Thwarting predators? A three-dimensional perspective of morphological alterations in the freshwater crustacean Daphnia

Predation is a major selective agent, so that many taxa evolved phenotypically plastic defensive mechanisms. Among them are many species of the microcrustacean genus Daphnia, which respond to an increased predation risk by developing inducible morphological alterations. Some of these features are obvious and easily recognized, e.g., crests in D. longicephala, while others are rather hidden, such as the bulkier shape of D. magna induced by the presence of the tadpole shrimp Triops. In this study we investigated the extraordinary diversity of morphological adaptations in the presence of predators with different foraging strategies in six predator-prey systems. For the first time we were able to analyze the unexposed and predator-exposed morphs comprehensively using three-dimensional scanning and reconstruction. We show that morphological changes are manifold in appearance between species and predators, and go beyond what has been known from previous 2D analyses. This further demonstrates the enormous trait flexibility of Daphnia. Interestingly, we found that among this variety some species share morphological strategies to counter a predator, while others use a different strategy against the same predator. Based on these intra- and interspecific comparisons, we discuss the mechanisms by which the respective defense might operate. These data therefore contribute to a deeper understanding of the inducible defenses’ morphology as well as their diversified modes of operation in Daphnia, being a cornerstone for subsequent investigations, including the determination of costs associated with morphological change.


Introduction
Predation is a major selective agent in ecosystems [1]. Therefore, many taxa evolved inducible morphological defenses to thwart predator attacks [2][3][4]. Microcrustaceans of the genus Daphnia, which are globally found in many freshwater ecosystems, have been intensely investigated and respond to an increased risk of predation conveyed by so-called kairomones and develop adaptive morphological features [5][6][7][8][9][10]. Some of these alterations are obvious and easily recognized, e.g. neckteeth in D. pulex [11,12], crests in D. longicephala [13,14], helmets in D. cucullata or D. lumholtzi [15,16], while other shape changes are obscure such as the abandoned body symmetry in D. barbata [8,17,18]. Moreover, different types of hidden morphological a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 With these data, we are able to confirm or refute hypotheses on the functional strategies of potential morphological defenses and enable new insights to deepen our understanding of the mechanisms behind.

Daphnia culture
Daphnia reproduce by cyclical parthenogenesis, so that experiments can be done on multiple individuals of the same genotype (i.e. clonal isolates). Clones of the investigated Daphnia species (D. magna, D. barbata, D. longicephala, D. pulex, D. lumholtzi) originated from various locations ( Table 1). All animals were cultured in our institute for several generations under laboratory conditions (16:8-hour day-night cycle, 20˚C +/-1˚C). Animals were reared in 1 L Procrustes analysis allows to bring these two landmark sets in the best overlapping fit without changing their relative positions, using at least pan and rotation. This is called partial Procrustes fit and allows comparisons of form. If in addition to pan and rotation also scaling is adjusted, the result is a full Procrustes fit and allows shape analysis. Double-headed arrows give point-wise Procrustes distances, which are minimized during the analysis. The average Procrustes distance is the mean length of all arrows. Landmarksets that differ only in scaling would overlap perfectly using full Procrustes, leading to a Procrustes distance of zero.

Predator culture
Chaoborus obscuripes larvae and Notonecta glauca were caught in the ponds of the botanical garden of the Ruhr-University Bochum. Chaoborus larvae of the 4 th juvenile instar were kept in 1.5 L glass jars at 4˚C in charcoal-filtered tap water. Adult N. glauca were kept at 15˚C in 1 L glass beakers (J. Weck GmbH und Co. KG, Wehr-Ö flingen, Germany) and fed with five adult daphniids and three defrosted chironomid larvae daily. Triops sp. were hatched from sand in 1.5 L tanks and each was fed daily with five adult daphniids, three defrosted chironomid larvae and one pellet of commercial fish food (Multisticks, Tetra GmbH, Melle, Germany). All Triops used for experiments had a body length larger than 1 cm. Gasterosteus aculeatus (obtained from Zoo Zajac GmbH, Duisburg, Germany) with a length of 5-7 cm were fed frozen chironomid larvae and a range of Daphnia sp. in the culture aquaria every other day.

General induction procedures
Morphological alterations in the investigated Daphnia species were induced by exposing them to the predators, respectively remains of consumed conspecifics. We avoided predation on experimental organisms by keeping predators in net cages made of acrylic cylinders of 12 cm height and 7 cm diameter, lined with 125 μm gauze (Hydrobios, Germany), still allowing kairomone exposure. Depending on the predators, we fed the predators with conspecific daphniids during the exposures, as mentioned in the explicit descriptions below. In order to keep the concentration of predator kairomones stable we did not exchange water. Controls were handled similarly, but without the predators. To reveal maximal defense expression for the functional analysis, in all species, we selected strongly induced individuals. Morphological changes were assessed when Daphnia deposited eggs in the brood pouch, except for D. pulex, where we investigated the second juvenile instar, as this is the instar of strongest morphological alteration. For all predator treatments and the respective controls we used 7-16 individuals, which were taken from three replicates per species.

Predator exposures
Triops-exposed Daphnia magna. We induced morphological alterations by exposing 10 D. magna neonates (first instar juveniles) to chemical cues from one Triops sp. We raised D. magna until animals deposited eggs in the brood pouch. Once these offspring were released, adult females were removed, while the neonates remained in the glass jars. This procedure was repeated for three generations [19,26], always keeping the number of animals per 1.5 L jar at a maximum of 10 by randomly removing surplus neonates.
Triops-and Notonecta-exposed Daphnia barbata. D. barbata was shown to develop contrasting features against Notonecta and Triops [17,21]. We therefore exposed 20 D. barbata from the 5 th embryonic stage [36] onwards to 1 individual of both predators independently, again using net cages placed in 1.5 L beakers. Notonecta and Triops were fed as described above in predator culture. Notonecta-exposed Daphnia longicephala. D. longicephala was exposed in groups of 10 individuals to predator cues of the heteropteran backswimmer Notonecta glauca from the first juvenile instar onwards using 1 L glass beakers. We used one Notonecta per net cage fed with 5 adult D. longicephala initially and three D. longicephala on subsequent days. Chaoborus-exposed Daphnia pulex. Inductions were started with 5 female D. pulex carrying embryos in the last embryonic stage, showing black eyespots [36], placed in 1 L glass beakers. For predator exposure, we placed 10 Chaoborus larvae and once 100 juvenile D. pulex as food source for the predator in the net cage. Gasterosteus-exposed Daphnia lumholtzi. D. lumholtzi were exposed to cues from the three-spined stickleback Gasterosteus aculeatus from the 5 th embryonic stage onwards in groups of 20 individuals in a 1 L glass beaker. For that, one stickleback was placed in a net cage positioned inside the beaker for 8 hours per day, without feeding during that time.

Animal fixation, staining and scanning
After reaching the respective instar, animals were fixed, stained and processed as described previously [29]. In short, we fixed the animals with 4% formaldehyde (J.T. Baker, Germany) diluted in phosphate buffered saline (PBS; pH 7.4, 0.1 M), stained them with 0.1% Congo red (Carl Roth GmbH + Co. KG, Karlsruhe, Germany) and scanned them with confocal microscopes (Leica TCS SP5II (Leica Microsystems, Wetzlar, Germany), Zeiss LSM 710 (Carl Zeiss Ltd., Cambridge, UK) and Nikon A1R (Nikon Instruments Europe B.V., Amsterdam, Netherlands)). We used the automatic image stitching when necessary for large specimens.

Modelling
Three-dimensional meshes were extracted from the stacks using MorphoGraphX [37]. After projecting a grid onto the meshes using Blender (Blender, version 2.73, Blender Foundation, Blender Institute Amsterdam, https://www.blender.org/, 2015), the resulting individual models consisting of comparable vertices, which means points on the surface, were averaged to a single model per treatment, using an automated Matlab Script [29] (Matlab R2014b, The Mathworks Inc., Natick, MA, 2015). Here, Procrustes analysis was used to superimpose the models in a best overlapping fit (Fig 1). With this script, we also determined displacement vectors, i.e. vectors between respective vertices in both morphs' models. We projected their length with color gradients on the 3D surface reconstructions in summation (subfigures C) or directions-wise (subfigures D, F, H). Also, Wilcoxon-tests were calculated for the position of vertices in all three dimensions of space between specimens of the unexposed and predatorexposed treatment. To correct for error based on multiple testing with the False Discovery Rate (FDR)-analysis, we calculated q-values, which give a value that allows to evaluate the trustworthiness of each derived p-value [38]. These q-values are based on the FDR, which is estimated from the distribution of p-values of the conducted Wilcoxon tests (subfigures E, G, I Using this 3D morphometric approach, we obtained a comprehensive overview of morphological alterations in the investigated Daphnia species. We describe the significant changes in adaptive morphological features in detail using the nomenclature of body parts listed in S1 Fig. In the following, shifts are described with 'strong tendency' when the Wilcoxon-tests are represented with p-values smaller than 0.05 and regarded significant when respective q-values calculated with the FDR-analysis are smaller than 0.01. To display directional changes, we show point translocation in a direction-specific manner across the x-, y-, and z-axis.

Comparison of intra-and interspecies differences with Procrustes distances
In addition, we calculated Procrustes distances (in micrometer), which represent the sum of the distances of respective vertices after a Procrustes analysis using Matlab (Matlab R2014b, The Mathworks Inc., Natick, MA, 2015). While partial Procrustes allows comparisons of form, including size effects, full Procrustes removes rotation, translation and scaling from shapes ( Fig 1A). For further information refer to [29].The Procrustes distances were calculated between the two treatments of each species to determine the alterations' magnitude (Table 1). Furthermore, we determined the Procrustes-distances between all models of predator-exposed specimens, which enables to evaluate, how strong shape and form differ between these morphs ( Table 2). For the latter, we again adjusted and projected a grid suitable to fit onto all models. This grid contained 432,321 vertices, from these, 384,019 vertices could be projected due to the irregular shape and were accordingly used for the calculation of the interspecies-Procrustesdistance. For the partial Procrustes distances, the size adjustment of the models before the calculation was excluded in the analyses via Matlab, for the determination of full Procrustes distances it was included. As Procrustes distances are dependent on the morphological distance between two morphs, but also influenced by the number of vertices used for the comparison, we divided all Procrustes distances by the number of vertices on which they are based, giving average partial and full Procrustes distance, which represent an average offset between the vertices of the point clouds of both morphs (Tables 1 and 2). Despite the magnitude of alterations which can be examined thoroughly, the actual morphological change of these alterations is not captured. Accordingly, the same numbers do not necessarily describe the same morphologies.

Ethics approval and consent to participate
All animals used in this study were maintained according to animal welfare regulations.

Results
We here describe the relevant, significant alterations between the predator-unexposed and -exposed morphs. For a detailed description of all changes, please see (S1-S6 Text in S1 File).

Daphnia magna: Triops-induced morphological alterations
Triops-exposed (n = 10) compared to -unexposed (n = 9) D. magna (Fig 2) have maximal alterations at the body margins as well as at the tail spine. In the lateral direction, we observed an increased body width at a rim on the carapace, as well as on the fornices in Triops-exposed animals. At the same time, the posterior ventral margin and the dorsal margin are slightly slimmer in predator-exposed D. magna. Summing up all alterations, D. magna confronted with the predator shows an overall increase in body dimensions.

Daphnia barbata: Triops-induced morphological alterations
Maximal changes of form in D. barbata exposed to Triops (n = 7) in comparison to unexposed individuals (n = 7) occur at the tip of the tail spine and at the head (Fig 3). In the lateral dimension, the shape is asymmetric. While the head is shifted to the animal's right, the tail spine is displaced to the left. Additionally, the carapace fold is more pronounced in the lateral dimension in predator-exposed than in unexposed D. barbata. Not only this fold, but also the whole carapace is expanded by 20 μm.

Daphnia barbata: Notonecta-induced morphological alterations
Maximal changes between unexposed (n = 7) and predator-exposed animals (n = 8) occur at the tail spine and at the tip of the helmet, which are more pointed, the rostrum, and the region between the head capsule and the carapace (Fig 4). In the lateral dimension, the width of the fold on the head and the tip of the helmet is increased, while the carapace is not or just slightly thickened. Attenuation of lateral width is found on the tail spine, the ventral margin, and the ventral parts of the head capsule.

Daphnia longicephala: Notonecta-induced morphological alterations
In comparison to the unexposed morph (n = 14), the predator-exposed morph of D. longicephala (n = 16) is characterized by a pronounced crest (Fig 5). The crest and the tail spine show the highest degree of overall deformation. In the lateral direction the folds on the head capsule are increased in thickness. Apart from these very strong alterations, the dorsal and ventral margins are decreased in thickness.

Daphnia pulex: Chaoborus-induced morphological alterations
In Chaoborus-exposed (n = 8) in comparison to unexposed D. pulex (n = 7), strongest shifts in morphology occur in the neckteeth region (Fig 6). This displacement is mainly due to a shift in dorsal direction. In the lateral dimension we observe a decrease in the head's thickness. Accordingly, also the fornix is attenuated. The heart region is characterized by an alternating pattern of positive and negative displacement in this direction.

Daphnia lumholtzi: Gasterosteus-induced morphological alterations
Obvious alterations of D. lumholtzi in Gasterosteus presence (n = 8) compared to unexposed animals (n = 10) are an elongated tail spine and a pointed helmet (Fig 7). The strongest alterations are observable at the tip of the tail spine and the tip of the pointed head capsule. Strong lateral deformations occur at the fornices. These laterally extruded body appendages of the head capsule enhance the absolute width of exposed animals significantly by 200 μm through being more pointed (Fig 7C and 7H insert), while some areas appear thinned in the lateral dimension in predator-exposed animals, e.g. regions at the dorsal and ventral margins of the carapace.

Procrustes distances
Smallest Procrustes distances were calculated for D. pulex and largest distances for D. longicephala (Table 1). Marginal differences between partial and full Procrustes distance were found between D. barbata in the presence of Triops, D. lumholtzi and D. pulex. Most species showed partial Procrustes distances in the range of 1-3% and full Procrustes distances of 1-2% of their body size. The largest difference between partial and full Procrustes distances was measured in D. longicephala. Here, we also found the largest average full Procrustes distance, as well as the largest Procrustes distances in relation to body size (Table 1).
In addition, the partial and full Procrustes distances between the predator-exposed morphs of all species were calculated ( Table 2). To normalize these values, we again used the average distances over all points of a model. These values were generally larger than the values found for the intraspecific Procrustes distance. Typical values for the partial analysis comprising only smaller species range around 500 μm, if larger species are part of the comparison, we find values up to 2161 μm. For the full Procrustes distance, typical values are smaller, the larger portion of comparisons does not exceed 200 μm. In the case of comparisons including D. lumholtzi, often values of less than 100 μm resulted for the full Procrustes analyses. Even in the comparison with D. longicephala, the full Procrustes value is smaller than 100 μm, while in all other species we obtained values ranging from 174-488 μm. The predator-exposed morph of D. lumholtzi is due to the pointed head more similar to D. longicephala than all other species.

Discussion
Based on the three-dimensional models of selected predator-unexposed and -exposed morphs of different Daphnia species we analyzed the intra-as well as interspecific alteration occurring in predator presence. We especially emphasized on the overlapping predator regimes to find evidence for the relevant morphological features in the overall alteration.

Comparison of morphological alterations
At a first glance, morphological changes in the different predator-exposed animals appear specific to the predator. However, analyzing the three-dimensional alterations in more detail allows to detect adaptive patterns overarching several species.
Triops-induced morphologies. We recently described explicit morphological changes in shape and form of D. magna in the presence of Triops sp. [29]. In the current study we confirm our previous results and here also found that D. magna reacts by increasing its overall body size. We find that the body length and dorso-ventral width of the animals was not only locally increased as Rabus et al. [21,26] described, but extended in the overall outline. In addition, we also describe that the fornix' width is significantly increased.
In comparison to D. magna, D. barbata expresses different morphological adaptations in presence of Triops. In Triops-exposed D. barbata we see elongated and twisted helmets and tail spines. The animals' helmets are shifted to the right in Triops-exposed specimens, while the tail spines bend to the left body half, causing a significant asymmetry, which was also described by Herzog et al. [17,22]. In addition to this observation, we find in our 3D analysis that the  Fig 1B). (E, G, I) p-values lower than 0.05 displayed in yellow, regions additionally supported by qvalues lower than 0.01 based on the FDR-analysis are coloured red. n unexposed = 10, n predator-exposed = 9. All scale bars: 1 mm. https://doi.org/10.1371/journal.pone.0254263.g002

PLOS ONE
3D-shape of Daphnia's inducible alterations ventral margin is especially convex at the posterior end, while the tail spine and helmet are bent backwards. Overall, these adaptations follow the form of a sickle.
The described shifts of D. magna and D. barbata in the presence of Triops are similar in the overall body outline which becomes more circular. While D. magna simply grows bulkier, which is thought to hamper Triops predation and to influence the prey choice of Triops [26,[39][40][41], D. barbata bends its body, forming a sickle. As this circular arc is approximately closed under strong expression, it may resemble the 'filled' circular outline of D. magna. Both shapes may help to reduce the contact zone between the Triops' gnathobases forming the food groove [17]. It is possible that the asymmetric shape also helps D. barbata to be swirled out of the food groove. Similarly, the asymmetry could hamper an immediate firm grasp of the prey during free swimming. Triops also hunt by trapping the prey between its body and the ground [42]. This may be impeded through the circular body outline, and thereby increase the daphniids survival chances. Furthermore, the increase in the lateral width may help both species to outgrow the mouth opening of this gape limited predator.
Notonecta-induced morphologies. In contrast to Triops-exposed D. barbata, in Notonecta-exposed D. barbata we found especially pronounced helmets and tail spines, which were previously reported by [17,18,22]. Furthermore, in the area around the center of the dorsal margin and centrally on the carapace we observe an increased lateral width. It has previously been discussed that animals exposed to Notonecta are well-protected also against Triops. This form change appears effective in multiple environments [22].
Forming a pronounced and straight helmet in combination with a pointed body shape could disable predator handling. Furthermore, the dorsal body margin could be the region where Notonecta pierces its proboscis through the carapace. Therefore, an increased lateral thickness may serve as a regional fortification, based not on the material properties, but its form, as this could aid a diversion of the proboscis away from this vulnerable region. A more rounded shape furthermore can help to distribute forces more evenly and therefore improves stability [43]. This mechanism is probably similar in the morphological change of D. barbata in the presence of Triops.
We speculate that the reduced lateral width at e.g., the ventral margin, which is often accompanied by a longer narrowed channel into the ventral gap, contributes to a defense by hampering the piercing into this gap, as it is harder for the predator to find. Furthermore, such narrowing may also help to divert the proboscis of the predator during piercing. Especially as Daphniidae are able to actively narrow their ventral cleft [44], this special shape appears as a useful support to avoid punctures at this position.
Notonecta also elicits a morphological response in D. longicephala [13,[45][46][47], which appears totally different from Notonecta-exposed D. barbata at a first glance. Considering the overall shape of this species, we find an extension of the body in the sagittal plane. In principle, the overall shape is comparable with that of predator-exposed D. magna, as the three-dimensional shape with a central bulge and a thinned body periphery occurs in both species.
In the lateral dimension, the extensions are very thin. The head capsule's lateral width is increased centrally, mainly because of the fornices forming a rim dorsal at the antennas' base. We hypothesize that these changes function to increase the structural stability of the head capsule and carapace, as suggested previously [48]. Since the fornix is elongated as a rim in the  axes (x, y, z respectively). Shades of blue colours represent alterations in negative direction, while shades of red display shifts in positive direction on the respective axis. Note the altered colour scale. The coordinate system gives the relevant axis marked in bold (see also Fig 1B). (E, G, I) p-values lower than 0.05 displayed in yellow, regions additionally supported by qvalues lower than 0.05 based on the FDR-analysis are coloured red. n unexposed = 7, n predator-exposed = 7. All scale bars: 600 μm. Increased crests could allow the organisms to grow larger antennal muscles that are necessary for stronger antennal strokes [49,50]. This is contradictory to recent research that found Notonecta-exposed D. longicephala swim more slowly [51]. In predation experiments, we often observe that Notonecta reorients captured prey D. longicephala until it can insert the proboscis into the ventral gap. In this regard, Notonecta-exposed D. barbata and D. longicephala probably share a common defensive feature, which is the narrowed ventral margin. In contrast, the unexposed shape actually could be exploited to direct the proboscis into the ventral gap.
Furthermore, prey capture seems to be complicated by the increased crest size. With a crest larger than the manipulating appendages of Notonecta, the grasp is weaker, and escape chances increase. This may be especially potent against smaller Notonecta instars. As previously observed, predation by various Notonecta instars affects Daphnia populations differently, i.e. through their choice of size [52]. To reach around the fully expressed crests, longer legs of the predator are necessary, which only later instars possess. Therefore, we hypothesize large Notonecta to be able to prey upon all D. longicephala, while smaller instars are dependent on smaller or unexposed specimens.
We anticipate that the different size classes of the two species reacting to Notonecta, D. barbata and D. longicephala, explain the partly contrasting morphological strategies. Furthermore, it is possible that small-sized D. barbata predominantly fall prey to juvenile Notonecta and are less attractive for later instars or even adult backswimmers in accordance with the theory of optimal foraging [53]. If that is the case, it may explain the varying morphologies in the two species. Maybe, the different alterations seen in D. barbata and D. longicephala are like excerpts of developing defensive traits as seen in D. cucullata, which counter various predators simultaneously in the different size classes of the ontogenetic development [24].
Generally, Notonecta prefers larger prey, suggesting a visual prey detection [54]. Nevertheless, Notonecta is thought to hunt using mechanoreception in addition to visual cues, as they are continuously active and also feed during the night [55,56]. Therefore, apart from beneficial interactions of the Notonecta-exposed morphological features with the feeding apparatus, hydromechanical advantages for D. longicephala are suggested, allowing them to swim disguised [50].
Chaoborus-induced morphology. In the predator-prey system Chaoborus-D. pulex, we observed the prey to form a pronounced head region at the base of the neckteeth. We also noticed an increase in body depth, i.e. the dorso-ventral length was enlarged, as Tollrian already stated [23]. Neckteeth have been described in numerous publications [11,12,36,57]. What has not been described so far, are the changes in lateral width observed here. In fact, the head capsule itself is comparatively thin, while the region of the neckteeth becomes significantly thicker. This is also mirrored in the central body, which is thickened in the predator exposed morph. Together with the increased tail spine length, these alterations corroborate the common hypothesis that neckteeth serve to outgrow the predator's gape, and to impair the predator's handling procedures, rendering D. pulex less graspable, independent of Chaoborus' strike efficiency [4,5,12,23,58]. Pastorok found that Chaoborus handled prey usually head showing the sites of cumulated alteration with blue giving zero and red maximal alteration. (D, F, H) Vertex displacements along the coordinate axes (x, y, z respectively). Shades of blue colours represent alterations in negative direction, while shades of red display shifts in positive direction on the respective axis. Note the altered colour scale. The coordinate system gives the relevant axis marked in bold (see also Fig 1B). (E, G, I) p-values lower than 0.05 displayed in yellow, regions additionally supported by q-values lower than 0.05 based on the FDR-analysis are coloured red. n unexposed = 7, n predator-exposed = 8. All scale bars: 600 μm.  Fig 1B). (E, G, I) p-values lower than 0.05 displayed in yellow, regions additionally supported by q-values lower than 0.05 based on the FDR-analysis are coloured red. n unexposed = 14, n predator-exposed = 16. All scale bars: 2 mm. https://doi.org/10.1371/journal.pone.0254263.g005

PLOS ONE
3D-shape of Daphnia's inducible alterations first, suggesting an interaction of the neckteeth with the mouthparts, which could not be confirmed in a recent study [20,59]. Therefore, we suggest the neckteeth may serve another purpose, e.g. altering the streamlining properties, as suggested for D. longicephala.
Also, the slimmed head morphology together with the increased thickness in the region around the heart may serve to improve stiffness as technically employed in cardboard or undulated surfaces of, e.g., tins [60]. In a previous publication [48], the model's resolution was not adequate to reflect the undulating structure. With our results, we are now able to enhance the precision of such models.
Overall morphological alterations in D. pulex appear, in comparison to the other described species, rather minor. Nevertheless, they have repeatably been shown to correlate significantly with an increase of survival chances when encountering Chaoborus predation and therefore have a strong ecological effect [12,61]. This underpins the hypothesis of a very 'technical' attack sequence, that can be disturbed with rather minor alterations of this 'lock-and-key'-like system [5].
Furthermore, the altered morphology may not only complicate predator handling time, but also the detection, possibly by altered hydrodynamic trails [62]. The hypothesis of altered streamlining properties is especially supported by the finding that Chaoborus-exposed animals are also less frequently attacked [20].
Gasterosteus-induced morphologies. The morphological changes of D. lumholtzi are quite unique compared to the other defensive features in this study, as in this species we found the formation of a pointed helmet and an elongation of the tail spine, which was reported previously [15,46,63,64]. Further, we were able to detect a previously unreported and potentially defensive structure similar to the pointed head and spine: the fornices on the sides of the head capsule are very pronounced in fish-exposed animals, extending the overall lateral width by up to 245 μm. This strategy to elongate structures orthogonal to each other was found in other Daphnia species (e.g. Hyalodaphnia hypsicephala, D. magna, D. cucullata, Scapholeberis mucronata) [65,66].
Since small fish and especially juvenile stages (0+ juveniles) are generally regarded as gapelimited predators, it is possible that this cross-shaped body outline blocks the predator's mouth and therefore significantly impedes fish suction feeding, even though the longitudinal axis is most relevant according to [67]. Even if a predator-exposed D. lumholtzi is being fully sucked into a fish's mouth, there are still many interfering points e.g. the gills slits in which the prey item could get caught prior to full ingestion. Indeed, fish spit out and subsequently avoid induced D. lumholtzi more often (personal observation, [64,68]).
Apart from this specific shape against fish predation, D. lumholtzi's helmets function as a multitool also against Chaoborus predation, similar to the helmets of D. cucullata [24,64].
We hypothesize the reduced lateral and dorso-ventral width of the carapace to be due to a trade-off between energetic and material costs. Furthermore, the shrinking in the dimension of the sagittal plane may allow the flow of water around the daphniids body, while the animal is wedged to the predator's mouth by suction. Moreover, it is imaginable that drag forces are decreased by this shape alteration, reducing the forces pulling the D. lumholtzi into the fish's mouth. Further, smaller bodies reduce visibility to fish [53,[69][70][71].
Apart from D. barbata, which forms much less pronounced helmets, D. lumholtzi is the only species in our study forming pointed helmets. In addition to this species, helmets are  Fig 1B). (E, G, I) p-values lower than 0.05 displayed in yellow, regions additionally supported by qvalues lower than 0.05 based on the FDR-analysis are coloured red. n unexposed = 7, n predator-exposed = 8. All scale bars: 400 μm. https://doi.org/10.1371/journal.pone.0254263.g006

PLOS ONE
3D-shape of Daphnia's inducible alterations known as morphological feature in D. cucullata, D. longispina and D. retrocurva, just to name a few [16,72,73]. Whether similar three-dimensional alterations like fornices also occur in these and other species remains to be investigated.

Procrustes distance as a measure of similarity
While morphological features are not necessarily detectable among species that share the same predator, Procrustes distance, especially average Procrustes distance, provides an objective measure of similarity.
The difference between the unexposed and predator-exposed morph of D. pulex is the smallest average partial Procrustes distance we measured. For the average full Procrustes distance, we determined values of 10.5 μm, i.e., the size has almost no influence on the altered form. Due to the calculation-order in Procrustes analyses, the full Procrustes distance is even slightly larger than the partial Procrustes distance here. This result is in accordance with the proposed anti-lock-and-key mechanism in this species [5], but is also influenced by the fact that D. pulex is the species in which we investigated the smallest individuals (Table 1). D. barbata, D. lumholtzi and D. magna have almost equally large intraspecific average full Procrustes distances (~30 μm), meaning that the pure shapes are altered with comparable effort in reaction to different predators. Nevertheless, as D. magna is obviously larger, the extent of alteration in comparison to body size in this species is lower (1% of body size) compared to D. barbata and D. lumholtzi (2-3% of body size). D. barbata induced by Triops as well as D. lumholtzi show almost no alteration between partial and full Procrustes distance. Mostly shape, not size, is important in the morphological changes of these species. In contrast, the intraspecific average Procrustes distance of D. magna decreases from 80 μm in the partial analysis to 28 μm in the full Procrustes fit, rendering the size-effect obvious. D. longicephala has a strong shape as well as a size component in its alterations, as the partial average Procrustes distance is 261 μm and the full average Procrustes distance is 139 μm.
Between the species, smallest average partial distances occur in comparisons including D. lumholtzi. This species is potentially 'ordinary', i.e., has a combination of features found in the other species, like a roundish central body or an average sized and shaped head. Therefore, the average full Procrustes distance to the other species is comparatively small. Furthermore, both D. barbata morphs show a quite similar partial as well as full average Procrustes distance to D. longicephala. Not only the Notonecta-, but also the Triops-exposed morph of D. barbata has this small Procrustes distance to D. longicephala. Previous studies have shown that also Triops-defended morphs had a reduced mortality under predation by Notonecta [22], therefore the calculated Procrustes distances may indicate a shape similarity.
Moreover, the partial Procrustes distances measured for D. magna in comparison to all other species are quite large. As they decrease substantially for the full Procrustes distances, it becomes clear that size here explains differences to a major extent. Similar to D. lumholtzi, D. magna's overall shape is not that different from the other species. For example, the average full Procrustes distance between D. pulex and D. magna is only 34.934 μm, a value similar to some intraspecific Procrustes distances. lumholtzi, Gasterosteus-exposed. (C) Overall deformation plot showing the sites of cumulated alteration with blue giving zero and red maximal alteration. The insert shows a detail of the head and especially the region of the fornices. (D, F, H) Vertex displacements along the coordinate axes (x, y, z respectively). Shades of blue colours represent alterations in negative direction, while shades of red display shifts in positive direction on the respective axis. Note the altered colour scale. The coordinate system gives the relevant axis marked in bold (see also Fig 1B). Subfigure H additionally gives an insert with a detail of the laterally strongly extended fornices. (E, G, I) p-values lower than 0.05 displayed in yellow, regions additionally supported by q-values lower than 0.05 based on the FDR-analysis are coloured red. n unexposed = 10, n predator-exposed = 8. All scale bars: 400 μm. ©Fig 7A and 7B: Graeve/Janssen. https://doi.org/10.1371/journal.pone.0254263.g007 Accordingly, Procrustes distance as a measure of difference is not able to predict predators, as the amount of alteration in form or shape is not specific for a predator. Nevertheless, Procrustes distance can indicate similar strategies, especially overall growth vs specific, local alteration. This can also give a hint on the costs of the potential defensive structures.

Conclusions
Each encounter situation where the prey passes the predator undetected, increases prey survival chances. Likewise, each attack that is not followed by an instant ingestion provides the chance for the prey to escape. We showed that eco-responsive flexibility with the potential to enhance survival through multiple three-dimensional morphological alterations evolved. Based on them we were able to draw hypotheses on the actual function. A somewhat general strategy seems to be the formation of an elongated tail spine. In addition, we show previously unknown changes in the lateral dimension in all species investigated.
While some of the investigated Daphnia species seem to follow a similar strategy against one predator, others appear to use alternative tactics against the same threat. However, similar morphological features are also formed against different predators. Therefore, generalizations are limited. We speculate that different preferred body sizes potentially require different morphological adaptations. Our results therefore show that, despite their different size classes and evolutionary origins of Daphnia species, similar strategies evolved, which is in accordance with [35].
Apart from the above-mentioned results, average Procrustes distances can point to the degree and type of alteration and energetic investments that are necessary to form the altered morphologies. Accordingly, knowing the three-dimensional alterations is a cornerstone to answer not only functional questions but also considering inducible defenses' trade-offs. For example, with our models, studies investigating the single mechanisms in depth, virtual simulations of the predatory attack or measurements of drag force alterations triggered by the observed three-dimensional alterations, just to name a few, can be undertaken.