Shell crushing resistance of alien and native thiarid gastropods to predatory crabs in South Africa

The successful invasion of freshwater and coastal lakes of South Africa by the recently introduced thiarid snail Tarebia granifera may be due in part to release from predatory pressure. This study aimed to determine the comparative vulnerability of T. granifera and the widespread native aquatic thiarid Melanoides tuberculata to predation. These species also account for many thiarid invasions in the Americas, Europe and parts of Africa. We quantified the shell crushing resistance of these snails, as well as the maximal shell crushing capability of native freshwater crab predators, Potamonautes sidneyi and P. perlatus. Using an Instron isometric transducer, we showed that Tarebia granifera shells were significantly stronger than Melanoides shells, and exceeded the crushing strength we documented for both potential predatory crabs. The greater shell strength of Tarebia granifera was due to shape, sculpture and thickness characteristics. Shell strength of Melanoides, however, remained within the range of crushing strength of their potential predators. Assuming crushing to be the main form of crab predation on snails, we inferred T. granifera to be less vulnerable to durophagous attack and that their population growth is thus not limited by predation pressure.


Introduction
Interspecific interactions, such as predation, can affect the invasion success of alien species (see MacNeil et al. 2013).The enemy release hypothesis (i.e., escape from predators, parasites and pathogens) is one mechanism often proposed to explain why alien species exhibit higher competitive ability and therefore successfully invade certain habitats (Elton 1958;Tilman 1999).However, native predators can affect the success of invasions by feeding less on the alien species and more on the native competitors (Shinen et al. 2009;Lopez et al. 2010).The practical usefulness of the concept of biotic resistance to invasion has long been criticised and debated in invasion ecology because factors such as propagule pressure and repeated introductions from multiple source populations greatly increase the chance of establishment (Alpert 2006;Heger et al. 2013;Ricciardi et al. 2013).To address how predation may affect the invasion process, the strength of interactions between native predators and alien species, as well as their native competitors and native prey, must be determined.This information will provide evidence for whether predator-prey interactions can control an alien invasive population (Grason and Miner 2012;Macneil et al. 2013).
As species are increasingly introduced to new habitats across the world through human activities, more opportunities arise to study the strength of novel interspecific interactions.Native predators may be able to easily feed on or even prefer alien species, quickly incorporating them into their diet and thus potentially increasing biotic resistance to invasion (Lopez et al. 2010;Barrios-O'Neill et al. 2014a).However, if alien prey are harder to handle, predators may ignore the alien, and continue to focus on native species, which may also have to deal with additional new competition from the invasive (Lopez et al. 2010).For example, if the newly introduced alien species has previously developed defence mechanisms, such as an unusually hard shell, it will remain shielded from predation pressure.Shell crushing predation by crabs has been shown to affect shell morphology of prey gastropods (Brookes and Rochette 2007;Cox 2013;Weigand and Plath 2014) and is considered a driver of snail biodiversity and evolution (Vermeij 1987(Vermeij , 1994;;West and Mitchel 2000;Harper 2006;Covich 2010 and references therein).The co-evolutionary relationship between crab predators and snail prey has been shown to vary in relation to heavier predation pressure leading to the formation of stronger shells (West andCohen 1994, 1996).
The thiarid gastropod Tarebia granifera (Lamarck, 1822) is originally from South-East Asia but has successfully invaded rivers and lakes in North, Central, and South America, the Caribbean, and Africa (Appleton et al. 2009).In the 1990s, T. granifera was accidentally introduced to South Africa and has rapidly invaded eastern parts of the country, including the Kruger National Park and iSimangaliso Wetland Park (Appleton 2003).It has been suggested that T. granifera invasion success is attributed to release from predatory pressure due to its thick shell and operculum (Appleton et al. 2009).However, the predation release of T. granifera in South Africa may be only temporary and due to naïve responses of native crabs to this newly introduced alien species.Crabs may learn to use other methods to extract snails from their shells, such as peeling or hooking through the aperture.Over longer time periods, if the selection pressure is strong enough they might evolve stronger chelae to capitalise on the new resource presented by growing T. granifera populations.
It has been reported that T. granifera tends to outcompete or displace other widespread native gastropods in South Africa, such as Melanoides tuberculata (O.F. Müller, 1774) (Miranda and Perissinotto 2014a, b).However, T. granifera and M. tuberculata are sometimes sympatric and both are considered global invasive species (Pointier et al. 2003;Karatayev et al. 2009;López-López et al. 2009;Work and Mills 2013).Moreover, there is strong evidence that what is referred to as M. tuberculata is in fact a species complex (Genner et al. 2004(Genner et al. , 2007)), with some clades clearly polyphyletic.
This study assessed the comparative vulnerability to native crab predation of alien and a native aquatic snail species by: 1) determining if there are significant differences in shell morphology and crushing resistance between T. granifera and the native thiarid

Species selection and field collections
In 2014, P. sidneyi crabs, M. tuberculata and T. granifera snails were collected from Kosi Bay's Lake Nhlange (26º57′37″S 32º49′'36″E), Lake Sibaya (27º22′11″S 32º42′56″E) and False Bay (27º57′7″S 32º22′37″E) in the St Lucia Estuary, KwaZulu-Natal.Potamonautes perlatus crabs were collected from the vicinity of Port Elizabeth, Eastern Cape (34º2′42″S 25º34′7″E).Potamonautes sidneyi occurs in Swaziland and Mozambique as well as in the Northern Cape, Eastern Cape, Mpumalanga, North-West, Gauteng, Free State, and KwaZulu-Natal provinces of South Africa.This species was also selected because of its common occurrence in freshwater bodies and observed crushing predation potential.The range of P. perlatus is restricted to the Western, Eastern and Northern Cape provinces of South Africa and possibly Namibia (Barnard 1935(Barnard , 1950)).Potamonautes perlatus is morphologically similar to P. sidneyi but attains greater sizes.Melanoides tuberculata was selected because it occurs within the combined range of both native crab species and is morphologically similar and closely related to T. granifera (Figure 1).Melanoides tuberculata originates in subtropical and tropical Africa and southern Asia (Brown 1994;Genner et al. 2004Genner et al. , 2007)).However, its invasion history is complex, as it has been repeatedly introduced across the world (Pointier et al. 2003;Facon et al. 2003Facon et al. , 2005;;2006, 2008;Genner 2004Genner , 2007;;De Kock and Wolmarans 2009;Oscoz et al. 2010;Strayer 2010;Van Bocxlaer et al. 2015).Although we are not in a position to resolve the nomenclatural and taxonomic issues for this group, we are confident that our work is addressing a single taxonomic unit, which is also considered a native morph in South Africa (see Raw et al. 2016; Figure 1).Here we use the name M. tuberculata sensu lato, with the recognition that this may be updated with future research (see also Raw et al. 2013Raw et al. , 2015Raw et al. , 2016;;Appleton and Miranda 2015;Van Bocxlaer et al. 2015).Tarebia granifera has a wide native distribution in South-East Asia and has been introduced in North, Central and South America, as well as throughout the Caribbean islands, and most recently to Africa (Abbott 1958;Pointier et al. 2003;Ben-Ami 2006;Appleton et al. 2009;Karatayev et al. 2009).

Gastropod shell morphometrics
Snail shell height (SH) and width (SW) were measured with Vernier callipers (to the nearest 0.01 mm).The thickness of shells broken in the laboratory was measured at various locations along the height of the mid-dorsal region of the body whorl, using a Nikon SMZ25 microscope with NIS elements measuring software (to the nearest 0.01 mm).The presenceabsence of shell sculpture which could influence shell breaking strength was also recorded.Analysis of covariance (ANCOVA) was used to compare size-standardized (covariate = shell width) differences in shell thickness between species.Untransformed data were used as model residuals conformed to assumptions of normality and equal variances.All statistical analyses were performed using the open source software R, version 2.14.1.

Gastropod shell breaking resistance
Forces were measured with an isometric transducer (type 9023, Kistler Inc., Winterthur, Switzerland) connected to a charge amplifier (type 5058a, Kistler Inc.) and set in a customised rig attached to a servohydraulic testing system (type 8801, Instron, Norwood, U.S.A.) in the laboratory.Live snails were positioned with the aperture down between two steel plates and subjected to increasing downward crushing forces at a constant rate (20 N/sec).The maximum force in newtons (N) resulting in shell failure (i.e., breaking of the body whorl) was recorded (Edgell and Rochette, 2008).Fifty snails of each species, with a similar SH and collected from the same locality (i.e., Lake Nhlange) were used.ANCOVA was used to test for differences in shell breaking resistance between species (covariate = shell width).Data were log transformed to satisfy assumptions of normality and equal variances.

Crab morphometrics
Crab carapace widest width (CWW), and the propodus height (PH) of the crushing (or major) chela were measured with Vernier callipers (to the nearest 0.1 mm).Digital images of the crushing chela were analysed with ImageJ software to estimate relative percentage of occlusion type according to Brown et al. (1979) and also to measure the distance from the dactyl fulcrum to the insertion of the dactyl closer muscle apodeme (L1) and the distance from the fulcrum to the dactyl tip (L2), so that mechanical advantage (MA) could be calculated for each sex (Elner and Campbell 1981).

Crushing chela closing force
Measurements were done in vivo using the Kistler system and a protocol described in detail in other studies (Herrel et al. 1999;Singh et al. 2000;Lailvaux et al. 2009).Crabs were induced to pinch down with their crushing chela on plates set at a gape of 6 mm (based on the average SW of available gastropod prey) and positioned on the proximal region of the chelae.Closing force was measured five times per individual, with a resting period of 20 minutes between measures.Because the objective of this study was to estimate maximal capabilities of crabs, only the maximum closing force measurements (N) of the most cooperative individuals were considered (see Losos et al. 2002).Hence the data from 13 P. sidneyi crabs and 15 P. perlatus crabs are presented.

Prey handling time
To gain some insight into the behaviour and handling ability of crabs exhibiting the highest maximal shell crushing capabilities, a simple post hoc experiment was also conducted.Four large (60 -80 mm CWW) male P. perlatus crabs were individually acclimated to 10 L buckets for a week, kept in a temperature controlled room (25 ± 1°C) with 12:12 photoperiod.Water was changed every 48 h and crabs were fed 20 g (wet weight) of fish muscle tissue daily.Crabs were then starved for 72 h before being presented with 10 live snails of 5 mm SW over a period of 60 minutes whilst being filmed with a small digital camera mounted on the edge of the bucket.Two of the crabs were presented with 10 T. granifera snails as prey and two were presented with 10 M. tuberculata snails (SW of all snails ≈ 5 mm).The footage was analysed and handling times were recorded with a stopwatch.Handling time in a successful attack was recorded as the time from first clasp with the chela to the time the crab consumed the prey entirely.If the attack was unsuccessful (i.e.ending in the rejection of the prey which remained intact or suffered only minor damage to the shell lip or tip), handling time was recorded as the time from first clasp to the time the prey was dropped.The number of attempted attacks and percentage successful attacks was also recorded (Rheinallt and Hughes 1985;Yamada and Boulding 1998).

Gastropod shell morphometrics and breaking resistance
The body whorl shell thickness of M. tuberculata snails used in the study was significantly different from that of T. granifera snails (ANCOVA: F 1,50 = 32.91,P < 0.001; Figure 2a).Tarebia granifera snails had thicker shells, with an average body whorl shell thickness of 0.16 mm ± 0.01 SE, versus 0.11 mm ± 0.003 SE in M. tuberculata.
In terms of shell sculpture, T. granifera shells have conspicuous knobs whereas M. tuberculata shells are smoother.In this study, Melanoides tuberculata shell height to shell width ratio (SH/SW) ranged from 2.17 to 4.71 mm, average 3.09 mm ± 0.05 SE, whereas the shells of T. granifera were comparatively less elongated with SH/SW range from 1.64 to 2.71 mm, average 2.14 mm ± 0.03 SE.The shell crushing resistance of T. granifera was significantly different from that of M. tuberculata (ANCOVA: F 1,50 = 269.53,P< 0.001; Figure 2b).Tarebia granifera shells resisted an average crushing force of 100 N ± 6 SE, while M. tuberculata only resisted an average of 31 N ± 4 SE.

Crab morphometrics, crushing chela closing force and prey handling time
Potamonautes perlatus crabs (47.0 -86.1 mm CWW, average 60.5 mm ± 3.4 SE) exhibited crushing chela maximum closing forces ranging from 18 to 598 N, average 130 N ± 39 SE (Figure 3).Crushing chela maximum closing forces of smaller P. sidneyi crabs (26.8 -49.2 mm CWW, average 37 mm ± 1.9 SE) ranged from 8 to 43 N, average 20 N ± 3 SE (Figure 3).The crushing chelae of P. perlatus and P. sidneyi exhibited similar occlusive geometry (Figure 4), consisting of 32 to 36 % rounded or molariform area, 59 to 62 % asymmetrical occlusive area and ending in a pointed tip.The average mechanical  advantage (MA) of the crushing chela for both species was also similar: 0.26 for females (n = 10) and 0.25 for males (n = 10) in P. perlatus, 0.25 for both females (n = 10) and males (n = 10) in P. sidneyi.Large male P. perlatus crabs were able to completely crush the shells of M. tuberculata and T. granifera in successful attacks.However, P. perlatus had more difficulty in handling T. granifera, often dropping these snails or spending time trying to crack the shell of certain individuals with no success (Table 1).Evidence of crab attack included damage to the shell lip and tip which were often broken off.The overall attack success of large crabs on T. granifera was low despite a slightly elevated attack rate when compared with M. tuberculata (Table 1).

Discussion
Tarebia granifera snails exhibit thick rotund shells with knobs which are significantly more crushresistant than shells of M. tuberculata (Figures 1 and  2).Thick shells can deter crab crushing predation (Trussell 1996;West and Cohen 1996).Furthermore, knobs may spread crushing force over an increased surface area or reduce the muscular leverage of the predator (West et al. 1991;SäLgeback and Savazzi 2009).Similarly, more rotund shells, with a lower SH/SW ratio, may be more crush-resistant (DeWitt et al. 2000).The crushing force resistance estimates in the current study are within the range of those reported by West et al. (1991) for M. tuberculata.Tarebia granifera has been estimated to have diverged from the Melanoides clade around 8.6 Ma in Asia (Genner et al. 2007) where the diversity of freshwater crabs is the largest in the world (Yeo et al. 2008;Shih and Ng 2011).Tarebia granifera has developed thicker shells compared to African M. tuberculata populations.However, thicker shells require additional calcium which may be limiting or difficult to sequester in certain habitats, thus making snails more vulnerable to shell crushing predators.Consistent with this, unusually small or eroded shells have been reported by Miranda et al. (2011) in the St Lucia Estuary and Lake Sibaya.Further studies may address intraspecific differences in crushing resistance in different habitats.
Shell damage, particularly scars, can represent failed crushing or peeling attacks by predators, so their frequency in a population can be used to assess predation pressure (see Edgell and Rochette 2008).No noticeable damage of this kind has been observed in T. granifera shells from natural environments in South Africa (Miranda et al. 2011).Indeed, West et al. (1991) commented that regeneration scars are only rarely found in African thiarids, with the exception of specimens from Lake Tanganyika which are also thought to have been involved in a co-evolutionary arms race with native crabs, resulting in unusually strong shells and armament (West and Cohen 1994).However, scars on shells are likely to be size-specific.Smaller shells are more likely crushed, whereas intermediate sizes may escape if dropped during attempted predation, and the largest individuals may only be vulnerable to attacks from the largest crabs.Furthermore, it is expected that predation pressure from shell-breaking predators such as crabs will be most intense on smaller individuals.
Although they overlap in general distribution and can be found in the same habitat, the two native freshwater crabs used in this study can be inferred as not having a significant impact on T. granifera populations.The current study demonstrates that native crabs have difficulty in overcoming the stronger shell defences of T. granifera.The strength estimates in the current study are in line with those reported by Marijnissen (2007) for P. platynotus (Cunnington, 1907) and the strongest P. perlatus crabs exhibit a crushing chela maximum closing force similar to that of mud crabs (genus Scylla) and lobsters (Elner and Campbell 1981;Yap et al. 2013).Large P. sidneyi crabs are able to crush the weaker shelled M. tuberculata, but they are unlikely to successfully prey on T. granifera.However, even some of the largest and strongest P. perlatus have a much lower attack success rate on T. granifera than on M. tuberculata (Table 1).Strength, chela gape and time limitations may cause hungry crabs to reject prey (Yamada and Boulding 1998).In accordance with optimal foraging theory, even if peeling is employed as an alternative handling technique to overcome stronger shell defence, it may be more cost effective to spend that handling time consuming other prey or food items (Hughes and Seed 1995;Yamada and Boulding 1998).Further noteworthy preliminary observations were made during the current study.After the prey handling experiment, the large P. perlatus crabs were presented with both T. granifera and M. tuberculata in the lab.Interestingly, crabs seemed to attack the closest snail and move on to the next if unsuccessful.No preference or selection for either species was apparent despite the continued high attack success on M. tuberculata.Like P. lirrangensis (Rathbun, 1904) of Lake Malawi (Weigand and Plath 2014), P. sidneyi and P. perlatus could be considered opportunistically carnivorous scavengers.Their occlusive geometry, indicative of a serrate rather than molariform dentition, as well as low mechanical advantage, suggest that they are adapted to an omnivorous diet (Yamada and Boulding 1998).However, there is evidence that T. granifera is part of P. sidneyi diet in Lake Sibaya (Peer et al. 2015).It is likely that M. tuberculata was predated upon by P. sidneyi in the past, but the native snail appears to have either disappeared from that area or its population is currently below detection threshold (Miranda and Perissinotto 2012).Further studies on the ecological interactions that occur in the field between predators and thiarids are needed.Population-level predator impacts should be addressed, as well as local adaptations in different habitats, such as the development of different armament and handling strategies.
The spread of T. granifera, like that of other alien invasive gastropods (Alonso and Castro-Díez 2008), is undoubtedly driven by a combination of factors such as changes in its migration regime as a result of human activity (Facon et al. 2006;Appleton and Miranda 2015), high rate of parthenogenetic reproduction (Facon et al. 2008;Miranda and Perissinotto 2012), ability to dominate benthic invertebrate assemblages in various habitats (Dussart and Pointier 1999;Miranda and Perissinotto 2014a, b), the production of chemical cues which deter other potential competitive snails (Raw et al. 2013(Raw et al. , 2015) ) and escape from native predators.According to Appleton et al. (2009), the strong shell defences of T. granifera may even facilitate endozoochorous dispersal by waterfowl.Although M. tuberculata is native to South Africa, it has a complex global invasion history and alien morphs may be present in South Africa (Genner et al. 2004;Van Bocxlaer et al. 2015).Although they can be found in the same habitat, the interactions between T. granifera and M. tuberculata seem to result in mutual exclusion, where one displaces the other (Pointier et al. 1998;Contreras-Arquieta and Contreras-Balderas 1999;Karatayev et al. 2009;López-López et al. 2009).
Some aspects of the potential for biotic resistance to alien invasion can be quantified in terms of the mechanisms involved in interspecific interactions, e.g., the crushing strength of predators and crushing resistance of prey.This information could be incorporated into a functional response approach addressing the strength of predator-prey interactions involving native and alien species (MacNeil et al. 2013;Barrios-O'Neill et al. 2014b).Further insight into the invasion success and ecological effects of alien invasive species can be gained by continuing to monitor these interactions in different habitats, whilst also taking evolutionary trajectories and relationships into consideration (Sakai et al. 2001;Smith 2004;Shine 2012).

Figure 2 .Figure 3 .
Figure 2. Comparison of a) shell width by body whorl shell thickness allometry, and b) log shell width by log shell crushing resistance, between sympatric populations of M. tuberculata and T. granifera.

Figure 4 .
Figure 4. Occlusive geometry of the crushing chela of female and male P. perlatus and P. sidneyi (total n = 40, n = 10 for each sex) presented as average percentage rounded or molariform area (rm), asymmetrical occlusive area (as) and pointed tip (pt).