Mechanical properties of the venomous spines of Pterois volitans and morphology among lionfish species

Spines are multi-functional biological materials found in nature that can greatly benefit organisms in terms of gripping, injection, damage and defense (Anderson, 2018). For example, cacti use spiny modified leaves that prevent water loss in their dry desert habitat and protect against herbivores (Koch et al., 2009). Hedgehogs use their quills for protection against predators and the quills absorb energy during impact from high falls (Vincent and Owers, 1986). Stonefish have a lachrymal saber that is an elongation of an anterior spine, which they are able to rotate into a locked lateral position possibly for defense (Smith et al., 2018). In addition, triggerfish have a modified anterior dorsal fin spine that has several purposes including self-defense, anchoring into crevices in the coral reef when sleeping and providing protection against a strong ocean surge or waves (Cleveland and Lavalli, 2010).

Similar to differences in anatomy, spine material varies, and affects the overall mechanics. Both lionfish and stingray spines are made of mineralized collagen, a combination of hydroxyapatite and collagen (Halstead and Modglin, 1950; Halstead et al., 1955). However, the mechanical properties of lionfish and stingray spines remain unknown. Spines in porcupines, hedgehogs and echidnas are made of keratin (Vincent and Owers, 1986) and have Young's moduli (E) ranging from 5.56 GPa in porcupines to 11.56 GPa in hedgehogs (McKittrick et al., 2012; Vincent and Owers, 1986). Biomechanical properties have only been examined for stingers (bees, wasps and scorpions), where venom is delivered through the middle of the spine (Zhao et al., 2015; Zhao et al., 2016). Lionfish spines, similar to those of stingrays, have venom glands and grooves that line the sides of the spine, whereas in bees, wasps and scorpions, venom flows through the middle (Halstead and Modglin, 1950; Halstead et al., 1955). Venom delivery morphologies in combination with material composition may affect the properties of the spine under various loading regimes.

In several organisms, mechanical properties vary along the length of the structure. In wasp stingers, the elastic modulus and hardness decrease along the length from the base to the tip (Das et al., 2018). In contrast, Young's modulus increases towards the tip of owl feather shafts (Bachmann et al., 2012). The tapered morphology of porcupine fish spines changes the location of maximum stress to the distal end (tip) of the spine, but does not change spine stiffness or toughness (Su et al., 2017). By focusing spine damage toward the distal ends, porcupine fish may conserve the energy required for regrowth.

The red lionfish, Pterois volitans, has 13 dorsal fin spines, three anal fin spines and one spine on each pelvic fin (Fig. 1A). In cross-section, P. volitans spines are solid and have a tri-lobed morphology, thought to be exclusive to lionfish (Halstead et al., 1955; Fig. 2A). This tri-lobed shape is formed by a pair of lateral grooves along the outer two-thirds of the length and these grooves contain glandular tissue that houses venom (Fig. 2B). Both the spines and the glandular tissue are covered by a thin membrane, which ruptures when the spines penetrate an object, releasing the venom (Halstead et al., 1955). The length of the refractory period between venom delivery events and whether the presence of venom in the lateral grooves affects the mechanical properties remain unclear.

The tri-lobed cross-section of the lionfish spine is reminiscent of I-beams used in building design and construction. Engineering beam theory demonstrates that the I-beam shape is able to carry both bending and shearing loads because most of the material is distributed away from the neutral axis. As a result, I-beams have a high second moment of area and span-to-depth ratio, meaning that this shape effectively resists bending (Vogel, 2013). The lionfish spine tri-lobed cross-section also has a large portion of the material located away from the center of the structure (Fig. 2A).

The goal of this study was to investigate the mechanical (bending) properties of the venomous spines of the red lionfish, P. volitans. Our specific goals were to: (1) determine Young's modulus (material stiffness; E), elastic energy storage (ability to absorb energy and restore the sample to its original shape), second moment of area (shape characterization; I) and flexural stiffness (bending resistance; EI) of dorsal, pelvic and anal spines; (2) determine whether mechanical properties differ along the length of the spines; and (3) compare second moment of area among seven lionfish species to determine spine structural variation. First, we hypothesized that mechanical properties differ among dorsal, pelvic and anal spines as a result of differences in morphology (Fig. 1B). Specifically, we hypothesized that pelvic and anal spines will be stiffer and can absorb more energy than the long, thin dorsal spines. Second, we hypothesized that mechanical properties differ along the length of the lionfish spines because of the tapering morphology, which results in venom grooves being absent at the spine tips. We predicted that the base of the spines will be stiffer and bend less, reducing the likelihood of damage at this location and preventing the energetic cost of regrowing an entire spine. Lastly, we hypothesized that the second moment of area will differ among Pterois species, indicating a difference in cross-sectional shape that may be due to varying ecological constraints of native habitats.

Adult Pterois volitans (Linnaeus 1758) specimens (n=6 females, 4 males) were obtained dead from local fishermen and lionfish tournaments on the Western Atlantic coast of Florida, USA. Lionfish are invasive and no permit is required for fishing this species in the state of Florida (Florida Wildlife Commission). For each specimen (total length, TL=188–370 mm; N=30 spines), the fourth dorsal, left pelvic and third anal spines had respective spine lengths in the range 45–90 mm, 24–39 mm and 16.5–37 mm. Spines were extracted from their proximal attachments on the body. The fourth dorsal spine is one of the longest, and was always intact in specimens collected and used in this study. The third anal spine is also the longest on that fin, and there is only one pelvic spine on each fin.

The proximal base of freshly dissected spines was potted in approximately 5 mm of marine epoxy, which was determined to be the smallest amount of epoxy necessary to secure the spines (Loctite, Westlake, OH, USA). We standardized the preparation process by measuring spine length before and after potting in epoxy. Point loads were applied and measurements were obtained at 60% and 0% (tip) of the unpotted spine length. Previous work has shown that embedding stiff materials in epoxy does not significantly alter the Young's modulus (Hoffler et al., 2005; Zysset, 2009). Potted spines were left to set in epoxy and cure for 48 h. To assess the impact of curing time on spine mass and shape, we conducted a pilot experiment on a separate subset of dorsal spines. We found that wet mass did not change for 14 days (Table S1). After 14 days, spine elastic energy storage remained the same, but stiffness increased. We also found there was no significant difference in second moment of area after 14 days (Table S2). As mass and shape remained the same for the first 13 days (Tables S1 and S2), we assumed that the spine mechanical properties presented here were not impacted by the 48 h curing time.

A clamp on an adjustable steel stand secured potted spines at a 90 deg angle to the stand. Point loads were applied at 60% and 0% of the spine length (Fig. 2C) to examine the effects of tapering on mechanical properties (Young's modulus, elastic energy storage, second moment of area and flexural stiffness). A tapered cantilever bending test was modified for an irregular cross-sectional shape on an Instron E1000 with a 250 N load cell (Su et al., 2017). A dissection pin was used to apply the point load along the spine at a displacement rate of 0.3 mm s⁻¹, and all spines were deflected to 10% of their total unpotted length. A lateral point load (Fig. 2A) was chosen for the following two reasons: (1) we were able to place the dissection pin in a single position within the lateral groove and (2) we could assess spine loading simulating a predator attacking from the side. Common displacement speeds for hard biomaterials range dramatically from 0.003 to 400 mm s⁻¹ (Halstead et al., 1955; Galloway et al., 2016; Su et al., 2017; Summarell et al., 2015; Whitenack and Motta, 2010). We decided to load the mineralized collagenous lionfish spines at a slower speed within this range because there are no previous studies on their mechanical behavior, and testing at faster speeds may have resulted in fracture before we were able to test at other point load locations. A previous study on the compressive modulus of keratinous horse hooves compared speeds from 0.167 to 1.667 mm s⁻¹ and found no significant effect between displacement rate and compressive properties (Landeau et al., 1983).

We obtained specimens from The Smithsonian (Washington, DC, USA) to collect comparative spine second moment of area in the anteroposterior orientation (I, mm⁴) from six additional lionfish species [Pterois andover G. R. Allen and Erdmann 2008, Pterois antennata (Bloch 1787), Pterois lunulata Temminck and Schlegel 1843, Pterois radiata G. Cuvier 1829, Pterois russelii E. T. Bennett 1831 and Pterois sphex D. S. Jordan and Evermann 1903]. These six species, combined with P. volitans, allowed for the investigation of seven out of 12 recognized species in the lionfish genus (FishBase; USNM Fish Catalog). At University of Washington's Friday Harbor Laboratories, fish were wrapped in cheesecloth and cling wrap, and scanned in a Bruker Skyscanner 1173 (Kontich, Belgium) at 70 kVp (kilovoltage potential), 114 mA (x-ray intensity), 35.7 μm slice resolution and 2K resolution for larger species and a lower resolution for smaller species. Multiple fish were scanned in a single canister, producing a DICOM file with data for several species. Individual fish were then digitally dissected (segmented) into separate files and reconstructed using Bruker DataViewer. We used Horos software (Horosproject.org; sponsored by Nimble Co LLC d/b/a Purview in Annapolis, MD, USA) to digitally segment spines and each spine was saved as a DICOM file. Dorsal spines ranging from the fourth to sixth spine (dependent on which spines were not damaged from museum specimens) were digitally dissected to obtain cross-sections in the anteroposterior orientation, and I was measured using BoneJ at 60% and 30% of spine length. We were unable to obtain the cross-section of spines at 0%. Lionfish spines narrow at the distal tip and it was not possible to obtain a cross-section at that location using Horos software.

We evaluated morphological relationships among spine length, fish total length and spine region using a two-way ANOVA, where fish total length and spine region were main effects. To evaluate data for goals 1 and 2, we used two-way ANOVA to examine mechanical properties: Young's modulus (E), elastic energy storage, second moment of area (I) and flexural stiffness (EI). Spine region (dorsal, anal and pelvic) and testing location (proximal and distal) were main effects, and we examined their interaction term. Fish total length was also included in the models as a covariate. If a main effect was significant, we used pairwise Tukey tests to evaluate those differences. If the interaction term was significant for a mechanical property, we present those data in the respective figure. If the interaction term was not significant, we present data for only significant main effects. All statistics were analyzed using JMP (SAS Institute Inc., Cary, NC, USA).

Using a separate subset of spines in a pilot study, we found that second moment of area (I) did not differ between hydrated and dehydrated spines (Table S2). Data for dehydrated spines are presented here because obtaining the cross-sections for second moment of area calculations was destructive and needed to be completed after all mechanical testing. Based on these results, we present I as a hydrated spine property for goals 1 and 2 above. In addition, we treated flexural stiffness (EI) as a hydrated spine property when we multiplied hydrated Young's modulus (E) with dehydrated second moment of area (I).

Spine length increased with fish total length (P<0.0001; n=10 fish; N=30 spines, 10 spines from each region); as fish got larger the spines got longer. Based on P. volitans used in this study, as fish increased from 188 to 370 mm, dorsal spine length doubled. Over the same range of fish sizes, anal spines increased in length 66% and pelvic spines increased 40%. Dorsal spines were significantly longer than pelvic and anal spines (P<0.0001).

A two-way ANOVA showed that elastic energy storage varied significantly (F_6,59=13.85, P<0.0001) among spine regions (P<0.0001) and testing locations (P=0.012), and increased with increasing fish total length (P<0.0001). The interaction between spine region and testing location was not significant (F_1,54=0.6591, P=0.52). The anal and pelvic spines can store about 85% more energy than the dorsal spines (Fig. 3A). The proximal end of the spine (base) can absorb about 42% more energy than the distal (tip) end (Fig. 3B). These data support our hypothesis that mechanical properties vary along the length of the tapered lionfish spines.

A two-way ANOVA showed that Young's modulus (E) varied significantly (F_6,59=78.56, P<0.0001) among spine regions and testing locations (P<0.0001), and increased with increasing fish total length (P<0.0001). The interaction between spine region and testing location was significant (P=0.0057). The anal and pelvic spines at the proximal end (base) were stiffer than the pelvic and anal spines at the distal end (tip) (Fig. 4A). The dorsal spines were significantly less stiff at both testing locations in comparison to the pelvic and anal spines (Fig. 4A). These data align with our hypothesis predicting that the base of the spines will be stiffer, which may reduce the likelihood of damage towards the base preventing the energetic cost of regrowing an entire new spine.

A two-way ANOVA showed that second moment of area (I) varied significantly (F_6,59=176.58, P<0.0001) among spine regions and testing locations (P<0.0001), and second moment of area increased with increasing fish total length (P<0.0001). The interaction between spine region and testing location was significant (P=0.0003). The distal ends, or tips, of lionfish spines had significantly lower I compared with the base of all spines (Fig. 4B). This aligns with our hypothesis that the spines are tapered and there are no venom grooves at the tip of the spines, and I becomes smaller towards the tip. The anal spines at the proximal region (base) had the highest I (Fig. 4B), and the anal spine at the proximal region is the stiffest (Fig. 4A).

A two-way ANOVA showed that flexural stiffness (EI) varied significantly among spine regions and testing locations (F_6,59=49.50, P<0.0001), and increased with increasing fish total length (P<0.0001). The interaction between spine region and testing location was significant (P<0.0001). The anal and pelvic spines at the proximal end (base) had a higher EI (Fig. 4C), which indicates they are more resistant to bending than the anal and pelvic distal ends, and the dorsal spines. These data align with our hypothesis that the base of the anal and pelvic spines can better resist bending, and will incur less damage than the tips of the spines. Overall, the dorsal spines do not resist bending as well, do not absorb a high amount of elastic energy and are not as stiff (Fig. 4A–C).

Second moment of area (I) of dorsal spines ranged from 0.0029 to 0.9267 mm⁴ (Table 1). There are no statistics for these comparative data (n=1 from each species). Pterois russelii had the highest I in comparison to the other species, suggesting that P. russelii dorsal spines are stiffer and more resistant to bending. Pterois radiata had the lowest I in comparison to the other species, suggesting that P. radiata dorsal spines are the least stiff and are less resistant to bending. Interestingly, P. russelii and P. radiata overlap in some native ranges, but P. radiata exhibits a broader native habitat range. Pterois sphex had a relatively high second moment of area compared with the other species examined here, and is only native to the Hawaiian Islands (Table 1).

Mechanical properties of P. volitans spines differed among spine regions and testing locations. Pelvic and anal spines were stiffer, could store more elastic energy and could resist bending more than dorsal spines (Figs 3A, 4A,C). Dorsal spines did not have a high resistance to bending and were not stiff structures, and we would expect them to be able to bend substantially under large lateral loads (Fig. 4A,C). Proximal regions of all spines were stiffer, could store more elastic energy and could resist bending more effectively than the distal ends (Figs 3B, 4A,C). Second moment of area (I) differed among lionfish species, and there may be relationships among spine morphology, second moment of area and native versus invasive locations (Fig. 4B and Table 1). For example, in P. volitans, the pelvic and anal spines had similar lengths, but their shape (as measured by second moment of area) was significantly different in the proximal region. Overall, spine length and mechanical properties increased with fish total length.

Lionfish spines, composed of mineralized collagen, ranged in Young's modulus from 0.11 to 10.88 GPa (Fig. 4A; Halstead et al., 1955). Lionfish spine stiffness falls within the range of other bony fish structures such as porcupine fish spines (6.8–20.5 GPa), which are made of nanocrystalline hydroxyapatite, collagen and water (Su et al., 2017). Teleost rib material stiffness (tilapia: 4.1–11.0 GPa, and carp: 3.6–14.5 GPa) is also similar to that of lionfish spines (Cohen et al., 2012; Horton and Summers, 2009). As lionfish spine stiffness is comparable to bone stiffness, we hypothesize that they are also heavily mineralized. Flexural stiffness of lionfish spines ranged from 2.28 to 8257 N mm² (Fig. 4C), which is much higher than the mineralized cartilage of batoid propterygia (33.74–180.16 N mm²; Macesic and Summers, 2012). These data suggest that the mineralized collagen spines of lionfish resist bending better than skeletal elements in cartilaginous fishes and are functioning similar to bony fish skeletal elements. In addition to being able to resist bending effectively (EI), lionfish spines can store a high amount of energy (particularly the pelvic and anal spines) (Fig. 3). We hypothesize that the combination of high bending resistance and elastic energy storage (especially in the pelvic and anal spines) is important for puncturing predators and prey, and reducing damage, which has potential energetic savings of avoiding spine regrowth. Future studies could examine puncture mechanics, the amount of mineralization in lionfish spines and detailed histology to empirically determine tissue type.

It has been suggested that the lionfish dorsal spines are important for defense in terms of predator avoidance and prey capture, because of their location on the dorsal surface, their length and the number (13) of spines. The dorsal spines are significantly longer and probably contain more venomous glandular tissue than the pelvic and anal spines. Dorsal spines are less stiff and can absorb less energy than the pelvic and anal spines (Figs 3 and 4A), which suggests that they may be less effective at puncturing and prone to breaking. A broken dorsal spine may not be detrimental to the fish, as they would have 12 other spines in the same location. Based on our results, we hypothesize that the dorsal spines primarily serve as an intimidation strategy, making the fish look larger to predators, rather than being strictly a defense apparatus.

The many native habitats of Pterois species, in combination with differing spine morphology, suggest a potential variation in eco-morphological pressures. Our data from seven lionfish species show that second moment of area (I) of dorsal spines does not relate to the body size range. For example, P. sphex was the smallest species examined here (22 cm TL; Randall, 1985) but had the second highest I (Table 1). In contrast, P. lunulata can reach up to 35 cm TL (Kuiter and Tonozuka, 2001) and had a low I (Table 1). The maximum size of P. volitans is documented at 38 cm and it had an intermediate I (Table 1). However, these comparative data suggest that spine morphology may vary with habitat. Pterois radiata inhabits shallower rocky crevices and reefs and has a low spine second moment of area, while P. russelii dwells on muddy/brackish substrates in shallow and offshore depths and has a high spine second moment of area (Kuiter and Tonozuka, 2001). Further relationships investigating eco-morphology are needed on lionfish in their native habitats – information that is currently limited.

Lionfish, specifically P. volitans, have become a successful invasive predator for various reasons. First, lionfish can survive without food for extended periods of time, although they have been shown to have a very healthy appetite and are generalist feeders (Albins and Hixon, 2008). In a starvation study, lionfish lost only 5–16% of their body mass over a period of 3 months without food (Fishelson, 1997). In addition, lionfish have specialized bilateral swim bladder muscles that provide pitch control for prey strike behavior (Hornstra et al., 2003). This novel buoyancy adaptation may aid in stability in the water column as well. Lionfish also have an extremely high reproductive rate of two million eggs per year and an early maturity size of 4–6 inches (Morris and Whitfield, 2009). Finally, they have bacterial communities in skin mucus that provide disease resistance, a novel trait that may have facilitated their successful invasion in the Western Atlantic and Caribbean (Stevens et al., 2016). These characteristics combined with 18 venomous spines make lionfish an aggressive and intimidating marine invader, although we show here that the numerous dorsal spines are not stiff, energy-absorbing structures. As a result, not all of the venomous spines of P. volitans may be effective at defense.

All lionfish spines that we examined lacked grooves at the tips and had tapered structures; second moment of area (I) decreased towards the tips of the spines for all species, similar to P. volitans (Table 1). As we suspect that the spines of all lionfish species are composed of the same material (mineralized collagen), the material stiffness or Young's modulus (E) should remain constant, and I would be the critical variable determining the spine bending properties among species. We predict that flexural stiffness (EI, bending resistance) decreases towards the tips of the spines for all species of lionfish, as it does in P. volitans investigated in this study. The shape of lionfish spines among species is also informational for groove size and possibly glandular tissue and venom quantities. For structures with venom or poison, it is important to take into account the biological structure as a whole (the material, shape, and venom or poison). To better understand the relationships between mechanical properties and spine shape, future studies could focus on these properties over a range of fish and spine sizes.

Several aquaria have observed that lionfish can regrow their spines, particularly damaged dorsal spines. There are currently no data discussing regrowth rate and whether spine regrowth always occurs. Regrowth of mineralized collagen spines may be influenced by water pH or temperature, which can vary among aquaria. From dissections for this study, we documented instances where regrowth may have occurred in approximately 10% of dorsal spines (Fig. 5). In these cases, regrowth always occurred between 30% and 60% of total spine length, suggesting this is the area of the spine where fracture is likely to occur. These predictions are supported by our data, which found that testing location was a significant effect impacting mechanical properties. We argue that the location of spine loading is important because of possible tapering or change in shape of biological materials.

Biomimetic designs draw inspiration from nature to engineer devices for human use. Fish scales and shark skin have gained attention in recent years for applications such as impact-resistant armor and self-cleaning materials (Nishimoto and Bhushan, 2013; Naleway et al., 2016; Zhu et al., 2012). Here, we present experimental data on lionfish spines and quantify their variations in shape, length and mechanical properties. Many researchers studying spine mechanics have argued that spines can be used for biomedical inspiration for devices such as hypodermic needles (Bai et al., 2015; Cho et al., 2012; Sahlabadi et al., 2017). Lionfish spines are not hollow, serrated or barbed, and they instead deliver venom from grooves along the sides (Fig. 2B). The lionfish spine design may be useful in creating reusable syringe needles and plungers that can be sterilized, which would decrease biomedical waste and sharps disposal costs. We hypothesize that the second moment of area and tapering of lionfish spines will reduce puncture forces. Further collaborations among biologists, medical doctors and engineers would be beneficial for such bioinspired applications.

We thank REEF and FWC for lionfish specimens. We thank Kent Wallace for Instron support. We thank Dr Justin Grubich (Field Museum) for lionfish expertise, commentary on experimental design, and thoughtful conversation. We thank the Florida Atlantic Biomechanics lab members for support and comments. We thank Dr Adam Summers and the Friday Harbor Laboratories at University of Washington for access to the micro-CT scanner. We thank Danielle Ingle for scanning specimens and the Smithsonian National Museum of Natural History for various lionfish species. We thank Breanna Nelson for help with data collection. We thank the JEB reviewers and editors for their thoughtful critiques of our manuscript.