Long-term study of behaviors of two cohabiting sea urchin species, Mesocentrotus nudus and Strongylocentrotus intermedius, under conditions of high food quantity and predation risk in situ

Background In the predator–sea urchin–macrophyte trophic cascade, the ecological effect of sea urchins as grazers depends both on their density and the changes in foraging activity, which are influenced by various disturbing factors. However, the complete duration of the alarm reactions of echinoids has not been studied until now. Here, we tested a hypothesis that two cohabiting sea urchins, Mesocentrotus nudus and Strongylocentrotus intermedius, which differ morphologically, might display different behavioral responses to high hydrodynamic activity and predation. Methods We used continuous time-lapse video recording to clarify behavioral patterns of M. nudus and S. intermedius in presence of a large quantity of food (the kelp Saccharina japonica) but under different weather conditions and different types of predation threat: (1) calm weather conditions, (2) stormy weather conditions, (3) predation risk associated with the presence of several sea star species and (4) predation risk associated with an alarm stimulus (crushed conspecifics or heterospecifics). Three separate video recording experiments (134 days in total) were conducted under field conditions. Video recording analysis was performed to determine the number of specimens of each sea urchin species in the cameras’ field of view, size of sea urchins’ groups, movement patterns and the duration of the alarm responses of both sea urchin species. Results We showed that in the presence of kelp, M. nudus and S. intermedius exhibited both similar and different behavioral responses to hydrodynamics and predation threat. Under calm weather, movement patterns of both echinoids were similar but M. nudus exhibited the higher locomotion speed and distance traveled. Furthermore, S. intermedius but not M. nudus tended to group near the food substrate. The stormy weather caused a sharp decrease in movement activity followed by escape response in both echinoids. Six starfish species failed to predate on healthy sea urchins of either species and only a few attacks on ailing S. intermedius specimens were successful. The alarm response of S. intermedius lasted approximately 90 h and 20 h for starfish attacks on ailing conspecifics and for simulated attacks (crushed conspecifics or heterospecifics), respectively and involved several phases: (1) flight response, (2) grouping close to the food, (3) leaving the food and (4) return to the food. Phase three was the more pronounced in a case of starfish attack. M. nudus only responded to crushed conspecifics and exhibited no grouping behavior but displayed fast escape (during 4 h) and prolonged (up to 19 days) avoidance of the food source. This outcome is the longest alarm response reported for sea urchins. Discussion The most interesting finding is that two cohabiting sea urchin species, M. nudus and S. intermedius, display different alarm responses to predation threat. Both alarm responses are interpreted as defensive adaptations against visual predators.

154 conspecifics or heterospecifics) in sea urchin species with different morphological 155 characteristics. 156 157 Video recording analysis 158 The recorded videos were viewed frame by frame. We counted the numbers of specimens of 159 each sea urchin species in the each camera field of view both in the absence and in the presence 160 of disturbing factors. To access the grouping behavior of the sea urchins, we calculated the mean 161 group size as the ratio of the total number of individuals in the cameras' field of view to the 162 number of associations (Hagen & Mann, 1994). Following Vadas et al. (1986), we distinguished 163 between sea urchin associations and aggregations. Each group of sea urchins in two-dimensional 164 groupings, including individuals suspected of being in tactile contact (there was no visible space 165 between them) and single individuals, was considered a separate association. 166 To determine the mean group size for both sea urchin species, the video frames were 167 randomly selected over the periods of calm weather. Only the video frames showing not more 168 than 30 specimens (43 frames for each species) were chosen for the mean group size calculation 169 in order to avoid crowding effects when sea urchins might be in tactile contact due to their high 170 density on the feeders. Under such a limitation, no groups of 3 or more sea urchins in cohesive 171 three-dimensional groupings (aggregations, according to Vadas et al., 1986) were observed in 172 our study. 173 Sea urchins' movement was analyzed using the free software, 'Tracker', for video 174 analysis (www. open sourcephysics.org/items/detail.cfm?ID=7365). The cell size (2 × 2 cm) of 175 the feeder mesh was used as a scale. We tracked and measured sea urchin displacement with an 176 interval of 1 min. Following Lauzon- Guay et al. (2006), we defined a step as the distance 177 between two successive positions of the sea urchin (1 min apart), a stop as an interval when sea 178 urchin remains stationary during at least 1 min (2 successive frames) and a move as the distance 179 between two successive stops which can be composed of one or more steps. The mean 180 locomotion speed was calculated as total distance passed divided by total time. To determine movement patterns of sea urchins in response to crushed con-and 263 heterospecifics, we measured step length and locomotion speed before and after treatment.    Manuscript to be reviewed 309 ± 7.5 mm (mean ± SD) were present in the cameras' field of view, whereas among the adult S. 310 intermedius with test diameters from 37 to 74 mm (64.4 ± 4.3 mm), there was a small number 311 (up to 12%) of juveniles with test diameters of 10-15 mm. On the surface of the feeders, both sea 312 urchin species were relatively evenly distributed in one plane and did not form aggregates (three-313 dimensional groups); however, they formed associations (dense two-dimensional groups). 314 Grouping behavior in S. intermedius was expressed to a greater extent than in M. nudus: when 315 from 14 to 25 of individuals were present on the surface of the feeders, the mean group size of S. 316 intermedius was approximately 2 times higher than that of M. nudus (2.27 ± 0.4 versus 1.12 ± 317 0.1, Mann-Whitney test, U = 0, P < 0.0001; see Table S4 for raw data and statistics).  Table S1 for raw data). On the eve of the storm periods, there 328 were 54 ± 9 (mean ± SD) of M. nudus specimens and 76 ± 37 of S. intermedius specimens 329 whereas during the storms, the average numbers for both species (26 ± 15 and 34 ± 21 for M. 330 nudus and S. intermedius, respectively) were significantly lower (Fig. 2, see Tables S5, S6 for 331 raw data and statistics). Approximately one day after the storm, sea urchins of both species 332 restored their numbers on the feeders (Fig. 1A-C, Fig. 2, Tables S5, S6).  (Tables 1 and 2, see also Table S7 for interspecies comparison).

338
Both sea urchin species responded to increased wave activity by a sharp decrease in the 339 number of steps, length of one move and entire distance travelled (Tables 1 and 2, see also Table   PeerJ (Fig. S2B). The number of sea urchins in the cameras' 366 field of view was relatively stable during the first 12 h, and sea urchin distances from the site of 367 attack did not change much (Fig. 3A). Then, the distances began to increase sharply, and their Manuscript to be reviewed 371 3A and S2C). The number of sea urchins on the feeder began to increase between 16 and 28 h 372 after the starfish left the sea urchin remains, and this coincided with gradual decrease of sea 373 urchin distances from the site of attack (Fig. 3A). In general, the alarm reaction of S. intermedius 374 from the onset of the starfish attack to restoration of the initial sea urchin population on the 375 surface of the feeder (Figs. 3A and S2D) lasted for approximately 90 h.

376
On September 21, a starfish attack occurred at the short edge of the feeder. Sea urchin 377 distances from the site of attack were almost unchanged during the first 9 h and then sharply 378 increased, and this coincided with maximum number of sea stars (P. pectinifera and L. fusca) 379 consuming an ailing specimen (Fig. 3B). After 24 h, no S. intermedius specimens remained 380 closer than 40 cm from the site of attack. They formed associations on the feeder and the nearest 381 stones. Eight hours after the beginning the attack, the number of sea urchins in the cameras' field 382 of view began to decrease, and after 55 h, there remained approximately 50% of sea urchins.
383 Consumption of the prey by the sea stars lasted 70 h. Restoration of sea urchin abundance and 384 distribution on the feeder began 10 h after the sea stars left the remains of the prey. The total 385 duration of the sea urchin alarm reaction was 88 h (Fig. 3B).

386
It should be noted that the sea stars P. pectinifera and L. fusca were constantly present on 387 the feeders. The starfish Asterias amurensis Lutken, 1871, Distolasterias nipon (Döderlein, 388 1902), Lysastrosoma anthosticta Fisher, 1992 and Aphelasterias japonica Bell, 1881 also often 389 appeared on the feeders. With a few exceptions, these starfish did not cause visible reactions in 390 healthy sea urchins. The behavior of P. pectinifera was the most aggressive. In one case, during 391 28 min, P. pectinifera attacked an S. intermedius specimen, which lost approximately 20% of its 392 spines, but finally, it was left alone. In two cases, sea stars P. pectinifera completely crawled on 393 S. intermedius individuals in such a way that starfish mouth was located directly above the sea 394 urchin's anal orifice. After 11 and 15 min in the first and second cases, respectively, the sea stars 395 left the potential prey, which indicates that a healthy sea urchin can effectively resist the 396 penetration of a starfish stomach through the anus.  Table S8 for raw data 403 and statistics). During this period, a sharp increase in the average step length just after the 404 treatment was recorded (Fig. 5). There were 2 time intervals with the highest locomotion speed, 405 the first 55 min after the treatment when a half of M. nudus specimens left the cameras' field of 406 view, and the last 128 min when the rest of sea urchins escaped (Fig. S3A). These intervals were 407 interrupted by the relatively stable 1 hour period when sea urchins almost stopped moving (Fig.   408 5).

409
Nine hours after the beginning of the experiment, there were no M. nudus specimens on 410 the feeders (Fig. 4, Table S8), and then, during a much longer period (7-19 days), only single M. 411 nudus individuals appeared (Fig. 1A-C). The restoration of the initial sea urchin numbers on the 412 feeders occurred only after the next storm event.  Table S1).  Table S1) demonstrated that, in the absence of crushed conspecifics, sea urchins M. nudus came 420 back 2-3 days after their removal from the surfaces of the feeders (Fig. 6, see Table S9 for raw 421 data and statistics).  Table S1); moreover, 424 statistical analysis revealed small but significant increase in S. intermedius numbers in 3 cases 425 when M. nudus was absent on the feeders (see Table S10 for statistics). Spatial pattern of S.   Table S1 for original data). Triangles indicate the time points when sea urchins M. nudus were removed from the feeders. Upside down triangles denote the time points when sea urchins M. nudus were crushed near the feeders.
Green squares denote the time points when the feeders were changed. Green rhombuses indicate the time points when the feeders were changed after the mimicking of stormy weather conditions. Shaded areas denote storm periods. Solid violet horizontal lines indicate periods of poor visibility because of high water turbidity. X-axis: month and date.