11 Crawford and Peterman 2013). Clinging to and climbing up misted and wetted surfaces is likely 12 to be vital to lungless salamanders traversing their natural habitat. 13 We see varying effects of moisture on attachment in other taxa of climbing reptiles and 14 amphibians, depending on the specific mechanism used and the substrate properties (Persson Stark et al. 2012Stark et al. , 2013Stark et al. , 2015)). In these cases, as in salamanders, attachment mechanisms 5 may be tuned to the roughness and wetness of their preferred habitat or represent a multi-faceted 6 approach that operates under diverse conditions.

Here we examine the effects of surface roughness and wetness on the ability of plethodontid 8 salamanders to cling at a range of substrate inclinations (i.e., cling angles). We investigate how 9 an increase in surface roughness affects species' maximum cling angle. In addition, we test how 10 misted and flowing water affect attachment on smooth and roughened surfaces. We predict that 11 roughened surfaces will allow species to engage in gripping behavior that will expand the range 12 of angles on which clinging is possible. We also predict that species may show enhanced 13 performance in wetter conditions that better match their microhabitat, due to tuning of the 14 biological adhesive for high performance in their natural environment, but that flowing water 15 could decrease cling performance by creating drag or disrupting the salamander's mucus coating. 9 and a depth of 1 -2 mm on the surface (Fig. S1).

10

For the misted treatment, salamanders were placed on a pre-misted substrate, such that 11 the droplets were between the salamander mucus layer and the substrate. For the flowing water 3 trials from high performing species. Analyses of significant differences were conducted between 4 select treatments (Roughness S to I2, I2 to R2, and S to R2, as well as dry condition to misted 5 condition and dry condition to wetted condition for all three roughness pairings) within species 6 using pairwise, two-sample, two-tailed Wilcoxon Signed Rank tests for non-parametric data and 7 corrected for false discovery rate (Benjamini and Hochberg, 1995) (Table S4 andS5).

8 Treatments were selected because cling angles on roughness levels I1 and R1 were not 9 significantly different from I2 and R2, respectively. Due to the non-parametric testing used, and 10 the non-linear response of cling performance to increasing roughness, phylogenetic comparative 11 methods and a comparative analysis of cling performance across species were not conducted.

12 Instead all analyses conducted compared performance within a species across treatment 13 conditions. Data from E. guttolineata were excluded from analyses due to limited numbers of 14 individuals (N = 2).

15

16 Roughness 17 Individuals from all 16 species tested were able to cling to dry surfaces across the full 18 range of roughness treatments (asperity size 0 -4000 µm) at maximum angles ranging from 76 ± 19 2º (mean and s.e.m.) to 180 ± 0º degrees from the horizontal (Table S1). On the smooth and 20 intermediate roughness plates (S, I1, I2) salamanders were unable to engage in gripping with the 21 toes and apparently attached via their mucus layer. In some cases, large salamanders (A. gracile, 1 attaches them to the substrate. At higher roughness (R1 and R2) some salamanders were 2 observed augmenting their mucus attachment with gripping behaviors with the feet, toes, or tail, 3 inserting them into gaps between asperities or interlocking with projections.

The relationship between maximum cling angle and roughness was non-linear and the 5 effect of roughness on maximum cling angle varied among species in magnitude but not 6 direction (Fig. 1). Phylogenetic comparative analyses of performance across species were not 7 conducted due to issues of normality, heteroscedascity, and non-linearity, but differences in 8 maximum cling angle within species across treatments were analyzed, and when the findings for 9 each species were considered, certain trends emerged. Small species (body mass 0.6 g to over 5

10 g) clung better to smooth surfaces than large species (5g to 29.0 g, Table S2, Fig. 1). Overall, the 11 effect of increasing roughness of the substrate resulted in significantly decreased cling angle 12 (Fig. 1, Table S2) in nine of the 16 species when moving from smooth substrates to intermediate 13 substrates, with significantly improved maximum cling angle from intermediate to highly 14 roughened substrates in 12 of 16 species.

15 Species differed in whether maximal cling angle occurred on the rough (R2) or on the 16 smooth (S) substrate (Fig. 1, Table S2) and in eight of 16 species, maximum cling angle on 17 smooth (S) and highly rough (R2) substrates did not significantly differ (Table S4). For nine of 18 the 16 species, the highest cling angle was achieved on smooth substrates, but large-bodied or 19 dexterous species (A. gracile, A. maculatum, A. flavipunctatus, A. lugubris, A. vagrans, and E.

20 eschsholtzii) clung significantly better to the roughest substrates (Fig. 1, Table S2, S4).

21 Wetness 22 All 14 species tested under wetness conditions were able to cling to dry, misted, and 23 wetted roughness plates (S, I2, R2) at angles from 47º ± 6º to 180º ± 0º (Table S3, Fig. 234). In 2 engage in mucus attachment with their ventral surface pressed against the surface, but with their 3 feet not always in contact with the surface (Fig. S1). On the roughest substrate, we observed 4 interlocking the feet and toes with projections on the surface, and in some cases, the ventral trunk 5 was completely off the surface (Fig. S1). Water negatively impacted maximum cling angle on 6 smooth surfaces in eight of 14 species, although seven of 14 species were more impacted by 7 flowing water, and only four of 14 species were significantly impacted by misted water (Fig. 2).

8

Water significantly improved cling angle in six of 14 species on the intermediate (I2)

9 substrate (Fig. 3, Table S3), although the magnitude of the effect varied by species (Table S5).

10 Cling angle on the roughest plate (R2) was not significantly affected by any wetness treatment 11 for 12 of the 14 species, except D. aeneus which clung better to flowing-water-covered surfaces 12 and A. maculatum which clung poorly to misted roughened surfaces (Fig. 4, Table S5). S2, S4, Fig. 1). In addition to being 9 the largest species, these species also have some of the largest feet. The size of the grit elements 10 relative to foot and toe size enable these species to engage in interlocking toes in crevices to 11 enhance attachment (as in small species), but potentially also to use the entire foot to create 12 attachment by antagonistic clasping of the toes around large granules (Cartmill, 1985).

Extremely small salamanders, such as B. attenuatus, E. wilderae, and D. aeneus showed 14 an adhesive contact attachment on smooth (S) substrates where frequently their feet were not in 15 contact with the surface during clinging. They were also able to engage in interlocking 16 attachment on rough (R1 and R2) substrates to support their body weight (Fig. S1). Despite their 17 smaller body size, feet, and toes, they did not show an ability to engage in interlocking at smaller 18 levels of roughness (I1 and I2). Even with miniatured and reduced limbs, as in B. attenuatus, all 19 tested species could fully support their weight on rough substrates when hanging from only the 20 feet (Fig. S1). Broadly, both a reduction in cling angle on substrates of intermediate (I1)

21 roughness and the ability to engage in interlocking attachment on substrates of sufficient 3 smooth, dry substrates, likely due to water filling in and bridging gaps between the body surface 4 and the surface. For large species, misted water negatively affects performance on smooth 5 substrates, and species are more likely to fail by sliding at angles between 0º and 90º. Flowing 6 water, despite exerting larger drag forces than the misted condition, has a similar negative effect 7 on smooth substrates to the misted condition, and has no effect on rough substrates, where 8 interlocking is the major mechanism of attachment. 9

The robust performance of salamanders on many rough and wetted surfaces suggests that, 10 in a range of moisture conditions, attachment on natural surfaces will be unaffected. Further 11 research into the effect of substrate on clinging and climbing performance is needed to fully 12 understand the mechanisms of attachment used by salamanders and the effect of naturally 13 occurring degrees of roughness, wetness, and fouling. These data represent an effort to broadly S4). S3) on smooth substrates (S, 0 µm) under three wetness conditions (dry (D), misted (M), flowing water (W)). Error bars indicate s.e.m., brackets denote pairwise comparisons between dry and misted, and between dry and wetted conditions, asterisks denote significant differences (P < 0.05, Table S5). S3) on intermediate substrates (I2, 300 -355 µm) under three wetness conditions (dry (D), misted (M), flowing water (W)). Error bars indicate s.e.m., brackets denote pairwise comparisons between dry and misted, and between dry and wetted conditions, asterisks denote significant differences (P < 0.05, Table S5). S3) on rough substrates (R2, 2000 -4000 µm) under three wetness conditions (dry (D), misted (M), flowing water (W)). Error bars indicate s.e.m., brackets denote pairwise comparisons between dry and misted, and between dry and wetted conditions, asterisks denote significant differences (P < 0.05, Table S5).