(a) Variation and morphological convergence in Diplodactylidae maxillae.
Phylogenetic position is a highly significant predictor of maxilla shape diversity in New Zealand Diplodactylidae, with all genera (Dactylocnemis, Hoplodactylus, Mokopirirakau, Naultinus and Woodworthia) being morphologically distinct. These results contrast previous long-held notions of skeletal conservatism in New Zealand’s geckos (e.g. (27,37)) through identification of taxonomically informative morphological variation within a single skeletal element. This retention of genus-level phylogenetic signal is remarkable given pronounced ecological species radiations since the early Miocene (14). However, similar trends of reduced disparity in maxillae (relative to rate of evolution) are observed across both extant and extinct squamates (excluding snakes; (38)), suggesting constrained evolution in this cranial region.
Diplodactylidae maxilla shape is predominantly characterized by two character-states, described by the first two axes of both PCA and CVA: (1) posterior extension/reduction of the nasal margin; and (2) increase/decrease in dorsoventral extent of the facial process. The separation of genera along PC1 appears to reflect broad habitat use of the New Zealand Diplodactylidae, with terrestrial-arboreal (Dactylocnemis, Hoplodactylus and Woodworthia) and exclusively arboreal (Naultinus) genera occupying positive and negative regions respectively (39,40). This morphological signature of habitat use extends to species-level comparison, most notably in the discrimination of the terrestrial-arboreal M. ‘southern North Island’ from the arboreal M.granulatus (41), characterized by a shift to more positive values.
In gekkotans, arboreal forms tend towards broad, pointed and dorsoventrally shallow skulls, enabling faster climbing speeds on non-horizontal surfaces (42,43). While cranial modifications associated with habitat use are undocumented in the New Zealand Diplodactylidae, extension of the nasal margin in arboreal species appears to be linked to two superficial morphological changes in the adjacent prefrontal margin: (1) a reduction in anterior extent (observed in other Gekkota; (44)); and (2) formation of a thickened ridge along the prefrontal orbital margin (Supplementary Figure 7). While the function of these structures remains unclear, association with arboreality provides strong evidence for ecomorphological convergence between phylogenetically independent lineages. Despite describing similar shape change, separation of genera along CV1 reflects broad phylogenetic relationships, distinguishing broad (Hoplodactylus, Woodworthia) and narrow (Dactylocnemis, Mokopirirakau, Naultinus) toed clades at positive and negative values respectively; supporting previous morphological classification (20).
In addition to habitat use, skull-shape evolution in lizards is strongly influenced by diet, with shape variation concentrated in the premaxilla, nasal and jaw joint, reflecting their roles in rostral prey capture and feeding biomechanics (38,42). Herbivorous lizard skulls tend towards reduced snouts and high temporal regions relative to carnivorous lizards, contributing to an increased bite strength required for processing fibrous and tough foliage (45–47). Conversely, omnivorous gekkotans represent intermediate forms not specialized to particular feeding behaviors, and consequently lack unique morphological adaptations (48). New Zealand geckos are predominantly omnivorous, consuming a wide variety of food items including plant matter (fruit, honeydew and nectar) and arthropods (40). Such extensive dietary overlap effects the performance of diet as an explanatory variable of maxilla shape diversity, given categories (omnivorous and insectivorous) are not discrete.
(b) Efficacy of size-based discrimination
Maxilla size was significantly correlated with phylogenetic affinity, however, only Hoplodactylus could be fully differentiated (under post-hoc comparisons), with the remaining Diplodactylidae genera exhibiting variable degrees of overlap. This highlights the inefficiency of previous size-based taxonomic identification of non-Hoplodactylus Holocene subfossil geckos, especially intermediate-sized genera (Dactylocnemis, Mokopirirakau and Naultinus), which exhibit complete size overlap. Similarly, while large relative size proves reliable in discriminating extant H. duvaucelii, applications in Holocene subfossil identification are limited given assumptions of temporal taxonomic homogeneity (or “covert biases”; (49)).
Previous analyses of squamate genera including Anolis (50,51) and Iguana (52) have shown maxillae to be effective predictors of snout-vent length (SVL). Our results exhibit similar trends both between and within Diplodactylidae genera, with mean genus centroid size reflecting relative SVL (53), and larger species (N. punctatus, D. ‘three kings’) having increased centroid sizes relative to congeners (N. elegans, D. pacificus; (54,55)).
(c) Increased Holocene diversity of large geckos
Our results provide evidence for increased morphological diversity of large geckos during the Holocene in New Zealand, with declines in both shape and size variation following Polynesian and European colonization.
Combined Procrustes and Mahalanobis distance comparisons provide support for previous size-based classification of five Holocene subfossils (A, D, E, J, K) as H. duvaucelii, confirming assumed prehuman distribution across both the North and South Islands. The remaining six Holocene subfossil specimens (B, C, F, G, H, I) exhibited classification discrepancies and/or reduced assignment probabilities (below relevant thresholds), reflected in their unique position across CV1/CV2. These distinct Holocene subfossil maxillae (“unknown taxa”) are not reflective of differential adaptation to mainland and island habitats (see above), therefore reflecting either increased morphological diversity of mainland large species (not encompassed by extant populations) or the presence of at least one extinct, large, broad-toed Diplodactylidae species.
Based on digit morphology, the extinct giant H. delcourti was positioned within the broad-toed clade, sister to H. duvaucelii (24), suggesting these “unknown taxa” could potentially represent small or even juvenile H. delcourti (with respect to the latter hypothesis). However, this seems unlikely given the paucity of reported subfossil remains of H. delcourti (56), despite extensive collections of other Diplodactylidae taxa (37). Accurate phylogenetic affinities of both H. delcourti and “unknown taxa” could be determined through future ancient DNA analysis.
During the Holocene, mainland H. duvaucelii (and “unknown taxa”) reached larger sizes than extant populations, reflected in a reduction in maximum maxilla size (a proxy for body size; e.g. (50)). Such sized-biased extinction is well-documented in Quaternary lizards globally (51,57–59), including the extinction of two large-bodied Eugongylinae skink species (Oligosoma northlandi and Oligosoma sp.) in northern New Zealand (12,31,60). This reflects the inherent vulnerability of New Zealand’s large-bodied, nocturnal herpetofauna towards high-predation rates and ecological displacement by exotic mammals (including the Pacific rat (kiore); (61,62)), particularly in forest-cleared environments (63). Smaller lizards can escape predation during periods of inactivity through utilizing narrow retreats, given limited overlap in body diameter with small mammalian predators (39). Conversely, refugia utilized by large-bodied lizards can be accessed by mammalian predators, evidenced by reductions in body weight, tail width and recruitment of H. duvaucelii on kiore-inhabited islands (35,64).
Similar to extant H. duvaucelii populations (65), Holocene subfossil H. duvaucelii also exhibit a latitudinal cline in maxilla size opposing Bergmann’s rule (i.e. increased size at high latitudes), with individuals from northern localities being noticeably larger than those from southern localities. For diurnal lizards, reduced body size appears to be an advantageous thermoregulatory strategy in cooler climates, with high surface-area to volume ratio permitting rapid heat gain whilst sun-basking (66,67). Despite being nocturnal, H. duvaucelii occasionally emerge from retreats to thermoregulate through cryptic sun-basking (68,69), suggesting small body size provided an adaptive advantage at high latitudes.