Comparative analyses of micro- and macro-scale surface structures in the convergent evolution of rain-harvesting behaviour in lizards
Editor: Anthony Herrel
Associate Editor: Donald Miles
Abstract
Rain-harvesting behaviour represents an adaptation for water collection that has evolved in some species of iguanian reptiles inhabiting arid environments. To date, such behaviour has been observed only in members of the families Agamidae and Phrynosomatidae. A common set of integumentary features characterizes these rain-harvesting species, including scale surface microstructures (SSMs), capillary channels and hinge joints. The influence of variations in these features on rain-harvesting has been a subject of discussion for many decades. Nevertheless, a comprehensive comparative study of similarities and differences between harvesting and non-harvesting species on a broader scale remains lacking. In this study, we classify scale surface microstructures into three categories: large hexagons (SSM1), smaller nested hexagons (SSM2) and hinge pits (SSM2H). As the first two SSM types are widespread, they do not appear to be directly linked to the adaptation for rain-harvesting. Conversely, the presence or absence of hinge pits distinguishes harvesters from non-harvesters. Additionally, channel hierarchy, width and structure determine the effectiveness of the rain-harvesting architecture. Only Moloch horridus exhibits distinct integumentary features in comparison to other agamids and rain-harvesting species. Ancestral character state reconstruction suggests that rain-harvesting behaviour was likely absent in the ancestor of Iguanians, even though overlapping scales and SSM1 were present. Our findings illustrate that rain-harvesting species have independently converged upon similar structural solutions to address their water acquisition challenges, building upon shared pre-existing features.
Introduction
Collecting available water from the environment and transporting it for ingestion has been the vital achievement throughout terrestrial animal evolution. One unique method is ‘rain-harvesting’, an adaptation in which an animal collects environmental water by integument and transports it to its mouth. This behaviour has been documented in beetles, lizards, snakes and tortoises living in arid or semi-arid habitats (Comanns et al., 2011, 2016, 2017; Joel et al., 2017; Phadnis et al., 2019).
These distinctive linked features and behaviours, first discovered in lizards in 1923, were initially thought to involve water absorption through the lizard's skin (Davey, 1923). Further research revealed that water was transported to the jaws via scale hinges (Bentley & Blumer, 1962; Gans et al., 1982). Sherbrooke (1990, 1993) used the name ‘rain-harvesting’ and referred to other hypothetically possible methods as ‘fog- or dew-harvesting’. Subsequently, Comanns et al. (2011) suggested using more comprehensive term ‘moisture harvesting’ for such behaviour. However, considering that ‘moisture’ has a broad meaning and that no fog or dew-based harvesting by lizards has been observed in nature to date, we prefer to use the term ‘rain-harvesting’.
To date, rain-harvesting has only been observed in two lizard families: Agamidae and Phrynosomatidae. The geographic and phylogenetic distances between species exhibiting rain-harvesting suggest that this behaviour has evolved convergently on four continents (Peterson, 1998; Sherbrooke, 1990, 1993, 2003; Sherbrooke et al., 2007). What enables this extraordinary behaviour are microstructures of the scale integument, particularly the asymmetrical, semi-tubular, interscalar channels. These channels are formed in scale hinges and connect with adjacent features of scales spread throughout the skin (Comanns et al., 2017; Sherbrooke et al., 2007). When water is available, it enters these channels, fills them and is then transported to the mouth through capillary action (Comanns, 2016; Malik et al., 2014). Interscalar channels have been studied in terms of their distribution across the body, microstructure and surface protrusions. These studies have inspired novel research of biomimicry and applied fluid dynamics engineering inspired not only drawing from reptiles (e.g. Buchberger et al., 2015, 2018; Comanns, 2016; Comanns et al., 2015, 2017) but also from other animals as natural models (e.g. Hischen et al., 2018; Lifka et al., 2021; Plamadeala et al., 2017).
Besides the interscalar channels, the influence of variation in scale surface microstructure on rain-harvesting has been discussed for decades (Price, 1982, 1990). Reptile skin has an outermost epidermal layer, called oberhäutchen, composed of corneous beta proteins (CβPs) (Holthaus et al., 2018). Oberhäutchen features consist of macro- and microstructures such as sharp spines, leaf-like or hexagonal micro-ornamentations and micro-pores (e.g. Alibardi, 2000; Sherbrooke et al., 2007; Spinner et al., 2013). Rain-harvesting lizards often possess hexagonal scale surface microstructures (SSM), also referred to as honeycomb ornamentations by Peterson (1984). Later, even smaller hexagonal SSM on a nano-scale have been discovered within the honeycombs in some lizards (Comanns et al., 2011; Sherbrooke et al., 2007; Yenmiş et al., 2016). These features have been discussed by Joel et al. (2021) in the context of surface energy in rain-harvesting and skin pollution, for example, retention of dirt particles. However, to date, no comprehensive study has compared the presence of both interscalar channels and SSMs in the two families of lizards, Agamidae and Phrynosomatidae, containing rain-harvesting and non-rain-harvesting species.
Lizard skin sense organs were first described by Leydig (1868) and studied as indicators of geographical distribution (Scortecci, 1941) and taxonomy (Allam et al., 2019; Dowling et al., 1972). Although Sherbrooke and Nagle (1996) described them in Phrynosoma mentioned, they might be used as mechanoreceptors, perhaps, in one sense, to alert buried lizards to come to the surface to rain-harvest at night, an observed behaviour.
Here, we compared the structures, micro- and nano-topography of interscalar channels and scale surfaces of 17 species, 8 of Agamidae and 9 of Phrynosomatidae, of which 5 are rain-harvesters (3 and 2, respectively). We determined morphological similarities and variations between and within families, as well as among rain-harvesters, non-rain-harvesters and species with unknown behaviour using stereo, light and electron microscopy. For the first time, we address the question of whether convergent evolution is the dominant selective force in determining scale microstructure for rain-harvesting by lizards.
Materials and methods
We studied skin samples of 17 species from four different continents. The designation ‘Uk’ (Unknown) in Table 1 refers to species from the genus Phrynosoma whose rain-harvesting behaviour has not yet been determined, and thus, they are categorized among the non-rain-harvesting for statistical analysis.
| Family | Species | Distribution | Rain-harvesting | Skin specimens |
|---|---|---|---|---|
| Agamidae | Phrynocephalus horvathi | Turkey, Armenia | Yes (Yenmiş et al., 2016) | TÜBİTAKa Project, Grant No: 114Z055 |
| Moloch horridus | Australia | Yes (e.g. Withers, 1993) | DZUP 9228b | |
| Trapelus pallidus | Syria, Jordan | Yes (Veselý & Modrý, 2002) | NMP6V 34 905 | |
| Trapelus sanguinolentus | Kazakhstan, Kyrgyzstan, Uzbekistan | No (Veselý & Modrý, 2002) | DZUP 0112 | |
| Trapelus ruderatus | Near East | No (Veselý & Modrý, 2002) | NMP6V 34 856/1 | |
| Laudakia stellio | Turkey | No (Yenmiş et al., 2016) | TÜBİTAK Project, Grant No: 114Z055 | |
| Laudakia caucasica | Turkey | No (Yenmiş et al., 2016) | TÜBİTAK Project, Grant No: 114Z055 | |
| Trapelus lessonae | Turkey | No (Yenmiş et al., 2016) | TÜBİTAK Project, Grant No: 114Z055 | |
| Phrynosomatidae | Phrynosoma cornutum | USA (SE), Mexico (NE) | Yes (Sherbrooke, 1990) | AMNH (WCS collection) (No: 4294; 4297) |
| Phrynosoma modestum | USA (SW), Mexico (N) | Yes (Sherbrooke, 2002) | AMNH (WCS collection) (No: 6658; 6686) | |
| Phrynosoma asio | Mexico (SW) | Uk | DZUP 9511 | |
| Phrynosoma orbiculare | Mexico (Central) | Uk | DZUP 9208 | |
| Phrynosoma hernandesi | Canada, western USA, into north central Mexico | Uk | AMNH (WCS collection) (No: 6610; 5604) | |
| Phrynosoma solare | USA (Arizona), Mexico (NW) | Uk | AMNH (WCS collection) (No: 4215; 4299) | |
| Phrynosoma mcallii | USA (Arizona, California), Mexico (NW) | No (Mayhew & Wright, 1971) | AMNH (WCS collection) (No: 4100; 4085) | |
| Phrynosoma blainvillii | USA (SW), Mexico (NW) | Uk | AMNH (WCS collection) (No: 4019; 4064) | |
| Phrynosoma goodei | USA (SW Arizona), Mexico (NW, Sonora) | Uk | AMNH (WCS collection) (No: 4095; 4066) |
- Uk: Unknown: It is not yet known whether the species is a rain-harvester or not (species with unknown behaviour).
- N, North; NE, North-east; NW, North-west; SE, South-east; SW, South-west.
- a The Scientific and Technological Research Council of Turkey.
- b Collection number.
Skin samples
Species distributed in Anatolia were collected from the wild (Laudakia caucasica and Phrynocephalus horvathi from Iğdır, Trapelus lessonae from Konya and Laudakia stellio from İzmir) and were euthanized using barbiturate (Ege University Animal Experimentations Local Ethics Board; No: EUHADYEK-2013-043). Skin samples measuring 2 × 2 cm were extracted from the mid-dorsal area of each specimen and fixed in Bouin's solution, 70% ethanol or glutaraldehyde, depending on the procedure. Other samples were obtained from the Prague National Museum (NMP6V), the Zoological Department of Palacký University collection (DZUP) and the American Museum of Natural History (AMNH) (from special collection of Dr. W. C. Sherbrooke) (Table 1). These samples were formalin fixed and ethanol preserved.
Sample preparation
For examination under stereomicroscope Euromex DZ5040, skin samples were air-dried for 10 min. For light microscopy, specimens fixed in Bouin or ethanol were processed using standard protocol and embedded in paraffin. Six-μm-thick sections were prepared for histological studies, stained with Ehrlich's haematoxylin–eosin (H&E) and examined using Olympus CX31 light microscope.
Scanning electron microscopy (SEM) samples of skin were prepared following the protocol of Yenmiş et al. (2016). Two sets of samples were prepared, the first one was used to scan scale surfaces and the second one to scan scale cross-sections. Samples were analysed and measured by Thermoscientific apreo8 (Thermo Fisher Scientific, Massachusetts, USA).
Scale measurements
Skin samples under the stereomicroscope Euromex DZ5040 were photographed using Euromex HD II digital camera, recorded, analysed and measured by LCmicro (version 5.2) and ImageJ (Version 1.53m) software. Variations in scale size, homogeneity and arrangement were recorded. The smallest and largest scales were determined using an ocular micrometre. Scale size was measured from microphotographs as the diameter of the smallest possible circle into which the scale could fit. For the imaging and measurement of the interscalar channels, the inner surface of the epidermis was studied after separation from the dermis using a scalpel and forceps. The width and length of both longitudinal and transversal interscalar channels were measured for rain-harvesting species (see Comanns, 2016). Channel widths were determined by measuring from three different locations of each channel and averaging them. For each species, more than 30 channels were measured for these parameters.
Scanning electron microscopy microphotographs were used to measure scale microstructures. Randomly chosen 20 scale surface honeycombs (SSM1), 20 smaller hexagons nested (SSM2) and 20 hinge micro-ornamentations (SSM2H) were measured using the SEM micrographs with the same technique as scale size measurements (Fig. 1). Skin sense organs per scale and per 6 mm2 skin sample were counted for each species.
Statistical analyses
The width and the length of both longitudinal and transversal interscalar channels measured were indexed as width/length and analysed with student's t-test (P < 0.05). We tested the normality of the diameter values of SSM1, SSM2 and SSM2H with a Kolmogorov–Smirnov test. SSM1 and SSM2H were not distributed normally (P < 0.05), therefore we performed nonparametric Mann–Whitney U-test. In contrast, SSM2 was distributed normally (P > 0.05) so a Student's t-test was performed. All morphometric data are presented in mean ± standard deviation format. We compared the measurements in terms of families and the ability to harvest rain. All tests were performed using the statistical software SPSS 25.0.
Ancestral character state reconstruction
We used previously published data (see Data S1) of three mitochondrial (ND2, ND4 and cytb) and three nuclear (BDNF, NKTR and RAG1) loci for 27 species (10 Agamidae, 15 Phrynosomatidae and as outgroups, 1 Scincidae, Scincus, and 1 Lacertidae Takydromus sexlineatus). GenBank accession numbers are provided in the Table S1 along with trait states and base pair information. Mesquite v3.81 (Maddison & Maddison, 2023) was used to concatenate and analyse the sequences, and the alignment was performed using MAFFT v7.52 (https://mafft.cbrc.jp 2023/March). Before generating a majority-rule consensus tree from a series of 100 randomly resolved trees, the trees were pruned to overcome the problem of different terminals. We reconstructed the ancestral states using both maximum parsimony (MP) (unordered) and maximum likelihood (ML) (Markov k-state one-parameter, Mk1) with the ‘trace character history’ function of Mesquite.
Results
A closer look at skin topography
Agamids have heterogeneous scale pattern in terms of shape and size (Fig. 2). All agamid species examined possess overlapping scales. The presence and the number of ridges (carina, keel) on the scales are variable through agamids. Ph. horvathi, T. pallidus and T. lessonae tend to have smoother scale surfaces with reduced or no ridges, while L. caucasica and T. ruderatus might have up to four ridges per scale. Moloch horridus has bulges (knob-like obtuse protrusions) rather than ridges. SSOs tend to appear on larger scales. The agamid with the maximum number of SSOs – up to 5 per scale – is L. stellio which also has the maximum number in an area of 6 mm2 (Fig. 3, Table 2). All agamids have hair-like SSO with the exception of M. horridus which has lenticular ones. Scale edges are either straight or undulated except for L. stellio which has zigzag border of margins. Scale sizes of agamids range from 0.1 to 2.9 mm in diameter, with M. horridus having both minimum- and maximum-sized scales.
| Family | Species | Ridge number/scale | SSO number/scale | SSO number/6 mm2 | SSO type | Scale edge type |
|---|---|---|---|---|---|---|
| Agamidae | Phrynocephalus horvathi | 0 | 0–1 | 12 | Hair-like | Straight |
| Moloch horridus | 0 (has bulges instead) | 0–1 | 12 | Lenticular | Zigzag | |
| Trapelus pallidus | 0–1 | 0–1 | 6 | Hair-like | Straight | |
| Trapelus sanguinolentus | 3 | 0–1 | 9 | Hair-like | Undulated | |
| Trapelus ruderatus | 1–4 | 1–3 | 18 | Hair-like | Undulated | |
| Laudakia stellio | 1–3 | 0–5 | 30 | Hair-like | Zigzag | |
| Laudakia caucasica | 2–4 | 0–1 | 9 | Hair-like | Straight | |
| Trapelus lessonae | 0 | 0–1 | 12 | Hair-like | Straight | |
| Phrynosomatidae | Phrynosoma cornutum | 1 | 0–1 | 6 | Lenticular | Straight |
| Phrynosoma modestum | 1 | 0–1 | 12 | Lenticular | Straight | |
| Phrynosoma asio | 1–3 | 0–1 | 6 | Lenticular | Straight | |
| Phrynosoma orbiculare | 1–3 | na | na | na | Straight | |
| Phrynosoma hernandesi | 1 | 0–1 | 3 | Lenticular | Straight | |
| Phrynosoma solare | 1 | 0–1 | 6 | Lenticular | Straight | |
| Phrynosoma mcallii | 3–4 | 0–1 | 8 | Lenticular | Straight | |
| Phrynosoma blainvillii | 1 | 0–1 | 3 | Lenticular | Straight | |
| Phrynosoma goodei | 1 | 0–1 | 20 | Lenticular | Straight |
- Rain-harvester given in bold font.
- na, not applicable; SSO, skin sense organ.
As for phrynosomatids, Phrynosoma hernandesi, P. goodei and P. solare have some slightly overlapping scales (Fig. 4). All the species have at least one ridge on the scale, and it can be up to four in P. mcallii. As in agamids, SSOs tend to appear on larger scales. None of the species have more than one SSO per scale. P. goodei has the maximum number of SSOs in an area of 6 mm2. All phrynosomatids have lenticular SSOs and they all have straight-scale edges (Fig. 2, Table 2). Scale sizes range from 0.1 to 1.7 mm in diameter with P. modestum having the smallest and P. solare having the largest scales.
Carrier channels
The capillary channels encircle irregularly hexagonal scales of the dorsal body, so the broad capillary network has an almost hexagonal structure in harvesters (Fig. 5).
The longitudinally oriented interscalar channels are longer than the length of transversal ones in all the rain-harvesting species (t-tests, P < 0.05). Among them, M. horridus has the longest longitudinal and transversal channels (long.: 361.89 ± 10.2 μm, trans.: 319.9 ± 9.01 μm; t-test, P = 0.036) (see Table 3 for indexes). The transversally oriented channels are wider than the longitudinal channels in all the rain-harvesters, yet this width difference was not statistically significant for M. horridus and T. pallidus (t-test, P > 0.05). M. horridus has the widest longitudinal and transversal channels (long.: 182.8 ± 7.41 μm, trans.: 185.87 ± 9.09 μm; t-test, P = 0.059). Evaluation of the width/length ratios of both longitudinal and transversal channels showed no significant difference among families of the harvesters: t-test value of longitudinal channel width/length (L-w/l) is 0.63 while t-test value of the same ratio of transversal channels (T-w/l) is 0.22. Among species, longitudinal channel w/l of M. horridus is significantly different from other rain-harvesters (L-w/l; t-test = 0.003).
| Family | Species | Channel size indexes | |
|---|---|---|---|
| L-w/l (mm) | T-w/l (mm) | ||
| Agamidae | Phrynocephalus horvathi | 0.201 ± 0.003 (40) | 0.531 ± 0.004 (33) |
| Moloch horridus | 0.505 ± 0.008 (42) | 0.581 ± 0.007 (41) | |
| Trapelus pallidus | 0.209 ± 0.002 (39) | 0.345 ± 0.004 (32) | |
| Phrynosomatidae | Phrynosoma cornutum | 0.212 ± 0.003 (36) | 0.314 ± 0.003 (33) |
| Phrynosoma modestum | 0.262 ± 0.003 (34) | 0.366 ± 0.004 (35) | |
- Note: Values are given as mean ± SD (n).
In non-harvesters and species with unknown behaviour, the longitudinal interscalar channels are longer than the transversal ones as well (t-tests, P < 0.05). Among them, L. stellio has by far the longest and widest longitudinal and transversal channels (length: long.: 1794 ± 45.2 μm, trans.: 926.7 ± 16.2 μm; width: long.: 294.6 ± 10.2 μm, trans.: 296 ± 7.9 μm; t-test, P = 0.042). T. lessonae has the shortest channels (length: long: 284 ± 8.8 μm; trans.: 192 ± 6.5 μm; t-test, P = 0.05) while the narrowest channels are found in P. mcallii (width: long: 102 ± 4.9 μm; trans.: 64.7 ± 3.8 μm; t-test, P = 0.04) (Fig. 6).
Transversal width in T. sanguinolentus and P. solare is smaller than the longitudinal width, while they are nearly equal in P. hernandesi. Contrarily, in P. goodei, L. stellio and T. lessonae, the width of the longitudinal channels is smaller than that of the transversal ones. Measurements could not be taken from T. ruderatus, L. caucasica, P. asio, P. orbiculare and P. blainvillii due to the insufficient separation of the connective tissue and, accordingly, the inability to obtain a clear view of the channels. The indexes are given in Table 4.
| Family | Species | Channel size indexes | |
|---|---|---|---|
| L-w/l (mm) | T-w/l (mm) | ||
| Agamidae | Laudakia stellio | 0.163 ± 0.005 (40) | 0.319 ± 0.007 (40) |
| Trapelus lessonae | 0.288 ± 0.007 (33) | 0.2 ± 0.006 (31) | |
| Trapelus sanguinolentus | 0.192 ± 0.002 (35) | 0.567 ± 0.01 (35) | |
| Phrynosomatidae | Phrynosoma solare | 0.511 ± 0.009 (30) | 0.344 ± 0.006 (30) |
| Phrynosoma hernandesi | 0.216 ± 0.003 (32) | 0.382 ± 0.005 (33) | |
| Phrynosoma goodei | 0.407 ± 0.01 (37) | 0.725 ± 0.02 (40) | |
| Phrynosoma mcallii | 0.252 ± 0.01 (30) | 0.372 ± 0.01 (31) | |
- Values are given as mean ± SD (n).
Analyses of the data from both Tables 3 and 4 showed that there is no significant difference in either longitudinal or transversal channel indexes among families or harvesting behaviour (t > 0.05).
Cross-sectional SEM analyses of the hinges showed that channels have narrow openings and wider bases. Here, skin folds of varying sizes have been observed. M. horridus has connecter subchannels that are narrower and deeper than the linked hinge bases. These sub-channels have not been found in the rest of the species (Fig. 7).
Micro-ornamentations
Detailed SEM analyses showed that all the rain-harvesters possess SSM1 throughout the scale surface except for P. cornutum which bears SSM1 at the edges of the scales only. T. pallidus has the smallest SSM1 and Ph. horvathi has the largest on average. SSM1 of the Ph. horvathi are overlapped and have sharp borders and one corner is elevated, while in other harvesters, SSM1 cover the surface side by side and their joint borders are elevated which makes their centres appear hollow (Fig. 8).
All the harvesters, except for M. horridus, have SSM2 nested in SSM1 as well as SSM2H on the surface of skin folds (island-like protrusions) within the channels. M. horridus only has SSM1. P. modestum has the smallest and T. pallidus has the largest SSM2 while SSM2H is smallest in T. pallidus and largest in Ph. horvathi. SSM2 are hexagonal in Ph. horvathi and T. pallidus while they are pit-like in P. cornutum and P. modestum. SSM2H resembles a ‘honeycomb’ much more than both SSM1 and SSM2 and are alike in all the harvesters that have them.
Ten of the twelve non-rain-harvesters and species with unknown behaviour studied have SSM1 (5 agamids and 5 phrynosomatids). In P. solare and P. mcallii, SSM1 are rounded, not honeycomb shaped, additionally, do not have SSM2. In L. stellio, SSM1 resemble those of Ph. horvathi in terms of imbricated pattern. P. orbiculare and P. goodei do not have the typical SSM (neither SSM1 nor SSM2 or SSM2H), instead they have non-geometrical, root-like and porous ornamentations at the scale surface and the hinge regions, which is why they have not been included in the statistical analyses. In the other species, SSM1 have similar shapes and they cover the surface side by side with elevated joint borders. It was not possible to detect SSM2H in P. asio, T. ruderatus and L. stellio due to the insufficient SEM micrographs. In other species, comb-like SSM2H are alike in terms of shape (Fig. 9).
Statistical analyses showed that the size of SSM1, SSM2 and SSM2H were significantly different among families (SSM1, U = 9227, P < 0.01; SSM2H, U = 3450.5, P < 0.01; SSM2, t = 10.525, P < 0.01). Yet, when comparing harvesters with non-harvesters and species with unknown behaviour, only SSM2H differs significantly among two groups (U = 3418.5, P < 0.01); however, neither SSM1 (U = 8761.5, P > 0.05) nor SSM2 (t = −0.963, P > 0.05) differs significantly (Table 5).
| Family | Species | SSM1 (μm) | SSM2 (μm) | SSM2H (μm) |
|---|---|---|---|---|
| Agamidae | Ph. horvathi | 25.16 ± 2.87 | 0.728 ± 0.116 | 0.735 ± 0.103 |
| T. pallidus | 14.87 ± 2.183 | 0.761 ± 0.184 | 0.415 ± 0.057 | |
| M. horridus | 19.77 ± 3.3 | Naa | Na | |
| T. ruderatus | 20.51 ± 2.81 | 0.85 ± 0.156 | 0.367 ± 0.065 | |
| T. sanguinolentus | 28.88 ± 8.27 | 0.847 ± 0.118 | Xb | |
| T. lessonae | 19.67 ± 3.29 | 0.689 ± 0.1 | 1.824 ± 0.228 | |
| L. stellio | 21.01 ± 5.47 | 0.69 ± 0.104 | X | |
| L. caucasica | 25.19 ± 5.10 | 0.753 ± 0.180 | 0.462 ± 0.094 | |
| Phrynosomatidae | P. cornutum | 25.08 ± 3.94 | 0.569 ± 0.107 | 0.459 ± 0.085 |
| P. modestum | 22.23 ± 2.055 | 0.478 ± 0.094 | 0.628 ± 0.011 | |
| P. hernandesi | 21.77 ± 3.04 | 0.733 ± 0.171 | 2.19 ± 0.227 | |
| P. solare | 18.58 ± 2.06 | Na | 0.539 ± 0.088 | |
| P. asio | 22.16 ± 3.3 | 0.182 ± 0.047 | X | |
| P. blainvillii | 17.98 ± 3.8 | 0.538 ± 0.099 | 7.110 ± 1.136 | |
| P. mcallii | 6.51 ± 0.81 | Na | 0.913 ± 0.164 |
- Known harvesters are given in bold font.
- a Na: Not applicable = Species do not have the mentioned characteristics.
- b X: It was not possible to get proper micrographs (SEMs) for measurement.
Cross-sectional analyses
Histological and SEM cross-sections showed that all the agamids have interscalar channels in between the two adjacent scales regardless of rain-harvesting (Fig. 10a–c,a′–c′) or not (Fig. 10d–h,d′–h′). There are skin folds with various sizes within the channels. SSM1 are visible on the oberhäutchen in sections stained with H&E (‘ssm’ within images), if it is protected during the preparation (Fig. 10c,e). H&E staining separated the dermis from epidermis as well as the epidermal regions as lowermost stratum germinativum, keratin-corneous layers (‘k’ within images) and the oberhäutchen as light yellowish pink.
Among Phrynosoma, P. goodei, P. asio and P. orbiculare showed slight overlapping compared to others. The tip of one scale does not lean on the base of the next one (Fig. 11e,e′,g,g′). Rain-harvesters (Fig. 11a,a′,b,b′), non-rain-harvesters and species with unknown behaviour (Fig. 11c–i,c′–i′) of Phrynosoma members have interscalar channels in between the two adjacent scales. In all the species scale, hinges are made of all the layers of the epidermis. There are skin folds with various sizes within the channels. SSM are not visible due to the loss of full oberhäutchen during preparation. H&E staining indicated epidermis and dermis, live and dead cells of stratum germinativum and keratinocytes–corneocytes (indicated as ‘k’ within the images) on top of it respectively.
Trait evolution
According to the Mesquite reconstruction (Fig. 12), rain-harvesting could have evolved independently at least twice within Agamidae over the course of evolutionary history. In terms of maximum parsimony analysis, it suggests that the common ancestor of phrynosomatids had 50% probability of exhibiting the rain-harvesting behaviour. However, the maximum-likelihood model (ML) indicates that rain-harvesting behaviour was more likely to be absent than present (60% vs. 40%, respectively) in the ancestor. This ML result is strongly supported by a decision threshold of 2.0 (Fig. 12). Regarding the presence of overlapping scales, both parsimony and likelihood analyses unequivocally (100% probability) reconstructed the trait as the ancestral state. This is also congruent with the reconstruction of the SSM1 trait, with MP indicating 100% probability and ML 93% probability.
Discussion
Scale morphology
Scale size and shape variability indicate that species do not share common scale features based on family affiliation or rain-harvesting behaviour. Therefore, scale shape alone is not a distinctive characteristic. However, it is worth noting that scale edge shape is consistently straight in all Phrynosoma members, while it varies in agamids as straight, undulated and zigzag shapes (Fig. 13). This variation could be attributed to phylogenetic position and distance among the taxa compared, as the studied agamids are not as closely related as the phrynosomatids. These results are in agreement with Comanns et al. (2011), which suggested that the scale morphology, for example, size, shape and regularity, differs significantly between species, families and even different body parts within an individual.
Skin sense organs
Ananjeva et al. (1991) studied the morphological and structural differences in 21 species of agamids and iguanids, 2 of which were rain-harvesters and 5 of which were also studied in the present study. They found that, except for one species, all agamids possess hair-like sensilla. Sherbrooke and Nagle (1996) examined the ultrastructural features in P. modestum and gross morphology in seven Phrynosoma species, three of which were rain-harvesters. They did not find the hair-like sensilla in any of the species studied and discovered that in majority of species examined were the skin sense organs (SSOs) confined to the dorsal scales. Notably, neither of these studies explored a direct link between the species and rain-harvesting behaviour. All the species examined in this study bear SSO (except for P. orbiculare) regardless of the family they belong to or rain-harvesting behaviour. As Sherbrooke and Nagle (1996) hypothesized that one of the SSO's functions in Phrynosoma might be detection of the rain droplets, we looked for a link between the number and shape of SSOs and rain-harvesting behaviour. However, our results showed no correlation for these features, which is in agreement with Ananjeva et al. (1991) and Sherbrooke and Nagle (1996), suggesting that they may be indicative of phylogenetic relationship.
Micro-ornamentation
The present study revealed that the large hexagonal micro-ornamentations (SSM1) on scale surfaces are adorned with smaller ornaments (SSM2) in some species, and in addition to these, two scale hinge regions bear pit-shaped microstructures (SSM2H). Statistical analysis of the size of these three microstructures showed significant differences between the studied agamid and phrynosomatid species. Mean SSM1 is more than 1.5 times larger in agamids than in Phrynosoma while the mean SSM2H is contrastingly more than 2.5 times larger in Phrynosoma than in agamids. However, there was a significant difference only in terms of SSM2H between the species that showed rain-harvesting behaviour and those that did not. Mean SSM2H is more than 1.5 times larger in non-rain-harvesters/species with unknown behaviour.
SSM1 has been discussed in terms of surface wettability by several authors (Comanns et al., 2011; Gans et al., 1982; Schwenk & Greene, 1987). Pre-wetting of the skin would make it easier and faster for water to be captured from its source, kept on scale surfaces and thus retained to enter the interscalar channels during rain-harvesting behaviour (Comanns et al., 2015, 2016).
However, it is worth noting that there is no consensus in the literature regarding the positive or negative effects of microstructures on scale surface wettability. In the present study, we found no correlation between rain-harvesting behaviour and surface microstructures.
Regarding the SSM2H, previous studies have shown that it is found in both rain-harvesting (e.g. P. cornutum) and non-rain-harvesting (e.g. Callisaurus draconoides) species, and is absent in rain-harvesting M. horridus (Sherbrooke et al., 2007). This is also the case for M. horridus in this study. These findings may initially seem contradictory, but the presence of SSM2H-like structures in other reptiles, such as snakes (see Yenmiş et al., 2022), suggests that understanding their importance may extend beyond rain-harvesting behaviour to wider morphological and phylogenetic considerations.
Channels
In the system consisting of longitudinal and transversal channels, the lengths and widths of the channels are crucial, particularly in terms of the quantity of water held. In the current study, we show that in all rain-harvesting species, longitudinal channels are longer than transversal ones while transversal channels are wider. This architectural pattern aligns with the expected flow direction of water.
All the harvesters have longer longitudinal channels than vertical ones. However, despite the physical expectations, width/length indices of both longitudinal and transversal channels of rain-harvesters showed no significant difference among harvesters and non-harvesters/species with unknown behaviour, both within and between families. Notably, M. horridus significantly differs from members of both families. This lack of a statistically significant result across samples grouped by families and behaviour suggests that the width/length index remains similar in all individuals, regardless of snout–vent length (SVL) and phylogenetic relationship.
Channels are interconnected with scale hinges, which have narrow openings between each two adjacent scales and a wider base. In M. horridus, these channels connect to narrower and smaller sub-channels. We found these sub-channels only in M. horridus, which supports Comanns et al.'s (2017) suggestion that the hierarchy of channels might enhance the movement of water. Additionally, the narrowing of longitudinal channels towards the head may facilitate the initial flow of water (Comanns et al., 2017). Our findings are in agreement with previously reported fact that Moloch has the most efficient capillary system to uplift water against gravity (Bentley & Blumer, 1962; Sherbrooke, 2004; Withers, 1993).
Rain-harvesting behaviour is typically accompanied by a specific posture in most lizards, involving bending the head down and lifting the tail base. However, this stereotypical posture is absent in Moloch (Sherbrooke, 1993), that is, while other species benefit from capillary forces as well as gravity, Moloch does not utilize the same stance advantage. Instead, it displays a different posture with stretched legs and lifted head, which might result from its highly effective interscalar rain-harvesting system (Comanns et al., 2017). Since this behaviour is a passive (energy-free) process, the spreading of the wide channels into sub-channels and the narrowing of the longitudinal channels towards the head may be facilitating the transport of water against gravity while the mouth behaves like a sink and forms a pumping mechanism with jaw opening and closing movements creating a negative pressure. A clearer understanding of the water transfer from the capillary systems of rain-harvesting lizards into the lizard's mouth, facilitating the movement of water in capillary hinge channels, awaits further investigation.
Directional capillary flow within the capillary system appears to be an advanced adaptation of rain-harvesting behaviour. It has been observed that the capillary flow of water is directional in P. cornutum and Ph. horvathi, while water entering the hinge channel system moves symmetrically in M. horridus and P. arabicus (Comanns et al., 2015; Yenmiş et al., 2016). The presence of skin folds along the hinge channel walls (island-like protrusions) may enhance capillary flow which has been suggested in previous studies (Sherbrooke et al., 2007). This may be enhanced by establishing a secondary narrower tube system (Comanns et al., 2011, 2017). These skin folds that widen the surface area of the channels have been shown in the integument of rain-harvesting lizards. However, these skin folds have also been found in non-harvesters and species with imbricated scales in the present study, confirming former results of Yenmiş et al. (2016), (but see Sherbrooke et al., 2007). The ancestral state reconstruction analyses support conclusion that the ancestor of iguanian clade containing the rain-harvesting species probably possessed overlapping scales and channels, which supports the hypothesis that rain-harvesting lizards convergently adapted to arid conditions using similar tools they already possessed. Further research may elucidate the connection between skin fold architecture and directional capillary flow, shedding light on its significance in rain-harvesting lizards' drinking mechanisms.
Conclusion
In arid habitats with limited access to water, ability to collect water from skin and channel it to the mouth represents a crucial adaptation for animals to survive in such environment. Phylogenetically distant groups of Squamata distributed across the globe have independently evolved similar features to facilitate the direct capture of rainwater between their scales for drinking (e.g. Sherbrooke et al., 2007) involving existing scale structures.
In this study, we compared scale surface microstructures of three categories: large hexagons (SSM1), smaller nested hexagons (SSM2) and hinge pits (SSM2H) in relation to rain-harvesting behaviour. We found that of the three types of scale surface microstructures SSM1 and SSM2 are widespread and seemingly unrelated to rain-harvesting adaptation, while SSM2H is exclusively found in rain-harvesters. We also confirmed a special position of the iconic rain-harvesting agamid species Moloch horridus, which exhibits non-directional water flow. We have identified a hierarchical transport system comprising sub-channels in addition to the main channels, which appear to enhance its capillary forces in moving water within the scale hinges. The combination of this hierarchical arrangement with channels that narrow towards the head provides insights into why Moloch possesses the most powerful water uplift system among rain-harvesting lizards. To date, only Phrynosoma cornutum (Phrynosomatidae) and Phrynocephalus horvathi (Agamidae) exhibit a capillary directional transport system.
We eagerly anticipate future research that can expand our understanding of rain-harvesting across a broader phylogenetic spectrum than is currently available to us. Various aspects of these lizard's adaptations have broadened our understanding of the role of channel surfaces' architecture in an unanticipated understanding of directional flow forces in capillary action (Comanns, 2018; Comanns et al., 2014, 2015). The scale surface microarchitecture has already given insights for humans designing systems to extract water from fog and flowing air (Aliabadi et al., 2018; Gejdos et al., 2017; Wan et al., 2021). Further study of rain-harvesting in diverse lizards could not only bring understanding of its evolution in various groups but also a deeper knowledge of fluid flow dynamics in tiny biologically created channels and across diverse surfaces.
Acknowledgements
This study was financed by The Scientific and Technological Research Council of Turkey (TÜBİTAK) (Grant number: 118Z022). We thank the Ege University Central Research Test and Analysis Laboratory and Dr. Pınar Sözer Bahadır for the help with SEM analyses; Dr. Jirí Moravec, Curator of Herpetology at Prague National Museum, for help in obtaining some of the skin samples; JKU Institute of Biomedical Mechatronics for the infrastructural support; Dr. Yusuf Bayrakcı, Dr. Gerda Buchberger and Melisa Avcı for proof reading; MSc. Christoph Wolf for the help in obtaining data for Ta. sexlineatus. We thank the anonymous reviewers and the editors for their valuable time and comments.
Author contributions
MY, MV, WCS and DA conceived the ideas and designed methodology; MY, MV, WCS and DA collected the data; MY analysed the data; MY, MV, WCS and DA led the writing of the manuscript. All authors contributed critically to the drafts and gave final approval for publication.
