Avian Behavior, Ecology, and Evolution

INTRODUCTION

Numerous bird species differ only subtly in external phenotype but produce strikingly different sounds. Acoustic traits are principal signals used in mate choice and species recognition, making vocal traits central to understanding reproductive isolation and speciation (Catchpole and Slater 1995). Vocalizations are also essential for field identification: For example, previous work has shown that acoustic traits can distinguish cryptic species of Rallus rails (Stiffler et al. 2018), Glaucidium owlets (Gwee et al. 2019), Bubo owls (Movin et al. 2022), Tanysiptera kingfishers (Sin et al. 2022), Sclerurus leaftossers (Cooper and Cuervo 2017), Troglodytes wrens (Toews and Irwin 2008), Camptostoma tyrannulets (Lima and Vaz 2024), and Phylloscopus warblers (Irwin 2000), as well as species of fish (Parmentier et al. 2022), frogs (Wilczynski et al. 1993), and nocturnal primates (Braune et al. 2008). Characterizing vocal traits therefore provides a powerful tool for identifying cryptic taxa, guiding research on ecology, taxonomy, and evolutionary divergence (Araya-Salas and Wright 2013).

The hummingbirds (family Trochilidae) are one of three bird families that independently evolved vocal learning (Jarvis et al. 2000, Hackett et al. 2008). Yet, hummingbird vocalizations are seldomly studied, perhaps because most species produce largely single-note songs (e.g., White-bellied Emerald (Chlorestes candida), Atwood et al. (1991), and Long-billed Hermit (Phaethornis longirostris), Araya-Salas and Wright (2013). However, a number of hummingbird species have intermediate vocal complexity (Baptista and Schuchmann 1990, Gaunt et al. 1994) and some rival passerines in their phonological and syntactic song intricacy (e.g., Ficken et al. 2000, Jarvis et al. 2000, Ornelas et al. 2002, González and Ornelas 2005, 2009, Ferreira et al. 2006, González et al. 2011). The diversity of vocal repertoires, independent evolution of song learning, and ability to learn complex songs makes the hummingbirds an interesting family in which to study vocal differences between sister taxa.

The giant hummingbird clade (genus Patagona) contains two species that diverged 2.1–3.4 Mya and that differ in breeding latitude, migratory behavior, and respiratory physiology (Williamson and Witt 2021a, Williamson et al. 2023, 2024, 2025b). While the two differ in average plumage patterns and external measurements, these differences may be insufficient to identify all individuals (Williamson et al. 2025a). Both giant hummingbird species are found in arid and semi-arid habitats and in intermontane valleys of the Andes, including scrub, hedgerows, agricultural areas, open woods (including Polylepis), and gardens, often with Agave, Puya, and Cactaceae spp. Across their ranges, they tend to associate with fast-flowing streams. The Northern Giant Hummingbird (Patagona peruviana) is a high-elevation resident known to breed in Ecuador, Peru, and northern Chile that occurs at ~1800–4300 meters (m) throughout the north-central Andes (Ridgely and Greenfield 2001, Jaramillo 2003, Schulenberg et al. 2010, Williamson et al. 2025a). The Southern Giant Hummingbird (Patagona gigas) breeds from sea level to ~2500 m in central Chile, above ~2500 m in northwest Argentina, and ~2100–4100 m in Bolivia. Southern latitude populations are long-distance migrants: During the austral winter (~March–Sept), Chilean-breeding Southern Giant Hummingbirds migrate north to tropical latitudes of the central Peruvian Andes, making a physiologically-taxing 4000-m ascent (Williamson et al. 2024, Ivy and Williamson 2024). During migration, at least some Southern Giant Hummingbird breeding populations from northwest Argentina also migrate as far as central Peru, but further details about migratory behavior and connectivity remain to be studied. During the nonbreeding season, the two species frequently co-occur at the same sites and elevations.

While the giant hummingbirds have diverged in many aspects of their ecology and physiology, they are phenotypically cryptic and difficult to distinguish—a challenge that has hindered attempts to characterize their range limits and seasonal co-occurrence (Williamson et al. 2024, 2025a). The two species have near-identical body mass distributions and differ only subtly in plumage, with a few key features permitting species identification. Throat color and pattern, eye-ring, post-ocular spot, wing length, bill length, and tail length tend to be informative (Williamson et al. 2025a). However, even among adults, the range of plumage variation, particularly in throat color and pattern, overlaps within and between species (Williamson et al. 2024). While the Northern Giant Hummingbird averages larger in bill length, wing chord, tail length, and tarsus length, there is substantial overlap with the Southern Giant Hummingbird, such that only ~65–85% of individuals can be identified to species using measurements when comparing sexes separately and with a trained linear discriminant model (Williamson et al. 2024, 2025a). Many individuals may be best left unidentified when observed at sites and during seasons of co-occurrence.

Although the two species are largely allopatric during their respective breeding seasons, Williamson et al. (2024) reported a male Northern x Southern Giant Hummingbird F1 hybrid from Lima Department, Peru, in September. The hybrid had Peruvian and Argentinian ancestry, and its occurrence suggests breeding range overlap between the two species (Williamson et al. 2024). These findings raise questions about each species’ seasonal range limits and the potential role of vocalizations in species recognition.

The principal vocalizations of both giant hummingbird species are comprised of a repeated series of simple, single notes that have been described previously in the literature as “a high-pitched cwueet” (Ridgely and Greenfield 2001), “a squeaky, loud heeee” (Schulenberg et al. 2010), a “characteristic melancholy pipping note” (Koepcke 1983), “a squeaky zeet” (Herzog et al. 2016), “a piercing SIP whistle” (Pearman and Areta 2020), and “a loud single chip note” (Jaramillo 2003). These vocalizations are given by both species when advertising to mates, defending territories, and indicating presence in an area (in a stationary position from perches, while feeding, and in flight) on both the breeding and non-breeding grounds. These principal vocalizations have been referred to in the literature variably as either songs or calls. The two species additionally make a variety of other irregular “chatter” vocalizations during the breeding and non-breeding seasons, most often during aggressive in-flight chases with congeners, which have been described as “squeaky whistles and trills” (Velásquez-Noriega et al. 2023). However, vocal repertoire has never been studied in the giant hummingbirds, and Patagona remains the only major hummingbird clade for which no systematic characterization of vocalizations has ever been conducted (Duque and Carruth 2022). Previous work has suggested that some species of hummingbirds are open-ended song learners, capable of modifying songs into adulthood (Araya-Salas and Wright 2013, Johnson and Clark 2020, 2024). Whether Patagona vocalizations constitute “songs” or “calls,” as well as whether Patagona species are capable of song learning, remains unknown. We thus refer here to the principal, single-note vocalizations given by both Patagona species as “vocalizations,” rather than songs, despite their apparent role in territorial aggression, resource defense, and mate advertisement.

Here, we sought to characterize variation in the principal vocalizations of the two giant hummingbird species. We collected field recordings of both species during both the breeding and nonbreeding seasons from Peru, Bolivia, and Chile, which we combined with recordings from public repositories to study vocal trait variation across the range of the genus. We first conducted a range-wide, qualitative spectrographic analysis of Patagona recordings to classify vocal differences, identifying two distinct vocal types that corresponded to the two species: Type 1 to the Northern Giant Hummingbird and Type 2 to the Southern Giant Hummingbird. We next quantified trait variation to evaluate the extent to which the two species differ in their vocalizations, using both breeding range and full annual data. Finally, we tested whether vocalizations can be used to distinguish the Northern and Southern Giant Hummingbirds—particularly in non-breeding zones of overlap (i.e., sympatry). We predicted finding subtle variation in both species’ principal vocalizations that mirrors the level of divergence observed in plumage and morphological comparisons. Accordingly, we posited that models trained to classify vocalizations to species level would have similar low to moderate predictive power as models previously trained using plumage and morphology (Williamson et al. 2024). Alternatively, divergent vocalizations would indicate that vocal traits may be a principal signaling mechanism, as has been shown for other cryptic complexes (Wilczynski et al. 1993, Irwin 2000, Toews and Irwin 2008, Braune et al. 2008, Ng and Rheindt 2016, Cooper and Cuervo 2017, Stiffler et al. 2018, Gwee et al. 2019, Movin et al. 2022, Parmentier et al. 2022, Sin et al. 2022, Lima and Vaz 2024).

METHODS

Field recording

In 2023 and 2025, J.L.W. collected 53 recordings of the two giant hummingbird species during both opportunistic and targeted fieldwork in Peru, Chile, and Bolivia. Fieldwork took place in Lima and Ancash Departments of Peru in July–Aug 2023, Lima Department in Jan 2025, in Valparaíso Region of central Chile during Jan–Feb 2025, and in La Paz and Cochabamba Departments of Bolivia during Feb 2025. During field recording, J.L.W. visited sites where giant hummingbirds had been recently seen or reported (per the eBird database, personal knowledge of field sites, and suggestions from local guides), as well as sites whose habitat appeared suitable for one or both species. Upon finding a focal giant hummingbird, J.L.W. carefully observed the individual and took photographs and recordings, to the extent possible. Most recordings (>81%) were captured with a Zoom F3 Field Recorder and Sennheiser ME67 shotgun microphone, but several were taken using an iPhone 14 Pro with either the Voice Memos or Merlin apps. All recordings, as well as representative photographs, were archived in the Macaulay Library (ML; https://www.macaulaylibrary.org/) at the Cornell Lab of Ornithology.

Vocal data collection

We compiled 217 total vocalization recordings of Northern and Southern Giant Hummingbirds made between 1976–2025, including 53 recordings collected in the field by the authors, 189 recordings from the Macaulay Library, and 28 recordings from Xeno-canto (XC; https://www.xeno-canto.org). We used all data available as of 15 February 2025. The compiled vocal dataset spanned the breeding and non-breeding seasons of both species. Once aggregated, we compared recordist names, recording dates, and localities to ensure each recording was unique between ML and XC and within the same eBird checklist (if applicable). We georeferenced missing latitudes and longitudes for 14 recordings missing coordinates, as described in Williamson and Witt (2021b). Each recording contained at least one single-note, principal vocalization (Fig. 1); many also contained a variety of flight and chatter calls. Because calls varied substantially in length, pitch, and frequency and were often given in response to interactions with conspecifics or other species, we analyzed only principal vocalizations.

Processing and annotation of recordings

We manually measured the following variables from spectrograms of original .wav, .mp3, and .m4a files of each vocalization recording: Start time (secs), end time (secs), minimum or lowest frequency in kilohertz (kHz), maximum or highest frequency (kHz), and peak frequency, or the frequency at which the sound had the highest amplitude (kHz). We scored three vocalizations (i.e., unique notes) per recording of each individual, and when possible, scored three consecutive vocalizations. When fewer than three vocalizations were available or scorable, we analyzed all scorable vocalizations. We assigned each recording a coarse quality score (“low,” “medium,” “high”) and noted whether audio files were compressed or noise reduced. We excluded recordings from our analysis that contained disruptive background noise or distortion that was not part of the original signal vocalization, strong echoes, and/or atypical chatter notes (57 recordings total, or ~26.2%). To ensure consistency, all recordings were resampled to a common sample rate of 48 kHz and spectrograms were analyzed using a standardized 512-sample Hann window (3dB filter, bandwidth 135 Hz), with 50% overlap and a hop size of 256 samples. All time and frequency measures were derived from spectrograms. Peak frequency measurements were automatically calculated by Raven once the annotations were drawn. To ensure consistency, R.J.Z. conducted all scoring in Raven Pro v.1.6.5 (K. Lisa Yang Center for Conservation Bioacoustics at the Cornell Lab of Ornithology 2024).

Our quality-filtered vocalization dataset included recordings from 160 individuals of both species. Using raw data scores for each recording, we calculated the duration of each vocalization (end time minus start time; in secs) and bandwidth, also referred to as frequency range (maximum frequency minus minimum frequency); two commonly used parameters in vocal analysis (Odom et al. 2021). For each individual, we then calculated a mean from the measurements of all scored vocalizations for each parameter, including: 1) minimum frequency, 2) maximum frequency, 3) peak frequency, 4) duration, and 5) bandwidth, as described in Toews and Irwin (2008). For 34 individuals, it was necessary to calculate mean trait values from only two scored vocalizations; for nine individuals, we took scores from a single vocalization. We then used mean measurements from each individual in downstream analyses.

Qualitative classification of vocal types

We qualitatively assessed spectrograms based on shape and sound to gain an understanding of giant hummingbird vocalizations, as in (Araya-Salas and Wright 2013). We identified two principal vocal types: The first, designated as “Type 1,” appeared as an initial upward-sloping note leading to a short, horizontal segment. By ear, the Type 1 vocalization was recognizable as a high-pitched, dry and thin, “tsee!” (Fig. 1; Appendix 1: Fig. S1). The second vocal type, designated as “Type 2,” appeared long and straight on the right with a tight, curved bend to the left at the top of the note (e.g., the shape of an inverted capital letter “J” or candy cane). By ear, the Type 2 vocalization was recognizable as a loud “tsiP!” with an abrupt ending (Fig. 1, Appendix 1: Fig. S2). Shapes and frequencies were consistent within each Type 1 and Type 2 vocalizations, with minimal within-type variation (Appendix 1: Figs. S1-S2). We classified all but one recording as Type 1 or Type 2. The unclassified individual, represented by ML410550051, had an anomalous shape, possibly due to recording environment and sound quality, and this individual was thus conservatively classified as “unknown” (Appendix 1: Fig. S3).

We next analyzed the distributions of Type 1 and Type 2 vocalizations across space and time. This effort revealed connections among vocal type, season, and geography consistent with species identity (Appendix 1: Figs. S4–S5). Specifically, Type 1 vocalizations were recorded from Ecuador (roughly year-round), Peru (Jan–Mar, May–Aug, and Oct), west-central Bolivia (Feb), and northern Chile (Aug (once), Appendix 1: Figs. S4–S5). Type 1 vocalizations were not observed from central Chile or Argentina during any month of the year. Type 2 vocalizations were recorded from Peru in March (once), May, July–Aug, Sept, Oct (once), and Dec (once); throughout Bolivia Jan–Feb, May, Sept, Dec; Argentina (Feb, March, Aug, Oct); northern Chile (July, Sept); and central Chile (Feb, Aug, Oct–Dec; Appendix 1: Figs. S4–S5). Type 2 vocalizations were not observed from Ecuador. Thus, Type 1 distribution patterns were consistent with those of the Northern Giant Hummingbird, and Type 2 with those of the Southern Giant Hummingbird (Williamson et al. 2025a).

To further test the connection between vocal type and species identity, we analyzed available extended data connected to a subset of recorded individuals: These data included photos taken by recordists, as well as morphological measurements, genetic samples, and real-time tracking data collected by the authors (Appendix 1: Figs. S1–S2). In all instances in which we had extended data from recorded individuals, Type 1 vocalizations corresponded to the Northern Giant Hummingbird and Type 2 vocalizations to the Southern Giant Hummingbird. As such, we classified vocal Type 1 as belonging to the Northern Giant Hummingbird and vocal Type 2 as belonging to Southern Giant Hummingbirds for downstream analyses of vocal variation.

Vocal trait datasets

We divided the data into breeding range and full annual range (the latter included data from both the breeding and non-breeding ranges) subsets for separate analyses. The more conservative breeding range subset enabled us to characterize the vocalizations of the giant hummingbirds using individuals of known species identity from known core areas of the breeding range, while data from the full annual range enabled us to test how our model performed when identifying individuals of unknown species identity. In the breeding range dataset, species identity was determined using a combination of month, locality, vocal type, and if possible, extended data associated with recordings (including plumage characteristics from accompanying photos taken in the field or available in eBird checklists, morphological measurements, genetic identifications, and movement tracking data). The Northern Giant Hummingbird breeding range subset included individuals from Ecuador (year-round), Peru (September–April; principal breeding months), and Arica y Parinacota Region of Chile (year-round). The Southern Giant Hummingbird breeding range subset included individuals from Chile (south of Arica y Parinacota; year-round) and Argentina (year-round). Although specific breeding periods may differ across latitudes, our intent was to capture individuals of known identity within the breeding range, not necessarily during breeding windows. For example, although Southern Giant Hummingbirds breed principally from Oct–Jan in central Chile, any giant hummingbird individual at any month of the year in central Chile is expected to be a Southern Giant Hummingbird; the Northern Giant Hummingbird has never been reported south of Arica y Parinacota in the extreme north of Chile. We conservatively excluded recordings from Bolivia from our breeding range dataset due to uncertainty about the distributions of the giant hummingbird species in the central Andes (Williamson et al. 2025a). Our final breeding range dataset included vocalizations from 24 Northern and 54 Southern Giant Hummingbirds.

Quantitative analysis of vocal traits

We used t-tests or non-parametric Wilcoxon signed-rank tests to quantitatively compare differences in mean vocal characteristics (minimum frequency, maximum frequency, peak frequency, duration, and bandwidth) between the two species. Next, to test the reliability of species recognition by vocalization, we standardized mean vocal traits and ran a Principal Components Analysis (PCA) using means of minimum vocalization frequency, maximum vocalization frequency, and bandwidth. To simplify the PCA and its interpretation, we excluded peak frequency, due to its high similarity with maximum frequency, and bandwidth, given that this measure was calculated from minimum and maximum song frequency scores. With these same three standardized traits, we then conducted a linear discriminant analysis (LDA), using a randomized 80% of breeding range data for model training and 20% for model testing, without replacement. All statistical tests and models were carried out using R (R Core Team 2025).

Our analyses confirmed strong differences between the vocalizations of Northern and Southern Giant Hummingbirds in all measured characteristics and the trained LDA model was able to distinguish the two species with 100% accuracy using breeding range data. We therefore used our trained LDA model, checked against qualitative vocal type assessment, to classify each non-breeding range recording as either a Northern or Southern Giant Hummingbird. We then combined breeding and non-breeding range data from across the annual ranges of both species. Our combined dataset included a total of 153 individuals (39 Northern and 113 Southern Giant Hummingbirds). We repeated comparative analyses of trait differences, as well as PCA and LDA analyses.

We additionally conducted LDA analyses using published morphological data (lengths of bill, wing, and tail) from Williamson et al. (2025) to compare the utility of vocalizations versus external phenotypic characters in distinguishing the two species. Our morphology dataset included 166 adult females and 161 adult males from across the annual range.

RESULTS

Comparative analysis of vocal traits

The principal vocalizations of the Northern and Southern Giant Hummingbird differed significantly in all measured traits (Fig. 2, Table 1, Appendix 1: Fig. S7). Northern Giant Hummingbirds had significantly higher minimum (p<2.2x10^-16, Wilcoxon rank sum test), maximum (p=2.2x10^-16, t-test), and peak (p=2.2x10^-16, t-test) frequencies than Southern Giant Hummingbirds. The principal vocalizations of the Northern Giant Hummingbird were also 2x longer (p=2.2x10^-16, Wilcoxon rank sum test). The vocalizations of the Southern Giant Hummingbird spanned a 2.6x greater bandwidth than those of the Northern Giant Hummingbird (p=2.2x10^-16, Wilcoxon rank sum test). These differences were consistent when comparing breeding range and full annual data (Fig. 2, Appendix 1: Fig. S7). Quantitative song differences between species were consistent with visual inspections of the spectrograms (Fig. 1, Appendix 1: Figs. S1–S2) and were easily audible when listening to recordings.

Vocal divergence between species

The vocalizations of each the Northern and Southern Giant Hummingbird reliably identified the two cryptic species with consistently high confidence (Fig. 3). In PCA analyses of three vocal traits (minimum frequency, maximum frequency, and duration), PC1 corresponded strongly to species identity, explaining 83.4% of the variation in breeding range data comparisons and 81.1% of the variation in full annual data comparisons (Fig. 3). PC2 captured variation in each minimum and maximum frequencies relative to song duration, explaining 9.2% of the variation in breeding range data comparisons and 9.6% of the variation in full annual range comparisons.

PCA results were consistent with the results of a LDA using minimum frequency, maximum frequency, and duration. The trained LDA model predicted species identity with 100% classification accuracy (Fig. 3). Two individuals were misclassified (2.7% error rate) when the model was applied to full annual range data. The two individuals in question were represented by Southern Giant Hummingbird individuals recorded in Cochabamba, Bolivia, confirmed in the field with visual evidence by J.L.W. ML631524705 was recorded on 12 February 2025 and ML631491349 on 16 February 2025; both were classified by the model as Northern Giant Hummingbirds. In all other cases, species identifications assigned by the trained LDA model matched those determined by assessment of vocal types, season, locality, plumage, and molt pattern and timing.

Vocalizations, more than morphology, reliably distinguish species

In contrast to vocalizations, the morphological trait distributions of Northern and Southern Giant Hummingbirds overlapped substantially (Fig. 3C-D). LDA models trained on previously published data of three morphological characteristics—and conducted separately for each adult females and males—could only distinguish Northern and Southern Giant Hummingbirds in 70.59% and 69.7% of instances, respectively (Williamson et al. 2024; Fig. 3 F-G). These results support previous findings that adult Northern and Southern Giant Hummingbirds may not always be reliably identified using morphological characteristics (Williamson et al. 2024, 2025a) but highlight the relative ease with which principal vocalizations can distinguish the two.

Clarified range limits and geographic co-occurrence in Bolivia

We recorded 38 individual giant hummingbirds of both species in Bolivia at 11 different localities, growing the existing collection of giant hummingbird vocalizations archived in repositories by ~75%. Our field recordings provided definitive evidence that the Northern Giant Hummingbird occurs in west-central Bolivia; specifically, in the vicinity of Copacabana on the shores of Lake Titicaca (~3900 m), where we encountered five Northern Giant Hummingbirds (Fig. 1a; See ML accession numbers 631522147, 631522523, 632814915, 631522608). Four were adults and one was a juvenile; of the four adults, three were in fresh plumage and one bird was molting its outer primaries. Despite documenting 62 Southern Giant Hummingbirds in suitable habitat in La Paz Department, including Sorata, Pongo, the city of La Paz, Cañon de Palca, Bosquecillo Auquisamaña, and Achocalla, as well as throughout Cochabamba Department, no Northern Giant Hummingbirds were detected in Bolivia outside the vicinity of Copacabana.

We confirmed the occurrence of the Southern Giant Hummingbird in all sampled regions of Bolivia (Fig. 1A). Notably, during our February 2025 fieldwork, we observed the Southern Giant Hummingbird around Copacabana, Bolivia, in the same sites where the Northern Giant Hummingbird was observed and recorded (at ~3900-4000 m). All observed Southern Giant Hummingbirds were molting and were identifiable by their earlier molt stage (i.e., molting inner primaries) and plumage. In several instances, Northern and Southern Giant Hummingbirds were seen and heard vocalizing simultaneously, which is reflected in recordings.

DISCUSSION

The Northern and Southern Giant Hummingbird differ markedly in their migration, genetics, and physiology, yet vary only subtly in external characteristics (Williamson et al. 2024, 2025a). Until now, their vocal traits had not been formally compared. Our analyses revealed that the two species differ significantly in all measured vocal traits. Importantly, principal vocalizations appear to be the single most definitive diagnostic trait for identifying the two species in the field. These findings are directly useful for ornithologists, birders, and naturalists observing and studying giant hummingbird ecology and evolution, as well as for management professionals conducting population-level censuses and conservation assessment. Because the vocal differences described here can be detected by ear (i.e., in the field or from recordings) and by eye (i.e., through visual assessment of spectrograms; Fig. 1, Appendix 1: S1-S2), real-time or later inspection of spectrograms provides an accessible way for hearing-impaired birders and researchers to identify the giant hummingbirds.

That vocalizations can so easily distinguish the giant hummingbirds contrasts with the difficulty of identification using plumage or morphology (Fig. 3). Due to within-species trait variation and between-species overlap in external phenotypic characteristics, non-vocal individuals can be challenging to identify in the hand or in the field, even with the aid of trained statistical models, and even when considering data for each sex separately (Williamson et al. 2024, 2025a). Whereas LDA models of morphological traits correctly identified ~70% of birds, LDA models trained on three vocal characteristics classified individuals correctly in 97.3–100% of cases. Only two individuals of 153 were misidentified by the trained model when applied to full annual data (Appendix 1: Fig. S6), but the context, length, and quality of the recordings indicates that they may have been aberrant. One individual, represented by ML631524705, had only one scorable vocalization, while the second, ML631491349, had only two scorable vocalizations. During scoring, R.J.Z. noted that the recording for ML631524705 appeared modulated and atypical; consistent with that observation, J.L.W.’s field notes indicate that they saw and heard this focal individual in an aerial chase with a second Southern Giant Hummingbird shortly after the single clear recorded vocalization. These results highlight how field identification of the giant hummingbirds may require additional context clues, even with diagnostic vocalizations. Additionally, whereas morphological measurements and plumage may only be useful for distinguishing adult individuals (juvenile plumage of both species may be indistinguishable), vocal characteristics appear applicable to both juveniles and adults.

The ranges of the giant hummingbirds have remained difficult to characterize using occurrence records and photos alone, partly due to the previously undescribed migratory routes of some populations of the Southern Giant Hummingbird (Williamson et al. 2025a). Our study provides new data on the ranges of both species in Bolivia and indicates a previously unknown zone of overlap. While preliminary given the limited scope of our field recording efforts, we nonetheless provide definitive evidence that the Northern Giant Hummingbird occurs in west-central Bolivia. The southernmost extent of the Northern Giant Hummingbird range remains uncertain; however, we have not yet detected the Northern Giant Hummingbird south of Copacabana, Bolivia (Lake Titicaca region), along the eastern slope of the Andes. Interestingly, we documented that the Southern Giant Hummingbird occurs at the same sites as the Northern Giant Hummingbird around Copacabana, Bolivia. In Copacabana, where both species overlap—sometimes on the same hillside and within mere meters of each other—they are distinguishable using visual and audio evidence. The Southern Giant Hummingbird most likely breeds at high elevations in La Paz Department, based on its occurrence there in February, a finding which awaits further substantiation. The northernmost extent of the Southern Giant Hummingbird breeding range remains unknown. Observers in the Lake Titicaca region of Peru and Bolivia can help expand knowledge by recording songs of individuals observed during all months of the year. Follow-up research will be necessary to refine breeding and non-breeding range and elevational limits, particularly during periods of the year when key floral resources (e.g., genera Eucalyptus, Mutisia, Agave, Puya, and Cactaceae species) are in bloom.

Diagnostic vocalizations additionally present an opportunity to illuminate aspects of giant hummingbird ecology and evolution. Collecting new field recordings from regions where few exist (e.g., northwest Argentina, northern Chile), as well as from zones and periods of sympatry (e.g., ML642932325, https://macaulaylibrary.org/asset/642932325 from 30 September 2025 in Lima, Peru; not analyzed in this study), will refine our understanding of seasonal ranges, elevations, and migratory phenology of the giant hummingbirds. For example, we identified several possible cases of apparent flexibility in the migration departure dates of the Southern Giant Hummingbird, relative to known tracked individuals from Chile: ML511071761 (recorded on 25 October 2022 in Cusco, Peru) may document a late-departing migrant Southern Giant Hummingbird, while XC17922 (recorded 9 March 2006 from Cusco, Peru) may document an early-arriving migrant Southern Giant Hummingbird; alternatively, these records may suggest that individuals originated outside of Chile, which remains to be confirmed. Similarly, the recording ML515620121 (from 13 December 2022 in Cusco, Peru) had a clear Type 2 vocal shape corresponding to the Southern Giant Hummingbird; however, to the ear, this individual sounded quite high-pitched and it is therefore not clear whether it was an early migrant, late migrant, or “overwintering” Southern Giant Hummingbird, or, an unusual-sounding Northern Giant Hummingbird.

Further research is needed to understand the function of the principal vocalizations analyzed here, which have been variably described in the literature as “songs” and “calls.” In the field, the authors have observed that both species perform these vocalizations—on both the breeding and nonbreeding grounds—when defending territories and resources and when advertising to potential mates. During fieldwork in 2023, the authors recorded both male and female Southern Giant Hummingbirds, whose identifications were confirmed with genetic sequencing, giving these vocalizations on the non-breeding grounds in Peru. These observations suggest that the vocalizations may function as songs, particularly in light of growing evidence that diverse bird species sing during both the breeding and nonbreeding seasons (e.g., Odom et al. 2016, Kipper et al. 2017, Proudfoot and Norton 2025). While song is typically considered a male trait, it is now widely recognized that females of many species sing (e.g., Odom et al. 2016, 2025, Riebel et al. 2019). Song is widespread in the sister clade of Patagona, occurring in at least 20 species of bee hummingbirds, as well as in mountain gems and emeralds (Ficken et al. 2000, Clark et al. 2018). It also occurs in more distantly related hummingbird lineages, such as the hermits, leading to the inference that the common ancestor of extant hummingbirds would have also had song (Clark et al. 2018). It therefore follows that the most recent common ancestor of Patagona and its sister clade, occurring ~14 Mya (McGuire et al. 2014), would have possessed song. Additional research is needed to clarify the functional role of Patagona vocal signals to ascertain whether they represent a derived form of song or whether song was lost in the Patagona lineage, as has occurred repeatedly in various other branches of the hummingbird phylogeny.

In this study, we showed that strikingly divergent vocalizations identify the two giant hummingbird species with high confidence and provide a robust method for elucidating range limits and areas of seasonal co-occurrence. Our work additionally highlights the value of open media repositories, such as Macaulay Library and Xeno-canto, in ecological and evolutionary research. Through the contributions of many recordists, we were able to analyze vocal data spanning 49 years, >36° of latitude, and >4300 meters in elevation. The oldest recordings in our dataset were collected by ornithologist Ted Parker during a May 1976 trip to Ancash, Peru (Stap 1991). On this trip—which occurred during the giant hummingbird non-breeding season—Parker incidentally recorded both a Northern Giant Hummingbird (ML11057) and at least one Southern Giant Hummingbird (ML11062 and ML11063). That evidence of species-level divergence had remained hidden in the world’s largest repository of bird sounds dating back to 1976 evokes curiosity about potential cryptic species that have yet to be uncovered.

LITERATURE CITED

Araya-Salas, M., and T. Wright. 2013. Open-ended song learning in a hummingbird. Biology Letters 9(5):20130625. https://doi.org/10.1098/rsbl.2013.0625

Atwood, J. L., V. L. Fitz, and J. E. Bamesberger. 1991. Temporal patterns of singing activity at leks of the White-bellied Emerald. The Wilson Bulletin 103:373-386.

Baptista, L. F., and K. Schuchmann. 1990. Song learning in the Anna Hummingbird (Calypte anna). Ethology 84(1):15-26. https://doi.org/10.1111/j.1439-0310.1990.tb00781.x

Braune, P., S. Schmidt, and E. Zimmermann. 2008. Acoustic divergence in the communication of cryptic species of nocturnal primates (Microcebus ssp.). BMC Biology 6:19. https://doi.org/10.1186/1741-7007-6-19

Catchpole, C. K., and P. J. B. Slater. 1995. Bird song: biological themes and variations. Cambridge University Press, Cambridge, UK. https://doi.org/10.1017/CBO9780511754791

Clark, C.J., J.A. McGuire, E. Bonaccorso, J.S. Berv, and R.O. Prum. 2018. Complex coevolution of wing, tail, and vocal sounds of courting male bee hummingbirds. Evolution 72(3):630-646. https://doi.org/10.1111/evo.13432

Cooper, J. C., and A. M. Cuervo. 2017. Vocal variation and species limits in the Sclerurus mexicanus complex. The Wilson Journal of Ornithology 129(1):13-24. https://doi.org/10.1676/1559-4491-129.1.13

Duque, F. G., and L. L. Carruth. 2022. Vocal communication in hummingbirds. Brain, Behavior and Evolution 97(3-4):241-252. https://doi.org/10.1159/000522148

Ferreira, A. R. J., T. V. Smulders, K. Sameshima, C. V. Mello, and E. D. Jarvis. 2006. Vocalizations and associated behaviors of the Sombre Hummingbird (Aphantochroa cirrhochloris) and the Rufous-breasted Hermit (Glaucis hirsutus). The Auk 123(4):1129-1148. https://doi.org/10.1093/auk/123.4.1129

Ficken, M. S., K. M. Rusch, S. J. Taylor, and D. R. Powers. 2000. Blue-throated Hummingbird song: a pinnacle of nonoscine vocalizations. The Auk 117(1):120-128. https://doi.org/10.1093/auk/117.1.120

Gaunt, S. L. L., L. F. Baptista, J. E. Sánchez, and D. Hernandez. 1994. Song learning as evidenced from song sharing in two Hummingbird species (Colibri coruscans and C. thalassinus). The Auk 111(1):87-103. https://doi.org/10.2307/4088508

González, C., and J. F. Ornelas. 2005. Song structure and microgeographic song variation in Wedge-Tailed Sabrewings (Campylopterus Curvipennis) in Veracruz, Mexico. The Auk 122(2):593-607. https://doi.org/10.1093/auk/122.2.593

González, C., and J. F. Ornelas. 2009. Song variation and persistence of song neighborhoods in a Lekking Hummingbird. The Condor 111(4):633-640. https://doi.org/10.1525/cond.2009.090029

González, C., J. F. Ornelas, and C. Gutiérrez-Rodríguez. 2011. Selection and geographic isolation influence hummingbird speciation: genetic, acoustic and morphological divergence in the wedge-tailed sabrewing (Campylopterus curvipennis). BMC Evolutionary Biology 11:38. https://doi.org/10.1186/1471-2148-11-38

Gwee, C. Y., J. A. Eaton, E. Y. X. Ng, and F. E. Rheindt. 2019. Species delimitation within the Glaucidium brodiei owlet complex using bioacoustic tools. Avian Research 10:36. https://doi.org/10.1186/s40657-019-0175-4

Hackett, S. J., R. T. Kimball, S. Reddy, R. C. K. Bowie, E. L. Braun, M. J. Braun, J. L. Chojnowski, W. Andrew Cox, K.-L. Han, J. Harshman, et al. 2008. A phylogenomic study of birds reveals their evolutionary history. Science 320(5884):1763-1767. https://doi.org/10.1126/science.1157704

Hao, Y., Y. Xiong, Y. Cheng, G. Song, C. Jia, Y. Qu, and F. Lei. 2019. Comparative transcriptomics of 3 high-altitude passerine birds and their low-altitude relatives. Proceedings of the National Academy of Sciences 116(24):11851-11856. https://doi.org/10.1073/pnas.1819657116

Herzog, S. K., R. S. Terrill, A. E. Jahn, J. V. Remsen, O. Maillard, V. H. García-Soliz, R. MacLeod, Ross, A. Maccormick, J. Q. Vidoz, C. C. Tofte, et al. 2016. Birds of Bolivia field guide. Asociación Armonía, Santa Cruz de la Sierra, Bolivia.

Irwin, D. E. 2000. Song variation in an avian ring species. Evolution 54(3):998-1010. https://doi.org/10.1111/j.0014-3820.2000.tb00099.x

Ivy, C. M., and J. L. Williamson. 2024. On the physiology of high-altitude flight and altitudinal migration in birds. Integrative and Comparative Biology 64(6):1766-1779. https://doi.org/10.1093/icb/icae062

Jaramillo, A. 2003. Birds of Chile. Princeton University Press, Princeton, New Jersey, USA.

Jarvis, E. D., S. Ribeiro, M. L. Da Silva, D. Ventura, J. Vielliard, and C. V. Mello. 2000. Behaviourally driven gene expression reveals song nuclei in hummingbird brain. Nature 406:628-632. https://doi.org/10.1038/35020570

Johnson, K. E., and C. J. Clark. 2020. Ontogeny of vocal learning in a hummingbird. Animal Behaviour 167:139-150. https://doi.org/10.1016/j.anbehav.2020.07.010

Johnson, K. E., and C. J. Clark. 2024. Open-ended vocal learning in Costa’s hummingbird. Animal Behaviour 218:207-216. https://doi.org/10.1016/j.anbehav.2024.10.012

K. Lisa Yang Center for Conservation Bioacoustics at the Cornell Lab of Ornithology. 2024. Raven Pro: Interactive Sound Analysis Software (Version 1.6.5) - Computer software. Cornell Lab of Ornithology, Ithaca, NY. https://www.ravensoundsoftware.com/

Kipper, S., P. Sellar and C. R. Barlow. 2017. A comparison of the diurnal song of the Common Nightingale (Luscinia megarhynchos) between the non-breeding season in The Gambia, West Africa and the breeding season in Europe. Journal of Ornithology 158:223-231. https://doi.org/10.1007/s10336-016-1364-0

Koepcke, M. 1983. The birds of the department of Lima, Peru. Harrowood Books, Newtown Square, Pennsylvania, USA.

Lima, R. D., and R. V. Vaz. 2024. Divergence in vocalizations indicates cryptic speciation in Camptostoma tyrannulets. Ornithology 142(2):ukae058. https://doi.org/10.1093/ornithology/ukae058

McGuire, J. A., C. C. Witt, J. V. Remsen, A. Corl, D. L. Rabosky, D. L. Altshuler, and R. Dudley. 2014. Molecular phylogenetics and the diversification of hummingbirds. Current Biology 24(8):1-7. https://doi.org/10.1016/j.cub.2014.03.016

Movin, N., T. Gamova, S. G. Surmach, J. C. Slaght, A. A. Kisleiko, J. A. Eaton, and F. E. Rheindt. 2022. Using bioacoustic tools to clarify species delimitation within the Blakiston’s Fish Owl (Bubo blakistoni) complex. Avian Research 13:100021. https://doi.org/10.1016/j.avrs.2022.100021

Ng, N.S.R., and F.E. Rheindt. 2016. Species delimitation in the White-faced Cuckoo-dove (Turacoena manadensis) based on bioacoustic data. Avian Research 7:2. https://doi.org/10.1186/s40657-015-0036-8

Odom, K. J., M. Araya-Salas, L. Benedict, K. Lim, J. Dale, W. H. Webb, C. Sheard, J. A. Tobias, G. F. Ball, M. L. Hall, N. E. Langmore, et al. 2025. Global incidence of female birdsong is predicted by territoriality and biparental care in songbirds. Nature Communications 16:6157. https://doi.org/10.1038/s41467-025-60810-5

Odom, K. J., M. Araya‐Salas, J. L. Morano, R. A. Ligon, G. M. Leighton, C. C. Taff, A. H. Dalziell, A. C. Billings, R. R. Germain, M. Pardo, et al. 2021. Comparative bioacoustics: a roadmap for quantifying and comparing animal sounds across diverse taxa. Biological Reviews 96(4):1135-1159. https://doi.org/10.1111/brv.12695

Odom, K. J., K. E. Omland, D. R. McCaffrey, M. K. Monroe, J. L. Christhilf, N. S. Roberts, and D. M. Logue. 2016. Typical males and unconventional females: songs and singing behaviors of a tropical, duetting Oriole in the breeding and non-breeding Season. Frontiers in Ecology and Evolution 4. https://doi.org/10.3389/fevo.2016.00014

Ornelas, J. F., C. Gonzalez, and J. Uribe. 2002. Complex vocalizations and aerial displays of the Amethyst-Throated Hummingbird (Lampornis amethystinus). The Auk 119:1141-1149.

Parmentier, E., R. Scalbert, X. Raick, C. Gache, B. Frédérich, F. Bertucci, and D. Lecchini. 2022. First use of acoustic calls to distinguish cryptic members of a fish species complex. Zoological Journal of the Linnean Society 195(3):964-975. https://doi.org/10.1093/zoolinnean/zlab056

Pearman, M., and J. I. Areta. 2020. Birds of Argentina. Princeton University Press, Princeton, New Jersey, USA.

Proudfoot, C., and W. H. J. Norton. 2025. Non-breeding European robins adjust their song in noisy environments. Behavioral Ecology 36(4):araf070. https://doi.org/10.1093/beheco/araf070

R Core Team. 2025. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://doi.org/10.32614/R.manuals

Ridgely, R. S., and P. J. Greenfield. 2001. Birds of Ecuador, Volume 2: field guide. Cornell University Press, Hong Kong.

Riebel, K., K. J. Odom, N. E. Langmore, and M. L. Hall. 2019. New insights from female bird song: towards an integrated approach to studying male and female communication roles. Biology Letters 15(4):20190059. https://doi.org/10.1098/rsbl.2019.0059

Schulenberg, T. S., D. F. Stotz, D. F. Lane, J. P. O’Neill, and T. A. P. III. 2010. Birds of Peru. Princeton University Press, Princeton, New Jersey, USA.

Sin, Y. C. K., J. A. Eaton, R. O. Hutchinson, and F. E. Rheindt. 2022. Re-assessing species limits in a morphologically cryptic Australasian kingfisher lineage (Coraciiformes: Halcyonidae) using bioacoustic data. Biological Journal of the Linnean Society 136(4):489-505. https://doi.org/10.1093/biolinnean/blac047

Stap, D. 1991. A parrot without a name. First University of Texas Press, Austin, Texas, USA.

Stiffler, L. L., K. M. Schroeder, J. T. Anderson, S. B. McRae, and T. E. Katzner. 2018. Quantitative acoustic differentiation of cryptic species illustrated with King and Clapper rails. Ecology and Evolution 8(24):12821-12831. https://doi.org/10.1002/ece3.4711

Toews, D. P. L., and D. E. Irwin. 2008. Cryptic speciation in a Holarctic passerine revealed by genetic and bioacoustic analyses. Molecular Ecology 17(11):2691-2705. https://doi.org/10.1111/j.1365-294X.2008.03769.x

Velásquez-Noriega, P., M. J. S. Mallea, F. Medrano, I. Heynen, and P. Pyle. 2023. Giant Hummingbird (Patagona gigas), version 3.0. Pages in F. Medrano, and B. K. Keeney, editors. Birds of the world. Cornell Lab of Ornithology, Ithaca, New York, USA. https://doi.org/10.2173/bow.giahum1.03

Wilczynski, W., B. E. McClelland, and A. S. Rand. 1993. Acoustic, auditory, and morphological divergence in three species of neotropical frog. Journal of Comparative Physiology A 172:425-438. https://doi.org/10.1007/BF00213524

Williamson, J., E. F. Gyllenhaal, S. M. Bauernfeind, E. Bautista, M. J. Baumann, C. R. Gadek, P. P. Marra, N. Ricote, T. Valqui, F. Bozinovic, N. D. Singh, and C. C. Witt. 2024. Extreme elevational migration spurred cryptic speciation in giant hummingbirds. Proceedings of the National Academy of Sciences 121(21):1-11. https://doi.org/10.1073/pnas.2313599121

Williamson, J. L., C. R. Gadek, B. W. Robinson, E. Bautista, S. M. Bauernfeind, M. J. Baumann, E. F Gyllenhaal, P. P Marra, N. Ricote, N. D Singh, et al. 2025a. Taxonomy, nomenclature, and identification of the giant hummingbirds (Patagona spp.) (Aves: Trochilidae). Zoological Journal of the Linnean Society 204(2):1-16. https://doi.org/10.1093/zoolinnean/zlaf036

Williamson, J. L., E. B. Linck, E. Bautista, A. Smiley, J.A. McGuire, R. Dudley, and C.C. Witt. 2023. Hummingbird blood traits track oxygen availability across space and time. Ecology Letters 26(7):1223-1236. https://doi.org/10.1111/ele.14235

Williamson, J. L., T. Valqui, and C. Witt. 2025b. Proposal 1059: American Recognize Patagona gigas and Patagona peruviana as separate species, with revised nomenclature (PASSED). Proposal to the South American Classification Committee (SACC):1-5.

Williamson, J. L., and C. C. Witt. 2021a. A lightweight backpack harness for tracking hummingbirds. Journal of Avian Biology 52(9). https://doi.org/10.1111/jav.02802

Williamson, J. L., and C. C. Witt. 2021b. Elevational niche-shift migration: Why the degree of elevational change matters for the ecology, evolution, and physiology of migratory birds. Ornithology 138(2):1-26. https://doi.org/10.1093/ornithology/ukaa087