Patterns of territorial space use by Shining Sunbeams(Aglaeactis cupripennis), tropical montane

hummingbirds

Lucas I. Pavan,1,4 Jill E. Jankowski,1,3 and Jenny A. Hazlehurst2

1Biodiversity Research Centre and Department of Zoology, University of British Columbia, Vancouver, BC V6T 1Z4,Canada

2Department of Biological Sciences, California State University, East Bay, Hayward, California 94542, USA

Received 22 November 2019; accepted 7 January 2020

ABSTRACT. For many territorial hummingbirds, habitat use is influenced primarily by the interactionbetween resource acquisition and non-foraging behaviors such as territory advertisement and defense. Previousresearch has highlighted the importance of foraging-associated habitat features like resource density anddistribution in determining the space-use patterns of hummingbirds. Less is known, however, about howhabitat selection associated with non-foraging behaviors influences space use by territorial species. We usedradio telemetry to examine patterns of territorial space use by Shining Sunbeams (Aglaeactis cupripennis) inhigh Andean montane forests near Manu National Park, Peru, and Bosque Comunal “El Carmen” nearChordeleg, Ecuador. We quantified within-territory habitat characteristics related to resource acquisition andnon-foraging behaviors such as territory advertisement and defense. We found that Shining Sunbeams showedhigh use of core areas in territories where foraging effort was relatively low. We found no relationship,however, between the position of core areas and habitat characteristics associated with territory defense,predator avoidance, or other non-foraging behaviors. We also found no relationship between use of non-coreareas and habitat use based on resource acquisition. Thus, patterns of territorial space use by ShiningSunbeams may be characterized by core areas not determined by foraging behavior. Further studies examiningterritorial behaviors and the influence of intrusion pressure will help identify the underlying determinants ofterritory space use by this and other species of Andean hummingbirds.

RESUMEN. Patrones de uso del espacio territorial por el Colibr�ı cobrizo (Aglaeactis cupripennis),colibr�ı montano tropicalPara muchos colibr�ıes territoriales, el uso del h�abitat est�a influenciado principalmente por la interacci�on

entre la adquisici�on de recursos y los comportamientos no forrajeros, como el anuncio y la defensa territorial.Investigaciones previas han resaltado la importancia de las caracter�ısticas del h�abitat asociadas con el forrajeo,como la densidad y la distribuci�on de los recursos para determinar los patrones de uso del espacio de loscolibr�ıes. Sin embargo, se sabe menos acerca de c�omo la selecci�on del h�abitat asociada con loscomportamientos de no alimentaci�on influye en el uso del espacio por parte de las especies territoriales.Utilizamos la radiotelemetr�ıa para examinar los patrones de uso del espacio territorial por el Colibr�ı cobrizo(Aglaeactis cupripennis) en los bosques monta~nosos altoandinos cerca del Parque Nacional Manu, Per�u, y elBosque Comunal "El Carmen" cerca de Chordeleg, Ecuador. Cuantificamos las caracter�ısticas del h�abitatdentro del territorio relacionadas con la adquisici�on de recursos y los comportamientos no forrajeros, como elanuncio y defensa territorial. Descubrimos que el Colibr�ı cobrizo mostraba un alto uso de �areas centrales enterritorios donde el esfuerzo de b�usqueda de alimento era relativamente bajo. Sin embargo, no encontramosrelaci�on entre la posici�on de las �areas centrales y las caracter�ısticas del h�abitat asociadas con la defensa delterritorio, la evitaci�on de depredadores u otros comportamientos que no son de forrajeo. Tampocoencontramos relaci�on entre el uso de �areas no centrales y el uso del h�abitat basado en la adquisici�on derecursos. Por lo tanto, los patrones de uso del espacio territorial por el Colibr�ı cobrizo puede caracterizarse por�areas centrales no determinadas por el comportamiento de b�usqueda de alimento. Otros estudios queexaminen los comportamientos territoriales y la influencia de la presi�on de intrusi�on ayudar�an a identificar losdeterminantes subyacentes del uso del espacio territorial por esta y otras especies de colibr�ıes andinos.

Key words: Andes, montane forest, Manu National Park, Neotropics, Oreocallis, radio telemetry, Trochilidae

The territorial behavior of hummingbirdscan be influenced by factors such as resourceabundance and quality, prior-residence effects,and the identity and density of co-occurringcompetitors (Norton et al. 1982, Marches-seault and Ewald 1991, Temeles et al. 2004,

3Corresponding author. Email: jankowsk@biodi-versity.ubc.ca

4Present address: Bass Biology Centre, StanfordUniversity, Stanford, CA 94305,USA

© 2020 Association of Field Ornithologists

1

J. Field Ornithol. 91(1):1–12, 2020 DOI: 10.1111/jofo.12321

Kokko et al. 2006, Rousseu et al. 2014, Men-diola-Islas et al. 2016). The results of previ-ous studies have highlighted how these factorscan influence territory size, or the thresholdsneeded to maintain territorial behavior insome species (Norton et al. 1982, Eberhardand Ewald 1994, Justino et al. 2012). Forexample, natural variation and experimentalmanipulation of competitor density, flowerabundance, nectar volume, and nectar qualityhave been used to demonstrate that territorialhummingbirds respond to environmental cuesby adjusting the size of defended areas (Ewaldand Carpenter 1978, Hixon 1980, Hixonet al. 1983, Tamm 1985, Marchesseault andEwald 1991, Eberhard and Ewald 1994,Temeles et al. 2004, Hazlehurst and Karubian2018). Such effects can interact with reducedresource availability and lead to territoryexpansion and increased intrusion pressurethat may impose limits on the upper boundsof territory size (Hixon 1980, Tamm 1985,Eberhard and Ewald 1994).In addition to their impact on territory

size, similar environmental factors can influ-ence the within-territory dynamics of hum-mingbirds (Tamm 1985, Temeles et al. 2004,Garc�ıa-Meneses and Ramsay 2012, Justinoet al. 2012). The relationship betweenresource acquisition and territory defense mayproduce different space-use patterns by terri-tory owners. For example, elevated levels ofaggression and high visitation rates by resi-dent owners are correlated with patches con-taining many flowers (Stiles 1971, Wolf andHainsworth 1991, Garc�ıa-Meneses and Ram-say 2012, Missagia and Alves 2016). Territo-rial aggression also depends on the size ofcompetitors and the extent of resource overlap(Lyon et al. 1997, Dearborn 1998), whichmay vary across seasons or with time of day(Stiles 1971, Paton and Carpenter 1984). Forexample, some species of hummingbirdsspend more time near territory boundariesduring peak activity hours to aid in territorydefense and then later consolidate foragingefforts near territory cores (Paton and Car-penter 1984). Ultimately, overall patterns ofspace use by territorial hummingbirds arerelated to resource acquisition, territorydefense, and how these factors interact withfeatures of the habitat.To describe space-use patterns by territorial

hummingbirds, understanding how different

behaviors might be associated with particularhabitat characteristics is necessary. Previousstudies of the relationship between habitatand territoriality have largely focused on for-aging behavior, correlating hummingbird visi-tation rates to resource-dense patches (Justinoet al. 2012, Garc�ıa-Meneses and Ramsay2012, Jim�enez et al. 2012). However, less isknown about how non-foraging behaviorsand related habitat features might influencepatterns of space use, particularly structuralhabitat characteristics. For example, some veg-etation in territories may provide greater visi-bility or maximize signal transmission interritory advertisement (Garc�ıa-Meneses andRamsay 2012, Jim�enez et al. 2012, Justinoet al. 2012, Rousseu et al. 2014). Incorporat-ing such data into our understanding of theterritorial behavior of hummingbirds has beenchallenging due to the difficulty of trackingindividual hummingbirds and obtaining pre-cise, behaviorally explicit data on space use inhummingbird territories.We used radio telemetry to address this

limitation and describe patterns of space useby Shining Sunbeams (Aglaeactis cupripennis),a territorial hummingbird found throughouthigh-elevation regions in the tropical Andes(Schulenberg et al. 2007). Although not obli-gate specialist nectarivores (i.e., feeding fromfew flower species), much of their diet con-sists of nectar from the flowering tree, Oreo-callis grandiflora (Proteaceae), in regionswhere their ranges overlap (C�espedes et al.2019). This system provides a unique oppor-tunity to examine foraging and non-foragingdeterminants of territory space use by allow-ing quantification of resource availabilitywhile simultaneously tracking individuals todescribe the spatial distribution of behaviors(e.g., foraging, vocalizations, and territorialaggression). Using this approach, we correlatehummingbird behaviors with habitat features,specifically resource density and vegetationstructure, to evaluate space use in non-breed-ing territories. By considering habitat struc-ture, our goal was to describe the relativecontribution of non-foraging behavior indetermining patterns of hummingbird territo-riality. Specific objectives were to determine(1) if there are core areas in territories thatare used more frequently and, (2) if so,whether core areas and non-core areas havedifferent habitat structural features and

L. I. Pavan et al.2 J. Field Ornithol.

patterns of use. An important goal of ourstudy was to evaluate new ways to quantifythe local distribution of and habitat use byhummingbirds, which can help us betterunderstand interactions among species, forcesstructuring local communities, and conse-quences for pollination in flowering plantcommunities in the high Andes.

METHODS

Study site and focal species. Our studywas conducted on the eastern slope of theAndes at the Wayqecha Biological Stationlocated at the southern edge of Manu NationalPark in Peru (13°10029.6″S 71°35013.8″W),and the Bosque Comunal “El Carmen” in theAzuay province of central Ecuador (2°59005.1″S 78°44045.2″W). Both sites are at the bound-ary between tropical montane forest and eitherhigh-elevation Andean puna (Peru) or p�aramo(Ecuador). In these transitional regions, habitatstructure was relatively heterogeneous, with dis-tinct patches of montane forest separated byopen grasslands. These isolated patches of forestwere composed of trees belonging to commonmontane forest genera, such as Eugenia andWeinmannia, and Chusquea bamboo. Oreocallisgrandiflora was the most abundant ornithophi-lous flowering species at both locations and, inmany areas of our study sites, was the onlyavailable floral resource during our study.Shining Sunbeams were the most common

hummingbird observed during our study atboth locations. All individuals captured dur-ing our study displayed territorial behavior,but their sex was not determined. Flight dis-plays, vocalizations, and aggression towardother hummingbirds were observed for alltracked individuals. Territorial behavior wasoften initiated from perches in defendedO. grandiflora trees (L. Pavan, pers. observ.).Many Violet-throated Starfrontlets (Coelignaviolifer) and Sparkling Violetears (Colibri cor-uscans) were present at our Peruvian studysite, whereas Green-tailed Trainbearers (Lesbianuna) and Purple-throated Sunangels (Helian-gelus viola) were present at the Ecuadoriansite. Tyrian Metaltails (Metallura tyrianthina)were common at both locations. All non-tar-get species were observed intruding into theterritories of Shining Sunbeams and feedingon O. grandiflora flowers, but territorial intru-sions by conspecifics were more common

than intrusions by the non-target species(C�espedes et al. 2019).

Behavioral space use. We used radiotelemetry to obtain data on the distributionof feeding and non-feeding activities in terri-tories. Shining Sunbeams (N = 19) were cap-tured using mist-nets between 15 August2013 and 9 December 2015 (11 in Peru and8 in Ecuador). Territory owners were targetedby placing mist-nets in areas where individu-als exhibited territorial behavior (e.g., vocal-izations, visual displays, and aggressiontoward other hummingbirds). All 19 birdswere likely adults based on plumage and thedegree of corrugations on their bills. Radio-transmitters (0.25 g; Blackburn Transmitters,Nacogdoches, TX) were attached ~ 1 cmbelow the interscapular space on the back byapplying eyelash adhesive to the skin (DUO,American International Industries, Los Ange-les, CA), allowing transmitters to fall off andbe retrieved when feathers started to regrow.No apparent behavioral or flight differenceswere observed between individuals with andwithout transmitters; other investigators usingradio telemetry to study hummingbirds havealso reported no apparent negative effects(Hadley and Betts 2009, Zenzal et al. 2014,Volpe et al. 2016, Zenzal and Moore 2016).Shining Sunbeams are relatively large hum-mingbirds (mean = 7.04 � 0.37 [SD] g,N = 19), and tags were only placed on birdsif the weight of tags was < 5% of their bodymass (Kenward 2001).Tracking began one day after transmitter

deployment, and birds were then tracked for2–4 days using a TRS-10X receiver with a0.1-MHz bandwidth and 3-element Yagiantenna (Wildlife Materials). Tracking tookplace in two sessions each tracking day, eitherfrom 6:30 to 8:30 and 13:00 to 15:00, orfrom 9:00 to 11:00 and 15:30 to 17:30. Allbirds were tracked on consecutive days andduring the same time period each day. GPSlocations, elevation, and behavioral observa-tions were recorded every 10–15 min duringthe 2013 field season, and every 5 min in2014 and 2015. Behaviors were categorized asfeeding, vocalizing, displaying, aggression, orperching/movement (movement refers to indi-viduals observed moving between percheswhen the start and end locations were notobserved). Visits to either individual flowersor inflorescences were recorded as feeding

Space Use by an Andean HummingbirdVol. 91, No. 1 3

behavior. Displays were defined as non-vocalterritory advertisement and were most oftenobserved as flight displays. Aggression wasdefined as an antagonistic interaction withanother hummingbird (regardless of species)such as direct fighting or chasing behavior.All other observations were recorded as perch-ing/movement if they could not be catego-rized as another behavior type. Observationswere made using binoculars from no closerthan 10 m to avoid disturbance. We recorded50 locations, elevation measurements, andbehavioral observations for each of the fivebirds tracked in 2013. Because the frequencyof observations was higher for the 14 birdstracked in 2014 and 2015, we randomlyselected and analyzed 50 of these observationsto ensure equal sample sizes across individu-als.

Microhabitat use. Vegetation surveyswere conducted immediately after trackingeach of the 19 individuals in this study toquantify within-territory habitat variation. In2013, 10-m diameter sample plots were ran-domly positioned at 10 points in each terri-tory, as established using tracking data. Thelocation of each plot was determined by gen-erating randomly intersecting UTM coordi-nates bounded by territory dimensions.Coordinates were restricted so that sampleplots were non-overlapping and locatedentirely within territory boundaries. In 2014and 2015, surveys were conducted by estab-lishing one transect along the longest axis ofeach territory and a second transect perpen-dicular to the first. Similar 10-m diametersample plots were established every 15 malong each axis. In total, 112 vegetation plotsin 11 territories (seven from Peru and fourfrom Ecuador) were included in our study.Due to differences in sampling methods, veg-etation data from the territories of eighttracked individuals were not included in ouranalyses.Environmental variables measured in each

sample plot included canopy height, canopycover, number of O. grandiflora inflores-cences, and total number of all inflorescenceswith flowers larger than ~ 1 cm in corollalength. These habitat characteristics wereselected based on previous studies of habitatvariation and space use by songbirds (Barget al. 2006, Anich et al. 2012, Jimenez et al.2012). They provided a reasonable

approximation of habitat structure andresource density that allowed comparison offoraging and non-foraging (all behaviors notclassified as feeding) determinants of space-use patterns. Canopy cover was visually esti-mated as the percentage of visible skyimpeded by vegetative growth within the 10-m diameter sample circle at a point 2 mabove ground. Observations were treated as anominal variable and estimated as 5% cate-gories. Canopy height was estimated from thebase of the tallest tree in each sampling plot.We measured the distance from the groundto the highest accessible point and then visu-ally extrapolated to the highest visible pointto estimate overall canopy height. Given theconspicuousness and relatively low number ofO. grandiflora inflorescences in each sampleplot (10.8 � 4.5 [SD]), the open habitat,and low mean canopy height (4.4 � 0.4[SD] m), O. grandiflora inflorescences werecounted from the center point of each sampleplot. Inflorescences of all other species (in-cluding single flower inflorescences) werecounted by systematically moving in 1-mwide bands through entire 10-m diameterplots.

Statistical analysis. The total area ofterritories was calculated with a 90% kerneldensity estimate (KDE) for each of the 19territories using the package “adehabitat HR”(Calenge 2006) in R version 3.2.3 (R CoreTeam 2015). The 90% isopleth (the contourline describing 90% of the observed variationin the model) was selected as the upperboundary because estimates based on isopleths< 50% and > 90% have been shown to biasarea calculations (Borger et al. 2006). Thesmoothing parameter was calculated indepen-dently for each 90% KDE using a fixed refer-ence bandwidth (href). This smoothingparameter was selected because it bestresponds to the unimodal data distribution inour study (Worton 1989). Separate 90% ker-nel density estimates using smoothing param-eters calculated from a least squares crossvalidation function (hLSCV) resulted in statis-tically similar area estimates and so wereexcluded from this analysis.Behavioral differences in use of space at

varying distances from core areas were ana-lyzed using the mean 90% KDE calculatedfrom pooled territory sampling data. Each90% KDE was divided into a series of

L. I. Pavan et al.4 J. Field Ornithol.

concentric rings representing 15% isopleths(contours signifying the percent variationdescribed by the KDE model; in this case,each contour represents an additional 15% ofvariation explained). We calculated the areaof each 15% isopleth for all individuals andcompared the differences in area between suc-cessive isopleths. We defined the core area asthe first point where consecutive isopleths atleast doubled in area. We then took the meanacross all individuals and rounded to thenearest increment of 5%. The number ofoccurrences of Shining Sunbeams feeding andnon-feeding observations (vocalization, dis-playing, aggression, and perching/movement)in each 15% isopleth ring was standardizedfor the area of each ring by calculating anobservation density. The variance betweenobservation densities calculated for each ofthe six isopleths was unequal (Levene’s test,P < 0.05 for both observation types) so weused two separate Welch’s ANOVA tests withsix groups each to examine differences in thedensity of observations at different distancesfrom territory centroids. One test included allnon-feeding observations (vocalizing, display-ing, aggression, or perching/movement), andthe second included observations where birdswere feeding on O. grandiflora flowers orinflorescences. Both Welch’s ANOVA analy-ses were followed by Games-Howell multiplecomparison tests. All pairwise comparisonsbetween feeding and non-feeding observationsin both ANOVA models, however, were non-significant and so were not reported in thisanalysis.To evaluate the environmental characteris-

tics most associated with the position of coreareas, measured environmental variables wereanalyzed using four separate general mixedeffect models where individuals were treatedas a random effect. Each of the four environ-mental variables was independently comparedagainst the isopleth on which the center ofeach sample plot occurred (as determined bythe KDE isopleths). The goal of this compar-ison was to determine whether there were sig-nificant differences in the characteristics ofcore and non-core areas. Specifically, thisanalysis was focused on whether resourceabundance or habitat characteristics weremore strongly correlated with the location ofcore areas. Each general mixed effect modelwas evaluated using normally distributed

residual analysis and marginal goodness-of-fit(Nakagawa and Schielzeth 2013).Finally, to examine possible differences in

patterns of habitat use inside and outside ofcore areas, we subsetted the total number ofhummingbird observations (50 per individual)into those that directly overlapped with vege-tation survey plots. Overlapping humming-bird observations were divided into thoseinside (< 20% KDE) and outside (> 20%KDE) of core areas. The 20% KDE thresholdwas selected as the core area using the previ-ously described method outlined by Barget al. (2005). General mixed effect modelswere again used to compare the relationshipbetween the four measured habitat variablesand the proportion of overlapping humming-bird observations inside and outside of thecore area (N = 80 inside core area, N = 33outside core area). Each mixed effect modelwas evaluated using a normally distributedresidual analysis and marginal goodness-of-fit(Nakagawa and Schielzeth 2013). Values areprovided as means � 1 SD.

RESULTS

Behavioral space use. The distributionof the kernel probability densities calculatedfor all 19 territories indicated that space useby territorial hummingbirds was aggregatedaround central points or core areas (Fig. 1).The mean 90% KDE size of the 19 territorieswas 1559.8 � 465.3 m2 (range = 604.4–3923.0 m2). The mean size of core areas(20% isopleth) was 102.5 � 23.5 m2, indi-cating that 20% of the variation in each ker-nel density estimate could be described by anaverage of only 7% of overall territory size(Table S1). The mean offset distance betweenthe centroid and geometric centers of territo-ries was 7.9 � 2.0 m. This resulted in theprobability density function in each of the 19territories being distributed asymmetricallyaround the center point of core areas (Fig. 1).The mean density of non-feeding observa-

tions varied significantly at different distancesfrom the center of territories (Welch’sANOVA, F5, 747 = 38.8, P = 0.0001, N = 6).Mean densities of non-feeding observationswithin the 15% isopleth were 225% higher(0.28 � 0.03 observations/m2) than the over-all mean, whereas non-feeding observationsbetween the 30 and 90% isopleths were

Space Use by an Andean HummingbirdVol. 91, No. 1 5

significantly lower (Fig. 2). We also found asignificant difference in the mean densities offeeding observations at different distances fromthe center of territories (Welch’s ANOVA, F5,120 = 3.9, P = 0.0027, N = 6). This was dri-ven largely by fewer feeding observations in the

90% isopleth ring; pairwise comparisonsrevealed no significant difference in the densityof feeding observations between the 15 and60% isopleths (Fig. 2).

Microhabitat use. Comparisons of themean values of all four environmental

Fig. 1. Ninety percent kernel density estimates for five representative hummingbird territories out of thetotal 19. Territory estimates based on all observations for each radio-tagged individual (50 locations perindividual). Color corresponds to probability density, with contour lines indicating 10% isopleths. An adhoc smoothing parameter (href) was calculated independently for each estimate. Estimate parameters,areas, and positions are summarized in Table 1. [Colour figure can be viewed at wileyonlinelibrary.com]

L. I. Pavan et al.6 J. Field Ornithol.

variables and the distance to territory cen-troids (as determined by the KDE isopleths)revealed no significant differences (Table 1).This included both abundance of O. grandi-flora inflorescences (R2 = 0.16, t110 = �1.0,P = 0.27, N = 112) and canopy cover(R2 = 0.04, t110 = 0.7, P = 0.52, N = 112;Fig. 3), two habitat characteristics predictedto be important in microhabitat use. Thehabitat characteristic most strongly associatedwith core areas was canopy height, but thisregression also had a low fit to the observeddata (R2 = 0.27, t110 = �2.6, P = 0.09,N = 112).Patterns of microhabitat use outside of core

areas were also poorly explained by habitatcharacteristics (Table 2). Use of peripheral

territory habitat was not related to Oreocallisabundance (R2 = 0.11, t31 = �1.3, P = 0.21,N = 33), canopy cover (R2 = 0.07,t31 = �0.3, P = 0.78, N = 33), or canopyheight (R2 = 0.08, t31 = �1.5, P = 0.16,N = 33; Fig. 4).

DISCUSSION

Based on the concepts of space use by birds(Barg et al. 2006, Tomasevic and Marzluff2018), we expected that Shining Sunbeamterritories would be characterized by centralcore areas with high overall use, but low for-aging activity, and that core areas would beassociated with particular habitat structuralfeatures. We also expected that territory useoutside core areas would be associated withforaging behavior and related environmentalvariables. We found that Shining Sunbeamterritories were characterized by highly uti-lized central core areas, where foraging effortwas disproportionately low. Core areas werenot, however, associated with particular struc-tural habitat features, except for a weak asso-ciation with canopy height, and habitat useoutside of core areas was not related toresource density.By examining the distribution of non-feed-

ing and feeding observations relative to terri-tory centroids, we show that ShiningSunbeam territories were centered on coreareas with lower proportional foraging effort.The mean density of non-feeding observationsof Shining Sunbeams was significantly higherwithin the 15% isopleth than at any otherdistance from territory centers. In contrast,the mean density of feeding observations didnot show any differences until the 75% iso-pleth. If the likelihood of a foraging eventoccurring were equivalent for all points in aterritory, the distribution of feeding observa-tions would resemble the distribution of over-all observations (i.e., feeding plus non-feedingobservations), with foraging activity beingdetermined solely by the density of generalactivity. Instead, uniformity in the density offeeding observations suggests that relative for-aging effort was independent of the distribu-tion of overall observations and skewedtoward territory boundaries. This providesevidence for a territory structure defined byactivity clustered around a central point withproportionally less foraging effort.

A

B

Fig. 2. Mean density of (A) non-feeding and (B)feeding observations that occurred at different dis-tances (KDE isopleths) from each territory cen-troid. There was a significant difference betweennon-feeding groups (F5, 747 = 38.8, P = 0.0001)and also between feeding groups (F5, 120 = 3.9,P = 0.0027). Games-Howell post-hoc analysisindicates significant pairwise comparisons at the15% isopleth for non-feeding observations and forthe 75% isopleth for feeding observations. Signifi-cant pairwise comparisons are indicated by uniqueletters. Error bars are 95% confidence intervals,and each dashed line represents the overall meanbetween groups for each observation type.

Space Use by an Andean HummingbirdVol. 91, No. 1 7

Many studies of habitat use by birds haverevealed core areas of concentrated use in ter-ritories (e.g., Samuel et al. 1985, Barg et al.2006, Anich et al. 2012, Tomasevic and Mar-zluff 2018) and, in some cases, the location

of core areas has been found to be related toeither habitat characteristics associated withstructural complexity, such as canopy coverand vegetation density (Walsberg 1993, Anichet al. 2012), or the location of nesting androosting sites (Tomasevic and Marzluff 2018).In our study, however, most observationswithin the 15% isopleth involved perching/movement rather than other territorial orfeeding behaviors. In addition, core areas inthe territories of Shining Sunbeam were notassociated with any variables reflecting eitherstructural complexity or resource acquisition.This suggests that the location of core areasin Shining Sunbeam territories may be influ-enced by factors unrelated to vegetation struc-tural components associated with mitigationof environmental stressors such as predatoravoidance and territory defense (Lima 1998,Rousseu et al. 2014).The only environmental characteristic that

varied relative to the position of core areaswas canopy height, although this relationshipwas weak (Table 1). In some species, spaceuse in territories can be influenced by theneed to transmit auditory and visual signals(Krams 2001, Barg et al. 2006). In thesecases, core areas with more exposed, elevatedperches may be favored (Hunter 1980, Barget al. 2006). Although not explicitly tested inour study, use of auditory and visual signalsfor territory advertisement by hummingbirdscould influence the relationship between loca-tion of core areas and canopy height (Ewaldand Bransfield 1987, Ornelas et al. 2002). Inaddition, use of exposed perches by hum-mingbirds is often associated with detectionof intruders, leading to a correlation betweenterritorial movements and visibility (Rousseuet al. 2014, Lanna et al. 2016). Further directcomparisons between the density of exposedperches and space-use patterns in territorialhummingbirds are needed to test these rela-tionships.

Table 1. Summary of the linear regression analyses performed on the relationship between four measuredhabitat variables and the distance from the territory center.

Habitat variable Linear regression P t N R2

Canopy height y = �0.016x + 4.77 0.09 �2.6 112 0.27Abundance of Oreocallis inflorescences y = �0.13x + 12.60 0.33 �1.0 112 0.16Canopy cover y = 0.068x + 18.67 0.52 0.7 112 0.04Abundance of total inflorescences y = �0.0033x + 1.92 0.82 �0.2 112 0.43

A

B

Fig. 3. Linear regression of (A) number of Oreo-callis inflorescences and (B) canopy cover com-pared to the distance from the territory center(KDE Isopleth). Neither environmental variablewas significantly related to the distance from theterritory center (number of Oreocallis inflores-cences: R2 = 0.16, t = �1.0, P = 0.33; canopycover: R2 = 0.04, t = 0.7, P = 0.52). Of the fourmeasured environmental variables, canopy heightshowed the strongest relationship, although theestimated fit of the regression was low (canopyheight: R2 = 0.27, t = �2.6, P = 0.09). A com-plete summary of results is provided in Table 1.Lines represent estimated lines of best fit.

L. I. Pavan et al.8 J. Field Ornithol.

We predicted, based on models of micro-habitat use in territories by songbirds, thatthe density of floral resources would be mostcorrelated with points of high hummingbirduse outside of core areas (Anich et al. 2012,Jimenez et al. 2012), but no measured envi-ronmental variable varied significantly withuse of areas on the periphery of territories.Although we found that, proportionally, for-aging effort was skewed toward territoryboundaries, habitat characteristics associatedwith foraging may not be sufficient todescribe use of peripheral territory space. Theobserved patterns of microhabitat use outsideof territory core areas may represent a trade-off between resource acquisition and minimiz-ing risk associated with environmental stres-sors such as predation pressure (Lima 1998).This may result in a pattern of peripheralhabitat use where environmental variablesassociated with foraging or pressures like pre-dation risk or aggressive interactions may notbe the sole determinants of habitat use. Moredetailed information about the flowers andinsects used as food resources may betterinform these territorial patterns. Alternatively,peripheral space use may be influenced byinteractions with neighboring hummingbirds.Intrusion pressure influences the size andshape of hummingbird territories and canlead to the periphery of territories being usedfor territory defense and vigilance (Nortonet al. 1982, Paton and Carpenter 1984, Rous-seu et al. 2014). We are unable to saywhether this is the case for Shining Sunbeamterritories because this would require directcomparisons between peripheral territory

space-use patterns observed under differentconditions of territory intrusion.Few investigators have attempted to use

radio telemetry with hummingbirds. Trans-mitters have been successfully employed totrack movements of Green Hermits(Phaethornis guy) in fragmented lower mon-tane forest areas of Costa Rica, and no differ-ences were noted in flight or other behaviorsof tagged and untagged individuals (Hadleyand Betts 2009, Volpe et al. 2016; also seeZenzal et al. 2014). Thus, for larger-bodiedhummingbirds (e.g., where tags are < 3–5%of body mass), telemetry is a highly tractablemethod for monitoring movements of indi-viduals, quantifying space use in territories,and assessing foraging behavior. Our resultssuggest that this technique could be employedfor a number of other tropical hummingbirdspecies across a diversity of habitats and eleva-tions to understand variation in territorialand traplining behaviors.Our results have important implications

not only for hummingbird behavioral studies,but for broader plant–pollinator interactionsand the role that territorial behavior may playin shaping the movement of plant genes.Resource monopolization by territorial hum-mingbirds can limit overall pollen movement,leading to inbreeding and reduced reproduc-tive output of defended plants (Waddington1983, Garc�ıa-Meneses and Ramsay 2012,Rousseu et al. 2014). Territorial intrusions bycompeting hummingbirds is one way thesebarriers are overcome. Because territory intru-sion is most likely to be successful in areasless frequented by territory owners, the

Table 2. The summary of the analyses performed on the relationship between four measured habitat vari-ables and the proportion of hummingbird observations overlapping vegetation plots inside (< 20% KDE;N = 80) and outside (> 20% KDE; N = 33) territory core areas.

Linear regression P t R2

Habitat variableCanopy height y = 0.0016x + 0.0068 0.06 1.9 0.07Canopy cover y = 0.000078x + 0.015 0.38 0.9 0.06Abundance of total inflorescences y = 0.0000057x + 0.016 0.51 0.7 0.06Abundance of Oreocallis inflorescences y = �0.0000059x + 0.016 0.93 �0.1 0.04

Habitat variableCanopy height y = �0.0064x + 0.043 0.16 �1.5 0.08Abundance of Oreocallis inflorescences y = �0.00051x + 0.015 0.21 �1.3 0.11Abundance of total inflorescences y = �0.000016x + 0.0090 0.62 �0.5 0.08Canopy cover y = �0.000089x + 0.011 0.78 �0.3 0.07

Space Use by an Andean HummingbirdVol. 91, No. 1 9

probability of intrusion is often reciprocal toterritorial space-use patterns (Paton and Car-penter 1984, Franceshinelli and Kesseli 1999,Garc�ıa-Meneses and Ramsay 2012). Addi-tional studies employing radio telemetry will,therefore, be key to understanding the struc-ture of hummingbird territories and possibleimplications for both the movement of plantgenes and the maintenance of genetic diver-sity in high-elevation Andean plant communi-ties (Hazlehurst et al. 2016).Overall, we found that space use by Shin-

ing Sunbeams was similar to that described

for some species of songbirds, especially withthe presence of core areas in territories, butwe also found reduced foraging effort inthese core areas. We found no evidence thatthis core area pattern was due to differentialhabitat use based on either maximizingresource acquisition or other non-foragingbehaviors. Although somewhat unexpected,these results may be broadly applicable tohummingbird foraging biology, especiallywhen considering pollination interactionsand the effects of pollinator behavior on pol-len movement.

Fig. 4. Proportion of total Shining Sunbeam observations inside (< 20% KDE isopleth) and outside(> 20% KDE isopleth) the core areas of each territory compared to three different environmental vari-ables: canopy cover, Oreocallis abundance, and canopy height (N = 80 inside core area, N = 33 outsidecore area). Proportion of Shining Sunbeam observations outside of the 20% KDE isopleth was not cor-related with Oreocallis abundance following linear regression analysis (R2 = 0.11, t = �1.3, P = 0.21), afactor predicted to be a major driver of use of peripheral habitat. None of the four measured environ-mental variables was significantly correlated with patterns of habitat use by Shining Sunbeams in non-core areas (Table 2). Lines represent estimated lines of best fit.

L. I. Pavan et al.10 J. Field Ornithol.

ACKNOWLEDGMENTS

We are indebted to all of our field crews thatassisted with data collection throughout this project.We are especially grateful to Esther Vallejo Santamariaand Laura C�espedes for their expertise. Previous draftsof this manuscript were greatly improved by commentsby Gary Ritchison and two anonymous reviewers.Research funding was provided by a Dean of ScienceUndergraduate Research Award (to L. I. Pavan) and bya Natural Science and Engineering Research CouncilDiscovery Grant F11-05365 (to J. E. Jankowski). Cap-ture and handling procedures were approved by UBCAnimal Care Protocol #A15-0067.

LITERATURE CITED

ANICH, N. M., T. J. BENSON, AND J. C. BEDNARZ.2012. What factors explain differential use withinSwainson’s Warbler (Limnothylpis swainsonii)home ranges? Auk 129: 409–418.

———, D. V. AIAMA, J. JONES, AND R. J.ROBERTSON. 2006. Within-territory habitat useand microhabitat selection by male CeruleanWarblers (Dendroica cerulea). Auk 123: 795–806.

———, ———, and ———. 2005. Describingbreeding territories of migratory passerines:suggestions for sampling, choice of estimator, anddelineation of core areas. Journal of AnimalEcology 74: 139–149.

BORGER, L., N. FRANCONI, G. DE MICHELE, A. GANTZ,F. MESCHI, A. MANICA, S. LOVARI, AND T.COULSON. 2006. Effects of sampling regime on themean and variance of home range size estimates.Journal of Animal Ecology 75: 1393–1405.

CALENGE, C. 2006. The package “adehabitat” for theR software: a tool for the analysis of space andhabitat use by animals. Ecological Modelling 197:516–519.

C�ESPEDES, L. N., L. I. PAVAN, J. A. HAZLEHURST, AND

J. E. JANKOWSKI. 2019. The behavior and diet ofthe Shining Sunbeam (Aglaeactis cupripennis): aterritorial high-elevation hummingbird. WilsonJournal of Ornithology 131: 24–34.

DEARBORN, D. C. 1998. Interspecific territoriality by aRufous-tailed Hummingbird (Amazilia tzacatl):effects of intruder size and resource value.Biotropica 30: 306–313.

EBERHARD, J. R., AND P. W. EWALD. 1994. Foodavailability, intrusion pressure and territory size:an experimental study of Anna’s Hummingbirds(Calypte anna). Behavioral Ecology andSociobiology 34: 11–18.

———, and R. J. BRANSFIELD. 1987. Territory qualityand territorial behavior in two sympatric species ofhummingbird. Behavioral Ecology andSociobiology 20: 285–293.

———, and F. L. CARPENTER. 1978. Territorialresponses to energy manipulations in the AnnaHummingbird. Oecologia 31: 277–292.

FRANCESHINELLI, E. V., AND R. KESSELI. 1999.Population structure and gene flow of theBrazilian shrub Helicteres brevispira. Heredity 82:355–386.

GARC�IA-MENESES, P. M., AND P. M. RAMSAY. 2012.Pollinator response to within-patch spatial contextdetermines reproductive output of a giant rosetteplant. Basic and Applied Ecology 13: 516–623.

HADLEY, A. S., AND M. G. BETTS. 2009. Tropicaldeforestation alters hummingbird movementpatterns. Biology Letters 5: 207–210.

———, and J. O. KARUBIAN. 2018. Impacts of nectarrobbing on the foraging ecology of a territorialhummingbird. Behavioural Processes 149: 27–34.

———, B. TINOCO, S. C�ARDENAS, AND J. KARUBIAN.2016. Pollination ecology of Oreocallis grandiflora(Proteaceae) at the northern and southern ends ofits geographic range. Journal of PollinationEcology 19: 71–80.

HIXON, M. A. 1980. Food production and competitordensity as the determinants of feeding territorysize. American Naturalist 115: 510–530.

———, F. L. CARPENTER, AND D. C. PATON. 1983.Territory area, flower density, and time budgetingin hummingbirds: an experimental and theoreticalanalysis. American Naturalist 122: 366–391.

HUNTER, M. L. 1980. Microhabitat selection forsinging and other behaviour in Great Tits (Parusmajor): some visual and acoustical considerations.Animal Behaviour 28: 468–475.

JIMENEZ, L., S. NEGRETE-YANKELEVICH, AND R.MACIAS-ORDONEZ. 2012. Spatial associationbetween floral resources and hummingbird activityin a Mexican tropical montane cloud forest.Journal of Tropical Ecology 28: 497–506.

JUSTINO, D. G., P. K. MARUYAMA, AND P. E.OLIVIERA. 2012. Floral resource availability andhummingbird territorial behaviour on aNeotropical savanna shrub. Journal ofOrnithology 153: 189–197.

KENWARD, R. E. 2001. A manual for wildlife radiotagging. Academic Press, London, UK.

KOKKO, H., A. LOPEZ-SEPULCRE, AND L. J. MORRELL.2006. From hawks and doves to self-consistentgames of territorial behavior. American Naturalist167: 901–912.

KRAMS, I. 2001. Perch selection by singingChaffinches: a better view of surroundings and therisk of predation. Behavioral Ecology 12: 295–300.

LANNA, L. L., C. S. DE AZEVEDO, R. M. CLAUDINO,R. OLIVIERA, AND Y. ANTONINI. 2016. Perchusage by hummingbirds in a fragment of AtlanticForest in Brazil. Wilson Journal of Ornithology128: 453–459.

LIMA, S. L. 1998. Nonlethal effects in the ecology ofpredator-prey interactions. BioScience 48: 25–34.

LYON, D. L., J. CRANDALL, AND M. MCKONE. 1997.A test of the adaptiveness of interspecificterritoriality in the Blue-throated Hummingbird.Auk 94: 448–454.

MARCHESSEAULT, L., AND P. W. EWALD. 1991. Effectof territory quality on intrusion rate innonbreeding hummingbirds. Behavioral Ecologyand Sociobiology 28: 305–308.

MENDIOLA-ISLAS, V., C. LARA, P. CORCUERA, AND P.L. VALVERDE. 2016. Residency in White-earedHummingbirds (Hylocharis leucotis) and its effectin territorial contest resolution. PeerJ 4: e2588.

Space Use by an Andean HummingbirdVol. 91, No. 1 11

MISSAGIA, C. C. C., AND M. A. S. ALVES. 2016.Territorial foraging behavior in the male Violet-capped Woodnymph is dependent on the densityof patches of inflorescences of Heliconiaspathocircinata Aristeg (Heliconiaceae) in theBrazilian Atlantic Forest. Revista Brasileira deBotanica 39: 1145–1150.

NAKAGAWA, S., AND H. SCHIELZETH. 2013. A generaland simple method for obtaining R2 fromgeneralized linear mixed-effects models. Methodsin Ecology and Evolution 4: 133–142.

NORTON, M. E., P. ARCESE, AND P. W. EWALD. 1982.Effect of intrusion pressure on territory size inBlack-chinned Hummingbirds (Archilocusalexandri). Auk 99: 761–764.

ORNELAS, J. F., C. GONZ�ALEZ, AND J. URIBE. 2002.Complex vocalizations and aerial displays of theAmethyst-throated Hummingbird (Lampornisamethystinus). Auk 119: 1141–1149.

PATON, D. C., AND F. L. CARPENTER. 1984. Foragingby territorial Rufous Hummingbirds: defense byexploitation. Ecology 65: 1808–1819.

R CORE TEAM. 2015. R: A language and environmentfor statistical computing. R Foundation forStatistical Computing, Vienna, Austria.

ROUSSEU, F., Y. CHARETTE, AND M. BELISLE. 2014.Resource defense and monopolization in a markedpopulation of Ruby-throated Hummingbirds(Archilocus colubris). Ecology and Evolution 4:776–793.

SAMUEL, M. D., D. J. PIERCE, AND E. O. GARTON.1985. Identifying areas of concentrated use withinthe home range. Journal of Animal Ecology 54:711–719.

SCHULENBERG, T. S., D. F. STOTZ, D. F. LANE, J. P.O’NEILL, AND T. A. PARKER III. 2007. Birds ofPeru. Princeton University Press, Princeton, NJ.

STILES, F. G. 1971. Time, energy, and territoriality ofthe Anna Hummingbird (Calypte anna). Science173: 818–21.

TAMM, S. 1985. Breeding territory quality andagonistic behavior: effects of energy availability andintruder pressure in hummingbirds. BehavioralEcology and Sociobiology 16: 203–207.

TEMELES, E. J., A. B. MUIR, E. B. SLUTSKY, AND M.N. VITOUSEK. 2004. Effect of food reductions on

territorial behavior of Purple-throated Caribs.Condor 106: 691–695.

TOMASEVIC, J. A., AND J. M. MARZLUFF. 2018. Spaceuse of suburban Pileated Woodpeckers (Dryocopuspileatus): insights on the relationship betweenhome range, core areas, and territory. Oecologia187: 15–23.

VOLPE, N. L., W. D. ROBINSON, S. J. K. FREY, A. S.HADLEY, AND M. G. BETTS. 2016. Tropical forestfragmentation limits movements, but notoccurrence of a generalist pollinator species. PLoSONE 11: e0167513.

WADDINGTON, K. D. 1983. Pollen flow and optimaloutcrossing distance. American Naturalist 122:147–151.

WALSBERG, G. E. 1993. Thermal consequences ofdiurnal microhabitat selection in a small bird.Ornis Scandinavica 24: 174–182.

WOLF, L. L., AND F. R. HAINSWORTH. 1991.Hummingbird foraging patterns: visits to clumpsof Ipomopsis aggregata inflorescences. AnimalBehaviour 41: 803–812.

WORTON, B. J. 1989. Kernel methods for estimatingthe utilization distribution of home-range studies.Ecology 70: 164–168.

ZENZAL, T. J., R. H. DIEHL, AND F. R. MOORE. 2014.The impact of radio-tags on Ruby-throatedHummingbirds (Archilochus colubris). Condor116: 518–526.

———, and ———. 2016. Stopover biology ofRuby-throated Hummingbirds (Archilochuscolubris) during autumn migration. Auk 133:237–250.

SUPPORTING INFORMATION

Additional Supporting Information may befound in the online version of this article atthe publisher’s website.

Table S1. Summary of kernel density esti-mated areas, parameters, and the position ofthe core area for hummingbird territories 1-19.

L. I. Pavan et al.12 J. Field Ornithol.