This dataset contains data for sequence analysis described in the paper: “Lemazina et al., (2021) The multifaceted vocal duets of White-browed sparrow weavers are based on complex duetting rules. Journal of Avian Biology, doi: 10.1111/jav.02703”. 

In our study, we recorded duet songs produced by six wild P. mahali pairs in South Africa. Our study further reveals the existence of answering rules in P. mahali, adhered to by both sexes. Even though answering rules can be shared between pairs of different colonies, pair-specificity in ordering of syllable types within the most commonly used duet trains exists. Most importantly, we found that a single answering rule was never used twice in one P. mahali duet song, but answering rules switched constantly, which contrasts the song structure in this species with the song structure of most other duetting bird species studied so far.

For sequence analysis, we used KOE (https://koe.io.ac.nz). Koe discoveres commonly-occurring transitions between syllables in a database using the cSPADE (constrained Sequential Pattern Discovery using Equivalence classes) algorithm (Zaki 2001). The credibility of sequence rules is asserted via the following factors: Support (proportion of songs in the database that contain a certain sequence at least once), Confidence (strength of association; the proportion of songs containing the syllable A that also contain the sequence A⇒B), and Lift (measure of the strength of the association relative to chance; ratio of the proportion of songs in which the transition A⇒B occurred versus the proportion of songs expected to contain the transition A⇒B by chance). Chain length referees to the number of items in the current squence, while Count refers to the occurance number of each particular sequence.

Our study further reveals the existence of answering rules in P. mahali, adhered to by both sexes. Even though answering rules can be shared between pairs of different colonies, pair-specificity in ordering of syllable types within the most commonly used duet trains exists. Most importantly, we found that a single answering rule was never used twice in one P. mahali duet song, but answering rules switched constantly, which contrasts the song structure in this species with the song structure of most other duetting bird species studied so far.

To investigate duetting rules in P. mahali, we prepared sound files for a sequence analysis by mixing synchronous four-hour recordings from both partners of a pair into a single render using the software Audacity (version 2.3.2, Audacity Team, www.audacityteam.org). The mixed files were uploaded to a free available web-based software (KOE, version: 5.4.1, https://koe.io.ac.nz/; Fukazawa et al. 2020) and manually segmented into separate duet bouts.

To determine answering rules in P. mahali duet songs, we filtered the database of syllable transitions provided by KOE for sequences with a chain length of two, which represented a single transition between a male and a female syllable. We included only transition types with a Lift value above two (probability of transition occurrence is two times higher than expected by chance) in the analysis of answering rules, even when a particular transition type occurred only once in the data set. 

To determine duet trains in P. mahali duet songs, we filtered the database provided by KOE for sequences with a chain length of at least three syllables and a Support value larger than 0.06. At a Support value >0.06, the song syntax of a pair was transforming from an entangled network to distinct directional sequences. 

The readme file contains an explanation of each of the variables in the dataset, its measurement units, and the way it was calculated from the primary data. Information on how the measurements were done can be found in the associated manuscript referenced above.