Fig 1.
The major elements of IDcall framework.
The recordings of sounds are delineated into identity clades based on the identity calls that characterize them.
Fig 2.
The major elements of IDcallPP.
The output of IDcall is used to produce stacked barplots of the posterior probabilities that each repertoire is a member of each identity clade.
Fig 3.
Identity clades, identity song types, and posterior probabilities of repertoire assignment for crickets.
Output from IDcall (top; taken from [14]) depicts similarity among male Teleogryllus cricket calling songs recorded from individuals derived from 16 field sites in Australia (data from [17]). ‘Oce’ indicates song repertoires recorded from crickets belonging to the oceanicus species (in teal). ‘Com’ denotes song repertoires recorded from crickets belonging to the commodus species (in brown). The letters in parentheses denote field sites (see [17] for site abbreviations). For each song, we created an interval vector comprised of four traits: chirp pulse length, chirp interpulse interval, chirp-trill interval, and trill pulse length (see [17] for details on how song traits were measured). Each repertoire (i.e. branch in the dendrogram) contains all the songs recorded from first-generation crickets that were derived from wild-caught individuals from each field site. (A) The average linkage hierarchical clustering dendrogram thus depicts similarity among song interval vectors of male crickets from the 16 sites. (B) The heatmap shows identity song type usage (rows) for each field site (columns) in shades of grey, with usage calculated based on probabilistic assignment of songs to types. Identity song type codes are on the left of the heat map and centroid song interval vector plots are on the right (with the spaces between the dots representing chirp pulse length, chirp interpulse interval, chirp-trill interval, and trill pulse length, and the scale bar in seconds). (C) The output from IDcallPP shows the posterior assignment probabilities of each repertoire belonging to each identity clade (i.e. species) as a stacked bar plot. See [14] and [17] for additional details.
Fig 4.
Identity clades, identity song types, and posterior probabilities of repertoire assignment for wrens.
Output from IDcall (top; taken from [14]) depicts similarity among male songs (data from [18]) from two subspecies of grey-breasted wood-wren: Henicorhina leucophrys hilaris (salmon) and Henicorhina leucophrys leucophrys (navy). Genotyping abbreviations are: Hil, parental H. l. hilaris; Leu, parental H. l. leucophrys; F1, first-generation hybrid; BC-hil, backcross between Hil and F1; and BC-leu, backcross between Leu and F1. For each song, we created an interval vector comprised of three traits: averaged note peak frequency, minimum song frequency, and maximum song frequency (see [18] for details on how song traits were measured). Each repertoire (i.e. branch in the dendrogram) contains all the songs recorded from a single individual. (A) The average linkage hierarchical clustering dendrogram thus depicts similarity among song interval vectors of 41 male wrens. (B) The heatmap shows identity song type usage (rows) for each wren (columns) in shades of grey, with usage calculated based on probabilistic assignment of songs to types. Identity song type codes are on the left of the heat map and centroid song interval vector plots are on the right (with the spaces between the dots representing averaged note peak frequency, minimum song frequency, and maximum song frequency, and the scale bar in Hertz). (C) The output from IDcallPP shows the posterior assignment probabilities of each repertoire belonging to each identity clade (i.e. subpecies) as a stacked bar plot. See [14] and [18] for additional details.
Fig 5.
Identity clades, identity coda types, and posterior probabilities of repertoire assignment for Atlantic/Mediterranean sperm whales.
Output from IDcall (top; taken from [14]) depicts similarity among coda repertoires of sperm whale groups (data from [14]) recorded in the Atlantic Ocean and Mediterranean Sea. Colored identity clades correspond to three sperm whale clans: Mediterranean (cyan), EC2 (gold), and EC1 (purple). Location abbreviations are: AtPAN = Atlantic coast of Panama, BAL = Balearic Islands, CAR = eastern Caribbean islands, and GOM = Gulf of Mexico. Each coda was represented as a vector of inter-click intervals. Each repertoire (i.e. branch in the dendrogram) contains all the codas recorded from a known social unit of whales in a year or, if the identity of the recorded whales was unknown, all the codas recorded on a single day. (A) The average linkage hierarchical clustering dendrogram thus depicts similarity among 82 sperm whale coda repertoires. (B) The heatmap shows identity coda type usage (rows) for each repertoire (columns) in shades of grey, with usage calculated based on probabilistic assignment of codas to types. Identity coda type codes are on the left of the heat map and centroid coda interval vector plots are on the right (with the spaces between the dots representing the inter-click intervals and the scale bar in seconds). (C) The output from IDcallPP shows the posterior assignment probabilities of each repertoire belonging to each identity clade (i.e. vocal clan) as a stacked bar plot. See [14] for additional details.
Fig 6.
Identity clades, identity coda types, and posterior probabilities of repertoire assignment for Pacific sperm whales.
Output from IDcall (top; taken from [14]) depicts similarity among coda repertoires of sperm whale groups (data from [14]) recorded in the Pacific Ocean. Colored identity clades correspond to a putative new sperm whale clan (orange) and four known clans: Short (red), Four-Plus (pink), Plus-One (blue), and Regular (green). Location abbreviations are: BAK = Baker Island, CHI = Chile, EAS = Easter Island, ECU = Ecuador, GAL = Galápagos Islands, JAR = Jarvis Island, NEW = New Zealand, PacPAN = Pacific coast of Panama, PER = Peru, and TON = Tonga. Each coda was represented as a vector of inter-click intervals. Each repertoire (i.e. branch in the dendrogram) contains all the codas recorded from a single photo-identified group of sperm whales in a year. (A) The average linkage hierarchical clustering dendrogram thus depicts similarity among 106 sperm whale coda repertoires. (B) The heatmap shows identity coda type usage (rows) for each repertoire (columns) in shades of grey, with usage calculated based on probabilistic assignment of codas to types. Identity coda type codes are on the left of the heat map and centroid coda interval vector plots are on the right (with the spaces between the dots representing the inter-click intervals and the scale bar in seconds). (C) The output from IDcallPP shows the posterior assignment probabilities of each repertoire belonging to each identity clade (i.e. vocal clan) as a stacked bar plot. See [14] for additional details.
Fig 7.
Stacked barplots showing posterior probability distributions of repertoires to identity clades using all calls (above) and just identity calls (below) for the four example data sets.