Short term changes in zooplankton community during the summer-autumn transition in the open NW Mediterranean Sea: species composition, abundance and diversity

Abstract. Short term changes in zooplankton community were investigated at a fixed station in offshore waters of the Ligurian Sea (DYNAPROC 2 cruise, September–October 2004). Mesozooplankton were sampled with vertical WP-II hauls (200 μm mesh-size) and large mesozooplankton, macrozooplankton and micronekton with a BIONESS multinet sampler (500 μm mesh-size). Temporal variations of total biomass, species composition and abundance of major taxa were studied. Intrusions of low salinity water masses were observed two times during the cruise. The first one, which was the most intense, was associated with changes in zooplankton community composition. Among copepods, the abundance of Calocalanus, Euchaeta, Heterorhabdus, Mesocalanus, Nannocalanus, Neocalanus, Pleuromammaand also calanoid copepodites increased markedly. Among non-copepod taxa, only small ostracods abundance increased. After this low salinity event, abundance of all taxa nearly returned to their initial values. The influence of salinity on each zooplankton taxon was confirmed by a statistical analysis (Perry's method). The Shannon diversity index, Pielou evenness and species richness were used to describe temporal variations of large copepod (>500 μm) diversity. The Shannon index and Pielou evenness decreased at the beginning of the low salinity water intrusions, but not species richness. We suggest that low salinity water masses contained its own zooplankton community and passed through the sampling area, thus causing a replacement of the zooplankton population.


Introduction
Organic carbon is synthesised by phytoplankton in the surface layer, via photosynthesis. Afterwards, a part of this carbon is exported to deep water, where it can be sequestered during many years. Intensity and quality of vertical particulate organic matter flux are related with physical and biological processes. For example, a gust of transition from stratified and oligotrophic summer conditions, to mixed and mesotrophic autumnal conditions. Sampling was done at an offshore station in the central part of the Ligurian Sea where horizontal advection is assumed to be negligible. The positioning of the Time Series Station (TSS, 28 miles offshore, 43 • 25 N, 8 • 00 E) was decided on the basis of a transect from coast to offshore waters. In addition, a grid of 16 stations 10 centred on the TSS was sampled three times during the cruise in order to describe the hydrological environment of the TSS (Fig. 1).

Environmental data acquisition
Wind speed was measured onboard with a meteorological station (sampling every 30 s and smoothing with a moving average with a 1 h window). Between the two legs, dur- 15 ing port call, wind speed data are taken from records by Meteo-France buoy located near the TSS, at the DYFAMED site (43 • 25 N, 7 • 52 E). CTD profiles (SBE 25) were performed with a time interval of about 3 h (255 profiles, temperature, salinity, pressure, fluorescence, O 2 , irradiance). Water sampling was done with a 12 bottles rosette simultaneously to get the profiles of nutrients, chlorophyll, and others chemical param-20 eters. In situ fluorescence was calibrated with chlorophyll-a concentration measured on rosette samples by HPLC. Using the method developed by Andersen and Prieur (2000), fluorescence (F , arbitrary units) was converted to chlorophyll concentration (Chl, µg L −1 ) with the following relationships: Leg 1 : Chl = 2.0740 * (F − 0.00785) (n = 453, r = 0.97) Short-term changes in the zooplankton community were investigated with two types of nets: (i) a multiple opening and closing net with 500 µm mesh nets, BIONESS 5 (Sameoto et al., 1980); the sampled community corresponds therefore to large-sized copepods, macroplankton and micronekton; (ii) a WP-II net (200 µm mesh size), the sampled community corresponding to mesozooplankton (copepods mainly). The BIONESS was obliquely hauled over the 0-250 m water column (9 different strata) in the vicinity of the time-series station. WP-II sampling was performed with 0-200 m ver-10 tical tows at the time series station with a triple WP-II net: two samples were used for biomass analysis (see Mousseau et al., 2008), the third one was formalin preserved for counting and taxonomic identification.

Preservation, counting and taxonomic identification
Samples were preserved with 5% borax-buffered formalin-seawater before counting 15 and identification. For copepod taxonomy, reference was made to the species inventory for Mediterranean Sea from Razouls and Durand (1991) and the web site of Razouls et al.: http://copepodes.obs-banyuls.fr. The species identification was not possible for all copepods, taxonomic determination is presented here at genus level. When the species could be recognized with absolute certainty, the name of the species is 20 specified. Non-copepod taxa are counted at a taxonomic level of family or order. Preserved WP-II samples were not available for the first part of leg 1 (17-22 September). Frozen samples, initially collected for biomass analysis were used for taxonomic identification. To defrost the samples, they were put in a beaker filled with room temperature water. As some organisms were damaged by the freezing, the taxonomic Introduction identification was less accurate. WP-II data from 17-22 September are also presented in this paper but these data are drawn in grey in the graphs (Figs. 4 to 7).

Data analysis
2.4.1 Abundance of zooplankton Raw data (from BIONESS and WP-II sampling), in number of individuals per net, were 5 standardized to number of individuals per square meter, depending on the section of the water column sampled (0-200 m for WP-II; 0-250 m for BIONESS). Abundance data from the BIONESS depth stratified hauls were integrated through the 0-250 m water column. In this study, we have separated copepods from the rest of zooplankton. For copepods, we only present the temporal abundance variation of main cope-10 pod genera, (i.e. genera whose abundance represents more than 1% of total copepod abundance). For the other organisms, we present temporal abundance variation of main non-copepod taxa, (i.e. taxa whose abundance represents more than 1% of total non-copepods abundance). However, a list of total individuals identified (copepods and other taxa) is presented in Appendix A. 15

Diversity indices
The computation of species diversity indices requires a taxonomic identification at species level. In WP-II samples, only 42% of total number of organisms could be determined at this level, making the calculation of species diversity indices impossible. In contrast, in BIONESS samples, 99% of copepods could be identified to species 20 level. Consequently, species diversity indices were only calculated using copepod data obtained with BIONESS net. Three different indices were computed: Shannon index (Shannon, 1948), Pielou evenness (Pielou, 1966), species richness. The comparison of these three indices will allow reveal if diversity variations are due to a change of the number of species, or Introduction Interactive Discussion a modification in the relative contributions of taxa, or a combined effect of these two parameters. Shannon diversity index (H ′ ) was computed from Eq. (3) where s is the number of species and p i is the relative frequency of the species i .
Pielou evenness (J) was computed by dividing H ′ by ln(s), as shown in Eq. (4): Species richness is defined as the number of species.

Day-night differences
10 Wilcoxon-Mann-Whitney test (p≤0.05) for non-paired samples was applied on zooplankton abundance and diversity data to see if there was a significant difference between night and day.

Relationship between zooplankton abundance and environmental parameters
Perry's method was used to determine if there was a relationship between zooplankton abundance and environmental parameters (Perry and Smith, 1994). This method allows identification of associations between each zooplankton group and an 20 environmental factor (in this study, the integrated water column salinity). Interactive Discussion sum of zooplankton abundance from all samples in each salinity class is computed, and this distribution is also cumulated. The cumulative distribution of abundance of each zooplankton group, g(t), was plotted against the cumulative distribution of salinity, f (t). If these two distributions are almost similar, there is no significant dependence of this zooplankton group on the environmental parameter, whereas the 5 greater their difference, the stronger is the association. A Monte Carlo randomization test was set after 10 000 permutations in order to test the significance of association between g(t) and f (t). This method is explained in detail in Perry and Smith (1994).

Relationship between zooplankton diversity and salinity 10
The method of cumulative sum of deviations from the mean, called "Cumsum" (Ibañez et al., 1993) is used for (i) detecting changes which occurred in the average level of a series, (ii) determining the date when changes appear, (iii) and estimating the average value of homogenous intervals. In the present study, this method was 15 used to determine if there was a relationship between diversity among large copepods and water column salinity during the cruise. The temporal variations of salinity and zooplankton diversity indices (day and night) are considered as three distinct chronological series. For each series x(i )of p values, the variable Sp, which is the cumulated sum of deviations from the mean k, is computed 20 as shown in Eq. (5): When x i is equal to the mean k over a period of time, the Sp curve is horizontal. When x i remains greater than k, Sp curve shows a positive slope and inversely. So, the moments when the series is changing relatively to the mean can be detected by slope 25 reversals.

Meteorological and environmental conditions
Temporal variations of wind speed (Fig. 2a) was characterised by several strong wind events (>25 knots). During the first part of the cruise, two from NE occurred (17 and 25 September). At the end of the cruise there was a succession of three gust of wind 5 from opposite directions: SW, NE and SW. The time-depth distribution of temperature ( Fig. 2b) shows highly stratified water column from the beginning of the cruise to 10 October. The thermocline was strongly marked, with a mixed-layer temperature higher than 20 • C (22 • C during weak wind periods). This thermocline was located at approximately 25 m depth throughout the cruise, except at the end, where it deepened to 40 m depth during the period of successive strong wind events (11-16 October). The thermocline deepening was accompanied by a strong cooling of the mixed-layer water and suggests the beginning of autumnal de-stratification.
The time-depth distribution of salinity ( Fig. 2c) shows the occurrence of two intrusions 15 of Low Salinity Water (LSW) during the cruise. This water has a coastal origin and crossed the Ligurian front along isopycnals by a barocline instability (Andersen et al., 2008 1 ). The first intrusion (LSW-1), which occurred from 21 to 30 September, was very important as well as by its size and by its intensity. LSW-1 was located between 15 m and 75 m depth. The lower value recorded was less than 38.05, whereas average 20 salinity at this depth lies between 38.30 and 38.40 outside the intrusion. The second intrusion (LSW-2), which occurred from 9 to 12 October, was weaker and restricted to the layer 20-40 m. A salinity less than 38.30 was recorded during two days, and Interactive Discussion minimum salinity was not lower than 38.20. The time-depth distribution of chlorophyll-a (Fig. 2d) shows a bimodal distribution on the vertical at the beginning of the cruise. The deeper peak (80 m depth) was mainly composed of senescent diatoms, which quickly sedimented. The upper one, which was located at about 50 m depth, was mainly composed of nanophytoplankton (Lasternas 5 et al., 2008 2 ). The 50 m peak persisted until the end of the cruise but the maximum concentration occurred at the beginning of the cruise (19-22 September). The decline coincided with the arrival of LSW-1.
3.2 Zooplankton abundance 3.2.1 Total zooplankton biomass 10 As temporal changes in the biomass of total zooplankton biomass are detailed in Mousseau et al. (2008) 3 , we will give only few comments. Total zooplankton biomass integrated over the 0-200 m water column varied between 0.15 g m −2 and 3.79 g m −2 ( Fig. 3). As expected, night data were generally higher than day ones, except for one point (night between 18 and 19 September). This was due to migratory organisms 15 which are located in deep layers during day and move to the surface layer during night. In spite of a strong variability in the data, it is noticeable that average zooplankton biomass appeared higher during LSW-1. between 10 000 and 45 000 ind m −2 (Fig. 4a). It reached a maximum during LSW-1, after which it nearly returns to initial values. In contrast, there were no visible effects of LSW-2 on total copepod abundance. Copepodits, which represents more than 48% of total copepod numbers, showed the same pattern as total copepods, with a maximum of 22 000 ind m −2 during LSW-1 (Fig. 4b). When considering abundance of adults av-10 eraged over the sampling period, the genus Clausocalanus ranked first, followed by Oithona, Pleuromamma, Calocalanus and Neocalanus. The sum of these five genera represents nearly 90% of the abundance of adults. Clausocalanus spp. was mainly C. pergens (43%). Its abundance did not vary a lot during the cruise but one maximum was recorded during the night between 27 and 28 September (Fig. 4c). Most of the small copepods and copepodits collected with WP-II net in the size range 20 200-500 um did not appear in the BIONESS samples. Total abundance of large copepods sampled with this net, fluctuates around 500 ind m −2 ( Fig. 5a) but shows a strong increase on 21 September, at the beginning of LSI-1 (until 3000 ind m −2 ). Afterwards, concentrations declined until the end of LSW-1 to come back nearly to the initial values. As with WP-II samples, there was no increase of total large copepods during 25 LSW-2. The abundance increase during LSW-1 was observed for most of the principal copepod genera, especially the dominant one: Neocalanus (Fig. 5b). This genus consisted of a single species, N. gracilis (as is WP-II samples) and represented more Interactive Discussion than 50% of total copepod numbers sampled with BIONESS net. It ranked first by average abundance, followed by Nannocalanus (exclusively N. minor ), Pleuromamma (32% P. abdominalis and 68% P. gracilis), Euchaeta, Scolecithricella, Heterorhabdus and Mesocalanus (exclusively M. tenuicornis). The abundance of all these taxa clearly increased with LSW-1, except for Euchaeta and Scolecithricella, for which abundance 5 increases were less evident ( Fig. 5c-h). Among the non-copepod taxa sampled in WP-II, the most abundant one were the appendicularians, followed by pteropods, ostracods, hyperiids, chaetognaths and euphausiids ( Fig. 6a-f). For most of these taxa, abundance fluctuated randomly without any strong relationship with either LSI-1 or 2 ( Fig. 6a-f). The most striking feature was the occurrence of short term abundance peaks (each time constituted with only one point): Appendicularians (night between 28 and 29 September), Pteropods (15 October), Ostracods (night between 28 and 29 September), Hyperiids (night between 19 and 20 September), Chaetognaths (25 September). These short term variations could have been related to horizontal patchiness. 15 Among non-copepod taxa sampled with BIONESS net, the most abundant were euphausiids (50% Nematoscelis megalops, 28% Meganyctiphanes norvegica and 14% Stylocheiron longicorne), followed by chaetognaths, hyperiids, ostracods and pteropods ( Fig. 7a-e). As in WP-II samples, there was no clear effect of LSW-1 or 2 on these taxa. Their abundances fluctuated randomly, mostly dominated by day-night 20 variations.

Day-night variations in zooplankton abundance
Vertical samples integrating zooplankton organisms over the upper layer (0-200 m) hide any migration into this depth range, so variations between day and night will reveal only taxa which are migrating out of this superficial layer during day. Among all 25 organisms sampled with WP-II, only hyperiids and euphausiids showed a significant difference between night and day abundances (Table 1). Among large-sized organisms (BIONESS samples), the difference between day and night abundance was sta-2248 Introduction

Back Close
Full Screen / Esc

Printer-friendly Version
Interactive Discussion tistically significant for euphausiids, pteropods and hyperiids and also for the copepod genera Euchaeta, Pleuromamma and Scolecithricella. These organisms crossed the low salinity layer during night, confronted a 0.2 salinity decrease and did not modify their behavior. Pteropods and the copepods Pleuromamma are known for their strong migratory be-5 havior (Andersen, 2001b) but in this study, they showed a significant day-night abundance variations only in BIONESS samples. This could be the consequence of two facts: first, the large proportion of juveniles in WP-II sampled, which do not migrate out of the 0-200 m layer, and second the patchiness inducing large variability in successive samples.

Relationship between zooplankton abundance and salinity
The results of Perry's test, which we used to examine the relationship between salinity and abundance of the different groups, are presented in Table 2 and Figs. 8 and 9. For the groups whose day-night abundance was not significantly different, Perry's test was made by merging night and day data. In contrast, day and night data were tested 15 separately for the others.
Most of the copepods from WP-II samples were significantly influenced by salinity (Table 2): total copepods, copepodits, Calocalanus and Neocalanus. These organisms were mainly sampled during low salinity periods (Fig. 8a). About 40% of total copepods, copepodits and Neocalanus were sampled in the two first salinity classes 20 and 50% of Calocalanus.

25
The non-copepod taxa sampled with WP-II and BIONESS nets seemed less influenced by salinity. Only the small ostracods (<200 µm, WP-II samples) showed a significant relationship with salinity ( Interactive Discussion the two first salinity classes (Fig. 8b).

Day-night variations of diversity
The results of Wilcoxon-Mann-Whitney test (Table 3) showed that night values of Shannon diversity index and Pielou evenness were significantly higher than day values.

5
However, day and night species richness were not significantly different. In other terms, during the night, Shannon index and Pielou evenness values were higher but the number of species did not change. This could have been due to the migratory taxa (Euchaeta spp., Pleuromamma spp. and Scolecithricella spp.) whose abundance were low in 0-250 m layer during day, and are increased considerably at night.

Temporal variations of large copepods diversity
Shannon diversity index strongly varied during the cruise, between 1.10 and 3.00 (Fig. 10). Lowest values were recorded during LSW-1, during day as well as during night. We can thus suggest that there was an impact of the LSW-1 on the copepod community structure, but this perturbation had a short duration time. 15 Pielou evenness varied between 0.24 and 0.64 and paralleled the Shannon diversity index. Decreases in Shannon index and Pielou evenness during LSW-1 were due to marked increases in the abundance of N. gracilis, N. minor, which dominated the copepod community.
The species richness (i.e. number of species) fluctuated in the range 18 to 30, with a 20 strong random variations from day to day. It did not decrease at the beginning of LSI-1 which confirms that shifts in diversity indices reflected changes in relative abundances of taxa within a stable community.  Figure 11 shows the cumulated sum of deviations from the mean (Cumsum) for salinity and night and day Shannon index. All three variables showed the same pattern: slope reversals occur at the same time, which suggests that diversity changes are related to changes in salinity. Figure 11 also suggests that the sampling period can be divided in 5 four parts:

BGD
-Part 1 (17-20 September): slopes are positives, which means that successive values are above the mean as well as for salinity than for Shannon index.
-Part 2 (20-30 September): negative slopes, which indicate values under the mean for salinity and diversity. This is the LSW-1 period.

10
-Part 3 (4-9 October): slopes become positives again, which indicates the end of LSW-1. Copepods community is returning to its undisturbed state.
-Part 4 (9-16 October): slopes are close to zero. There is no effect of LSW-2. Copepods community structure comes back to its initial values; salinity and diversity are stable. Interactive Discussion spring-summer and summer-autumn. We will present here the similarities and the differences between the two zooplankton communities observed.
In the study of Andersen et al. (2001a), total copepod abundance sampled with WP-II fluctuated between 15 000 and 50 000 ind m 2 . During Dynaproc 2, the range of values is very close: 10 000-45 000 ind m 2 .

5
The comparison of major taxa sampled during Dynaproc 1 (late spring-summer) and Dynaproc 2 (summer-autumn) reveals that the two periods shared a great number of taxa: Clausocalanus, Euchaeta, Heterorhabdus, Neocalanus, Oithona and Pleuromamma. Andersen et al. (2001a) reported the presence of Calanus helgolandicus, Centropages typicus and Monacilla typica among the major species during Dynaproc 1. Although these three taxa were found during Dynaproc 2, their abundance was very low (0.25% C. helgolandicus, 0.20% C. typicus and 0.03% M. typica). C. helgolandicus and M. typica are deep-living species (Andersen et al., 2001a), which could explain their low abundance in the 0-250 m layer. C. typicus is a spring species whose abundance decreases during summer (Mazzocchi et al., 2007), and it becomes rare in 15 autumn.
Mesocalanus is the only genus which appears among the major taxa found during Dynaproc 2 but not during Dynaproc 1. The abundance of this species is low outside LSW-1 (<10 ind m −2 ) but it increased during the salinity event. Without the increase during LSW-1, Mesocalanus would not have been among the major taxa in Dynaproc 20 2 cruise.
Large copepods diversity has calculated in the present study, for Dynaproc 2 cruise but unfortunately, Andersen et al. (2001a, b) have not calculated it for Dynaproc 1. Therefore, it is not possible to compare the dynamic of large copepods diversity between the two periods.

Impact of LSW on zooplankton community
The sampling site of Dynaproc 2 cruise was located near the permanent DYFAMED time-series station. For many years, this offshore site was thought to be protected from  , 1983;Sournia et al., 1990;Marty and Chiaverini, 2002). Recently, Stewart et al. (2007) formulated the possibility of lateral processes at DYFAMED site (transport of particles along isopycnals or intrusion of shelf waters to the site) to explain the disparity in their sediment traps data. The Dynaproc 2 cruise data brings some arguments in 5 favor of shelf water intrusions hypothesis. These observations are the first ones which show clearly the dynamics of such intrusion in the central part of the Ligurian Sea.
The results of our study showed that the arrival of LSW-1 in the sampling area was associated with changes in the copepod community. These changes are summarised in Fig. 12. The temporal segmentation of the cruise was obtained from the cumsum on salinity (Sect. 3.3.3). Nannocalanus and large Neocalanus strongly increased at the beginning of LSW-1 but their abundance decreased quickly. Euchaeta also increased at the beginning of LSW-1 but its abundance stayed high throughout the intrusion. Mesocalanus increased at the middle of the intrusion but decreased immediately. The abundance increase of undetermined copepodits, Heterorhabdus, small Neocalanus 15 and Pleuromamma occurred at the end of LSW-1 and had a short duration. A decrease in measures of the diversity of large copepods diversity (Shannon index and Pielou evenness) was visible only at the beginning of LSW-1.
So, we suggest that LSW-1 contained its own zooplankton community and passed through the sampling area, thus causing a community replacement. There were no 20 taxonomic changes but rather only an abundance increase of some groups and a decrease in the diversity, in terms of evenness, of large copepods. The LSW-1 did not bring any new group of zooplankton: all taxonomic groups found during LSW-1 were also sampled outside the intrusion. Moreover, the different lags in the timing of several copepod taxa variations suggest different characteristics at the beginning, in the middle 25 and at the end of LSW-1.
The increase of zooplankton abundance during LSW-1 cannot be explained by reproduction for two reasons. First, the increase occurred too fast and second, high abundance does not last a long time and zooplankton community comes back nearly Interactive Discussion to its initial structure a few days after LSW-1, before the end of the cruise. Although we observed an increase in copepod abundance during LSW-1, the increase is unlikely to represent a preference for low salinity waters. Rather, zooplankton is strongly influenced by currents and hydrodynamic. Salinity is, in fact, a marker which indicates the arrival of different water masses containing different populations.

Conclusion
Dynaproc 2 cruise was initially devoted to study, at short time scales, how ecosystems switch from summer oligotrophy to autumnal mesotrophy in the Ligurian Sea, and notably the effect of wind forcing on mixing. Monthly data acquired since 1991 at DYFAMED station, showed that summer-autumn shift generally occurred between 10 mid-September to mid-October (Marty and Chiaverini, 2002). In 2004 (the year of Dynaproc 2 cruise), the seasonal shift occurred late and the destratification due to gust of wind started only five days before the end of the cruise, which is too short to study its effect on zooplankton community.
However, a marked phenomenon was been recorded during the cruise: the intrusion 15 of coastal LSW two times in the sampling area, which was thought to be protected from coastal water by Ligurian current flow. Although the authors of a recent study (Stewart et al., 2007) venture the hypothesis of such coastal intrusions existence at DYFAMED station, they have never been observed before Dynaproc 2. The cruise lasted only one month but two coastal water intrusions were observed: these phenomena may be 20 more frequent that one can think previously. Our study documents a marked effect of coastal LSW intrusion on the offshore zooplankton community of the Ligurian Sea, and therefore its potential effect on matter flux. So, it seems necessary to multiply high frequency studies or automatic measurements in this area in the aim (i) to determine the frequency occurrence of LSW intrusions in                      . 8. Cumulative frequency distribution of different zooplankton groups sampled with WP2 net (g(t), in black) in relation to salinity levels (f (t), in grey). (a) copepods (b) other groups. Only taxa for which Perry's test showed a significant relationship between zooplankton abundance and salinity were plotted ( Table 2). The arrow indicates the salinity class for which the greatest difference between g(t) and f (t) was founded. For example, in (a) more than 50% of Calocalanus spp. were sampled in the two first salinity classes. 5,2008 Short term changes in zooplankton community  Fig. 9. Cumulative frequency distribution of copepods sampled with BIONESS net (g(t), in black) in relation to salinity levels (f (t), in grey). (a) Copepods for which day and night abundances were not significantly different (day and night data were merged). (b-c) Copepods for which day and night abundances were significantly different: (b) day data, (c) night data. Only taxa for which Perry's test showed a significant relationship between zooplankton abundance and salinity were plotted ( Table 2). The arrow indicates the salinity class for which the greatest difference between g(t) and f (t) was founded.