Page  159

2020 THE GREAT LAKES BOTANIST

RESPONSE OF 25 RARE PLANT SPECIES ON ROCKY SHORELINES OF ISLE ROYALE NATIONAL PARK INTHE FACE OF EXTREMEWATER LEVELS IN LAKE SUPERIOR

SuzanneSanders1,JessicaKirschbaum GreatLakesInventoryandMonitoringNetwork,NationalParkService 2800LakeShoreDr.East,Ashland,WI54806

SarahE.Johnson

NorthlandCollege,CenterforScienceandtheEnvironment 1411EllisAve.,Ashland,WI54806

ABSTRACT

Arctic and alpinerare plant species populate wave-splashedrocky shorelines ofIsle Royale NationalPark, wheresummertemperaturesaremoderatedbyLake Superior. Usingdata from the mid1990s and resurvey data from 1998, 2003, and 2016, we examined trajectories of change in occurrencefor25speciesat28sitescoincidentwithrisinglake levels thatfollowed aperiod of sustained low levels. We analyzed changes in site occupancy of species individually and by functional, geographic, and microhabitatgroupings.We also assessed change in population structure for four focal species.Ofthe25species,siteoccupancyincreasedfor13,andremainedsteadyforsix,decliningin another six.Site occupancydid not change over time within functional,geographic, and microhabitatgroupings. Thefourfocalspeciesincreasedinallmeasuresofabundancefromlow-tohigh-water periods, reflectingdynamic and systematically changingpopulations respondingto similar ecologicalexposures. ThesefindingssupporttheideathatthemoderatinginfluenceofLakeSuperioronair temperatures benefits these populations, despite warmingtemperatures and a15-year sustainedlow water period, and contribute to our understanding ofthe responses of at-risk species to extreme climateevents.

KEYWORDS:Disjunct,Pinguicula vulgaris,Saxifraga paniculata,Saxifraga tricuspidata,Vaccinium uliginosum

INTRODUCTION

The cold, upwelling waters ofLake Superior provide refugia for small suites ofbothArctic and alpine disjunct plant species (Given and Soper 1981). Populationsofthesespeciesarefoundinalimitednumberofrockyshorelinehabitats onislandsandalongthenorthshoreofthelake(SoperandMaycock1963;Marr etal.2009;ZlonisandGross2018)(Table1).Perhapsmostnotably,IsleRoyale NationalPark(“IsleRoyaleâ€orthe“Parkâ€),locatedinnorthwestLakeSuperior, ishosttoseveraldozenrarespecieswithdistributionstypicallymorecommonat higher latitudes and altitudes.

On Isle Royale and elsewhere around Lake Superior, many of these species opportunisticallygrowinnarrowveinsofsoilthatforminrockcracks,oronthin

1Author for correspondence (suzanne_sanders@nps.gov)

Page  160 TABLE 1. Conservation status of the 25 species in this study according to Michigan Natural Features Inventory (MNFI 2017) (MNFI) for Michigan and to NatureServe (2017) for Michigan, Minnesota, Wisconsin, Ontario, the US, Canada, and globally.The MNFI categories are endangered (E), threatened (T), and special concern (SC). The NatureServe categories are critically imperiled (1), imperiled (2), vulnerable (3) apparently secure (4), and secure (5). These numbers are prefaced by the letters S, N, or G for state, national, or global ranks, respectively.A “?†indicates an inexact or uncertain rank.A range rank (N#N# or S#S#) indicates uncertainty about the exact status; aT# following the G# or an asterisk in a state or province ranking indicates that the status applies to subspecies or varieties.

NatureServe Status

Scientific name and authority MNFI Mich Minn Wis Ont USA Canada Global

Allium schoenoprasum L. T S2 S2 unranked S4 N3N5 N5 G5 Bistorta vivipara (L.)Delarbre T S1S2 S3 notpresent S5 unranked N5 G5 Carex atratiformis Britton T S2 notpresent notpresent S2 unranked N4N5 G5 Carex gynocrates Wormsk.ex Drejer notlisted unranked unranked S4 S5 unranked N5 G5 Carex media R.Br. T S2S3 unranked S2 S4S5 unranked N5 G5T5 Castilleja septentrionalis Lindl. T S2S3 S1 notpresent S5 unranked N5 G5 Cryptogramma acrostichoides R.Br. T S2 notpresent notpresent S2S3 unranked N5 G5 Draba arabisans Michx. SC S3 S3 S2 S4 unranked N4N5 G4 Drosera anglica Huds. SC S3 S3 S1 S5 unranked N5 G5 Empetrum nigrum L. T S2 S1 notpresent S5 N5 N5 G5 Euphrasia hudsoniana Fernald&Wiegand T S1 S3 notpresent S4? unranked N4N5 G5? Huperzia selago (L.)Bernh.ex schrank&Mart. SC S3 unranked S1S2 S4 unranked N5 G5 Lonicera involucrata (Richardson)Banks exSpreng. T S2* notpresent S1 S5 unranked N5 G5T4T5 Packera indecora (Greene)A.Löve&D. Löve T S1 S3 S1 S5 unranked N5 G5 Parnassia palustris L. T unranked unranked unranked unranked unranked N5 G5 Pinguicula vulgaris L. SC S3 S3 S1 S5 unranked N5 G5 Poa alpina L. T S1S2 unranked notpresent S4 unranked N5 G5 Sagina nodosa (L.)Fenzl T S2 S1 notpresent S4 N3N4 N5 G5 Saxifraga paniculata Mill. T S1 S2 notpresent S4 N2 N4N5 G5 Saxifraga tricuspidata Rottb. T S2 notpresent notpresent S4 unranked N5 G5 Tofieldia pusilla (Michx.)Pers. T S2 S1 notpresent S5 unranked N5 G5 Triantha glutinosa Baker notlisted unranked S4S5 S2S3 S4? unranked N5 G5 Trisetum spicatum (L.)K. Richt. SC S2S3 S4S5 S2 S4 unranked N5 G5 Vaccinium uliginosum L. T S2 S2 notpresent S5 unranked N5 G5

Vaccinium vitis-idaea L. E S1 unranked S1S2* S5 unranked N5 G5T5

160 THE GREAT LAKES BOTANIST Vol. 59

Page  161 2020 THE GREAT LAKES BOTANIST

FIGURE1.Inset:LocationofIsleRoyaleNationalParkinLakeSuperiorandtheGreatLakes.Main map:Isle Royale withthestudyarea,shown circledtothenortheast(ESRIBasemap2019).

cryptobioticcrusts.Somespeciesoccuratthemarginsofrockpoolwetlandsthat have formed in rock depressions (Judziewicz 1999). Isle Royale’s rocky headlands, whichdominateboththenortheastcoastofthemainislandandnumerous barrierislands,aredirectlyexposedtothewindandwaveactionoftheopenlake (Figure 1).As aresult, changes in lake level and storm events impactthe availability of waterto plants via the height of wave splash and rockpoolrecharge.

Lake Superior water levels are dynamic and are largelyinfluencedbythe regional climate (Stow et al. 2008). The Lake Superior region experienced warming temperatures, increased evaporation, and intermittent declines in precipitationfromthelate1990stotheearly2010sduetoastrongElNiñoevent, sothat Lake Superior levels dropped (Assel et al. 2004; Gronewold and Stow 2014). Water levels remainedbelow the 100-year mean from 1998to 2013,the longest suchperiod(15years) since 1918, when the collection oflong-term data begins (USACE2020)(Figure 2).

Coincident withthis extreme low water period, three censuses ofthe distribution and abundance of Isle Royale’s rocky shoreline rare plant communities were made in 1993–1994, 1998, and2003(Judziewicz 1995, 1999,2004). Subsequent to the 2003 census, the height of Lake Superior dropped to its lowest level in 85 years before rising to levels above the 100-year mean in 2014 (Figure 2), setting a record for the most rapid rise in levels over a 2-year period

Page  162 162 THE GREAT LAKES BOTANIST Vol. 59

FIGURE2.LakeSuperiorwaterlevels,expressedasmetersabovesealevel,duringtheperiodfrom 1918to 2018(USACE2020).Thehorizontaldashedline representsthe 100-year mean;the vertical solidlinesrepresentthefoursamplingevents.Graybarshighlightthedurationofperiodsaboveand below thelong-term meanwater level.

(Gronewold et al. 2016). Water levels were still above the long-term average in 2016, coincidingwith our mostrecentplantrecensus as reportedin this paper.

Manyofthe rare rocky outcropspeciesatIsle RoyaleNationalParkare ator near their distribution limits; as such, conditions may already be marginal for them(Lawton1993;Curnuttetal.1996),andanyadditionalstressesbroughton by low lake levels could lead to losses of populations and extirpation from the Park.We undertookthe present workto examine changes in populations of rare ArcticandalpinespeciesofLakeSuperiorrockyshorelinesonIsleRoyale,concurrent with long-term low lake levels. We examined site-level trajectories of change in occurrence for 25 species, individually and in functional and microhabitatgroupings based on pollination mechanism,dispersaldistance,degree of clonality, vertical distribution above the lake, and geographic range position. These groupings were selectedbecause they mayinform us aboutthe ability of speciestopersistandspreadinresponsetowarmingandfuturelakelevelvariation. We also tested for differences in measures of abundance of four focal species between the initial and final censuses. These four species were chosen, becausetheyaregenerallymoreapparentandwidespreadacrossthesurveyarea thantheotherspeciesinthisstudy.UnderstandingtheirresponsesonIsleRoyale couldprovide information to managers, conservationists, and researchers thatis applicable across the broader distributions ofthese species.

MATERIALSAND METHODS

Study Area

Isle Royale (Figure 1) is an archipelago in Lake Superior composed of one large island(72km long,14kmwide)andmorethan400smallerbarrierislands.Theirgeneralorientationisanortheast

Page  163 2020 THE GREAT LAKES BOTANIST

southwestdirection withparallel, erosion-resistantbasalt ridges (Thornberry-Erlich2008). Jurisdictionally, theParkis apart ofKeweenawCounty,Michigan.

ManyofthePark’srareplantspeciesarelocatedontheexposedrockyheadlandsandcliffsofthe mainislandandthebarrierislandsatthenortheastendofthepark(SlavicandJanke1987;Figure1). Theseheadlandsprovidesunnymicrohabitatsexposedtothetemperature-moderatingeffectsofLake Superior (Given and Soper 1981), water recharge from the lake, and shelter from prevailing west winds due to the land mass ofthe main island(Judziewicz 1995). Soil formation is negligible here andis restricted to cracks, crevices, and other low depressions. Cryptobiotic crusts provide thin organicsurfaceswhichareoftencolonizedbyasuiteofmorewidespreadspecies( e.g.,Campanula rotundifolia L., Achillea millefolium L., Solidago hispida Muhl. ex Willd.) that are adapted to the stressesofsunexposure, iceheaving,andrelativelylownutrient availability.

Xeric species occupy the most exposed faces of these rocky headlands, while hydrophytes occupyrockpoolsthatformincrevicesanddepressions. Themostxericmicrohabitatssupportspecies dependentonoccasionalwavesplash,precipitation,andprobablyfog(Larsonetal.2000;Fischeret al. 2009;Marr et al. 2009), such as Saxifraga paniculata and Trisetum spicatum.Rockpools within mostheadland areas provide habitatfor Sphagnum spp. and wetland species such as Trichophorum cespistosum (L.)Hartm.,Drosera rotundifolia L., and Pinguicula vulgaris (Slavik andJanke1987). Due to the heterogeneity among rock pool locations, the water budgets of these rock pools vary in their hydrologic input sources (Smith1983). Some pools are precipitation-dominated via snowmelt andrainfall,someareexposedtoLakeSuperiorwavesplash,somereceiverunoffinputsfromhigher pointsontheisland,someareingroundwaterseepageareas,andsomeareinfluencedbymultiplehydrologicinputs (Egan etal. 2015).Waterlevels are the most consistentin seepagearea pools andin poolsatlower elevationsrelative to LakeSuperior.

While some species in the rocky shoreline community of Isle Royale have distributions well south ofLake Superior, severalhave broad northerlydistributions and are circum-Arctic or circumboreal. SometaxarepresentdisjunctpopulationsofArcticoralpinespeciesatthefarsouthernextent oftheirrange (Table2),while othersaredisjunctfromwestern ornorthwestNorthAmericapopulations. Climate-drivenglaciationandglacialretreatsduringtheQuaternaryperiodfacilitateddisjunctions in some species (Comes and Kadereit 1998). For example, research indicates that two of our focalspecies,Vaccinium uliginosum (Alsosetal.2005)andSaxifraga paniculata (Reisch2008),expanded post-glaciation northward into theArctic via dispersalfrom refuge sites they occupied during glaciation. Be it due to dispersal events after glaciation or to refuge sites that remained after range contractions duringglaciation, the timing and causes ofdisjunction should notbe assumedto be thesame amongalltaxa inourstudy(Thorne1972).

Plant Censuses

Judziewicz(1995)conductedbaselinefieldcensusesfor102rareplantspeciesin1993and1994. Hechoselocationsbasedonknownoccurrencesofrareplantspeciesfromearlierbotanicalsurveys, knownhabitatpreferences,andinterpretationofaerialimagery.Thesearea-widecensuseswereconductedinbothearlyandlateseason; thus, hisdataencompassthefruitingandfloweringperiodsfor mostofthetarget species.Judziewicz'sapproachtoquantifyingabundance dependedonthespecies and its growth habitat. For example, both Saxifraga paniculata and S. tricuspidata form cushions which arecomposedofmultiple rosettes (McGuireandArmbruster 1991;Reisch2008;Medeiroset al. 2012). While uncertainty aboutgenetic identity ofindividual rosettes exists (McGuire andArmbruster 1991), Judziewicz (2004) followed Reisch et al. (2003) and recognized cushions as geneticallyuniqueindividuals, or“genetsâ€,andtherosettescomprisingthesecushionsasclonesofone-another, or “ramets†(Dr. E. Judziewicz, personal communication, May 17, 2016). In these instances, dataincludecountsofbothrametsandgenets,aswellasreproductivestems.InthecaseofPinguicula vulgaris,vegetativereproductionisprolific,andthereisnovisualmethodtodiscerngeneticdifferences. Inthisinstance,allrosetteswereconsideredramets.Forotherspeciesthathaveaprostrate orspreadinggrowthhabit(e.g.,Empetrum nigrum,Vaccinium uliginosum),hemadeocularestimates ofthe area occupiedin m2.In severalinstances, Judziewicz noted onlypresence or absence. In collaboration with the National Park Service, he established and permanently marked 28 locations wherehefoundtargetrarespecies.Siteswerelooselydefinedasareassupportingrareplantsandlocatedonseparateislandsfromoneanother, oronthemainland,separatedbyatleast300meters.The sizes of the sites varied to include the entirety of the rare plants present, in most cases, unless he

Page  164 164 THE GREAT LAKES BOTANIST Vol. 59

TABLE2.Rarespeciesincludedinthepresentstudyandtheirfunctionalgroupsforthreeplanttraits, formicrohabitattype,andforpositioningeographicrange.Apositionintherangeisnotassignedto Cryptogramma acrostichoides, because Isle Royale is the only documented location within the study’s rangecriteria.

Species Pollination mechanism Dispersal distance Clonality Microhabitat type Position in range Allium schoenoprasum Biotic Bistorta vivipara Biotic Carex atratiformis Abiotic Carex gynocrates Abiotic Carex media Abiotic Castilleja septentrionalis Biotic Cryptogramma acrostichoides Abiotic Draba arabisans Biotic Drosera anglica Biotic Empetrum nigrum Abiotic Euphrasia hudsoniana Biotic Huperzia selago Abiotic Lonicera involucrata Biotic Packera indecora Biotic Parnassia palustris Biotic Pinguicula vulgaris Biotic Poa alpina Abiotic Sagina nodosa Biotic Saxifraga paniculata Biotic Saxifraga tricuspidata Biotic Tofieldia pusilla Biotic Triantha glutinosa Biotic Trisetum spicatum Abiotic Vaccinium uliginosum Biotic Vaccinium vitis-idaea Biotic Local Local Widespread Widespread Widespread Local Widespread Widespread Widespread Widespread Local Widespread Widespread Widespread Widespread Widespread Widespread Widespread Local Local Local Local Widespread Widespread Widespread Short distance Short distance Notclonal Longdistance Notclonal Notclonal Notclonal Longdistance Short distance Longdistance Notclonal Short distance Notclonal Notclonal Notclonal Longdistance Notclonal Clonal Short distance Notclonal Short distance Notclonal Notclonal Longdistance Longdistance Splash zone Splash zone Splash zone Peatyshore Rock pool Forestedge Lichenzone Lichenzone Rock pool Lichenzone Rock pool Lichenzone Forestedge Lichenzone Peatyshore Rock pool Forestedge Splash zone Lichenzone Lichenzone Rock pool Rock pool Lichenzone Rock pool Lichenzone Northern Southern Southern Southern Southern Central Central Central Southern Southern Southern Central Central Central Southern Southern Central Southern Southern Southern Central Southern Southern Southern

noted a constrained census area. He revisited and collected similar data at these 28 plots in 1998 (Judziewicz1999)and2003(Judziewicz 2004).

Weresampledthe28permanentplotsbetweenJuly6andAugust1,2016tomaximizetheprobabilitythatplantswerefloweringorfruitingandappliedar ea-widecensussearchestocountormeasure all individuals present for any of the 25 target species (Table 2). We a priori chose four focal species for more intensive population metrics. Judziewicz's surveys in 1998 and 2003 most consistently included demographic data across all plots for these four species, which are generally more abundant and apparent than the others. Applying similar methods as Judziewicz, we counted the number of ramets (rosettes), genets (cushions), and reproductive stems of Saxifraga paniculata and

S. tricuspidata atallsiteswheretheywerelocated.ForPinguicula vulgaris,ahighlyclonalspecies, werecordedthe total numberoframets(rosettes).Weestimatedarealcoveragefor Vaccinium uliginosum. Data Summaries and Analyses

The availability of quantitative data from the earlier surveys varies among species, years, and sites.Insomeinstances,onlypresenceisnotedwhileinothers,detailedcountsofbothreproductive and non-reproductive individuals were recorded. Site occupancy data, however, are standard and comparableamongyearsandspeciesinthe long-term dataset.Foreachofthe 25targetspecies,we summedthe numberofsitesoccupiedbyeach species duringeach census and examinedthemagnitude anddirection ofchange in site occupancybetweentheinitialandfinalcensuses.

Wethenidentifiedfive,broad,functional,geographic,andmicrohabitatgroupsthatcouldpotentiallyfurther explain site occupancy:pollination mechanism, dispersaldistance, degree of clonality,

Page  165 2020 THE GREAT LAKES BOTANIST

vertical distribution above Lake Superior, and geographic range position (Table 2). Some levels withinthesecategoricalgroups(e.g.,abioticandbioticpollinationorlongandshortdistancedispersal; see below) may indicate species traits that contribute to the persistence, expansion, or contraction of apopulation over time. Usingliterature searches and our knowledge ofthe species'biology, weclassifiedspeciesintolevelsforeachgroup(Table2).Foreachofthefivegroupsofinterest,we summedsiteoccupancy(thenumberofsitesagivenspecieswaspresent)acrossspecies,withineach level.Ineachgroup,wethencomparedthesesumsacrossthefourcensuses,thethreebyJudziewicz in 1993–1994, 1998,and2003, andour samplingin 2016.

Togroupspecies bypollinationmechanism, we classifiedspecies as bioticallypollinatedifthey containedanyspecializedattractants,suchascolorfulflowers.Otherwise,specieswereclassifiedas abioticallypollinated.Fordispersaldistance,weconsideredspeciestobecapableofwidespreaddispersalifthe fruits have adaptations for wind, water, fur, or animalgutdispersal. Otherwise, dispersalwasconsideredlocal. Wecategorizedthedegreeofclonalityasshortdistanceifthespeciesiscapable of vegetative reproduction with only a very limited distance from the parent plant (e.g., adventitious buds). We categorized some taxa as having longdistance clonalityif there are adaptations for further spread(e.g., rhizomes, stolons, bulbils). Species were considered not clonalifthey werecapableofonly sexualreproduction.

Most of the species here occupy key niches within the rocky shorelines, so we categorized species bytheir microhabitat or verticalposition and relative influence byLake Superior (Table 2). Forexample,amongourtargetspecies,Sagina nodosa islargelylimitedtorockcervicesinthelowest elevation habitat<4 meters (slope distance)from the calm weather waterline, which we categorizedasthe“ splashzoneâ€ofLakeSuperior.Otherspeciesalsogrowingnearthewaterlinebutinisolated pockets of perpetually wet soil were placed in the “peaty shore†microhabitat. Pinguicula vulgaris and Drosera anglica are largely confined to the edges of small rock pools. For these and other species occupying a similar niche, we designated the microhabitat as “rock pool.†Some species,such as Trisetum spicatum,occupycracks inthebasaltinrelativelymore upland microhabitats that receive less frequent splash from Lake Superior. These upland areas of exposed rock are lichen-covered, and soilformation is minimal, so werefer to this microhabitat as the “lichen zone.†Finally, the lichen zone transitions to a shrub (i.e., Juniperus communis L.) and forested zone, and species such as Saxifraga tricuspidata occupythis transition zone, which we refer to as the “forest edge.â€

To categorize species within their broader geographical ranges, we determined whether the Isle Royale populations were in the southern, central, or northern third of the species ranges in eastern NorthAmerica (Table 2). Using the online data portals of Canadensys (2017), Consortium ofMidwestern Herbaria (2017), and Consortium of Northeastern Herbaria (2017), we recorded the most northern and southern records for each species. Because populations in mainlandEurope and westernNorthAmericaaresubjecttovastlydifferenttemperaturepatternsfromthoseincentralandeastern North America and Greenland, we considered only North American records east of the 100° meridian west and in Greenland. We computed range thirds from UTM northing values for each species.Insomeinstances,herbariumreportsatthesouthernendsuggestadventiveindividuals(e.g., arecordofLonicera involucrata inthe cityofWashington D.C.,“nearfence gateâ€).Thesewerenot included. We also did not include Cryptogramma acrostichoides in this classification, because the Isle Royale populations, which are disjunct from the main range of this species in western North America, weretheonlyones withinthestudyarea.

For the four focal species with complete abundance records, we evaluated the change in abundancefrompriortothe15- yeardropinlakelevel(1990s)to2016afterthelakelevelsroseagain.For Saxifraga spp.,weassessedtheabundanceoframets,genets,andreproductivestems,whileforPinguicula vulgaris,we assessed onlythe abundance of ramets due tolimiteddataonreproductive status in the historical records. For Vaccinium uliginosum,we testedfor change in areal coverage. We computed the annual percentage change for most metrics using the formula ((2016 value–1993 value)/1993 value *100))/22years (in some cases 1994 was the baseline year).We applied nonparametric permutation tests (R, coin package)to testfor differences in abundances between years. Thesetestsdonotrequireassumptionsofequalvariancebetweengroupsandnormalityoferrors;as such,theyareapplicablehere,wheresamplesizesaresmallandassumptionsofparametrictestscannot be met. In most instances, the baseline time period was from the 1993/94 census, however in three instances where earlier data were not available, once for P. vulgaris and twice for V. uliginosum, we instead used 1998 as the baseline year. The number of site-census year pairs tested for

Page  166 166 THE GREAT LAKES BOTANIST Vol. 59

TABLE 3. The number of sites in which rare species were observed in 1993/94 and their site occupancy in those same sites in subsequent years. Numbers in parentheses indicate the total site occupancy including additional sites where those species were not recorded as observed in previous censuses. Directionarrowsindicateeithergain, loss, ornonetchange betweenthefirstandlastsampling event, including any additional sites.

Species 1993/94 1998 2003 2016 Direction Allium schoenoprasum Bistorta vivipara Carex atratiformis 1 3 2 1 2 1 1 2 3 1(3) 2 2 ↑ ↓ ↔ Carex gynocrates Carex media Castilleja septentrionalis Cryptogramma acrostichoides Draba arabisans Drosera anglica Empetrum nigrum Euphrasia hudsoniana Huperzia selago Lonicera involucrata 1 3 3 2 2 1 8 2 2 1 0 1 3 1 2 1 8 1 2 1 0 1 1 2 3(4) 1 8 0 1(2) 1 0 2(4) 3(5) 2 4(5) 1 8(10) 0(1) 1(3) 1 ↓ ↑ ↑ ↔ ↑ ↔ ↑ ↓ ↑ ↔ Packera indecora 4 2 2 3 ↓ Parnassia palustris Pinguicula vulgaris Poa alpina Sagina nodosa Saxifraga paniculata Saxifraga tricuspidata Tofieldia pusilla Triantha glutinosa Trisetum spicatum Vaccinium uliginosum Vaccinium vitis-idaea 2 11 4 2 6 9 3 2 8 13 1 2 9 2 2(3) 7 7 3 3 6(8) 12 1 0 12(15) 2 2 7 10 2 3 7(12) 14 1 0 11(16) 2(3) 2(4) 8(9) 9 1(3) 5(9) 12(19) 13(16) 1(2) ↓ ↑ ↓ ↑ ↑ ↔ ↔ ↑ ↑ ↑ ↑

each species was eight (Saxifraga paniculata), nine (S. tricuspidata), nine (P. vulgaris), and 14 (V. uliginosum).Continuedpersistenceofthesepopulationsisofmanagementsignificance,sodetecting any potential decline is paramount, and we are less concerned about making a Type I error (concluding that population decline exists even when it does not) than a Type II error (concluding that thereisnochangeinabundancewhen,infact,changehasoccurred).Forthisreason,weconsidered significance asP ≤0.1.

RESULTS

The25rareplantspeciessurveyedin2016were dispersed over the 28 monitoringsites( n=3.4rarespecies/site).Siteoccupancyincreasedfor13speciesand remained steady for six species. Six species underwent losses in the number of sitesoccupied(Table3).Inmostcases,functional,microhabitat,andgeographic groupings of taxa showed little change in site occupancy between the censuses (Figure3).Bioticallypollinatedspeciescomposedovertwo-thirdsofthesiteoccupancy, regardless ofyear (Figure 3a). Likewise, species with widespreaddispersalrepresentednearlytwo- thirdsofthespecies-sitecombinationsforallyears (Figure3b).Atallcensuses,specieswhicharenotclonalweregenerallyslightly

Page  167 2020 THE GREAT LAKES BOTANIST

FIGURE3.Siteoccupancyacrosssampleyears,groupedbylifehistorytraits(a-c),theirlocalhabitat niches (d), andgeographic range position atIsle Royale NationalPark(e).Y-axis represents the number of sites agiven species was located, summed across all species within thatlevel-year combination. Cryptogramma acrostichoides isnotincludedinrangelocationsinceitistheonlylocation ofthis species withinthestudyarea.

morecommonatsitesthanspeciescapableofeitherlongorshortdistanceclonality (Figure 3c). The lichen zone and rock pools are more frequently occupied byrarespeciesthanothermicrohabitatswithin sites(Figure3d).Throughoutall sample periods, mostoftherarespecies occurrencesareofspecies in thesouthern third oftheir ranges (Figure3e).

While change over time was notlinear, basal rosette (ramet) numbers for the threeherbaceousfocalspeciesincreasedbetweentheearly1990sand2016(Figure 4,Table 4) at an annualgrowth rate offrom 9.8% (Saxifraga paniculata)to almost14% (Saxifraga tricuspidata andPinguicula vulgaris).Despitefourtimes as manytotal ramets observedfor S. tricuspidata in 2016 comparedto the early 1990s (Figure 4,Table 4), mean ramet number among sites was not statistically greater (Table 5). The only decline in abundance that occurred during the low water period(2003) was in the areal coverage of Vaccinium uliginosum (Figure

Page  168 168 THE GREAT LAKES BOTANIST Vol. 59

FIGURE4. Mean number of ramets, genets, and reproductive stems of Saxifraga paniculata and S. tricuspidata,mean numberof ramets ofPinguicula vulgaris,andmeanarea occupiedby Vaccinium uliginosa for sites ofthe four focal species across all studyyears. Error bars represent±1 s.e. Permutation tests compared only one pre-low water level time (typically 1993/94) against the 2016 value.

Page  169 2020 THE GREAT LAKES BOTANIST

TABLE 4. Population metrics across all sites in each sample period for the four focal species for which consistent quantitative abundance data were available.

Total Total Numberof Percentageof TotalAreal Sample Period Number ofRamets Number of Genets Reproductive Stems Reproductive Stems Coverage (m2) Saxifraga paniculata 1993/94 4,757 186 263 5.53 1998 2,807 190 170 6.06 2003 5,642 450 610 10.81 2016 15,009 1,060 1,230 8.20 Saxifraga tricuspidata 1993/94 6,245 113 1,755 28.10 1998 3,346 82 515 15.39 2003 11,714 277 3,977 33.95 2016 24,977 221 4,046 16.20 Pinguicula vulgaris 1993/94 2,417 222 9.18 1998 506 164 32.41 2003 4,799 1,332 27.76 2016 9,829 2,789 28.38 Vaccinium uliginosum 1993/94 22.75 1998 24.40 2003 13.97 2016 63.07

4, Table 4), but this species had an overall trend of increase at a 8% annual growth rate from the early1990s to 2016.While average areal coverage of Vaccinium uliginosum was 2m2 greater in 2016 than in 1993/94, high variability among populations precluded significant differences (Figure 4; Table 5). The meanplantpopulationsize(numberofgenets)ofthelichenzonespeciesS. paniculata was five times larger in 2016 than was observed in 1993/94 (Figure 4, Table 4), representing a 21.4% annualgrowth rate. Similarly, the average numberofreproductivestemswasalmostfivetimesgreaterin2016forS. paniculata (Figure4,Table4),butlessthan10% ofallrametswerereproductive.ThenumberofreproductivestemsoftherockpoolspeciesP. vulgaris increasedfrom9% to28% duringthistimeperiod(Table4).TheforestedgespeciesS. tricuspidata hadgreater variability among populations in some years, so increases observed

TABLE 5. P-values for permutation tests for site-year pairs by species.An asterisk indicates significance at a = 0.1.

Species Ramets Genets Reproductive stems Area

Saxifraga paniculata 0.0551* 0.0324* 0.0781* — Saxifraga tricuspidata 0.1377 0.3300 0.4006 — Pinguicula vulgaris 0.0230* — — — Vaccinium uliginosum — — — 0.1312

Page  170 170 THE GREAT LAKES BOTANIST Vol. 59

inaveragerametnumber,genets,andcountsofreproductivestemsin2016were not statisticallygreater than baseline numbers (Figure 4;Table 5).

DISCUSSION

OurworkdetailingthepopulationtrendsofrarespeciesonIsleRoyaleshows thatthe majority ofthese species (19 of25) are either remaining stable or experiencing expanding population structure. This must be interpreted cautiously, however,asthenumberofindividualsandsitessupportingeachofthesespecies remainssmall.Themoredetaileddataonpopulationstructuresthatwecollected on the four focal species tendto supportthe idea of modestgrowth. While permutationtestsshowedchanges (increases)in onlyfouroftheeight species-metricmeasuresofabundance, werecognizethatthesmallnumberofthesepopulationscanlimitstatisticalpowertodetectchange. Actualvaluesofsiteoccupancy and abundance are generally of more interest to managers; in our case, we observedsizeableincreasesinallmeasuresofabundanceforallfourfocalspecies. Acrossall25species,thesuccessofthemajorityofspeciesweobservedin2016 after a 13-year sampling hiatus was surprising given a 15-year low water level concomitant with rising air temperatures, increased evaporation, and intermittentlylowerprecipitationregionally( Gronewoldetal.2016;Zhongetal.2016), coupledwiththefactthatthesespeciesgenerallygrowfarthernorthorathigher altitudes than Isle Royale.Air and water temperatures have both risen over the 25years ofthis monitoring, withlake temperatures warmingfaster (Austin and Colman2008).Nonetheless,temperaturemoderationofcoastalhabitatsbyLake Superior maybe limitingthe effects of warmingtemperatures.

Lake Superior hydrology may be playing a large role regulating the occurrenceandpopulationsizesoftheserarespeciesviachangesinlakelevels. Water levels are determined by a number of natural inputs and outputs (precipitation, stream input,groundwater,surfacewaterrunoff,andevaporation (Gronewold et al. 2016))aswellas limitedwater levelregulation,whichismanagedbythe International Joint Commission and currently follows the Lake Superior RegulationPlan2012( InternationalJointCommission2012).LakeSuperiorwaterlevels have fluctuated since 1918, with an average water height of 183.4 m above sealevel.Theselevelswerebelowaveragefrom1998through2013,afterwhich levels rose to above-average heights (Figure 2). While the difference between the 1986high(183.73m)andthe mostrecentlow (182.98in 2007)isonly0.75 m, this can significantly impact the amount of bedrock either exposed or submerged, astheseheadlandsarelowandwithgradualslopesintothelake.Theincrease in lake level and increases in precipitation between 2013 and 2016 (Gronewold et al. 2016) could explain the relatively successful response of species in 2016, as long-established populations are once again in closer proximityto water recharge andto the temperature-moderatinginfluence of moister coastal conditions.

Rocky shoreline species, while presumablyweak competitors,generallypossessphysiologicalandanatomicaladaptationsthatconfertoleranceagainststress

Page  171 2020 THE GREAT LAKES BOTANIST

(Maestre et al. 2009; Pellissier et al. 2010). Saxifraga paniculata, in particular, displaysanumberofthesetraitsassociatedwithwaterstress,includingleafcupping( Neuner et al. 2008) and athick cuticle (Hegi1975).This species also displays leaf hydathodes, an adaptation to, among other things, high summer temperaturesoftenpresent on limestone substrates (Andrei andParaschivoiu2008). This adaptation creates acooler, more humid atmosphere aroundthe plant(Andrei and Paraschivoiu 2008), but also allows direct uptake of snowmelt (see Hacker andNeuner 2006) andpossibly rainwater and moisture from fog.

The ability to absorb moisture from the air may also exist in Pinguicula,a genus of carnivorous species with specialized glands on the lower leaf surfaces (Lloyd 1942;Adlassnig et al. 2005), although we are unaware whether this has been demonstrated. The favorable performance of P. vulgaris may be more a function of its dispersal mechanism. Like most carnivorous plants, P. vulgaris possesses only a weak root system (Adlassnig et al. 2005); in autumn, a winter bud(hybernaculum)isproducedwithseveralgemmaearoundthebase(Heslop- Harrison 2004). Running water or, in the case of Isle Royale, wave action can dislodge gemmae, allowing them to disperse to and colonize new rock pools (Legendre 2000).The 0.75 mrise in water level mayhave allowed wave action to reach extant colonies whichhad experienced either limited, or nohighvelocitywave actionfor severalyears. Expandedrockpoolhabitat mayalsohaveoccurred withincreased water levels.

Despitethenumerouspositivechanges,sixspeciesexperiencedoveralllosses in the number of sites occupied(population contraction) over the entirety ofthe study period. It is likely that several factors contributed to changes for each species. One ofthe declining species, Bistorta vivipara,is known to have alow seed set and slow, conservative growth(Diggle 1997);five growing seasons are required between leaf and inflorescence initiation to reach functional maturity. As a consequence, the current season's above-ground appearance reflects environmental conditions of the previous four years; likewise, poor environmental conditionsduringoneyearcouldimpactfiveyearsofplantperformance.Reproduction is primarily asexual, by bulbils, and successful fruit set has only rarely been reported in the literature (see Diggle et al. 2002 for citations). In North America, fruit set in the subalpine environment has been reported only in Wyoming(Bliss1958).Reasonsforthelowfruitsetincludelowpollenviability (Engell 1978; Diggle et al. 2002) and high rates of embryo abortion, possibly due to genetic abnormalities (Diggle et al. 2002). Reproduction by bulbils may be adaptive in cold environments (Billings and Mooney 1968) by reducing relianceonpollinatorsandallowingdispersionofsuccessfulgenotypesawayfrom the parent plant. Like many of the species in this study, however, Isle Royale populations are highly isolated from other populations outside of the Park, and oftenevenfromoneanotherwithinthearchipelago.LackofgeneticrecombinationonaregularbasiswouldlimittheabilityofBistorta vivipara toadaptatthe southern edge of its distribution where conditions are likely sub-optimal (Reed andFrankham 2003;Spielman et al.2004).

Like Bistorta vivipara, Poa alpina also reproduces by both bulbils and seed andalsoexperiencedadeclineoverthe24-yearstudyperiodwithtwoofthefour sites extirpated. Research in NorthAmerica (Hermesh andAcharya 1987) has

Page  172 172 THE GREAT LAKES BOTANIST Vol. 59

demonstratedtemperature-specificadaptationforanumberofreproductivecharacteristics, includingthenumberoffloretsperpanicle.InanAlpineenvironment in Europe, Steiner et al. (2012)foundplastic responses of P. alpina in response to transplantations across elevations. Although we are unaware of the relative proportionofseedvs.bulbilproductionforIsleRoyalepopulations,collectively, these suggest Poa alpina shouldbe able to either adapt or acclimate to environmental conditions there.

Theabilitytoadaptmaybegreaterforannualspecies,asaresultofmorefrequentgeneticrecombination, althoughdistancesbetweenpopulationscouldalso be prohibiting this. One of the species that fared the poorest over the 24-year study period is Euphrasia hudsoniana, originally present attwo locations. This appearstohavebeenextirpatedfromthesiteswhereitwasoriginallyknown,although it was subsequently found in 2016 at a site not previously occupied by this species. Zlonis and Gross (2018) examined the genetic structure of this species on the rocky shoreline of Lake Superior in northern Minnesota. They demonstratedahighdegreeofheterozygositywithinpopulations,whichmaybe aresultofitstetraploidgenome(MeirmansandVanTienderen2013).Whilethis level of heterozygosity should promote population persistence, the distances separating known populations will inhibit, or even prevent, gene flow between them.Ultimately,suchisolationmayberenderingthisspeciesunabletoadaptto changing environmental conditions.

Ofthespeciesthatperformedwell,two—Sagina nodosa andTrisetum spicatum— wereparticularlysurprisingtous.Sagina nodosa growsjustabovethewaterline andtypicallyoccupiesthe lowestnicheon the rocks. Becauseofthis,we mayhaveanticipatedadownwardmigrationofexistingpopulationsoverthelow water period, followed by a loss of these populations as the water level rose to above-average depths after 2013. Instead, we found a net increase in populations. In addition, although abundance data are not available for all sites at all time periods, where data are available, we found marked increases in the number of stems in total, as well as those that were reproductive (data not shown). Sagina nodosa produces vegetative buds on the stem which disarticulate; these float and are carried by wind and water to potential new colonization sites (Wright1953).Thisformofreproductionanddispersalishighlyadaptiveinthis environmentandwilllikelyservethisspecieswellinthefaceoffuturepotential fluctuating water levels.

The grass species Trisetum spicatum is also performing well in the Park. While we do nothave abundance data for all sites and censuses, the abundance ofbothgenetsandreproductivestemsgenerallyincreasedten-foldatsiteswhere data are available (data not shown). In addition, we often noticed this species growing in several areas along the rocky coastlines outside of established sampling sites. Unfortunately, little is known about the biology of this species that couldinform us ofthe causes for this increase.

Piecing together the biological basis for change in rare species can be challenging, and rare species sampling efforts themselves also present unique challenges. In earlier sampling efforts in the absence of aerial imagery, maps were often hand drawn, and GPS was often either unavailable or unreliable, making relocationinexact,although,Judziewicz’s(1999)mapsandnotationsaboutsam

Page  173 2020 THE GREAT LAKES BOTANIST

ple locations werequitedetailed.Differencesbetweenobservers also introduces uncertainty(Alexanderetal.2012;MorrisonandYoung2016);thesedifferences canbebetweensamplingeventsorfromwithinthesameevent.Finally,although rare,thefewpopulationswhereaspeciesispresentaresometimescomprisedof largenumbersofgenetsorramets,whichalsointroducessomedegreeoferrorin completecensuscounts,especiallywhen,asinsomeinstances,hundredsoframets are present. For the current work, we acknowledge a degree of uncertainty when counting ramets and assessing areal coverage, includingfor allfour focal species. However, counts of genets and reproductive stems were much more straightforward, since these were fewer and more clearlydefined. We note here that the relative patterns of abundance of both genets and reproductive stems largely mirror those of ramet abundance for the first three time periods of this study (Figure 4, a small dip in abundance in 1998, followed by a sizeable increase in 2003). In our current census (2016), we see a similar pattern, where marked increases in ramets are concordant with more genets and reproductive stems, providingfurther assurance thatthe large increases we observedin 2016 are genuine. Finally, the three earlier censuses were completed by Judziewicz (1995, 1999, 2004), who author S. Johnson assisted elsewhere to monitor rare plantsusingsimilarmethods.BecauseJohnsonwasatallsamplinglocationsand timesin2016,wehaveassurancethattechniquesandpracticesofthe2016event were fairly consistent withthose of earlier censuses.

We recognize thatfactors besides lake level maybe playinginto species performance, particularlylocally. Unfortunately, we lack site-leveldata on temperature and solar radiation;likewise, the recordfor precipitation atthe parkis incomplete. Because of these shortfalls, we are unable to test for correlations of these metrics with population response. However, what this work does show is that, despite a 16-year record of low water, the majority of these species are as abundant,orevenmoreabundant,thanpriortothedropinwaterlevel.Ourwork here,whichrepresentsasinglesnapshotintime,suggeststhatthesespeciespossess a degree of resilience to ecosystem stressors that we had, perhaps, not anticipated. The continuedfate ofthese populations, of course, remains unseen.

Research Needs and Management Recommendations

For the majority of species studied here, we lack detailed information on basic biology(Godefroid et al. 2011). Understandingthe breeding system, populationgeneticstructure, anatomy(asitrelatestogrowthanddevelopment),and physiology (responses to temperature and solar irradiation) would help guide any future management action and/or development of restoration plans. Thorough, updatedinventories of allpotentially suitable habitatis also desired.

Atthispointintime,weurgeparkmanagerstoexploreseedbankingforsome of these species. Retaining seed not only conserves genetic material, but provides sources for ex-situ research, as suggested above.We also suggestthatIsle Royale managers initiate dialog with their counterparts at other parks around Lake Superior that support these species. For populations with limited genetic diversity, reciprocal outplanting of seed (or vegetative material) between them maybe requiredto avoidlocal extinction.

Page  174 174 THE GREAT LAKES BOTANIST Vol. 59

While the idea of working to preserve these species is noble, in reality, options to manage these species will be limited in the face of continuing climate change. By andlarge, these species are adaptedto cooler, moist conditions with lesssolarirradiation,concomitantwithnorthernlatitudes.Theyhavemostlikely persisted due to the moderating effects of Lake Superior and available rocky shorelinehabitat.Asairandwatertemperaturescontinuetorise,suitablehabitat withintheParkmayceasetoexist.Managingfoottrafficinhighvisitoruseareas (e.g.,ScovillePointonthemainisland)couldlimittramplingof speciessuchas Saxifraga tricuspidata and Empetrum nigrum. Removing non-native wetland species(T. angustifolia L.anditshybridwiththenativeT. latifolia L.)fromrock poolsmayservetoprotecthabitatforrarespeciessuchasCarex media R.Br.At the upper edge of their habitat, succession to shrubs and forest may overcome habitatforpopulationsofspeciessuchasS. tricuspidata andE. nigrum,somaintaining more open conditions in some sites is an option to retain regenerating populations oftheserare species. Managingfor other species is not as clear-cut. WhilewehaveeveryinterestinpreservingnaturalresourceswithintheNational Park system, one may question the prudence of this, given the current climate forecasts, the difficulties encountered during rare species reintroductions, and thefiscaldemandsofsuchefforts(Godefroidetal.2011).Nonetheless,whatwe propose here is still of value. Several ofthese species have widespreaddistributions and documenting their changes in occurrence in Isle Royale may signal species-scale distributional shifts related to climate change. Documenting changes in rare species may cue managers of the risk of deeper ecosystem changes.Knowledgegainedofbreeding systems,physiology, andgenetic structure ofthese species could be applied not only to managers on Isle Royale, but to practicing conservationists elsewhere.

ACKNOWLEDGEMENTS

We thank Lynette Potvin, Mark Romanski, and other natural resource staff at Isle Royale NationalParkfor arranginglogistics during our field work. We are indebtedto Dr. EmmetJudziewicz for his impressively expansive floristic surveys on Isle Royale NationalParkinthe 1990s and early 2000sandforhavingtheforesighttoidentifypermanentmonitoringsitesfortrackingpopulationsof numerous rare species. We are grateful toAnton Reznicek for providing natural history insight for several species of Carex.Finally, Rebecca Key was instrumentalin database development anddata management.

LITERATURE CITED

Adlassnig, W., M. Peroutka, H. Lambers, and I. K. Lichtscheidl. (2005). The roots of carnivorous

plants. PlantandSoil274:127–140. doi:10.007/s11104-004-2725-7. Alexander,H. M.,A.W. Reed,W. D. Kettle, N.A. Slade,S.A. BodbylRoels, C. D.Collins,andV.

Salisbury.(2012).Detection andplantmonitoringprograms:Lessons from anintensivesurvey of

Asclepias meadii withfiveobservers.PLOSONE7:e52762. doi:10.1371/journal.pone.0052762. Alsos, I. G., L. Engelskjøn, L. Gielly, P.Taberlet, andC. Brochmann. (2005). Impact ofice ages on

circumpolar molecular diversity:Insights from an ecologicalkey species. Molecular Ecology14:

2739-2753.doi:10.1111/j.1365-294X.2005.02621.x Andrei,M.,andR.M.Paraschivoiu.(2008).Anatomicalresearchesontheovergroundvegetativeor

gansof Saxifraga mutata L. subsp. demissa (Schott&Kotschy)D.A.Webb andSaxifraga panic

Page  175 2020 THE GREAT LAKES BOTANIST

ulata Miller.AnaleleÅžtiinÅ£ificealeUniversităţii“Al.I.Cuzaâ€IaÅŸi,Sect.IIa.BiologieVegetală54: 5–14.

Assel,R.A.,F.H.Quinn,andC.E.Sellinger.(2004).Hydroclimaticfactorsoftherecentrecorddrop inLaurentialGreatLakeswaterlevels.BulletinoftheAmericanMeteorologicalSocity85:1143– 1151.doi:10.1175/BAMS-85-8-1143.

Austin,J.A.,andS.M.Colman.(2008).AcenturyoftemperaturevariabilityinLakeSuperior.Limnology andOceanography53:2724–2730. doi:10.4319/lo.2008.53.6.2724.

Billings, W. D., andH.A. Mooney. (1968). The ecology of arctic and alpine plants. BiologicalReviews43: 481–529. doi:10.1111/j.1469-185X.1968.tb00968.x.

Bliss, L. C. (1958). Seed germination in arctic and alpine species. Arctic 11(3): 180–188. doi:10.14430/arctic3743.

Canadensys. (2017).Availableathttp://data.canadensys.net/explorer.(AccessedOctober 212017).

Comes,H.P.,andJ.W.Kadereit.(1998).TheeffectsofQuaternaryclimaticchangesonplantdistribution andevolution.TrendsinPlantScience3:432–438. doi:10.1016/S1360-1385(98)01327-2.

Consortium of Midwest Herbaria. (2017). Data Portal. Available at http//:midwesternherbaria.org/ portal/.(AccessedOctober21,2017).

Consortium of Northeastern Herbaria Data Portal. (2017).Available at portal.neherbaria.org/portal. (AccessedOctober21,2017).

Curnutt,J.L.,S.L.Pimm,andB.A.Maurer.(1996).Populationvariabilityofsparrowsinspaceand time. Oikos76:131–144.doi:10.2307/3545755.

Diggle,P.K.(1997).ExtremepreformationinalpinePolygonum viviparum:Anarchitecturalanddevelopementalanalysis. American JournalofBotany84:154–169.doi:10.2307/2446077.

Diggle,P.K.,M.A.Meixner,A.B.Carroll,andC.F.Aschwanden.(2002).Barrierstosexualreproduction in Polygonum viviparum:A comparative developmental analysis of P. viviparum and P. bistortoides.AnnalsofBotany89:145–156.doi:10.1093/aob/mcf020.

Egan, A. T., L. T. Ferrington Jr., T. Lafrançois, M. B. Edlund, and J. McCullough. (2015). Spatial arrangementandmetricsoffreshwatercoastalrockpoolsappliedtoamphibianconservation.Limnologica51: 101–109. doi:10.1016/j.limno.2014.12.007.

Engell, K. (1978). Morphology and cytology of Polygonum viviparum in Europe. I. The Faroe Islands. BotaniskTidsskrift72:113–118.

Fischer,D.T.,C.J.Still,A.P.Williams,andG.MacDonald.(2009).Significanceofsummerfogand overcastfordroughtstressandecologicalfunctioningofcoastalCaliforniaendemicplantspecies. Journal ofBiogeography36:783–799.doi:10.1111/j.1365-2699.2008.02025.x.

Given, D. R., andJ. H. Soper. (1981).The arctic-alpine element ofthe vascular flora atLake Superior. Publications in Botany10:1–70.

Godefroid, S., C. Piazza, G. Rossi, S. Buord,A.-D. Stevens, R.Aguraiuja, C. Cowell, et al. (2011). How successful are plant species reintroductions? Biological Conservation 144: 672–682. doi:10.1016/j.biocon.2010.10.003.

Gronewold, A. D., and C. A. Stow. (2014). Water loss from the Great Lakes. Science 343: 10841085. doi:10.1126/science.1249978.

Gronewold,A. D., J. Bruxer, D. Durnford, J. P. Smith,A. H. Clites, F. Seglenieks, S. S. Qian, et al. (2016). Hydrological drivers of record-setting water level rise on Earth's largest lake system. Water ResourcesResearch52:4026–4042.doi:10.1002/2015WR018209.

Hacker, J., and G. Neuner. (2006). Photosynthetic capacity and PSII efficiency of the evergreen alpine cushion plant Saxifraga paniculata during winter at different altitudes. Arctic, Antarctic, andAlpineResearch38:198–205.

Hegi, G. (1975). Illustrierte Flora von Mitteleuropa. Band IV, 2. Teil, Teilband, A. bearbeitet von Huber H.VerlagPaulParey,Berlin., Germany.

Hermesh,R.,andS.N.Acharya.(1987).Reproductiveresponsetothreetemperatureregimesoffour Poa alpina populations from the Rocky Mountains of Alberta, Canada. Arctic and Alpine Research19: 321–326.doi:10.1080/00040851.1987.12002607.

Heslop-Harrison, Y. (2004). Biological flora of the British Isles: Pinguicula L. Journal of Ecology 92:1071–1118.doi:10.1111/j.0022-0477.2004.00942.x.

International Joint Commission. (2012). Lake Superior regulation:Addressing uncertainty in upper Great Lakes water levels. March 2012.Available online at https://www.ijc.org/sites/default/files/ Lake_Superior_Regulation_Full_Report.pdf

Judziewicz,E.(1995).Inventory and establishment of monitoringprogramsfor specialfloristicele

Page  176 176 THE GREAT LAKES BOTANIST Vol. 59

ments at Isle Royale National Park, Michigan. Technical report for National Park Service Great Lakes InventoryandMonitoringNetwork,Ashland,Wisconsin.

Judziewicz, E. (1999). Inventoryandestablishmentof monitoringprograms forspecialfloristic elements at Isle Royale NationalPark, Michigan:1998 resurvey. Technical reportfor NationalPark ServiceGreatLakesInventory andMonitoringNetwork,Ashland,Wisconsin.

Judziewicz, E. (2004). Inventoryandestablishmentof monitoringprograms forspecialfloristic elements atIsle Royale NationalPark, Michigan:2003 resurvey. Technical reportfor NationalPark ServiceGreatLakesInventory andMonitoringNetwork,Ashland,Wisconsin.

Larson,D.W.,U.Matthes,andP.E.Kelly.(2000).Cliffecology:Patternandprocessincliffecosystems. CambridgeUniversityPress,Cambridge,UnitedKingdom.

Lawton,J.H.(1993).Range,populationabundanceandconservation.TrendsinEcologyandEvolution8: 409–413. doi:10.1016/0169-5347(93)90043-O.

Legendre, L. (2000).The genus Pinguicula L. (Lentibulariaceae):An overview.Acta Botanica Gallica 147:77–95.doi:10.1080/12538078.2000.10515837.

Lloyd, F.E.(1942).The CarnivorousPlants.RonaldPress,NewYork,N.Y.

Marr, J. K., M. R. Penskar, andD.A.Albert. (2009). Rare plant species andplant communitytypes ofManitou Island andGullRock, Keweenaw County, Michigan. The Michigan Botanist48:97– 120.

Maestre, F. T., R. M. Callaway, F. Valladares, and C. J. Lortie. (2009). Refining the stress-gradient hypothesisforcompetitionandfacilitationinplantcommunities.JournalofEcology97:199–205. 10.1111/j.1365-2745.2008.01476.x.

McGuire,A.D.,andW.S.Armbruster.(1991).Anexperimentaltestforreproductiveinteractionsbetween two sequentiallyblooming Saxifraga species (Saxifragaceae).American Journal ofBotany 78:214-219.

Medeiros,A.S.,R.G.Biastoch,C.E.Luszczek,X.A.Wang,D.C.G.Muir,andR.Quinlan.(2012). Patternsinthelimnologyoflakesandpondsacrossmultiplelocalandregionalenvironmentalgradients in the eastern CanadianArctic. InlandWaters 2:59–76.

Meirmans,P.G.,andP.H.VanTienderen.(2013).Theeffectsofinheritanceintetraploidsongenetic diversity andpopulation divergence.Heredity110:131–137.doi:10.1038/hdy.2012.80.

MNFI.(2017).Available atmnfi.anr.msu.edu/species/plants. (AccessedOctober18, 2017).

Morrison, L.W., andC.C.Young.(2016). Observer errorin sampling arare plantpopulation. Plant Ecology andDiversity9:289–297.doi:10.1080/17550874.2016.1220989.

NatureServe. (2017).Availableat explorer.natureserve.org.(AccessedOctober18,2017).

Neuner, G., V. Braun, O. Bushner, and D. Taschler. (2008). Leaf rosette closure in the alpine rock species Saxifraga paniculata Mill.: significance for survival ofdrought andheat under highirradiation. Plant,cell andEnvironment.22:1539–1548.

Pellissier,L.,K.A.Bråthen,J.Pottier,C.F.Randin,P.Vittoz,A.Dubuis,N.G.Yoccoz,etal.(2010). Species distribution models reveal apparent competitive and facilitative effects of a dominant species on the distribution of tundra plants. Ecography 33: 1004–1014. doi:10.1111/j.16000587.2010.06386. x.

Reed, D. H. andR. Frankham. (2003). Correlation between fitness andgenetic diversity. ConservationBiology17: 230–237.doi:10.1046/j.1523-1739.2003.01236.x.

Reisch,C.(2008).GlacialhistoryofSaxifraga paniculata (Saxifragaceae):Molecularbiogeography of adisjunctarctic-alpinespeciesfromEuropeandNorthAmerica.BiologicalJournaloftheLinneanSociety93: 385–398.doi:10.1111/j.1095-8312.2007.00933.x.

Reisch, C., P. Poschlod, and R. Wingender. 2003. Genetic variation of Saxifraga paniculata Mill. (Saxifragaceae): molecular evidence for glacial relict endemism in central Europe. Biological JournaloftheLinnean Society80:11-21.

Slavik,A. D., andR.A. Janke. (1987).The vascular flora ofIsle Royale NationalPark.The MichiganBotanist26: 91–134.

Smith, D. C. (1983). Factors controllingtadpole populations ofthe chorus frog (Pseudacris triseriata) onIsleRoyale,Michigan.Ecology64:501–510. doi:10.2307/1939970.

Soper, J. H., and R. F. Maycock. (1963). Acommunity of arctic-alpine plants on the east shore of LakeSuperior.CanadianJournalofBotany41:183–198.doi:10.1139/b63-016.

Spielman, D., B.W. Brook, andR. Frankham. (2004). Most species are notdriven to extinction before genetic factors impactthem. Proceedings ofthe NationalAcademy ofSciences 101:15261– 15264.doi:10.1073/pnas.0403809101.

Page  177 2020 THE GREAT LAKES BOTANIST

Steiner,B.L.,G.F.J.Armbruster,J.F.Scheepens,andJ.Stöcklin.(2012).Distributionofbulbil-and seed-producingplantsofPoa alpina (Poaceae)andtheirgrowthandreproductionincommongardens suggest adaptation to different elevations. American Journal of Botany 99: 2035–2044. doi:10.3732/ajb.1200213.

Stow, C.A., E. C. Lamon,T. K. Kratz, C. E. Sellinger. (2008). Lake level coherence supports commondriver. EOS89:389–404.doi:10.1029/2008EO410001.

Thornberry-Erlich,T.(2008).IsleRoyaleNationalParkgeologicresourceevaluationreport.Natural ResourcesReportNPS/NRPC/GRD/NRR—2008/037.NationalParkService, Denver,Colorado.

Thorne, R. F. (1972). Major disjunctions in the geographic ranges of seedplants. QuarterlyReview ofBiology47:365–411.doi:10.1086/407399.

USACE. (2020). Great Lakes Long-term water level datasets. Available at: https://www.lre.usace. army.mil/Missions/Great-Lakes-Information/Great-Lakes-Information-2/Water-Level-Data. U.S. ArmyCorpsofEngineers,DetroitDistrict,Detroit,MI48226USA.(AccessedAugust30,2020).

Wright, F. R. E. (1953). Notes on the dispersal of Sagina nodosa var. moniliformis lange. Watsonia 2:369–370.

Zhong, Y., M. Notaro, S. J. Vavrus, and M. J. Foster. (2016). Physical accelerated warming of the Laurentian Great Lakes: Physical drivers. Limnology and Oceanography 61: 1762–1786. doi: 10.1002/lno.10331.

Zlonis, K. J. and B. L. Gross. (2018). Genetic structure, diversity, and hybridization in populations of the rare Arctic relict Euphrasia hudsoniana (Orobanchaceae) and its invasive congener Euphrasia stricta.ConservationGenetics19:43–55.doi:10.1007/s10592-017-0995-x.