UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Insect pollination and experimental warming in the High Arctic Robinson, Samuel Victor Joseph 2014

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


24-ubc_2014_spring_robinson_samuel.pdf [ 6.25MB ]
JSON: 24-1.0167419.json
JSON-LD: 24-1.0167419-ld.json
RDF/XML (Pretty): 24-1.0167419-rdf.xml
RDF/JSON: 24-1.0167419-rdf.json
Turtle: 24-1.0167419-turtle.txt
N-Triples: 24-1.0167419-rdf-ntriples.txt
Original Record: 24-1.0167419-source.json
Full Text

Full Text

Insect Pollination and ExperimentalWarming in the High ArcticbySamuel Victor Joseph RobinsonB.Sc., The University of British Columbia, 2006A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF SCIENCEinThe Faculty of Graduate and Postdoctoral Studies(Geography)THE UNIVERSITY OF BRITISH COLUMBIA(Vancouver)April 2014c© Samuel Victor Joseph Robinson 2014AbstractAs climate change causes retreats in Arctic glaciers, it is important to un-derstand the mechanics of growth and community change in Arctic plantcommunities. Arctic plants have been shown to respond to observed and ex-perimental changes in temperature by altering their reproductive strategies,growth, and phenology. Researchers have used open-top chambers (OTCs)to experimentally alter the near-surface air temperatures of tundra plantcommunities over long periods of time, but these devices may exclude in-sect pollinators to flowers during crucial periods of pollen reception. Insectpollination in the context of OTCs and Arctic plants is therefore impor-tant to understand, but has been poorly researched. I altered pollination ofSalix arctica, Dryas integrifolia, and Papaver radicatum inside and outsideof OTCs in a High Arctic shrub community, and conducted targeted insectnetting to understand the dynamics of the visiting insect community. I alsoconducted bowl trapping inside and outside of OTCs to gauge their effect oninsect availability to receptive flowers. OTCs altered the timing of floweringin Arctic plants, and significantly reduced the availability of pollinators toavailable flowers. However, I found that while both warming and pollina-tion can alter flower and seed production in the study species, pollinationis largely independent of OTC warming. Early-flowering species have thepotential to be most affected by OTC-induced insect exclusion. The mostcommon visiting insects were flies of the families Syrphidae and Muscidae,with occasional bumblebees (Bombus polaris). Papaver radicatum was byfar the most heavily-visited flower, and I showed that the Syrphidae visit theflower preferentially at low temperatures, likely for warmth as well as pollen.I discuss these results in context with the current literature on Arctic plantand insect communities, and make recommendations for future research.iiPreface• Chapters 2 and 3 are based on work conducted by Samuel Robinsonand supervised by Dr. Greg Henry. I was responsible for field work,lab work, statistics, and writing.• No publications have yet arisen from this.• No approval was required from the UBC Research Ethics Board.iiiTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . ivList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiAcknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . viiiDedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 History of Pollination Ecology . . . . . . . . . . . . . . . . . 11.1.1 Pollination in General . . . . . . . . . . . . . . . . . . 11.1.2 Studies in Arctic Pollination Biology . . . . . . . . . 21.2 Plant-Pollinator Interactions . . . . . . . . . . . . . . . . . . 21.2.1 Benefits for Plants . . . . . . . . . . . . . . . . . . . . 21.2.2 Benefits for Insects . . . . . . . . . . . . . . . . . . . 31.3 High Arctic Ecosystems and Climate Change . . . . . . . . . 31.3.1 The ITEX Program . . . . . . . . . . . . . . . . . . . 31.4 Review of Arctic Pollination and Insect Studies . . . . . . . 41.4.1 Confounding Effects of OTCs . . . . . . . . . . . . . 41.4.2 Insect Community Composition . . . . . . . . . . . . 61.4.3 The Effects of Seasonality and Temperature . . . . . 61.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71.6 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 Pollen Limitation in Open-Top Chambers . . . . . . . . . . 92.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 92.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11ivTable of Contents2.2.1 Site Description . . . . . . . . . . . . . . . . . . . . . 112.2.2 Plant Species . . . . . . . . . . . . . . . . . . . . . . . 112.2.3 Experimental Design . . . . . . . . . . . . . . . . . . 122.2.4 Statistical Analyses . . . . . . . . . . . . . . . . . . . 142.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162.3.1 Flower Response . . . . . . . . . . . . . . . . . . . . . 162.3.2 Seed Response . . . . . . . . . . . . . . . . . . . . . . 172.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182.4.1 Salix arctica . . . . . . . . . . . . . . . . . . . . . . . 182.4.2 Dryas integrifolia . . . . . . . . . . . . . . . . . . . . 192.4.3 Papaver radicatum . . . . . . . . . . . . . . . . . . . 202.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222.6 Figures and Tables . . . . . . . . . . . . . . . . . . . . . . . . 243 Effects of Open-Top Chambers on Flowering Patterns andPotential Insect Visitation . . . . . . . . . . . . . . . . . . . . 393.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 393.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413.2.1 Flowering Community . . . . . . . . . . . . . . . . . . 413.2.2 Insect Community . . . . . . . . . . . . . . . . . . . . 413.2.3 Statistical Analyses . . . . . . . . . . . . . . . . . . . 423.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 443.3.1 Flower Availability . . . . . . . . . . . . . . . . . . . 443.3.2 Insect Community . . . . . . . . . . . . . . . . . . . . 453.3.3 Interaction Rates . . . . . . . . . . . . . . . . . . . . 463.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473.4.1 Flower Availability . . . . . . . . . . . . . . . . . . . 473.4.2 Insect Community . . . . . . . . . . . . . . . . . . . . 493.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 543.6 Figures and Tables . . . . . . . . . . . . . . . . . . . . . . . . 564 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72AppendixA Model Fitting . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90vList of Tables1.1 Summary of studies . . . . . . . . . . . . . . . . . . . . . . . 52.1 Experimental Design . . . . . . . . . . . . . . . . . . . . . . . 372.2 Flower Response . . . . . . . . . . . . . . . . . . . . . . . . . 382.3 Seed Response . . . . . . . . . . . . . . . . . . . . . . . . . . 383.1 Tests for flower response to warming . . . . . . . . . . . . . . 653.2 Differences in OTC and contol catches . . . . . . . . . . . . . 663.3 Ranking of Visitors . . . . . . . . . . . . . . . . . . . . . . . . 663.4 Total flower-insect interactions . . . . . . . . . . . . . . . . . 67A.1 Model Parameters . . . . . . . . . . . . . . . . . . . . . . . . 90A.2 Flower Model Selection . . . . . . . . . . . . . . . . . . . . . 91A.3 Seed Model Selection: Part 1 . . . . . . . . . . . . . . . . . . 92A.4 Seed Model Selection: Part 2 . . . . . . . . . . . . . . . . . . 93viList of Figures2.1 Location Map . . . . . . . . . . . . . . . . . . . . . . . . . . . 252.2 Site Photo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262.3 Salix arctica (Pall) . . . . . . . . . . . . . . . . . . . . . . . . 272.4 Dryas integrifolia (Vahl) . . . . . . . . . . . . . . . . . . . . . 282.5 Papaver radicatum (L.) . . . . . . . . . . . . . . . . . . . . . 292.6 Flower output from hand-pollinated and warmed plants . . . 302.7 Flower output from pollen-manipulated plants outside warm-ing treatments . . . . . . . . . . . . . . . . . . . . . . . . . . 312.8 Flower output from warmed and control D. integrifolia . . . . 322.9 Seeds per flower and seed mass per flower from hand-pollinatedand warmed plants . . . . . . . . . . . . . . . . . . . . . . . . 332.10 Seeds per flower and seed mass per flower from pollen-manipulatedplants outside warming treatments . . . . . . . . . . . . . . . 342.11 Average seed mass and germinability from hand-pollinatedand warmed plants . . . . . . . . . . . . . . . . . . . . . . . . 352.12 Average seed mass and germinability from pollen-manipulatedplants outside warming treatments . . . . . . . . . . . . . . . 363.1 Stellaria longipes Goldie . . . . . . . . . . . . . . . . . . . . . 563.2 Ordination of OTC and control flower communities . . . . . . 573.3 Flower densities . . . . . . . . . . . . . . . . . . . . . . . . . . 583.4 Ordination of OTC and control insect communities . . . . . . 593.5 Bowl trap catches (Time Series) . . . . . . . . . . . . . . . . . 603.6 Bowl trap catches (Total) . . . . . . . . . . . . . . . . . . . . 613.7 Net catches (Time Series) . . . . . . . . . . . . . . . . . . . . 623.8 Net catches (Total) . . . . . . . . . . . . . . . . . . . . . . . . 633.9 Factors influencing Papaver radicatum interaction rate . . . . 64viiAcknowledgementsThis work was funded by the National Science and Engineering ResearchCouncil of Canada (NSERC), ArcticNet, Polar Continental Shelf Program(PCSP), the Northern Scientific Training Program (NSTP), the CanadianInternational Polar Year (IPY - CiCAT), and the University of BritishColumbia. Permission for research was granted from the Nunavut Depart-ment of Environment, and the use of the buildings at Alexandra Fiord wasgranted by the Royal Canadian Mounted Police (RCMP).I foremost wish to thank Dr. Greg Henry for his help, advice, and encour-agement during the last three years. Thank you to all of my field assistantswho have helped me during the summers of field work, notably Christo-pher Greyson-Gaito, Doug Curley, Darcy McNicholl, Meagan Grabowski,and Matt Huntley. I thank Anne Bjorkmann for her statistical help andpractical advice. Thank you to Fred Stride, for letting me continue to playin UBC Jazz Ensemble I throughout the past three years, and exercise acompletely different part of my brain!I thank my committee members, Dr. Elizabeth Elle and Dr. Roy Turk-ington, for their support and direction. I also thank Dr. Marwan Hassanfor his support and encouragement.Finally, I wish to thank my parents and my brothers for all of their loveand support during my M.Sc. I can never thank you enough for how you’vesupported me.viiiDedicationThis work is dedicated to the memory of the people killed during the crashof First Air Flight 6560 on August 20, 2011, in Resolute Bay, Nunavut.ixChapter 1IntroductionPollination of flowering plants by insects (entomophily) in Arctic ecosystemsis poorly understood. Visitation of flowers is beneficial for both the insectsthat visit them, as well as the plants, and is essential for fertilization andproduction of seeds in many plants. Climate change has already begun toalter plant community composition in the Arctic, and will continue to alter itinto the next centuries. Understanding the speed and extent of these changespartially depends on their seed production. However, very few studies havebeen conducted that have identified changes in insect communities that theplants are associated with. In this literature review, I will examine theresearch that has currently been conducted on Arctic pollination biology,and the study of climate change effects on plant and insect communities.Finally, I will outline my specific research goals.1.1 History of Pollination Ecology1.1.1 Pollination in GeneralPollination of plants has been examined since antiquity by natural philoso-phers such as Virgil (and certainly as much by agrarians), but Rudolf Cam-erarius (1665-1721) is generally credited with the first studies of the sexu-ality of plants (Faegri and Van der Pijl, 1979). However, Camararius onlyexamined self-fertilization, and Joseph Ko¨lruter (1733-1806) and ChristianSprengel (1750-1816) were the first to examine pollination of flowers by in-sects (Proctor et al., 1996), and found that many plants required insectpollinators to obtain full seed set (fully-formed, fertile seeds). Charles Dar-wins work built on the concepts laid down by Ko¨lruter and Sprengel in hisobservations of numerous floral structures and pollination techniques, andsuggested a mechanism (natural selection) by which the structure of flowerscould ultimately be controlled by their interactions with pollinators (Darwin,1859, 1862). His work also lead to the general thesis that “nature abhors self-fertilization” (Waser and Ollerton, 2006). This hypothesis (“Darwins Law”)continued to dominate literature on plant-pollinator interactions until the11.2. Plant-Pollinator Interactionsearly 1900s.1.1.2 Studies in Arctic Pollination BiologyAfter Darwins publication of Species and Orchids, naturalists in the HighArctic such as Aurivillius noted that flowers such as Pedicularis lanata andP. hirsute appeared to be completely self-pollinating in stark contrast to“Darwins Law (Kevan, 1973a). Other authors continued this trend into the20th century (Warming, 1888; Mathiesen, 1921), and the hypothesis thatstressed environments, such as the High Arctic, favoured self-fertilizationcontinued. This was generally accepted until the work of Chernov (1966,1985) and Hocking (1968) revealed that insect pollination was far more com-mon than previously thought. Kevan (1972b) examined this in even greaterdetail by using acetate cones and netting to exclude pollinators, demonstrat-ing dependence (and independence) of several pan-Arctic flowering plants oninsect visitation. In recent years, plant-pollinator interactions in the HighArctic have seen an even greater interest in the scientific community, inthe face of global climate change and an increased interest in networks and“connectivity in species interactions (Elberling and Olesen, 1999; Lundgrenand Olesen, 2005; Hegland et al., 2009).1.2 Plant-Pollinator Interactions1.2.1 Benefits for PlantsIt is well known that plants benefit from insect visitation. Visitation offlowers by insects, particularly bees and flies, has been shown to be essen-tial to fertilization and fruit formation in many species. Many families ofplants have self-incompatibility mechanisms that do not allow pollen fromthe same plant to fertilize itself (autogamy, or selfing). Transfer of pollenbetween individuals (heterogamy or outcrossing) can be facilitated either bywind or by pollinating animals such as insects. This outcrossing increasesgenetic variation within populations, meta-populations, and contributes toongoing process of natural selection by allowing gene transfer. This is alsoimportant in the development of resilient genotypes which can withstandenvironmental change, and to avoiding the pairing of deleterious alleles (in-breeding) (Proctor et al., 1996). Pollinating insects are often a much moreefficient mechanism because they direct pollen transfer directly to flowers,often to individuals of the same species, thus reducing the amount of pollenrequired by a plant to produce a successful mating. This has been found21.3. High Arctic Ecosystems and Climate Changeto be essential in pollen-limited Arctic plants such as Saxifraga oppositifolia(Stenstro¨m and Molau, 1992; Gugerli, 1998).1.2.2 Benefits for InsectsPlants are not the only beneficiary in this relationship. It has been arguedthat the pollen would have served as an attractant to the first pollinating in-sects, because it pre-dates angiosperm plants. It is one of the most nutrient-rich parts of a plant, and is commonly used by bees and other insects to reartheir broods (Simpson et al., 1983). However, since it is energetically expen-sive to produce, many orders of plants have evolved nectaries to serve as analternative attractant. The nectar produced by flowers represents a majorsource of energy for flower-visiting insects. Ko¨lruter was arguably the firstto study its significance (and to discover that this was the source of honey)(Proctor et al., 1996), followed by (Darwin, 1862). More recently, Hock-ing (1968) compared nectar concentrations between the high and low Arctic(Churchill, NWT and Lake Hazen, NU) and found that northern flowersproduce smaller quantities of high-concentration nectar. He also found thatunlike more southern ecosystems, flowers compete for insect visitors ratherthan insects competing for nectar. Energetics also play a significant rolein the plant-pollinator relationship. Heinrich (1983) writes about the rolethat optimal foraging plays in nectar and pollen feeding, and also how shel-ter by the flower contributes to the energy balance of pollinators. Hockingand Sharpin (1965) as well as Kevan (1975) show the corolla of heliotropicflowers such as Dryas integrifolia and Papaver radicatum act as parabolicreflectors, and suggest that added heat from this is an added attractant topollinating insects.1.3 High Arctic Ecosystems and Climate Change1.3.1 The ITEX ProgramGlobal climate change has been shown to have altered air temperatures andprecipitation patterns over the past twenty years (Hinzman et al., 2005;Alexeev et al., 2012), and is predicted to have even larger effects on meantemperatures in Arctic regions over the next twenty years (Hassol, 2004;Stocker et al., 2013). This will have large effects on the structure and func-tion of future plant communities (Walker et al., 2006), and has already beenshown to be occurring (Hudson and Henry, 2009; Hill and Henry, 2011; El-mendorf et al., 2012a). It is well known that both insects (Downes, 1964;31.4. Review of Arctic Pollination and Insect StudiesDanks et al., 1994) and plants (Chapin, 1983; Jones et al., 1997; Kladyet al., 2011) can dramatically change their behaviour and growth underhigher temperatures. The International Tundra Experiment (ITEX) wascreated to monitor the changes in plant phenology, growth and reproductionacross the Arctic resulting from passive warming using Open-Top Chambers(OTCs) (Marion et al., 1997; Henry and Molau, 1997). Part of the ITEXprogram also involves monitoring populations of larval Gynaephora groen-landica and G. rossii (Mølgaard and Morewood, 1996), but unlike the plantcommunities, no long term sampling scheme has been undertaken for otherland-dwelling arthropods. Insects are also key indicators of environmen-tal change (Danks, 1992), and as their populations should be expected tochange with a warming Arctic, so should their interactions (both negativeand positive) with Arctic plants.1.4 Review of Arctic Pollination and InsectStudiesTable 1.1 provides a summary of some of the major works in Arctic pollina-tion ecology to date. Below, I will briefly review the most recent work, andidentify knowledge gaps that this work will operate in.1.4.1 Confounding Effects of OTCsResearchers have shown that OTCs can alter seed production in both highand low Arctic sites for Dryas octopetala (Alatalo and Totland, 1997; Welkeret al., 1997). Stenstro¨m et al. (1997) noted latitudinal differences in theeffect of experimental warming on seed set in Saxifraga oppositifolia, andsuggest that warming by OTCs may reduce pollinator activity. Jones et al.(1997) also note that reduced seed set in OTC-warmed Salix arctica may bedue to either reduced wind or insect pollination. However, the confoundingeffects of the OTCs on seed set are not well understood, and the confoundingeffect may alter predictions made for seed production in flowering plants ina warming High Arctic.To quantify changes, observed pollination rates within the OTCs shouldbe compared with control plots, but rates of background pollination arenot well known across high Arctic plants. Bocher (1996) suggested thatunderstanding this was necessary for the ITEX experiment to understandthe degree of wind pollination and insect pollination that occurs within theOTCs. It has not been demonstrated that OTCs directly create a pollina-41.4. Review of Arctic Pollination and Insect StudiesTable 1.1: Summaries of some key studies that have contributed to themodern study of Arctic pollination ecologyAuthor Year Location FindingsY. Chernov 1963 YugorskiPeninsula,RussiaFound that many flowering plants dependon insect pollination for seed-setB. Hocking 1968 Lake Hazen,NunavutExamined differences in pollination strate-gies between low and high Arctic floweringplants; found that high Arctic pollinationnetworks tend to be pollinator-limitedL. Bliss 1971 Various Reviewed all literature on plant reproduc-tion in tundra ecosystems; described thecharacteristics of plant reproductive sys-temsP. Kevan 1972 Lake Hazen,NunavutExamined dependence on pollination in 13Arctic flower plants; identified pollinationdeficits present as well as dependence oninsect pollination in 5 speciesH. Elberling &J. Olesen1999 Latnajaurje,SwedenExamined the composition of an Arcticpollinating community; identified the pre-dominance of Diptera as pollinatorsR. Ring 2001 AlexandraFiord,NunavutExamined abundance of insects inside andoutside of OTCs in the context with theITEX program; found differences in abun-dance in some insect familiesR. Lundgren &J. Olesen2005 Uummannaq,GreenlandConducted a network study on an Arcticpollinating community; found that insectand plant species were very closely linked51.4. Review of Arctic Pollination and Insect Studiestion deficit, but it is implied any pollination deficit created is due to alteringflowering phenology in relation to emergence of pollinating insects (Alataloand Totland, 1997). Bliss (1958) showed historical rates of seed germinationwithin Arctic plants, and other authors have shown that OTCs can changeseed germination rates (Wookey et al., 1995; Klady et al., 2011). The mech-anism for this is not understood, but has been suggested to be a combinationof temperature alteration and changes in pollination rates.1.4.2 Insect Community CompositionRing (2001) completed pit-trap surveys that indicate lower numbers of in-sects are trapped in OTCs than in control plots, and has shown how thisvaries between taxa. However, actual visitation to flowers by insect taxa isnot known, so it is difficult to say whether or not all pollinating taxa areexcluded from OTCs (or attracted to them). Also, the temporal scale ofmeasurement (collections were made weekly) was not appropriate for thescale of the ecosystem, which can undergo large changes in a phenology inonly a matter of days. Identifying visitors to Family (or Genus) will help toassign degrees of pollination-importance to different taxa of insects, and ul-timately help to better understand rates of seed set and plant establishmentwithin OTCs. Some of this work has already been done at the ITEX site atAlexandra Fiord, Nunavut, and indicates that Hoverflies (Order: Diptera,Family: Syrphidae) may be responsible for a majority of the pollination inthe High Arctic (Robinson, 2011), which was also observed by both Hocking(1968) and Kevan (1972b) at Lake Hazen Camp, Nunavut.1.4.3 The Effects of Seasonality and TemperatureLevels of seed set have been found to be higher in late-blooming alpine flow-ers due to higher numbers of pollinators (Kudo, 1993). Earlier snowmelts,and hence earlier flowering times, in the High Arctic may have a similareffect on seed set, but this depends on the synchrony between insect emer-gence and flower emergence. Besides snowmelt timing, other environmentalvariables such as wind speed, cloud cover, and ambient temperature can con-trol day-to-day variation of insect visitation in alpine zones (Kudo, 1993),and operates similarly in Arctic zones (Totland, 1994; Bergman et al., 1996).Typically, pollinating insects minimize their energy consumption by only vis-iting under energetically favourable conditions (Heinrich, 1975, 1983), whichis usually visible in a diurnal cycle in mid-latitude ecosystems, but may notexist in the High Arctic where insects may be more opportunistic with re-61.5. Conclusionsgards to visitation. It has also been noted that smaller insects are moreaffected by changes in temperature and insolation (Gilbert, 1985), and maybe “released from summer limitations on feeding and growth as the HighArctic undergoes warming. Bergman et al. (1996) found that activities ofbumblebees and butterflies were highly correlated with air temperature andincoming radiation, and that seed-set in bumblebee-pollinated plants wasreduced during colder periods due to lack of visitation.1.5 ConclusionsOther than the work of Bergman et al. (1996), almost nowhere else in theArctic ecological literature has such a direct connection been made betweenweather, insects, and seed production in flowering plants. The shortness ofthe flowering season makes these connections very important for both flow-ering plants and theirassociated pollinators. There are few studies examin-ing the diversity and numbers of insects during the flowering season (Ring,2001), but few examining pollinating insects (Høye et al., 2013). Changesin pollination have been inferred in some species (Chapin, 1983; Stenstro¨met al., 1997; Jones et al., 1997), but very few studies have explicitly deter-mined levels of pollination (Kevan, 1972b), and none have examined pol-lination in the context of the ITEX program. Finally, long-term changesin the plant community have been observed (Walker et al., 2006; Hudsonand Henry, 2009), but only recently has the importance of their respectivepollination strategies been realized (Lundgren and Olesen, 2005). Insectcommunities may exhibit unpredictable responses because of individualisticchanges in behaviour due to rising temperatures and lengthening growingseasons (Danks, 2004), highlighting the need for research that emphasizeshow insect communities have changed along with plant communities. Insectcommunities, in general, are poorly sampled across the Canadian Arctic,and while changes in other ecosystem components (such as sea ice and snowmelt) have been documented, changes in insect populations have not beenstudied.1.6 ObjectivesThe work I have undertaken has multiple goals. Briefly:• it aims to understand how OTCs interact with rates of pollination, inorder to understand how reproductive changes within OTCs representfuture warmed ecosystems.71.6. Objectives• it aims to understand the specific nature of a High Arctic insect com-munity, with reference to insect taxonomy, seasonal timing, and itsstructure over the flowering season. This includes ambient bowl trap-ping and targeted netting, which will help to gauge the importance ofspecific groups of flower visitors.• it aims to provide more accurate estimates of insect availability insideand outside of OTCs, and relate those to the dominant visitors foundin the surrounding community.These goals will provide a more accurate picture of how OTCs affect theflying insect community, the pollination regimes of Arctic plants, and howthose two factors interact in the context of global change experiments.8Chapter 2Pollen Limitation inOpen-Top Chambers2.1 IntroductionDuring the past decades, northern residents and the scientific communityhave witnessed large changes in Arctic plant communities (Hinzman et al.,2005; Elmendorf et al., 2012a). They have seen expansion into previouslyunoccupied areas (Tape et al., 2006; Morton et al., 2012), as well as changesin dominance and abundance within monitored communities (Elmendorfet al., 2012a). Climate change is predicted to bring greater changes to Arc-tic regions compared to mid-latitude regions, with mean air temperaturesincreasing 7-9◦ C by 2100, along with an increase in rainfall (Stocker et al.,2013). Many of these changes are already occurring (Graversen et al., 2008;Serreze et al., 2009; Alexeev et al., 2012). Plant community responses toclimate change in the Arctic are not well understood, and vary highly de-pending on the region (Elmendorf et al., 2012b; Brochmann et al., 2013).Arctic plant communities can influence feedbacks to climate change (Wookeyet al., 2009), such as CO2 flux from the soil (Oechel et al., 2000; Lund et al.,2012) and the near-surface radiation budget (Blok et al., 2010; Myers-Smithet al., 2011).Climate change in the High Arctic is predicted to be even more severe(Hassol, 2004; Stocker et al., 2013). Because of their isolation, sparse plantcover, and short growing season (8-12 weeks), High Arctic plant commu-nities are studied far less than other ecosystems, and represent an area oflarge future changes and great uncertainty. For example, early snowmeltmay result in an increased length of growing season, but only if mid-seasondroughts are offset by rain.Since its inception in 1990, members of the International Tundra Ex-periment (ITEX) have studied the effects of climate warming on Arctic andalpine plant communities. The main experimental technique used by theITEX is open-top chambers (OTC), which rely on a passive heating ef-92.1. Introductionfect by trapping incoming solar radiation and limiting convective heat loss(Marion et al., 1997; Hollister and Webber, 2000). OTCs increase the near-surface temperature over the snow-free season, and have been shown to drivechanges in plant communities over a wide range of latitudes (Walker et al.,1999; Klanderud and Totland, 2005; Walker et al., 2006; Hudson and Henry,2010; Klady et al., 2011). OTCs have also been shown to change the repro-ductive effort (flower numbers, seed numbers per flower) of Arctic plants(Stenstro¨m et al., 1997; Klady et al., 2011). These changes have usuallybeen attributed to the effect of increased air temperature, but some authorshave empirically found that OTCs decrease natural insect and wind pollina-tion (Molau and Shaver, 1997; Stenstro¨m et al., 1997; Hollister and Webber,2000).Asexual reproduction plays a major role in the maintenance of Arcticplant communities (Billings and Mooney, 1968; Bell and Bliss, 1980), butdispersal into unvegetated areas is accomplished by sexual reproduction andseed dispersal. As glaciers recede due to prolonged warming, the processby which plants make and disperse their seeds is important to understand(Stenstro¨m et al., 1997; Bjorkman et al., 2013). Most Arctic plants havefairly robust pollination systems and are capable of producing at least someviable seeds by self-pollination. However, most of these plants increase seed-set when hand-pollinated or exposed to insect visitors, indicating that insectpollination is beneficial to plants (Stenstro¨m and Molau, 1992; Molau, 1993;Totland and Eide, 1999). The relative importance of wind, insect, and self-pollination in sexual reproduction are not well studied in the High Arctic.In light of this, it is important to understand how pollination influences seedproduction and viability in Arctic plants, and hence, their ability to colonizenew areas.The understanding of the mechanisms behind Arctic plant reproductionis at this point, mainly speculative. There are relatively few studies in-vestigating the flower and seed production of Arctic plants under differentpollination regimes, and even fewer examining how these regimes interactwith warming. There are equally few studies examining how OTC warmingand pollination interact with seed viability. Most studies use passive meth-ods such as seed size, but few conduct actual germination trials because ofthe tedious nature of the experiment. To make accurate predictions aboutthe state of plant reproduction and spread, it is necessary to quantify howthe effect of warming interacts with any induced changes in the pollinationregime. In this study, I investigate how pollination changes interact withOTC warming in three Arctic flowering plant species.102.2. Methods2.2 Methods2.2.1 Site DescriptionThe study was conducted at Alexandra Fiord, Ellesmere Island, Nunavut, inthe Canadian High Arctic (Map in Figure 2.1). Alexandra Fiord is describedas an “oasis in a polar desert” (Svoboda and Freedman, 1994a), and there areITEX experimental sites located at seven distinct community types withinthe valley (Svoboda and Freedman, 1994b). For this study, I used the XericShrub site, which is characterized by early snowmelt, peaty and sandy soils,and a deep active layer(Muc et al., 1989; Svoboda and Freedman, 1994a;Jones et al., 1997, 1999). Salix arctica is the most dominant plant at thesite, followed by graminoids such as Poa arctica and Festuca brachyphylla.2.2.2 Plant SpeciesArctic Willow (Salix arctica (Pall)) is a woody, prostrate shrub present inthe circumpolar tundra, and can be found across a large range of latitudesand ecological conditions (Dawson and Bliss, 1989b; Jones et al., 1997).Individual plants are dioecious and can exhibit differing growth patternsbetween sexes (Dawson and Bliss, 1989a; Jones et al., 1999). Inflorescencesare dull red vertical catkins, with nectaries present on both male and fe-male catkins. Catkins can have hundreds of individual flowers, and becauseflowers can produce multiple seeds a single female catkin can produce wellover 500 seeds (this work). In the High Arctic S. arctica set their nextyear’s flower buds at the end of the snow-free season (late July-August),and bud break commences a few days after snow melt the following year.Along with Saxifraga oppositifolia, they are some of the earliest floweringplants at Alexandra Fiord (Kevan, 1990). Warming has been shown to ad-vance their flowering phenology and their seed development (Jones, 1995;Jones et al., 1999). Both male and female plants produce nectar, and maleplants also produce copious amounts of bright yellow pollen. The polli-nation strategies of other species in the genus Salix tend to be a mixturebetween wind and insect pollination (ambophily) (Tamura and Kudo, 2000;Totland and Sottocornola, 2001; Culley et al., 2002). Salix arctica is ableto be wind-pollinated, but requires insect visitation for full seed-set (Kevan,1972b; Peeters and Totland, 1999).Mountain Avens (Dryas integrifolia (Vahl)) is a semi-evergreen dwarfshrub forming prostrate cushions, present in northern North America andGreenland (Porsild and Cody, 1980). It is closely related to its widespreadsister species, Dryas octopetala, and there is some evidence of hybridization112.2. Methodsbetween the two species (Philipp and Siegismund, 2003). Like S. arctica, itproduces flower buds during the autumn of the previous flowering season,but bud burst occurs later than S. arctica (Philipp et al., 1990). It produceswhite, 8-11 petalled perfect (bisexual) flowers with deep nectaries. Smallproportions of male-only or female-only (unisexual) flowers have been notedby some workers (Molau and Mølgaard, 1996; Wada and Kanda, 2000).Some of the recorded unisexual flowers may be also due to flower herbivoryby noctuid caterpillars such as Sympistis nigrita (Greg Henry, pers. comm.).The flowers are often visited by insects, and (Kevan, 1972b) has shown itto be dependent on insect pollination for full seed set. The flowers of D.integrifolia are heliotropic, and temperatures can be elevated in the flowercorolla by 6-8◦C, which can benefit both reproductive development withinthe flower as well as insects that bask or feed within the corolla (Hockingand Sharpin, 1965; Kevan, 1975; Krannitz, 1996). I used hand-held thermo-couples to test temperature differences.Arctic Poppy (Papaver radicatum (L.)) is an annual flowering forb, presentacross the circumpolar tundra and alpine zones. It produces yellow, cup-shaped perfect flowers in which the female parts are fused into a capsulewhich swells after fertilization. Warming has been shown to increase bothplant biomass and flower production in P. radicatum, as well as acceler-ate flowering phenology (Le´vesque et al., 1997; Mølgaard and Christensen,1997). These changes are thought to be caused by an increased lengthof growing season rather than elevated temperature (Mølgaard and Chris-tensen, 1997). Kevan (1975) found that air temperatures within the corollacould be elevated by approximately 7◦C compared to ambient temperatures,but I found that temperatures were elevated by an average to 5◦C.2.2.3 Experimental DesignTo test how OTCs interact with pollination, I manipulated pollen levelsinside and outside of OTCs using the three aforementioned plant species.Treatment factors included Warming (2 levels: Warmed, Control) and Pol-lination (4 levels: Hand-pollinated, Net Excluder, Cloth Excluder, Control;explained below), with 10 replicates within each combination of factors. Itagged individual plants in both OTC and Control treatments, then ran-domly assigned Pollination treatment to them. The population size of flow-ering plants within OTCs at the Xeric Shrub site was limited, I could notconduct a 4-by-2 complete cross with sufficient replication, so I removedsome levels of the experiment (Table 2.1). Because finding sufficient flower-ing plants outside of OTCs was not an issue, I was able to use all 4 levels of122.2. Methodspollination (Hand pollinated, Net Excluder, Cloth Excluder, Control) out-side the Warming treatments. The final structure of the experimental designis shown in Table 2.1.Hand-pollination was conducted by taking pollen from the anthers of thesame species of flowering plants using a small, soft paintbrush and transfer-ring it to the receptive pistils of fully emerged flowers on treatment plants.I made an attempt to use pollen from plants both within and outside ofOTCs randomly where possible, but this was sometimes limited becauseof advanced flowering phenology within OTCs or site characteristics. NetExcluders were used to exclude pollinators from individual flowers, using a1mm nylon mesh attached using thin twist-ties. This mesh was intended toexclude pollinators without causing large changes to any wind-borne pollenload. Cloth Excluders were used to gauge the importance of self-pollination,using bags of 10µm cloth to exclude any wind-borne pollen from D. integri-folia and P. radicatum flowers. All Excluder bags were attached before theflowers were receptive to pollen, and removed once the flowers had senescedand were dispersing seeds. Cloth Excluders were not used on S. arctica be-cause the species is dioecious and cannot be self-fertilized. Rain may havecreated some problems later in the season by washing pollen from late-seasonflowers (mainly P. radicatum), but this was confined to only a few days inlate July to early August when most flowers were beyond pollen receptivityand had begun to senesce.OTCs are hexagonal, plexiglass chambers approximately 1.5m across,designed to contain a 1-m2 plot, and have been found to raise the ambientair temperature by 1-3◦C. Molau and Mølgaard (1996) and Jones (1995)describe the use of OTCs used by the ITEX group, including constructionand design of the chambers, and the efficacy of OTCs has been evaluatedby Marion et al. (1997). In general, OTCs increase the near-surface airtemperature during sunny days, and cause the date of snowmelt to advance.Hollister and Webber (2000) found that OTCs mimic an above-average warmsummer, and during the summer of 2012, I found that the temperaturewas elevated by 1.79◦C within the OTCs. During the summer of 2012,the temperature was elevated by 1.79◦C. Ten OTC/control plot pairs wererandomly assigned in 1992 when the treatments began, and four extra pairswere later added for monitoring CO2 flux from the tundra.To control for plant size, I sampled random plants from the site, collectedsize metrics (described in Equation 2.1,2.2 and 2.3), dried and weighed thesampled plants, and developed regression models to estimate above-groundbiomass (AGB). Size metrics measured include: length of longest branch,width of largest leaf, mat length and width, rosette length and width.132.2. MethodsIn order to apply the Excluder bags before the flowers were receptive,plants were checked and pollination treatments were applied every day dur-ing the early growing season. After the majority of flowering was underwayfor each species and Excluder bags were in place, the plants were checkedand pollinated using a paintbrush every 2 days. Once flowers had senescedand had begun to disperse their seeds, the entire floral head was collectedand stored in individual envelopes. Some of the flowers failed to produceseeds, either due to flower abortion or being damaged by caterpillars. Insome cases, I was able to see insect frass or the actual larvae, but this wasnot always the case.Upon return from the field site in late August, seeds were stored in afreezer at -4◦ C for 27 weeks (September-March). Batches of seeds wereweighed to the nearest 0.001g, counted, and a random sample from eachflowering head was taken for germination, as numbers permitted (up to 100seeds for S. arctica, 50 for D. integrifolia and P. radicatum). Germinationtrials took place in a greenhouse which provided 20 hours of light daily,and maintained temperatures from 15-25◦ C. Seeds were placed on filterpaper in 10 cm petri dishes and kept moist with distilled water. Seedswere checked every 2 days from May 15 - June 14, 2013, and germinatedseeds were removed from each dish and counted. I defined “germinated” asshowing exposed cotyledons (S. arctica) or an exposed rootlet of >5mm (D.integrifolia and P. radicatum).2.2.4 Statistical AnalysesTo compare numbers of flowers produced per plant, I used Generalized Lin-ear Models (GLM) with a Zero-Truncated Negative Binomial (ZTNB) distri-bution and a log-link function. This was necessary because the data did notfit assumptions of normality and heteroskedasticity for a traditional ANOVAmodel. Overdispersion in the data was too large to use a Poisson GLM, soI used a Negative Binomial GLM (Hilbe, 2011). A Zero-Truncated distri-bution (sometimes referred to as a “hurdle” model) was necessary becauseplants that failed to produce flowers could not receive pollination treatments,making it impossible for a treated plant to have zero flowers. Such modelsare described in Zuur et al. (2009) and Hilbe (2011). I selected the “best”model using Akaike’s Information Criteria (AIC), using second-order bias-corrected AIC to account for sample size (Burnham and Anderson, 2002;Anderson, 2008). AIC is considered a predictor of information entropy in amodel, and is used to select the most parsimonious model (Anderson, 2008).To examine the AGB:flower production relationship, I used Zero-Truncated142.2. MethodsNegative Binomial GLMs. I included inferred AGB as a predictor to reduceany confounding effect of plant biomass on flower production. All predictormetrics were included in the original model, then AIC was used to deter-mine the “best” model. Inferred biomass was calculated as follows for eachspecies:logBsalix = b0 logW + b1 logL+ b2 logLB + b3 logL× logLB +  (2.1)Where B = Above-ground biomass, W = Mat Width (cm), L = Mat Length(cm), LB = Longest Branch Length (cm), and  = errorlogBdryas = b0 + b1 logW + b2 logL+  (2.2)Where B = Above-ground biomass, W = Mat Width (cm), L = Mat Length(cm), and  = errorlogBpapaver = b0 logW + b1 logL+ b2 logLW + b3 logW × logLW +  (2.3)Where B = Above-ground biomass, W = Rosette Width (cm), L = RosetteLength (cm), LW = Width of largest Leaf(cm), and  = errorTo model the seed response across treatments, I used generalized lin-ear mixed-effects models (GLMM), with seed response for each harvestedflower nested within plant. GLMMs are relatively new in statistical litera-ture, and are sensitive to the same assumptions that underlie both GLMs(Hilbe, 1994; Neuhaus and McCulloch, 2011) and linear mixed-effects mod-els (LMM) (Laird and Ware, 1982; Pinheiro and Bates, 2000; Zuur et al.,2009). Bolker et al. (2009) gives an excellent overview of assumptions andfitting methods associated with GLMMs.For each model, seed response was modeled using Pollination and Warm-ing as Fixed Effects, with each individual plant as a Random Effect (nec-essary to avoid pseudoreplication). This structure of mixed model was re-peated for the following responses:• Seed Number• Total Seed Mass (g)• Average Seed Mass (Total Seed Mass/Seed Number) (g)• % Germinability (Number Germinated/Number Sampled) x 100152.3. ResultsAll statistical analyses were done using R 3.0.2 (R Core Team, 2013). Bi-nomial GLMs were fit using the base stats library, the zero-truncated GLMswere fit using the VGAM library (Yee, 2013), negative binomial GLMMswere fit using the glmmADMB library (Fournier et al., 2012), Gaussian, Pois-son and binomial GLMMs were fit using the lme4 library (Bates et al., 2013).The seed germination data were found to be highly overdispersed, and wasmodelled using a Markov-Chain Monte Carlo (MCMC) approach throughthe MCMCglmm library (Hadfield, 2010), which allows for overdispersionand is more robust than likelihood-based methods of parameter estimation.Wald z-tests or t-tests of significant effects (i.e. p<0.05) are not reliablefor small sample sizes in GLMs and are not defined for LMMs or GLMMs.Likelihood-ratio tests are also unreliable for small sample sizes, and theirperformance in a GLMM context is not well studied (Pinheiro and Bates,2000; Bolker et al., 2009). I cite differences between the nested candidatemodels using ∆AIC, which is a measure of model parsimony. For citationsof differences between candidate models, a negative ∆AIC indicates thatadding the term improves the model (Anderson, 2008). Table 2.6 containsan overview of flower response to pollination and warming.2.3 ResultsUsing a handheld thermocouple probe, I found that temperatures were ele-vated by only 1◦C on average within the corolla of D. integrifolia. However,the temperature within the corolla of P. radicatum was elevated by an av-erage of 5◦C.2.3.1 Flower ResponseSalix arctica exhibited no response in flower number to warming treatment(∆AIC = 1.23). The effect of pollen was negligible both inside (∆AIC= 2.08, Figure 2.6.1) and outside the warming treatments (∆AIC = 2.64,Figure 2.7a). AGB was an important predictor of flower number (∆AIC =-6.22), but the model was not improved by adding warming, pollination, orany higher-order combination of the terms.Dryas integrifolia forms flower buds the previous year similar to S. arc-tica, but responded completely differently to warming treatment. Out-side the warming treatments, pollen did not alter flower numbers (∆AIC2.73, Figure 2.7b). Flower number within warming treatments responded topollen and showed an interaction between warming and pollination (∆AIC= -6.88), where plants that were warmed but unpollinated produced the162.3. Resultsgreatest number of flowers per plant (Figure 2.6b). However, when I usedinferred AGB as a covariate to correct for the size of the Dryas mats, awarming term improved the model (∆AIC = -5.70), but pollination did not(∆AIC = 0.69). This can be seen in Figures 2.6e and Figure 2.8 which showsthat biomass influences flower production, but only if the plant is warmed.Papaver radicatum also exhibited no response in flower number to pol-lination (∆AIC = 12.56) or warming (∆AIC = 1.95). AGB was found toimprove AIC the most (∆AIC = -15.36), but the model was not improvedany further by adding warming, pollination, or any higher-order combina-tion of the terms. Figures 2.6 and 2.7 show that the flower numbers andflower numbers per unit AGB for P. radicatum are very similar.2.3.2 Seed ResponseSalix arctica exhibited a dramatic response to pollination manipulation, in-creasing the number of seeds produced (∆AIC = -4.28) and total biomass ofseeds (∆AIC = -1.2) when pollinated, while not increasing the overall seedsize (∆AIC = 22.40). Pollinated catkins produced an average of 100 moreseeds inside and 145 more seeds outside the OTCs, compared to their un-pollinated counterparts (Figures 2.9a, 2.10a). OTC warming did not affectseed number (∆AIC = 2.02), nor did the interaction between warming andpollination, indicating that S. arctica may be generally subject to pollenlimitation, but that this is not altered by OTCs (Figure 2.9a).Pollination and OTC warming appears to have not altered germinationin any important way, but this is difficult to interpret because of the verylow germinability of S. arctica seeds (Figures 2.11, 2.12). I found a marginaleffect of warming, pollen, and a warming:pollen interaction on seed germi-nation in S. arctica using a MCMC approach to model fitting (∆DIC =1.08). However, the Deviance Information Criteria (DIC), a Bayesian ana-logue of AIC, showed very little difference between the full model (Warming+ Pollen + Warming:Pollen) and the null model, indicating that the differ-ence between candidate models is very small. Warming was only significantif high-order combinations of terms were included.Dryas integrifolia did not change total seed number or biomass in re-sponse to OTC warming or pollination. For seed germination, warming,pollination, and a warming:pollen interaction were identified as being impor-tant (∆DIC = -3.35). ∆DIC values between candidate models were small,but a MCMC modelling framework also identified the full model as havingthe best DIC score, although the interaction term was not significant (pmcmc= 0.136). Outside the OTC treatments, pollen was important in predicting172.4. Discussiongerminability (∆DIC = -1.84), with hand-pollinated treatments producingmore viable seeds (pmcmc = 0.026).Papaver radicatum also did not change seed number or biomass withOTC warming or pollination. I found a marginal effect of warming, pollen,and a warming:pollen interaction on seed germination using a MCMC ap-proach (∆DIC = -1.08), but the DIC differences between models was verysmall. This is most likely due to small number of control plants used, aswell as the low overall germination values. Similar to S. arctica, this makesthe results difficult to interpret, but in general there was very little effecton germinability by OTC warming (∆DIC = 3.17) or pollination (∆DIC =0.47). Within the “best” (full) model, no terms were found to be signifi-cantly different from 0 (pmcmc >0.2). However, outside the OTC treatments,I found that the selfing treatment caused a decrease in germinability (pmcmc< 0.001), but hand pollination and insect exclusion did not. See Table 2.3for an overview of the seed response.For detailed tables of AIC values for all candidate models, see Tables A.2,A.3, and A.42.4 Discussion2.4.1 Salix arcticaNeither warming nor pollination altered flower production did not change inS. arctica. This is not surprising given that S. arctica produces flower budsthe previous season.Pollen addition was the main factor constraining production of seeds in S.arctica. I did not find evidence to suggest that OTCs alter seed production,or that they alter the pollination of S. arctica flowers. Jones et al. (1997)found that the ratio of seeds:flower in individual Salix catkins was decreasedby OTCs, which they used to imply that pollination was reduced withinthe treatments. While not directly counting flower numbers per catkin, Irecorded growth of the terminal shoot (length of catkin + length of stem)for each catkin, which has been shown to be a linear predictor of flowernumber (Jones, 1995). I found that neither warming nor pollination hadany effect on the length of the terminal shoot (i.e. number of flowers percatkin), but Jones et al. (1999) has shown that this can vary from year-to-year and habitat-to-habitat. Pollen treatment did not alter the total numberof catkins produced by S. arctica, because flower buds are grown at the endof the previous season.Salix arctica germination was not largely affected by either pollination182.4. Discussionor OTC warming. Both inside and outside the OTCs, I found that all nestedmodels had a very similar DIC, indicating that the effects of pollination andOTC warming are not largely important in germination. Much of this isprobably due the inherently low germinability of S. arctica seeds. This alsomay be due to the sample size used, as well as the structure of the binomialGLMM required to analyze nested, overdispersed, binomial data. In thefuture, larger numbers of plants per level of treatment (>15) should be usedif analyzing germination under a GLMM framework.My low recorded germination rates for Salix arctica are consistent withother studies on germination (Bliss, 1958; Billings and Mooney, 1968; Daw-son and Bliss, 1989a). However, Klady et al. (2011) tested seed germinationacross sites in the Alexandra Fiord Lowland, and recorded much higher ger-mination rates for S. arctica (23%) than what were found in my study. Thismay have occurred for a few reasons: Klady et al. (2011) used a much shortercold stratification period (1 month) before testing germinability, and useda longer photoperiod (24hrs). The length of stratification is more likely thecause of the difference between my treatments, because Dawson and Bliss(1989a) used a similar cold stratification period (-30◦ C. for 18-24 weeks)to ours, and reported similar overall germination (2%). Salix seeds are alsovery small, on the order of 0.001 g per seed. The very small amount ofendosperm along with each fruit means that viability decreases very quicklyover time (Densmore and Zasada, 1983).2.4.2 Dryas integrifoliaAs a species, the reproductive responses of D. integrifolia have not beenstudied as much as its more widespread sister taxa D. octopetala.I found that OTC warming increased flower production per unit AGBin D. integrifolia. This is most likely caused by a decreased rate of budabortion in warmed plants. Wookey et al. (1993) also found that artificialwarming treatments can stimulate flower production in D. octopetala, de-spite the fact that Dryas sets reproductive buds the previous year. Theyattribute this to a decreased rate of flower abortion and unsuccessful bud-burst within warmed treatments. Similarly, I found that warming treat-ments increased the Flower Production:AGB ratio for individual plants,whereas control plants produced approximately the same number of flowersper plant regardless of the biomass. Pollination treatments did not alterthis relationship.Although flower number responded to warming, I did not find any changesin seed number or seed mass resulting from OTC warming or pollination192.4. Discussiontreatment. In this respect, my findings differ from the results of Welker et al.(1997) who found that OTCs increased both seed number and seed mass inD. octopetala . However, their between-site findings were very different, in-dicating that D. octopetala has a strong dependence on local environmentalconditions. I found that that germinability of D. integrifolia was affected byOTC warming, pollination, as well as their interaction. Because I also foundincreased germinability in hand-pollinated plants outside OTCs, it is likelythat seed viability is related to both warming and pollination. Because Ifound lower seed viability outside of the OTCs, it is not likely that OTCsare interfering with pollination in any significant way, and that the increasedviability is caused by the elevated air temperatures.My pollen treatments outside of the OTCs revealed that pollen additionincreases germinability in D. integrifolia, but that exclusion of insect polli-nators or wind has no large effect on germinability. Kevan (1972b) foundthat insect exclusion caused a reduction of seed-set in D. integrifolia, andthat emasculated flowers were still capable of producing some seeds in aninsect-free environment, although his sample size was small (1 plant, 5 flow-ers). I also found that D. integrifolia can produce seeds in the absence ofpollinators. A possible explanation for this is that the flowers require smallquantities of pollen relative to the amount of pollen they produce. Philippet al. (1990) found that in D. integrifolia only 1% of the total pollen pro-duced by a single flower was necessary to pollinate all of the stigmas onthe flower. In this way, D. integrifolia can be thought of as a pollination“opportunist”: extra pollen enhances germinability, but it is not dependanton these external vectors.2.4.3 Papaver radicatumI found that experimental warming did not increase flower production, orflower production per unit AGB in P. radicatum. AGB was also not differentbetween control and OTC for P. radicatum.Seed number and seed mass were unaffected by either warming or pol-lination treatments. This indicates that the mating system of P. radicatumis fairly resistant to changes in pollination rates and ambient temperature.Kevan (1972b) found that P. radicatum exhibited no response in seed pro-duction to exclusion of pollinators, meaning that self-pollination is commonin this system. At the level of the flower, the temperature effect of OTCsmay be relatively small, because P. radicatum are capable of maintaininga high-temperature microclimate within their corolla by tracking the sun(Kevan, 1975).202.4. DiscussionThe germination response ofPapaver radicatum is difficult to analyze.Because I did not exclude insects from hand-pollinated blossoms, I cannotexclude the possibility that the action of hand-pollination either imperfectlydeposited pollen or damaged the stigmas, confounding a change in over-all germinability. However, I believe that my treatments were not overlyunrealistic, and conclude that the pollination system of P. radicatum is ro-bust enough to not require large amounts of cross-pollination. Given thatmany plants at the edge of their range retain some degree of self-pollination,this is not surprising. I exclude germination conditions as a possible factor,because Olson and Richards (1979) found that final germination rates werenot largely affected by temperature conditions. Karlsson and Milberg (2007)found that the time of seed collection was an important factor in seed ger-mination. I did not find any effect of the day of seed collection throughthe season, most likely due to the short range of collection days late in theseason.In the control plants, germinability was reduced strongly by the self-ing treatment (pmcmc <0.001), but not strongly altered by either hand-pollination or insect exclusion (pmcmc¡0.1). This is more likely due to changesin temperature caused by the selfing treatments, rather any kind of self-incompatibility mechanism. Papaver radicatum is heliotropic, and temper-atures can be strongly elevated within the corolla (Hocking, 1968; Kevan,1975). Corbett et al. (1992) found that obstruction of flower petals can causethe blossoms to not track the sun, and have lower ovary temperatures, lead-ing to smaller and fewer seeds. I suspect that this may have been the casewith the selfing treatments, as the cloth required to exclude any wind-bornepollen necessarily covered the entire corolla, which probably caused a dropin temperature. While I did not find smaller seeds within the selfing treat-ment, it is conceivable that it may have altered germinability. The exclusiontreatment (1mm net) is probably a more realistic estimate of self-fertilizedseed production in P. radicatum, given that wind-pollination events are rarein Papaver.OTCs do not strongly alter the magnitude of (seed) reproductive re-sponse of P. radicatum. Other species of Papaver, such as P. somniferum,are generally insect-pollinated or self-pollinated because of their sticky, densepollen (Patra et al., 1992; Miller et al., 2005). Papaver radicatum is similar,but has much higher rates of self-pollination than lower-latitude Papaverspecies (Nordal et al., 1997). Papaver radicatum can be considered to be acomplex of different subspecies (Solstad et al., 1999), so it is possible thatindividual subspecies could have slightly different responses to temperatureand pollination. Year-to-year variability was also not considered within this212.5. Conclusionsstudy, but can strongly influence both the timing and magnitude of flowerproduction (Le´vesque et al., 1997).2.5 ConclusionsMany studies have shown that OTC warming can have variable effects onsexual reproduction in Arctic and alpine plants, but I found that OTCwarming and changes in pollination are generally independent. Of the threeplant species I examined:• Salix arctica altered seed production in response to pollen levels, pro-ducing more seeds when hand-pollinated, and fewer when insect visi-tors were excluded.• Dryas integrifolia produced more flowers per unit of AGB in OTCwarming treatments, and displayed higher germinability in both warmedand pollinated treatments.• Papaver radicatum reproductive output was not largely altered byOTC warming or pollen manipulation.Seedling establishment is very rare in the Arctic, mainly due to lowsummer temperatures and low water availability for seedlings (Bell and Bliss,1980). Establishment typically happens in climatic “windows” for long-livedspecies (Bliss, 1958; Eriksson and Fro¨borg, 1996). Almost all High Arcticplants can be thought of as stress tolerant C -selected species, sensu Grime(1977), or highly competitive K -selected species, sensu MacArthur (1967).Molau (1993) re-defined the systems of MacArthur (1967) and Grime (1977)in the context of tundra communities, which can be described in terms oftheir flowering phenology alone. Rather than r and K -selected, he defined“pollen-risking” and “seed-risking” strategies, which exist along a continuumof early-flowering to late-flowering, with corresponding rates of low selfing tohigh selfing. Examples of “pollen risking” strategists include S. arctica andSaxifraga oppositifolia, while “seed-risking” strategists include P. radicatumWith Papaver radicatum, a conflicting pattern emerges: a mid-seasonflowering plant that is apparently unaffected by selfing or pollination. Thismay be more of an artifact of the treatments used rather than an actualreflection of how the plants reproduce. Papaver radicatum is highly depen-dent on insolation to produce seeds (Corbett et al., 1992), and assumedly,to produce viable seeds. Kjellberg et al. (1982) found a similar pattern inD. octopetala. The 10µm cloths that I used to exclude wind-borne pollen222.5. Conclusionsfrom the flowers of P. radicatum would also have had a shading effect withinthe corolla, which may have caused temperature to be confounded with self-ing. This does not appear to have had a large effect in D. integrifolia.Unfortunately, there are relatively few methods to directly test rates of self-pollination aside from genetic analysis. If rates of insect visitation were tobe continuously monitored over the season, it may be possible to indirectlyassess how rates of visitation influence germination for Papaver radicatum.Because of the tedious nature of germination studies, there is a lack ofstudies of seed germinability related to warming in the Arctic. Many studiessimply use seed size as a proxy for germinability, but my results show thatseed size is generally independent of germinability. Cooper et al. (2004) andMu¨ller et al. (2011) suggest that greenhouse germination alone is insufficientto demonstrate germinability, and that field trials are necessary to measurethe actual germination probabilities. To my knowledge, there are no studieson field trials of seed germination inside OTCs in the Arctic. Even thoughseedling establishment is rarer in the Arctic, it is still an important factorin many Arctic ecosystems (Freedman et al., 1982; Nordal et al., 1997), andits importance will continue to increase as glaciers retreat, creating newhabitats for establishment.232.6. Figures and Tables2.6 Figures and Tables242.6. Figures and TablesFigure 2.1: Location of Alexandra Fiord, Ellesmere Island (78.882◦N,75.782◦W). Source: James Hudson.252.6.FiguresandTablesFigure 2.2: The Xeric Shrub site at Alexandra Fiord, with several ITEX OTCs visible. Many Salix arctica catkins(white, fluffy flowers) are dispersing seeds.262.6.FiguresandTablesFigure 2.3: Salix arctica (Pall) ♂. Plants are dioecious, with male plants maturing earlier and producing copiousamounts of pollen. Both male and female forms produce nectar.272.6.FiguresandTablesFigure 2.4: Dryas integrifolia (Vahl). Their flowers are mainly complete (hermaphrodidic), but some flowers areentirely staminate or pistillate. All forms of flowers have deep nectaries, and produce some of the largest volumesof nectar of all high Arctic flowers (Hocking, 1968). The flowers are moderately heliotropic.282.6.FiguresandTablesFigure 2.5: Papaver radicatum (L.) All Papaveraceae lack nectaries, but produce large amounts of sticky, densepollen. Their flowers are highly heliotropic, and the temperature within the bowl-shaped corolla can be significantlyhigher than the surrounding air. The fly basking in the corolla is a male Phaonia spp. (Diptera: Muscidae)292.6.FiguresandTablesFigure 2.6: Flower output from hand-pollinated and warmed plants. Flowers per plant for a) S. arctica, b)D.integrifolia, and c) P. radicatum. Flowers per gram AGB shown for d) S. arctica, e) D.integrifolia, and f) P.radicatum. Letters a, b indicates ∆AIC >2 between candidate models.302.6.FiguresandTablesFigure 2.7: Flower output from pollen-manipulated plants outside warming treatments. Flowers per plant areshown for a) S. arctica, b) D.integrifolia, and c) P. radicatum. Flowers per gram AGB shown for d) S. arctica, e)D.integrifolia, and f) P. radicatum. Pollination treatments are described in Methods. Letter a indicates indicates∆AIC ≈ 0 between candidate models.312.6. Figures and TablesFigure 2.8: Flower output from warmed and control D. integrifolia plantsas a function of above-ground biomass (AGB). All measurements are perplant.322.6.FiguresandTablesFigure 2.9: Seeds per flower and seed mass per flower from hand-pollinated and warmed plants. Seeds per flower isshown for a) S. arctica, b) D.integrifolia, and c) P. radicatum. Mass of seeds per flower is shown for d) S. arctica,e) D.integrifolia, and f) P. radicatum. Letters a, b indicates ∆AIC >2 between candidate models.332.6.FiguresandTablesFigure 2.10: Seeds per flower and seed mass per flower from pollen-manipulated plants outside warming treatments.Flowers per plant are shown for a) S. arctica, b) D.integrifolia, and c) P. radicatum. Flowers per gram AGB shownfor d) S. arctica, e) D.integrifolia, and f) P. radicatum. Letters a, b indicates ∆AIC >2 between candidate models.342.6.FiguresandTablesFigure 2.11: Average seed mass and germinability from hand-pollinated and warmed plants. Average seed massis shown for a) S. arctica, b) D.integrifolia, and c) P. radicatum. Seed germinability is shown for d) S. arctica, e)D.integrifolia, and f) P. radicatum. Letters a, b indicates ∆AIC >2 between candidate models.352.6.FiguresandTablesFigure 2.12: Average seed mass and germinability from from pollen-manipulated plants outside warming treat-ments. Average seed mass is shown for a) S. arctica, b) D.integrifolia, and c) P. radicatum. Seed germinability isshown for d) S. arctica, e) D.integrifolia, and f) P. radicatum. Letters a, b indicates ∆AIC >2 between candidatemodels.362.6. Figures and TablesTable 2.1: Factors used in final design of experiment. A complete crossbetween Warming and Pollination was not possible due to the small num-ber of plants available within the Warmed (OTC) treatments. Pollinationtreatments: Control plants were not altered. Pollinated plants were hand-pollinated. Excluded plants had 1mm mesh net Excluder bags placed overflowers to exclude pollinators. Selfed plants had 10µm cloth Excluder bagsplaced over flowers to exclude all external pollen. Selfing treatments werenot applied to S. arctica because the plants are dioecious.Salix arcticaPollinationControl Pollinated Excluded Selfed TotalWarmingControl 10 Plants 10 Plants 10 Plants - 30OTC 10 Plants 10 Plants - - 20Total 20 20 10 0 50Dryas integrifoliaPollinationControl Pollinated Excluded Selfed TotalWarmingControl 10 Plants 10 Plants 10 Plants 10 Plants 40OTC 10 Plants 10 Plants - - 20Total 20 20 10 10 60Papaver radicatumPollinationControl Pollinated Excluded Selfed TotalWarmingControl 10 Plants 10 Plants 10 Plants 10 Plants 40OTC 10 Plants 10 Plants - - 20Total 20 20 10 10 60372.6. Figures and TablesTable 2.2: Models explaining flower number in response to warming andpollen manipulation. - indicates no important effect by any treatment fac-tors. For specific AIC values, see Table A.2.Treatment Flower# Flower#BiomassSalixOTC∗pollen- -Salix pollen - -DryasOTC∗pollenWarming +Pollen+Warming:PollenWarmingDryas pollen - -PapaverOTC∗pollen- -Papaver pollen - -Table 2.3: Models explaining seed number, mass, and germination in re-sponse to warming and pollen manipulation. - indicates no important effectby any treatment factors, while subscripted text indicates marginal effect(∆AIC<2). For specific AIC values, see Table A.3 and A.4.Treatment Seed# TotalSeedMassMass perSeedProportionGerminationSalixOTC∗pollenPollen Pollen - -Salix pollen Pollen Pollen - PollenDryasOTC∗pollen- - - Warming+Pollen+Pollen:WarmingDryas pollen - - - PollenPapaverOTC∗pollen- - - -Papaverpollen- - - Pollen38Chapter 3Effects of Open-TopChambers on FloweringPatterns and Potential InsectVisitation3.1 IntroductionInsects are key components of nearly all terrestrial ecosystems worldwide,acting as consumers, predators, parasites, and decomposers (Price, 1997;Schowalter, 2006), and their activities are well-known to contribute to ecosys-tem function (Holling, 1973; Losey and Vaughan, 2006). Arctic insects areno less important (Strathdee and Bale, 1998), and act as consumers (More-wood and Ring, 1998), decomposers, predators (Oliver, 1963), parasitoids(Kukal and Kevan, 1987), and pollinators (Kevan, 1972b). Some authorsindicate that the importance of insects in Arctic ecosystems is understated(Roslin et al., 2013; Gillespie et al., 2013). Insects represent the bulk of thetotal animal diversity of the Arctic (Chernov, 1995), but in general are notwell studied.As climate change increases air temperature and advances summer snowand ice melt in the Arctic, the structure and function of most ecosystemswill be altered. This will be seen in a variety of ways, including increasednet ecosystem production (Welker et al., 2000; Oberbauer et al., 2007; Lundet al., 2012), nutrient cycling (Epstein et al., 2000), changes in plant biomass(Hill and Henry, 2011), and timing of reproductive events (Wookey et al.,1993; Oberbauer et al., 2013). Animals will alter the timing of their re-production, migration, and hibernation, notable examples being birds (Gas-ton et al., 2005), ungulates (Post and Forchhammer, 2008), and insects(Franze´n and O¨ckinger, 2012).Niche-based models of range expansion arelikely insufficient to represent actual transition, and experimental methodsare necessary to predict responses to climate change (MacDonald, 2010;393.1. IntroductionVan der Putten et al., 2010). Warming experiments such as those done byresearchers involved in the International Tundra Experiment (ITEX) are anattempt to study the effect of climate change on plant communities, mainlyusing open-top chambers (OTCs). While these methods are generally suf-ficient to represent climate change scenarios in plant communities (Marionet al., 1997; Hollister and Webber, 2000), their effects on animals, specifi-cally insects, are not well-quantified. Some authors have found that OTCschange the activity of soil and plant-dwelling insects (Ring, 2001; Dolleryet al., 2006; Gillespie et al., 2013). It is generally assumed that the greaterair temperatures within OTCs accelerates the growth and reproduction ofthese insects, given that the activities of insects are highly dependent ontemperature (Danks et al., 1994; Danks, 2004). However, Moise and Henry(2010) suggest that plot-based manipulations are unrealistic for estimatinganimal effects within global change experiments, and others have implicatedthat OTCs exclude pollinating insects from plants during crucial parts oftheir flowering period (Jones et al., 1997; Stenstro¨m et al., 1997).The detailed composition of plant-pollinator networks in the Arctic isnot well known, except for a few studies (Mosquin and Martin, 1967; Ke-van, 1972b; Olesen et al., 2008; Franze´n and O¨ckinger, 2012). Changes tothe network over the growing season are often unstudied, as the continuousmonitoring of flowering plants and insects is tedious work. Most studiesfocus only on a single species of plant (Philipp et al., 1990, 1996; Sten-stro¨m et al., 1997) or a single group of visiting insects (Kevan and Kevan,1970; Richards, 1973; Pont, 1993). There are very few studies examininghow climate change has altered plant-pollinator networks, and even fewerin the Arctic (Franze´n and O¨ckinger, 2012; Høye et al., 2013). Dipterans(true flies) are often the dominant pollinator in Arctic plants (Mosquin andMartin, 1967; Elberling and Olesen, 1999; Robinson, 2011), but bumblebees(Bombus spp.) are often cited as being important pollinators (Richards,1973; Lundberg and Ranta, 1980), as they can withstand fluctuations inair temperature, wind, and precipitation more readily. Furthermore, therehave been no studies other than Ring (2001) that attempt to systematicallyquantify the effect of OTCs on ambient insect populations.In this study I quantify the structure of a High Arctic plant-pollinatornetwork, and examine the evolution of the interactions over the growingseason. In particular, I examine how OTCs alter both the level of availableflowers and the levels of available visiting insects over the course of the grow-ing season. I discuss how well these OTC-induced changes may representfuture ecological interactions, particularly plant-pollinator interactions. Fi-nally, I examine the plant-visitor interaction rates, and show how differences403.2. Methodsin resource availability between Arctic flowers can change the magnitude ofinteraction.3.2 Methods3.2.1 Flowering CommunityThe site and location are described in Chapter 2. The flowering plant com-munity (non-graminoid) is composed primarily of Salix arctica, Dryas in-tegrifolia, Stellaria longipes and Papaver radicatum, with a small numberof Draba lactea, Saxifraga oppositifolia, Saxifraga cernua, Saxifraga tricuspi-data, Cassiope tetragona, and Pedicularis capitata individuals. Salix arctica,D. integrifolia, and P. radicatum are described in Chapter 2.Stellaria longipes Goldie subsp. longipes is a long-lived perennial thatgrows in a variety of habitats, and is a successful colonizer (Chinnappa et al.,2005). In primary successional sequences in the Arctic, S. longipes is oftenone of the first colonizers, along with Salix arctica and Saxifraga oppositi-folia (Breen and Le´vesque, 2006), but is also present in undisturbed com-munities. Flowers are small, 10-petaled, 5-7mm white flowers (Figure 3.1),and plants can be gynodioecious (female-only as well as bisexual flowers).Often their ratio of bisexual to female-only flowers changes from year-to-year or within the flowering season, depending on environmental conditions(Philipp, 1980). They are phenotypically very plastic, and can adapt to alarge variety of environmental conditions due to their polyploidy and largegenotypic variability (Dang and Chinnappa, 2007).To gauge flower availability to insect visitors, I monitored flowers inplots over the growing season (14 Control, 14 OTC plots), counting all openblossoms with non-withered anthers and stigmas, where the petals had notyet begun to wither. This was done in both control and OTC plots, onthe same days that insect sampling was conducted. Any receptive, open,non-graminoid flower was counted. Individual catkins of Salix arctica werecounted as a single blossom.3.2.2 Insect CommunityTo sample the overall flying insect community, I followed the CANPOLIN(2009) bowl trap protocol using sets of white, yellow, and blue bowls. These15 cm plastic bowls were attached to the ground using metal tent pegs. Tentpegs were necessary because wind storms of up to 80 km/h occasionallyoccur in the valley. For the 15 treatment bowl traps, a single bowl was413.2. Methodsplaced in each of the 14 OTCs (one of the OTCs contained 2 bowls), thecolour of the bowl being systematically assigned along the length of thesite. For the control bowl traps, 5 of each colour was placed outside theOTCs in a 140 m transect along the site, where bowl colour was repeatedsystematically along the transect. This sampling procedure took place atthe xeric shrub community described in Muc et al. (1989) and Svoboda andFreedman (1994a). I conducted insect sampling every two days, reducingthis to three days during the latter part of the season when insect activitywas visibly decreased, and no receptive flowers were available.The bowls were filled with (unscented) soapy water during the morning ofthe sampling days. The morning following the sampling day, the contents ofthe bowls were poured through a 0.5 mm mesh strainer to catch arthropodspresent in the bowls. Mites and Collembola may have escaped through thissize of mesh, but these were not the focus of my study.Bowl traps do not give any information about taxa of insects that visitflowers, so I conducted targeted netting of visiting insects along the siteoutside the OTCs. Field workers walked a patrol route across the site,capturing any visiting insects they saw on flowers. The length of time duringthe patrol was recorded, and patrols were repeated over the course of theday by different workers. Insects I failed to capture were identified as far asvisually possible (usually to Family).All arthropods were stored in 70% ethyl alcohol and transported to thelab at UBC in mid-August. Insects were pinned, identified, and labeledas precisely as possible, usually to Genus (Chironomidae, Mycetophillidae,Sarcophagidae, Tachinidae, and Ichneumonidae were only identified to Fam-ily). Soft-bodied insects, minute insects, or arthropods such as spiders werenot identified, as they were rarely encountered on flowers, their abundanceswere low, and their populations were considered to be beyond the scope ofthe project.3.2.3 Statistical AnalysesTo test if the composition of flowers was different inside and outside of theOTCs, I used non-metric multi-dimensional scaling (NMDS) to reduce thedimensionality of my flower count data and a multivariate analysis of vari-ance (MANOVA) to test for differences between OTC and control “flowercommunities”. This was done using a permutation test of Bray-Curtis sim-ilarity, which partitions variance between variables. For my analysis, I usedwarming treatment and day of year as factors. Because there were manyplots that had no flowers and NMDS techniques are ill-suited for dealing423.2. Methodswith “empty communities”, flower density for OTC and control was takenas an average of the 14 warmed and control plots. Days where no flowerswere recorded in any plots were not used.To test whether OTCs influence overall flower density at the plot scale,I used Generalized Estimating Equations (GEE). GEEs are described inLiang and Zeger (1986) and Hardin and Hilbe (2003), and are essentially aversion of Generalized Linear Models (GLM) that allow for more elaboratecorrelation structures, such as between subject or between measurementperiod. In this case I used Plot as a grouping structure, and used an auto-regressive model of order 1 (AR-1) to model correlation between the sameplot on different days.Tests for day of maximum flowering were done in the following man-ner: the days of peak flowering for all non-zero plots were tallied, then aStudent’s T-test of equal means was used to compare between warmed andcontrol days. If the data did not meet the assumptions of a T-test, a non-parametric Kruskal-Wallis test of parameter equal distribution was used. Itested induced changes in day of first flower, the day of last flower, and thelength of the flowering season (Day of Last Flower − Day of First Flower).If the day of first flower for a plot was on first day of observations, that plotwas not used, and similarly if the day of last flower was on the last day ofobservations, that plot was also not used.Similar to the examination of the flower community, I used MANOVA totest changes in insect catches between OTC and Control plots. All commu-nity analyses were done at the taxonomic level of Family, because this wasthe common level to which I was able to identify all specimens. I used countsof bowl-trapped insects as the “community” data, and compared change incatches from OTC and Control plots, between bowl colours, and across thelength of the season. To examine the effect of OTCs on bowl-trap catcheswithin a single family, I also used GEEs, with an AR-1 structure to modelcorrelation between catch rates on different days.Even though patterns of preference appear clear in terms of catch rates,these catch rates may be conflated by flower and insect abundance. In otherwords, if there were a higher number of receptive flowers available, or a largernumber of insects present on that day, I would expect a higher catch rate.This can also be viewed in light of MacArthur’s “principle of equal oppor-tunity”, where limited resources are used in proportion to their abundance(MacArthur, 1972). In order to examine the preference of flower visitorsin relation to abundance, I scaled the catch rate by insect availability toexamine the interaction rate (catch rate/bowl trap catches, used as an esti-mate of population size) for each group of pollinators. Interaction rates are433.3. Resultsnot directly comparable between groups of pollinators because individualgroups of pollinators will be differentially attracted to bowl traps, but theyare roughly comparable within each group. The data met the assumptions ofsimple linear regression once transformed, so I modeled interaction rates us-ing flower availability, insolation and wind speed (air temperature was foundto be collinear with insolation). No Bombus polaris were caught in the bowltraps, so I could not get any general estimates of abundance throughout theseason, so I used the catch rate instead of the interaction rate. I could nottransform the data to meet the assumptions of a linear model, so I comparedcatch rates to individual variables using a Kendall rank-correlation test. Irealize that interspecific competition (Hocking, 1968) or facilitation (Heg-land and Totland, 2005; Ghazoul, 2006) may occur between simultaneouslyflowering plants, but a network study of this nature is complex and beyondthe scope of this work.GEE computations were done using the geepack library (Halekoh et al.,2006), NMDS and community analysis were done using the vegan library(Oksanen et al., 2013), rank-order tests were done using the Kendall library(McLeod, 2011), and the remainder was done in the base library in R (RCore Team, 2013).3.3 Results3.3.1 Flower AvailabilityThe flower community was significantly altered by OTC warming (p=0.05)and date (p=0.004), but OTC warming explained only 4.4% of the variation,while date explained 50.3%. This indicates that OTCs do alter the overallcomposition of the flowering communities, but not nearly as much as day-to-day changes over the season. Warming was only a significant predictorif permutation tests were constrained by day of year, indicating that whilethere are differences between OTC and control plots, these differences arelargely not detectable at the scale of the entire growing season. Figure 3.2shows how the flower community changes over the course of the floweringseason using NDMS scores, which represents the composition of availableflowers in a multi-species space. The composition follows a clockwise arc asthe community progresses from Salix to Dryas to Papaver -Stellaria. Bothcommunities generally follow the same progression, but the warmed com-munity progresses faster.Similar to other phenological studies (Wookey et al., 1993; Alatalo andTotland, 1997; Jones et al., 1997), the timing of flower phenology was altered443.3. Resultsby the OTC treatments. The Day of First Flower (DFF) was significantlyearlier in both S. longipes (p=0.05) and P. radicatum (p=0.0007) (Table3.1), similar to the results of Mølgaard and Christensen (1997). Based onmy past observations, and the findings of Hocking (1968) and Jones et al.(1997), I suspect that the day of first flower for S. arctica would also havebeen significantly earlier, but I was unable to begin surveys early enough inthe season to formally test this.In S. arctica ♀(p=0.0005), and D. integrifolia (p=0.004), I found thatexperimental warming accelerated the Day of Peak Flowering (DPF) (Table3.1). In P. radicatum and S. arctica ♂, I found changes in peak floweringwere only marginally significant (p=0.09). In S. longipes, it was not shiftedat all (p=0.41), but the senescence period was reduced. In other words,senescence occurred more quickly, but at approximately the same time ofthe season. The Day of Last Flower (DLF) was not altered in any speciesexcept for S. arctica ♂(p=0.0008) and D. integrifolia (p=0.0002), in whichit occurred earlier.I found that the length of the flowering season of warmed plots wassignificantly shortened in D. integrifolia, and lengthened for S. longipes(marginally), and P. radicatum. Warmed plots had similar densities of S.arctica catkins (p=0.7, p=0.41) and D. integrifolia flowers (p=0.8) through-out the season. However, I found that warmed plots had a significantlyhigher flower density of S. longipes (p=0.002) and P. radicatum (p=0.02)(Figure 3.3).3.3.2 Insect CommunityGeneralThe flying insect community was altered by OTC warming (p<0.001), bowltrap colour (p<0.001), and Day of Year (p<0.001). However, warming andbowl trap colour only accounted for only 3.2% and 2.8% of the variation, re-spectively, while day accounted for 50.4%. This indicates that while OTCshave an effect on the insect community composition, it is not very largein comparison to day-to-day variation over the growing season (see Figure3.4a). To illustrate the seasonal trend better, I roughly divided the com-munities found in Control and OTC bowl traps into Early, Mid, and Lateseason (Figure 3.4b). In general, the insect community follows a clockwisepattern, but the variance is much more constrained during the peak of flow-ering (“Mid-Season”). The pattern becomes much more variable after the“Late Season”, especially in the OTC bowl traps.453.3. ResultsUsing the bowl-trap data over the entire season, I found that OTCstended to reduce bowl catch rates of all insect families sampled, includingthe dominant floral visitors (pmcmc <0.001), but not by a large amounton a per-bowl basis. However, this gives only a general sense of decreases,and does not address differences between families or genera. When I testedday-to-day bowl catch rates using a GEE, I found that OTCs significantlyreduced the bowl catch rates (p<0.0001) for the five main families of insectsobserved, particularly for the Ichneumonidae, Noctuidae, and Syrphidae.This can be seen in Figures 3.5, 3.6, and Table 3.2, which show lower catchesof all the dominant families of flying insects. In particular, Syrphidae catcheswere reduced by 80%. Ichneumonidae and Noctuidae were reduced by 70%and 50%, respectively. Finally, Dolichopodidae and Musidae were reducedby 25% and 20%, respectively.Flower visitorsThe most dominant flower visitors were flies (Diptera) in the families Syr-phidae and Muscidae, followed by Bombus polaris (Hymenoptera: Apidae).Eupeodes curtus and Phaonia spp. were by far the most common visitors,followed by Platycheirus spp. and Drymeia spp. Bombus polaris, Melang-yna arctica, and Eupeodes nigroventris were also found in abundance (Table3.3).In terms of hourly visitation rates, Muscidae were the dominant flowervisitors during the early part of the flowering season (June to mid-July),while Syrphidae appeared more frequently during the late season (mid-Julyto August). Figure 3.7 displays the peaks in catch numbers graphically, andMuscidae seem to show a preference for D. integrifolia while Syrphidae seemto show a preference for P. radicatum (Table 3.4).3.3.3 Interaction RatesSyrphidaeNet catch rates (Catches/hr) of both Syrphidae and Muscidae on Dryas in-tegrifolia were correlated with Muscid population (Kendall τ=0.24, p=0.02)and Syrphid population (Kendall τ=0.55, p<0.0001). The catch rate on allother flowering plants showed no relationship with insect population. Inter-action rates (Net Catch Rate/Bowl Catch Rate) for Syrphidae were largelyunaffected by overall floral density, except for P. radicatum (p<0.0001) andStellaria longipes (p=0.03), which showed an increase in interaction ratesat higher flower densities.463.4. DiscussionI also found that increased insolation tended to decrease Syrphidae inter-action rates with P. radicatum (p=0.02), implying that the Syrphidae seekout blossoms of P. radicatum when insolation is lower, possibly for bask-ing (Figure 3.9). Insolation and air temperature are correlated (r2=0.39,p<0.0001), but I found that the model involving insolation was superior tothe one involving air temperature (∆AIC -4.1).MuscidaeFor the Muscidae, interaction rates were unaffected by overall floral avail-ability, and tended to show comparable interaction rates at any surroundingdensity of flowers. Their interaction rates were also largely unaffected bywind speed and insolation.Bombus polarisFor Bombus polaris, relatively few numbers of insects were netted in compar-ison to other pollinator clades (Figure 3.8). However, I found that similar tothe Syrphidae, B. polaris catches on P. radicatum were positively influencedby overall flower density (p=0.02), but not by wind speed or insolation. Ihad too few observations to examine S. longipes visitation by B. polaris,and B. polaris was not netted on the other dominant flower species, D. in-tegrifolia or S. arctica. Figure 3.9 shows the interaction rate for Papaverradicatum against flower density and insolation for both the Syrphidae andfor Bombus polaris.3.4 Discussion3.4.1 Flower AvailabilityDuring the summer of 2012, the OTC plots at the xeric shrub site were intheir 21st year of continuous warming treatment. 2012 also represented oneof the warmest years on record for the Canadian High Arctic, and also forAlexandra Fiord. Despite fairly consistent results across warmed Arctic sites(Klady, 2006; Elmendorf et al., 2012b; Oberbauer et al., 2013), interannualvariability is an important component of studying Arctic ecosystems (Jones,1995; Henry and Molau, 1997; Jones et al., 1997; Høye et al., 2007).Some authors have examined “snapshot phenology” of plots during thepeak flowering season (Wookey et al., 1995). Others have examined it atthe plant scale over the season (Wookey et al., 1995; Stenstro¨m et al., 1997;473.4. DiscussionTotland, 1999), but to my knowledge, there are no other Arctic studiesthat have examined flower availability at a plot scale continuously across anentire flowering season. However, several mid-latitude studies have examinedflower availability at varying time scales (Holzschuh et al., 2007; Elliott,2009). I found that OTCs altered flower availability in a variety of ways, butthat generally differences between OTC and control plots were associatedwith the timing of flowering, and sometimes with overall density of flowers.Because we did not arrive soon enough to begin measurements of S.arctica, I could not observe changes in the day of first flower or length ofseason. Hocking (1968) found that male plants tended to flower earlier thanfemale plants. Other authors have found that OTC warming did not increasethe day of first mature flower in plants of either sex of S. arctica (Jones et al.,1997), but this is subject to much interannual variation. I found that theOTC warming accelerated the day of peak flowering significantly in bothmale and female plants, which suggests that the day of first flower may alsobe earlier. The day of last flower was not altered, and overall flower density(per plot) was not altered, indicating that OTC warming affects the timingof flower production, but does not influence overall flower production in S.arctica.OTC warming accelerated the phenological development of Dryas in-tegrifolia. Other authors have reported similar results in Dryas octopetala(Wookey et al., 1993, 1995; Welker et al., 1997). My results were similarto results of the authors mentioned, in that the timing of first flower wasnot affected, but that advances in flowering became apparent soon after theearly season. I also found that the day of last flower was advanced by sixdays in OTCs, and that overall flowering season was shortened by four days,indicating that visiting insects that rely on D. integrifolia may be forced toalter their diet under future scenarios.OTC warming increased the per-meter flower density of Stellaria longipes,but did not alter the timing of flower production. Stellaria longipes has neverbeen studied in the context of OTC warming, so these can be taken as pre-liminary predictions of future S. longipes flower densities, at least at similarxeric sites. This is relevant for the pollinator community, as S. longipes ap-pears to be an important food source for the Syrphidae. Although I did notspecifically examine feeding habits, I did observe large numbers of Dipter-ans drinking nectar in the corolla of S. longipes. It produces large amountsof sweet nectar (Hocking, 1968), and Kevan (1973b) has shown that theirnectar is a food source of parasitic wasps in the Arctic (Chalcididae, Bra-conidae, Ichneumonidae). To my knowledge, there is currently no literatureon how S. longipes responds phenologically, structurally, or reproductively483.4. Discussionto induced warming, but this is an indication that it will likely respondpositively under future warmed climates.Warmed plots significantly shifted the flowering in Papaver radicatumearlier by six days on average, and showed an increased length of the overallflowering season by eight days. I also found that the overall density ofPapaver radicatum flowers was increased in the warmed plots, but withoutmeasurements of plants per plot it is impossible to know whether this isfrom increased P. radicatum establishment, or rather increased reproductiveeffort. Klady et al. (2011) found that P. radicatum did not increase per-plant reproductive effort when experimentally warmed, which I also found inChapter 2. My measurements were in flowers/m2 while Klady et al. (2011)measured flower biomass/plant, but I suspect that some of the increase inflower density during the growing season is due to greater establishment inthe warmed plots.3.4.2 Insect CommunityInsects and OTCsWhile OTCs have been found to drive changes in Arctic plant communitiesby passive warming (Hudson and Henry, 2010; Elmendorf et al., 2012a),these changes are not expected to be seen to the overall insect commu-nity, because any accelerated emergence or feeding patterns of insects withinOTCs would be dwarfed by the greater population of “unwarmed insects”from outside the chambers. Some authors have studied the behaviour ofHemiptera and larval Lepidoptera within OTCs, but this was only possiblethrough use of a corral which confined the insects within the OTCs , and isessentially impossible to do with flying insects. Bocher (1996), Stenstro¨met al. (1997), and Jones et al. (1997) all recognized that passive warmingexperiments such as OTCs may exclude insects from plants, and I have pre-sented quantitative evidence to suggest that they were correct. Primarilythis affects low-flying, medium-sized insects such as wasps and flies. I haveobserved flies and wasps encounter the angled plexiglass of the OTCs, rightthemselves after the collision, and fly away in a different direction after ashort period. This appears to cause substantial reductions in catch rates,with an 80% reduction in Syrphidae catches and a 70% reduction in Ich-neumonidae catches (see Results, and Table 3.2). However, the impacts ofthis exclusion are not well known. Of the five dominant insect families, Iwill review possible confounding effects found within OTCs that may beunrepresentative of future Arctic plant communities.493.4. DiscussionThe reduction of Muscidae and Syrphidae is significant mainly in termsof reduction to pollination within OTCs. Both families of Diptera havebeen shown to be the dominant floral visitors in most Arctic regions (Ke-van, 1972b; Pont, 1993), and their reduction in availability to plants withinOTCs may provide unrealistic estimates of pollination rates within futureecosystems (Alatalo and Totland, 1997). However, reduction in visitation ismore likely to affect the early season pollen-risk strategists such as Saxifragaoppositifolia (Molau, 1993), as plants that flower later tend to be associatedwith lower rates of self-incompatibility. I report similar results in Chapter2, which indicates that late-flowering plants such as P. radicatum are fairlyrobust to changes in pollination.The behaviour of other Arctic insects within warmed environments hasbeen studied relatively little. Because Arctic insects are highly dependentupon their surrounding for thermoregulation (Danks, 1996, 2004), ground-dwelling invertebrates, particularly Arachnids, Collembola, Hemiptera, andlarval Lepidoptera can be found in greater densities within OTCs (Richard-son et al., 2002; Gillespie et al., 2013). They will stay inside of OTCs whengiven the opportunity, and Ring (2001) has found evidence that smallerDipterans (Nematocera, Chironomidae, Empididae) also prefer OTCs. Ifound that OTCs influence the composition of the insect community on aday-to-day basis. In particular, larger insects such as Muscidae, Ichneu-monidae, and Dolichopodidae are found less within OTCs. This is particu-larly important because Ichneumonidae and Dolichopodidae are ecologicallyimportant in many ecosystems as parasitoids and predators, respectively(Jervis and Kidd, 1986; Gelbic˘ and Olejn´ıc˘ek, 2011). In Arctic insect com-munities this is no less true. For example, at the Alexandra Fiord site,the dominant Lepidopteran grazer Gynaephora groenlandica is subjected tomortality rates of up to 95% due to Ichneumonid and Tachinid parasitoids(Kukal and Kevan, 1987).From this, I imply that OTCs likely cause artificial trophic imbalancesby excluding ambient predatory or parasitic insects, while promoting theactivity of smaller soil, plant, or ground-dwelling insects. Because I wasprimarily interested in pollinating insects within OTCs, smaller unidentifiedarthropods such as Nematocera, small Braconidae, and Acari (Mites) werelargely overlooked for the purpose of this study, but may be highly influentialon the Hemiptera (Aphids) that feed on the plants within the OTCs. Dolleryet al. (2006) have shown that OTC warming increases populations of soil-dwelling mites, Chironomid larvae, and plant-feeding aphids, while reducingHymenopteran parasitoids. While not explicitly sampling for parasitoids,Gillespie et al. (2013) also found that aphids increased dramatically in the503.4. Discussionfirst years that OTCs were present, but then were reduced to normal levels inlater years. They attribute this to a corresponding increase in soil-dwellingpredators such as mites and spiders, but it is uncertain if flying predatorswould increase their growth and phenology in the same way in future climatescenarios. Much of the emergence and life history of Dolichopodidae andother predatory Arctic insects is essentially unknown, and should be a topicof future study.Flower-visiting InsectsArctic flowering plants have relatively robust pollination strategies, and canproduce seeds in absence of pollinating insects (Chapter 2). I suspect thatthe insects are more heavily dependent on the food and energy gained fromflowers.I found that the dominant visitors of flowers are overwhelmingly Dipteraof the family Muscidae and Syrphidae, followed by the Bombus polaris (Hy-menoptera: Apidae). While B. polaris is well-noted as an Arctic and alpinepollinating species (Bergman et al., 1996; Philipp et al., 1996), it is relativelyuncommon in terms of actual catch rates. The per-visit efficacy of B. polariscompared to the common Arctic Syrphidae and Muscidae such as Eupeodescurtus or Phaonia spp. is not known, but is assumed to be higher in B.polaris simply due to body size and foraging behaviour (Sahli and Conner,2007). While it is generally accepted that bees are more effective pollinatorsthan flies (Herrera, 1987), visitation by flies can provide ample pollination,provided enough visitors are available (Mesler et al., 1980; Berenbaum et al.,2007), and this appears to be the case here. Other authors have identifiedthat flies are often an over-looked component of many pollination systems(Larson et al., 2001; Kearns, 2001; Lundgren and Olesen, 2005), and I sug-gest that further study should be devoted to them, especially in Arctic andalpine ecosystems that are relatively free of bee pollinators.I did not find evidence to suggest that S. arctica are dependent on in-sects for pollination, as I observed very low rates of visitation to female S.arctica plants. The female catkins are also capable of making seeds (albeitlower numbers) when insects are excluded from them (Chapter 2). However,insects may be more dependent on them for a source of food, as queens ofBombus polaris were observed gathering pollen from male S. arctica early inthe season when no other pollen sources were available. The queens made nosystematic attempt to visit female catkins, and concentrated their foragingon pollen-laden male catkins (>30 consecutive male catkins were visited,and approximately equal numbers of male/female catkins were available;513.4. Discussionpersonal observation). By the time that Muscidae and Syrphidae emerge,S. arctica are largely senesced and non-receptive to pollen. It is possiblethat early-season visits are important for S. arctica (Molau, 1993), but I didnot find large numbers of insects present that would facilitate pollinationother than the occasional B. polaris.I observed Syrphidae and Muscidae feeding from the deep nectaries of D.integrifolia, as well as feeding directly on the pollen from stamens. Theserepresent an important food source for insect visitors, especially becausethey have some of the most nutrient-rich nectar of all Arctic flowering plants(Hocking, 1968), but without detailed observations it is difficult to assesswhether pollen or nectar is the main attractant for the visiting insects. Hock-ing (1968) found that within the same plant species, nectar in Arctic flowersis much more concentrated at high latitudes, and that equal quantities werepresent in covered and uncovered flowers at high latitudes. He used thisto show that between-plant competition for pollinators is higher in Arcticplants than in mid-latitude plants (Hocking, 1968). Kevan (1973a) viewsnectar in Arctic flowers as more of an attractant for Lepidoptera, and Chi-ronomidae, but has noted that about 40% of the insects in the Arctic havebeen found imbibing nectar.Catch rates on D. integrifolia scaled with insect abundance (bowl trapcatches) for the Syrphidae and Muscidae. Large numbers of Muscidae visitedD. integrifolia, most likely to forage for nectar and pollen reserves, as wellas for warmth. However, I found that for both catch rates and interactionrates were not related to flower abundance for D. integrifolia, implying thattheir resources are not critical for their insect visitors.The more likely attractant for most flower visitors is pollen and warmth.Papaver radicatum, does not produce nectar, yet I caught almost twice thenumber of visitors on P. radicatum than D. integrifolia. Some of this isdue to timing of insect and flower emergence, but even at low densities, P.radicatum appear to be the most attractive blossom for insect visitors, es-pecially the Syrphidae. Pollen production and availability in D. integrifoliaand P. radicatum has not been quantified, but corolla temperatures of D.integrifolia are much less than P. radicatum (Kevan, 1972b, 1975). Kevan(1973a) discussed floral attraction in Arctic plants, and indicated that flowersize is a key feature of insect attraction in Arctic plants because it is linkedto both visual attraction and to the thermal regime within the corolla. I alsofound that interaction rates of P. radicatum with the Syrphidae decreasedat higher rates of insolation, indicating that they likely visit the flowers atlower temperatures to gain thermal energy. This is contrary to the results ofTotland (1994) who found that visitation rate was correlated with temper-523.4. Discussionatures. This is likely due to climatic differences between Alexandra Fiord,Nunavut and Mount Sandalsnut, Norway. Alexandra Fiord is in the HighArctic (78.9◦ N), and diurnal insolation and temperature are more constantthan southern latitudes, even during the “night time”, whereas Mt. Sandal-snut (60.4◦ N) is subject to large variation in solar radiation and temperatureover the course of the day.I also found that interaction rates for Syrphidae and the catch rate forBombus polaris tended to scale with abundance for flowers of P. radicatum,implying that they are limiting resources (sensu MacArthur (1972)) for theseinsects. Syrphidae are known to feed on pollen during the time of egg forma-tion (Kevan, 1973a). Bombus polaris queens and workers forage for nectarpollen to provide for their brood, and have been shown to feed on all theflowers regarded in this study (Richards, 1973). Richards (1973) reportedthat the queens of B. polaris tended to visit Salix arctica and Saxifraga op-positifolia, while the workers visited Dryas integrifolia, Pedicularis capitataand Cassiope tetragona. Interaction rates for the Syrphidae with Stellarialongipes also scaled with floral abundance, implying they are also limitingresources. S. longipes is known to produce large amount of sweet-smellingnectar and pollen. However, the flowers are much smaller and shorter thanthe flowers of P. radicatum, making them less attractive. They also providevery little thermal benefits in comparison to P. radicatum. In this case, it ismost likely that pollen (and possibly nectar) are limiting resources for theSyrphidae.One of the reasons for the lack of B. polaris visits to anything but P.radicatum at my site may have been that the composition of available flowersat other sites was more attractive to B. polaris, but this is beyond the scopeof my study. Other authors have found that insolation and air temperaturetended to increase visitation in other Arctic species of Bombus (Lundberg,1980; Bergman et al., 1996), but I failed to find any trend, possibly because Ifocused on overall patterns rather than diurnal patterns in visitation, as wellas climatic differences between study sites. Both Bergman et al. (1996) andRichards (1973) found that Bombus queens were able to forage for longerat lower temperatures, simply because of the higher internal temperaturesassociated with their larger body size. I did not catch any queens of B.polaris. Because of the relatively small sample size of B. polaris netted(18 total), and the lack of overall population estimates, the relationship ofbetween overall visitation and flower density is tenuous, and merits furtherstudy.533.5. Conclusion3.5 ConclusionFlower production at a per-plot basis was altered strongly for most flow-ering plant species, and has implications for the structure of future plant-pollinator interactions. I demonstrate that:• OTCs significantly change the structure of the flowering plant com-munity on a day-to-day basis.• The day of peak flowering for Salix arctica and Dryas integrifoliatended to be earlier in the warmed plots.• The length of the overall flowering season was shorter for Dryas inte-grifolia and longer for Papaver radicatum in the warmed plots.• Flower density was higher for Stellaria longipes and Papaver radicatumin the warmed plots.Insect activity was also altered within OTCs, particularly for the domi-nant visitors. I show that:• OTCs significantly change the structure of the flying insect commu-nity on a day-to-day basis, and likely exclude insect predators andparasites.• Bowl-trap catches of nearly all groups of flying insects were signifi-cantly reduced over the growing season in the warmed plots.• The most dominant insect visitors were Syrphidae and Musicidae, witha few Bombus polaris.• Syrphid flies and Bombus polaris demonstrated temperature and availability-dependent visitation to Papaver radicatum. They tended to be caughtat higher rates when flower density was high and temperatures werelow.OTCs have been shown to strongly alter the dynamics of flowering inArctic plants (Wookey et al., 1993; Welker et al., 1997; Mølgaard and Chris-tensen, 1997; Stenstro¨m et al., 1997; Alatalo and Totland, 1997; Klady et al.,2011), and similarly I found that OTCs alter the timing and magnitude offlowering in Arctic plants. While this is of interest in the context of futuregrowth and range extensions, it is also of interest in the context of futureecological interactions. Climate change is expected to result in trophic mis-matches (Post and Forchhammer, 2008; Post et al., 2009), including insect543.5. Conclusionoutbreaks (Jepsen et al., 2008, 2011). It has been suggested that changesin insect emergence may result in plant/pollinator mismatches (Donnellyet al., 2011; Straka, 2012) and it is possible that mismatches could occuracross the Arctic (Høye and Forchhammer, 2008; Hegland et al., 2009; Høyeet al., 2013). However, the extent of this is not known, given that bothinsect and flower phenology in the Arctic tends to be related to snowmelt(Danks et al., 1994; Høye and Forchhammer, 2008; A´vila-Jime´nez and Coul-son, 2011; Oberbauer et al., 2013). Long-term records of insect emergenceand phenology are lacking, especially in the Arctic, but Høye et al. (2013)have recently demonstrated that a phenological mismatch between Arcticplants and pollinators is indeed occurring, and show that the consequencesare more severe for late-emerging flower visitors. Their observations do notinclude actual visitor observations, but are still valuable in terms of com-munity data.I found that Papaver radicatum flowers are a limiting resource to visitingBombus polaris and Syrphidae. Additionally, I found evidence that Syrphi-dae visit flowers for warmth, as well as pollen resources. This adds quanti-tative evidence to the observations of Hocking and Sharpin (1965), Kevan(1975), Totland (1996), and others (Philipp et al., 1996; Danks, 2004) whoidentified flower microclimate as an important insect attractant along withscent, colour, and food resources. Considering that the corollas offer insectsa food-laden microclimate on par with that of an OTC, it is no wonder thatthey seek out these blossoms! Lo¨ve (1969), Kevan (1972a), and Mølgaard(1989) and have studied how flower colour, temperature, and heliotropismare related in P. radicatum, but have not explicitly linked their studies to in-sect attraction. Papaver radicatum, and other similar heliotropic cup-shapedflowers, should be further studied at additional Arctic and alpine sites tounderstand how selection operates on warmth and colour.553.6. Figures and Tables3.6 Figures and TablesFigure 3.1: Stellaria longipes Goldie subsp. longipes produce 5-7mm 10-petaled bisexual white flowers. They produce sweet smelling nectar andyellow pollen. The are visited by Diptera and occasionally Bombus polaris.563.6. Figures and TablesFigure 3.2: NDMS ordination of flower communities over the growing seasonin warmed and control plots.573.6. Figures and TablesFigure 3.3: Daily flower densities for a) Salix arctica male, b) Salix arcticafemale, c) Dryas integrifolia, d) Stellaria longipes, and e) Papaver radicatum,in both warmed and control plots. 14 warmed and 14 control plots weremonitored.583.6. Figures and TablesFigure 3.4: NDMS ordination of insect communities showing a) changes inNMDS scores over the season in warmed and control plots, and b) groupedinto early, mid and late season communities. Abundances were averaged foreach day between bowl colours, and split by warming treatment.593.6. Figures and TablesFigure 3.5: Daily bowl catches of a) Muscidae, b) Ichneumonidae, c)Dolichopodidae, d) Noctuidae, and e) Syrphidae over the flowering seasonin warmed (OTC) and control plots.603.6.FiguresandTablesFigure 3.6: Total catch of all insects over the growing season in warmed (OTC) and control plots.613.6. Figures and TablesFigure 3.7: Hourly catch rate of a) Musidae, b) Syrphidae, and c) Bombusvistors on flowers of the four main plant species.623.6.FiguresandTablesFigure 3.8: Total net catches of insect visitors outside of OTCs over the entire flowering season.633.6.FiguresandTablesFigure 3.9: Factors influencing the Papaver radicatum interaction rate (Netting Catch Rate/Population Estimate)for the a,b) Syrphidae and c) Bombus polaris. P-value for B. polaris uses a Kendall rank-test (r2 not defined).643.6.FiguresandTablesTable 3.1: Results of tests for flower response to warming, along with p-values for tests of significant difference(Student’s T or Kruskal-Wallis). Changes in the day of first flower (DFF), peak flower (DPF), last flower (DLF),and season length (Len.) of all species are shown, as well as overall changes in density (Dens (flowers/m2). ForDFF, DPF, and DLF, numbers indicate how many days the event was shifted, with negative numbers indicatingan earlier date. Len. represents the change in length of the flowering season.Plant Species DFF p DPF p DLF p Len. p Dens. pSalix arctica(♂) n/a n/a -3.7 0.0008 -6.0 0.0008 n/a n/a -0.70 0.70Salix arctica(♀) n/a n/a -4.6 0.0005 -1.1 0.37 n/a n/a -3.22 0.41Dryas integrifolia -1.7 0.18 -3.7 0.004 -6.0 0.0002 -4.2 0.04 -0.97 0.80Stellaria longipes -4.8 0.05 -1.6 0.41 0.3 0.63 5.1 0.09 4.79 0.002Papaver radicatum -5.9 0.0007 -2.7 0.09 1.8 0.61 7.7 0.006 1.51 0.02653.6. Figures and TablesTable 3.2: Differences in average catches between Control and OTC BowlTraps over the five most abundant insect families (all other families hadoverall numbers too low to test using a GEE). Numbers under Treatmentare in catches per pay over the entire growing season.GroupTreatmentp-value Total CatchesControl OTCDiptera: Muscidae 58.33 47.25 <0.0001 2188Hymenoptera: Ichneumonidae 25.27 7.14 <0.0001 661Diptera: Dolichopodidae 4.56 3.38 <0.0001 189Lepidoptera: Noctuidae 5.86 2.91 <0.0001 135Diptera: Syrphidae 2.18 0.42 <0.0001 56Table 3.3: Abundance of netted insect species over the entire growing season.Order: Family Species ♀ ♂ TotalDiptera: Syrphidae Eupeodes curtus (Hine, 1922) 180 0 180Diptera: Muscidae Phaonia spp. 37 63 100Diptera: Syrphidae Platycheirus spp. 34 0 34Diptera: Muscidae Drymeia spp. 27 5 32Hymenoptera: Apidae Bombus polaris Curtis, 1835 19 0 19Diptera: Syrphidae Melangyna arctica (Zetterstedt, 1838) 12 0 12Diptera: Syrphidae Eupeodes nigroventris (Fluke, 1933) 11 0 11Diptera: Empididae Rhamphomyia spp. 3 1 4Diptera: Culicidae Aedes spp. 1 2 3Diptera: Syrphidae Eupeodes lapponicus (Zetterstedt, 1838) 3 0 3663.6. Figures and TablesTable 3.4: All flower-insect interactions observed throughout the floweringseason, ranked by frequency.Flowering PlantInsect Visitor PapaverradicatumDryasintegrifoliaStellarialongipesSalixarctica♀Salixarctica♂SaxifragatricuspidataCerastiumalpinumDrabalacteaSaxifraganivalisTotalDipteraEupeodes curtus 126 11 35 1 2 3 2 0 0 180Phaonia spp. 25 72 0 1 1 0 1 0 0 100Platycheirus spp. 27 6 0 0 0 0 0 1 0 34Drymeia spp. 23 9 0 0 0 0 0 0 0 32Melangyna arctica 11 0 0 0 1 0 0 0 0 12Eupeodes nigroventris 8 0 2 0 0 0 0 0 1 11Rhamphomyia spp. 1 3 0 0 0 0 0 0 0 4Eupeodes lapponicus 3 0 0 0 0 0 0 0 0 3Aedes spp. 0 2 0 0 1 0 0 0 0 3Dolichopus spp. 1 0 0 0 0 0 0 0 0 1Helophilus groenlandicus 1 0 0 0 0 0 0 0 0 1Helophilus lapponicus 1 0 0 0 0 0 0 0 0 1Platycheirus groenlandicus 0 0 0 1 0 0 0 0 0 1Mycetophilidae 1 0 0 0 0 0 0 0 0 1Scathophagidae 0 0 0 1 0 0 0 0 0 1Chironomidae 0 0 0 1 0 0 0 0 0 1Hym.Bombus polaris 17 0 1 0 0 0 0 0 0 18Ichneumonidae 0 1 0 0 0 0 0 0 0 1Lepidoptera Psychophora sabini 0 2 0 0 0 0 0 0 0 2Agriades glandon 0 0 0 1 0 0 0 0 0 1Lycaena phlaeas 0 1 0 0 0 0 0 0 0 1Sympistis nigrita 0 1 0 0 0 0 0 0 0 1Boloria chariclea 0 1 0 0 0 0 0 0 0 1Total 245 109 38 6 5 3 3 1 1 41167Chapter 4ConclusionsExperimental studies on flower pollination in the High Arctic are rare (Ke-van, 1972b), and very few of them deal explicitly with the topic of warmingand pollination changes (Stenstro¨m and Molau, 1992). In this study, I exper-imentally altered pollination of flowering plants, and demonstrated that anypollination deficits induced by open-top chambers (OTCs) are likely to beinsignificant in an Arctic ecosystem. Salix arctica demonstrated increasedseed production with hand-pollination, and decreased seed production wheninsects were excluded, but OTCs do not appear to alter this in any signifi-cant way. Warming and pollination did not alter S. arctica germination ratessignificantly. Dryas integrifolia altered its flower production under warmedconditions, producing more flowers/g of above-ground biomass (AGB) whenwarmed. I also found that germination rates of D. integrifolia increasedwhen pollinated outside of the warming treatments, or when warmed by anOTC, but that pollination and warming effects are not additive. Papaverradicatum proved resilient to changes in pollination and OTC warming, anddid not alter its flower production, seed production, or germination rates inresponse to any treatment. By demonstrating this, I further validate theuse of OTCs as a technique for studying climate change effects on reproduc-tion in Arctic plant communities, and suggest that reproductive data fromOTC-warmed Arctic plant communities are likely representative of futurewarmed climates.The only exception to this may be early flowering self-incompatiblyplants such as Saxifraga oppositifolia. (Stenstro¨m et al., 1997) showed thateven minor changes in S. oppositifolia reproductive phenology occur at suchearly periods that pollination is reduced due to mismatches between pol-linator emergence and flowering time. Other authors have implied thatother early-flowering plants may be subjected to artificial pollen limitationwithin OTCs (Stenstro¨m and Molau, 1992; Jones et al., 1997). Both insectand flower emergence in Arctic ecosystems is generally tied to snowmeltrather than temperature (Molau, 1993; Walker et al., 1999; Totland andAlatalo, 2002; Dollery et al., 2006; Høye et al., 2007; Høye and Forchham-mer, 2008), but Høye et al. (2013) identified that climate-induced changes in68Chapter 4. Conclusionsphenology tend to shift phenology in late-flowering populations more thanearly-flowering populations.I also examined the temporal structure of an Arctic plant-pollinator net-work, showing that the Syrphidae and the Muscidae constituted the mostcommon floral visitors at the site, and that Bombus polaris plays a muchreduced role (in terms of catches/hr). I showed that Dryas integrifolia doesnot appear to constitute a limiting resource for insect visitors, while Papaverradicatum is much more important to insect visitors, especially the Syrphi-dae. Using bowl-trapping, I demonstrated that OTCs significantly reducedcatches of visiting insects, as well as other common flying insects. However,in the context of the results of experimental pollination manipulation, thisreduction is probably not ecologically significant (except for early-floweringplants). Finally, I showed that OTC warming tended to shorten the flow-ering season for D. integrifolia, while lengthening the flowering season andincreasing the flower density of P. radicatum. This implies that insect visi-tation could be enhanced later in the season under a future warmed climate,but much of this depends on the timing and duration of pollinator emer-gence.Whether shifts in plant phenology will actually decouple Arctic plant-pollinator networks has yet to be seen, as most previous studies did notconduct actual netting or visitor observations, but used bowl or pit trapcatch rates as a proxy for visitation. Bowl-trapping and pit-trapping arerelatively easy ways to collect insects but yield no actual information onplant-insect interactions, and are more useful as a measure of relative pop-ulation size. I did not find that insect population was a proxy of catch ratesfor any species except Dryas integrifolia, implying that visitation rates de-pend on other factors such as floral availability and environmental variables.Thus, future studies should examine actual visitation rates to better under-stand pollination rates, and avoid using bowl trapping rates as a “proxy ofa proxy of pollination”.Netting was not conducted inside of OTCs because of the danger it posedto the long-term study plants, the instrumentation present in the plots, aswell as the inherent inaccuracy in comparing large area captures to smallarea captures. Measuring the actual visitation rate of insects to flowers in-side and outside of OTCs is highly tedious work involving manually observ-ing clusters of flower for long periods of time, recording visit rates of variousinsects. Video capture of visiting insects is a relatively new field, but hasbeen used by some authors to investigate detailed insect visitation patterns(Steen and Thorsdatter, 2011; Lortie et al., 2012). I conducted time-lapsephotography (5-second interval) of groups of flowers both inside and outside69Chapter 4. Conclusionsthe OTCs, using GoPro R©HERO 1 cameras with an attached battery pack.These cameras were able to capture 12 hours of continuous data, and I con-ducted observations over the span of the growing season, concurrently withnetting and bowl trapping. I have also devised an image-processing programthat is able to determine threshold ed changes in intensity associated withdark-coloured insects landing on a brightly-coloured flower. The problem offlower motion due to wind is difficult to handle, but there are several classi-fication schemes that may be able to handle this process. However, due totime constraints I was not able to complete an analyses of this method, andit is not part of this thesis.In addition to monitoring plant-pollinator interactions, an effort shouldbe made to further understand the ecology of Arctic insect communities,rather than solely descriptive studies. Aphids, midges, gall-wasps, stem-borers, leaf-miners, thrips, and caterpillars are all part of the Arctic herbiv-orous insect community, but their role has been far less studied than thatof ungulate or avian grazers. Predatory flies and parasitic wasps most likelyplay a large role in regulating their populations (Kukal and Kevan, 1987).Spiders, in particular, are well-noted in Arctic arthropod collections (Leechand Ryan, 1972), but their ecological role has been studied to a very smallextent. Hodkinson et al. (2001) links their activities to establishment andearly plant community development, they form the base of the food web formany Arctic birds, and they are seen hunting insects throughout all partsof the snow-free season (personal observation). Clearly they play a role inthe overall community, but this role is almost completely unstudied.The effects of OTCs on plant growth and development are fairly well-established (Wookey et al., 1995; Jones et al., 1997; Welker et al., 1997;Mølgaard and Christensen, 1997; Stenstro¨m et al., 1997), along with changesin community structure (Wookey et al., 1993; Hudson and Henry, 2010;Elmendorf et al., 2012a). However, changes in life history and populationdynamics are not well understood in the context of OTC warming. Assuggested by Cooper et al. (2004) and Mu¨ller et al. (2011), germinationshould be examined in the context of OTC warming to examine how inducedwarming changes rates of establishment. This is a key piece of ecologicalknowledge that has heretofore been unstudied. Seed germination studiesin the context of Arctic warming often yield dissimilar results, which canbe dependent on year-to-year and site-to-site variation (Molau and Shaver,1997; Welker et al., 1997; Molau, 1997; Mølgaard and Christensen, 1997;Mu¨ller et al., 2011; Klady et al., 2011). Milbau et al. (2009) found thatseed germination occurred earlier, but not at higher rates, in warmed soils,implying that future seed establishment will take place earlier in the year.70Chapter 4. ConclusionsDepending on local moisture conditions, this has the potential to increase ordecrease survival, as most seedling mortality occurs due to summer drought(Bliss, 1958; Bell and Bliss, 1980).While changes to established plant communities are of great importance,new habitats will become available for colonization by plants as the Arcticwarms, especially in the High Arctic. How these ranges will be colonizeddepends on reproduction and establishment by adjacent plants. In this way,it is important to understand the mechanisms that lead to establishment,including pollination, seed formation, and dispersal. Insects have the po-tential to alter patterns of pollination, especially in early-flowering plants.Thus, the continued study of plant-pollinator relationships is important tounderstanding aspects of future Arctic ecosystems.71BibliographyAlatalo, J. and Totland, O. 1997. Response to simulated climatic changein an alpine and subarctic pollen-risk strategist, Silene acaulis. GlobalChange Biology, 3(S1):74–79.Alexeev, V. A., Esau, I., Polyakov, I. V., Byam, S. J., and Sorokina, S.2012. Vertical structure of recent Arctic warming from observed data andreanalysis products. Climatic change, 111(2):215–239.Anderson, D. R. 2008. Model based inference in the life sciences: a primeron evidence. Springer.A´vila-Jime´nez, M. and Coulson, S. 2011. Can snow depth be used to predictthe distribution of the high Arctic aphid Acyrthosiphon svalbardicum(Hemiptera: Aphididae) on Spitsbergen? BMC Ecology, 11(1):25.Bates, D., Maechler, M., Bolker, B., and Walker, S. 2013. lme4: Linearmixed-effects models using Eigen and S4. R package version 1.0-4.Bell, K. L. and Bliss, L. C. 1980. Plant reproduction in a high Arcticenvironment. Arctic and Alpine Research, 12(1):1–10.Berenbaum, M., Bernhardt, P., Buchmann, S., Calderone, N., Goldstein, P.,Inouye, D., Kevan, P., Kremen, C., Medellin, R., Ricketts, T., and others.2007. Status of pollinators in North America. The National AcademiesPress.Bergman, P., Molau, U., and Holmgren, B. 1996. Micrometeorological im-pacts on insect activity and plant reproductive success in an alpine envi-ronment, Swedish Lapland. Arctic and Alpine Research, 28(2):196–202.Billings, W. D. and Mooney, H. A. 1968. The ecology of Arctic and alpineplants. Biological Reviews, 43(4):481–529.Bjorkman, A., Henry, G., and Vellend, M. 2013. Migration potential oftundra plant species in a warming Arctic: responses of southern ecotypes72Bibliographyof three species to experimental warming in the high Arctic. In EGUGeneral Assembly Conference Abstracts, volume 15, page 548.Bliss, L. 1958. Seed germination in Arctic and alpine species. Arctic,11(3):180–188.Blok, D., Heijmans, M., Schaepman-Strub, G., Kononov, A., Maximov, T.,and Berendse, F. 2010. Shrub expansion may reduce summer permafrostthaw in Siberian tundra. Global Change Biology, 16(4):1296–1305.Bocher, J. 1996. Further insect studies. In The ITEX Manual, chapter 10,pages 37–38. Danish Polar Center, Copenhagen, 2nd edition.Bolker, B. M., Brooks, M. E., Clark, C. J., Geange, S. W., Poulsen, J. R.,Stevens, M. H. H., and White, J.-S. S. 2009. Generalized linear mixedmodels: a practical guide for ecology and evolution. Trends in ecology &evolution, 24(3):127–135.Breen, K. and Le´vesque, E. 2006. Proglacial succession of biological soilcrusts and vascular plants: biotic interactions in the High Arctic. Cana-dian Journal of Botany, 84(11):1714–1731.Brochmann, C., Edwards, M. E., and Alsos, I. G. 2013. The Balance ofNature and Human Impact, chapter The dynamic past and future of Arcticplants: climate change, spatial variation, and genetic diversity, pages 133–152. Cambridge University Press.Burnham, K. P. and Anderson, D. R. 2002. Model selection and multi-modelinference: a practical information-theoretic approach. Springer.CANPOLIN 2009. CANPOLIN Sampling Protocols.Chapin, F. S. 1983. Direct and indirect effects of temperature on arcticplants. Polar Biology, 2:47–52.Chernov, Y. 1995. Diversity of the Arctic terrestrial fauna. In Chapin, F. S.and Ko¨rner, C., editors, Arctic and alpine biodiversity: patterns, causesand ecosystem consequences, volume 113 of Ecological Studies, chapterDiversity of the Arctic Terrestrial Fauna, pages 81–95. Springer BerlinHeidelberg.Chernov, Y. I. 1966. Kompleks antofil’nykh nasekomykh v tundrovoy zone[Complex of anthophilous insects in the tundra zone]. Voprosy geografii,69(1-2):279–307.73BibliographyChernov, Y. I. 1985. The Living Tundra. Studies in polar research. Cam-bridge University Press, Cambridge; New York.Chinnappa, C., Donald, G. M., Sasidharan, R., and Emery, R. N. 2005. Thebiology of Stellaria longipes (Caryophyllaceae). Botany, 83(11):1367–1383.Cooper, E. J., Alsos, I. G., Hagen, D., Smith, F. M., Coulson, S. J., and Hod-kinson, I. D. 2004. Plant recruitment in the high Arctic: seed bank andseedling emergence on Svalbard. Journal of Vegetation Science, 15(1):115–124.Corbett, A. L., Krannitz, P. G., and Aarssen, L. W. 1992. The influence ofpetals on reproductive success in the arctic poppy (Papaver radicatum).Canadian Journal of Botany, 70(1):200–204.Culley, T. M., Weller, S. G., and Sakai, A. K. 2002. The evolution of windpollination in angiosperms. Trends in Ecology & Evolution, 17(8):361–369.Dang, T. T. T. and Chinnappa, C. 2007. The reproductive biology of Stel-laria longipes Goldie (Caryophyllaceae) in North America. Flora - Mor-phology, Distribution, Functional Ecology of Plants, 202(5):403–407.Danks, H., Kukal, O., and Ring, R. 1994. Insect cold-hardiness: insightsfrom the Arctic. Arctic, 47:391–404.Danks, H. V. 1992. Arctic Insects as Indicators of Environmental Change.Arctic, 45(2):159–166.Danks, H. V. 1996. The wider integration of studies on insect cold-hardiness.European Journal of Entomology, 93:383–404.Danks, H. V. 2004. Seasonal Adaptations in Arctic Insects. Integrative andComparative Biology, 44(2):85 –94.Darwin, C. 1859. The origin of species by means of natural selection. JohnMurray.Darwin, C. 1862. The various contrivances by which orchids are fertilisedby insects. John Murray.Dawson, T. E. and Bliss, L. 1989a. Intraspecific variation in the water rela-tions of Salix arctica, an Arctic-alpine dwarf willow. Oecologia, 79(3):322–331.74BibliographyDawson, T. E. and Bliss, L. 1989b. Patterns of water use and the tissue waterrelations in the dioecious shrub, Salix arctica: the physiological basis forhabitat partitioning between the sexes. Oecologia, 79(3):332–343.Densmore, R. and Zasada, J. 1983. Seed dispersal and dormancy patternsin northern willows: ecological and evolutionary significance. CanadianJournal of Botany, 61(12):3207–3216.Dollery, R., Hodkinson, I. D., and Jo´nsdo´ttir, I. S. 2006. Impact of warm-ing and timing of snow melt on soil microarthropod assemblages associ-ated with Dryas-dominated plant communities on Svalbard. Ecography,29(1):111–119.Donnelly, A., Caffarra, A., and ONeill, B. F. 2011. A review of climate-driven mismatches between interdependent phenophases in terrestrial andaquatic ecosystems. International Journal of Biometeorology, 55(6):805–817.Downes, J. A. 1964. Arctic Insects and Their Environment. The CanadianEntomologist, 96:279–307.Elberling, H. and Olesen, J. M. 1999. The structure of a high latitude plant-flower visitor system: the dominance of flies. Ecography, 22(3):314–323.Elliott, S. E. 2009. Subalpine bumble bee foraging distances and densities inrelation to flower availability. Environmental entomology, 38(3):748–756.Elmendorf, S. C., Henry, G. H., Hollister, R. D., Bjo¨rk, R. G., Boulanger-Lapointe, N., Cooper, E. J., Cornelissen, J. H., Day, T. A., Dorrepaal, E.,Elumeeva, T. G., et al. 2012a. Plot-scale evidence of tundra vegetationchange and links to recent summer warming. Nature Climate Change,2(6):453–457.Elmendorf, S. C., Henry, G. H. R., Hollister, R. D., Bjrk, R. G., Bjorkman,A. D., Callaghan, T. V., Collier, L. S., Cooper, E. J., Cornelissen, J.H. C., et al. 2012b. Global assessment of experimental climate warmingon tundra vegetation: heterogeneity over space and time. Ecology Letters,15(2):164–175.Epstein, H. E., Walker, M. D., Chapin III, F. S., and Starfield, A. M. 2000.A transient, nutrient-based model of Arctic plant community response toclimatic warming. Ecological Applications, 10(3):824–841.75BibliographyEriksson, O. and Fro¨borg, H. 1996. ” Windows of opportunity” for recruit-ment in long-lived clonal plants: experimental studies of seedling estab-lishment in Vaccinium shrubs. Canadian Journal of Botany, 74(9):1369–1374.Faegri, K. and Van der Pijl, L. 1979. The Principles of Pollination Ecology.Pergamon international library of science, technology, engineering andsocial studies. Pergamon Press, Oxford; New York, 3rd revised edition.Fournier, D. A., Skaug, H. J., Ancheta, J., Ianelli, J., Magnusson, A.,Maunder, M. N., Nielsen, A., and Sibert, J. 2012. AD Model Builder:using automatic differentiation for statistical inference of highly param-eterized complex nonlinear models. Optimization Methods and Software,27(2):233–249.Franze´n, M. and O¨ckinger, E. 2012. Climate-driven changes in pollinatorassemblages during the last 60 years in an Arctic mountain region inNorthern Scandinavia. Journal of Insect Conservation, 16(2):227–238.Freedman, B., Hill, N., Svoboda, J., and Henry, G. 1982. Seed banks andseedling occurrence in a high arctic oasis at Alexandra Fjord, EllesmereIsland, Canada. Canadian Journal of Botany, 60(10):2112–2118.Gaston, A. J., Gilchrist, H. G., and Hipfner, J. M. 2005. Climate change, iceconditions and reproduction in an Arctic nesting marine bird: Brunnich’sguillemot (Uria lomvia L.). Journal of Animal Ecology, 74(5):832–841.Gelbic˘, I. and Olejn´ıc˘ek, J. 2011. Ecology of Dolichopodidae (Diptera) in awetland habitat and their potential role as bioindicators. Central EuropeanJournal of Biology, 6(1):118–129.Ghazoul, J. 2006. Floral diversity and the facilitation of pollination. Journalof Ecology, 94(2):295–304.Gilbert, F. S. 1985. Diurnal activity patterns in hoverflies (Diptera: Syrphi-dae). Ecological Entomology, 10(4):385–392.Gillespie, M. A. K., Jo´nsdo´ttir, I. S., Hodkinson, I. D., and Cooper, E. J.2013. Aphid-willow interactions in a high Arctic ecosystem: responsesto raised temperature and goose disturbance. Global Change Biology,19(12):3698–3708.76BibliographyGraversen, R. G., Mauritsen, T., Tjernstro¨m, M., Ka¨lle´n, E., and Svens-son, G. 2008. Vertical structure of recent Arctic warming. Nature,451(7174):53–56.Grime, J. P. 1977. Evidence for the Existence of Three Primary Strategiesin Plants and Its Relevance to Ecological and Evolutionary Theory. TheAmerican Naturalist, 111(982):1169–1194.Gugerli, F. 1998. Effect of elevation on sexual reproduction in alpine popu-lations of Saxifraga oppositifolia (Saxifragaceae). Oecologia, 114(1):60–66.Hadfield, J. D. 2010. MCMC Methods for Multi-Response Generalized Lin-ear Mixed Models: The MCMCglmm R Package. Journal of StatisticalSoftware, 33(2):1–22.Halekoh, U., Højsgaard, S., and Yan, J. 2006. The R package geepack forgeneralized estimating equations. Journal of Statistical Software, 15(2):1–11.Hardin, J. W. and Hilbe, J. M. 2003. Generalized estimating equations.Wiley Online Library.Hassol, S. J. 2004. Impacts of a warming Arctic: Arctic Climate ImpactAssessment. Cambridge University Press.Hegland, S. J., Nielsen, A., La´zaro, A., Bjerknes, A.-L., and Totland, Ø.2009. How does climate warming affect plant-pollinator interactions?Ecology Letters, 12(2):184–195.Hegland, S. J. and Totland, Ø. 2005. Relationships between species flo-ral traits and pollinator visitation in a temperate grassland. Oecologia,145(4):586–594.Heinrich, B. 1975. Energetics of Pollination. Annual Review of Ecology andSystematics, 6:139–170.Heinrich, B. 1983. Insect Foraging Energetics, pages 187–214. Scientific andAcademic Editions.Henry, G. H. R. and Molau, U. 1997. Tundra plants and climate change:the International Tundra Experiment (ITEX). Global Change Biology,3(S1):1–9.Herrera, C. M. 1987. Components of Pollinator ”Quality”: ComparativeAnalysis of a Diverse Insect Assemblage. Oikos, 50(1):79–90.77BibliographyHilbe, J. M. 1994. Generalized linear models. The American Statistician,48(3):255–265.Hilbe, J. M. 2011. Negative binomial regression. Cambridge UniversityPress.Hill, G. B. and Henry, G. H. 2011. Responses of High Arctic wet sedge tundrato climate warming since 1980. Global Change Biology, 17(1):276–287.Hinzman, L. D., Bettez, N. D., Bolton, W. R., Chapin, F. S., Dyurgerov,M. B., Fastie, C. L., Griffith, B., Hollister, R. D., Hope, A., Huntington,H. P., et al. 2005. Evidence and implications of recent climate change innorthern Alaska and other arctic regions. Climatic Change, 72(3):251–298.Hocking, B. 1968. Insect-flower associations in the High Arctic with specialreference to nectar. Oikos, 19(2):359–387.Hocking, B. and Sharpin, C. D. 1965. Flower Basking by Arctic Insects.Nature, 206(4980):215.Hodkinson, I. D., Coulson, S. J., Harrison, J., and Webb, N. R. 2001. What awonderful web they weave: spiders, nutrient capture and early ecosystemdevelopment in the high Arctic some counter-intuitive ideas on commu-nity assembly. Oikos, 95(2):349–352.Holling, C. S. 1973. Resilience and stability of ecological systems. Annualreview of ecology and systematics, 4:1–23.Hollister, R. D. and Webber, P. J. 2000. Biotic validation of small open-topchambers in a tundra ecosystem. Global Change Biology, 6(7):835–842.Holzschuh, A., Steffan-Dewenter, I., Kleijn, D., and Tscharntke, T. 2007.Diversity of flower-visiting bees in cereal fields: effects of farming system,landscape composition and regional context. Journal of Applied Ecology,44(1):41–49.Høye, T. T., Ellebjerg, S. M., and Philipp, M. 2007. The Impact of Climateon Flowering in the High Arcticthe Case of Dryas in a Hybrid Zone. Arctic,Antarctic, and Alpine Research, 39(3):412–421.Høye, T. T. and Forchhammer, M. C. 2008. High-arctic Ecosystem Dynam-ics in a Changing Climate, volume 40, chapter Phenology of High-ArcticArthropods: Effects of Climate on Spatial, Seasonal, and Inter-AnnualVariation, pages 299–322. Academic Press.78BibliographyHøye, T. T., Post, E., Schmidt, N. M., Trøjelsgaard, K., and Forchhammer,M. C. 2013. Shorter flowering seasons and declining abundance of flowervisitors in a warmer Arctic. Nature Climate Change, 3(8):759–763.Hudson, J. M. and Henry, G. H. 2010. High Arctic plant community resists15 years of experimental warming. Journal of Ecology, 98(5):1035–1041.Hudson, J. M. G. and Henry, G. H. R. 2009. Increased plant biomass in aHigh Arctic heath community from 1981 to 2008. Ecology, 90:2657–2663.Jepsen, J., Kapari, L., Hagen, S., Schott, T., Vindstad, O., Nilssen, A.,and Ims, R. 2011. Rapid northwards expansion of a forest insect pestattributed to spring phenology matching with sub-Arctic birch. GlobalChange Biology, 17(6):2071–2083.Jepsen, J. U., Hagen, S. B., Ims, R. A., and Yoccoz, N. G. 2008. Climatechange and outbreaks of the geometrids Operophtera brumata and Epir-rita autumnata in subarctic birch forest: evidence of a recent outbreakrange expansion. Journal of Animal Ecology, 77(2):257–264.Jervis, M. A. and Kidd, N. A. C. 1986. Host-feeding Strategies In Hy-menopteran Parasitoids. Biological Reviews, 61(4):395–434.Jones, M. 1995. Experimental investigations into effects of climate changeon High Arctic plants. PhD thesis, University of Alberta, Edmonton.Jones, M., Bay, C., and Nordenha¨ll, U. 1997. Effects of experimental warm-ing on arctic willows (Salix spp.): a comparison of responses from theCanadian High Arctic, Alaskan Arctic, and Swedish Subarctic. GlobalChange Biology, 3(S1):55–60.Jones, M. H., MacDonald, S. E., and Henry, G. H. 1999. Sex-and habitat-specific responses of a high arctic willow, Salix arctica, to experimentalclimate change. Oikos, 87:129–138.Karlsson, L. M. and Milberg, P. 2007. A comparative study of germinationecology of four Papaver taxa. Annals of Botany, 99(5):935–946.Kearns, C. A. 2001. North American dipteran pollinators: assessing theirvalue and conservation status. Conservation Ecology, 5(1):5.Kevan, P. 1990. Sexual differences in temperatures of blossoms on a dioeciousplant, Salix arctica: significance for life in the arctic. Arctic and AlpineResearch, 22:283–289.79BibliographyKevan, P. and Kevan, D. 1970. Collembola as pollen feeders and flower vis-itors with observations from the high Arctic. Quaestiones Entomologicae,6:311–326.Kevan, P. G. 1972a. Floral colors in the High Arctic with reference toinsect-flower relations and pollination. Canadian Journal of Botany,50(11):2289–2316.Kevan, P. G. 1972b. Insect pollination of High Arctic flowers. Journal ofEcology, 60(3):831–847.Kevan, P. G. 1973a. Flowers, insects, and pollination ecology in the Cana-dian High Arctic. Polar Record, 16(104):667–674.Kevan, P. G. 1973b. Parasitoid Wasps as Flower Visitors in the CanadianHigh Arctic. Anzeiger fu¨r Scha¨dlingskunde, Pflanzen- und Umweltschutz,46(1):3–7.Kevan, P. G. 1975. Sun-tracking solar furnaces in High Arctic flowers: sig-nificance for pollination and insects. Science, 189(4204):723 –726.Kjellberg, B., Karlsson, S., and Kerstensson, I. 1982. Effects of heliotropicmovements of flowers of Dryas octopetala L. on gynoecium temperatureand seed development. Oecologia, 54(1):10–13.Klady, R. A. 2006. Long-term experimental warming effects on tundra plantsexual reproduction in the high Arctic. Master’s thesis, University ofBritish Columbia.Klady, R. A., Henry, G. H. R., and Lemay, V. 2011. Changes in HighArctic tundra plant reproduction in response to long-term experimentalwarming. Global Change Biology, 17(4):1611–1624.Klanderud, K. and Totland, Ø. 2005. Simulated climate change altered dom-inance hierarchies and diversity of an alpine biodiversity hotspot. Ecology,86(8):2047–2054.Krannitz, P. 1996. Reproductive ecology of Dryas integrifolia in the highArctic semi-desert. Canadian journal of botany, 74(9):1451–1460.Kudo, G. 1993. Relationships between flowering time and fruit set of the en-tomophilous alpine shrub, Rhododendron aureum (Ericaceae), inhabitingsnow patches. American Journal of Botany, 80(11):1300–1304.80BibliographyKukal, O. and Kevan, P. G. 1987. The influence of parasitism onthe life history of a high arctic insect, Gynaephora groenlandica(Wo¨cke)(Lepidoptera: Lymantriidae). Canadian Journal of Zoology,65:156–163.Laird, N. M. and Ware, J. H. 1982. Random-effects models for longitudinaldata. Biometrics, 38:963–974.Larson, B., Kevan, P., and Inouye, D. 2001. Flies and flowers: taxonomicdiversity of anthophiles and pollinators. The Canadian Entomologist,133(04):439–465.Leech, R. and Ryan, J. 1972. Notes on Canadian Arctic spiders(Araneida), mainly from Devon island, NWT. The Canadian Entomolo-gist, 104(11):1787–1791.Le´vesque, E., Henry, G., and Svoboda, J. 1997. Phenological and growth re-sponses of Papaver radicatum along altitudinal gradients in the CanadianHigh Arctic. Global Change Biology, 3(S1):125–145.Liang, K.-Y. and Zeger, S. L. 1986. Longitudinal data analysis using gener-alized linear models. Biometrika, 73(1):13–22.Lortie, C., Budden, A., and Reid, A. 2012. From birds to bees: apply-ing video observation techniques to invertebrate pollinators. Journal ofPollination Ecology, 6(0):125–128.Losey, J. E. and Vaughan, M. 2006. The economic value of ecological servicesprovided by insects. Bioscience, 56(4):311–323.Lo¨ve, D. 1969. Papaver at high altitudes in the rocky mountains. Brittonia,21(1):1–10.Lund, M., Falk, J. M., Friborg, T., Mbufong, H. N., Sigsgaard, C., Soe-gaard, H., and Tamstorf, M. P. 2012. Trends in CO2 exchange in a highArctic tundra heath, 2000–2010. Journal of Geophysical Research: Bio-geosciences, 117(G2).Lundberg, H. 1980. Effects of weather on foraging-flights of bumblebees(Hymenoptera, Apidae) in a subalpine/alpine area. Ecography, 3(2):104–110.Lundberg, H. and Ranta, E. 1980. Habitat and food utilization in a subarcticbumblebee community. Oikos, 35(3):303–310.81BibliographyLundgren, R. and Olesen, J. 2005. The Dense and Highly Connected Worldof Greenland’s Plants and Their Pollinators. Arctic, Antarctic, and AlpineResearch, 37(4):514–520.MacArthur, R. H. 1967. The theory of island biogeography, volume 1. Prince-ton University Press.MacArthur, R. H. 1972. Geographical Ecology. Harper and Row.MacDonald, G. M. 2010. Global warming and the Arctic: a new worldbeyond the reach of the Grinnellian niche? The Journal of ExperimentalBiology, 213(6):855–861.Marion, G., Henry, G., Freckman, D., Johnstone, J., Jones, G., Jones, M.,Levesque, E., Molau, U., Mølgaard, P., Parsons, A., Svoboda, J., andVirginia, R. 1997. Open-top designs for manipulating field temperaturein high-latitude ecosystems. Global Change Biology, 3(S1):20–32.Mathiesen, F. 1921. The structure and biology of arctic flowering plants.15. Scrophulariaceae. Meddelelser om Gronland, 37:359–507.McLeod, A. 2011. Kendall rank correlation and Mann-Kendall trend test.R package version 2.2.Mesler, M. R., Ackerman, J. D., and Lu, K. L. 1980. The Effectiveness ofFungus Gnats as Pollinators. American Journal of Botany, 67(4):564–567.Milbau, A., Graae, B. J., Shevtsova, A., and Nijs, I. 2009. Effects of awarmer climate on seed germination in the subarctic. Annals of Botany,104(2):287–296.Miller, J., Henning, L., Heazlewood, V., Larkin, P. J., Chitty, J., Allen, R.,Brown, P. H., Gerlach, W., and Fist, A. 2005. Pollination biology of oilseedpoppy, Papaver somniferum L. Crop and Pasture Science, 56(5):483–490.Moise, E. R. D. and Henry, H. A. L. 2010. Like moths to a street lamp:exaggerated animal densities in plot-level global change field experiments.Oikos, 119(5):791–795.Molau, U. 1993. Relationships between flowering phenology and life historystrategies in tundra plants. Arctic and Alpine Research, 25:391–402.Molau, U. 1997. Responses to natural climatic variation and experimentalwarming in two tundra plant species with contrasting life forms: Cassiopetetragona and Ranunculus nivalis. Global Change Biology, 3(S1):97–107.82BibliographyMolau, U. and Mølgaard, P., editors 1996. International tundra experiment(ITEX) manual. Danish Polar Center, Copenhagen, 2nd edition.Molau, U. and Shaver, G. 1997. Controls on Seed Production and Seed Ger-minability in Eriophorum vaginatum. Global Change Biology, 3(S1):80–88.Mølgaard, P. 1989. Temperature relations of yellow and white floweredPapaver radicatum in north Greenland. Arctic and Alpine Research,21(1):83–90.Mølgaard, P. and Christensen, K. 1997. Response to experimental warm-ing in a population of Papaver radicatum in Greenland. Global ChangeBiology, 3(S1):116–124.Mølgaard, P. and Morewood, D. W. 1996. ITEX Insect: Gynaephora groen-landica/ G. rossii. In The ITEX Manual, page 85. Danish Polar Center,Copenhagen, 2 edition.Morewood, D. W. and Ring, R. A. 1998. Revision of the life history ofthe High Arctic moth Gynaephora groenlandica (Wocke) (Lepidoptera:Lymantriidae). Canadian Journal of Zoology, 76(7):1371–1381.Morton, D. C., Masek, J. G., Wang, D., Sexton, J. O., Nagol, J. R., Ropars,P., Boudreau, S., et al. 2012. Satellite-based evidence for shrub andgraminoid tundra expansion in northern Quebec from 1986 to 2010. GlobalChange Biology, 18(7):2313–2323.Mosquin, T. and Martin, J. 1967. Observations on the pollination biologyof plants on Melville Island, NWT, Canada. Canadian Field Naturalist,81:201–205.Muc, M., Freedman, B., and Svoboda, J. 1989. Vascular plant communi-ties of a polar oasis at Alexandra Fiord(79 degree N), Ellesmere Island,Canada. Canadian Journal of Botany, 67(4):1126–1136.Mu¨ller, E., Cooper, E. J., and Alsos, I. G. 2011. Germinability of Arcticplants is high in perceived optimal conditions but low in the field. Botany,89(5):337–348.Myers-Smith, I. H., Forbes, B. C., Wilmking, M., Hallinger, M., Lantz, T.,Blok, D., Tape, K. D., Macias-Fauria, M., Sass-Klaassen, U., Le´vesque,E., et al. 2011. Shrub expansion in tundra ecosystems: dynamics, impactsand research priorities. Environmental Research Letters, 6(4):045509.83BibliographyNeuhaus, J. and McCulloch, C. 2011. Generalized linear models. WileyInterdisciplinary Reviews: Computational Statistics, 3(5):407–413.Nordal, I., Hestmark, G., and Solstad, H. 1997. Reproductive biology anddemography of Papaver radicatum: a key species in Nordic plant geogra-phy. Opera Botanica, 132(132):77–87.Oberbauer, S., Elmendorf, S., Troxler, T. G., Hollister, R. D., Rocha, A. V.,Bret-Harte, M. S., Dawes, M. A., Fosaa, A. M., Henry, G. H. R., Høye,T. T., Jarrad, F. C., Jo´nsdo´ttir, I. S., Klanderud, K., Klein, J. A., Molau,U., Rixen, C., Schmidt, N. M., Shaver, G. R., Slider, R. T., Totland, Ø.,Wahren, C. H., and Welker, J. M. 2013. Phenological response of tun-dra plants to background climate variation tested using the InternationalTundra Experiment. Philosophical Transactions of the Royal Society B:Biological Sciences, 368(1624):20120481.Oberbauer, S. F., Tweedie, C. E., Welker, J. M., Fahnestock, J. T., Henry,G. H., Webber, P. J., Hollister, R. D., Walker, M. D., Kuchy, A., Elmore,E., et al. 2007. Tundra CO2 fluxes in response to experimental warm-ing across latitudinal and moisture gradients. Ecological Monographs,77(2):221–238.Oechel, W. C., Vourlitis, G. L., Hastings, S. J., Zulueta, R. C., Hinz-man, L., and Kane, D. 2000. Acclimation of ecosystem CO2 exchangein the Alaskan Arctic in response to decadal climate warming. Nature,406(6799):978–981.Oksanen, J., Blanchet, F. G., Kindt, R., Legendre, P., Minchin, P. R.,O’Hara, R. B., Simpson, G. L., Solymos, P., Stevens, M. H. H., andWagner, H. 2013. Vegan: Community Ecology Package. R package version2.0-10.Olesen, J., Bascompte, J., Elberling, H., and Jordano, P. 2008. Temporaldynamics in a pollination network. Ecology, 89(6):1573–1582.Oliver, D. R. 1963. Entomological Studies in the Lake Hazen Area, EllesmereIsland, including Lists of Species of Arachnida, Collembola, and Insecta.Arctic, 16(3):175–180.Olson, A. R. and Richards, J. H. 1979. Temperature Responses of Germi-nation in Arctic Poppy (Papaver radicatum Rottb.) Seeds. Arctic andAlpine Research, 11(3):343–348.84BibliographyPatra, N., Ram, R., Chauhan, S., and Singh, A. 1992. Quantitative studieson the mating system of opium poppy (Papaver somniferum L.). Theo-retical and Applied Genetics, 84(3-4):299–302.Peeters, L. and Totland, Ø. 1999. Wind to insect pollination ratios andfloral traits in five alpine Salix species. Canadian Journal of Botany,77(4):556–563.Philipp, M. 1980. reproductive biology of Stellaria longipes Goldie as re-vealed by a cultivation experiment. New Phytologist, 85(4):557–569.Philipp, M., Bo¨cher, J., Mattsson, O., and Woodell, S. R. J. 1990. A quanti-tative approach to the sexual reproductive biology and population struc-ture in some arctic flowering plants: Dryas integrifolia, Silene acaulis andRanunculus nivalis. Meddelelser om Grønland, 34:1–60.Philipp, M. and Siegismund, H. R. 2003. What can morphology and isozymestell us about the history of the Dryas integrifolia–octopetala complex?Molecular Ecology, 12(8):2231–2242.Philipp, M., Woodell, S. R. J., Bcher, J., and Mattsson, O. 1996. Repro-ductive Biology of Four Species of Pedicularis (Scrophulariaceae) in WestGreenland. Arctic and Alpine Research, 28(4):403–413.Pinheiro, J. C. and Bates, D. M. 2000. Linear mixed-effects models: basicconcepts and examples. Springer.Pont, A. 1993. Observations on anthophilous Muscidae and other Diptera(Insecta) in Abisko National Park, Sweden. Journal of natural history,27(3):631–643.Porsild, A. E. and Cody, W. J. 1980. Vascular Plants of Continental North-west Territories, Canada. Ottawa: National Museums of Canada.Post, E. and Forchhammer, M. C. 2008. Climate change reduces re-productive success of an Arctic herbivore through trophic mismatch.Philosophical Transactions of the Royal Society B: Biological Sciences,363(1501):2367–2373.Post, E., Forchhammer, M. C., Bret-Harte, M. S., Callaghan, T. V., Chris-tensen, T. R., Elberling, B., Fox, A. D., Gilg, O., Hik, D. S., Høye, T. T.,et al. 2009. Ecological dynamics across the Arctic associated with recentclimate change. Science, 325(5946):1355–1358.85BibliographyPrice, P. W. 1997. Insect ecology. John Wiley & Sons.Proctor, M. C. F., Yeo, P., and Lack, A. 1996. The Natural History ofPollination. Timber Press, Portland, OR.R Core Team 2013. R: A Language and Environment for Statistical Com-puting. R Foundation for Statistical Computing, Vienna, Austria.Richards, K. 1973. Biology of Bombus polaris Curtis and Bombus hyper-boreus Schonherr at Lake Hazen, Northwest Territories (Hymenoptera:Bombini). Quaestiones Entomologicae, 9:115–157.Richardson, S. J., Press, M. C., Parsons, A. N., and Hartley, S. E. 2002.How do nutrients and warming impact on plant communities and theirinsect herbivores? A 9-year study from a sub-Arctic heath. Journal ofEcology, 90(3):544–556.Ring, R. A. 2001. Arctic Insects, Global Warming and the ITEX Program.Newsletter of the Biological Survey of Canada (Terrestrial Arthropods),20(1):17–23.Robinson, S. V. 2011. Plant-Pollinator Interactions at Alexandra Fiord,Nunavut. Trail Six: An Undergraduate Journal of Geography, 5:13 – 20.Roslin, T., Wirta, H., Hopkins, T., Hardwick, B., and Va´rkonyi, G. 2013.Indirect interactions in the high arctic. PloS one, 8(6):e67367.Sahli, H. F. and Conner, J. K. 2007. Visitation, effectiveness, and efficiencyof 15 genera of visitors to wild radish, Raphanus raphanistrum (Brassi-caceae). American Journal of Botany, 94(2):203–209.Schowalter, T. D. 2006. Insect ecology: an ecosystem approach. AcademicPress.Serreze, M. C., Barrett, A. P., Stroeve, J. C., Kindig, D. N., and Holland,M. M. 2009. The emergence of surface-based Arctic amplification. TheCryosphere, 3(1):11–19.Simpson, B., Neff, J., Jones, C., and Little, R. 1983. Evolution and diversityof floral rewards. In Handbook of Experimental Pollination Biology, pages142–159. Scientific and Academic Editions.Solstad, H., Elven, R., and Nordal, I. 1999. Are there too many speciesand subspecies in the Papaver radicatum complex? In Nordal, I. and86BibliographyRazzhivin, V., editors, The Species Concept in the High North - a Panar-ctic Flora Initiative, volume 38, pages 281–294. Det Norske VidenskapsAkademi. I. Matematisk - Naturvitenskapelig Klasse, Skrifter, Ny Serie,Oslo, Sweden.Steen, R. and Thorsdatter, A. 2011. Portable digital video surveillancesystem for monitoring flower-visiting bumblebees. Journal of PollinationEcology, 5:90–94.Stenstro¨m, M., Gugerli, F., and Henry, G. H. R. 1997. Response of Saxifragaoppositifolia L. to simulated climate change at three contrasting latitudes.Global Change Biology, 3(S1):44–54.Stenstro¨m, M. and Molau, U. 1992. Reproductive ecology of Saxifraga op-positifolia: phenology, mating system, and reproductive success. Arcticand Alpine Research, 24(4):337–343.Stocker, T., Qin, D., and Platner, G. 2013. Climate Change 2013: The Phys-ical Science Basis. Working Group I Contribution to the Fifth AssessmentReport of the Intergovernmental Panel on Climate Change. Summary forPolicymakers (IPCC, 2013).Straka, J. 2012. Humming along or buzzing off? The elusive consequencesof plant-pollinator mismatches and factors limiting seed set in the CoastRange of British Columbia. Master’s thesis, University of Victoria.Strathdee, A. and Bale, J. 1998. Life on the edge: insect ecology in arcticenvironments. Annual review of entomology, 43(1):85–106.Svoboda, J. and Freedman, B., editors 1994a. Ecology of a Polar Oasis:Alexandra Fiord, Ellesmere Island, Canada. Captus University Publica-tions, Toronto.Svoboda, J. and Freedman, B., editors 1994b. Ecology of a Polar Oasis:Alexandra Fiord, Ellesmere Island, Canada, chapter Plant Communitieson the Uplands in the Vicinity of the Alexandra Fiord Lowland, pages97–110. In Svoboda and Freedman (1994a).Tamura, S. and Kudo, G. 2000. Wind pollination and insect pollination oftwo temperate willow species, Salix miyabeana and Salix sachalinensis.Plant Ecology, 147(2):185–192.87BibliographyTape, K., Sturm, M., and Racine, C. 2006. The evidence for shrub expansionin northern Alaska and the Pan-Arctic. Global Change Biology, 12(4):686–702.Totland, Ø. 1994. Influence of climate, time of day and season, and flowerdensity on insect flower visitation in alpine Norway. Arctic and AlpineResearch, 26(1):66–71.Totland, Ø. 1996. Flower heliotropism in an alpine population of Ranunculusacris (Ranunculaceae): effects on flower temperature, insect visitation,and seed production. American Journal of Botany, 83(4):452–458.Totland, Ø. 1999. Effects of temperature on performance and phenotypicselection on plant traits in alpine Ranunculus acris. Oecologia, 120(2):242–251.Totland, Ø. and Alatalo, J. M. 2002. Effects of temperature and date ofsnowmelt on growth, reproduction, and flowering phenology in the arc-tic/alpine herb, Ranunculus glacialis. Oecologia, 133(2):168–175.Totland, Ø. and Eide, W. 1999. Environmentally-dependent pollen limita-tion on seed production in alpine Ranunculus acris. Ecoscience, 6:173–179.Totland, Ø. and Sottocornola, M. 2001. Pollen limitation of reproductivesuccess in two sympatric alpine willows (Salicaceae) with contrasting pol-lination strategies. American Journal of Botany, 88(6):1011–1015.Van der Putten, W. H., Macel, M., and Visser, M. E. 2010. Predictingspecies distribution and abundance responses to climate change: why it isessential to include biotic interactions across trophic levels. PhilosophicalTransactions of the Royal Society B: Biological Sciences, 365(1549):2025–2034.Wada, N. and Kanda, H. 2000. Notes on floral traits and gender expres-sion of Dryas octopetala under a simulated environmental change. PolarBioscience, 13:147–151.Walker, M., Walker, D., Welker, J., Arft, A., Bardsley, T., Brooks, P.,Fahnestock, J., Jones, M., Losleben, M., Parsons, A., et al. 1999.Long-term experimental manipulation of winter snow regime and sum-mer temperature in arctic and alpine tundra. Hydrological Processes,13(1415):2315–2330.88Walker, M. D., Wahren, C. H., Hollister, R. D., Henry, G. H. R., Ahlquist,L., Alatalo, J., Bret-Harte, M., Calef, M., Callaghan, T., Carroll, A.,et al. 2006. Plant community responses to experimental warming acrossthe tundra biome. Proceedings of the National Academy of Sciences of theUnited States of America, 103(5):1342–1346.Warming, E. 1888. Biologiske optegnelser om gronlandske planter. 3. Scro-phulariaceae. Svensk Botanisk Tidskrift, 17:202–227.Waser, N. M. and Ollerton, J. 2006. Plant-Pollinator Interactions: FromSpecialization to Generalization. University of Chicago Press, Chicago.Welker, J., Fahnestock, J., and Jones, M. 2000. Annual CO2 flux in dry andmoist Arctic tundra: field responses to increases in summer temperaturesand winter snow depth. Climatic Change, 44(1-2):139–150.Welker, J. M., Molau, U., Parsons, A. N., Robinson, C. H., and Wookey,P. A. 1997. Responses of Dryas octopetala to ITEX environmental ma-nipulations: a synthesis with circumpolar comparisons. Global ChangeBiology, 3(S1):61–73.Wookey, P., Robinson, C., Parsons, A., Welker, J., Press, M., Callaghan, T.,and Lee, J. 1995. Environmental constraints on the growth, photosynthe-sis and reproductive development of Dryas octopetala at a high Arcticpolar semi-desert, Svalbard. Oecologia, 102(4):478–489.Wookey, P. A., Aerts, R., Bardgett, R. D., Baptist, F., Br˚athen, K. A.,Cornelissen, J. H., Gough, L., Hartley, I. P., Hopkins, D. W., Lavorel, S.,et al. 2009. Ecosystem feedbacks and cascade processes: understandingtheir role in the responses of Arctic and alpine ecosystems to environmen-tal change. Global Change Biology, 15(5):1153–1172.Wookey, P. A., Parsons, A. N., Welker, J. M., Potter, J. A., Callaghan, T. V.,Lee, J. A., and Press, M. C. 1993. Comparative responses of phenologyand reproductive development to simulated environmental change in Sub-Arctic and High Arctic plants. Oikos, 67(3):490–502.Yee, T. W. 2013. VGAM: Vector Generalized Linear and Additive Models.R package version 0.9-2.Zuur, A., Ieno, E., Walker, N., Saveliev, A., and Smith, G. 2009. Mixedeffects models and extensions in ecology with R. Springer.89Appendix AModel FittingTable A.1: Model parameters used to infer Above-Ground Biomass (AGB).Model ParametersEquation 2.1 (Salix arctica) b0=-5.02, b1=0.82, b2=0.50, b3=0.98Equation 2.2 (Dryas integrifolia) b0=-3.74, b1=1.39, b2=1.02Equation 2.3 (Papaver radicatum) b0=-5.02, b1=0.82, b2=0.50, b3=0.9890AppendixA.ModelFittingTable A.2: Likelihood criteria for flower production models. AICc is AIC corrected for finite sample sizes (Burnhamand Anderson, 2002). Terms in bold indicate the “best” model.Plant Species Factors Model TermsFlower Number Flower Number /BiomassAICc ∆AICc AICc ∆AICcSalix arctica OTCxPollenWarming * Pollen 262.58 5.10 250.73 6.39Warming + Pollen 260.88 3.40 248.50 4.16Warming 258.71 1.23 246.47 2.13Pollen 259.57 2.08 246.31 1.97- 257.48 0 244.34 0Dryas integrifolia OTCxPollenWarming * Pollen 180.41 0 179.31 0.23Warming + Pollen 185.25 4.85 181.15 2.07Warming 184.05 3.64 179.08 0Pollen 187.14 6.73 185.47 6.39- 187.28 6.88 184.78 5.70Papaver radicatum OTCxPollenWarming * Pollen 156.52 3.52 138.41 4.42Warming + Pollen 154.32 1.32 136.38 2.39Warming 154.99 1.99 134.57 0.58Pollen 165.57 12.56 135.76 1.77- 153.00 0 133.99 0Salix arctica Pollen Pollen 192.10 2.64 195.40 1.11- 189.46 0 194.28 0Dryas integrifolia Pollen Pollen 167.52 2.73 220.16 2.48- 164.80 0 217.68 0Papaver radicatum Pollen Pollen 137.74 2.92 125.54 2.48- 134.82 0 123.06 091AppendixA.ModelFittingTable A.3: Likelihood criteria for seed production and seed mass models. Terms in bold terms indicate the “best”model.Plant Species Factors Model TermsSeed Number Seed mass per FlowerAIC ∆AIC AIC ∆AICSalix arctica OTCxPollenWarming * Pollen 3385.02 4.03 -735.91 3.38Warming + Pollen 3383.01 2.02 -737.42 1.87Warming 3387.29 6.30 -736.23 3.06Pollen 3380.99 0 -739.29 0- 3385.26 4.28 -738.09 1.20Dryas integrifolia OTCxPollenWarming * Pollen 1739.01 5.49 -838.02 29.89Warming + Pollen 1737.47 3.95 -847.46 20.44Warming 1735.47 1.95 -859.02 8.89Pollen 1735.52 2.00 -856.41 11.49- 1733.52 0 -867.91 0Papaver radicatum OTCxPollenWarming * Pollen 1231.36 4.63 -509.77 29.27Warming + Pollen 1229.32 2.59 -518.86 20.18Warming 1227.22 0.48 -528.47 10.58Pollen 1228.76 2.03 -529.36 9.69- 1226.73 0 -539.04 0Salix arctica Pollen Pollen 2599.48 0 -643.53 0- 2625.18 25.71 -622.09 21.44Dryas integrifolia Pollen Pollen 1505.62 0 -743.38 25.81- 1506.00 0.39 -769.19 0Papaver radicatum Pollen Pollen 887.97 3.66 -354.29 27.50- 884.31 0 -381.79 092AppendixA.ModelFittingTable A.4: Likelihood criteria for seed mass and germination models. DIC is a Bayesian analogue of AIC used inMonte Carlo Markov Chain model selection. Terms in bold indicate the “best” model.Plant Species Factors Model TermsAverage Seed Mass Germination RateAIC ∆AIC DIC ∆DICSalix arctica OTCxPollenWarming * Pollen -4500.28 63.24 3932.39 0Warming + Pollen -4521.90 41.61 3932.77 0.38Warming -4541.11 22.40 3933.52 1.13Pollen -4544.38 19.14 3934.15 1.75- -4563.51 0 3933.47 1.08Dryas integrifolia OTCxPollenWarming * Pollen -2165.58 56.80 4727.51 0Warming + Pollen -2184.89 37.49 4729.09 1.58Warming -2205.52 16.86 4729.30 1.78Pollen -2201.82 20.56 4730.46 2.95- -2222.38 0 4730.86 3.35Papaver radicatum OTCxPollenWarming * Pollen -1653.04 69.98 1539.65 0Warming + Pollen -1676.02 46.99 1540.10 0.45Warming -1700.47 22.54 1542.82 3.17Pollen -1698.72 24.30 1540.12 0.47- -1723.02 0 1540.73 1.08Salix arctica Pollen Pollen -4544.38 19.14 3260.62 1.12- -4563.51 0 3259.51 0Dryas integrifolia Pollen Pollen -2084.99 21.08 3212.12 0- -2106.08 0 3213.96 1.84Papaver radicatum Pollen Pollen -1753.24 24.20 754.58 0- -1777.44 0 755.92 1.3493


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items