Habitat orientation alters the outcome of interspecific competition: A microcosm study with zooplankton grazers

Abstract Habitat orientation has recently been demonstrated to affect the foraging behavior, growth, and production of plankton grazers. Because the orientation effect may vary with species, we hypothesize that habitat orientation may alter interspecific interactions between animal species. We experimentally investigated how habitat orientation (placing cuboid chambers in three orientations with long, medium, and small side as the chamber height) affected the interaction between two common cladoceran species, Daphnia magna and Moina micrura, which competitively exploited green algae of Chlorella pyrenoidosa at two volume scales (64 and 512 ml). Results show that chamber orientation and volume additively affected the behavior and species performance of the grazers. Specifically, both grazer species generally decreased their average swimming velocity, grazing rate (on algal cells), body size, and survival and reproduction rates with increasing chamber height for both chamber volumes and with decreasing chamber volume regardless of chamber orientation. Nevertheless, the decrease magnitude was greater for M. micrura with increasing chamber height but was greater for D. magna with decreasing chamber volume. Correspondingly, when cocultured, the density ratio of D. magna to M. micrura increased with increasing chamber height but decreased with decreasing chamber volume. At the end of the experiment, none of D. magna individuals survived in the small and short (large‐based) chambers, and few M. micrura individuals survived in large and tall (small‐based) chambers. These results indicate that both habitat orientation and size affect the outcome of interspecific competition between grazer species. We suggest that variation in habitat orientation may improve community coexistence and species diversity in nature.

Even for habitats having the same size and shape, they may differ in how they are placed in space; this geometric feature is so-called habitat orientation (Pan & Sun, 2016). Habitat orientation is diverse in nature and can have ecological consequences at different organization levels. For example, narrow terrestrial forest patches could be placed either perpendicular or parallel to the migratory direction of bird species, which have differential effects on bird abundance (Hollenbeck & Ripple, 2007). Similarly, narrow aquatic plant patches could be situated perpendicular or parallel to water current (Tanner, 2003), harboring different levels of fish abundance and diversity.
Recent studies provide direct experimental evidence that habitat orientation affects the foraging rate (Pan & Sun, 2016), survival, growth, and reproduction, as well as population dynamics of zooplankton grazers (Zhang, Pan, Chen, Hu, & Sun, 2017), showing its ecological consequences at the individual and population level.
Here, we further propose that habitat orientation may exert species-specific effects on species performance and hence affect interspecific competition of aquatic organisms. Taking cuboid chambers for phytoplankton-zooplankton systems as an example, if the chamber volume and the ratio of length: width: height are constant, varying habitat orientation will change vertical length (in the direction of gravity), which is longer in tall (small-based) chambers than those in short (large-based) ones (see Figure 1). Thus, tall chambers can lead to higher proportion of vertical movement compared to short chambers for zooplankton grazers (e.g., cladoceran species) due to the random walk strategy of these organisms in aquatic environments (Visser & Thygesen, 2003). Given that cladoceran grazers generally have to spend more energy to swim upward the same distance than individuals swimming horizontally as a result of additional costs of overcoming gravity, swimming velocity of these grazers should be lower in tall chambers than in short ones (Pan et al., 2015).
Such behavioral responses of cladocerans to habitat orientation would transmit to affect their grazing rate and thus body growth, and survival and reproduction rates due to the significant positive relationships between swimming velocity and grazing rate (Christensen, Lauridsen, Ravn, & Bayley, 2005;Visser & Kiørboe, 2006) and between grazing rate with body growth, and survival and reproduction rates (Rinke & Vijverberg, 2005;Zhang et al., 2017).
Importantly, cladoceran grazers may have differential responses to habitat orientation among species that coexist in the same aquatic community. For example, body shape has a greater influence in determination of the drag and energy consumption of aquatic organisms when they moving through water. As a result, cladocerans (e.g., Daphnia sp.) with higher ratio of body depth to width should be more appropriate to swimming vertically than horizontally (Weber & Noordwijk, 2002;Zhang et al., 2017) and thus would be more adapted to narrow and deep waters; while in wide and shallow habitats, the species with lower ratios of body depth to body width should be better adapted. If the interspecific difference in the responses due to body shape is large enough, the habitat orientation would possibly affect the outcome of the interspecific competition.
We in this study determined the effects of habitat orientation (three levels: placing cuboid chambers in three orientations with long, medium, and small side as the chamber height, respectively) on interspecific competition between two cladoceran species Daphnia magna and Moina micrura, which were fed with the chlorophyte Chlorella pyrenoidosa at two chamber volumes (64 and 512 ml). We collected the data on the velocities of horizontal swimming, upward F I G U R E 1 Illustration showing three levels of habitat orientation and two levels of chamber volume. The side length ratio was 1:2:4 for each chamber swimming. and downward swimming, and the time allocation among horizontal swimming, upward swimming, downward swimming, and quiescence, as well as species performance (e.g., grazing rate, body size, and survival and reproduction rates), and then, we determined the relative competitive ability between D. magna and M. micrura.
We predicted that (1) D. magna would be more appropriate to swimming and grazing in tall chambers than short ones (because it had a higher ratio of body depth to width compared to M. micrura), and (2) thus, the survival, growth and reproduction rates, as well as the competitive ability of D. magna, would be higher in the tall chambers.

| Experimental organisms
Our experimental organisms included two cladoceran species (D. magna and M. micrura) and one algal species (Chlorella pyrenoidosa). D. magna and M. micrura were collected from Lake Taihu (31.5°N, 120.1°E), East China. These two cladoceran species are reported to coexist in many freshwater ecosystems worldwide, such as lakes, reservoirs, rivers, and ponds (Chen et al., 2016;Marcé, Comerma, García, Gomà, & Armengol, 2005;Zhao, Sun, Huang, & Dai, 1996). Moreover, D. magna adults are much larger (adult size 2.3-6.0 mm vs. 0.5-1.3 mm) and seem to prefer vertical movement by spending more time on vertical movement than M. micrura adults (Pan et al., 2015). Green alga C. pyrenoidosa (FACHB-28) was obtained from the Freshwater Algae Culture Collection of the Institute of Hydrobiology, the Chinese Academy of Sciences. The two cladoceran species were fed with C. pyrenoidosa at a rate of 10 5 cells/ ml per day for more than 3 months prior to the experiments. Both grazer and algal species were grown in COMBO in an incubator at 20 ± 0.5°C with a 14-hr light (at 50 μmol photons m −2 s −1 ): 10-h dark cycle.
Prior to the experiments, the grazers (about 2-3 days old) were transferred to clean COMBO medium and starved for at least 4 hr to increase motivation to forage. Only medium-sized individuals of each grazer species (1.10 ± 0.09 mm long for D. magna and 0.83 ± 0.07 mm long for M. micrura; the data denote mean ± SD, n = 12) were used as experimental materials because large individuals might bear embryos and reproduce during short-time experiments and small individuals were hard to trace during behavioral observations. Reynolds numbers of D. magna and M. micrura were always greater than 2.5 according to empirical formulas: Reynolds numbers = UL/v, where U is the three-dimensional instantaneous velocity (ranging from 0.26 to 0.46 cm/s and from 0.33 to 0.56 cm/s for D. magna and M. micrura, respectively; calculated from the results of this study); L is the body length (cm); and v is the coefficient of kinematic viscosity (cm 2 /s, Svetlichny & Hubareva, 2005).

| Experimental design
The experiment was designed to include six chamber treatments, that is, three levels of habitat orientation (varied the orientation of cuboid chambers with the small, medium, and large side as base, respectively) × two levels of chamber volume (64 and 512 ml), with each treatment having 18 cuboid chambers (transparent polyethylene containers, see Figure 1). The side length ratio was 1:2:4 for each chamber. The 18 cuboid chambers (for each treatment) were used for three different (sub) experiments on grazer behavior, grazing rate, and interspecies competition, with each experiment having six replicate chambers. Each chamber was set as a cocultured plankton system, in which cocultured D. magna and M. micrura were grazers and the alga C. pyrenoidosa was their exclusive diet. The grazer activity and grazing rate were measured 4 hr after the start of the experiment. However, the competition experiment lasted for 32 days. Furthermore, in both grazing rate and competition subexperiments, we set up an additional treatment that included the algal species only, such that the grazing rate of the cocultured grazer spe-  Lürling, 2003). The algal density was set relatively low (2.5 × 10 4 cells/ml) to minimize the reaction of positive phototaxis, but it was still above the incipient limiting level for grazer species and has been used by many other microcosm experiments (e.g., Gilbert, 1990;Pan, Zhang, & Sun, 2014).
All the chambers were capped with breathable polyethylene films and then transferred to incubators (TS-2102GZ, Shanghai Anjing laboratory equipment Co., Ltd) that were set at 20 ± 0.5°C with light intensity being 50 μmol photons m −2 s −1 . They were mechanically stirred (at 90 rpm) and gently aerated with sterile filtered air (Sartorius, Midisart 2000) for 5 min every 1 hr throughout the experiment to facilitate gas exchange (to homogenize DO concentration between treatments) and keep the algae in suspension.
Measurements indicated that algal density was indistinguishable between the top and bottom layers of the experimental chambers an hour after a stirring event (see Appendix 1).

| Behavior experiment
Behavior experiment was carried out in a laboratory at 20 ± 1°C, which was achieved by an air conditioner. Cool-white fluorescent bulbs were installed approximately 2 m away around (including the top and all sides) the experimental setup providing illumination (at 40 μmol photons m −2 s −1 ) to homogeneous light condition following our previous studies (Pan et al., 2015). Such a physical setting may not only minimize the grazer reaction of positive phototaxis that confounds the random walk pattern of zooplankton grazers (Garcia, Moss, Nihongi, & Strickler, 2007;Komin, Erdmann, & Schimansky-Geier, 2004), but also allow for our results applicable to the animals that do not show phototaxis in aquatic systems (Zhang et al., 2017).
Swimming activities of grazer individuals were recorded 4 hr after the beginning of the experiment with two video cameras (1,920 × 1,080 pixels), which were placed orthogonally at a distance of 26 cm from the centre of the projective plane (i.e., the bottom or the backside of the chamber). This physical setting allowed for concurrently recording the swimming behavior of a zooplankton individual in both horizontal and vertical directions. After 15 min of acclimation, the recording began and finished after 5 min of filming at 60 frame/s. Video recordings were transferred to computer, and four types of behaviors were recorded using an image measurement tool (Adobe After Effects CS4) for the grazers, that is, horizontal swimming, upward swimming (swim vertically upward), downward swimming (swim directly downward), and quiescent status. Firstly, we recorded the time for the four types of behaviors. Then, we chose the fragments that contained swimming trajectories away from the chamber walls (in the middle of both views of the camera) and were longer than 2 s (following Gorski & Dodson, 1996;Moison, Schmitt, & Souissi, 2012) to calculated instantaneous swimming velocity as the distance swum by the grazer individuals between two frames (i.e., 16.9 ms) using ImageJ 1.46 and MTrackJ plugin (Artells et al., 2013;Manenti, Denoël, & Ficetola, 2013). Each video was first calibrated to convert pixels into real distances (mm) using reference marks when analyzing the instantaneous swimming velocity. Finally, the average swimming velocity (V average , mm/s) was approximated according to the formula that was used in our previous studies (Pan & Sun, 2016;Pan et al., 2015): and T q were the durations of horizontal swimming, upward swimming, downward swimming, and quiescence (s), respectively; V h , V u , and V d were the velocities of horizontal, upward, and downward swimming (mm/s). Three individuals of each grazer species were randomly chosen to determine the above-mentioned metrics in each chamber, and thus, a total of 108 individuals per grazer species (3 individuals per chamber × 6 chambers per treatment × 6 treatment) were followed. All these metrics of each grazer obtained during the behavioral investigation was averaged for each chamber before data analyses. We additionally calculated swimming velocity ratio as the ratio of average swimming velocity of D. magna to M. micrura.

| Grazing experiment
Investigation of grazing rate (i.e., clearance rate) was conducted in the incubators (TS-2102GZ, Shanghai Anjing laboratory equipment Co., Ltd) that were set at 20 ± 0.5°C with light intensity being 50 μmol photons m −2 s −1 . The algae were sampled 4 hr after the beginning of the experiment. During sampling, we first measured the concentration of DO using a Hach HQ40d oxygen probe (Hach, Loveland, Colorado, USA) as soon as the chambers were taken out of the incubator. Results showed that DO was unaffected by habitat orientation and chamber volume and was always higher than 6.7 mg/L (see Appendix 2), indicating that oxygen availability was not a potential limitation for the growth of grazer individuals after an intermittent stirring event. Then, 2 ml of the sampled solution was removed to a 10 ml tube that contained 0.1 ml of Lugol's preservative for microscopic enumeration of algal cells. At the end of the experiment, algal density was determined using a light microscope at 400× magnification.
The grazing rate (G, ml animal individual −1 hr −1 ) was calculated as the difference in algal density between the experimental treatments (with grazer) and the corresponding controls (without grazers) following the formula of Pace, Porter, and Feig (1983): where V is the chamber volume (ml); C 0 and C 1 are the algae density at the end of the measurement in the control and experimental chambers (cell/ml), respectively; N is the number of the grazer individuals within the chamber; and T is the duration of the experiment (4 hr).

| Competition experiment
This experiment was designed to evaluate whether habitat orientation affected interspecific competition between the two grazer species after long-term cocultures (32 days). The experiment was carried out in incubators that were controlled under the same conditions as for the investigation of grazing rate, except for using a light-dark cycle of 14 hr:10 hr. Microcosms were sampled on days 2, 4, 6,8,10,12,14,17,20,24,28, and 32 at 8:00 after the treatments were initiated. The number of individuals was counted for each grazer species in each chamber before each sampling. During sampling, 5% of the volume was removed from each chamber, and 2 ml of these removed solutions was transferred into a 10 ml tube that contained Lugol's preservative for the measurement of C. pyrenoidosa. Then, 5% volume of fresh medium was added to each chamber to replenish nutrients and prevent metabolic waste buildup. Subsequently, each chamber was capped and replaced back into the incubator.
Additionally, body size, reproduction, and survival rates of D. magna and M. micrura were estimated on day 6 when the size differences between the parental and neonatal grazers were most obvious and when the first-generation offspring of the grazers had not begun producing new neonates according to previous studies (e.g., Pan et al., 2014;Zhang et al., 2017). Body size of each grazer species was determined using all (<5) survived adult females in small chambers and five randomly chosen adult females in large chambers following Zhang et al. (2017) as the length from top of the head to tip of the abdomen, which was measured using an inverted light microscope at ×40 magnification. Body size was averaged for each chamber before data analyses because the number of grazer individuals differed largely among chambers. Reproduction rate (No/ Ni, %) and survival rate (Ns/Ni, %) were estimated as the number of offspring (No) and survived adults (Ns) to the total number of individuals added to the chamber at the initial of experiment (Ni), respectively. Body size ratio and survival rate ratio were defined as the ratio of body size and survival rate of D. magna to M. micrura, respectively.

| Data analysis
All data were tested for normality and variance homogeneity prior to analyses. Data on swimming velocities were log-transformed to achieve the normality and homogeneity. Two-way ANOVAs were used to determine the effect of habitat orientation and chamber volume on swimming behavior, average swimming velocity, and grazing rate, density, and other species performance variables (including body size, reproduction, and survival rates, as well as the swimming velocity ratio, body size ratio, survival rate ratio, and density ratio of D. magna to M. micrura) for both grazer species and the cell density of the algal species on a specific day, followed by the Tukey test for multiple comparisons once a significant effect was detected. Three-way ANOVAs were used to determine the effects of habitat orientation, chamber volume, and grazer species on average swimming velocity. Two-way repeated measures ANOVA (RM-ANOVA) were used to test the effects of treatment factors (habitat orientation and chamber volume) on algal cell, grazer density, and density ratio of D. magna to M. micrura during the competition experiment, followed by the Tukey test for multiple comparisons once a significant effect was detected. The sphericity assumption was evaluated with the Mauchly's test, and in the case of violation, the Greenhouse-Geisser correction was applied to recalculate the Fvalue. In addition, linear regression analyses were conducted to determine the relationships among treatment factors mentioned above. All the analyses were carried out using IBM SPSS19.0 package (SPSS Inc., USA).

| Swimming activity
Both chamber orientation and volume, but not their interaction, significantly affected swimming activity and average swimming velocity for both D. magna and M. micrura (Figures 2 and 3; Tables 1 and 2). For both chamber volumes, increasing chamber height generally led to increases in the durations of quiescence and upward swimming (Table 2), thereby decreasing the time ratio of horizontal to vertical swimming in both grazer species (Figure 2c,d).
Nevertheless, increasing chamber height led to a decrease in both horizontal and upward swimming velocities (Table 2). Consequently, average swimming velocity decreased with increasing chamber height for both grazers species (Figure 3a,b). For all three orientations, increasing chamber volume generally led to increase in the duration of horizontal movement and both horizontal and upward swimming velocities. As a result, average swimming velocity was generally higher in large chambers than small ones for both grazer species (Figure 3a,b).
However, the effects of chamber orientation and volume differed between grazer species (Tables 1 and 2

| Grazing rate
Both chamber orientation and volume significantly affected the algal density in the presence of grazers (as reflected by grazing rate) ( Table 1) but not in the absence of grazers (Figure 3c).
The grazing rate consistently decreased with increasing chamber height for both chamber volumes, and it also appeared to increase with increasing chamber volume for all chamber orientations ( Figure 3d). Additionally, the grazing rate was significantly and positively associated with average swimming velocity of D. magna (r 2 = .767, p < .05; Appendix 4a) and M. micrura (r 2 = .923, p < .01; Appendix 4b), as well as the average swimming velocities (r 2 = .957, p < .01; Appendix 4c).

| Population dynamics
Chamber orientation and volume showed contrasting effects on population density between the two grazer species (Figure 6). Both chamber orientation and volume significantly affected the algal density in the presence of grazers (as reflected by grazing rate) but not in the absence of grazers throughout the experiment (Table 3). In the presence of grazers, despite the large variation in algal density with the experiment progressing, the algal density was generally lower in the short chambers than tall ones for both chamber volumes, and it was generally lower in the large chambers than small ones regardless of chamber orientation in the later stage of the experiment (Appendix 5).

| D ISCUSS I ON
The spatial competition theory that patchy environments may allow more species to coexist has achieved a great success in ecology (Mcclain & Barry, 2010;Tilman, 1994;Yeager et al., 2016). The present study is the first to show that habitat orientation affects the outcome of interspecific interaction between zooplankton grazers. Our results suggest that habitat orientation, in addition to size and shape, is an important abiotic factor that should be considered when trying to understand how spatial heterogeneity affects species coexistence.
F I G U R E 2 Time ratios (means ± 1 SD, n = 6) of quiescent status to total recording time (a, b) and of horizontal to vertical swimming (c, d) of grazer species Daphnia magna and Moina micrura when they were cultured in combination with the green alga Chlorella pyrenoidosa as the exclusive diet under three levels of habitat orientation and two levels of chamber volume. Different letters indicate significant differences among treatments. Multiple comparisons of treatment means were performed using Tukey test at the 0.05 significance level

| Effects of habitat orientation on zooplankton behavior and grazing rate
Spatial storage effect is one of potential mechanisms contributing to coexistence, particularly between closely related species (Schäffler, Saborowski, & Kappeler, 2015). The spatial storage effect operates when competing species exhibit different responses to spatial variation and the strength of competition varies spatially (Abrams & Holt, 2002;Jiang & Morin, 2007). Morphological differences can be one of the important causes accounting for the storage effect as they can result in a differential locomotor pattern (Emlet, 1994) and thus in the use of space between competing species (Basset, 1995;Westphal et al., 2006). For example, D. magna individuals are usually found in the upper layer of the experimental vessel (Weber & Noordwijk, 2002) and move vertically more frequently compared with M. micrura (Pan et al., 2015). This might be attributed to higher ratio of body depth (the maximum dorsoventral depth) to body width in D. magna TA B L E 1 Summary of ANOVA results (F and p values) showing the effects of habitat orientation, chamber volume, and/or grazer species on average swimming velocity and grazing rate (n = 6) in behavior and grazing experiments, respectively, under three levels of habitat orientation and two levels of spatial scales F I G U R E 3 Average swimming velocities (means ± 1 SD, n = 6) of Daphnia magna (a) and Moina micrura (b) and their combined effects (d, means ± 1 SD, n = 6) in foraging algal Chlorella pyrenoidosa (using the treatment without predators as controls, c) when they were cultured in combination under three levels of habitat orientation and two levels of chamber volume. Different letters indicate significant differences among treatments. Multiple comparisons of treatment means were performed using Tukey test at the 0.05 significance level TA B L E 2 Swimming behavior (means ± SD, n = 6) of Daphnia magna and Moina micrura when they were cocultured under three levels of habitat orientation and two levels of chamber volume 3.0 ± 0.6c 2.2 ± 0.2b 3.0 ± 0.6a SS, small volume and short chamber height; SM, small volume and medium chamber height; ST, small volume and tall chamber height; LS, large volume and short chamber height; LM, large volume and medium chamber height; LT, large volume and tall chamber height. Different letters indicate significant differences among treatments. Multiple comparisons of treatment means were performed using Tukey test at the 0.05 significance level.
F I G U R E 4 Variations (means ± 1 SD, n = 6) in body size (a and b), reproduction rate (c and d) and survival rate (e and f) of grazer species Daphnia magna and Moina micrura with green alga Chlorella pyrenoidosa as the exclusive diet when they were cocultured in the chamber on day 6 under three levels of habitat orientation and two levels of chamber volume. Different letters indicate significant differences among treatments. Multiple comparisons of treatment means were performed using Tukey test at the 0.05 significance level than M. micrura (especially between juveniles of the two grazer species), which permitted to reduce the pressure drag in vertical movement in D. magna. Consequently, D. magna individuals would be more adapted to narrow and deep habitats compared to M. micrura.
As expected, increasing chamber height led to deceases in average swimming velocity in both grazers species, but the decrease in magnitude was more obvious for M. micrura than that for D. magna individuals. Aquatic organisms (e.g., large zooplanktons with Reynolds numbers >1) have to spend more energy to swim upward for the same distance than individuals swimming horizontally as a result of additional costs of overcoming gravity (Garcia et al., 2007;Pan et al., 2015). Therefore, zooplankton individuals usually decrease average swimming velocity with increasing duration of upward swimming in tall spaces in order to allocate more energy to the survival and growth of grazers (Pan & Sun, 2016;Seuront, 2006;Svetlichny & Hubareva, 2005). As a result, in the present study, F I G U R E 5 Linear regressions between average swimming velocity with body size (a and b), reproduction rate (c and d) and survival rate (e and f), and between body size with reproduction rate (g and h), of Daphnia magna and Moina micrura individuals, as well as between swimming velocity ratio with body length ratio (i) and survival rate ratio (j), of Daphnia magna to Moina micrura, respectively, under three levels of habitat orientation and two levels of chamber volume (n = 6) D. magna individuals should have allowed more energy and time in swimming activity in tall chambers because of a higher morphological adaptation to reduced water drag force, as evidenced by the less reduced average swimming velocity for D. magna compared to M. micrura in the tall chambers (Figure 3a,b). Because the grazing rate of zooplanktons is often positively correlated with swimming velocity (Visser & Kiørboe, 2006;Pan et al., 2015; and the present study, see Appendix 4), it is not surprising that the differential effects of habitat orientation on swimming velocity led to the significant difference in grazing rate between the two grazer species.
F I G U R E 6 Densities (means ± 1 SD, n = 6) of Daphnia magna and Moina micrura when they were cocultured in the chamber throughout the month-long experiment under three levels of habitat orientation and two levels of chamber volume. SS, small volume and short chamber height (a); SM, small volume and medium chamber height (b); ST, small volume and tall chamber height (c); LS, large volume and short chamber height (d); LM, large volume and medium chamber height (e); LT, large volume and tall chamber height (f). The symbol *, **, and *** indicate statistical significance at the .05, .01, and .001 level, respectively, between densities of the two grazers species through one-way RM-ANOVA

| Effects of habitat orientation on interspecies competition
The altered grazing rate by spatial variation might have further transmitted to affect the development of the study grazers. Previous studies show that low grazing rate often limits body growth and induce high mortality rate in zooplankton individuals (Ashforth & Yan, 2008;Ismail, Qin, & Seuront, 2011), and zooplankton organisms often reduce reproduction rate under starvation so as to improve body reserves to avoid possible future food shortages (Bradley, Perrin, & Calow, 1991). In the present study, the lower values of Small space can inhibit the swimming activity of zooplankton species, especially for large ones (Dodson, Ryan, Tollrian, & Lampert, 1997). Therefore, swimming velocity might be stimulated by increasing chamber volume in both grazer species, but should be more obvious in large-sized D. magna than small-sized M. micrura. Consistently, increasing chamber volume increased the average swimming velocity of both grazer species, but was more prominent in D. magna than in M. micrura due to a higher increase in the velocity of both upward (vertically ascent) and horizontal movements. Consequently, the grazing rate, body growth, survival and reproduction rate were more significantly improved in D. magna relative to M. micrura, resulting in a higher density ratio of D. magna to M. micrura in large chambers than small ones, regardless of chamber orientation.
TA B L E 3 Results of two-way repeated measures ANOVA (F and p values) showing the effects of time, orientation, and chamber volume on algal and grazer densities (n = 6) in different phytoplankton-zooplankton systems

| Summary
In summary, our results provide experimental evidence that changes in habitat orientation and size affect species differentially likely because of the difference in their body morphology and body size, which further affected the outcome of interspecies competition.
This finding indicates that variation in habitat orientation can serve as an additional dimension of spatial heterogeneity, permitting more species to coexist and promote species diversity within biological communities (Davies et al., 2005;Kneitel & Chase, 2004). Indeed, habitat orientation may contribute to species coexistence in nature.
For instance, leopards better coexist with lions while competing for captured prey in treed African savanna than untreed grassland; vertically and horizontally flat bodied fishes coexist in tropical reefs bearing various oriented habitats that are possibly similar in both size and shape (Gardiner & Jones, 2005;Swanson, Arnold, Kosmala, Forester, & Packer, 2016). Thus, the effect of habitat orientation on species interaction should be further studied in various ecosystems to fully understand the importance of habitat orientation to community coexistence and species diversity.

ACK N OWLED G M ENT
This study was supported by grants of National Natural Science Foundation of China (31670440, 31540007, and 31325004) and Yunnan Province Key Research Program for Science and Technology (S2017BC041).

CO N FLI C T O F I NTE R E S T
None declared.

APPENDIX 1
Cell densities (means ± SD, n = 6) of Chlorella pyrenoidosa in treatment containing only algae cells across different layers of water column in the large volume and small-based experimental chambers. Algal density was monitored at the 1st, 2nd, and 3rd hour in the experiment with a frequency of 5 min stirring per hour (see more details in text). Different letters indicate significant differences (p < .05) among treatments. Multiple comparisons of means were performed using Tukey test at p = .05, following one-way ANOVAs.

APPENDIX 2
Dissolved oxygen concentration (means ± SD, n = 6) in different plankton systems under different microcosm conditions during the grazing experiment. Different letters indicate significant differences (p < .05) among treatments. Multiple comparisons of means were performed using Tukey test at the 0.05 significance level.

APPENDIX 3
Swimming velocity ratio of Daphnia magna to Moina micrura when they were cultured in combination under three levels of habitat orientation and two levels of chamber volume. Different letters indicate significant differences among treatments. Multiple comparisons of treatment means were performed using Tukey test at the 0.05 significance level.