Aquat. Sei. 66 (2004) 315-326 \t i \~7 (\jy\fj\ 1015-1621/04/030315-12 V L I t \ v ^ ¥ l n c 7 t=P C « ------DOi i o. 1007/s00027-004-0682-0 VLAAMS INSTITUUT VOOR U t ¿ t e ¡Aquatic S ciences e EAWAG, D M » * * 2004 FLANDERS MARINE INSTITUTE

Oostende - Belgium

ó 1 7 8 5 Research Article

Temporal variation in feeding rhythms in atidal marsh population of the common goby (Kroyer, 1838)

Henrietta Hampel1* and Andre Cattrijsse2

1 University of Gent, Department of Biology, Marine Biology Section, Krijgslaan 281, S8, B-9000 Gent, Belgium 2 Flemish Marine Institute, VLIZ (Vlaams Instituut van de Zee), Vismijn, Pakhuizen 45-52, Oostende B-8400, Belgium

Received: 15 May 2003; revised manuscript accepted: 31 October 2003

Abstract. Pomatoschistus microps (Teleostei, Gobiidae) of the diel cycle is inferior in comparison with the tidal intensively uses the mesohaline marsh of Westerschelde influence on the feeding behaviour of the common goby. estuary as a nursery and foraging ground. The sampling A significant difference in foraging activity occurred be­ campaign covered the semi-lunar, diel and tidal cycles. tween the spring and neap tide. The common goby mi­ The density of P. microps and potential hyperbenthic prey grated in lower abundance into the creek during spring species in the marsh creek, fullness index, evacuation tide but foraged more intensively. At both spring and neap rates and daily ration of common goby were calculated. tide, a significant difference was found in the fullness in­ Mesopodopsis slabberi, Neomysis integer and Corophium dex between day and night. At spring tide, gobies feed volutator were the most dominant prey items in terms of more during the day, while they forage more intensively biomass. Numerically, copepods dominated the diet. Mi­ at night at neap tide. All the three cycles (tidal, diel and grating fish enter the marsh creek with a relatively empty semi-lunar) influenced the feeding rhythm of the com­ stomach and leave the marsh with a higher stomach con­ mon goby. The tidal influence is superior over the diel tent. Pomatoschistus microps seemed to feed more inten­ variation, while the explanation of the combined effect of sively during the day than the night, however the influence diel and semi-lunar cycle needs further studies.

Key words. Common goby; feeding habit; tidal diel and semi-lunar cycle; marsh.

Introduction The biology and ecology of Pomatoschistus microps has been intensively studied. The population structure The common goby, Pomatoschistus microps, is a small (Bouchereau et al., 1989; 1993; Pampoulie et al., 2001), gobiid fish commonly found in all European coastal wa­ breeding behaviour (Magnhagen, 1998; Pampoulie et al., ters, estuaries, fjords, salt marshes and high shore pools 2001), age-structure, growth and reproduction (Arruda (Petersen, 1919). In the intertidal area of the Wester- et al., 1993), food selection (Magnhagen, 1985) and the schelde estuary (SW Netherlands), the common goby is diet spectra of newly hatched gobies (Menher, 1992) all one of the most abundant fish species (Maes et al., 1997; received attention. Despite of these intensive investiga­ Hostens and Mees, 1999). tions, no work has been published on the influence of different cycles (tidal, diel and lunar) on the migration or on feeding rhythm of the common goby. * Corresponding author phone: +32 2 74 62 139; e-mail : [email protected]; [email protected] Three main functions have been ascribed to intertidal Published on Web: August 24, 2004 migration of fish: reproduction, avoidance of predation, 316 A. Cattrijsse and H. Hampel Short term feeding habit of the common goby

and feeding (Gibson, 1993). In the Westerschelde estuary, This study aims to describe the feeding habit of the common goby uses the marsh creeks between June Pomatoschistus microps during the semi-lunar period, to and October (Cattrijsse et al., 1994). During this period, investigate the separate and combined effects of the tidal, several waves of early postlarval gobies enter the marsh diel and semi-lunar cycle on the feeding behaviour of creeks and remain there to spend their early life stage. common goby in a salt marsh creek, and to quantify the Predator avoidance may result from the huge numbers of change in tidal food consumption. juveniles that stay in small pools and seepage waters (Cattrijsse et al., 1997; Kneib, 1997). The marsh and the tidal flats must harbour favourable conditions for nesting, Materials and methods but. Cattrijsse et al. (1994) never found nests. Cattrijsse Study area et al. ( 1994) also mentioned that the common goby utilizes The studied marsh ‘Het Verdronken Land van Saeftinghe’ the marsh creeks as feeding grounds, preying mainly lies in the mesohaline part of the Westersehelde estuary upon amphipods and mysids. The common goby shifts its (Fig. 1). It is the largest estuarine brackish marsh in Eu­ feeding preference from copepods towards mysids and rope with a surface area of 2800 ha (Dijkema et al., 1984). infaunal species during its life span. All samples were collected in an intertidal creek measur­ In general, three different cycles may influence the ing 1600 m long and 4 m deep at the sampling point. Dur­ migration and feeding habits of intertidal fish species. ing tidal cycle the creek falls completely dry for at least The tidal effect is the most obvious for the movement of 5 hours. Water starts entering the marsh 3 hours before fish into intertidal areas (Gibson, 1988). The availability high water and leaves the creek 4 hours after high water. of intertidal feeding areas depends upon the tidal phase. There is no connection of any other major creek in the Several studies have investigated the tidal influence on marsh, and there are no tidal flat and pools at the mouth of the feeding behaviour of fish in the salt marshes (Weis- the creek. The channel showed a sigmoid shape close to berg et al., 1981; Rozas et al., 1988; Rozas and LaSalle, the estuary, however the creek was straight and the slopes 1990; Rountree and Able, 1992). of both banks were symmetrical at the sampling point. A second important cycle is diurnal variation. The ac­ tivity of most marine fishes is synchronised with the daily cycle of light and darkness (Gibson, 1993). The diurnal in­ Sampling fluence on the feeding activity of fish in tidal marshes has Sampling took place from a bridge spanning the creek. received some attention (Kneib and Stiven, 1978; Weis- A stow net with a 1 x 1 mm mesh size, an opening of berg et al., 1981; Cadigan and Fell, 1985; Antholz et al., 1 X 1 m and a length of 5 in was used. The net was placed 1991). Weisberg et al. (1981) argued that die! rhythms on the bottom to passively sample migrating fish and po­ would result in reduced feeding on the marsh surface and tential hyperbenthic prey species in the first metre of the an increased utilization of subtidal prey. lower water column. Hyperbenthic species were defined The semi-lunar phases form a third cyclic effect on the as animals found in the first metre of the lower water col­ migration and feeding behaviour of intertidal fish how­ umn, larger than 1 mm and capable of actively migrating ever its influence is poorly studied. Changes in abundance with the moving water. Two weights attached to the sides of fish species in marsh areas related to the semi-lunar of the frame prevented the net from being lifted by the cycle of spring and neap tides have been documented by currents, while ropes kept the net in place during sam­ Rooker and Dennis (1991) and Kneib and Wagner (1994). pling. The opening size of the net, the mesh size and the The influence of the semi-lunar cycle on the feeding ac­ thickness of the fibre were used to calculate the necessary tivity of fish has not often been investigated. Only Hamer- length of net to maintain the pressure inside the net under lynck et al. (1993) observed a semi-lunar feeding rhythm all current conditions (Tranter, 1979). This method re­ in juvenile Pomatoschistus lozanoi from a sandy beach. duced the probability that animals avoided the net. Integrated work about the influence of the tidal, diel and The net fished for one hour, was rinsed and then re­ semi-lunar cycle on the feeding habit of intertidal fish placed in the same position. After high water, the net was does not exist. turned to sample the ebb current until all water had left Marsh creeks and other intertidal habitats are believed the creek. Sampling lasted a complete tidal cycle, yield­ to play a significant role in the early life stages of many ing 3 flood (FI, F2 and F3) and 4 ebb (E4, E5, E6 and E7) fish and crustacean species as feeding grounds. A full samples on every sampling occasion. Water current and appreciation of the value of such habitats necessitates water height were measured every 15 minutes. Current more information than tidal feeding rhythms. Day-night speed was measured every 15 minutes with a current me­ changes in feeding activity and fluctuations in stomach ter placed about 20 cm above the bottom next to the net. contents over a semi-lunar cycle are needed to fully un­ Water height was measured with a ruled stake placed next derstand the functioning of such feeding grounds and to to the net in the deepest part of the creek. Temperature quantify their importance in the food web of the marsh. was recorded every hour. Aquat. Sei. Vol. 66, 2004 Research Article 317

Vttalivtgeft Hartsweert | ‘I LEGEND: Cr WTL * 5 m // / A \ \ North /T*0^ M a v .Bath t \ sea v j i . i\> \ \ ^ * x j ¡ V » . ■S/ v- v*-~* 1 \ Temeustf^--*^*' w** v.*^ j ^ L'" BEL. ¿5 Q 5 Km 4** -* ' " r. Doei \ \ 4* E 4^15’ Ç-/- N ÿï.\ y Saeftinghe 'V À 'A "T?...... 0 200 m N, V y y y t y . , ¿ m í \ f Æ r> | U.)J Aj *, /■ /?aç-

I ¿A-V /V ~ vf jh a " S f t sampling point

Figure 1. The Westerschelde estuary in the SW Netherlands and the positioning of the marsh “Het Verdronken Land van Saeftinghe” and the sampling site.

August was chosen for sampling because, in Saeft­ high water always occurred around midnight or midday, inghe, the density and biomass oí Pomatoschistus microps hence taking full night or daytime tidal samples did not peaks during this month (Cattrijsse et al., 1994). To cover present a problem. During spring tide, high water was the semi-lunar cycle, observations were made between always in late afternoon or early morning. Complete night 7 and 19 August 1999 from full moon until new moon, sampling was thus not possible during spring tide. Eight covering the possible combinations of the diel, tidal and and 16 August were considered as night samples since the semi-lunar cycles (Table 1). Eight and 10 August were organisms moved into the creek during night. On 10 and considered as spring tide, while 16 and 17 August were 13 August, night hours fell when ebb had started already defined as neap tide according to the predicted water and evening hours did not influence the migration, hence height. Due to weather conditions, the spring tide period these sampling days were considered as complete day slightly shifted and peaked on 10 August. samplings. At both spring and neap tide, day and night samples in the field, all the fishes were anaesthetised in a ben- were taken, however, with some restrictions. At neap tide, zocaine (Ethyl amino-4-benzoate)-water solution to pre­

Table 1. Selected sampling days and periods during the semi-lunar cycle with predicted water heights (1) (cm above the Dutch ground level NAP) and observed maximum water height (2) (cm above the Dutch ground level NAP) in the marsh creek in August, 1999. Filled blocks in diurnal phases indicate night hours during the sampling campaign. Tidal phase is labelled with F 1 F3 as flood hours and E4 E7 as ebb hours. The moment of high water (HW) is indicated by a vertical line.

Date Time Diurnal Phase Moon Phase Lunar Phase Waterheight Tidal Phase 1. 2

HW FI-F3 E4--E7 7 Aug 13:30-20:45 L„4__ j day lull 297 340 8 Aug 02:20-09:35 ■ M O night spring 319 350 I O Aug 15:30-23:00 crtrai da-v spring 319 380 13 Aug 17:05-01:15 ¡—‘rwÊ day 289 360 16 Aug 20:55-04:25 ■ M B night neap 235 290 17 Aug 09:25-17:00 ■ e n nday neap 224 295 19 Aug 12:00-19:30 n — 1 day new 276 340 318 A. Cattrijsse and H. Hampel Short term feeding habil of the common goby vent regurgitation of stomach contents. The gobies and Table 2. Taxa occurring in the stomachs of Pomatoschistus microps. captured hyperbenthic species were preserved in 10% Stars label the species captured also in the marsh creek and consid­ ered as prey species forP. microps. formalin-brackish water solution. Amphipoda Brachyura Corophium volutator * Carcinus maenas (juvenile) * Data processing and analysis Gammarus salinus Carcinus maenas (megalopa)* Gammarus zaddachi * During one sampling hour of the tidal cycle, the number Melitta pellucida Copepoda of gobies that entered or left the creek was expressed as Orchestia spp. Calanoides a percentage of the total number of individuals that mi­ Harpacticoidea grated in or out of the creek. The total number of indi­ Mysidacea Mesopodopsis slabberi* Polychaeta viduals entering the creek during the 3 flood hours was Neomysis integer * Nereis diversicolor * considered to be 100%. The number of individuals cap­ Schistomysis kervillei Spio spp. tured during each hour of flood was expressed as a per­ Isopoda Others centage of the total number captured during the entire Lekanesphaera rugicauda* Acarina spp. flood tide. Likewise, the total number of individuals cap­ Eurydice pulchra * C urnae can spp. tured on leaving the creek with ebb was taken as 100%, Cyathura carinata Insect * and the number captured during each ebb hour was ex­ Idotea spp. Fish eggs Nematoda spp. pressed as a percentage of that total. Caridea Oligochaeta For each day of the semi-lunar cycle, the density was Crangon crangon * Syphon of bivalve estimated in the lower metre of the water column. First, density was estimated using the total number of individ­ uals caught and the total volume of water passing through in a dessicator for 2 hours before weighing to calculate the the net during each hour. These densities were extrapo­ stomach fullness index. lated to the volume of the water passing through the lower For statistical analysis, stomachs were classified. one metre of the creek to obtain a total number of animals Night samples were 8 and 16 August and day samples that passed the sampling point during that hour. For each were 7, 10, 13, 17 and 19 August. According to the water hour of flood, these total numbers were summed and used height prediction, spring tide samples were taken on 8 to calculate the density that was present at high water and 10 of August and neap tide samples on 16 and 17 Au­ using the total volume of water in the lower 1 m that passed gust. Stomachs samples on 7 August were considered as during the flood. A similar calculation, using numbers spring tide-day and on 8 August as spring tide-night sam­ and volumes of water that passed the sampling point dur­ ples. Neap tide-day samples were taken on 17 August and ing the four hours of ebb, was used to obtain a second es­ neap tide-night samples on 16 August. timate of density at high water. The abundances reported For quantitative analysis of stomach contents, the full­ for each sampling day were obtained by averaging the ness index (FI), was used: flood and ebb estimates of that day. This calculation was Si performed for the common goby and for its prey. Species F I =r i- X 100, which were found in the creek and in the stomachs of Wi P. microps, were considered as prey species (Table 2). where Sj is the ash-free dry weight (ADW) of the stomach The densities of these species were summed and defined content in milligram (mg) and W¡ is the ash-free dry as the total prey density. weight of the fish in mg. The mean FI and standard error A maximum of 20 fish between 20 and 30 mm standard (se) were determined for every hour of the tidal cycle length (SL) was selected at random from all sampling hours (F1-E7) and every day of the semi-lunar cycie (7-19 of the tidal cycle and their stomach contents analysed. In the August). laboratory, all prey items encountered were counted, identi­ For qualitative stomach analysis, the food composi­ fied, if possible to species level, and the length of each in­ tion was expressed as gravimetric percentage (G%): dividual measured. The biomass (mg ADW) of fish and ADW of prey type¡ each prey item present in the stomachs was computed with G o/o = ------L ...LPJ J------X 100 length-ADW (ash-free dry weight) regressions established Total ADW of the ingested food formerly in the laboratory by Flamerlynck and Cattrijsse (1994) and Beyst et al. (1999) to estimate the gravimetric and as a numerical percentage (N %) (Hyslop 1980): percentage of prey items in stomachs. Stomach contents Number of prey type. were placed in pre-weighed aluminium foil cups, dried at N % = ------i — ------X 100 Total number of prey items ingested 110°C for 5 hours and then samples were placed in pre­ weighed ceramic dishes and ashed at 550 °C for 15 minutes Two different models were used to calculate the daily ra­ to obtain ADW. Samples were cooled to room temperature tion during the tidal cycles to obtain comparable results. Aquat. Sei. Vol. 66, 2004 Research Article 319

According to Eggers’ (1977) model, daily ration (Rd) was Results determined as: Densities of the common goby and its prey species Rd = Fl X E X T, During all sampling occasions the common gobies exhib­ ited a similar tidal migration pattern (Fig. 2). Highest where FI is the average fullness index of all fish col­ numbers were observed in the earliest flooding water lected, E is the instantaneous rate of gastric evacuation (46-87%) and during the last hours of the ebbing tide and T is time. Calculating the daily ration for the tidal (14-86%). Gobies migrated in low numbers during the cycle T equalled 7. high water (minimum 0.12 %). A significant and negative The model of Elliott and Persson (1978) was followed relation existed between the number of gobies and the to estimate the daily ration by summing the ration esti­ recorded water height of each hour (Spearman-Rank cor­ mates obtained for each hourly interval during the com­ relation, p < 0.001) (Table 3). plete tidal cycle. According to this model, the daily ration Night-time densities of common goby were higher (Rd) was determined as: than the previous or the following daytime densities (Fig. 3a). On 7 August, during day at high tide, 4.2 ind. n r3 „ _ v,ti(Fit+1- Fl.e ^ E T were found, while the density at night on 8 August was estimated to be 8.8 ind. n r3. The night time density of 16 August was 11.5 ind. nr3, while it reached only where FI is the average fullness index of all fish collected 9.4 ind. m 3 the following day. in t and in t + 1, E is the instantaneous rate of gastric evac­ Over the semi-lunar cycle, the density of Pomato­ uation and T is time which equalled 1. Andersen (1984) schistus microps seemed to be higher at neap tide than calculated gastric evacuation forPomatoschistus microps at spring tide (Fig. 3a), but no correlation was found with as: water height (Spearman-Rank correlation, p = 0.21) E - 7.385 X TL '0832 x e0-0639*1’ 2Ü>, (Table 3). On 10 August when the water height reached a maximum, the density dropped to 1.3 ind. n r3. where TL is the total length of the fish and T is the ambi­ The total density of potential prey species followed ent temperature. TL was calculated from TL-SL regres­ the change in water height during the semi-lunar cycle sions established in the laboratory. (Spearman-Rank correlation, p = 0.003) (Table 3) with Spearman-Rank tests were used to analyse the corre­ higher values at spring tide (maximum 181.9 ind. nr3) lation between the number of gobies and the average FI’s than at neap tide (minimum 2.5 ind. n r3) (Fig. 3c). At the of each hour sample with the recorded water height. To highest water level (10 August), a drop in density also oc­ evaluate the semi-lunar influence, the same analysis tool curred (88.4 ind. n r3). The average fullness index ofPo­ was used to test the relationship between daily densities, matoschistus microps showed a positive correlation with daily average FI’s, daily total prey densities and daily ra­ the total density of potential prey during the semi-lunar tion vs. the predicted maximum water height during the cycle (Spearman Rank correlation, p = 0.04) (Table 3). semi-lunar cycle. The predicted water heights were used rather than the observed ones because these were more representative of the lunar cycle. Due to climatological Pomatoschistus microps conditions, the maximum water heights during 10 and 13 100 August were higher than predicted. This caused a shift of 80 the highest water levels after the true springtide. On 13 August the moon was already in its last quarter. Also, the relationship between average FI’s vs. prey density during the semi-lunar cycle was tested by Spear­ 40 o man-Rank correlation. Mann-Whitney U tests were applied to test the influ­ \ 20 cu ence of tidal, diel and semi-lunar cycles on the fullness index. The fullness indices were compared between flood F1 F2 F3 E4 E5 E6 E7 and ebb tide, day and night hours, and between spring and Hour of tidal cycle ______neap tide. Relative num ber of individuals Since ANOVA was not applicable for the dataset, the Average Fullness index interactions between the different cycles were investi­ Figure 2. Relative numbers of Pomatoschistus microps (full line), gated by using the non-parametric Kruskal-Wallis test. the average fullness index (dashed line) and standard error for each The results of the Kruskal-Wallis test were verified by a hour of the tidal cycle. F1-F3 represents the flood hours and post-hoc Multiple Comparisons test. (Conover, 1980). E4-E7 the ebb hours of the tidal cycle. 320 A. Cattrijsse and H. Hampel Short term feeding habit of the common goby

Table 3. Results of Spearman-Rank correlation comparing different variables. Significant values are labelled with bold numbers. Daily ration was estimated by Elliott and Persson (1) and Eggers model (2).

Tidal cycle Semi-lunar cycle

Variables p value Variables p value

No. of fish Recorded water height < 0.001 Daily density of fish Predicted water height 0.21 Average FI Recorded water height 0.02 Daily average FI Predicted water height 0.21 Daily total prey density Predicted water height 0.003 Daily ration (1) Predicted water height 0.3 Daily ration (2) Predicted water height 0.06 Daily average FI Daily total prey density 0.04

(a) Pomatoschistus microps Pomatoschistus microps 14 12 - ♦ 10 8 w 3.5 6 c 4 - #> o ♦ Û 2 ♦ 0 224 12351 276 289 297 [319] 319 224 12351 276 289 297 1319 I 319 Water height [cm] Water height [cm] date = 17 16 19 13 7 8 10 date = 17 16 19 13 7 8 10 neap tide spring tide neap tide spring tide

Prey species (c) 200 ♦ ♦ 150 - ■o Ê 100 - 4 to 50 _ ♦ # 4> 224 1235 I 276 289 297 1319 I 319 Water height [cm] date = 17 16 19 13 7 8 10 neap tide spring tide

Figure 3. Density of common goby (a), the average stomach fullness index (FI) with standard error (b) and the density of potential prey species occurring in the marsh creek (c) during the semi-lunar cycle. The predicted water height (cm) is representative for the phase of the semi-lunar cycle. Night samplings are labelled with rectangles.

Qualitative stomach analyses Numerically, copepods dominated the diet during most of All taxa found in the stomachs of Pomatoschistus mi­ the sampling days (16.3—65.7%). Corophium volutator crops and also sampled in the marsh creeks are listed in and Mesopdopsis slabberi were often taken by the com­ Table 2. Table 4 shows the number of stomachs analysed mon gobies (maximum 39.4 and 45.3%, respectively). and also indicates the number of empty stomachs in each Nereis diversicolor was not a common food item in the hour of the tidal cycles. marsh creek. The two mysid shrimps, M. slabberi and The qualitative stomach analyses showed that few Neomysis integer, and the infaunal amphipod,C. voluta­ species were important in the diet of gobies (Fig. 4). tor, were the most dominant prey items in terms of bio- Aquat. Sei. Vol. 66, 2004 Research Article 321

160% 100% □ Other M Insect larvae O) 80% ® 80% Ü M. slabberi @N integer 9 60% 2 60% □ C opepod ■ C. volutator * 40% is§A/. diversicolor

£ 20%

224 1 235 j 276 289 297 1 319 [ 319 224 1 235 1 276 289 297 [ 319 1 319 Water height [cmj Water height [cm] neap tide spring tide neap tide spring tide

= 111 1.24 140 117 109 83 90

Figure 4. Numerical (N%) and gravimetric (G%) percentages of the stomach content ofPomatoschistus microps over the semi-lunar cycle. Species are labelled as follows: Corophium volutator (C. volutator), Neomysis integer (N. integer), Mesopodopsis slabberi (M. slabberi), Nereis diversicolor (N diversicolor). Predicted water height (cm) is used to indicate the lunar phase. Night samples are labelled with rectangles and the number of stomachs analysed (n) is indicated under each sampling occasion.

Table 4. Number of stomachs (contained food/empty) of Pomato­ ness index reached a maximum of 5.7. Consumption de­ schistus microps analysed in each hour of the semi-lunar cycle. creased after that moment and the fullness index started Date Hour o f the tidal cycle to drop progressively over the ebbing tide. The tidal effect on the intertidal feeding of the com­ FI F2 F3 E4 E5 E6 E7 mon goby was tested with a Mann-Whitney U-test, lumping all flood FI’s and all ebb FI’s. The amount of 07 August 20/5 20/2 12/0 0/0 20/0 20/0 17/1 08 August 20/1 6/0 0/0 0/0 20/0 20/0 17/0 food present in the stomachs was significantly higher 10 August 20/4 20/0 0/0 0/0 10/0 20/0 20/0 during ebb than during flood period (Table 5, Fig. 5a). 13 August 20/0 20/0 18/0 0/0 19/0 20/0 20/1 Feeding activity was, thus, the most intense during flood 16 August 20/3 20/0 20/1 5/0 20/0 20/0 19/1 when gobies entered the creek with low food content in 17 August 20/1 20/0 20/0 8/0 20/1 20/1 3/0 19 August 20/1 20/1 20/0 20/0 20/0 20/0 20/0 stomachs and reached a maximum around high water. After that, the food items gradually passed the gastroin­ testinal canal keeping the total amount of food present during ebb higher than during flood. mass, reaching maximum of 46.6, 38.6 and 59 gravimet­ ric percentages, respectively. There were no clear changes Diel cycle. Analyses (Mann-Whitney U test and Kruskal- in diet composition detected over the semi-lunar cycle. Wallis test) aimed to test the difference of feeding habit between the day and night hours and the combined effect of tidal and diel cycle. Quantitative stomach analyses To detect the difference in diel feeding activity, fish Tidal cycle. Analyses were used to test the difference were collected both at daylight and night time tidal cy­ of feeding between the flood and ebb, and the change cles. A Mann-Whitney U test indicated a significant (p = in this activity in correlation with the change in water 0.03) higher feeding activity during day (FI = 3.41) than height. during night (FI = 3.19) (Table 5 and Fig. 5b). All fullness indices of the first hour of the seven sam­ Investigating the combined effect of tidal and diel cy­ pling days were averaged and similar calculations were cle on the feeding activity of the gobies, Kruskal-Wallis carried out for the other hours of the tidal cycle. A clear tests showed a significant difference existed (p < 0.001) change in the amount of food in the stomachs of the com­ between the four groups of day-flood, day-ebb, night- mon gobies was observed (Fig. 2) and Spearman-Rank flood and night-ebb (Table 5). Multiple Comparisons analysis showed a positive significant correlation with showed significant differences between day-flood and change in water height (p = 0.02) (Table 3). Gobies day-ebb and between night-flood and night-ebb samples started to feed upon entering the creek, and continued (Table 5). During both day and night, the fullness index doing so until high water when the overall average full­ was higher during ebb (Fig. 5d). There was no significant 322 A. Cattrijsse and H. Hampel Short term feeding habit of the common goby

Table 5. Mann-Whitney U test, Kruskal-Wallis and Multiple Comparisons tests showing the effects of tidal, diel and semi-lunar cycles on the fullness index of the common goby.

Mann-Whitney U test Kruskal- Wallis test Multiple comparisons

Compared groups p value Compared groups p value Compared groups Level of significance

Flood Ebb <0.001 Day-flood Night-flood <0.001 Day-flood Night-flood non significant Day Night 0.03 Day-ebb Night-ebb Day-ebb Night-ebb non significant Spring tide Neap tide <0.001 Day-flood Day-ebb significant Night-flood Night-ebb significant Spring-day Neap-night 0.002 Spring-day Spring-night significant Spring-night Neap-day Neap-day Neap-night significant Spring-day Neap-day non significant Spring-night Neap-night significant

Short term feeding habit, o f the common goby

(a) (b) 4,5 3,6 3,5 S 4>° Xf h 3,4 » .£ 3,5 3,3 V) 3,2 2 3,0 - £ 3,1 LL3 2,5 H 3,0 2,0 2,9 flood ebb day night spring neap Tidal phase Diel phase Lunar phase n * 337 411 n = 547 201 n = 168 227

(e)

Being more influenced by diel variation rather than the spring tide and this could encourage the fish to feed tidal cycle, several foraging opportunities would be lost. more intensively. Again, total potential prey density Weisberg et al. (1981) made similar observations and drops sharply at the highest water levels. Other prey found that tidal variation is more important than diurnal items, like benthic (e. g., Nereis diversicolor) and plank­ rhythms in controlling the feeding of Fundulus heterocli­tonic species (e.g., Copepod), represented only a small tus. These authors showed that when nocturnal feeding gravimetric part of the gobies’ diet. While hyperbenthic occurred, feeding activity was significantly less than the prey abundance may largely explain the observed feed­ feeding peaks found during daytime. ing patterns during the semi-lunar cycle some questions remain unanswered. At spring tide, higher prey density encourage fish to feed more than what the higher full­ Semi-lunar influence ness index and daily ration indicated, but the lower Scientific observations examining the effect of the density of fish migrating into the marsh creek is not semi-lunar cycle on the foraging activity of fish species explained. At neap tide, total density of potential prey are largely lacking. Campana and Neilson (1985) and densities did not differ between day and night. Yet, the Hamerlynck et al. (1993) pointed to the existence of feeding intensity of the gobies was higher during neap bands on the otoliths of fish and the relation of these tide at night. bands with a semi-lunar periodicity in growth and thus The daily ration varies with the age of fish and food consumption. These bands consisted of 14 daily with ambient temperature (Norte-Campos and Temming, growth rings. Narrow rings alternated with wider rings. 1994). Eggers measured a daily ration of 5 12 % of body Arrellano (pers. comm.) proved that the wider daily weight for Pomatoschistus minutus. In the Wadden Sea, growth rings in Pomatoschistus lozanoi observed by the estimated tidal ration for Pomatoschistus microps Hamerlyck et al. (1993) coincided with spring tide and was 27.3% at 18.5°C (Norte-Campos and Temming, the narrower ones with neap tide. These observations 1994). The result of this study (7.2- 14.5 % body weight) thus proved that food consumption followed a semi-lu­ falls between these two reported values and followed the nar pattern and feeding intensity would differ between same change during the semi-lunar cycle as the fullness neap and spring tide. Hamerlynck et al. (1993), indeed, index. observed in the su.btid.al o f a sandy beach in The Nether­ This study showed clear effects of tidal, diel, and lands that the feeding intensity of P. lozanoi was related semi-lunar cycles on the feeding habits of Pomatoschis­ to the spring-neap tide. Aiso, Morton et ai. (1987) re­ tus microps. The influence of these cycles is not equal. ported that feeding index values for most salt-marsh Because of the strong tidal influence, diel variation is pre­ visiting fishes were higher when water depth reached a sent but inferior to tidal effects. Semi-lunar cycle clearly maximum because then larger areas of salt marshes influences both the abundance of P. microps and its feed­ were available for extended periods. The data from the ing habits, however, day-night differences between spring present study indicated a significant higher fullness in­ and neap tide are still in question, requiring further in­ dex during spring tide. The drop in average FI at the vestigations. highest water level caused no correlation between the fullness index and water height. The highest water levels and currents may represent a stressful environment, in­ Acknowledgments fluencing the migration and feeding of the fish. Hamer­ lynck et al. (1993) also observed a similar drop in full­ The authors wish to thank the foundation “Het Zeeuws ness index of P. lozanoi at maximum water levels and Landschap” for allowing sampling in the marsh “Het currents. While fullness index and consumption gener­ Verdronken Land van Saeftinghe”. Dang Van Thi and ally increase with increasing water levels, above a certain Emanuelle Bulayi greatly assisted in performing the threshold value, feeding is inhibited. Currents indeed in­ stomach analyses as part of their masters training pro­ crease with rising water levels during the semi-lunar gram. cycle (authors, pers. obs.). The change in abundance of the common goby is in accordance with this assumption. In contrast to the full­ R eferences ness index and the daily ration, the density of the com­ mon goby was higher at neap tide and lower at spring tide. Allen, E. A., P. B. Fell, M. A. Peck, J. A. Gieg. C. R. Guthke and M. Fewer individuals entered the creek at spring tide but Newkirk, 1994. Gut contents of common mummichogs, Fun­ foraged more. dulus heteroclitus L., in a restored impounded marsh an in Prey abundance is also determined by the same cycle natural reference marshes. Estuaries 172: 462-471. Andersen, N. G., 1984. Depletion rates of gastrointestinal content (Morgan, 1990; authors, pers. obs.). The potential prey in common goby (Pomatoschistus microps (Kr.). Effects of species of the common goby were more abundant during temperature and fish size. Dana 3: 31-42. Aquat. Sei. Vol. 66, 2004 Research Article 325

Antholz, B., W. Meyer-Anholz and C. D. Zander, 1991. Feeding Hamerlynck, O., A. Cattrijsse and R. V Arellano, 1993. Daily ration activities of two euryhaline small-sized fish in a western of juvenile Pomatoschistus lozanoi de Buen (Pisces: Gobiidae). Baltic brackish fjord. Helgoländer Meeresunters 45: 285- ICES Journal of Marine Science 50: 471-480. 300. Healey, M. C., 1971. The distribution and abundance of sand gob­ Arruda, L. M., J. N. Azevedo and A. I. Neto, 1993. Abundance, age- ies, Gobius minutus, in the Ythan estuary. Journal of Zoology structure and growth, and reproduction of gobies (Pisces: Gob­ 163: 177 229. iidae) in the Ria de Aveiro lagoon (Portugal). Estuarine Coastal Hesthagen, Í. H., 1980. Locomotor activity in the painted goby, and Shelf Science 37: 509 523. Pomatoschistus pictus (Malm) (Pisces), in relation to light Berge. J. A., K. 1. Johannessen and L, O. Reiersen, 1983. Effects of intensity. Sarsia 65: 13 18. the water soluble fraction of the North Sea crude oil on the Hostens, K. and J. Mees, 1999. The mysids-feeding guild of demer­ swimming activity of the sand goby Pomatoschistus minutus sal fishes in the brackish zone of the Westerschelde. Journal of (Pallas). Journal of Experimental Marine Biology and Ecology Fish Biology 55: 704-719. 68:159-167. Hyslop, E. J., 1980. Stomach contents analysis - a review of Beyst, B., J. Mees and A. Cattrijsse, 1999. Early postlarval fish in methods and their application. Journal of Fish Biology 17: the hyperbenthos of the Dutch Delta (south-west Netherlands). 411-429. Journal of the Marine Biological Association of the United Kneib, R. T., 1997. The role of tidal marshes in the ecology of Kingdom 79:709-724. estuarine nekton. Oceanography and Marine Biology 35: 163- Bouchereau, J. R., J. C. Joyeux and J. P. Quignard, 1989. Structure of 220 . the Pomatoschistus microps population (Kroyer, 1838), Pisces, Kneib, R. T. and A. E. Stiven, 1978. Growth, reproduction and feed­ Gobiidae, in the lagoon of Mauguio (France). Vie et Milieu 39: ing of Fundulus heteroclitus (L.) on a North Carolina salt mash. 19-28. Journal of Experimental Marine Biology and Ecology 31: Bouchereau, J. R., J. P. Quignard, J. C. Joyeux and J. A. Tomasini, 121-140. 1993. Population structure of reproductive specimens ofPo­ Kneib, R. T. and S. L. Wagner, 1994. Nekton use of vegetated marsh matoschistus microps Kroyer, 1838 (Gobiidae), in the lagoon of habitats at different stages of tidal inundation. Marine Ecology Mauguio, France. Cybium 17: 3-15. Progress Series 106: 227-238. Cadigan, K. M. and P. E. Fell, 1985. Reproduction, growth and Kroyer, II. N., 1838. Danmarks Fiske. Kjobenhavn. 1: 1-616. feeding habits of Menidia menidia (Atherinidae) in a tidal Maes, J., A. Taillieu, P. Van Damme and F. 01 levier, 1997. The com­ marsh-estuarine system in southern New England. Copeia position of the fish and crustacean community of the Zee- 21-26. schelde Estuary (Belgium). Belgian Journal of Zoology 127: Campana S. E. and J. D. Neilson, 1985. Microstructure of fish 47-55. otoliths. Canadian Journal of Fisheries and Aquatic Sciences Magnhagen, C., 1985. Activity differences influencing food selec­ 42:10141032. tion in the marine fish Pomatoschistus microps. Canadian Jour­ Cattrijsse, A., E. S. Makwaia, H. R. Dankwa, O. Hamerlynck and nal of Fisheries and Aquatic Sciences 43: 223 221. M. A. Hemminga, 1994. Nekton communities of an intertidal Magnhagen, C., 1998. Alternative reproductive tactics and courtship creek of a European estuarine brackish marsh. Marine Ecology in the common goby. Journal of Fish Biology 53: 130 137. Progress Series 109: 195 208. Magnhagen, C. and A. M. Wiederholm, 1982. Food selectivity Cattrijsse, A., H. R. Dankwa and J. Mees, 1997. Nursery function of versus prey availability: a study using the marine fish Pomato­ an estuarine tidal marsh for the brown shrimp Crangon cran­ schistus microps. Oecologia (Berlin) 55: 311-315. gon. Journal of Sea Research 38: 10-121. Mehner, T, 1992. Diet spectra ofPomatoschistus microps (Kroyer) Conover, W. J., 1980. Practical Non-parametric Statistics, Wiley, and Pomatoschistus minutus (Pallas) (Teleostei, Gobiidae) dur­ New York, 229-237 pp. ing first weeks after hatching. Zoologischer Anzeiger 229: Dijkema, K. S., W. G. Beeftink, J. P. Doody, J. M. Gehu, B. Heyde- 13-20. mann and M. Rivas Martinez, 1984. Salt marshes in Europe, Morgan, S. G., 1990. Impact of planktivorous fishes on dispersal, European Committee for the Conservation of Nature and Nat­ hatching, and morphology of estuarine crab larvae. Ecology ural Resources, Strasburg, 178 pp. 71: 1639-1652. Eggers, D. M., 1977. Factors in interpreting data obtained by Morton, R. M., B. R. Pollock and J. P. Beumer, 1987. The occur­ diel sampling of fish stomachs. J. Fish. Res. Bd. Canada 34: rence and diet of fishes in a tidal inlet to a saltmarsh in south­ 290-294. ern Moreton Bay, Queensland. Australian Journal of Ecology Elliott, J. M. and L. Persson, 1978. The estimation of daily rates of 12:217-237. food consumption for fish. Journal of Animal Ecology 47: Norte-Campos, A. G. C. and A. Temming, 1994. Daily activity, 977-991. feeding and rations in gobies and brown shrimp in the north­ Evans, S., 1984. Energy budget and predation impact of dominant ern Wadden Sea. Marine Ecology Progress Series 115: epibenthic carnivores on a shallow solt bottom community at 41-53. the Swedish west coast. Estuarine, Coastal and Shelf Science Pampoulie, C., P. Chauvelon, E, Rosecchi, J. L. Bouchereau and 18: 651-672. A. J. Crivelli, 2001. Environmental factors influencing the go- Gibson, R. N., 1988. Patterns of movement in intertidal fishes, in: biid assemblage of a Mediterranean Lagoon: Empirical evi­ G. Chelazzi and M. Vannini (eds.), Behavioral adaptation to in­ dence from a long-term study. Hydrobiologia 445: 175 181. tertidal life, Plenum, New York, pp. 55 63. Petersen, C. G. J., 1919. Our gobies (Gobiidae). From the egg to the Gibson, R. N., 1993. Tidally-synchronized behaviour in marine adult stages. Rep. Dan. Biol. Stn. 26: 45 66. fishes. In: M. A. Alin (eds.), Rhythms in fishes, NATO-AS1 Rooker, J. R. and G. D. Dennis, 1991. Diel, lunar and seasonal series, Plenum Press, New York, pp. 67 86. changes in a mangrove fish assemblage off southern Puerto Gibson, R. N. and I. H. Hesthagen, 1981. A comparison of the ac­ Rico. Bulletin of Marine Science 49: 684-698. tivity pattern of the sand goby Pomatoschistus minutus (Pallas) Rountree, R. A. and K. W. Able, 1992. Foraging habits, growth, and from areas of different tidal range. Journal of Fish Biology 18: temporal patterns of salt-marsh creek habitat use by young-of- 669-684. year summer flounder in New Jersey. Transactions of the Amer­ Hamerlynck, O. and A. Cattrijsse, 1994. The food of Pomatoschis­ ican Fisheries Society 121: 765-776. tus minutus (Pisces, Gobiidae) in Belgian coastal waters, and a Rozas, L. P., 1995. Hydroperiod and its influence on nekton use comparison with the food of its potential competitor/?lozanoi. of the salt marsh: A pulsing ecosystem. Estuaries 18: 579- Journal of Fish Biology 44: 753-771. 590. 326 A. Cattrijsse and H. Hampel Short term feeding habit of the common goby

Rozas, L. P. and M. W. LaSalle, 1990. A comparison of the diets of Ryer, C. II., 1987. Temporal patterns of feeding by blue crab Gulf Killifish, Fundulus grandis Baird and Girard, entering and {Callinectes sapidus) in a tidal-marsh creek and adjacent sea- leaving a Mississippi brackish marsh. Estuaries 13: 332-336. grass meadow in the lower Chesapeake Bay. Estuaries 10: Rozas, L. P., C. C. Mclvor and W. E. Odum, 1988. Intertidal rivulets 136-140. and creekbanks: corridors between tidal creeks and marshes. Tranter, D. J., 1979. Monographs on oceanographic methodology 2. Marine Ecology Progress Series 47: 303 307. Zooplankton sampling, UNESCO Press, Paris, 174 pp. Rozas, L, P. and D. J. Reed, 1993. Nekton use of marsh-surface Weisberg, S. B., R. Whalen and V. A. Lotrich, 1981. Tidal and diur­ habitats in Louisiana (USA) deltaic salt marshes undergoing nal influence on food consumption of a salt marsh killifish submergence. Marine Ecology Progress Series 96: 147 157. Fundulus heteroclitus. Marine Biology 61: 243 346.

To access this journal online: http : //www. birkhauser. ch