Life in coastal pebble sediment: Unique interstitial organism community and selective feeding on meiobenthos by interstitial fishes (Luciogobius: Gobiidae)

Interstitial areas of coastal pebble sediment in the Japanese Archipelago are inhabited by extremely elongated gobies of the genus Luciogobius, which are characterized by an increased number of vertebrae and reduction of scales, eyes, and fins. To explore the little-known interstitial life of Luciogobius gobies, we investigated the diets of two interstitial Luciogobius species, L. elongatus and L. grandis, and the interstitial organism communities of the gobies’ microhabitats in an exposed gravelly coast in Shirahama, southern Japan. The interstitial organism community in pebbly sediment was dominated by minute arthropods such as harpacticoids, isopods, and ostracods, presenting a marked contrast to the communities in sandy sediments, which are dominated by nematodes and turbellarians. The gut contents of the two goby species were composed exclusively of interstitial organisms, especially harpacticoides and isopods. Although each prey assemblage was roughly similar to the interstitial organism community in the corresponding microhabitat, marked preferences for harpacticoids and flabelliferan isopods were detected in L. elongatus and L. grandis, respectively. Irrespective of their intense feeding of harpacticoids, rare catches of large isopods were suggested to be nutritionally important for the gobies. These results suggest that the Luciogobius gobies are the first known fishes that depend exclusively on interstitial organisms, and that selective feeding upon meiobenthos may facilitate the coexistence of several interstitial goby species in pebbly sediment.


Gut content
137 After removing formalin, we dissected the stomach of each goby, extracted all bodies and 138 fragments of prey individuals, sorted them into taxonomic groups, and counted them under a 139 binocular microscope. Thus, we obtained prey assemblage data sets for two goby species (L. 140 elongatus and L. grandis), for three seasons (February, June, and October), and for daytime and 141 nighttime. Based on the datasets, we calculated the percentage of fish that ate at least one 142 individual of the prey group (%F). Furthermore, we estimated the numeric proportion of each 143 prey group (%N), i.e., the numeric percentage of each prey item out of the total number of 144 individuals. To detect factors involving in interspecific, seasonal and diurnal variances of the 145 prey assemblages, we conducted nonmetric multidimensional scaling (NMDS) using Soft [9] 146 and obtained a two-dimensional graphical representation of the multivariate prey assemblages.

Interstitial organism community in goby microhabitats
149 To uncover the community structure of potential prey organisms, we sampled interstitial 150 organisms in the gobies' microhabitats, since preliminary observations suggested that their 151 stomach contents were composed mainly of interstitial organisms. We sampled about 2 L of 152 pebbly sediment from the same site as goby sampling, put it into a bucket, poured sea water into 153 the bucket, stirred the water with a shovel for 30 s, and then filtered the supernatant through a 154 plankton net (100-µm mesh). We repeated the filtering procedure three times. Collected 155 interstitial organisms were immediately fixed in 5% formalin. In the laboratory, the interstitial 156 organisms were dyed with Rose Bengal to more easily observe translucent bodies [8]. The 157 samples were sorted into taxonomic groups in the same manner as gut contents and were 158 counted.

159
To analyze sediment granularity, pebbly sediments from the gobies' microhabitats 160 were collected and sorted with standard sieves in running water. The sorted sediments were 161 dried and weighed. From the data set of the fractions, the mean particle diameter and sorting 162 index were calculated for each sediment sample.

Estimation of prey biomass
165 To estimate the biomass of prey organisms, we measured the length (L) and width (W) of 10-20 166 individuals from each taxonomic group (if fewer than ten individuals were available, we 167 measured as many individuals as possible), and calculated the average volume using the 168 following formula [8]:  (Fig 3), suggesting that most collected gobies were adults. In contrast, the mean total length of 189 collected L. grandis gobies increased from February to October (Fig 3). In L. grandis, most 190 gobies collected in October were adults, while gobies collected in February and June contained 191 juveniles. In addition to gut contents, we monitored ovarian maturation of female gobies. L. Comparison between prey assemblages and interstitial organism 279 communities 280 The prey assemblages were roughly similar to the interstitial organism communities, which 281 consisted of harpacticoids, flabelliferan and asellotan isopods, amphipods, brachyurans, 282 ostracods, caecid gastropods, and annelids (Figs 7, 8). The prey assemblages were dominated in 283 number by harpacticoids, which were the most abundant interstitial organisms. While annelids, 284 nematodes, and foraminiferans were common in the sediment, they were rare in gobies' guts. Irrespective of the rough similarity between the prey assemblages and the 298 interstitial organism communities, the proportions of some groups differed. In L. elongatus, the 299 proportions of harpactioids in gut were greater than those in sediment, especially in June and 300 October and in both daytime and nighttime (Fig 7), suggesting selective predation of 301 harpacticoids. In L. grandis, the proportions of flabelliferan isopods and amphipods in gut were 302 greater than in sediment in both daytime and nighttime (Fig 8), suggesting that they were also 303 selectively predated. Because L. grandis gut contents in June were frequently empty, the 304 proportions of amphipods and flabelliferan isopods were 100%, respectively (Fig 8).

305
The numerical dominance of harpacticoids in diets does not necessary indicate 306 nutritional importance, because harpacticoids are minute. The size distribution of interstitial 307 organisms collected at the gobies' habitats (Fig 9) showed that flabelliferan isopods were 308 exceptionally large among the interstitial organisms. Thus, the prey assemblages of the two 309 goby species were dominated in biomass by isopods (especially suborder Flabellifera), which 310 constituted less than 50% in number but 60-80% in terms of volume in the diet (Fig 10). 311 Numerical and volumetric comparisons between prey assemblages and interstitial organism 312 communities (Fig 10) showed that the numerical dominance of harpacticoids in the L. elongatus