EFFECTS OF EDGE TYPE AND PATCH SHAPE ON AVIAN COMMUNITIES .
The Auk 113(3):586-598, 1996EFFECTSOF EDGE TYPE ANDAVIANCOMMUNITIESPATCHSHAPEONIN A MIXEDCONIFER-HARDWOODFORESTRITA Y. HAWROT AND GERALDJ. NIEMI5013 Miller Trunk Highway,Natural ResourcesResearchInstitute,Universityof stbird communitieswere sampled along line transectsin northwesternWisconsinduring June of 1985 through 1992 to determine whether edge type and patchshapeaffect avian fied along thesetransectsincluded:(I) edgesthat defined the habitat patchesthey separated,(2) fractalstoquantify patch shapes,and (3) areal extent of different patches.Three multiple-regressionmodelswere constructedfor I0 bird speciesand the mean number of individuals and speciesusing the following setsof independent variables:(I) edge variablesand fractals,(2) areavariables,and (3) the first six componentsfrom a principal componentsanalysisbasedon sindicatedthat edgevariablesand fractaldimension explained the most variation in abundancefor stedNuthatches(Sittacanadensis),Hermit Thrushes(Catharusguttatus),and American Robins (Turdusmigratorius).In contrast,area variablesexplained the mostvariation in abundancefor Red-eyedVireos ensylvanica),and Ovenbirds(Seiurusaurocapillus).Abundancesof Yellow-bellied roats(Geothlypistrichas),andWhite-throated Sparrows(Zonotrichiaalbicollis)were equally correlatedwith both edgeand area variables.Resultsof our study show that, for selectedspecies,forest managementstrategiesmust bedevelopedthat considernot only standcharacteristics,but alsothe edgescreatedbetweenthese stands.Received8 March 1995, accepted30 August1995.specieshave focusedon abrupt edges createdby clearcuts(Conner and Adkisson 1975, Titterington et al. 1979, Hansson 1983) or powerline corridors (e.g. Gates and Gysel 1978,ture or composition(Krummel et al. 1987), thetype and amountof edgebetweenadjacenthabitat patchesmay becomemore important. Forexample,when artificial regenerationis usedinforest management, forest composition andhabitattypesadjacentto eachother are largelydetermined by selection of planting stock(Krummel et al. 1987).Within this context,edgemay be redefined as a habitat patch boundaryKroodsma 1987, Small and Hunter 1989, Askinsor "an outer band of a habitat patch that has an1994). In most cases,avian diversity and abundance have been found to be higher at theseedges (Conner and Adkisson 1975, Gates andGysel 1978, Strelke and Dickson 1980); however, some evidence suggeststhat edges havehigher rates of nest predation (Wilcove 1985,Andr6n and Angelstam 1988, Paton 1994) andincreased brood parasitism (Brittingham andTemple 1983, Paton 1994).As changesin forest-landscapemosaicsresulting from human activities(i.e. logging) continue to lead to replacement of natural vegetationwith managedsystemsof differentstruc-environment significantly different from theinterior of the patch"(Hansenand Urban 1992).Becauseunique patternsof biodiversitymay occur at edges, thereby influencing ecologicalflows between habitatpatches(Angelstam1992),a greaterunderstandingof the ecologyof edgesEDGE TRADITIONALLYHAS BEEN DEFINEDas anabrupt boundarybetween two structurallydistinct habitats, such as a forest and a clearcut(Johnston 1947, Odum 1971, Smith 1980). Manystudies examining the effectsof edges on birdE-mail: rhawrot@sage.nrri.umn.edu586is vital.Few studies have focusedon how edges ofvarying degreesof contrast(i.e. subtleand intermediate edges)affect forest birds. Bamford(1986) evaluated the importance of deciduousforestedgesto bird life within a conifer forest.These subtle edges were important to songbirds,particularly to thosespeciesthat are mostassociated with deciduous forests (Bamford1986). DeGraaf (1992) examined the effects of
July1996]EffectsofEdgeTypeandPatchShape587T n ,E1. Tota! areaand frequencyof eachhabitattype de!ineatedalong transectsbasedon the interpretationof 1982black-and-whiteinfrared 7Regenerating (REGEN)Upland 19 yearsold, 75% coniferous 19 yearsold, 75% deciduous 19 yearsold, 25-75% deciduous11-19 yearsold153881119133Mature, coniferous (LOCON)Mature, deciduous (LODEC)Mature, mixed (LOMIX) 19 yearsold, 75% coniferous 19 yearsold, 75% deciduous 19 yearsold, 25-75% deciduous4011Mature, coniferous (UPCON)Mature, deciduous (UPDEC)Mature, mixed (UPMIX)FrequencyßLowlandAlder shrub swamp (ALDER)Open sedge/grassyopening (GRASS)15459131881310Number of sites(of 20) in which habitattype found.abrupt, intermediate,and subtleedgesbetweeneven-agednorthern hardwoodstandsand foundthat standedges,even of greatly contrastingageor height, were different from field-forestedges. Breeding-bird assemblagesin stand edgesthat were beyondthe "abrupt" stage(e.g. pole/sawlog, sawlog/large sawlog) were similar toeachother, presumablybecauseof minimal differencesin foliage profilesamong standsthatwere more than 30 years old.The main objectivesof our study were to locateand quantify edges(both natural and manmade) of varying degrees of contrast (subtle,intermediate,and abrupt) and shape,and to de-May 1982 (when !eavesof deciduoustreeswere absent).The minimum mappingunit wasapproximate!y450 m2.Forestmapswere field-verifiedand updatedduringspring1993.Approximately70%of the de!ineated habitat patcheswere verified by ground-truthing andof these70%,over 90%were classifiedcorrectly.Becauseloggingwasminimizedin the transectarea,mostvegetationchanged!ittle over the 11 years.Forestmapswere digitizedusingthe ARC/INFO3.4DGIS(ESR11987).UsingARC/INFO software,eachtransectwas divided into seven 400-m segmentseachseparatedby a 50-m bufferfor a tota!of 28 segments.Spatial autocorrelationtests indicated that a 50-mbuffer was sufficientfor consideringeach400-m segment as an independent experimental unit (Hanowski et al. 1993). Seven segmentsthat had beenscribe statistical associations between these hab!oggedduring the studyperiod and one segmentthatitat variablesand bird community characteris- containedtwo pondswere excluded,!eaving 20 segtics found at these edges.ments for use in the analysis.A 100-m buffer wasplacedaround the 200 x 400-m censusarea of eachsegment,creatinga studysite of about24 ha.METHODSEdgeswithin eachsitewere describedby the typesFour 4.35-kmtransectswere chosenfrom a group of habitatpatchestheyseparatedand categorizedintoof randomly se!ectedbird surveytransectsorigina!!y one of three classes:subtle, intermediate, or abruptestablished in northwesternWisconsin withinthe(Table 2). An exampleof a subt!eedge was a matureChequamegon National Forest (46ø00'N, 91ø00'W; pine forestnext to a maturemixed hardwoodforest,Hanowski et al. 1993). The study area is a forested while an exampleof an intermediateedgewasa lowglacia!landscapewith numerouslakesand wetlands land mixed forest next to an alder shrub swamp. Thein primari!y up!and second growth forest. Forest one abruptedgeusedin our studywasmatureforestadjacentto a grassyopening.The tota! !engthof eachmanagement practices and natural events have resuitedin a variety of habitatpatchesand the creation edge type and distributionof each edge among theof many differentedgetypes(Hanowskiet al. 1993). 20 siteswas determined using the edge ana!ysisproVegetationsampling.--Agloba! positioning system gram from the analysispackageAPACK (BoederetPathfinderBasicreceiverwas used to provide tran- al. 1995; Table 2). Rare or infrequent types (thosesect-positioninformation to a ground resolution of edges!essthan 100 m in length and/or distributed20 to 30 m accuracy(August et al. 1994); these data on fewer than seven sites)were combined with otherwere transferredto a geographicinformation system edgesbasedon similaritiesin descriptionof the edges(GIS).Forestmapswere createdby de!ineatinghabitat anddegreeof contrast(Table2). Eachof the final ninepatchesalong the transectsinto one of nine categories edge typeshad a total length of more than 1,000 m(Table 1) basedon interpretationof 1:15,840black- and was found on a minimum of seven sites, withand-white infrared stereophotographytaken during the exceptionof matureforestto regeneratingforest,
588HAWROTAND NIEMI[Auk, Vol. 113T BLE2. Total length and frequencyof eachedge type. Edgescombinedinto one of nine edgetypes.Combinationsof edgesbasedon similaritiesin descriptionand degreeof contrast.Edgesincludedin eachedgetype are given in parentheses."Adjacentto" indicatedby UPCON/UPMIX, UPCON/UPDEC),2,480m (8); lowlandconiferous/lowlandmixed(LOCON/LOMIX, LOCON/LODEC, LOMIX/LODEC),3,644 m (9).Intermediate: upland mixed/lowland mixed (UPMIX/LOMIX, UPMIX/LODEC, UPDEC/LODEC, UPDEC/LOMIX, UPCON/LOMIX), 6,527 m (13); upland mixed/lowland coniferous(UPMIX/LOCON, UPDEC/LOCON), 3,893 m (9); upland mixed/alder (UPMIX/ALDER, UPDEC/ALDER), 1,269m (7); lowlandmixed/alder (LOMIX/ALDER, LOCON/ALDER, LODEC/ALDER), 5,126 m (8); mature/regenerating(UPDEC/REGEN, UPMIX/REGEN, LOCON/REGEN), 1,428 m CON/GRASS,LOMIX/GRASS,LOCON/GRASS), 5,329 m (10).Habitat codes as in Table 1.Totallengthof eachedgetype.Numberof sites(of 20) in whichedgetypefound.which wasfound on only three sites(Table2). How-terswere obtainedfor eachsite using the mean num-ever,this type wasretainedbecauseof its potential ber of individuals and speciesrecordedfor the eightyear nce.Areasof differenthabitatpatchesfound within eachsite also were calculated using the same nine cate-abundancewere detectedfor mostbirdsspecies(Blakeet al. 1994), but these fluctuationslikely occurredingoriesfrom Table1. Patchsize(area)wasquantified responseto environmentalconditions(e.g. droughtso that effects due to the amount of suitable habitatand temperature),not to changesin the landscape,found within the landscapecould be distinguished becausewe eliminatedloggedsitesfrom our analysis.from effectsdue to edgesor a combinationof edges Ten bird specieswere selecteda priorifor ki the transects,and their primary habitat associationsmajorhabitattypesfoundwithin the studyet al. 1993)wasusedto samplebirdcommunities.One representand habitatcensuswas conductedannually per transectduringJuneof 1985through 1992.Birdswere countedfrom0.5 h beforeto 4.5 h after sunriseon dayswith littlewind ( 15 kin/h) and no precipitation.One of threeexperiencedobservers( 3 yearsof experience)conductedthe bird censuseachyear;therefore,observerarea (Table 3).Statisticalanalyses.--Fractalanalysiswas used toquantifypatchshapeswithin a site.The box-countalgorithmmethod (Morse et al. 1985) was used tocalculate an overall fractal dimension estimate (D) foreachstudysite (Westman1993).Valuesof D near 1.0variabilitywas relativelylow. Eachbird detected indicatesimpleshapesapproachingthoseof a circle,within 100 m of the transect center line was identifiedwhereasvaluesapproaching2.0 describemore comand its positionwith respectto the transectrecorded. plex shapes(Morse et al. 1985).We limited the number of independentvariablesBirdsflying abovethe canopywere not counted.Estimatesof bird speciesand community parame- to one-halfthe samplesizein the selectionmultiple-T BLE3. Habitat associationsof bird speciesselectedfor detailed analysis.SpeciesHabitatassociationYellow-bellied sand bogs(Breweret al. uousor mixed open forests,suburbs(NiemiRed-breastedNuthatch (Sittacanadensis)Hermit Thrush (Catharusguttatus)and Pfannmuller 1979)Coniferous forests (Brewer et al. 1991)Coniferous,mixed, or deciduousforests(Niemi andHanowski 1984)American Robin (Turdustnigratorius)Red-eyedVireo (Vireoolivaceus)Chestnut-sidedWarbler (Dendroicapensylvanica)Forests,gardens,and parks(Dawson1979)Deciduous forests(Conner and Adkisson d Niemi 1978)Ovenbird trichas)White-throated Sparrow (Zonotrichiaalbicollis)Upland forests(Collins et al. us or mixed forests,thickets (Titteringtonet al. 1979)
nalysis.However,we did not includemorethan one-third the number of independent variablesto sitesin any of the final multiple-regressionanalyses.Moreover, we consider this an exploratory statisticalanalysisrather than a rigoroustest of specifichypotheses.Because20 study sites and a total of 19independentvariables(10 edgevariablesand 9 areavariables)were used in our study, principal components analysis(PCA) was also performed to reducethe 19 independent variables to a smaller set of orthogonalprincipal componentsscores(PCS;Tatsuoka1971)that combinedthe edgeand areavariables.ThePCSwere thenusedto completea multiple-regressionanalysisusing linear combinationsof the independentvariables.Hence, three multiple-regressionmodelswere constructedfor each bird speciesand each communityparameter using the following sets of independentvariables:(1) nine edgevariablesand fractaldimension, (2) nine area variables,and (3) the first six PCAcomponents.Models were constructedusing maximum R2improvement (MAXR), in which the bestonevariable model, the best two-variableforth was found based on the relativemodel, and socontributionofeach variable to the least squareslinear regressionmodel (SAS Institute 1988). The level of significancefor entering an independent variable into the equation wassetat a P - 0.05. Independent and dependentvariableswere transformedusing square-rootor logarithmic transformations,whichever best improvedthe model assumptionsof normality and homogeneity of variances.The final models included thosethat met model assumptionsand had independentvariablesthat were not highly correlated(P 0.05)witheach other.RESULTSBirds.--We observed111 specieson the studysitesduring the eight-year study period (for alist of species,seeHanowskiet al. 1991).Meansof 22 individualsand 10 specieswere observedper siteoverall years(Table4). Of the 10speciesselectedfor analysis,the Red-eyedVireo (Vireoolivaceus)and Ovenbird (Seiurusaurocapillus)were observed most often, whereas the Red-breasted Nuthatch (Sitta canadensis)and Chest-nut-sidedWarbler (Dendroicapensylvanica)werethe least common (Table 4).Vegetation.--The most common edge typefound within the sites was upland deciduousforestadjacentto upland mixed forest,and theleastcommonedgetype wasupland mixed forest adjacentto alder shrub swamp (Table 2).Upland deciduousforestwas the most commonhabitat type, and regenerating forest was theleast commonhabitat type (Table 1).589
590HAWROTANDNIEMI[Auk,Vol. 113TAnrE5. Quantificationof structuralcharacteristics.Fractal-dimensionvalue (D) and length of edges(m)found within eachstudy site.Dash indicatesedge type was not found in that site.Edge N/UPMIXvaluesforthe20sites(Table 5).PCA identified six components(eigenvalue !.0), which accountedfor 89%of the variance.The firsttwo componentsexplained46%of thein the 19 habitatvariablesand con-trastedtwo differentlandscapeswithin the studyarea.Sitesthat had high scoresof PC ! generallyhad high fractal-dimensionvaluesand a highproportionof lowlandedgesand habitats.Sitesthat had high values of PC 2 generally had ahigh proportionof upland edgesand habitats.Values for PC 1, fractal dimension,and theamount of lowland edges or habitats foundwithin a site were highly positively correlatedwith51codes as in Table 1.ranged from 1.05to 1.34 (Table 5). The lowestvalue describedtwo sites that were comprisedentirely of contiguous,uplanddeciduousforest(i.e. lackededgesasdefinedby this study).Sitesthat had a higher proportion of lowland edges,such as LOMIX/ALDER, had relatively higherfractaldimensionvaluesthan thosesiteshavinga higher proportionof upland edges,suchasvariance1,0091,226210each other.Speciesrelatedwith edges.--Modelsconstructed using edge variablesand fractal dimensionvalues explained the most variation in abundance for Black-cappedChickadees(Parusatri-capillus), Red-breasted Nuthatches, HermitThrushes (Catharusguttatus), and AmericanRobins(Turdusmigratorius),as well as the mostvariation in number of speciesobserved(Fig.!). The model constructedusing the amount ofedge habitat of upland mixed forestadjacenttolowland coniferousforestexplained 23%of thevariationand wasthe bestpredictorfor numberof Black-cappedChickadees(Table6, Fig. !A; P 0.03).The modelconstructedusingthe amountof edge habitat of upland mixed forestadjacentto lowland mixed forestand upland coniferousforestadjacentto upland mixed forestexplained35% of the variationin Red-breastedNuthatchnumbers (Table 6, Fig. lB; P 0.03). The bestpredictorfor number of Hermit Thrusheswasthe modelconstructedusingthe amountof edgehabitatof lowland mixed forestadjacentto aldershrub (Table 6, Fig. !C; P 0.03), which explained 25% of the variation. The model constructedusingfractaldimensionand the amountof edge habitat createdbetween upland mixedforest and alder shrub explained 49% of thevariation and wasthe bestpredictorfor numberof AmericanRobins(Table6, Fig. !D; P 0.003).Speciesrelatedwith area.--Models constructedusingarea variablesexplainedmostof the variation in abundance for the Red-eyed Vireo,
July1996]EffectsofEdgeTypeandPatchShape591T BLœ6. Bestmodelsfor multipleregression(usingall possiblesubsetsof n.AdjustedR2 is R2 for modelafteradjustmentmadefor correspondingdegreesof freedom.Model aR2bAbundanceYellow-bellied ed-breastedNuthatch (UPMIX/LOMIX [0.18] UPCON/UPMIX [0.17])Hermit Thrush (LOMIX/ALDER)American Robin (FD [0.27] UPMIX/ALDER [0.22])Red-eyedVireo (UPDEC [0.31] LOMIX [0.13])Chestnut-sidedWarbler (UPCON [0.15] LOCON [0.13] UPMIX [0.08] LOMIX [0.08])Ovenbird (UPDEC [0.51] ALDER [0.10])Common Yellowthroat (ALDER) cCommon Yellowthroat X [0.19])White-throatedSparrow(FD [0.26]UPDEC/UPMIX [0.15])0.45 (0.42)***0.23 (0.19)*0.35 (0.27)*0.25 (0.21)*0.49 (0.43)**0.44 (0.38)*0.44 (0.29)0.61 (0.57)***0.48 (0.45)***0.46 (0.43)***0.42 (0.34)**0.41 (0.34)**0.19 (0.14)0.26 (0.22)*No. of individuals (UPDEC)No. of species(UPMIX/ALDER)*, P 0.05; **, P 0.01; ***, P - 0.001.Habitatcodes as in Table 1.AdjustedR2 given in parentheses.Bothsetsof areaandedgevariablesresultedin significantmodelswith dSparrow.Chestnut-sided Warbler, and Ovenbird, as wellusingthe amountof lowland mixed and uplandas the most variationconiferous forest, and the amount of lowlandin numberof individualsobserved(Fig. 2). The model constructedusingthe amountof uplanddeciduousforestand theamountof lowland mixed forestexplained44%of the variation in Red-eyedVireo numbers(Table 6, Fig. 2A; P 0.01).The model constructed0.2(3coniferousand upland mixed forest,explained44%of the variation and was the bestpredictorfor number of Chestnut-sided Warblers (Table6
using the following sets of independent variables: (I) edge variables and fractals, (2) area variables, and (3) the first six components from a principal components analysis based on all independent variables. Multiple-regression analysis indicated that edge variables and fractal
Edge and Edge Gold Specifier's Guide OSB-4000 April 2014 2 Edge oriented strand board (OSB) floor panels help builders get an edge on their competition by delivering both the value and product reliability needed for solid, stable floors. Easily recognized by their green edge seal, each Edge panel is uniform in size to allow easy installation, minimal waste, and
EDGE COLLETS 8 EDGE COLLETS 9 EDGE COLLETS 10 EDGE COLLETS 11 EDGE COLLETS 12 EDGE COLLETS 13 EDGE ROLLERS AND BUSHINGS FOR ANTI-VIBRATION DEVICE 14 Bushings for Anti-Vibration Device for Patriot 338 / Patriot 551 22MM FMB COLLETS14 Rollers for Movable Anti-Vibration Device for Patriot 338 / Patriot 551 14 Bushings for Movable Anti-Vibration Device
Edge Computing is a new computing paradigm where applications operate at the network edge, providing low-latency services with augmented user and data privacy. A desirable goal for edge comput-ing is pervasiveness, that is, enabling any capable and authorized entity at the edge to provide desired edge services-pervasive edge computing (PEC).
Solid Edge solutions for designing better tooling Keywords "Solid Edge; Solid Edge tool design; tool design; fixture design; XpresRoute; Voyager partners; jig design; Solid Edge Mold Tooling; mold tooling; Solid Edge Electrode; electrode tooling; QuickSheet Templates; Solid Edge Engineering Reference; Insight" Created Date: 5/9/2008 12:46:43 PM
NATIONAL GYPSUM CO — Type FSK, Type FSK-G, Type FSW, Type FSW-3, Type FSW-5, Type FSW-G, Type FSK-C, Type FSW-C, Type FSMR-C, Type FSW-6, Type FSL 2K. Gypsum Board* — (As an alternate to Item 2) — 5/8 in. thick gypsum panels, with beveled, square, or tapered edges, applied either horizontally or vertically. Gypsum panels fastened to framing with 1-1/4 in. long Type W coarse
Much of Shadowrun is structured around gaining and using Edge. Your Edge rank indicates how many points you have to start, and the Edge Points tracker can be used as you gain and spend Edge. See the Edge Boosts table on the last page for ways to spend your edge on
Study and Comparison of Various Techniques of Image Edge Detection . better than all other edge detectors on various aspects such as it is adaptive in nature, performs better for noisy image, gives sharp edges, low probability of detecting false edges. Keywords – Canny detector, edge, edge detector, Prewitt operator, Roberts operator, Sobal opertor I. INTRODUCTION Edge detection is a basic .
12 Edge computing -Training vs. Inference Training Creating AI model from existing data Inference Applying AI model to new data New data into trained model Dec. 2020 FD-SOI Substrates for Edge Computing Soitec proprietary Source: Nvidia, Soitec Evolution Cloud Cloud and Edge Cloud and Edge On-device Edge Limitations Power ceiling due to thermal inefficiency Energy limited due to .
