Introduction

Predation is an important selective agent driving the evolution of prey species. To alleviate the tremendous effects of predation on survival community dynamics, many prey species have developed inducible defenses to impede or prevent capture and ingestion by a variety of predators1, 2. Inducible defenses, as striking examples of phenotypic plasticity3, can evolve because the presence of some reliable cues released by predators induces several kinds of effective but costly defenses that help prey species survive in a variable and unpredictable predation risk environment4. Many freshwater organisms, e.g., algae, rotifers, cladocerans, and amphibians, have developed various inducible defenses, including defensive morphology, diel vertical migration, swarming, life history shift, and diapause, to avoid predation5,6,7. Inducible defenses can not only generate a type of adaptive prey mechanism to defend against predation, but also promote coexistence of prey species and constitute community stability8, 9.

Monogonont rotifers, as an important part of zooplankton communities, were the first model organisms used to investigate the ecology and evolution of inducible defenses in aquatic ecosystems4, 10. Nearly 1,600 monogonont species exist worldwide11; however, since the discovery of inducible defenses in rotifers12, 13, only fourteen rotifer species in the genera Brachionus, Keratella, Plationus, and Filinia, have been reported to exhibit predator-induced defenses10. Almost all identified inducible defenses of these rotifers are morphological, involving the development and elongation of lorica spines10. Inducible defenses are common in freshwater ecosystems1, 5,6,7, hence the lack of evidence to determine the widespread existence of inducible defenses in rotifers has been confusing scientists14. At present, no satisfactory explanation for this phenomenon has been proposed.

In aquatic ecosystems, some species can change structures of cuticle under predation risk, e.g., increased thickness and rigidity of Daphnia carapace in response to the predatory tadpole shrimp15, increased cuticle thickness of the dragonfly Leucorrhinia larvae in response to fish16, and increased shell thickness of the snail Physa in response to crayfish17. Ultrastructural defenses are predicted to be adaptive because it might be difficult for predator to crush and ingest the prey with rigid cuticle18, 19. In general, rotifers are prey for a variety of predatory invertebrates who are usually equipped with the jaws or mandibles, e.g., predatory rotifers (Asplanchna), cyclopoid copepods (Mesocyclops), predatory cladocerans (Leptodora), and midge larvae (Chaoborus)20. It makes logical sense, therefore, that ultrastructural defenses in rotifers could be an effective grazing deterrent toward multiple predatory invertebrates. Moreover, development of ultrastructural defenses may be particularly favored by small-sized rotifer prey species, which are not large enough to offer a ‘size refuge’. By now, predator-induced ultrastructural defenses in rotifers have received little attention. Determining the distribution of such defenses across different predator-prey systems will enhance our understanding of the ecological consequences of inducible defenses which evolve in a complex and multi-predator environment.

The herbivore rotifer B. calyciflorus and B. angularis and their predator A. brightwellii are often sympatric and widely distributed in ponds, lakes, and rivers all over the world21,22,23,24. The large B. calyciflorus can develop elongated spines in response to kairomones of A. brightwellii 13; however, the smaller B. angularis does not exhibit such inducible defenses. The present study compares the morphological responses of B. calyciflorus and B. angularis to A. brightwellii in terms of lorica ultrastructure, including thickness and hardness. However, the responses may be more evident in small-sized B. angularis. Ultrastructural defenses are often predicted to be adaptive18, 19, but direct evidence demonstrating this is relatively scarce. Here we evaluate whether the induced traits of Brachionus act as a defense against Asplanchna. Moreover, we anticipate finding some fitness costs associated with development of ultrastructural defenses in Brachionus because one of the prerequisites for evolution of inducible defenses is the significant cost which prevents defenses from being fixed4. Lastly, we assess the fitness consequences to B. calyciflorus and B. angularis in two prey-one predator interactions.

Results

Morphological responses of Brachionus to Asplanchna

The biometric characters of Brachionus measured in the present study were shown in Fig. 1. In response to Asplanchna kairomones, B. calyciflorus developed remarkably elongated posterolateral spines (general linear mixed-effects model, F = 749.869, P < 0.001) and larger body size (general linear mixed-effects model, F = 14.883, P = 0.001); however, B. angularis did not develop any spines and maintained body size (general linear mixed-effects model, F = 0.028, P = 0.869) when exposed to kairomones (Table 1). Asplanchna could induce ultrastructural defensive morphology in both Brachionus species; this defensive morphology was expressed as an increase in lorica thickness (supporting information, Fig. S1). Both B. angularis (general linear mixed-effects model, F = 53.889, P < 0.001) and B. calyciflorus (general linear mixed-effects model, F = 57.528, P < 0.001) developed thicker lorica when they were exposed to Asplanchna kairomones (Table 1). B. angularis has many minute projections spread throughout the lorica, but such is not the case in B. calyciflorus (supporting information, Fig. S2). These characteristics could explain the smooth lorica surface of B. calyciflorus and the uneven lorica surface of B. angularis in transverse section view (Fig. S1). In accordance with lorica thickness, Asplanchna kairomones also affected the strength of the lorica surface (Table 1). Observations using atomic force microscopy (AFM) showed that B. angularis exposed to kairomones obtained higher cuticle hardness than unexposed ones (general linear mixed-effects model, F = 695.110, P < 0.001), whereas the unexposed B. calyciflorus had lower cuticle hardness than exposed ones (general linear mixed-effects model, F = 132.943, P < 0.001). The data of force-depth indentation curves confirmed the above mentioned results (supporting information, Fig. S3), indicating that the AFM tip was pushed much deeper into the lorica of the unexposed Brachionus prey than that of kairomones-exposed prey.

Figure 1
figure 1

Biometric characters of Brachionus calyciflorus (left) and Brachionus angularis (right) measured in the present study. (A) body length; (B) body width; (C) length of posterolateral spine. White filled circles indicate the tested areas on each rotifer under atomic force microscopic observations. Scale bar = 100 µm.

Table 1 Biometric characters of Brachionus calyciflorus and Brachionus angularis cultured in the medium with and without Asplanchna kairomones.

Benefits and costs of inducible defenses in Brachionus

The predatory A. brightwellii exerted disparate selectivity and preference on induced and noninduced rotifer prey (one-way ANOVA, df = 3, F ingestion rate = 8.649, P ingestion rate < 0.001, F ingestion time = 10.257, P ingestion time < 0.001, Fig. 2). The feeding behavior of Asplanchna indicated that the ingestion rate (percentage of ingested individuals after captured) for defended B. angularis was significantly lower than that of the three other rotifer morphotypes (Duncan post hoc tests, P < 0.05, Fig. 2). Approximately 40% of captured induced B. angularis were rejected unharmed by Asplanchna; however, almost all captured individuals of the three other rotifer morphotypes were ingested by Asplanchna (Fig. 2). The ingestion time (time from the successful capture of prey by jaw to ingestion of captured from mastax to stomach) reached the highest value (54.1 s) for defended B. angularis and the lowest value for undefended B. calyciflorus (3.6 s) but did not differ between undefended B. angularis and defended B. calyciflorus (Duncan post hoc tests, P < 0.05, Fig. 2).

Figure 2
figure 2

Prey ingestion rate (the percentage of the number ingested to number of successful captures) and prey ingestion time (time from the successful capture of prey by jaw to ingestion of captured from mastax to stomach) of Asplanchna brightwellii fed Brachionus calyciflorus (B.c.) and Brachionus angularis (B.a.) cultured in the medium with (K+) and without (K) Asplanchna kairomones. To minimize the potential influences of body size and presence of attached eggs on feeding preference of A. brightwellii, we conducted the experiments with newborn B.c. (<1 h old) and non-ovigerous adult B.a. (>24 h old). Body length of B.c. in K = 137.0 ± 1.2 µm (n = 30); Body length of B.c. in K+ = 150.8 ± 1.4 µm (n = 30); Body length of B.a. in K = 124.6 ± 1.2 µm (n = 30); Body length of B.a. in K+ = 122.3 ± 1.4 µm (n = 30). Prey ingestion rate = 100 × (number of ingestion/number of successful capture); Prey ingestion time = time from successful capture of prey by jaw to ingestion of captured from mastax to stomach. Data of prey ingestion rate are mean ± S.E. based on 15 replicated observations within 10 min. Data of prey ingestion time are mean ± S.E. based on 15 replicated recordings.

We detected significant fitness costs of inducible defenses in two Brachionus species, but occurring in different forms (Fig. 3). Compared with unexposed B. angularis, production of offspring in exposed B. angularis decreased significantly (Mann-Whitney U-test, Z = −2.666, P = 0.008). However, no variation of reproduction between B. calyciflorus individuals cultured in K+ and K environments (Mann-Whitney U-test, Z = −0.639, P = 0.523) was observed. The mixis ratio of offspring in relation to kairomones showed a reverse trend in two Brachionus species (Fig. 3), indicating that those with Asplanchna kairomones had a lower mixis ratio of offspring in B. calyciflorus (Mann-Whitney U-test, Z = −9.175, P < 0.001), but showed no difference in B. angularis (Mann-Whitney U-test, Z = −0.191, P = 0.849).

Figure 3
figure 3

Boxplots of production and mixis ratio of offspring in Brachionus calyciflorus and Brachionus angularis cultured in the medium with (K+) and without (K) Asplanchna kairomones. The solid horizontal line is the median and the triangle is the mean. The deviation bars represent the 5th and 95th percentiles, whereas the dots show the 1th and 99th percentiles. Data shown are based on 64 replicated Brachionus mothers.

Fitness consequences for Brachionus when interacting with Asplanchna

In most cases, when B. angularis and B. calyciflorus were cultured together with adult A. brightwellii, B. angularis obtained much higher short-term fitness than B. calyciflorus (Fig. 4). Regardless of morphotypes, B. angularis individuals constantly had a higher survival than B. calyciflorus individuals (Mann-Whitney U-tests, Z = −3.085 ~ −4.330, P < 0.01) except for the combination of induced B. calyciflorus and non-induced B. angularis, which showed no difference in survival (Mann-Whitney U-test, Z = −1.862, P = 0.078).

Figure 4
figure 4

Survival of Brachionus calyciflorus (B.c.) and Brachionus angularis (B.a.) cultured together with Asplanchna brightwellii. At each prey combination, B.c. and B.a. cultured in the medium with (K+) and without (K) Asplanchna kairomones are provided with proportion of 50%: 50% (25 individuals from each species). We conducted the experiments with newborn B.c. and non-ovigerous adult B.a. (see the legend in Fig. 3 for the detailed explanation). Data are mean ± S.E. based on 12 replicated observations.

The population dynamics of B. angularis and B. calyciflorus in coexisting cultures with or without A. brightwellii, are shown in Fig. 5. The dynamics of the two Brachionus populations showed no difference during the culture period when cultured together (repeated measures ANOVA, F 1,10 = 0.257, P = 0.623, Fig. 5). When the coexisting Brachionus were cultured together with adult A. brightwellii, population dynamics of the two Brachionus populations had no difference during the first half of coexistence experiments (repeated measures ANOVA, F 1,10 = 0.871, P = 0.373) but differed significantly during the second half of coexistence experiments (repeated measures ANOVA, F 1,10 = 28.162, P < 0.001). The presence of predatory Asplanchna resulted in near extinction of B. calyciflorus from the environments by the end of the experimental period (Fig. 5).

Figure 5
figure 5

Population dynamics of Brachionus angularis (B.a.) and Brachionus calyciflorus (B.c.) fed with Chlorella pyrenoidosa at a concentration of 1 mg C L−1. Left: B.a. and B.c. were co-cultured without predator. Right: B.a. and B.c. were co-cultured under predation risk (Asplanchna brightwellii = 100 ind. L−1). Data shown are mean ± standard error values based on six replicates.

Discussion

In our study we detected the presence of ultrastructural defenses in herbivorous B. angularis and B. calyciflorus, which are induced by the kairomones of the predatory rotifer A. brightwellii and manifested as increased lorica thickness and enhanced lorica hardness. Asplanchna can induce not only the development of longer posterolateral spines and larger body size in B. calyciflorus, but also shifts in the ultrastructure of lorica. However, B. angularis exhibits evident ultrastructural defenses in response to Asplanchna kairomones, but retains body shape. For B. angularis or B. calyciflorus, defended morphs are less susceptible to predation by A. brightwellii than basic morphs in terms of ingestion rate and ingestion time. Given the hard armor of defended B. angularis, individuals were physically better protected against predation, leading to higher fitness in predator-prey interactions. Compared with that of B. angularis, the elongated spine and strengthened lorica of induced B. calyciflorus provided less anti-predation capability. As a consequence of the development of inducible defenses, both B. angularis and B. calyciflorus experienced fitness costs; however, these costs were manifest in different forms. The costs of inducible defenses were expressed as decreased reproduction in B. angularis and reduced sex investment in B. calyciflorus.

Shifts of morphology, life history, and behavior in response to predatory vertebrates or invertebrates have been reported as inducible defenses in cladocerans5. However, the main inducible defenses in rotifers are morphological, including development or elongation of spines10. The exuberant phenotypes provide effective protection from small predatory invertebrates (e.g. Asplanchna)25, 26. This protection can be explained by the ‘anti-lock-and-key’ hypothesis27. However, defensive morphologies with spination might be ineffective against large predatory invertebrates, which are always equipped with strong mandibles25, 28. For defense against large predatory invertebrates, the development of hard armor would be more adaptive. Although both Brachionus prey species exhibit the combination of a harder and thicker lorica under predation risk, such inducible defenses are more evident and effective in the smaller sized B. angularis. The types of inducible defenses employed by prey species depend on the evolutionary history and the ecological environment of the organism29, 30. On the one hand, multi-predator environments are common in nature30, 31. The coexistence of a variety of predatory invertebrates results in heavy predation on rotifer prey species, particularly small rotifers20. On the other hand, spined morphs of small rotifers may still be not large enough to offer ‘size refuge’ from predation. Therefore, we hypothesize that the evolution of ultrastructural defenses may be more common in small rotifer species, and is adaptive to multiple predatory invertebrates. Small rotifer prey may be easily seized and swallowed whole by predatory invertebrates, but an induced rigid lorica can protect them from being completely ingested. Moreover, differences in lorica thickness and hardness will increase individual fitness without altering body shape, thereby avoiding changes in hydrodynamics (e.g., shifts in center of gravity)32 and graspability (e.g., loss of relatively smooth cuticle due to spination)33. These speculations are possible because some small but hard-armored rotifers (e.g., B. budapestinensis and K. cochlearis) are less vulnerable to predatory invertebrates20, 34.

In freshwater ecosystems, two types of defense strategies are recognized: (1) pre-capture defenses that enable prey to avoid observation or detection by a predator and (2) post-capture defenses that protect prey from capture and ingestion by a predator1. In our work, inducible ultrastructural defenses in Brachionus can be categorized as post-capture defenses, in which induced individuals with greater rigidity of their lorica are capable of impeding handling and ingestion by Asplanchna. Our results verify the general idea that architectural or ultrastructural defenses of prey species provide effective mechanical protection against many potential predators18, 35. Based on our data, we can further predict that rotifers with harder lorica will be less susceptible to a variety of predatory invertebrates because such ultrastructural defenses make prey difficult for predatory invertebrates to crush and ingest. Although development of a harder lorica together with long spines and large sizes may further impede capture or handling of B. calyciflorus by predator, the well-defended B. angularis obtains high short-term and long-term fitness when they coexist with A. brightwellii. Some may argue that the experimental designs with newborn B. calyciflorus in predation experiments may underestimate the performance of strengthened lorica and size effects of B. calyciflorus populations in defending against predation from Asplanchna. However, we argue that similar characteristics of the lorica may exist in both neonate and adult B. calyciflorus, because empirical data indicated that the handling time of prey for A. brightwellii did not differ between young and adult spined B. calyciflorus 25. Moreover, a range of sizes of Brachionus would coexist with the predator in long-term population experiment. Finally, B. angularis obtained higher survival in both short-time and long-time cultures with the predator. Thus the potential effects of varied sizes could be minimized, and prey with thicker and harder lorica would be better defended.

Commonly, resource competition between identical rotifer species causes the extinction of the inferior competitor36. However, some experimental evidences showed that inducible defenses could promote coexistence of identical species when they competed for a single food source37. The costs of inducible defenses create negative feedback loops that prevent strong population fluctuations under predation risk8. Our data verify this idea, showing that two Brachionus coexisted in the environments containing predator kairomones. In addition, the preferential selection of prey species by a common predator as a result of the disparity of prey availability and prey palatability can drive more susceptible species to extinction9, 38,39,40. Two Brachionus prey species have developed inducible defenses in response to Asplanchna kairomones, but the well-defended B. angularis obtains higher fitness when they coexist with A. brightwellii. Although B. calyciflorus was not completely outcompeted during the course of the experiment, its density was marginal in comparison with the superior competitor B. angularis. Considering that the intensity of predation and effectiveness of inducible defense frequently play vital roles in predator-prey interactions39, 40, future studies associated with influences of inducible defenses on multi-trophic communities should consider these ecological factors in their theoretical or empirical models.

Allocation costs in defended rotifers have been hypothesized and tested. Some empirical studies have supported this hypothesis26, 41, 42, whereas other research found no defense cost25, 43. Unsuccessful attempts to observe fitness costs in relation to inducible defense may be a consequence of the experimental design rather than the lack of tradeoffs14, 44. Here, we detected allocation costs, that is, decreased reproduction or reduced sexual reproduction, in defended Brachionus prey species, verifying the previous findings and proving the presence of fitness costs in rotifers26, 41, 42. In our work, neonates of the basic phenotypes are cultured in both K+ and K environments to assess costs of inducible defenses; thus, we may have underestimated the magnitude of cost in the two Brachionus species because we only evaluated the cost of producing defenses and neglected the cost of maintaining such defenses. In addition, we speculate that the maintenance costs of defended B. angularis will be much lower than that of B. calyciflorus because morphological changes can alter the hydrodynamics of B. calyciflorus and probably result in high energy requirements for maintenance. Interestingly, these two species have developed disparate allocation costs. For B. calyciflorus, inducible defenses cannot provide complete protection against predatory invertebrates25; accordingly, energy expenditure in sexual reproduction is saved and allocated to the production of more parthenogenetic offspring to offset predation loss42. In contrast, the relatively high survival rates of B. angularis compared with that of B. calyciflorus in the presence of predatory invertebrates may preserve the energy expenditure of sexual reproduction and maintain resting-egg production, which is important for the long-term fitness of the population45.

In conclusion, the results of this study confirmed the existence of predator-induced ultrastructural defenses, which are expressed as increased lorica thickness and enhanced lorica hardness, in Brachionus rotifers. Such inducible defenses are usually invisible under an optical microscope and are referred to as hidden defensive morphology. To our knowledge, this is the first time that the presence of inducible ultrastructural defenses in monogonont rotifers has been observed. Our results not only expanded the understanding of inducible ultrastructural defense to other planktonic taxa, which were previously only observed in cladocerans15, 19, but also verified the effective mechanical protection of such hidden defenses. Furthermore, our data suggest that the evolution of inducible ultrastructural defenses in prey populations may be adapted to a variety of predatory invertebrates. Hidden ultrastructural defenses are supposed to be taxonomically widespread30; thus, further empirical and field experiments should be designed to uncover the ecology and evolution of such inducible defenses in other rotifers and other planktonic taxa.

Methods

Rotifer culture

The herbivores B. angularis and B. calyciflorus and carnivorous A. brightwellii were obtained from wetting of dried sediment collected from a freshwater pond (39°57′N; 116°21′E) in Beijing, China, following the methods of previous studies42, 46. One newly hatched female Brachionus or Asplanchna was isolated and allowed to reproduce parthenogenetically in the COMBO medium47 enriched with 40 ml L−1 soil water extract40 (hereafter COMBO). Brachionus was cultured with the food alga Chlorella pyrenoidosa, which was also cultured in COMBO, at a concentration of 1.0 mg C L−1. Asplanchna was fed with a mixture of B. angularis and B. calyciflorus at a density of 10 ind. mL−1. The culture medium was renewed every two days, and the rotifer densities were maintained at low levels (≈200 ind. L−1). All rotifer and algal cultures, as well as experimental incubation conditions, were maintained at 20 ± 1.0 °C with a 14 L:10 D photoperiod. Before use in the experiments, all Brachionus and Asplanchna rotifer clones were cultured in the laboratory for at least a month in exponential growth phase.

Morphological responses of Brachionus to Asplanchna

The experiment was performed in 1.0 L glass beakers containing 0.5 L COMBO with (K+) or without (K) Asplanchna kairomones. The preparation of Asplanchna-conditioned COMBO was described in previous studies42, 46. In brief, 100 adult females of A. brightwellii (≈800–900 µm in length) were placed into a 1000 mL COMBO and deprived of food supply for 24 hours. Then, conditioned COMBO was filtered with fiber filters (0.45 µm). The predation risk of Asplanchna in the present work is relatively high (100 ind. L−1), but this is common in natural water environments21, 48, 49. The experiment was started by placing 10 egg-carrying amictic females of B. angularis or B. calyciflorus into beakers containing K+ or K COMBO. Brachionus were fed daily with 1.0 mg C L−1 of C. pyrenoidosa. The culture medium was renewed every day and the density of Brachionus population was controlled below 500 ind. L−1. The density of rotifer prey was relatively low, which would produce the maximum magnitude of defense in response to Asplanchna kairomones50. Brachionus are fast-growing monogonont species with the juvenile development time less than 24 h at 20 °C51. Furthermore, life span of experimental clones was less than 7.5 days in this study (supporting information, Table S1). Therefore, our experiments lasted for 15 days that can diminish the potential maternal effects on inducible defenses46. Then, adult B. angularis or B. calyciflorus was harvested and preserved in 70% ethanol to measure the thickness and hardness of lorica. For each Brachionus species, five replicates in each treatment (K+ or K) were conducted.

Transmission electron microcopy was used to evaluate lorica thickness of Brachionus. Brachionus individuals from each replicate were fixed in 2.5% glutaraldehyde for 2 h. Following three 20-min changes in 0.1 M phosphate buffer (pH 7.5), rotifers were post-fixed at 4 °C in 1% osmic acid (2 h), dehydrated in a graded ethanol series (50%, 70%, 80%, 90%, and 100%), and embedded in epoxy resin. Sections were cut with an ultratome (E + E Elektronik), mounted on grids, post-stained with 2% aqueous uranyl acetate (30 min) and lead citrate (15 min), and examined under an electron microscope (JEM-1200EX, JEOL). Microscopic images of transverse sections were used to measure lorica thickness. Micrographs of five different areas in a transverse section were taken at 10,000× magnification. For each replicate, three random transverse sections, belonging to three different rotifer individuals, were assessed. Based on the scale bar in the micrograph, we measured the lorica thickness with Auto CAD 2014 software. The value of lorica thickness in each replicate was based on the mean of the three samples. It should be noticed that we could not locate the sections used for the measurement of thickness in a certain part of Brachionus body due to their small body size. Hence, the influences of varying thickness in the different parts of body might not be excluded in the present work.

An atomic force microscope (MultiMode 3D, Veeco, USA) with a 125 × 125 µm lateral scan range and a 125-µm z-extension stage was used to evaluate lorica hardness of Brachionus. A microscope (Olympus SZ2-STS, 9F07795), connected to CCD camera (10x-A, Nikon 1102716), allowed precise positioning of the silicon nitride cantilever at the area of interest on the lorica. The silicon nitride cantilevers (14–26 KHz, SNL-10, Veeco, USA) with a nominal spring constant of 0.32 N/m and a pyramid tip (nominal tip radius = 2 nm) were used for all atomic force microscopy (AFM) measurements. A 9 × 9 nm2 force map containing one measurement point was probed and a force-depth indentation curve was obtained from each area of interest. Five different areas were analyzed on each rotifer (Fig. 1), and two randomly selected rotifers were measured per replicate. The maximum force exerted on the lorica was set to 3 nN, resulting in maximum indentation depth of approximately 15 nm, which ensured that only the hardness of the lorica surface was detected. The Young’s modulus (E) used to indicate the hardness E = F×(1 −  v 2)/(0.7453 × δ 2 × tanα), from the force-depth indentation curve with the equation described in previous works15, 19:

$$E=F\times (1-{v}^{2})/(0.7453\times {\delta }^{2}\times \,\tan \,\alpha ),$$

where E is the Young’s modulus, F is the indentation force applied to the sample, v is the Poisson’s ratio (set to 0.5 in this study), δ is the indentation, and α is the face angle of the pyramid (22.5° in this study). In each case, AFM was measured using a wet rotifer sample. The mean Young’s modulus of each rotifer was calculated to indicate the average hardness of lorica, and the value of lorica hardness in each replicate was based on the mean of the two samples.

We also evaluated the visible morphological changes of Brachionus in response to Asplanchna kairomones, e.g., spine development and elongation, and body size changes. Six Brachionus were randomly chosen from the preserved samples of each replicate. Biometric characteristics, such as body length and width and length of developed or elongated spines (e.g. posterolateral spines), of Brachionus rotifers were measured to the nearest 2.5 µm at a magnification of 400× (Fig. 1). Body size (volume) of Brachionus was calculated following the previous work52. The values of biometric characters in each replicate were based on the mean of the six samples.

Benefits and costs of inducible defenses in Brachionus

When exposed to Asplanchna kairomones, both Brachionus species developed morphological or ultrastructural defenses (see ‘Results’ for detailed information). Thus, we assessed the performance of inducible defenses of B. angularis and B. calyciflorus in defending against A. brightwellii predation. Following the methods described above, we obtained populations of B. angularis and B. calyciflorus in the K+ and K environments. Defense efficiency was determined based on Asplanchna ingestion rate (the percentage of the number ingested to number of successful captures) and Asplanchna ingestion time (time from the successful capture of prey by jaw to ingestion of captured from mastax to stomach). The criteria used to record the feeding behavior of Asplanchna were detailed in previous studies53. The main goal of this work was to evaluate the effective mechanical protection resulting from variations of lorica ultrastructure. Given the body size and oviposition of rotifers can influence the feeding behavior of Asplanchna 25, 53, we conducted the experiments with newborn B. calyciflorus (<1 h old) and non-ovigerous adult B. angularis (>24 h old) to minimize the potential influences of body size and presence of attached eggs on feeding preference of A. brightwellii. The detailed information of biometric characters of newborn B. calyciflorus and non-ovigerous adult B. angularis, which were measured based on the methods mentioned above, was shown in Fig. 3.

One starved adult Asplanchna (6-h starvation) was introduced into each cell of a 24-well plate containing 0.5 mL COMBO and 10 individuals of one particular herbivore rotifer species cultured in K+ or K. The number of captured and ingested prey within 10 min was recorded with a stereomicroscope. In a separate experiment, ingestion time (handling time) of starved adult Asplanchna when they were supplied with the same types of rotifer prey was recorded. When an Asplanchna ingested one prey item, it was replaced by another starved Asplanchna, and the number of rotifer prey was restored to 10 individuals per 0.5 mL. The ingestion time of Asplanchna was recorded to the nearest 0.1 s using a stopwatch on a smart-phone. For each of the prey species from either K+ or K, a total of fifteen replicated observations were conducted to measure ingestion rate and ingestion time.

At present, two forms of costs in association with inducible defenses have been reported in monogonont rotifers: decreased reproduction and reduced sex investment26, 42. In the present work, we conducted life table studies to determine the costs of inducible defenses of B. angularis and B. calyciflorus. A juvenile amictic female (<6 h old) B. angularis or B. calyciflorus which was cultured in K environment was placed into each cell of a 24-well plate containing 1.0 mL freshly prepared K+ and K COMBO with 1.0 mg C L−1 of C. pyrenoidosa. The kairomone concentration in the environment and the method to prepare K+ COMBO were the same as mentioned above. For each species-environment treatment combination, a total of 64 replicated females were evaluated. The culture medium in each treatment was renewed every 12 h. During this process, the number of newborn B. angularis and B. calyciflorus in each replicate was counted and then the newborns were transferred individually to 24-well culture plates containing 1 mL COMBO and 1.0 mg C L−1 C. pyrenoidosa. These newborn rotifers were cultured until they became ovigerous adults and were then classified as amictic (female-producing) female or mictic (male-producing) female. Experiments were terminated when all mother rotifers died. In cyclical parthenogens such as brachionids, sex investment is frequently defined as the production of male-producing females54, 55. Thus, mean number of offspring and the average mixis ratio of offspring (percentage of male-producing females) were used to assess the costs of inducible defenses in the two rotifer prey species.

Fitness consequences for Brachionus when interacting with Asplanchna

In this work, we evaluated fitness consequences of inducible defenses for B. angularis and B. calyciflorus to the presence of A. brightwellii based on short-term (40 min) and long-term (12 days) multispecies interactions. In the short-term studies, the susceptibility of the basic and induced morphs of B. angularis and B. calyciflorus to predation by A. brightwellii was assessed. Here again, we conducted short-term experiments with newborn B. calyciflorus (<1 h old) and non-ovigerous adult B. angularis (>24 h old). The combination of either 25 induced or 25 non-induced B. calyciflorus together with either 25 induced or 25 non-induced B. angularis was exposed to one starved adult A. brightwellii (6-h starvation) in each cell of a 6-well plate with 5.0 mL COMBO for 40 min. The Asplanchna was removed from the environment, and Brachionus populations in each combination were preserved with 2% formalin; then, the number of individuals of each Brachionus morph was recorded. Each experimental combination was run in 12 replicates.

In the long-term studies, coexistence experiments of the two Brachionus species with or without Asplanchna predation were performed. Experiment was conducted in each cell of a 6-well plate containing 10 ml COMBO and 1.0 mg C L−1 C. pyrenoidosa. The initial density of both B. angularis and B. calyciflorus in the coexistence experiment was 1 ind. mL−1. During the initiation of a separate coexistence experiment, we introduced one adult A. brightwellii into the environment to produce predation pressure. The COMBO used in coexistence experiments was conditioned with Asplanchna at a density of 100 ind. L−1 for a 24-h exposure. Every day, the culture medium was renewed with the appropriate algal food concentration, and the number of each Brachionus species in the environment was counted. During this period, newborn Asplanchna were removed from the environment, and dead Asplanchna were replaced with new ones. The experiments were terminated after 12 days when the trends of most populations became obvious and stable. Each experimental culture was run in six replicates.

Data analysis

We performed a general linear mixed-effects model to test the effects of Asplanchna kairomones on visible and hidden morphological changes in Brachionus. A one-way ANOVA using treatment (morphotype) as the fixed factor was conducted to analyze the prey ingestion rate and prey ingestion time of Asplanchna. When significant differences (P < 0.05) were detected, Duncan’s test was used for pairwise comparisons. Nonparametric Mann-Whitney U-tests were performed to compare: (i) the mean number of offspring and average mixis ratio of offspring in life table studies, and (ii) the survival of rotifer prey cultured together with Asplanchna in short-term fitness evaluations. Repeated measures ANOVA were performed to compare the performance of Brachionus populations with and without predation in long-term fitness evaluations. Assumptions for ANOVA were evaluated using the Levene’s test for homogeneity of variances and the Kolmogorov-Smirnov test for normality. The data, expressed as percentage or ratio, were arcsine transformed. All statistical analyses were performed using the SPSS statistical package version 21.0.