Modulation of behavioural laterality in wild New Caledonian crows (Corvus moneduloides): Vocalization, age and function

ABSTRACT The New Caledonian crow (Corvus moneduloides) is known for displaying a unique set of tool-related behaviours, with the bird’s bill acting as an individually consistently lateralized effector. However, we still fail to understand how such laterality develops, is modulated or even if its expression is consistent across other behavioural categories. Creating the first ethogram for this species allowed us to examine laterality and vocalisations in a population of wild, free-flying New Caledonian crows using detailed analyses of close-up video footage. We revealed the existence of an overall strong left-sided bias during object manipulation only and which was driven by the adult crows of our focal population, the stabilization of individual preferences occurring during the birds’ juvenile years. Individually, at least one crow showed consistent side biases to the right and left within different behavioural categories. Our findings highlight previously unknown variability in behavioural laterality in this species, thus advocating for further investigation. Specifically, we argue that a better understanding of the New Caledonian crow's biology and ecology is required if one wishes to pursue the promising comparative road that laterality could be connected to the evolution of tool-making.


Introduction
A species' comparatively enhanced sensory and/or motor abilities have often been linked back to asymmetrical hemispheric lateralization (see Rogers, 2021), a trait long seen as uniquely human (Corballis, 2002) and which these faculties reflect (Brown & Magat, 2011b;Hellige, 1993). Contrary to the idea of human uniqueness, however, reports of differential cerebral information processing, namely because of regional structural discrepancies (Bisazza, Rogers, & Vallortigara, 1998), have flourished in the last few decades across all kinds of taxa, both vertebrates (see Rogers & Andrew, 2002;Rogers, Vallortigara, & Andrew, 2013;Vallortigara & Versace, 2017) and invertebrates (see Frasnelli, 2013;Frasnelli, Vallortigara, & Rogers, 2012). Nonetheless, a certain form of "chimpocentrism", as identified by Beck (1982), still prevails, as the subject of laterality has most notably been discussed in relation to language and handedness in object manipulation, i.e., research topics heavily focused on primate observation and experimentation (see Fitch & Braccini, 2013;Marchant, 2018). However, whether one is interested in vertebrate laterality for the purpose of studying the evolution of human laterality, or to understand the evolution of laterality more broadly, it is beneficial to broaden the approach by considering taxa beyond the primates (Fairlie & Uomini, 2022). Here we focus on laterality in a species of corvid bird.
Behavioural laterality is defined as a side bias that can be manifested in any aspect of animal behaviour. Examples include the direction of turning in a school of fish (Miletto Petrazzini, Sovrano, Vallortigara, & & Messina, 2020), the choice of which cheek to kiss first in a French greeting (Chapelain et al., 2015), or the side of lunge rolling of a baleen whale while foraging which results in greater wear on one side of baleen plates (MacNeilage, 2013). We focus here on the sub-category of motor laterality, which refers to general organ-use preferences, whether those are sagittally paired or not (McGrew & Marchant, 1997), with pawedness (e.g., Bisazza, Cantalupo, Robins, Rogers, & Vallortigara, 1996), footedness (e.g., Izawa, Kusayama, & Watanabe, 2005), eyedness (e.g., Ventolini et al., 2005) or even tail (e.g., Quaranta, Siniscalchi, & Vallortigara, 2007) and tongue movements (e.g., Ying, 2019) being discussed in the literature. Of particular interest are organ-use preferences outside of handedness (i.e., in species who do not have hands) because they represent convergent evolution (Osvath, Kabadayi, & Jacobs, 2014) of the traits which are of interest. In this respect, the New Caledonian crow (Corvus moneduloides) makes for a highly interesting model for two reasons. Firstly, it would provide further input into our understanding of avian cognition, which was deemed one of "the best examples of brain asymmetry in a non-human species" (Rogers, 1980). Secondly, this corvid species also possesses both a larger relative brain size (Cnotka, Güntürkün, Rehkämper, Gray, & Hunt, 2008) and a unique tool-related repertoire (Hunt & Gray, 2004a;McGrew, 2013) when compared to other passerines. Patterns of asymmetry are exponentially complexified following an increase in brain size (Herculano-Houzel, 2009;Kaplan & Rogers, 2021;Uomini & Ruck, 2018). Therefore, the New Caledonian crow, with its relatively large brain size compared to other passerines, becomes a model of interest, especially when looking for laterality patterns outside of the more classically explored and lateralized tool repertoire. Extensive descriptions of tool manufacture (Hunt, 1996;Hunt & Gray, 2004b, 2006 and tool use (Hunt, 1996;Hunt & Gray, 2004b;Rutz & St Clair, 2012) have been made, highlighting how the bird's bill can act as an individually consistently lateralized effector (Hunt, 2000a;Rutledge & Hunt, 2004).
Given trends, however, do emerge within every field of research, and that of avian behavioural laterality is no stranger to it. Previous studies only focused on, and thus collected data for a few behaviours per study, all comprised within the same behavioural category, and no comprehensive ethogram exists (e.g., Baciadonna, Zucca, & Samour, 2021;Holzhaider, Hunt, & Gray, 2010a;Hunt & Gray, 2004b;Kenward, Rutz, Weir, & Kacelnik, 2006). Thus, with this study, we aimed to redress the imbalance in knowledge. Concretely, the New Caledonian crow's tool behaviour is being deeply explored whereas the biology and ecology of this species remain relatively unknown (Hunt, Holzhaider, & Gray, 2012;Kenward, Rutz, Weir, Chappell, & Kacelnik, 2004;McGrew, 2013;Rutz & St Clair, 2012) and even qualitative studies aiming to identify the factors driving tool-related behavioural laterality are scarce (see Table 1 for a summary of all data published to date on laterality in New Caledonian crows).
More specifically, we still fail to understand how the New Caledonian crow's laterality develops, is modulated and if its expression is consistent across other behavioural categories. Therefore, in the present study, we aimed to address these three questions by studying a population of wild, free-flying New Caledonian crows. These questions were motivated by the striking parallels recently found with human ancestor Homo erectus (Hunt & Uomini, 2016) which have already been addressed within other prone-tocomparison taxa, such as primates. For instance, tool use in immature chimpanzees (Pan troglodytes) is not as strongly lateralized as it is in adult individuals (Humle & Matsuzawa, 2009). Thus, our first prediction was that New Caledonian crow juveniles, who go through a similarly extended immature period offering opportunities to both observe and practice tool behaviour (Pereira & Fairbanks, 1993;Tomasello, Davis-Dasilva, Camak, & Bard, 1987;Uomini, Fairlie, Gray, & Griesser, 2020), will display an analogous instability in lateralizationat least in the tool domain. Secondly, many primates and birds show individually consistent hand / foot preferences for particular action types (Allen et al., 2018;Prieur, Pika, Barbu, & Blois-Heulin, 2018;Prieur, Lemasson, Barbu, & Blois-Heulin, 2019;Regaiolli et al., 2021a;Rodgers & Cain, 2019), leading to our second prediction that adult crows will maintain, given a certain behavioural category, an individual side bias as well. Thirdly, in order to explore the hypothesis about one possible evolutionary pathway to language that has been proposed specifically for the human lineage, namely that behavioural motor lateralization in the toolmaking activities of early hominins led to increased motor skill laterality, and then brain laterality in these species (Prieur et al., 2019;Ruck & Uomini, 2021), which in turn would have facilitated the emergence of language (Meguerditchian, Cochet, & Vauclair, 2011;Mosquera, Geribàs, Bargalló, Llorente, & Riba, 2012;Uomini & Meyer, 2013), we explored the potential cooccurrence of vocalisation and action in New Caledonian crows. Just as chimpanzees' right-hand lateralization is enhanced during such instances (Hopkins & Cantero, 2003), we predicted that, for a given behaviour, its combination with a vocal event would lead to a more pronounced lateralization than the same behaviour expressed alone.
Although previous studies of laterality in New Caledonian crows point to an overall trend towards a left-sided bias at the populational level and the existence of an often consistent lateralization in identified birds (see Table  1), these data alone are not sufficient. Indeed, they only cover quite limited populations throughout the archipelago. While large datasets on laterality in free-living crows behaving in their natural environment are mostly from unknown individuals, with only very few data from identified individuals in the wild, controlled experiments stem from a handful of well-known now permanently captive crows who were brought from New Caledonia to Oxford, UK between 2000 and 2010 and later moved to their current home in Seewiesen, Germany (Jacobs, von Bayern, & Osvath, 2021). In total, results on only 15 identified free-living crows and 13-26 permanently captive crows have been published (see Table 1).
In other words, these studies, although being challenged by a variety of contexts, situations, and motivations, have all demonstrated sizeable limitations, hereby illustrating the complex expression of laterality in this species. For instance, every captive study solely offered human-made tools to the crows, whereas the more naturalistic studies allowed the free-living birds to make their own. This difference might affect the results due to a potential effect of manipulating a familiar vs. unfamiliar object, e.g., laterality could be expressed either more or less strongly (Baragli, Scopa, Felici, & Reddon, 2021;Blazevich & Gill, 2006) depending on whether the crows use their own manufactured vs. a provided tool. In summary, our study fills three important gaps in the existing data on New Caledonian crow laterality by examining: (i) a new population of wild free-living crows who are partly identified and (ii) performing a full range of spontaneous actions of their natural daily repertoire (iii) with the objects of their choice.

Species ecology
In 2019, a fieldwork trip totalling 61 days was led between October and December in a 50-acre private land consisting of lowland dry forest in La Foa, on the main island of Grande Terre, New Caledonia (north-western point: 21°43'51.3"S 165°52'35.2"E; south-eastern point: 21°43'56.0"S 165°52'52.1"E) on which the New Caledonian crow is endemic (Kacelnik, Chappell, Weir, & Kenward, 2006). This Corvid species typically builds up small mixed-aged family units in which offspring can stay up to two years post-fledging, i.e., until about 30 months of age (Holzhaider et al., 2011;Kenward et al., 2004). However, these family groups of three-to-four individuals are also part of larger groups of up to 50 individuals (as registered in Rutz, Ryder, & Fleischer, 2012a) that only temporarily encounter each other  but make up genetically differentiated subpopulations (Abdelkrim, Hunt, Gray, & Gemmell, 2012;Rutz et al., 2012a). In our study, based on the largest number of individuals observed simultaneously during gatherings in which they made grouping calls, i.e., the emission at once of single and multiple calls by different individuals usually spanning several minutes , we estimate the La Foa group to consist of 12 individuals. Reports by the local landowners independently confirmed that this crow population has remained stable at about a dozen individuals for the last 30 years. Although subjects were untagged, some of them could be individually identified via their physical peculiarities (injuries, damaged plumage, spiderweb stuck to head and/or body after foraging under log piles, body size), as New Caledonian crows present a slight sexual dimorphism in weight, but not in shape . Juveniles were identified by pink mouth or throat coloration (Heinrich & Marzluff, 1992;Hunt, 2016), begging behaviours, and being fed by another bird (Holzhaider et al., 2011). Adults, on the other hand, present an all-black mouth coloration (Heinrich & Marzluff, 1992;Hunt, 2016) and/or were identified as such when seen feeding another bird (Holzhaider et al., 2011). We estimated age by the range of months in which breeding attempts were observed in New Caledonian crows (September to March) in an island-wide landbird survey from 1998 to 2011 (Barré et al., 2013). Thus, we estimated that the juveniles in our videos were between 7 and 23 months old.

Data collection
Video footage was obtained in the field using a set of automated and humanoperated cameras (see below for details). Filming took place ad libitum, each time a crow approached an experimental setup made of three baited logs similar to those used in Holzhaider et al. (2010aHolzhaider et al. ( , 2010bHolzhaider et al. ( , 2011. The apparatus consisted of naturally fallen tree logs from the same forest, attached to wooden legs which held the logs horizontally at a height of 30-50 cm above the surfaceeither the ground or a wood pile. They were drilled with one or more 10-centimetre deep and 1-2-centimetre diameter vertical holes, following previous protocols (Hunt & Gray, 2004a;Hunt et al., 2007 ;Hunt, Corballis, & Gray, 2006;Rutz & St Clair, 2012). A pea-sized amount of prey-like food item (either ground beef or a piece of beef heart) or one actual prey item, here a live longhorn beetle larva (Agrianome fairmairei) obtained locally so as to retain the same prey that the crows forage for themselves at this location, was placed at the bottom of each hole (Bluff, Troscianko, Weir, Kacelnik, & Rutz, 2010b;Rutz et al., 2010;Sugasawa, Klump, St Clair, & Rutz, 2017). The crows could only access this food by manipulating elongated tools, here twigs or leaf-stems from four tree species which are locally abundant in this forest (Albizia saman, Aleurites moluccana, Diospyros fasciculosa, and Syzygium cuminii; Figure 1). The crows used these twigs or leaf-stems either unmodified, or modified (using its bill, a crow snaps the leaves off, bites off a length of the tool, and/or bends the tool (Hunt & Gray, 2004a)). These tools were sourced from the ground near the feeding logs, plucked fresh from a nearby tree, or crows arrived already transporting a tool.
The baited logs were placed at a location crows were known to regularly visit, and which was visible from a secluded single point where two researchers sat in chairs. At times when crows were absent from the feeding area, one of the experimenters would walk around the forest to locate the birds and film them ad libitum while foraging freely. Filming was done using a Panasonic HC-VX989 4 K camcorder, a Sony XDCAM PXW-Z150 camcorder, a Sony α7s II mirrorless digital camera, and a GoPro Hero4 action camera. In addition, video footage was obtained from automatic trail cameras (12 MP WildBlick 3.0c RL, p. 2019 Browning Strike Force Pro X, 2020 Browning Patriot, Bushnell Aggressor HD Red Glow, Bushnell Core Dual Sensor No Glow, Spypoint Force 11 MP D) at nine different locations around the forest where crows repeatedly foraged. Most of these devices also provided a measure of time, date and temperature for each clip taken; when it was not the case, these metadata were manually collected.
From our total library of video clips, we selected those which were filmed close enough for the crows' limb and body movements to be visible. We thus retained 02:13:75 h of video footage for this study, during 01:45:51 h of which wild, free-flying New Caledonian crows individually effectively appear.

Coding for laterality
Identifying behaviours of interest for the present study was a several step process. Indeed, due to the ongoing tendency of avian behavioural laterality research to collect data on only a few behaviours per study, and not to use a comprehensive ethogram (e.g., Baciadonna et al., 2021;Holzhaider et al., 2010a;Hunt & Gray, 2004b;Kenward et al., 2006), no exhaustive New Caledonian crow ethogram has been published as of the start of this study (January 2021). Therefore, we first established an ethogram from a literature review of works existing and in preparation on mammalian, reptilian, avian and Corvid behaviours (Simmons, 1964;Goodwin & Gillmor, 1986;Brown & Veltman, 1987;Ruby & Niblick, 1994;Taylor et al., 2010;Auersperg et al., 2011;Alfredsson, 2012;Miller, Schiestl, Whiten, Schwab, & Bugnyar, 2014;Klump, van der Wal, St Clair, & Rutz, 2015), refined with the help of field researchers' personal observations and videos. More specifically, only behaviours effectively observed by researchers in the wild and potentially being lateralized, i.e., that may display differential limb or side preference during their execution, were considered (Nettle, 2003). Vocalisations, albeit not thoroughly understood as of yet (Bluff, Kacelnik, & Rutz, 2010a), were here identified and named on a descriptive basis (Uomini, unpublished). Our working ethogram consisted finally of 47 behavioural items, 15 of which were vocalisations, which we grouped into 5 behavioural categories (see Supplement A). The grouping was inspired by literature (e.g., Greenberg, 1977;Schleidt, Yakalis, Donnelly, & McGarry, 1984;Stanton, Sullivan, & Fazio, 2015) and, following Miller (1988), was designed to remain uninterpretive in terms of motivation and, thus, easily adaptable to and/or comparable with other species. The categorization process was functional in the sense that it was based on the observed outcome of the actions, while avoiding interpretations about the reason for the action unless these reasons or motivations are clearly known (for example, allopreening in birds is a documented affiliative behaviour (Radford & Du Plessis, 2006)).
Using video analysis software BORIS, version 7.9.24 (Friard & Gamba, 2016), laterality was coded according to the body side starting a behavioural bout, and from the bird's egocentric perspective (see Supplement A for details). A bout is understood as a behavioural sequence of identical actions, i.e., a succession of behavioural items whose lateralization could not be modified by the focal individual without interrupting the sequencetherefore, bouts mainly concern object manipulation and locomotion (Chapelain & Hogervorst, 2009;Martinho, Burns, von Bayern, & Kacelnik, 2014). Consequently, each data point encoded for our analysis represented the first event comprised within a behavioural bout. This unit was chosen over raw events as bouts ensure data point independence (McGrew & Marchant, 1997) and are most commonly used, thus facilitating interstudy comparisons (Harrison & Nystrom, 2008).
All statistical analyses were performed using R, version 4.0.4 (R Core Team, 2021).

Robustness and reliability in data collection
The descriptive robustness of the aforementioned ethogram, i.e., whether it allowed for both a consistent and cohesive identification of behaviours, was evaluated by calculating Cohen's (1960) kappa as part of an inter-rater reliability assessment performed on a randomly chosen set of 17 video clips independently coded by two scientists (Hallgren, 2012). Following the same procedure, using the package irr, version 0.84.1 (Gamer, Lemon, & Puspendra Singh, 2019) supported by the package lpSolve, version 5.6.15 (Berkelaar, 2020), the coder's own reliability was assessed using 10% of the 66 video clips included in the present study that were independently coded a month apart (Hallgren, 2012). Each time, the obtained values were well over the 0.67 threshold of substantial to near perfect agreement between coders (Krippendorff, 1980;Landis & Koch, 1977). Moreover, we verified that the coding session (main vs. reliability coding session) did not significantly influence the quantity of behavioural bouts collected (mean duration of registered state events), via Fisher's permutation test for paired data with a stratification for individual identity as included in the package coin, version 1.4-1 (Hothorn, Hornik, van de Wiel, & Zeileis, 2006. Together, these results testify for a substantial consistency between both coders and coding sessions, thus vouching for the quality of the data exploitation (Landis & Koch, 1977).

Statistical modelling
Events in which laterality was not displayed (i.e., symmetrical behaviours) were excluded from further analysis. Therefore, if such an event occurred as part of a behavioural bout, it ended that bout. Associations between age class, behavioural type and the co-occurrence of a vocal event during the remaining bouts (N = 1127; see Supplement A for a detailed account) were tested at both the populational and individual level using generalized linear mixed-effects models for binomial data (GLMMs) implemented by the glmer function of the lme4 package, version 1.1-26 (Bates, Maechler, Bolker, & Walker, 2015). We ruled out any dependencies the dataset might have included by entering individual identity and video clip as nested random factors. They respectively accounted for repeated measurements and the same origin of some crows, as some individuals did appear across different clips characterized by varying surrounding environments, i.e., arboreal vs. terrestrial, clearing vs. dense foliage, alone vs. in presence of inter-and/ or intraspecific individual(s). P-values for GLMMs were obtained using a linear regression model as defined by the Anova function of the car package, version 3.0-10 (Fox & Weisberg, 2019). Before beginning the analyses, all models had been submitted for diagnosis, with the coder checking that sample size was both large enough and not too imbalanced, as well as verifying the absence of multicollinearities by calculating variance inflation factors using the package MuMIn, version 1.43.17 (Barton, 2020). The latter values were always below the 2.5 threshold of notability (Faraway, 2006). The MuMIn package also provided the r.squaredGLMM function which allowed us to characterize covariate effects, and more specifically the proportion of explained variation by the fixed effects, by calculating a marginal pseudoR² (Nakagawa, Johnson, & Schielzeth, 2017). Additionally, the presence of intragroup differences was established via a two-tailed Z test (α = 0.05) performed by the prop.test function of the stats package (R Core Team, 2021).
The co-occurrence of vocalisation did not lead to any side bias (GLMM: χ² 1 = 0.060, P = 0.807) and no association between (a) age class and behaviour type or  Inter-groups comparisons did not reveal any significant differences. Asterisks indicate the significance level (**: P < 0.01) and n.s. that no significant differences have been found.
(b) behaviour type and the co-occurrence of vocalisation was revealed to significantly influence behavioural lateralization (GLMM: (a): χ² 33 = 22.640, P = 0.912; (b): χ² 9 = 4.658, P = 0.863). The absence of a significant effect of the interaction between the co-occurrence of a vocal event and age groups allowed us to pool the latter together (GLMM: χ² 2 = 1.930, P = 0.381). Further analyses at the populational level revealed that, if a significant left-sided behavioural bias was expressed, it was only in absence of a co-occurring vocalisation (two-tailed Z test: χ² 1 = 5.496, P = 0.019; vs. in presence: χ² 1 = 1.655, P = 0.193).

Discussion
The New Caledonian crow population of our site in La Foa is composed of a dozen individuals. The crows we filmed showed actions that were strongly lateralized when it came to tool behaviour and associated object manipulation, similarly to those of other studied groups of the same species (Table  1). More specifically, this left-sided bias was only found within this behavioural category, which has been described in non-human primates as the most strongly lateralized (see Uomini, 2009). In our case, this bias appeared to be driven by adults alone. Indeed, in contrast to adults, juveniles in our study seemed to fail to express any stable overall side preference. If confirmed, this finding would support our first hypothesis: laterality does develop during the New Caledonian crow's extended childhoodat least in regard to tool behaviour. This effect can be explained by young crows spending the first few years of their life in what can be seen as a "safe haven", with adults even outside their family unit tolerating their proactive presence (Holzhaider, Hunt, & Gray, 2010b;Uomini et al., 2020). Fledglings are then not only able to observe experienced individuals make and use tools, but also pick up discarded twigs and sticks, try using them to obtain food, and, ultimately, resort to begging if they cannot meet their nutritional needs (Uomini et al., 2020). Adult proficiency is, on average, only reached around 10-12 months of age in this species (Holzhaider et al., 2010b;Hunt et al., 2012). This intergenerational interaction is likely crucial for New Caledonian crows to learn tool skills: captive juveniles who never observed tool-use demonstrations, albeit able to both craft and handle simple tools, ended up less skilled than their wild counterparts .
Regarding our second hypothesis, i.e., the existence of a consistent side bias, given a certain behavioural category, in adult crows, we can neither support nor refute it. We did not find a consistent laterality in our adult population for any behavioural categories other than object manipulation, except in the unique case of crow male Spike who also showed a significant right-side bias in his "activity and exploration" behaviours. According to Ströckens, Güntürkün, and Ocklenburg (2013), non-human vertebrates show more variation than humans both across tasks and across individuals in terms of limb preferences. As Table 1 shows, there appears to be an overall left-side bias in New Caledonian crows as a species for object manipulation, which is reflected in our data. Regarding individual patterns, lateralization to the right or left is also consistent with previous findings in New Caledonian crows (Table 1). Likewise, it is consistent with the results on object manipulation with the feet in other birds such as parrots (Godinho, Marinho, & Bezerra, 2020). This result would support the theory that skilled motor actions are key to the expression of laterality, as proposed by Corballis (1998Corballis ( , p. 1148. If the development of a motor skill serves to strengthen pre-existing lateral biases, then less-skilled actions (such as head turning or stepping) would not be expected to reveal laterality. To our knowledge, ours is the first study to consider laterality for a wide range of behaviours in New Caledonian crows. As stated by Baciadonna et al. (2021), it is crucial to distinguish behavioural contexts of laterality, as they found that laterality varied per task in Gyrfalcons (Falco rusticolus). Thus, it would be judicious to repeat this analysis on a wider range of individuals and behaviours, as our current data do not yet allow to establish whether adult crows maintain individual side biases within a given behavioural category.
Our third hypothesis, which aimed to explore a potential evolutionary link between behavioural lateralization in tool-making, brain laterality and the emergence of language, was not supported by our data. The co-occurrence of a vocal event appeared to play no role in the strength of a behaviour's lateralization. A contrario, the previously observed left-sided bias seemed to resurface only in absence of any vocalisation. We can attribute these results either to the preponderant number of adults building up our list of subjects, as juvenile birds of some species vocalize significantly more than adults (Favaro, Ozella, & Pessani, 2014), or to the overall scarcity of vocalisations co-occurring with what appears as a very limited set of lateralized actions (only 14.02% of lateralized bouts, 73.42% of which were Scanning). Either way, it is very much possible that our dataset is, in itself, biased and we would advise to replicate the experiment with both a more balanced division of participants and a matching number of co-occurring vocal events if we wish to come to any conclusive statements about the potential link between actions, vocalisation, and laterality in this species, as it has been suggested in others (Uomini, 2009).
Another key aspect of our results needs to be highlighted: they described an overall bias in a given direction, i.e., biases were found at the populational level, whereas previous studies of laterality in New Caledonian crow behaviours had exclusively testified for individual preferences (Rutledge & Hunt, 2004;. Our results thus directly confirm previous findings of population-level lateralization on pandanus tool counterparts (Hunt, 2000b;Hunt, Corballis, & Gray, 2001, 2006. These studies reported that task complexity affected the direction and strength of laterality in pandanus counterparts . When examining their video footage,  found that crows preferentially use head movements to the right to begin tool-making, thus reaffirming the key role of birds' left cerebral hemisphere in object manipulation (Vallortigara & Rogers, 2005). Our study, however, highlighted the existence of a left-sided bias within our population.
One explanation for this left-side bias could be related to postural control, where the foot supporting posture determines which foot is free for manipulation (Tommasi & Vallortigara, 1999). While New Caledonian crows never manipulate objects with their feet, they could still have postural biases, but our data did not reveal any such biases in their locomotor movements. Another explanation could be related to eye-use preference, as in Australian parrots, where preference has been linked to foot-use laterality (Brown & Magat, 2011a). The study of Martinho et al. (2014) hinted towards this idea, as 8 of the 13 crows had the same side eye preference as their tool-use side. However, New Caledonian crows appear to heavily use their binocular vision when using a tool (Troscianko, Von Bayern, Chappell, Rutz, & Martin, 2012) Alternatively, the left-side bias could be related to right-hemisphere functions. The manufacture of twig tools, as was observed here, operates within a three-dimensional space whereas that of pandanus tools is fulfilled through a series of complex, and of importance here, two-dimensional cuts and rips (Hunt, 1996;Hunt & Gray, 2004a;Rutz & St Clair, 2012). Indeed, the plant material, if not left raw after selection to be used as a "straight stick probe" (Rutz et al., 2012a(Rutz et al., , 2012b, could then be snapped and eventually sculpted into a hook by crows (Hunt & Gray 2004a) or made into a curved shape as was specifically done by the La Foa crows (personal observations). In other words, the difference at stake may not be of task difficulty, but related to spatial cognition and recognition, a functional realm typically supported by the right hemisphere in the avian brain (Karenina & Giljov 2022;Vallortigara & Rogers 2005). The limits of such explanation are however reached when accounting for the indiscriminate categorization of both (potential) tools and food items as "objects" in the present study, thus rendering any power assessment impossible.
Similarly, we are currently unable to evaluate individual contributions to this directional tendency, thus not knowing whether the majority of our La Foa subjects actually presented such bias. Indeed, the number of observed crows in our study has artificially been raised to 81 due to an absence of individual identification: the crows in our study were untagged and most of them lacked any physical peculiarities that made them stand out, as this species tends to be phenotypically homogeneous ). However, two adult birds were both easily recognisable and regularly appeared on camera: Tracy, a female who bared an injured chest, and Spike, her partner, whose right-sided plumage was characterized by an upward bent feather shaft (see Supplementary Video).
Individual analyses for Tracy and Spike departed from overall trends in several ways: the type of behavioural bout expressed by either of these birds did not influence their behavioural lateralization, which did not display any side bias. In other words, the significant left-sided bias observed at a populational level could not be attributed to either one of these crows. Moreover, Tracy and Spike themselves did present different individual tendencies, as a categorical breakdown of their actions revealed. In reality, Spike alone showed significantly lateralized bouts, with a strong left-sided bias emerging during object manipulation. More surprisingly, a weaker albeit still significant right-sided bias characterized his active and explorative behavioural bouts. Such results echo those found in other avian species (e.g., flamingos) where males displayed stronger lateral preferences within given behavioural categories (Regaiolli, Spiezio, Ottolini, Sandri, & Vallortigara, 2021b)but we have yet to formally test for the existence of a similar trend within the New Caledonian crow, as these results may have just been mirroring the intriguing possibility that laterality could vary within individuals.
Another perspective we hope to see future studies explore is that laterality, even if not as strong as in tool behaviour, could (a) be found across several behavioural categories, (b) get reversed from one domain to the other, thus reflecting a complex form of cerebral lateralization and behavioural control (Rogers, 2021;Rogers et al. 2013), and (c) overall vary in strength, with task features acting as predictors . Bird locomotion, for instance, has already been thoroughly discussed; results show not only strong individual footedness preferences in a species like the Japanese jungle crow (Corvus macrorhynchos) (Izawa et al. 2005), but also highlight the need to take into account any evolutionarily relevant environmental factors that may shape these trends (Brown & Magat 2011b;Davies & Green 1991;Güntürkün et al. 1988). To completely rule out the existence of similar locomotor preferences in the New Caledonian crow, our lack of significant results within this behavioural category should be sought once again across ecologically different populations, as we know that crow groups reside in a wide range of territories on New Caledonia (Abdelkrim et al. 2012). Furthermore, the lack of lateralization within our categories of "activity & exploration", "locomotion", and "maintenance" may have been due to the large spectrum of behaviours that each category covers, and a more detailed analysis per behaviour would be welcomed, if one could obtain sufficient data points. As the video clips used here were originally taken to study tool behaviour, they may have introduced an additional bias regarding the types of behaviours expressed in our data.
Overall, our study supports the tight connection between laterality and the evolution of tool-making skills that has previously been argued by others (Haslam et al. 2019: Hunt, 2000b, Hunt et al., 2001, 2006Mosquera et al. 2012). What is yet uncertain is which came first. One major theory proposes that being a tool-using species is a driving factor in the evolution of laterality, as tool use causes the brain to become more strongly lateralized (Cashmore et al. 2008;Hunt et al. 2001;MacNeilage, 2014;Martinho et al. 2014;Mosquera et al. 2012;Steele & Uomini, 2005;Uomini, 2009;Uomini & Ruck 2018;Weir et al. 2004). Alternatively, tool-making skills could emerge in already lateralized species thanks to hemispheric specialization in vertebrates, which enhances the simultaneous processing of two tasks that draw on different hemispheres (Rogers, 2017;Versace & Vallortigara 2015). The lateralization of tool-use and tool-making skills would then be further supported through the learning process and an individual's lifelong practice (Biro et al. 2006;Boesch, 1991;Sousa et al. 2009;Uomini et al. 2020).
Given the individual laterality profiles suggested by our findings, the present study advocates for comparison, already at the species leveli.e., between populations, between individuals and between those two units. In order to achieve such a goal, the focal subjects need to be individually identified, for instance via the trapping and marking of birds with a leg ring, a wing tag or even a harness-mounted transceiver tag (Caffrey, 2002a;Hunt, 2016;Rutz et al. 2012aRutz et al. , 2012bSt Clair et al. 2016), keeping in mind the extreme time and luck required to catch crows (Caffrey, 2002b;Hunt, personal communication 2016;Uomini, personal observation), or simply by focusing our work on already labelled populations if we wish to avoid the anxiety-inducing, fitness-reducing procedures involved in capture and marking (Burley et al. 1982;Pietz et al. 1993). Moreover, a long-term monitoring would allow researchers to both take a special interest for individual behaviour outside of tool use and manufacture and follow the evolution of behavioural trends across generations. For instance, it seems that social life plays a crucial role in a New Caledonian crow's life, as already suggested by the peculiar structure of a highly-tolerant family unithowever, in this domain like in many others, studies remain scarce (Holzhaider et al. 2011;Kenward et al. 2004). In other words, there is an urgent need to biologically and ecologically better understand the New Caledonian crow, whose tool manufacture skills, unlike any other non-human species, present hominid-like features (Hunt & Gray 2004a;Hunt & Uomini 2016), if we wish to ever pursue this promising comparative road and perhaps uncover new co-evolutionary paths leading to the emergence of language.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Funding
This work was supported by Templeton World Charity Foundation: [Grant Number 0271].

Author contributions
CM conceived the study. CM and NU designed the study. NU collected video data. CM coded video data and performed all analyses. CM wrote the paper and prepared the supplementary material, with inputs from NU. All authors approved the final manuscript.

Ethics statement
All research activities complied with national and international ethical standards for wild animal observation (The Ornithological Council (Fair et al. 2010), the Association for the Study of Animal Behaviour (2012), the American Psychological Association (2010), and the International Primatological Society (2014)), and were carried out under permits from the Direction de l'Environnement, Province Sud, New Caledonia and with ethical approval from the Max Planck Society's Ethics Council (2016 and #2018_34). The authors report there are no competing interests to declare.

Data availability statement
The R script and associated data can be downloaded from the following OSF repository: https://osf.io/7x8ur.