university of zurich · predictions of increasing specialization of pollination systems during...
TRANSCRIPT
University of ZurichZurich Open Repository and Archive
Winterthurerstr. 190
CH-8057 Zurich
http://www.zora.uzh.ch
Year: 2010
Plant-pollinator network assembly along the chronosequence of aglacier foreland
Albrecht, M; Riesen, M; Schmid, B
Albrecht, M; Riesen, M; Schmid, B (2010). Plant-pollinator network assembly along the chronosequence of aglacier foreland. Oikos, 119(10):1610-1624.Postprint available at:http://www.zora.uzh.ch
Posted at the Zurich Open Repository and Archive, University of Zurich.http://www.zora.uzh.ch
Originally published at:Oikos 2010, 119(10):1610-1624.
Albrecht, M; Riesen, M; Schmid, B (2010). Plant-pollinator network assembly along the chronosequence of aglacier foreland. Oikos, 119(10):1610-1624.Postprint available at:http://www.zora.uzh.ch
Posted at the Zurich Open Repository and Archive, University of Zurich.http://www.zora.uzh.ch
Originally published at:Oikos 2010, 119(10):1610-1624.
Plant-pollinator network assembly along the chronosequence of aglacier foreland
Abstract
Forelands of retreating glaciers offer an ideal model system to study community assembly processesduring primary succession. As plants colonize the area that is freed from ice they should beaccompanied by their pollinators to successfully reproduce and spread. However, little is known aboutthe assembly of plant-pollinator networks. We therefore used quantitative network analysis to study thestructure of plant-pollinator interactions at seven sites representing a chronosequence from 8 to 130years since deglaciation on the foreland of the Morteratsch glacier (southeastern Switzerland). At thesesites, individual visits of plant fl owers by insects were recorded throughout the fl owering season.Species richness of insect-pollinated plants and plant-pollinating insects, together with measures ofinteraction diversity and evenness, increased along the chronosequence at least for the fi rst 80 yearsafter deglaciation. Bees were the most frequent fl ower visitors at the two youngest sites, whereas fl iesdominated in mature communities. Pollinator generalization (the number of visited plant speciesweighted by interaction strength), but not plant generalization, strongly increased during the primarysuccession. This was reflected in a pronounced decline in network level specialization (measured asBlüthgen's H2') and interaction strength asymmetry during the fi rst 60 years along the chronosequence,while nestedness increased along the chronosequence. Thus, our findings contradict niche-theoreticalpredictions of increasing specialization of pollination systems during succession, but are in agreementwith expectations from optimal foraging theory, predicting an increase in pollinator generalization withhigher plant diversity but similar flower abundance, and an increase in diet breadth at higher pollinatordensities during primary succession.
Albrecht et al. Plant–pollinator network assembly 1 _______________________________________________________________________
1 2 3 4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Plant–pollinator network assembly along the chronosequence of a glacier foreland
Matthias Albrecht 1, Matthias Riesen 1 and Bernhard Schmid 1
1 Institute of Evolutionary Biology and Environmental Studies, University of Zurich,
Winterthurerstrasse 190, 8057 Zurich, Switzerland
Corresponding author:
Matthias Albrecht
E-mail: [email protected]
Fax: +41 44 635 57 11
Albrecht et al. Plant–pollinator network assembly 2 _______________________________________________________________________
Abstract 22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
Forelands of retreating glaciers offer an ideal model system to study community
assembly processes during primary succession. As plants colonize the area that is
freed from ice they should be accompanied by their pollinators to successfully
reproduce and spread. However, little is known about the assembly of plant–
pollinator networks. We therefore used quantitative network analysis to study the
structure of plant–pollinator interactions at seven sites representing a chronosequence
from 8 to 130 years since deglaciation on the foreland of the Morteratsch glacier (SE-
Switzerland). At these sites, individual visits of plant flowers by insects were
recorded throughout the flowering season. Species richness and Shannon diversity of
insect-pollinated plants and plant-pollinating insects, together with measures of
interaction diversity and evenness, increased along the chronosequence at least for the
first 80 years after deglaciation. Bees were the most frequent flower visitors at the
two youngest sites, whereas flies dominated in mature communities. Pollinator
generalization (the number of visited plant species weighted by interaction strength),
but not plant generalization, strongly increased during the primary succession. This
was reflected in a pronounced decline in network level specialization (measured as
Blüthgen’s H2’) and interaction strength asymmetry during the first 60 years along the
chronosequence, while nestedness increased along the chronosequence. Thus, our
findings contradict niche-theoretical predictions of increasing specialization of
pollination systems during succession, but are in agreement with expectations from
optimal foraging theory, predicting an increase in pollinator generalization with
higher plant diversity but similar flower abundance, and an increase in diet breadth at
higher pollinator densities during primary succession.
Albrecht et al. Plant–pollinator network assembly 3 _______________________________________________________________________
Introduction 46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
How communities assemble is a central issue in ecology (e.g. Clements 1916, Hubell
2001, Lortie et al. 2004, Sargent and Ackerly 2008). Primary successions, in
particular glacier forelands with a well-known chronology of glacier retreats, have a
long history as natural model systems for the study of plant community assembly
(reviewed in Matthews 1992; Walker and del Moral 2003). The chronosequence
across a glacier foreland can be interpreted, with the required caution, as a space for
time substitution of primary succession (Foster and Tilman 2000, Kaufmann 2001,
Walker and del Moral 2003). Previous studies have focused on a single trophic level,
typically plant (Matthews 1992) and less often animal community assembly
(Kaufmann 2001, Hodkinson et al. 2001, Hodkinson et al. 2004). Very little is known
about multitrophic community assembly and how plant communities interact with
animal communities during this process (Sargent and Ackerly 2008). Here, we study
a plant–pollinator community assembly process during primary succession on a
glacier foreland.
Most of the research on plant–pollinator interactions has tended to focus on
single species or guilds of related plants and their pollinators, giving little attention to
the larger community context (Waser et al. 1996). Recently, however, plant–
pollinator interactions at the community level have attracted considerable attention
(Waser and Ollerton 2006 and references therein), largely fueled by new
developments in the analysis of complex networks (Proulx et al. 2005). Complex
network analysis has provided important insight into the structure of plant–pollinator
networks (e.g. Jordano 1987, Olesen and Jordano 2002, Jordano et al. 2003, Stang et
al. 2006, Bascompte and Jordano 2007, Blüthgen et al. 2008, Ings et al. 2009,
Vasquez et al. 2009a,b). For example, an increasing number of studies suggest that
Albrecht et al. Plant–pollinator network assembly 4 _______________________________________________________________________
the structure of most plant–pollinator networks is asymmetric (Vazquez and Aizen
2004, Bascompte et al. 2006) and nested (Bascompte et al. 2003), whereby
specialized species interact with a subset of the interaction partners of more
generalized species. Furthermore, large interaction networks may show some degree
of modularity (Olesen et al. 2007). Several ecological and evolutionary determinants
of these network patterns are currently discussed. The concept of forbidden links (e.g.
Jordano et al. 2003) refers to the possibility that some interactions may not occur
because of mismatches in the morphology, phenology or spatial distribution of
species (Stang et al. 2006, Bascompte and Jordano 2007, Santamaría and Rodríguez-
Gironés 2007). Conversely, the neutrality hypothesis claims that network patterns are
mainly driven by random encounters, so that abundant species interact more
frequently and with a higher number of species than rare species (Vazquez 2005).
Recent findings suggest that both of these mechanisms contribute to observed plant–
pollinator network structures (Blüthgen et al. 2006, Stang et al. 2006, Bascompte and
Jordano 2007, Vazquez et al. 2009a,b).
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
Much less is known about the dynamics of plant–pollinator networks (but see
e.g. Petanidou et al. 2008, Olesen et al. 2008, Bascompte and Stouffer 2009). A
fundamental open question is how network structure changes during the process of
network assembly (Olesen et al. 2008, Bascompte and Stouffer 2009). Some studies
used simulation models to predict network disassembly regarding robustness and
stability (see Bascompte and Jordano 2007, Bascompte and Stouffer 2009 and
references therein), concluding that the heterogeneous degree distribution and the
cohesive organization in nested systems, as typically found for mutualistic networks,
may primarily account for the high robustness predicted for networks that are
structured in such a way. In an empirical study Olesen et al. (2008) recorded the day-
Albrecht et al. Plant–pollinator network assembly 5 _______________________________________________________________________
to-day dynamics of an arctic pollination network to analyze the attachment process of
new species (plant species that initiated flowering and pollinator species that started
to actively forage) to the network over two seasons. Studying plant–pollinator
networks during real succession, as done in the present paper, represents a new
approach to get a deeper insight into network assembly processes.
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
Understanding the patterns of specialization and generalization in plant–
pollinator interactions is a central and hotly debated topic in pollination ecology and
plant–pollinator network research (Waser et al. 1996, Johnson and Steiner 2000,
Waser and Ollerton 2006, Blüthgen et al. 2008, Petanidou et al. 2008). However,
there is a paucity of theory predicting the determinants of the degree of specialization
in pollination systems and how the latter may change during assembly processes
(Johnson and Steiner 2000). Successional status has been considered as such a
determinant (Parrish and Bazzaz 1979). From a plant’s perspective, niche theory
predicts that mature plant communities should be more specialized (have narrower
pollinator niches) than species of early successional communities, because the
expected higher density and diversity of plant species should result in more
interspecific competition and thus reduced niche overlap, at least in saturated systems
(Parrish and Bazzaz 1979). Furthermore, higher unpredictability of pollinator
availability should favor more generalized use of pollinators by plants (Waser et al.
1996) in early- as compared with late-successional plant communities.
In contrast to plants, pollinators only profit from increased specialization if
this is associated with increased resource-use efficiency or reduced interspecific
competition, e.g. through reduced handling time (e.g. Heinrich 1976, Pyke 1984).
However, if resources are similar and accessible, optimal foraging theory predicts
that, with an expected increase in resource-plant diversity during succession,
Albrecht et al. Plant–pollinator network assembly 6 _______________________________________________________________________
pollinators should be more generalized in mature stages to minimize foraging time
(Heinrich 1976, Pyke 1984, Waser et al. 1996, Blüthgen et al. 2007). Furthermore,
optimal foraging theory predicts that a decrease in resources provided by single plant
species should result in an increase in diet breadth (MacArthur and Pianka 1966).
Furthermore, increasing pollinator density during succession may result in a more
rapid depletion of floral resources and, as a consequence, also lead to a higher diet
breadth of individual pollinators (Fontaine et al. 2008).
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
In this paper we quantify plant–pollinator interactions at seven sites along a
chronosequence from 8 to 130 years since deglaciation at the Morteratsch glacier
foreland (SE-Switzerland). We examine how (1) plant diversity and flower
abundance, (2) pollinator diversity and visitation rate, (3) diversity of plant–pollinator
interactions, and (4) different aspects of specialization in plant–pollinator interactions
change along the chronosequence of the glacier foreland.
Material and Methods
Study site
The glacier “Morteratsch” is located at the south-facing slope of the Bernina Range in
the subalpine belt of the Central Alps of Switzerland (Fig. 1). Its foreland (9°56' E,
46°26' N) is approximately 2.2 km long and extends over altitudes from 1900 to 2010
m above sea level. Since the beginning of monitoring the glacier dynamics in 1881,
the glacier has continuously retreated at a rate of approximately 18 m per year (VAW
2007). The entire glacier foreland lies below the upper tree line of Larix decidua
Miller and Pinus cembra L. forest which reaches approximately 2400 m above sea
level in the study region. The plant primary succession along the chronosequence of
the glacier foreland is dominated by herbaceous plants in a pioneer stage, followed
Albrecht et al. Plant–pollinator network assembly 7 _______________________________________________________________________
after about 50 years by a stage with shrubs (Ziefle 2006). An open forest stage is
reached after around 150 years. Annual mean temperature at Samedan, 10 km north-
west of the glacier foreland at an altitude of 1700 m above sea level, is 1.2 °C, and
yearly average precipitation is 700 mm. Snow cover on the glacier foreland usually
lasts from October to May.
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
The foreland of the Morteratsch glacier was selected as model system for this
study because pronounced side moraines separate the glacier foreland from the
surrounding vegetated slopes, thus reducing plant propagule pressure from the slopes,
and because the history of the glacier retreat since 1857 is well documented (Elsener
2006 and references therein). Due to these factors the foreland of the Morteratsch
glacier has been a preferred study site for previous research on plant succession
(Ziefle 2006).
Study design
Plant–pollinator interactions were studied along 14 transects located at 7 sites of
increasing chronological age across the Morteratsch glacier foreland. Maps providing
exact information on past glacier movements and snout positions (Elsener 2006) were
digitalized and imported into a GIS. Next, the entire chronosequence from the last
glacier expansion in 1878 until present was subdivided into 7 strata limited by
isochrones of age since deglaciation, separated by time steps of roughly 20 years. The
isoclines represent the glacier extensions from 1878 until 2000 (Fig. 2). This map was
then merged with a map providing data about the characteristic vegetation for each of
110 intersection points of grid with a mesh width of 40 m and covering the entire
glacier foreland (Ziefle 2006, Elsener 2006). Then the known flooding areas close to
the glacial stream were excluded as potential areas for choosing sites. Finally, the
Albrecht et al. Plant–pollinator network assembly 8 _______________________________________________________________________
most representative vegetation unit was identified for each site as the unit with the
most intersection points per stratum.
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
At each site two transects that were perpendicular to each other were
established at a randomly selected intersection point of the most representative
vegetation nearby the delimiting isocline of the corresponding stratum. The seven
sites in 2008 corresponded to a chronological age since deglaciation of 8, 28, 49, 63,
84, 109 and 130 years (Fig. 2). Each intersection point was the center of a cross
consisting of a 50-m long transect in the east–west direction and a 30-m long transect
in the north–south direction. At the second oldest site (109 years since deglaciation),
sampling was not possible along the shorter transect due to very dense tree and shrub
vegetation (dominated by Betula sp., Salix sp. and Rhododendron ferrugineum L.).
Therefore, only the longer transect was sampled and analyzed at this site. Each
transect was marked with cord during the entire sampling season.
Sampling protocol for plant–pollinator interactions
Plant–pollinator interactions were observed during the entire flowering season (10
June to 21 August 2008) at roughly 2-week intervals. This resulted in seven sampling
rounds for each transect. The snow melted at roughly the same time at each sampling
site. Flower-visiting insects were sampled between 9:30 and 17:30 h in sunny weather
conditions. Because of potential temporal differences in pollinator activity, each site
was sampled at least twice in the morning (9:30–12:30), twice in the early afternoon
(12.30–15:30) and twice in the late afternoon (15:30–17:30). Time of day in which
sampling took place was varied randomly across sites and sampling rounds. During
each sampling round, all flowering vascular plants and their insect visitors were
sampled within a belt of 1 m along the transects. While slowly walking along the
Albrecht et al. Plant–pollinator network assembly 9 _______________________________________________________________________
transects every insect visitor landing on an open flower was caught with a sweep-net
immediately after landing and put into a 50 ml tube containing ethyl acetate. Thus,
the number of individual flowers within the same inflorescence that an insect visited
was not considered in the analyses. Each transect was walked twice during each
sampling round. The time spent for each transect and sampling round was 15 min for
the short transects and 25 min for the long transect, resulting in total sampling time of
4 hours and 40 min for each site. If possible, flower visitors were identified to species
level, otherwise to genus or family level (see Appendix 1). Visitation rate was defined
as the total number of flower visits observed along a transect during one sampling
round. Flower visitors were assumed to be pollinators. Flower abundance of vascular
plants was recorded as the number of inflorescences of each flowering plant species,
including species that did not receive any pollinator visits.
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
Quantifying plant–pollinator network structure
The pooled, quantified plant–pollinator networks for each site were plotted. To detect
changes in the interaction structure of the plant–pollinator networks along the
chronosequence, the following quantitative, weighted properties (i.e. properties that
account for variation in interaction strength) were calculated for each transect
(sampling rounds pooled): linkage density (LDq), interaction diversity (IDq),
pollinator generality (Gpoll,q, the mean number of plant species visited by a pollinator
species weighted by interaction strength; also referred to as generalization in the
following text) and plant generality (Gplant,q, the mean number of pollinator species
visiting a plant species weighted by interaction strength, equivalent to vulnerability
Vq of Bersier et al. 2002). Plant species that were not visited by pollinators were not
included in these calculations. These metrics were calculated following Bersier et al.
Albrecht et al. Plant–pollinator network assembly 10 _______________________________________________________________________
221
222
223
(2002) and Albrecht et al. (2007). They are based upon the Shannon indices of
entropy to measure the diversity of flowering plants (HN) and pollinator individuals
(HP) for each taxon k:
Error! Bookmark not defined.HN,k = −
bik
b• ki=1
s
∑ log2bik
b• k 224
HP,k = −bkj
bk•j=1
s
∑ log2
bkj
bk• 225
226
227
228
229
230
231
Here b•k is the number of flowering plants visited by visitor species k, and bk• is the
number of flower visitors visiting plant species k. The network metrics are based on
the “reciprocals” of HN,k and HP,k (Bersier et al. 2002):
nN,k =2HN,k
0
⎧⎨⎪
⎩⎪if b•k = 0
nP,k =2HP,k
0
⎧⎨⎪
⎩⎪if bk• = 0
The weighted linkage density (LDq) was calculated for each) as:
LDq =12
bk•
b••
nP,k+b• k
b••
nN,kk=1
s
∑k=1
s
∑⎛⎝⎜
⎞⎠⎟ 232
233
234
where b•• is the total number of visited flowering plants. Weighted pollinator
generality (Gpoll,q) and plant generality (Gplant,q) were calculated as:
Gpoll,q =b• k
b••k=1
s
∑ nN,k
235
Gpoll,q =bk•
b••k=1
s
∑ nP,k
236
237
238
239
To measure network level generalization, we calculated the recently
developed index H2’ by Blüthgen et al. (2006). This index characterizes the degree of
specialization among the two trophic levels in bipartite networks as the deviation
Albrecht et al. Plant–pollinator network assembly 11 _______________________________________________________________________
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
from an expected probability distribution of interactions, evaluated by the
standardized two-dimensional entropy. For a detailed description of the mathematical
derivation of the index see Blüthgen et al. (2006). The index accounts for variability
in interaction frequencies and is not only robust against variation in the size of a
network, but also its shape, i.e. the skewness of the frequency distribution of the
recorded species (Blüthgen et al. 2006, 2008).
An additional aspect of generalization in weighted, bipartite networks is
interaction strength asymmetry, calculated as the grand mean of the imbalance
between the interaction strengths of each species pair within a network (Bascompte et
al. 2006). We used the modified version of interaction strength asymmetry proposed
by Dormann et al. (2009), in which all singleton species are omitted from the network
to avoid that these species receive a very high influence.
The measure of the diversity of interactions (IDq) was calculated using the
Shannon index:
IDq = pi∑ log2 pi( )
where pi is the proportion of the interaction i of the total number of interactions.
Furthermore, we adapted Hurlbert’s probability of inter-specific encounters (PIE) as a
measure for evenness in ecological communities, which is unbiased by the number of
species in a sample (Hurlbert 1971), to measure interaction evenness across networks
as:
PIE =L
L+1⎡⎣⎢
⎤⎦⎥
1−Li
L⎛⎝⎜
⎞⎠⎟
2
i∑
⎡
⎣⎢⎢
⎤
⎦⎥⎥ 260
261
262
263
where L is the total number of individual plant–pollinator interactions and Li the
number of individual interactions of the i-th link between a specific plant–pollinator
pair.
Albrecht et al. Plant–pollinator network assembly 12 _______________________________________________________________________
We believe that the above mentioned weighted network properties are the
most appropriate metrics currently available to investigate our research questions
(Bersier et al. 2002, Blüthgen 2006, Dormann et al. 2009). However, in order to make
it easier to compare the network properties of this study with those of other plant–
pollinator network studies that give only qualitative network metrics we also
calculated these traditionally used, unweighted properties (number of links per
species, connectance, average pollinator species degree, average plant species degree,
average network level degree and nestedness). Connectance is the realized proportion
of all possible links. Species degree is the number of different species a species
interacts with. Nestedness is a measure of departure from systematic arrangement of
species by niche width (Bascompte et al. 2003). The nestedness “temperature” T (0°–
100°) measures the departure from a perfectly nested interaction matrix, with
maximum nestedness defined as T = 0°.
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
For the calculation of the more complex network properties H2’, interaction
strength asymmetry and nestedness the data of the two transects and the seven
sampling rounds were pooled for each of the seven sites (to avoid small network sizes
potentially affecting the functioning of the more complex algorithms used for the
calculations of these metrics). To test whether pooling of the two transects of a site
would lead to different results for other measures, all analyses for those other
measures were also performed using these pooled data (fitting linear models with the
single continuous explanatory variable site age). The results of these analyses were
very similar to those obtained by the mixed-effect model analyses presented here. H2’
was calculated on the webpage http://nils.mib.man.ac.uk/~nils/stat/. For the
calculations of all other network metrics (except Hurlbert’s PIE) the bipartite package
supplied in the R-system of statistical computing was used (Dormann et al. 2009).
286
287
288
Albrecht et al. Plant–pollinator network assembly 13 _______________________________________________________________________
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
Data analysis
We analyzed variation in the dependent variables of plant and pollinator species
richness, plant and pollinator Shannon diversity indices, flower abundance, visitation
rate of pollinators and the above-mentioned network properties with linear mixed-
effects models. For the Shannon diversity indices we used the formula described
above for interaction diversity IDq, but with pi as the proportion of inflorescences or
flower visiting individuals belonging to the i-th plant or pollinator species,
respectively. Site age (number of years since deglaciation) was treated as fixed
continuous explanatory variable. The remaining variation among sites (i.e. site fitted
after site age), transect nested within site, sampling round and the site x sampling
round interaction were treated as random effects. For the analyses of the network
properties data from all sampling rounds were pooled, because some networks of
single sampling rounds were too small for a reliable analysis (thus, sampling round
was not included in the analyses of network properties). According to the
specification of the random terms for the mixed-model analysis the fixed term site age
was tested against the residual variation among sites. This avoided a potential pseudo-
replication problem that would have occurred if site age would have been tested
against variation among transects or among residuals (= transect x sampling round
interactions). Transect length was included as a covariate in the models to account for
any potential effects of transect length. Finally, to test whether variation in pollinator
species richness and visitation rate along the chronosequence were explained by
variation in plant species richness and flower abundance, these two measures were
used as further covariates in the models with pollinator visitation rate, pollinator
Albrecht et al. Plant–pollinator network assembly 14 _______________________________________________________________________
species richness and pollinator diversity as the dependent variables and sampling
round, site age and transect length as explanatory variables.
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
We analyzed similarity between the networks at the different sites in observed
plant–pollinator interactions by calculating the modified Simpson’s index proposed
by Koleff et al. (2003). This index measures the similarity once differences in number
of interactions have been removed. It is calculated as the number of interactions of
two sites divided by the minimum number of interactions present in one of the two
sites (Petanidou et al. 2008). This modified Simpson’s index for interaction similarity
was calculated for each of the 7 x 6 / 2 = 21 site pairs. To test whether plant–
pollinator networks of closer sites were more similar in interactions than those of sites
further apart from each other along the chronosequence we fitted a linear model with
the response variable modified Simpson’s index for interactions and the explanatory
variables closeness (in years since deglaciation, i.e. 20 years for the pair consisting of
the 8- and 28-year old sites), mean age of the site pair (i.e. 18 years for the
aforementioned site pair), indicating whether similarity changed along the
chronosequence, and their interaction.
Mixed effect models were fitted using the lme-function of the nlme package
supplied in the R-system of statistical computing (R Development Core Team 2008).
Arithmetic means ± 1 standard error (SE) are reported. Due to the low number of sites
the power of the F1,5-test for site age was low and thus marginal significances at P <
0.1 are also reported as a protection against making too many type-II errors (Toft and
Shea 1983). The effect sizes in these analyses can easily be calculated from the
reported F-values and the degrees of freedom by taking the square-root of the ratio r2
= (df numerator * F) / (df numerator * F + df denominator) (Rosenthal and Rosnow
Albrecht et al. Plant–pollinator network assembly 15 _______________________________________________________________________
1985), thus the square-root of F / (F + 5). For example, an F-value of 5 results in an
effect size of 0.50, whereas an F-value of 10 results in an effects size of 0.67.
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
Results
Pooled, quantified plant–pollinator networks for all sites along the chronosequence
are shown in Figure 2. Over the entire season a total of 54 plant species were
flowering and 37 of them (67%) were visited by a total of 92 pollinator species (Fig.
2; Appendix 1). In total, 506 plant–pollinator interactions (pollinator visits) were
recorded. Whereas flower abundance was not significantly higher at late- than at
early-succesional sites (F1,5 = 3.98, P = 0.103; Fig. 3a), species richness of flowering
plants (F1,5 = 12.89, P = 0.016), Shannon diversity of flowering plants (F1,5 = 8.77, P
= 0.032), visitation rate (F1,5 = 8.47, P = 0.033), species richness of pollinators (F1,5 =
8.37, P = 0.034) and network size (total number of interacting plant and pollinator
species; F1,5 = 7.87, P = 0.038) all increased along the chronosequence (Fig. 3b–e).
Species richness and Shannon diversity of pollinators peaked at intermediate
successional age (84 years since deglaciation; Fig. 3e,f).
To test if the observed pattern in pollinator species richness and visitation rate
along the chronosquence was related to plant species richness and flower abundance,
we introduced these covariates stepwise into the linear models. Because plant species
richness and flower abundance were highly correlated (F1,6 = 58.84, P < 0.001), we
only report results of the stronger explanatory variable, which was plant species
richness. Plant species richness was highly significant in explaining variation in
pollinator richness (F1,76 = 52.78, P < 0.001) and pollinator visitation rate (F1,76 =
48.83, P < 0.001). Moreover, site age was not significant anymore in explaining
variation in pollinator richness or visitation rate when plant species richness was
Albrecht et al. Plant–pollinator network assembly 16 _______________________________________________________________________
included in the model (results not shown). This suggests that increasing site age
influenced pollinators indirectly via its effects on plant species richness, although
pollinator species richness and pollinator diversity showed a saturating or peaked
pattern, respectively, while pollinator visitation rate continued to increase in the late-
succesional stages.
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
Diptera accounted for 74%, Hymenoptera (almost exclusively Apidae)
accounted for 24% and other insect orders accounted for 2% of the visits to flowers
(Fig. 2, Appendix 2). The number of visits by Diptera significantly increased along
the chronosequence (F1,5 = 14.77, P = 0.012), whereas that of Hymenoptera showed a
non-significant decrease (F1,5 = 3.74, P = 0.111). This resulted in a significant change
in the relative number of visits by Diptera and Hymenoptera along the
chronosequence (decline in log-transformed Hymenoptera/Diptera-ratio: F1,5 = 14.43
P = 0.013) from the youngest site (55% Hymenoptera, 45% Diptera) to the oldest site
(4% Hymenoptera, 96% Diptera).
At the youngest two sites (8 and 28 years since deglaciation) two plant
species, Epilobium fleischeri (Onagraceae; species code 1268: Fig. 2, Appendix 2)
and Saxifraga bryoides (Saxifragaceae; species code 909: Fig. 2, Appendix 2),
received by far the most pollinator visits. Bumblebees almost exclusively visited E.
fleischeri at these two sites (Fig. 2). The interaction structure at these sites was
characterized by the dominant interaction between the bumblebee Bombus sichelii
and E. fleischeri (Fig. 2). At site 3 (49 years since deglaciation), 44% of the pollinator
individuals were recorded on Saxifraga paniculata (Saxifragaceae; species code 887:
Fig. 2, Appendix 2), which produced flowers most abundantly (34%) at this site. The
percentage of flowering plant species for which pollinator visits were observed was
lower at the two pioneer sites (24% at the 8-years old site and 31% for the 28-years
Albrecht et al. Plant–pollinator network assembly 17 _______________________________________________________________________
old site), while at older sites this percentage was much higher (54% on average,
ranging from 47% to 59%; Fig. 2).
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
Interaction diversity increased with site age (F1,5 = 7.38, P = 0.041), at least up
to 84 years (Fig. 4a). Interaction evenness tended to increase throughout the entire
130-year period covered by the chronosequence (F1,5 = 6.38, P = 0.053; Fig. 4b).
However, weighted linkage density (F1,5 = 2.12, P = 0.205; Fig. 4c) remained more or
less constant along the chronosequence, while the unweighted number of links per
species increased with site age (F1,5 = 8.42, P = 0.034; Appendix 2). Connectance
tended to decrease with site age (F1,5 = 4.13, P = 0.098; Appendix 2).
Weighted network-level specialization measured as Blüthgen’s H2’
significantly decreased along the chronosequence (F1,5 = 8.62, P = 0.032; Fig. 4d).
The strongest decline occurred between the 28- and the 63-year old sites. This
decrease in network-level specialization reflected the higher average number of
pollinator species visiting a plant species (increase in pollinator generality with site
age; F1,5 = 39.08, P = 0.002; Fig. 4e), while the weighted average number of plants
visited by a pollinator did not change (F1,5 = 0.12, P = 0.747; Fig. 4f). The qualitative,
unweighted equivalents of the above mentioned weighted metrics showed
qualitatively identical patterns (increase in average network level and pollinator
species degree: F1,5 = 8.04, P = 0.037; F1,5 = 7.86, P = 0.001; no directional trend in
average plant species degree: F1,5 = 1.18, P = 0.327; Appendix 2). Examples of
pollinator species that occur at most sites and that visited more plant species at older
sites compared to earlier ones include Tricops sp., Paonia sp. or Bombus sichelii. For
most of the abundantly flowering plant species the highest degree values (number of
visitor taxa per plant species) were recorded at the site with highest flower abundance
of the corresponding plant species. For example, for Epilobium fleischeri this was the
Albrecht et al. Plant–pollinator network assembly 18 _______________________________________________________________________
8-year old site, for Saxifraga paniculata the 49-year old site or for Achillea erba-rotta
the 84-year old site.
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
The contrasting patterns of plant and pollinator generalization in early-
successional stages resulted in a decline in interaction strength asymmetry along the
chronosequence (F1,5 = 9.39, P = 0.028; Fig. 4g), with pollinators being more
specialized in these early assembly stages than plants. Conversely, networks at
mature successional stages showed a more nested interaction structure than those at
early stages (F1,5 = 9.39, P = 0.028; Fig. 4h).
Interaction similarity between networks (the modified Simpson’s index) was
higher for closer sites than for sites further apart from each other along the
chronosequence (F1,17 = 13.31,12 P = 0.002; Appendix 3). However, average
interaction similarity did not change with site age (F1,17 = 0.04, P = 0.854; Appendix
3).
Discussion
Our study shows pronounced changes in the structure of plant–pollinator networks
along the chronosequence of a primary glacier foreland succession. These changes
can be interpreted as community assembly processes leading to increasingly diverse
plant–pollinator systems at least during the first 80 years after deglaciation. With
increasing site age, the average pollinator visited more plant species (increasing
pollinator generalization, but not plant generalization), suggesting that plant
community assembly may have been the driver in this diversity-begets-diversity
assembly process. As a consequence, network level specialization (measured as
Blüthgen’s H2’) and interaction strength asymmetry declined up to a mid-successional
stage.
Albrecht et al. Plant–pollinator network assembly 19 _______________________________________________________________________
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
Plant and pollinator community assembly
Species richness and Shannon diversity of plants with open flowers increased from
the youngest to the mid- and late-successional sites along the chronosequence (see
Fig. 3). This confirms findings of previous studies on plant primary succession on the
foreland of the Morteratsch glacier (Ziefle 2006) and most other studied glacier
forelands (Matthews 1992, Walker and del Moral 2003).
Pollinator community assembly seemed to follow plant community assembly.
Thus, the increase in plant diversity along the chronosequence was paralleled by an
increase in visitation rate and species richness of pollinators and, at least up to a mid-
successional stages, pollinator Shannon diversity (see Fig. 3). Recent experimental
work also indicates that pollinator visitation rate increases linearly with plant species
richness, while pollinator diversity scales with plant species richness along a
saturating curve (Ebeling et al. 2008). Considering the relatively high mobility of
most pollinator taxa, colonization of freshly available, early-successional habitats by
pollinators was probably not dispersal limited. In addition, the increasing
generalization of pollinators suggests that the diversity of available floral resources
was the crucial factor affecting pollinator community assembly on the local scale of
our seven sites (Potts et al. 2003b). Evidence from studies on secondary successions
also indicates that pollinator diversity mainly mirrors plant diversity, which can lead
to increasing pollinator diversity with increasing age of meadows (Gathmann et al.
1994) or decreasing pollinator diversity along recovering vegetation gradients after
fire, with highest plant diversity levels in the first few years after a burn (Potts et al.
2003a).
Albrecht et al. Plant–pollinator network assembly 20 _______________________________________________________________________
In addition to the changes in pollinator species richness, the species
composition of pollinator communities also changed along the chronosequence of the
primary succession of our glacier foreland. Bees, particularly bumblebees, dominated
in pioneer stages and flies in mature stages. These changes in pollinator species
composition probably reflect those in species composition of the plant communities
along the chronosequence. For example, the abundance of bumblebees closely
mirrors the flower abundance of Epilobium fleischeri, a pioneer species with highest
abundances at the youngest two sites.
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
Plant–pollinator network assembly
Interaction diversity and less so interaction evenness increased along the
chronosequence up to mid-successional stages. These trends were consistent with an
increase in pollinator generality and a decrease in network-level specialization along
the chronosequence (see Fig. 4). In contrast to some measures used previously to
characterize generalization or specialization, such as the average species degree
(number of taxa interacting with one partner species; e.g. Herrera 2005) or
connectance (proportion of realized of all possible links; Jordano 1987, Olesen and
Jordano 2002), the quantitative, weighted network metrics used in this study are
largely independent of network size and sampling effort (Bersier et al. 2002, Blüthgen
et al. 2006). The metric H2’ additionally controls for variation in skewness of the
frequency distribution of the data, which may affect some weighted properties such as
generality or interaction strength asymmetry (Blüthgen et al. 2008). In spite of their
deficiencies, the qualitative metrics of plant and pollinator generality and network
level generalization (average plant, pollinator and network level degree) produced
similar results as their quantitative, weigthed counterparts. Only for connectance, a
Albrecht et al. Plant–pollinator network assembly 21 _______________________________________________________________________
qualitative metric that has often been used to describe generalization in networks,
there was a tendency to decline with site age, while the results for quantitative H2’
(and the qualitative average network degree) indicated a significant increase in
network level generalization. Connectance as a qualitative metric to characterize
generalization across networks has been criticized (Blüthgen et al. 2006, 2008 and
references therein) because it is not only sensitive to sampling intensity but also
because it is negatively correlated with network size (e.g. Olesen and Jordano 2002).
Since network size increased with site age in our study, a decrease in connectance
with site age is expected only due to this mathematically inherent scale dependence
(Blüthgen et al. 2006). This emphasizes the importance of using properties that are
more robust against such effects when comparing networks differing in size (Bersier
et al. 2002, Blüthgen et al. 2006).
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
Why does plant generality not decrease with successional age?
Our results do not support niche-theoretical predictions that plant species in mature
communities should be more specialized with respect to their pollinators than those in
early-successional communities because of increased interspecific competition and
reduced niche overlap (Parrish and Bazzaz 1979). Although some characteristic plant
species of the two pioneer sites, such as Epilobium fleischeri or Saxifraga bryoides
were visited by a relatively high number of different pollinator species (Fig. 2),
overall plant generalization did not show a significant directional trend along the
chronosequence. Interestingly, however, some abundantly flowering plant species
such as Arabis alpina at the pioneer sites received no or very few pollinator visits.
Most studies on plant–pollinator networks do not provide data regarding the
percentage of flowering plant species for which no pollinator visits are observed. In
Albrecht et al. Plant–pollinator network assembly 22 _______________________________________________________________________
our study, approximately one third of all flowering species received no pollinator
visits, although most of them are known to be normally pollinated by insects. This
percentage was roughly twice as high for the two youngest sites compared with the
more mature ones, indicating that a considerably lower number of flowering plant
species is integrated into the pollination network at early than at late stages of
network assembly.
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
Why does pollinator generality increase with successional age?
Because plant diversity increased yet total flower abundance remained relatively
constant along the chronosequence, the abundance of flowers of individual plant
species declined. Under these circumstances and according to optimal foraging
theory, pollinators should become generalists to minimize foraging time (MacArthur
and Pianka 1966, Heinrich 1976, Pyke 1984). Thus, the increase in pollinator
generalization and the decrease in network-level specialization found in this study
may reflect the increase in heterospecific floral resources during the primary
succession. Another prediction of optimal foraging theory that is supported by our
findings is that the increase in pollinator abundance during the succession, indicated
by the positive correlation between visitation rate and site age, should result in a
higher diet breadth of individual pollinators as a response to more rapid depletion of
floral resources at high pollinator abundance (Fontaine et al. 2008). Such a parallel
increase in abundance and generalization along the chronosequence was found for
example for several taxa of flies or the bumblebee species Bombus sichelii. However,
also many plant species showed a positive correlation between flower abundance and
generalization level (see Fig. 4). A simple mechanism of network assembly that is
compatible with such a pattern is the preferential attachment model (Barabasi and
Albrecht et al. Plant–pollinator network assembly 23 _______________________________________________________________________
Albert 1999). According to this model, new species have a higher probability to link
to species that have already many links in a network, leading to a “rich-get-richer”
process. Studying the day-to-day attachment of new species (i.e. species that started
to flower or actively forage) to an arctic plant–pollinator network during two seasons,
Olesen et al. (2008) found that new species indeed tended to attach to already well-
connected species, although this was constrained by some morphological and
phenological mismatching of plants and pollinators. Although we did not study such
factors as potential drivers of the observed patterns of network assembly in the
present study, we found at least some evidence that abundance was positively linked
with the degree of many plant and pollinator species.
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
Our findings are consistent with the view that plant–pollinator networks, at
least in later successional stages, may be not so much niche-structured, but may rather
reflect a high degree of opportunism with respect to resource use and thus a more
neutral way of plant–pollinator networks assembly (Vazquez 2005, Petanidou et al.
2008). Such opportunism is, however, obviously limited by phenological,
morphological and co-evolutionary constraints (Jordano et al. 2003, Stang et al. 2006,
Bascompte and Jordano 2007, Santamaría and Rodríguez-Gironés 2007, Vazquez et
al. 2009a,b). On glacier forelands and other systems structured by primary
succession, plants often occur in scattered and sparse populations, in particular during
pioneer stages, and thus may depend much more than those of later assembly stages
on specialized pollinators that show high levels of fidelity (Johnson and Steiner 2000
and references therein). On the other hand, a plant species typically found in early
stages of primary succession that relies to a large degree on animal pollination, such
as E. fleischeri (Albrecht, unpublished data), should attract a diverse pollinator
community considering the unpredictability of pollinator availability during this stage
Albrecht et al. Plant–pollinator network assembly 24 _______________________________________________________________________
(Parrish and Bazzaz 1979). This should lead to exactly the pattern found here, with
higher pollinator specialization but not plant specialization in the pioneer stages of the
succession, resulting in a higher interaction strength asymmetry in these early stages.
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
Increased nestedness of plant–pollinator networks with successional age
Most plant–pollinator networks that have been studied so far were characterized by a
highly nested interaction structure, i.e. more specialized species tend to interact with
interaction partners that are highly generalized, and with nestedness levels
significantly higher compared with those of null models (Bascompte et al. 2003,
Bascompte and Jordano 2007). Our finding of high nestedness levels in the mature
assembly stages is in a line with the existing studies typically investigating mature,
relatively stable systems. The early assembly stages, however, were significantly less
nested, indicating that nestedness is a typical feature of rather developed and stable
plant–pollinator networks, but not necessarily of still evolving ones at the beginning
of assembly processes. This view would also be in agreement with simulation models
suggesting that a highly nested interaction structure makes ecological networks more
cohesive and stable (Bascompte et al. 2003, 2009, Bascompte and Jordano 2007).
Conclusions
It is a well-known fact that ecological network data so far have always been
constructed of incomplete samples (Blüthgen et al. 2008). Also in the present study,
sampling thoroughness may not have been extremely high. However, here we were
primarily interested in the relative change of network properties and their pattern
along a primary successional gradient rather than estimate values, for example of
fundamental specialization, most accurately. Despite potential limitations inherent in
Albrecht et al. Plant–pollinator network assembly 25 _______________________________________________________________________
the space-for-time substitution approach and the above mentioned uncertainties in
sampling thoroughness typical for ecological network data, the analysis of networks
along chronosequences provide unique insights into the assembly of multitrophic
communities and their interaction structure that cannot be gained by short-term
experimental studies (Matthews 1992, Hodkinson et al. 2004). To our knowledge, this
is the first study that used a network approach to explore plant–pollinator assembly
during primary succession. We found pronounced changes in the interaction structure
of the plant–pollinator communities during the studied glacier foreland succession,
which was characterized by a strong increase in pollinator generality, but not plant
generality, during network assembly, resulting in a decline of network level
specialization and interaction strength asymmetry. An important implication of this
study is that the network structure of evolving ecological communities in early
assembly stages can be significantly different to that of the usually studied, rather
mature and stable ones. Further research is needed to unravel the exact mechanistic
processes underlying the observed patterns in the evolution of plant–pollinator
network properties during this type of primary succession. Ecological network
analysis along successional gradients is a promising tool to gain new insights in how
networks of multitrophic communities assemble.
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
Acknowledgements
We dedicate this paper to Christine Müller, who died after the start of this study; she
was a great inspiration for us. We thank Ulrich Schneppat for valuable advice on
insect preparation and the following taxonomic specialists for help with the species
identification: Jean-Paul Haenni (Diptera) and Hermann Blöchlinger (Diptera), Ruth
Bärfuss (Syrphidae), Rainer Neumeyer and Felix Amiet (Apidae), Seraina Klopfstein
Albrecht et al. Plant–pollinator network assembly 26 _______________________________________________________________________
and Hannes Baur (Ichnneumonidae), Jürg Schmid (Lepidoptera), Peter Herger
(Coleoptera) and Denise Wyniger (Heteroptera). We are grateful to Stefan Elsener for
providing GIS data and Roland Schmid for help with drawing the plant–pollinator
networks. We also thank Martina Stang and two anonymous reviewers for their
valuable comments on an earlier version of this manuscript. This work was supported
by the Swiss National Science Foundation (grant 631-065950 to Christine Müller).
611
612
613
614
615
616
Albrecht et al. Plant–pollinator network assembly 27 _______________________________________________________________________
References 617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
Albrecht, M. et al. 2007. Interaction diversity within quantified insect food webs in
restored and adjacent intensively managed meadows. – J. Anim. Ecol. 76: 1015–
1025.
Bascompte, J. et al. 2003. The nested assembly of plant–animal mutualistic networks.
– Proc. Natl Acad. Sci. 100: 9383–9387.
Bascompte, J. et al. 2006. Asymmetric coevolutionary networks facilitate biodiversity
maintenance. – Science 312: 431–433.
Bascompte, J. and Jordano, P. 2007. Plant–animal mutualistic networks: the
architecture of biodiversity. – Annu. Rev. Ecol. Evol. Syst. 38: 567–593.
Bascompte J. and Stouffer, D. B. 2009. The assembly and disassembly of ecological
networks. – Phil. Trans. R. Soc. B 364: 1781–1787.
Bersier, L. F. et al. 2002. Quantitative descriptors of food-web matrices. – Ecology
83: 2394–2407.
Blüthgen, N. et al. 2006. Measuring specialization in species interaction networks. –
BMC Ecol. 6: 9.
Blüthgen, N. et al. 2007. Specialization, constraints, and conflicting interests in
mutualistic networks. – Curr. Biol. 17: 341–346.
Blüthgen, N. et al. 2008. What do interaction network metrics tell us about
specialization and biological traits? – Ecology 89: 3387–3399.
Clements, F. E. 1916. Plant succession. Carnegie Institute of Washington.
Dormann, C. et al. 2009. Indices, graphs and null models: analyzing bipartite
ecological networks. – The Open Ecology Journal 2: 7-24.
Ebeling, A. et al. 2008. How does plant richness affect pollinator richness and
temporal stability of flower visits? – Oikos 117: 1808–1815.
Albrecht et al. Plant–pollinator network assembly 28 _______________________________________________________________________
Elsener, S. 2006. Besiedlungsdynamik von Arve (Pinus cembra L.) und Lärche (Larix
decidua Mill.) im Vorfeld des Morteratschgletschers. – Diploma thesis. Swiss
Federal Institute for Forest, Snow and Landscape Research.
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
Fontaine, C. et al. 2008. Generalist foraging of pollinators: diet expansion at high
density. – J. Ecol. 96: 1002–1010.
Foster, B. L. and Tilman, D. 2000. Dynamic and static views of succession: testing
the descriptive power of the chronosequence approach. – Plant Ecol. 146: 1–10.
Gathmann, A. et al. 1994. Trap-nesting bees and wasps colonizing set-aside fields –
succession and body-size, management by cutting and sowing. – Oecologia 98: 8–
14.
Heinrich, B. 1976. The foraging specializations of individual bumblebees. – Ecol.
Monogr. 146: 105–128.
Herrera, C. M. 2005. Plant generalization on pollinators: Species property or local
phenomenon? – Am. J. Bot. 92: 13–20.
Hodkinson, I. D. et al. 2001. What a wonderful web they weave: spiders, nutrient
capture and early ecosystem development in the high Arctic – some counter-
intuitive ideas on community assembly. – Oikos 95: 349–352.
Hodkinson, I. D. et al. 2004. Invertebrate community assembly along proglacial
chronosequences in the high Arctic. – J. Anim. Ecol. 73: 556–568.
Hubell, S. P. 2001. The unified neutral theory of biodiversity and biogeography. –
Princeton Univ. Press.
Hurlbert, S. 1971. Nonconcept of species diversity – critique and alternative
parameters. – Ecology 52: 577–586.
Ings, T. C. et al. 2009. Ecological networks – beyond food webs. – J. Anim. Ecol. 78:
253–269.
Albrecht et al. Plant–pollinator network assembly 29 _______________________________________________________________________
Johnson, S. D. and Steiner, K. E. 2000. Generalization versus specialization in plant
pollination systems. – Trends Ecol. Evol. 15: 140–143.
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
Jordano, P. 1987. Patterns of mutualistic interactions in pollination and seed dispersal:
connectance, dependence asymmetries, and coevolution. – Am. Nat. 129: 657.
Jordano, P. et al. 2003. Invariant properties in coevolutionary networks of plant–
animal interactions. – Ecol. Lett. 6: 69–81.
Kaufmann, R. 2001. Invertebrate succession on an alpine glacier foreland. – Ecology
82: 2261–2278.
Koleff, P. 2003. Measuring beta diversity for presence-absence data. – J. Anim. Ecol.
72: 367-382.
Lortie, C. J. et al. 2004. Rethinking plant community theory. – Oikos 107: 433–438.
MacArthur, R. H. and Pianka, E. R. 1966. On optimal use of a patchy environment. –
Am. Nat. 100: 603-609.
Matthews, J. A. 1992. The ecology of recently deglaciated terrain. – Cambridge Univ.
Press.
Olesen, J. M. and Jordano, P. 2002. Geographic patterns in plant–pollinator
mutualistic networks. – Ecology 83: 2416–2424.
Olesen, J. M. et al. 2007. The modularity of pollination networks. – Proc. Natl Acad.
Sci. 104: 19891–19896.
Olesen, J. M. et al. 2008. Temporal dynamics in a pollination network. – Ecology 89:
1573–1582.
Parrish, J. A. D. and Bazzaz, F. A. 1979. Difference in pollination niche relationships
in early and late successional plant communities. – Ecology 60: 597–610.
Albrecht et al. Plant–pollinator network assembly 30 _______________________________________________________________________
Petanidou, T. et al. 2008. Long-term observation of a pollination network: fluctuation
in species and interactions, relative invariance of network structure and
implications for estimates of specialization. – Ecol. Lett. 11: 564–575.
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
Potts, S. G. et al. 2003a. Response of plant–pollinator communities to fire: changes in
diversity, abundance and floral reward structure. – Oikos 101: 103–112.
Potts, S. G. et al. 2003b. Linking bees and flowers: how do floral communities
structure pollinator communities? – Ecology 84: 2628–2642.
Proulx, S. R. et al. 2005. Network thinking in ecology and evolution. – Trends Ecol.
Evol. 20: 345–353.
Pyke, G. H. 1984. Optimal foraging theory – a critical review. – Annu. Rev. Ecol.
Syst. 15: 523–575.
Rosenthal, R. and Rosnow, R. L. 1985. Contrast analysis: Focused comparisons in the
analysis of variance. – Cambridge Univ. Press.
Santamaría, L. and Rodríguez-Gironés, M. 2007. Linkage rules for plant–pollinator
networks: Trait complementarity or exploitation barriers? – PLoS Biol. 5: e31.
Sargent, R. D. and Ackerly, D. D. 2008. Plant–pollinator interactions and the
assembly of plant communities. – Trends Ecol. Evol. 23: 123–130.
Stang, M. et al. 2006. Size constraints and flower abundance determine the number of
interactions in a plant–flower visitor web. – Oikos 112: 111–121.
Toft, C. A. and Shea, P. J. 1983. Detecting community-wide patterns: estimating
power strengthens statistical inference. – Am. Nat. 122: 618-625.
VAW 2007. Gletscherberichte (1881–2007). – Laboratory of Hydraulics, Hydrology
and Glaciology (VAW).
Vazquez, D. P. and Aizen, M. A. 2003. Null model analyses of specialization in
plant–pollinator interactions. – Ecology 84: 2493–2501.
Albrecht et al. Plant–pollinator network assembly 31 _______________________________________________________________________
Vazquez, D. P. and Aizen, M. A. 2004. Asymmetric specialization: a pervasive
feature of plant–pollinator interactions. – Ecology 85: 1251–1257.
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
Vazquez, D. P. 2005. Degree distribution in plant–animal mutualistic networks:
forbidden links or random interactions? – Oikos 108: 421–426.
Vazquez, D. P. et al. 2009a. Uniting pattern and process in plant–animal mutualistic
networks: a review. – Ann. Bot. 103: 1445–1457.
Vazquez, D. P. et al. 2009b. Evaluating multiple determinants of the structure of
plant–animal mutualistic networks. – Ecology 90: 2039–2046.
Walker, L. R. and del Moral, R. 2003. Primary succession and ecosystem
rehabilitation. – Cambridge Univ. Press.
Waser, N. M. et al. 1996. Generalization in pollination systems, and why it matters. –
Ecology 77: 1043–1060.
Waser, N. M. and Ollerton, J. 2006. Plant–pollinator Interactions. – Univ. of Chicago
Press.
Ziefle, M. 2006. Gletscherschwund und Vegetationsdynamik in Morteratsch
(Pontresina, Schweiz). – Diploma thesis. Univ. of Zürich.
Albrecht et al. Plant–pollinator network assembly 32 _______________________________________________________________________
Figure legends 733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
Fig. 1. Location and map of the Morteratsch glacier foreland at the south-facing slope
of the Bernina Range (Engadine, Switzerland). Positions of sampling sites
representing a chronosequence from 8 to 130 years since deglaciation together with
isoclines of glacier extensions from 1857 to 2008 (grey lines) and the glacial stream
(black line).
Fig. 2. Quantified plant–pollinator networks at seven sites of increasing distance from
the glacier snout across the Morteratsch glacier foreland, representing a
chronosequence from 8 to 130 years since deglaciation. Lower bars represent flower
abundance of vascular plants and upper bars pollinator visitation rates drawn at
different scales. The width at the basis of the wedges represents interaction frequency.
Taxa are coded by numbers (see Appendix 2 for names).
Fig. 3. Flower abundance (a), plant species richness (b), plant diversity (Shannon
diversity index) (c), pollinator visitation rate (d), pollinator species richness (e) and
pollinator diversity (Shannon diversity index) (f) at seven sites of increasing distance
from the glacier snout across the Morteratsch glacier foreland, representing a
chronosequence from 8 to 130 years since deglaciation. The broken lines connect the
mean values of the two transects of each site (for better visualisation they are slightly
jittered; at the 109-year old site only one transect could be sampled). For the
calculation of (a) to (c), all flowering plant species, including species that did not
receive any pollinator visits, were used. Plant–pollinator interactions were sampled
during a total of four hours and 40 min per site during seven sampling rounds from 10
June to 21 August 2008.
Albrecht et al. Plant–pollinator network assembly 33 _______________________________________________________________________
758
759
760
761
762
763
764
765
766
767
768
769
770
771
Fig. 4. Changes in quantitative, weighted network properties at seven sites of
increasing distance from the glacier snout across the Morteratsch glacier foreland,
representing a chronosequence from 8 to 130 years since deglaciation: (a) interaction
diversity (Shannon diversity index), (b) interaction evenness (measured as Hurlbert’s
probability of interspecific encounters), (c) linkage density, (d) network level
specialization (measured as Blüthgen’s H2’ using the pooled networks of a site, see
“Material and Methods”), (e) pollinator generality and (f) plant generality. The broken
lines connect the mean values of the two transects of each site (for better visualisation
they are slightly jittered; at the 109-year old site only one transect could be sampled).
Plant species that were not visited by pollinators were not included in these analyses.
For description and calculation of the network properties see “Material and Methods”
section.
Albrecht et al. Plant–pollinator network assembly 34 _______________________________________________________________________
Figures 772
773
774
Figure 1
775
Albrecht et al. Plant–pollinator network assembly 35 _______________________________________________________________________
Figure 2 776
777
778
779
Albrecht et al. Plant–pollinator network assembly 36 _______________________________________________________________________
Figure 3 780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
Albrecht et al. Plant–pollinator network assembly 37 _______________________________________________________________________
Figure 4 798
799
800
801
Albrecht et al. Plant–pollinator network assembly 38 _______________________________________________________________________
Appendix 1. List of codes for pollinator and plant species used in Fig. 2. 802
Code Pollinator Code Plant 3001 Anthidium montanum Morawitz 1864 221 Papaver croceum Ledebour 3002 Anthomyiidae sp. 385 Cerastium fontanum ssp. fontanum Baumgarten
3003 Bombus monticola Smith 1849 390 Cerastium arvense ssp. strictum Koch
3004 Bombus mucidus Gerstaecker 1869 419 Silene vulgaris ssp. vulgaris Garcke 3005 Bombus pyrenaeus Perez 1879 447 Oxyria digyna Hill
3006 Bombus sichelii Radoszkowski 1859 472 Rumex scutatus L.
3007 Bombus sp. 583 Salix sp. 3008 Calliphoridae sp. 652 Cardamine resedifolia L.
3009 Cantharis pagana Rosenhauer 1847 667 Arabis alpina L.
3010 Cauchas fibulella Denis & Schiffermüller 1775 778 Myrica germanica L. 3011 Chamaesyrphus scaevoides Fallén 1817 786 Rhododendron ferrugineum L.
3012 Cheilosia albitarsis Meigen 1822 789 Vaccinium vitis–idaea L.
3013 Cheilosia sp. Meigen 1822 803 Pyrola minor L. 3014 Cheilosia variabilis Panzer 1798 818 Primula latifolia Lapeyrouse
3015 Coenosia sp. Meigen 1826 879 Sempervivum montanum L.
3016 Coenosiinae subfam. 880 Sempervivum arachnoideum L. 3017 Colletes impunctatus Nylander 1852 882 Sempervivum tectorum ssp. Alpinum L.
3018 Coreus sp. Fabricius 1794 887 Saxifraga paniculata Miller
3019 Cryptocephalus aureolus Suffrian 1847 901 Saxifraga aizoides L. 3020 Dasysyrphus friuliensis van der Goot 1960 909 Saxifraga bryoides L.
3021 Dilophus femoratus Meigen 1804 910 Saxifraga aspera L.
3022 Empididae sp. 936 Geum reptans L. 3023 Empis sp. L. 1758 969 Potentilla aurea L.
3024 Empria fletcheri Cameron 1878 1129 Trifolium pratense ssp. pratense L.
3025 Episyrphus balteatus De Geer 1776 1135 Trifolium pallescens Schreber 3026 Eristalis tenax L. 1758 1145 Trifolium badium Schreber
3027 Eupeodes corollae Fabricius 1794 1150 Anthyllis vulneraria ssp. alpestris Schultes
3028 Eupeodes latilunulatus Collin 1931 1158 Lotus alpinus Ramond
3029 Eupeodes nitens Zetterstedt 1843 1268 Epilobium fleischeri Hochstetter
3030 Halictus rubicundus Christ 1791 1297 Thesium alpinum L.
3031 Helina annosa Zetterstedt 1838 1499 Laserpitium halleri Crantz
3032 Hybomitra sp. Enderlein 1922 1585 Myosotis scorpioides L.
3033 Hybotidae sp. Enderlein 1922 1701 Thymus sp.
3034 Exochus sp. Gravenhorst 1829 1761 Linaria alpina ssp. alpina (L.) Miller 3035 Lasioglossum alpigenum Dalla Torre 1877 1796 Veronica fructicans Jacques
3036 Lasioglossum fratellum Perez 1903 1824 Euphrasia minima Schleicher
3037 Leucozona lucorum L. 1758 1846 Rhiantus minor L. 3038 Malthodes sp. Kiesenwetter 1852 1853 Melampyrum silvaticum L.
3039 Meliscaeva auricollis Meigen 1822 1900 Campanula barbata L.
3040 Miridae sp. 1908 Campanula cochleariifolia Lamarck
3041 Muscidae sp. 1917 Phyteuma hemisphaericum L.
3042 Myopa sp. Fabricius 1775 1923 Phyteuma betonicifolium Villars
3043 Nematus bohemani Thomson 1871 1954 Galium anisophyllon Villars
3044 Osmia inermis Zetterstedt 1838 2012 Valeriana tripteris L.
3045 Osmia loti Morawitz 1867 2028 Adenostyles leucophylla (Willd.) Reichenbach
3046 Paragus punctulatus Zetterstedt 1838 2030 Solidago virgaurea ssp. minuta (L.) Archangeli 3047 Phaonia hybrida Schnabl 1888 2051 Erigeron acer ssp. politus (Fr.) Lindberg
3048 Phaonia meigeni Schnabl 1888 2067 Anthennaria dioica (L.) Gaertner
3049 Phaonia serva Meigen 1826 2114 Achillea erba-rotta ssp. moschata (Wulfen) Vacc.
Albrecht et al. Plant–pollinator network assembly 39 _______________________________________________________________________
3050 Phaonia sp. Robineau-Desvoidy 1830 2139 Leucanthemopsis alpina (L.) Heywood
3051 Phoridae sp. Robineau-Desvoidy 1830 2271 Leontodon helveticus Mérat 3052 Phratora vitellinae L. 1758 2297 Taraxacum sp.
3053 Pipiza bimaculata Meigen 1822 2343 Hieracium staticifolium Allioni
3054 Pipizella sp. Rondani 1856 2357 Hieracium murorum L. 3055 Platycheirus albimanus Fabricius 1781
3056 Platycheirus melanopsis Loew 1856
3057 Platycheirus scutatus Meigen 1822 3058 Platypalpus sp. Le Peletier & Serville 1828
3059 Psilidae sp.
3060 Rhamphomyia sp. Meigen 1822 3061 Rhingia campestris Meigen 1822
3062 Sarcophagidae sp.
3063 Scaeva pyrastri L. 1758 3064 Scaeva selenitica L. 1758
3065 Sciaridae sp.
3066 Siphona sp. Meigen 1803 3067 Sphaerophoria bankowskae Goeldlin 1989
3068 Sphaerophoria laurae Goeldlin 1989
3069 Sphaerophoria scripta L. 1758 3070 Sphaerophoria sp. Le Peletier & Serville 1828
3071 Ichneumon sp. L. 1758
3072 Spilogona sp. Schnabl 1911 3073 Stomoxys calcitrans L. 1758
3074 Syrphus ribesii L. 1758
3075 Syrphus torvus Osten-Sacken 1875 3076 Syrphus vitripennis Meigen 1822
3077 Tachinidae sp.
3078 Tenthredo brevicornis Konow 1886 3079 Thricops aculeipes Zetterstedt 1838
3080 Thricops cunctans Meigen 1826
3081 Thricops longipes Zetterstedt 1845 3082 Thricops nigritellus Zetterstedt 1838
3083 Thricops rostratus Meade 1882
3084 Thricops semicinereus Wiedemann 1817 3085 Thricops separ Zetterstedt 1845
3086 Thricops sp. Zetterstedt 1845
3087 Thricops sudeticus Schnabl 1888 3088 Pipizella annulata Macquart 1829
3089 Bombus pratorum L. 1761
3093 Cryptinae subfam. 3094 Tryphoninae subfam.
3095 Campopleginae subfam.
803
804
Albrecht et al. Plant–pollinator network assembly 40 _______________________________________________________________________
805
806
807
808
809
810
T1 T2 Site T1 T2 Site T1 T2 Site T1 T2 Site T1 T2 Site T1 T2 Site T1 T2 Site
Qualitative network properties
Network size 17 15 25 16 18 27 31 27 42 24 24 37 33 37 54 26 NA 26 41 24 49Number of visited plant species 4 3 4 4 3 5 9 10 12 11 11 16 12 9 15 17 NA 17 26 14 20Number of not visited plant species 12 8 13 10 7 11 15 7 13 9 12 12 12 6 10 2 NA 2 17 11 3Number of pollinator species 13 12 21 12 15 22 22 17 30 13 13 21 21 28 39 9 NA 9 15 10 29Links per species 0.77 0.87 0.92 0.75 0.83 0.85 0.94 0.78 1.02 0.75 0.71 0.89 0.94 0.92 1.11 1.08 NA 1.08 1.07 0.96 1.2Connectance 0.25 0.36 0.27 0.25 0.33 0.21 0.15 0.12 0.12 0.13 0.12 0.10 0.12 0.13 0.10 0.18 NA 0.18 0.11 0.16 0.10Average degree 1.53 1.73 1.84 1.5 1.67 1.7 1.87 1.56 2.05 1.5 1.42 1.78 1.88 1.84 2.22 2.25 NA 2.25 2.15 1.92 2.41Plant degree 3.25 4.33 5.75 3 5 4.6 3.22 2.1 3.58 1.64 1.55 2.06 2.58 3.78 4 3.11 NA 3.11 2.93 2.3 2.95Pollinator degree 1 1.08 1.1 1 1 1.05 1.32 1.24 1.43 1.38 1.31 1.57 1.48 1.21 1.54 1.65 NA 1.65 1.69 1.64 2.03
130
Site age (years)
49 63 84 1098 28
Appendix 2. Qualitative (unweighted) network properties of plant–pollinator networks at seven sites of increasing distance from the glacier
snout across the Morteratsch glacier foreland, representing a chronosequence from 8 to 130 years since deglaciation. T1 = value for the 50 m
long transect; T2 value for the 30 m long transect; Site = value for the pooled data of the two transect for a site. At the 109-year old site only one
transect could be sampled. For the calculation of the network properties only the number of flowering plant species visited by pollinators were
used.
Albrecht ________
et al. Plant–pollinator network assembly 41 _______________________________________________________________
811
812
813
814
815
816
817
818
Site pair Modified Simpson's index
8 – 28 0.308 – 49 0.228 – 63 0.008 – 84 0.13
8 – 109 0.048 – 130 0.0028 – 49 0.1728 – 63 0.0928 – 84 0.17
28 – 109 0.1328 – 130 0.0449 – 63 0.1249 – 84 0.16
49 – 109 0.2049 – 130 0.0263 – 84 0.24
63 – 109 0.1263 – 130 0.1284 – 109 0.1784 – 130 0.08
109 – 130 0.14
Appendix 3. Values of the modified Simpson’s index as a measure of interaction
similarity of the plant–pollinator networks for each site pair. Plant–pollinator
interactions were sampled at seven sites of increasing distance from the glacier snout
across the Morteratsch glacier foreland, representing a chronosequence from 8 to 130
years since deglaciation. The numbers in the site pair column indicate the ages of the
two sites in years.