Animals
Six wild female Amphioctopus fangsiao (d’Orbigny 1841) were caught using a trawl net on 2 February 2021 in the central area of the Seto Inland Sea (34°25'N, 133°47'E), off Kurashiki City, Okayama Prefecture. They were packed individually in a plastic bag with seawater and oxygen gas, then transported by a bullet train (Shinkansen) to Tokyo University of Marine Science and Technology. Body wet weight (WW) of the females was 125 ± 32 g (range 75 − 162 g). They were individually maintained in a 40 L rectangular glass aquarium (450 × 300 × 300 mm) containing artificial seawater (Sealife 714; Nihonkaisui Co., Ltd., Tokyo, Japan) equipped with a water recirculation system (ProFit Filter Big; Kotobuki Kogei Co., Ltd., Nara, Japan). Frozen shrimp (Plesionika semilaevis) were supplied to the females once a day until the start of spawning. Water temperature, pH, dissolved oxygen concentration, light intensity, and salinity were measured once a day at 9 a.m. The respective mean values (± SD) were as follows: 11.4 ± 1.8°C, pH 8.08 ± 2.19, 11.0 ± 0.5 mg L− 1, 99.0 ± 43.7 lx, and 32.3 ± 1.3 g L− 1. Females started spawning from 25 March 2021, and were left (without further feeding) to incubate their eggs until hatching.
Seven hatchlings (mean WW 0.12 ± 0.03 g, range 0.08 − 0.15 g; hatched during the period 23 to 30 June 2021) were reared individually in cylindrical plastic containers (diameter 70 mm, length 105 mm) in which numerous small holes (diameter 1 mm) allow free circulation of seawater. These containers were placed together in a water bath filled with artificial seawater (500 × 807 × 207 mm, water volume 60 L). To prevent escapes, the upper aperture of the container was occluded with a polyurethane sponge. Small frozen shrimps (Acetes japonicus) were supplied twice a day as food for these juveniles. The rearing water was changed once a week, and the water temperature, pH, dissolved oxygen concentration, light intensity, and salinity were measured once a day at 9 a.m. during rearing periods, with the following respective values (mean ± SD): 22.4 ± 1.5°C, pH 8.26 ± 0.11, 9.3 ± 0.3 mg L− 1, 127.3 ± 120.7 lx, and 32.9 ± 1.37 g L− 1. Juvenile rearing was terminated at 50 d after hatching, by which time the WW of the juveniles had reached 4.69 ± 0.50 g (range 3.86 − 5.22 g, Fig. S1).
As experimental prey organisms for A. fangsiao juveniles, juvenile clams (the bivalve Ruditapes philippinarum), small marine snails (the gastropod Littorina brevicula), and mysids (the mysid shrimp species Neomysis awatschensis) were selected, because they are sympatric with A. fangsiao. The juvenile bivalves and gastropods were captured, respectively, at a tidal flat (35°20'N, 139°38'E) and a rocky shore (35°18'N, 139°29'E) in Tokyo Bay, Kanagawa Prefecture, Japan (permission number 3–40, 2021, Kanagawa Prefecture). The mysids caught from the coastal waters of Kyusyu Island of Japan were purchased from an aquarium trader. These prey organisms were maintained in aquaria with a supply of appropriate feed: microalgae Chlorella vulgaris (fortified with n-3 unsaturated fatty acids) for juvenile bivalves; pieces of kelp for the gastropods; and Artemia franciscana nauplii and artificial compound diet (Vital Prawn S0, Higashimaru Co., Ltd., Kagoshima, Japan) for the mysids.
Feeding experiment
To investigate ontogenetic changes in predatory capacity and prey choice of A. fangsiao juveniles, seven octopuses were used for the feeding experiment. The experiment was carried out in an experimental arena on the day of hatching (aged 0 d) using four juvenile octopuses (octopuses A − D, Fig S1) and then repeated every five days until the juveniles were 50 d old (11 times per juvenile, for a total of 44 times). Three additional octopuses (E − G) were used as substitutes on three occasions, following failure to obtain results (such as the loss of power supply to the video camera or escape of the octopus from the experimental arena).
The experimental arena consisted of a cylindrical acrylic container placed concentrically inside a larger container (Fig. 1). Both the outer (inner diameter 292 mm, height 100 mm) and inner (inner diameter 144 mm, height 50 mm) containers were filled with artificial seawater (water depth 55 mm). To prevent escapes, nylon mesh (height 70 mm) was attached to the inside wall of the inner container, and an air stone was placed in the outer container to provide oxygen while recirculating the water between the inner and outer containers.
Seven individuals of each of the three prey organisms (total 21 animals) were placed in the experimental arena and then the octopus was added inside a 100 mL glass beaker placed in the center of the arena for acclimation. During the acclimation period (10 min), prey organisms were able to move freely around the arena and the octopus could see them but could not contact them. Then, the octopus was released from the beaker and its behavior was recorded for 24 h. A video camera (HDR-SR12; Sony Corporation, Tokyo, Japan) was placed facing upward in a space beneath the arena and the octopus’s behavior was recorded through the translucent acrylic base of the arena (Fig. 1). White LED lights illuminated the roof (white PVC board) of the aquarium and were programmed to produce a photoperiod of 13L:11D. During the dark period (from 7 pm to 5 am), the white LEDs were replaced with an infrared LED. The video camera was set to infrared mode throughout the entire experimental period (for 24 h). The camera was connected to a video capture board (GV-HDREC; I-O Data Device Inc., Kanazawa, Japan) and HDD recorder (ELP-CED020U; Elecom Co., Ltd., Osaka, Japan). The octopus was removed immediately after the experimental period, and its WW was measured before returning it to the rearing container. Water temperature, pH, dissolved oxygen concentration, light intensity, photon flux density, and salinity of the experimental aquarium were measured immediately after finishing the 24 h video recording, producing the following respective mean values: 22.4 ± 1.0°C, pH 8.25 ± 0.12, 8.95 ± 0.35 mg L− 1, 102 ± 11.1 lx, 2.65 ± 0.51 µmol m− 2 sec− 1, and 32.3 ± 1.36 g L− 1.
As prey organisms, seven individuals each of bivalves and gastropods of various sizes were concomitantly supplied for the experiment. The mysids presented as prey were randomly sampled from the mysid-rearing tank. The size of prey organisms was measured using a stereomicroscope equipped with image analysis software (NIS-Elements D5.10.00; NIKON Corporation, Tokyo, Japan). The shell length (SL) of the bivalves supplied to the octopuses was 6.98 ± 2.34 mm (range 2.21–12.24 mm), and the shell height (SH) of the gastropods was 6.43 ± 0.95 mm (3.81–9.59 mm). The experimental period ended when the octopus was removed. The number of surviving prey organisms was counted and the size of consumed bivalves and gastropods was measured. The total length (TL) of surviving mysids was also measured (mean TL 8.20 ± 0.89 mm; range 6.06–10.64 mm).
Video-recorded behavior of the octopuses during 44 trials (total 1056 h) was observed using media player software (VLC ver. 3.0.15). The time (in seconds) that the octopus spent consuming individual prey organisms, from capture to discarding (shells or exoskeletons), was recorded as handling time (H). Where an octopus discarded a prey organism with a large amount of residual flesh, and H was obviously short, the value of H was excluded from the data to avoid overestimation of the predatory capacity of the octopus. The rank order in which each prey organism was consumed by each octopus was also recorded and the size of bivalves and gastropods was measured. To assess the predatory capacity of octopuses on mysids, the number of attacks on mysids was counted to calculate the proportion of successful attacks.
Force required to prey on bivalves and gastropods
To quantify the predatory capacity of octopuses needed to prey on bivalves and gastropods, the forces required to open bivalve shells and to detach gastropods from the substrate were investigated. Live bivalves (55 individuals with a size range of 2.21–12.88 mm SL) were used. After SL measurement, one valve was attached to a polyethylene thread and the other valve was firmly attached to an aluminium block using instant glue (Aron Alpha Extra; Toagosei Co., Ltd., Tokyo, Japan). The aluminium block was secured to a table with clamps and the thread was connected to a force gauge (AD-4932A-50N, A&D Co., Tokyo). The maximum force (unit N) required to open the bivalve was measured by pulling the force gauge. The force required to detach each gastropod was also measured using a force gauge. Live gastropods (27; size range 4.73–9.64 mm SH) were used. A polyethylene thread was attached to the top of the gastropod shell with instant glue, then the gastropod was put into the arena (the same as that used for the feeding experiment). After ensuring that the foot of the gastropod was adhering to the base of the arena, the force gauge connected to the thread was pulled up to detach the gastropod and the maximum force used was recorded.
Energy content of prey organisms
To estimate the energy content of the prey organisms, chemical analysis was conducted on their flesh. The whole body of mysids and the soft bodies of bivalves and gastropods were sampled and preserved at -60°C until analysis. The content of moisture, crude protein, total lipid, and ash were determined by, respectively, the air-oven method, combustion method, acid hydrolysis method, and direct method (ashing at 550°C). Carbohydrate content was calculated by the moisture, crude protein, total lipid, and ash content. Energy content (unit J) was estimated based on the protein, lipid, and carbohydrate contents in dry matter basis (Table S1). These were the standard analysis method for food composition in Japan (Ministry of Education, Culture, Sports, Science and Technology, Japan 2020).
The relationship between size and dry weight (DW) of gastropods and mysids was also assessed. For gastropods, the soft bodies of 20 individuals were dissected from the shell after measuring SH, and the DW of the soft body was measured by drying at 100°C for 24 h. For the mysid, the DW of the whole bodies of 5 individuals was determined for a total of 100 individuals. For the bivalves, the DW of the soft body was estimated from the shell size (SL) according to Hasegawa and Higano (2010): DW = 3.6×10− 6×SL3.30.
Data analysis
All statistical analyses were performed using R statistical software (R4.3.2 Core Team 2023) with a 5% significance level. Log-transformed values of weight (DW and WW), length (SL and SH), and force were used to standardize their additivity in the models.
To determine the relationship between gastropod DW and SH, a linear model (LM) was applied using the lm function, while DW and SH were used as response and explanatory variables, respectively. A LM was also applied to estimate the relationship between shell size (SL or SH) and force required to open bivalves or detach gastropods.
To estimate the relationships between the size of octopus and sizes (SL and SH) of the bivalves and gastropods consumed by the octopus, a linear mixed-effect model (LMM) was applied using the lmer function in the lme4 package (Bates et al., 2015). In this analysis, mean or maximum shell sizes, consumed by an octopus during the test period of 24 h, were used as response variables. When the octopus consumed one bivalve or gastropod, the size was treated as a mean (not a maximum) value. The explanatory variable was the WW of the octopus, and the identity (A − G) of the octopuses was included as a random intercept effect (Zuur et al. 2009). The statistical significance of the explanatory variable was evaluated using the Anova function (type II, F tests) in the ‘car’ package (Fox and Weisberg 2011). To assess the change in the success ratio of attacking to catching of mysids by the octopus, a generalized linear mixed effect model (GLMM) with the binomial family (logit link) was applied using the glmer function in the lme4 package. In this analysis, the number of successful and unsuccessful attacks were binary response variables, and the WW of the octopus was an explanatory variable, while the identity of the octopus was included as a random intercept effect.
To examine changes in time (H) spent to consume individual prey organisms, a GLMM with the Gamma family (log link) was used. In this analysis, handling time to consume individual bivalve (Hb), gastropod (Hg), and mysid (Hm), were used as separate response variables in three GLMMs, while octopus WW and prey size (SL or SH) were used as explanatory variables. For mysids, prey size was excluded from the model because individual mysids were closely similar in size. The Gamma GLMM was used to assess the octopus rate of energy intake per unit time (i.e., E/H). In this model, the E of an individual bivalve (Eb) or gastropod (Eg) were used as separate response variables in two GLMMs, while the Hb or Hg were used as an offset. The octopus WW was an explanatory variable. In these GLMMs, the identity of the octopus was included as a random intercept effect. The mysids were of similar size, so it was assumed that their energy content (Em) was similar and the Em/Hm ratio for mysids was estimated simply using the mean Em of individual mysids and the relationship between octopus WW and time Hm spent to consume the mysid.
To assess the effects of octopus size on total time spent to consume prey organisms (Ht, summation of all Hb, Hg, and Hm) and total intake of energy (Et, summation of all Eb, Eg, and Em) during the 24 h experimental period, GLMMs with the Gamma family (log link) were applied. In these analyses, values of Ht or Et were the response variables; and the WW and identity of the octopuses were an explanatory variable and a random effect, respectively. In the analysis to test the Et/Ht ratio throughout 24 h including all prey species and feeding events, values of Et and Ht were used as a response variable and an offset, respectively. The explanatory variables and the random intercept effect were the WW and identity of the octopus, respectively.
It can be expected that the prey choice of each juvenile octopus changes depending on its predatory capacity, manifested as: (1) the number of individuals of each prey species consumed (chosen) by the octopus; (2) the time spent in consuming each prey species; and (3) the energy it obtained from each prey species. To investigate (1), the change in proportion of prey species, a binomial GLMM (logit link) was used, in which the proportion of each type of prey among the total number of prey animals (Nt) consumed was a response variable (i.e., Nb/Nt, Ng/Nt, and Nm/Nt for bivalves, gastropods, and mysids, respectively). The explanatory variables were octopus WW, prey species (as a categorical fixed factor), and their interaction. To assess (2), the change in proportion of time spent consuming a particular prey species, a binomial GLMM was applied using the proportion of H for the prey species among the total Ht over 24 h as a response variable (i.e., Hb/Ht, Hg/Ht, and Hm/Ht for bivalves, gastropods, and mysids, respectively). The explanatory variables were octopus WW, prey species (as a categorical fixed factor), and their interaction. The binomial GLMM was also applied to estimate (3) the change in energy dependence of the octopus on each prey species. In this analysis, the proportion of E obtained from a certain prey species by the octopus among the total Et over 24 h was used as a response variable (i.e., Eb/Et, Eg/Et, and Em/Et for bivalves, gastropods, and mysids, respectively). The explanatory variables were octopus WW, prey species (as a categorical fixed factor), and their interaction. In these GLMMs, identity of the octopus was used as a random intercept, and the statistical significance of each explanatory variable was evaluated using the Anova function (type II, Wald χ2 tests) in the ‘car’ package.
The rank order in which each prey organism was consumed by each octopus can also be interpreted as a consequence of prey choice by the octopus. And the choice may depend on the criteria E (energy maximiser: how much energy can be obtained from a prey animal), H (time-minimizer: how fast the octopus can consume a prey animal), or E/H (rate maximiser: how fast energy can be obtained from a prey animal), following the assumptions of foraging theory (Pyke et al. 1977; Stephens and Krebs 1986; Stefens et al. 2007). For instance, if the octopus shows a higher priority for E/H, the E/H of a feeding event is expected to decrease among the animals subsequently preyed upon. This is because the octopus is expected to choose first the prey with the highest E/H. To test whether or not the juveniles followed these three criteria, the Gamma GLMM was used with the log link function. In this analysis, three criteria Hi, Ei, and Ei/Hi were used as separate response variables in three GLMMs, where Hi and Ei were handling time and energy of a prey animal selected i-th by an octopus, respectively. The explanatory variables were the order (i) of the prey organism consumed and the octopus WW. Taking into account the effect of ontogenetic development of predatory capacity on octopus prey choice (see Results section, Figs. 2, 3), this analysis was applied separately to early juveniles of WW < 2.5 g (mainly younger than 25 d) and grown juveniles of WW > 2.5 g (mainly 30 − 50 d). The boundary (2.5 g WW) between early and grown juveniles was chosen to bisect the test range of WW of juveniles (0.08 − 5.22 g), to secure enough data for the analyses of these two stages.