Skip to main content

Experimental shifts in exotic flowering phenology produce strong indirect effects on native plant reproductive success

Cite this dataset

Waters, Susan; Hille Ris Lambers, Janneke; Chen, Wei-Ling Cherry (2020). Experimental shifts in exotic flowering phenology produce strong indirect effects on native plant reproductive success [Dataset]. Dryad.


  1. By causing phenological shifts that vary among species, climate change is altering time envelopes for species interactions, often with unexpected demographic consequences. Indirect interactions, like apparent competition and apparent facilitation, are especially likely to change in duration because they involve multiple interactors, increasing the likelihood of asynchronous phenological shifts by at least one interactor. Thus, we might observe ecological surprises if intermediaries of indirectly interacting species change their mediating behavior.
  2. We explored this possibility in a plant-pollinator community that is likely to experience asynchronous phenological shifts. We advanced and delayed the flowering phenology of two ubiquitous exotic plants of western Washington prairies, Hypochaeris radicata and Cytisus scoparius, relative to seven native perennial forb species whose phenologies remained unmanipulated. These species interact indirectly through shared pollinators, whose foraging behavior influences plant reproductive success. We quantified impacts of experimental phenological shifts on seedset, pollinator visitation rates, and visiting pollinator composition relative to an unmanipulated control. We first verified that unmanipulated indirect interactions between native and exotic plants were strong, ranging from facilitative to competitive.
  3. Seedset of native plants was strongly affected by changes in exotic flowering phenology, but the magnitude and direction of effects were not predicted by the nature of the original indirect interaction (facilitative vs. neutral vs. competitive) or the change in interaction duration. The relationship between pollinator visitation and seedset changed for most species, though changes in pollinator visitation rate and pollinator composition were not as widespread as effects on native seedset.
  4. Synthesis. Changes in pollinator foraging behavior in response to changes in available floral resources are probably responsible for the unexpected effects we observed. Asynchronous phenological shifts have the potential to produce large and unexpected effects on reproductive success via indirect interactions.


Three datasets here include flowering phenology, pollinator visitation, and seed production data for prairie forbs native to western Washington, USA. Data were collected at Glacial Heritage Prairie in Littlerock, WA from July-August 2011. We experimentally altered flowering phenology for two exotic species (Cytisus scoparius and Hypochaeris radicata) and looked at effects on seedset of seven neighboring native forbs (Ranunculus occidentalis, Camassia quamash, Microseris laciniata, Lupinus lepidus, Eriophyllum lanatum, Prunella vulgaris, Campanula rotundifolia).

We established 48 5x5 m plots in a randomized block design, with eight blocks each containing one plot for each of six phenology treatments. Within blocks, plots were randomly located, with the constraint that they must contain at least 6 naturally occurring individuals of each of the native species. While other forb species were present in plots, they never occurred at high densities relative to focal species. Naturally occurring C. scoparius was absent from all plots, and naturally occurring H. radicata was present in all plots. We established plots to be at least 10 - 16 meters apart; plots are not separated by enough distance to represent non-overlapping pollinator communities.

            Control plots were unmanipulated, while flowering greenhouse-grown potted exotic plants were placed in experimental plots. Management actions had largely removed C. scoparius  from the surrounding plant community, so the potted plants represented a “reinvasion” for C. scoparius, while for H. radicata, treatments amplified an existing invasion. Pots were placed with the desired timing (‘Ambient’, ‘Advanced’ and ‘Delayed’), to impose phenological treatments (1 potted C. scoparius shrub = 152 ± 23 flowers added, or 4 pots of large H. radicata = 56 ± 4 capitulas added). There were thus two kinds of controls: exotic absent, and exotic present with “ambient” timing. The appropriate control is different depending on the species pair: for native species that had no initial flowering overlap with the neighboring exotic, the appropriate comparison is exotic absent (control) vs. ‘Advanced’ or ‘Delayed’ exotic phenology. By contrast, the appropriate comparison for plants that already overlapped with the exotic (before manipulation of exotic phenology) would be ‘Ambient’ exotic phenology vs. ‘Advanced’ or ‘Delayed’ exotic phenology.

Using potted plants allowed us to isolate pollinator-mediated effects from direct competitive/facilitative effects that occur belowground. Pots were typically at least ~0.75m away from focal native individuals, so shading was minimal. We altered exotic phenology by adding pots at the onset of the species’ natural flowering at nearby sites for Ambient treatments, ~3 weeks before ambient flowering for Advanced treatments, or ~3 weeks later than ambient flowering for Delayed treatments. We were unable to delay flowering of C. scoparius, so the Delayed treatments were used only for H. radicata. Note that in addition to manipulating presence/absence and timing of flowering, these treatments also affect the total floral density of the neighborhood.


We quantified seedset of 3 individuals of each focal native species within plots, representing three phenological stages to capture phenological spread in the native population: Early (first individuals of the species to bloom in plot), Peak (blooming during maximum conspecific floral density), and Late blooming (site-wide, >80% of conspecifics with mostly senescing flowers). We selected only plants that were estimated to be in “full bloom” (i.e., bearing multiple flowers/inflorescences), haphazardly chose one newly opening inflorescence on each of these individuals, and bagged the inflorescence when it had senesced to ensure seed capture. To quantify pollinator dependence, we also excluded pollinators on one additional individual of each native species using a breathable bag, and collected seed after senescence.  When counting seed from pollinator-excluded and naturally pollinated flowers, we also recorded evidence of predispersal seed predator activity that might reduce seed counts (insect presence or seed residues).

We observed pollinator visitation to each focal native species 3 times per plot per plant species during the bloom season: early, mid, and late. Observations were performed for fifteen minutes on sunny days > 15.5°C. Visiting insects were netted in sites once a week for three weeks and preserved in reference collections; specimens were classified into visually recognizable morphospecies, and further observations used these morphospecies designations without collecting specimens. We counted visits to the focal native individuals, identified each visitor, and recorded the number of conspecific floral units (flowers or inflorescences) within the 1x1m subplot where the focal plants were located to quantify floral patch size. Visits were counted only if there was contact between the visitor and flower stigmas or anthers. We surveyed floral richness and abundance biweekly in all plots as a covariate, by counting open inflorescences in 6 randomly chosen 1x1m subplots per 5x5m plot.

Usage notes

For all the .csv files:

phentrt=phenology treatment. NAT=control plot without any exotic forb added, ECYT=potted C. scoparius placed in plots ~2 weeks in advance of typical flowering peak, RCYT=potted C. scoparius placed in plots at typical flowering peak, EHYP=potted H. radicata placed in plots ~2 weeks in advance of typical flowering peak, RHYP=potted H. radicata placed in plots around typical flowering peak, LHYP=potted H. radicata placed in plots ~2 weeks later than typical flowering peak

block=randomized block. Site was divided into 8 blocks, A-H

plot=a 5x5 m plot; 1-7

species=six-letter codons representing the 7 native western Washington prairie forbs listed above: ACHMIL=Achillea millefolium; BROELE=Brodiaea elegans; CAMQUA=Camassia quamash; CAMROT=Campanula rotundifolia; DODHEN=Dodecatheon hendersonii; EMPTY=no species flowering in plot; ERILAN=Eriophyllum lanatum; FRAVIR=Fragaria virginiana; FRIAFF=Fritillaria affinis; HIECYN=Hieracium cynoglossoides; HYPPER=Hypericum perforatum; HYPRAD=Hypochaeris radicata; LEUVUL=Leucanthemum vulgare; LOMUTR=Lomatium utriculatum; LOTMIC=Lotus micranthus; LUPLEP=Lupinus lepidus; MICLAC=Microseris laciniata; PARVIS=Parentucellia viscosa; PRUVUL=Prunella vulgaris; RANOCC=Ranunculus occidentalis; SOLSPA=Solidago spathulata; TAROFF=Taraxacum officinale; TEENUD=Teesdalia nudicaulis; TRIDUB=Trifolium dubium; VICSAT=Vicia sativa; VIOADU=Viola adunca; ZIGVEN=Zigadenus venenosus.

phenstg=a qualitative category for when the focal native forb individual flowered relative to others in the population: EARLY, PEAK, LATE, or LL ("latelate").


For seed data:

polltrt=pollination treatnent imposed: either SELF (pollinators excluded with bag) or OPEN (no manipulation, pollinators have access to inflorescence)

totheads=total number of seedheads/pods/capitulas on plant

filledheads=number with filled seeds

FILLEDseednum=number of seeds filled (i.e., successfully pollinated) in the seedhead counted

UNFILLEDseednum=number of seeds unfilled in the seedhead counted

FILLEDseedwt=weight of filled seeds in the seedhead counted

UNFILLEDseedwt=weight of unfilled seeds in the seedhead counted

plwt=aboveground dry biomass of plant from which seed was harvested and counted

infested=evidence of seed predation (insect frass, seed fragments): binary variable

seedperhead=number of filled seeds counted


For phenology data:

phenobsdate=date flowering phenology was observed

phenobsjulianday=julian date flowering phenology was observed

obsperiod=segment of season during which observations took place 1-7

square=1-25. One of 25 1x1m squares making up the 5x5m plot.

species_counted= a six-letter codon representing the identity of forb species flowering in the plot at the time of survey

blooms=# flowering stems counted for that species


For pollinator visitation data:

patch_size: number of flowering units of the focal native forb species present in a 1x1m square chosen for observation

pollobsdate: date of observation for pollinator visitation

pollobsjuliandate: julian date of observation for pollinator visitation

species_watched: focal native forb species observed for visitation

rhsp=Rhamphomyia sp.

drsp2=Drymeia sp. 2

anthsp1=Anthomyiid sp.

unfly1=k=unknown fly 1

drsp1=Drymeia sp. 1

erhi=Erystalis hirsuta

boca=Bombus californicus

bovo=Bombus vosnesenskii

ossp1=Osmia sp.

bomi=Bombus mixtus

hese=Hemihalictus series

hasp=Halictid sp.

andsp1=Andrenid sp.

apme=Apis mellifera

nomsp1=Nomada sp.

duca=Dufourea campanulae

unfly2=unknown fly 2

unbee1=unknown bee 1

unwasp1=unknown wasp 1

beetle1=unknown beetle 1

boma=Bombus melanopygus

snakefly=unknown snakefly

unbee2=unknown bee 2

other= other taxon not identified, e.g. true bug, earwig, etc.


National Science Foundation, Award: DGE-0718124

Mary Gates Undergraduate Research Fellowship , Award: NA

Mary Gates Undergraduate Research Fellowship