117(12),
ics of plant–flower visitor networks. We demonstrated that the
1796–1807. https​://doi.org/10.1111/j.0030-1299.2008.16987.x
presence of alien species facilitates invader complexes and pro- Allaire, J. J., Ellis, P., Gandrud, C., Kuo, K., Lewis, B. W., Owen, J., …
motes changes in the organization of plant–flower visitor inter- Yetman, C. J. (2017). networkD3: D3 JavaScript network graphs from
actions over the flowering season. Additionally, we highlighted R. R Package version 0.4.
Allen, R. B., Duncan, R. P., & Lee, W. G. (2006). Updated perspective on
the importance of considering the role of different trophic levels
biological invasions in New Zealand. In R. B. Allen & W. G. Lee (Eds.),
when analysing the impact of alien species in plant–flower visi- Biological invasions in New Zealand (pp. 435–451). Berlin, Germany:
tor communities. Overall, our findings suggest that management Springer.
strategies should pay particular attention to the timescale at Anderson, B., Terblanche, J., & Ellis, A. (2010). Predictable patterns of trait
which interactions with alien species dissolve or form, and to the mismatches between interacting plants and insects. BMC Evolutionary
Biology, 10(1), 204. https​://doi.org/10.1186/1471-2148-10-204
consequences and drivers of such seasonal dynamics. In the light
Anderson, M. J., Crist, T. O., Chase, J. M., Vellend, M., Inouye, B. D., Freestone,
of our results, conservation plans should consider the potential A. L., … Swenson, N. G. (2011). Navigating the multiple meanings of
consequences of aggregating interaction data with an underly- β diversity: A roadmap for the practicing ecologist. Ecology Letters,
ing temporal structure when it comes to assessing the impacts of 14(1), 19–28. https​://doi.org/10.1111/j.1461-0248.2010.01552.x
Arroyo-Correa, B., Burkle, L. A., & Emer, C. (2019). Alien plant and
alien species in the community structure. In our study system, the
pollinator species configure the seasonal dynamics of mutualistic
transition to alien-dominated communities could simultaneously
ARROYO-CORREA et al. Journal of Ecology |
      11

networks – Complete dataset. Zenodo Digital Repository, https​:// Freeman, L. C. (1978). Centrality in social networks conceptual clarifica-
doi.org/10.5281/zenodo.3011651 tion. Social Networks, 1(3), 215–239. https​://doi.org/10.1016/0378-
Atkinson, I., & Cameron, E. (1993). Human influence on the terrestrial biota 8733(78)90021-7
and biotic communities of New Zealand. Trends in Ecology & Evolution, Gezon, Z. J., Wyman, E. S., Ascher, J. S., Inouye, D. W., & Irwin, R. E.
8(12), 447–451. https​://doi.org/10.1016/0169-5347(93)90008-d (2015). The effect of repeated, lethal sampling on wild bee abun-
Barrat, A., Barthelemy, M., Pastor-Satorras, R., & Vespignani, A. (2004). dance and diversity. Methods in Ecology and Evolution, 6(9), 1044–
The architecture of complex weighted networks. Proceedings of the 1054. https​://doi.org/10.1111/2041-210X.12375​
National Academy of Sciences of the United States of America, 101(11), González, A. M. M., Dalsgaard, B., & Olesen, J. M. (2010). Centrality
3747–3752. https​://doi.org/10.1073/pnas.04000​87101​ measures and the importance of generalist species in pollination net-
Bartomeus, I., Vilà, M., & Santamaría, L. (2008). Contrasting effects of works. Ecological Complexity, 7(1), 36–43. https​://doi.org/10.1016/j.
invasive plants in plant–pollinator networks. Oecologia, 155(4), 761– ecocom.2009.03.008
770. https​://doi.org/10.1007/s00442-007-0946-1 Goulson, D. (2003). Effects of introduced bees on native ecosystems.
Bascompte, J., Jordano, P., Melian, C., & Olesen, J. (2003). The nested Annual Review of Ecology, Evolution, and Systematics, 34(1), 1–26.
assembly of plant-animal mutualistic networks. Proceedings of the https​://doi.org/10.1146/annur​ev.ecols​ys.34.011802.132355
National Academy of Sciences of the United States of America, 100(16), Goulson, D., & Hanley, M. E. (2004). Distribution and forage use of exotic
9383–9387. https​://doi.org/10.1073/pnas.16335​76100​ bumblebees in South Island, New Zealand. New Zealand Journal of
Basilio, A., Medan, D., Torretta, J., & Bartoloni, N. (2006). A year-long Ecology, 225–232.
plant-pollinator network. Austral Ecology, 31(8), 975–983. https​://doi. Gurevitch, J., & Padilla, D. K. (2004). Are invasive species a major cause
org/10.1111/j.1442-9993.2006.01666.x of extinctions? Trends in Ecology & Evolution, 19(9), 470–474. https​://
Beckett, S. (2016). Improved community detection in weighted bipar- doi.org/10.1016/j.tree.2004.07.005
tite networks. Royal Society Open Science, 3(1), 140536. https​://doi. Kaiser-Bunbury, C. N., Muff, S., Memmott, J., Müller, C. B., & Caflisch,
org/10.1098/rsos.140536 A. (2010). The robustness of pollination networks to the loss of
Bjerknes, A., Totland, Ø., Hegland, S., & Nielsen, A. (2007). Do alien plant species and interactions: A quantitative approach incorporating
invasions really affect pollination success in native plant species? pollinator behaviour. Ecology Letters, 13(4), 442–452. https​://doi.
Biological Conservation, 138(1–2), 1–12. https​://doi.org/10.1016/ org/10.1111/j.1461-0248.2009.01437.x
j.biocon.2007.04.015 Kaiser-Bunbury, C. N., Valentin, T., Mougal, J., Matatiken, D., &
Blüthgen, N., Menzel, F., & Blüthgen, N. (2006). Measuring specializa- Ghazoul, J. (2011). The tolerance of island plant–pollinator
tion in species interaction networks. BMC Ecology, 6(1), 9. https​://doi. networks to alien plants. Journal of Ecology, 99(1), 202–213.
org/10.1186/1472-6785-6-9 https​://doi.org/10.1111/j.1365-2745.2010.01732.x
Burkle, L., & Alarcón, R. (2011). The future of plant-pollinator diversity: Kraft, N. J., Comita, L. S., Chase, J. M., Sanders, N. J., Swenson, N. G.,
Understanding interaction networks across time, space, and global Crist, T. O., … Cornell, H. V. (2011). Disentangling the drivers of β di-
change. American Journal of Botany, 98(3), 528–538. https​://doi. versity along latitudinal and elevational gradients. Science, 333(6050),
org/10.3732/ajb.1000391 1755–1758. https​://doi.org/10.1126/scien​ce.1208584
Burkle, L., Myers, J., & Belote, R. (2016). The beta-diversity of spe- Kuznetsova, A., Brockhoff, P. B., & Christensen, R. H. B. (2016). lmerTest:
cies interactions: Untangling the drivers of geographic variation in Tests in linear mixed effects models. R package version 2.0-33. https​
plant-pollinator diversity and function across scales. American Journal ://doi.org/10.18637/​jss.v082.i13
of Botany, 103(1), 118–128. https​://doi.org/10.3732/ajb.1500079 Lewis, A. C. (1989). Flower visit consistency in Pieris rapae, the cab-
CaraDonna, P., Petry, W., Brennan, R., Cunningham, J., Bronstein, J., bage butterfly. The Journal of Animal Ecology, 1–13. https​://doi.org/
Waser, N., & Sanders, N. (2017). Interaction rewiring and the rapid 10.2307/4982
turnover of plant-pollinator networks. Ecology Letters, 20(3), 385– Lopezaraiza-Mikel, M. E., Hayes, R. B., Whalley, M. R., & Memmott, J.
394. https​://doi.org/10.1111/ele.12740​ (2007). The impact of an alien plant on a native plant–pollinator net-
Chittka, L., & Schürkens, S. (2001). Successful invasion of a floral market. work: An experimental approach. Ecology Letters, 10(7), 539–550.
Nature, 411(6838), 653–653. https​://doi.org/10.1038/35079676 https​://doi.org/10.1111/j.1461-0248.2007.01055.x
Chrobock, T., Winiger, P., Fischer, M., & Kleunen, M. V. (2013). The cob- Maruyama, P. K., Vizentin-Bugoni, J., Sonne, J., Martín González, A. M.,
blers stick to their lasts: Pollinators prefer native over alien plant Schleuning, M., Araujo, A. C., … Dalsgaard, B. O. (2016). The inte-
species in a multi-species experiment. Biological Invasions, 15(11), gration of alien plants in mutualistic plant–hummingbird networks
2577–2588. https​://doi.org/10.1007/s10530-013-0474-3 across the Americas: The importance of species traits and insularity.
D’Antonio, C. M., & Dudley, T. L. (1993). Alien species: The insidious invasion Diversity and Distributions, 22(6), 672–681. https​://doi.org/10.1111/
of ecosystems by plants and animals from around the world has become ddi.12434​
a major environmental problem. Pacific Discovery, 1993(Summer), 9–11. McKinney, A. M., & Goodell, K. (2011). Plant–pollinator interactions be-
Dinno, A. (2017). dunn.test: Dunn's test of multiple comparisons using tween an invasive and native plant vary between sites with different
rank sums. R package version 1.3.5. flowering phenology. Plant Ecology, 212(6), 1025–1035. https​://doi.
Donovan, B. J. (1980). Interactions between native and introduced bees org/10.1007/s11258-010-9882-y
in New Zealand. New Zealand Journal of Ecology, 3, 104–116. Memmott, J. (2009). Food webs: A ladder for picking strawberries or
Dormann, C. F., Gruber, B., & Fründ, J. (2008). Introducing the bipartite a practical tool for practical problems? Philosophical Transactions
package: Analysing ecological networks. R News, 8, 8–11. of the Royal Society B: Biological Sciences, 364(1524), 1693–1699.
Dupont, Y. L., Padrón, B., Olesen, J. M., & Petanidou, T. (2009). Spatio- https​://doi.org/10.1098/rstb.2008.0255
temporal variation in the structure of pollination networks. Oikos, 118(8), Memmott, J., & Waser, N. M. (2002). Integration of alien plants into a
1261–1269. https​://doi.org/10.1111/j.1600-0706.2009.17594.x native flower–pollinator visitation web. Proceedings of the Royal
Emer, C., Vaughan, I. P., Hiscock, S., & Memmott, J. (2015). The impact of the in- Society of London. Series B: Biological Sciences, 269(1508), 2395–2399.
vasive alien plant, Impatiens glandulifera, on pollen transfer networks. PLoS https​://doi.org/10.1098/rspb.2002.2174
ONE, 10(12), e0143532. https​://doi.org/10.1371/journ​al.pone.0143532 Moragues, E., & Traveset, A. (2005). Effect of Carpobrotus spp. on the
Encinas-Viso, F., Revilla, T. A., & Etienne, R. S. (2012). Phenology drives pollination success of native plant species of the Balearic Islands.
mutualistic network structure and diversity. Ecology Letters, 15(3), Biological Conservation, 122(4), 611–619. https​://doi.org/10.1016/
198–208. https​://doi.org/10.1111/j.1461-0248.2011.01726.x j.biocon.2004.09.015
 |
12      
Journal of Ecology ARROYO-CORREA et al.

Morales, C. L., & Aizen, M. A. (2002). Does invasion of exotic plants when resources are scarce. Journal of Ecology, 106(6), 2409–2420.
promote invasion of exotic flower visitors? A case study from the https​://doi.org/10.1111/1365-2745.12978​
temperate forests of the southern Andes. Biological Invasions, 4(1–2), Stouffer, D. B., Cirtwill, A. R., & Bascompte, J. (2014). How exotic plants
87–100. https​://doi.org/10.1023/A:10205​13012689 integrate into pollination networks. Journal of Ecology, 102(6), 1442–
Morales, C. L., & Aizen, M. A. (2006). Invasive mutualisms and the struc- 1450. https​://doi.org/10.1111/1365-2745.12310​
ture of plant–pollinator interactions in the temperate forests of Timóteo, S., O’Connor, C. J., López-Núñez, F. A., Costa, J. M., Gouveia,
north-west Patagonia, Argentina. Journal of Ecology, 94(1), 171–180. A. C., & Heleno, R. H. (2018). Pollination networks from natural and
https​://doi.org/10.1111/j.1365-2745.2005.01069.x anthropogenic-novel communities show high structural similarity.
Norfolk, O., Gilbert, F., & Eichhorn, M. P. (2018). Alien honeybees in- Oecologia, 188(4), 1155–1165. https​://doi.org/10.1007/s00442-
crease pollination risks for range-restricted plants. Diversity and 018-4281-5
Distributions, 24(5), 705–713. https​://doi.org/10.1111/ddi.12715​ Traveset, A., Heleno, R., Chamorro, S., Vargas, P., McMullen, C. K., Castro-
Novotny, V. (2009). Beta diversity of plant-insect food webs in tropical Urgal, R., … Olesen, J. M. (2013). Invaders of pollination networks in
forests: A conceptual framework. Insect Conservation and Diversity, the Galápagos Islands: Emergence of novel communities. Proceedings
2(1), 5. https​://doi.org/10.1111/j.1752-4598.2008.00035.x of the Royal Society B: Biological Sciences, 280(1758), 20123040.
Olesen, J. M., Bascompte, J., Elberling, H., & Jordano, P. (2008). Temporal https​://doi.org/10.1098/rspb.2012.3040
dynamics in a pollination network. Ecology, 89(6), 1573–1582. Traveset, A., & Richardson, D. M. (2014). Mutualistic interactions
https​://doi.org/10.1890/07-0451.1 and biological invasions. Annual Review of Ecology, Evolution, and
Olesen, J. M., Eskildsen, L. I., & Venkatasamy, S. (2002). Invasion of pol- Systematics, 45, 89–113. https​://doi.org/10.1146/annur​ev-ecols​
lination networks on oceanic islands: Importance of invader com- ys-120213-091857
plexes and endemic super generalists. Diversity and Distributions, Trøjelsgaard, K., Heleno, R., & Traveset, A. (2019). Native and alien
8(3), 181–192. https​://doi.org/10.1046/j.1472-4642.2002.00148.x flower visitors differ in partner fidelity and network integration.
Olesen, J. M., Stefanescu, C., & Traveset, A. (2011). Strong, long-term Ecology Letters, 22(8), 1264–1273. https​://doi.org/10.1111/ele.13287​
temporal dynamics of an ecological network. PLoS ONE, 6(11), Trøjelsgaard, K., Jordano, P., Carstensen, D. W., & Olesen, J. M. (2015).
e26455. https​://doi.org/10.1371/journ​al.pone.0026455 Geographical variation in mutualistic networks: Similarity, turnover
Padrón, B., Traveset, A., Biedenweg, T., Díaz, D., Nogales, M., & Olesen, and partner fidelity. Proceedings of the Royal Society B: Biological
J. M. (2009). Impact of alien plant invaders on pollination networks Sciences, 282(1802), 20142925. https​://doi.org/10.1098/rspb.2014.
in two archipelagos. PLoS ONE, 4(7), e6275. https​://doi.org/10.1371/ 2925
journ​al.pone.0006275 Tylianakis, J. M., & Morris, R. J. (2017). Ecological networks across en-
Petanidou, T., Kallimanis, A. S., Tzanopoulos, J., Sgardelis, S. P., & Pantis, vironmental gradients. Annual Review of Ecology, Evolution, and
J. D. (2008). Long-term observation of a pollination network: Systematics, 48, 25–48. https​://doi.org/10.1146/annur​ev-ecols​
Fluctuation in species and interactions, relative invariance of network ys-110316-022821
structure and implications for estimates of specialization. Ecology Valido, A., Rodríguez-Rodríguez, M. C., & Jordano, P. (2019). Honeybees
Letters, 11(6), 564–575. https​://doi.org/10.1111/j.1461-0248.2008. disrupt the structure and functionality of plant-pollinator net-
01170.x works. Scientific Reports, 9(1), 4711. https​://doi.org/10.1038/
Poisot, T., Canard, E., Mouillot, D., Mouquet, N., & Gravel, D. (2012). The s41598-019-41271-5
dissimilarity of species interaction networks. Ecology Letters, 15(12), Venables, W. N., & Ripley, B. D. (2002). Modern applied statistics with S
1353–1361. https​://doi.org/10.1111/ele.12002​ (4th ed.). New York, NY: Springer.
R Core Team. (2017). R: A language and environment for statistical com- Vitousek, P. M., D'Antonio, C. M., Loope, L. L., Rejmanek, M., & Westbrooks,
puting. Vienna, Austria: R Foundation for Statistical Computing. R. (1997). Introduced species: A significant component of hu-
Retrieved from https​://www.R-proje​c t.org/ man-caused global change. New Zealand Journal of Ecology, 21(1), 1–16.
Rafferty, N. E., & Ives, A. R. (2011). Effects of experimental shifts in flow- Vizentin-Bugoni, J., Tarwater, C. E., Foster, J. T., Drake, D. R., Gleditsch, J.
ering phenology on plant–pollinator interactions. Ecology Letters, M., Hruska, A. M., … Sperry, J. H. (2019). Structure, spatial dynamics,
14(1), 69–74. https​://doi.org/10.1111/j.1461-0248.2010.01557.x and stability of novel seed dispersal mutualistic networks in Hawaii.
Rasmussen, C., Dupont, Y. L., Mosbacher, J. B., Trøjelsgaard, K., & Olesen, Science, 364(6435), 78–82. https​://doi.org/10.1126/scien​ce.aau8751
J. M. (2013). Strong impact of temporal resolution on the structure Wardle, P. (1991). Vegetation of New Zealand. Cambridge, New York:
of an ecological network. PLoS ONE, 8(12), e81694. https​://doi. Cambridge University Press.
org/10.1371/journ​al.pone.0081694 Waser, N. M., Chittka, L., Price, M. V., Williams, N. M., & Ollerton, J.
Revilla, T. A., & Křivan, V. (2018). Competition, trait–mediated facilitation, (1996). Generalization in pollination systems, and why it matters.
and the structure of plant–pollinator communities. Journal of Theoretical Ecology, 77(4), 1043–1060. https​://doi.org/10.2307/2265575
Biology, 440, 42–57. https​://doi.org/10.1016/j.jtbi.2017.12.019
Shea, K., & Chesson, P. (2002). Community ecology theory as a frame-
work for biological invasions. Trends in Ecology & Evolution, 17(4), S U P P O R T I N G I N FO R M AT I O N
170–176. https​://doi.org/10.1016/S0169-5347(02)02495-3 Additional supporting information may be found online in the
Simanonok, M. P., & Burkle, L. A. (2014). Partitioning interaction turn-
Supporting Information section.
over among alpine pollination networks: Spatial, temporal, and envi-
ronmental patterns. Ecosphere, 5(11), 1–17. https​://doi.org/10.1890/
ES14-00323.1
Simberloff, D., & Von Holle, B. (1999). Positive interactions of nonindig- How to cite this article: Arroyo-Correa B, Burkle LA, Emer C.
enous species: Invasional meltdown? Biological Invasions, 1(1), 21–32. Alien plants and flower visitors disrupt the seasonal
https​://doi.org/10.1023/A:10100​86329619
dynamics of mutualistic networks. J Ecol. 2019;00:1–12.
Souza, C. S., Maruyama, P. K., Aoki, C., Sigrist, M. R., Raizer, J., Gross, C.
L., & de Araujo, A. C. (2018). Temporal variation in plant–pollinator https​://doi.org/10.1111/1365-2745.13332​
networks from seasonal tropical environments: Higher specialization