Troglocaris

Jugovic, Jure, Prevorčnik, Simona, Aljančič, Gregor & Sket, Boris, 2010, The atyid shrimp (Crustacea: Decapoda: Atyidae) rostrum: phylogeny versus adaptation, taxonomy versus trophic ecology, Journal of Natural History 44 (41 - 42), pp. 2509-2533 : 2526-2530

Cave shrimps of the subgenus Troglocaris ( Atyidae ),

exhibit high variability in rostral length and dentition, the relative rostral length representing the only age-independent and sexually non-dimorphic character. In Troglocaris s. str. the observed pattern of rostral length variability is similar to the pattern of the atyid shrimp Xiphocaris ( Covich et al. 2009) . In Xiphocaris coexisting with predatory fish, rostra were significantly longer and the total shrimp population density was much lower than in Xiphocaris living where fish had no access. Additionally, lower abundances of shrimps with long and short rostra were reported in the localities without and with fish, respectively. Also in four species of the kakadukaridid shrimps, Leptopalaemon gagadjui Bruce and Short, 1992 , and three undescribed species, positive correlation between the length of the rostrum, the number of rostral teeth, and the presence of predatory fish was demonstrated (John Short, personal communication, 2009). Similar conclusions were drawn by some authors reporting analogous defence mechanisms (i.e. elongated spines) in Odonata larvae living with predatory fish ( Johansson and Samuelsson 1994; Arnquist and Johansson 1998; Johansson and Wahlström 2002; Mikolajewski and Johansson 2004; Mikolajewski et al. 2006; Brooks 2008). Johansson and Samuelsson (1994) wrote about lower total density of prey in the case of the co-occurrence with predators. Not only in macroinvertebrates but also in smaller invertebrates (e.g. Rotifera, Cladocera), similar cases of defence mechanisms are known ( Adler and Harvell 1990; Harvell 1990).

Possible causes for the coincidence between long rostrum, increased carapace size and the presence of Proteus in Troglocari s s. str.

An overall positive correlation between the shrimp’s relative rostral length and presence of Proteus , as well as the results of our feeding behaviour observations indicate that the elongation of the rostrum in Troglocaris represents its defence mechanism against the predation of Proteus . Four possible explanations for such an adaptation exist: (1) the shrimps are responding phenotypically to the presence of predatory kairomones in the water with intensified rostrum growth during ontogeny, (2) in polymorphic populations, the selection favours alleles for a longer rostrum in shrimps occurring with Proteus , (3) in polymorphic populations, selection acts against a welldeveloped rostrum where it has a cost but provides no advantage, (4) the elongated rostrum is a plesiomorphy, the reduction is caused by “mutational pressure”, where the selection does not counteract the “short rostrum” mutations. While in Cladocera and Rotifera the response of the prey is phenotypically induced by the presence of a predator ( Adler and Harvell 1990; Harvell 1990), in the studies of Xiphocaris and Odonata larvae (see above) the authors did not try to explain the mechanism of the prey adaptation. In the case of the mechanisms of eye reduction in subterranean animals the pros and cons of the last two options represent the topic of an ongoing discussion (e.g. Sket 1985).

To avoid biased conclusions owing to different samples sizes, as well as highly variable relative rostral lengths in some populations, the average relative rostral length of each sample has been used for determining a boundary between populations with short and long rostra. A boundary has been set at 43% considering the minimal number of individuals classified outside the predicted group (see Table 1 , under G/O).

Two gaps in the distribution area of Proteus have been reported by Kletečki et al. (1996): the gap between Slovenian populations and those in Croatian Istra and between populations of northwestern and southeastern Croatia (i.e. in the Velebit mountains and partially in Lika). Whereas in the northern Dinarides the majority of populations with short rostra have actually been living in the area without Proteus , i.e. without predatory pressure, in the southern Dinarides only two exceptions from the Adriatic phylogroup have short rostra: a population on the southern border of the Velebit mountains (62) and the nearby island population (64) (see Figure 4 ). Most populations from the latter phylogroup have long rostra despite a presumable absence of predatory pressure. In samples 69, 65, 66 and 63, however, many shrimps with short rostra are present within samples characterized by their high variation in relative rostral lengths: 28–57%, 36–57%, 36–56% and 30–50%, respectively (see Table 1 , under G/O). We have tried to explain deviation from the expected rostral length pattern in the Adriatic phylogroup with the distribution of cave shrimps’ putative cyprinid fish predators ( Zupančič and Bogutskaya 2002), but without success. High variation in samples mentioned above can be explained (1) by the increased variability in structures not influenced by selection pressure ( Darwin 1859), (2) by a deficient knowledge of actual Proteus localities, denoting higher prey–predator contact than estimated, or (3) because the response of the Adriatic populations on the presence of Proteus might be different to that in other phylogroups. Irrespective of either phenotypic or genetic induction of the shrimps’ response, the third option is supported by the genetic distinction of the Adriatic phylogroup. One has to notice that the combination of long rostra and absence of Proteus is not fatal for shrimps, whereas the opposite situation would probably decrease the shrimps’ survival rate. It would, however, be energetically costly to produce the long rostra without at least a small biological advantage.

Apart from the “outliers” from the Adriatic phylogroup, the E-Slo phylogroup population 42 (from Slovenia) with long rostra and single specimens from samples 81 and 90 (from Italy) with short rostra (data from Fabjan 2001, no molecular data available; see Table 1 ) also show deviation from the expected pattern. The former population is living in a locality with presumably no predatory pressure, whereas in the latter two samples, Proteus presence is confirmed for the first and questionable for the second location. We would like to point out that Proteus is, however, known from close proximity to the Slovenian sample 42 (see Figure 4 ), and that the Italian Troglocaris from nearby locations (i.e. 80, 82 for the first and 89 for the second sample mentioned above; Table 1 and Figure 4 ) all have distinctly longer rostra. It is also true that the Italian localities 81 and 90 are situated on the very border of the Proteus distribution area, Proteus being highly endangered ( Bressi 2004). This could mean that the predatory pressure here is lower than in the centre of the distribution area. Other populations examined by Fabjan ( Table 1 ), including 11 populations from the same localities as examined in the present study, all follow the anticipated rostral length pattern. According to results of our research, the relative rostral length represents an adequate tool for making assumptions about Proteus presence/absence, also in the samples of Troglocaris s. str. excluded from our statistical analyses (denoted with “#” in Table 1 because of doubtful absence of Proteus ).

The increased carapace size in Troglocaris s. str. co-occurring with the predator is the result not only of the elongation of the rostrum, but also of the significantly larger postorbital carapace length. On the other hand, the main factor for the increased carapace size in Xiphocaris elongata populations living with predatory fish seems to be the elongation of the rostrum (evident from fig. 5 in Covich et al. 2009). Also in shrimps from the Adriatic phylogroup that are on average smaller (in comparison to the shrimps from the W-Slo and the E-Slo phylogroups), body size increases upon co-occurrence with Proteus . An increase in size serving as a defence mechanism has already been reported in the green alga Scenedesmus ( Baldwin 1996) ; its consumer Daphnia ( Crustacea: Phyllopoda) elicits changes in Scenedesmus , which render the alga too large to be ingested.

Proteus impact on the morphology of the subgenera Spelaeocaris and Troglocaridella

In the Dinarides where Troglocaris s. str. is widespread, other species of cave shrimps also occur. Although they mainly exhibit similar morphometric response to the presence of Proteus (see Figure 1 ), the values are not consistent with values obtained for Troglocaris s. str. Insufficient data did not allow adequate statistical analysis. Troglocaris (Spelaeocaris) kapelana Sket and Zakšek, 2009 , from Lika ( Croatia) and its yet undescribed sister species, T. ( Spelaeocaris ) sp. ( Sket and Zakšek 2009), from Bosanska Krajina ( Bosnia and Herzegovina) coexist with Proteus . They both have long rostra bearing sparse, large spines. In T. (S.) prasence Sket and Zakšek, 2009 samples, specimens with long as well as with short rostra are found, although no data on Proteus presence is available. In samples close to the known Proteus localities (i.e. with known hydrological underground connections), relative rostral length surpasses that registered in localities isolated from known Proteus localities: 33–57% CL (n = 15) and 19–27% CL (n = 6), respectively. For some specimens from one of Proteus -free localities (Rijeka Crnojeviča, Montenegro), not only a shortened rostrum but also a rostrum without any teeth is characteristic. Specimens of T. (Spelaeocaris) neglecta Sket and Zakšek, 2009 , known from just a few Proteus -free localities, all have short rostra without teeth and they are smaller than Troglocaris s. str. specimens.

In the only species of the subgenus Troglocaridella , the situation is similar to that seen in T. (S.) prasence : the rostrum is longer in samples from caves inhabited by Proteus or hydrologically connected caves (25–57% CL, with 11% and 20% CL, exceptions; n = 22) than in the Proteus -free localities (11–23% CL, n = 5).

While the carapace length in adult Troglocaridella is similar to the carapace length of Troglocaris s. str., the carapace is on average longer in T. (S). kapelana , T. (S.) sp., T. (S.) prasence and in T. (S.) pretneri . Perhaps this is compensation for the lack of a long rostrum, particularly in T. (S.) pretneri , which is on average the largest species of the Dinaric cave shrimps. For such a conclusion, larger samples and further studies are needed.

All the above examples of the genera Troglocaris s. l., Xiphocaris and Leptopalaemon confirm the assumption that the relationship between predation and rostral length is a widespread phenomenon in shrimps, making this structure less useful for taxonomy.

Feeding efficiency of Proteus

Both Proteus recognized Troglocaris as prey although shrimps had never been offered to them during the last 20 years of captivity. Their behaviour did not change throughout the experiment. Furthermore, the small size of the experimental arena did not interact with the exhibition of their natural behaviour because the identical predation behaviour has already been observed in much larger laboratory pools (Aljančič, personal observation) and in nature (e. g. Briegleb 1963). Besides, the animal’s adaptation to occupy narrow crevices is well expressed in its shape and its behaviour.

In most cases the time needed for swallowing the shrimp with a rostrum was longer, but with obvious individual differences between both analysed predators. The spitting-out reaction was not recorded by the camera, but as it was witnessed by the first author in one Proteus , we believe such a reaction might occur in a negligible number of cases. Since predator–prey interactions are usually very complex, different equilibria between the prey and the predator may establish (see Adler and Harvell 1990, for review). The consequences of the shrimp’s armament can therefore be anticipated, but we cannot conceive its influence upon selection. The situation is similar to the results of a laboratory test reported by Johansson and Samuelsson (1994): the European perch ( Perca fluviatilis Linnaeus, 1758 ) succeeded in devouring the shortspined larvae of Leucorrhinia dubia ( Odonata ) significantly faster and more easily than the long-spined larvae. Furthermore, our results indicate that the shrimps are not easy prey to catch. None of the successful Proteus attacks on shrimps with long rostra were accomplished from the front of the shrimp whereas numerous unsuccessful attacks were recorded. In addition, the most frequent escape reaction of the shrimp approached by Proteus was quick backward swimming. Hypothetically this would increase the survival rate in shrimps, especially when their main defence is frontal as described in Troglocaris .

We acknowledge the possibility that the artificially shortened rostra might have affected the shrimps’ behaviour, leading to faster consumption by Proteus . Nevertheless, we believe it is highly improbable as shrimps were given time to recover before the experiment, and no visible change in their behaviour was observed after amputation. Moreover, shrimps collected in the field frequently had broken rostra, without other visible effects. No evidence of the unsuccessful Proteus attacks causing breakage could be gathered in the field. As only a high survival rate would be beneficial for the shrimps after the predator’s attack, we think that clipping the rostrum has little effect on the shrimps.

Taxonomical consequences of rostral variability in the genus Troglocaris s. l.

Rostral variability within each of the analysed phylogroups is high, so applying rostrum morphology to the taxonomy of Troglocaris s. str. as a major or even a sole character is inappropriate.

Although great importance was placed on the rostrum in caridean taxonomy (e.g. see De Grave 1999), separation of taxa by rostral length and dentition led to some incorrect taxonomic decisions and misidentifications. In Xiphocaris Bouvier (1925) as many as three species were defined solely by rostral lengths: Xiphocaris elongata Guerin - Meneville, 1855, Xiphocaris gladiator Pocock, 1889 and Xiphocaris brevirostris Pocock, 1889 . It is not impossible that these species are in fact identical with some long- and short-rostrum forms of Covich et al. (2009). The latter authors, however, did not indicate whether their Xiphocaris forms were phenotypically caused morphs or genetically fixed races.

Troglocaris schmidti Dormitzer var. intermedia Babič, 1922 of the subgenus Troglocaris s. str. ( Sket and Zakšek 2009) was described from Croatia (59) by its smaller size and shorter rostrum in comparison with the type population from the Slovene cave Kompoljska jama (35). Later authors (e.g. see Sket and Zakšek 2009) treated the Croatian population as a subspecies, T. anophthalmus intermedia . Considering the results of our study, diagnosis of the subspecies is highly inaccurate because most other Troglocaris s. str. populations living in caves without Proteus share the same morphological characteristics. In fact, many populations with short rostra are more related to populations with long rostra than to the population of T. a. intermedia (see Table 1 and Figure 4 ).

The genus Spelaeocaris (now a subgenus, Sket and Zakšek 2009) was also mainly defined by its very short rostrum ( Matjašič 1956a, b). Nowadays, species with long rostra are also known to belong to the aforementioned subgenus (see Sket and Zakšek 2009).

High rostral variability within Troglocaris s. str. phylogroups 1–5 does not allow universal separation among the phylogroups, neither by the rostral length, nor by shape. Rostral length and/or shape and dentition are proven to be adequate taxonomic characters only in T. (T.) bosnica , despite its variability in rostral length. Troglocaris bosnica has always at least 10, usually between 13 and 16 teeth (or even more) ventrally on its long rostrum. On the other hand, specimens from the Troglocaris s. str. phylogroups 1–5, have mainly 0–9 ventral rostrum teeth. In less than 1% of cases examined in our study, the number is 10–13 (see also Sket and Zakšek 2009). Rostral shape only allows recognition of certain populations within some phylogroups (see Results).

Figures

Figure 1. A simplified phylogenetic tree of the genus Troglocaris, redrawn and adapted from Sket and Zakšek (2009). Phylogenetic relationships among Troglocaris aggr. anophthalmus phylogroups are weakly supported, and T. anophthalmus – Soča has been added manually (dashed line; see also Zakšek et al. 2009). *Our specimens of the putative Istra phylogroup have not been investigated molecularly. Relative rostral length has been estimated for Proteus and Proteus-free localities, within each of the three Dinaric subgenera separately. For comments on few exceptions within Troglocaris s. str. and situation in T. (S.) prasence and T. (Troglocaridella) hercegovinensis see text.

Figure 4. Coincidence between the relative rostral length in Troglocaris s. str. and co-occurrence of Proteus sp. Sample numbers are as in Table 1; diamonds and plus symbols denote samples with relative rostrum average <43% and> 43%, respectively; asterisks (*) denote samples of T. bosnica. SLO, Slovenia; I, Italy; CRO, Croatia; BIH, Bosnia and Herzegovina.