Domenico D’Alelio, Maria Grazia Mazzocchi, Marina Montresor, Diana Sarno, Adriana Zingone, Iole Di Capua, Gayantonia Franze*, Francesca Margiotta, Vincenzo Saggiomo &
Maurizio Ribera d’Alcala
Stazione Zoologica Anton Dohrn, Villa Comunale, 80121 Naples, Italy
Keywords
Climate change; coastal systems; community ecology; long-term ecological research station (LTER); Mediterranean Sea; meta-analysis; networks; plankton; trophic-webs.
Correspondence
Domenico D’Alelio, Stazione Zoologica Anton Dohrn, Villa Comunale, 80121 Naples, Italy.
E-mails: [email protected], dom.dalelio@gmail.
com
*Present address: Department of Biology, University of Akron, Akron, OH 44325-3908, USA.
Accepted: 24 June 2014
doi: 10.1111/maec.12211
Abstract
The internal organization of plankton communities plays a key role in biogeo-chemical cycles and in the functioning of aquatic ecosystems. In this study, the structure of a marine plankton community (including both unicellular and multicellular organisms) was inferred by applying an ecological network approach to species abundances observed weekly at the long-term ecological research station MareChiara (LTER-MC) in the Gulf of Naples (Tyrrhenian Sea, Mediterranean Sea) in the summers of 2002–2009. Two distinct condi-tions, characterized by different combination of salinity and chlorophyll values, alternated at the site: one influenced by coastal waters, herein named ‘green’, and the other reflecting more offshore conditions, named ‘blue’. The green and blue ‘phases’ showed different keystone biological elements: namely, large dia-toms and small-sized flagellates, respectively. Several correlations amongst spe-cies belonging to different trophic groups were found in both phases (connectance ~0.30). In the green phase, several links between phytoplankton and mesozooplankton and within the latter were detected, suggesting matter flow from microbes up to carnivorous zooplankton. A microbial-loop-like sub-web, including mixo- and heterotrophic dinoflagellates and ciliates, was present in the green phase, but it was relatively more important in the blue phase. The latter observation suggests a more intense cycling of matter at the microbial trophic level in the blue phase. These results show that different modes of eco-logical organization can emerge from relatively small changes in the composi-tion of aquatic communities coping with environmental variability. This highlights a significant plasticity in the internal structure of plankton webs, which should be taken into account in predictions of the potential effects of climatic oscillations on aquatic ecosystems and biogeochemical cycles therein.
Introduction
Understanding the structure and functioning of plankton communities is a crucial step in tracking biogeochemical cycles and predicting future responses of aquatic ecosys-tems to environmental changes at different times and spatial scales (de Senerpont Domis et al. 2013; Behrenfeld
& Boss 2014).
The flux of matter, energy and information in the oceans largely depends on the structure of plankton communities,
which are, in turn, characterized by the species present, their abundance and functional roles, and their possible biological inter-connections (Sommer et al. 2012). The clear-cut, paradigmatic formalization of a planktonic tro-phic chain ruled by phytoplankton production and zoo-plankton grazing – dating back to G. A. Riley’s work (Anderson & Gentleman 2012)– and applied to any plank-tonic system, has been progressively questioned by the increasing levels of awareness of the fine-tuned mecha-nisms at the base of plankton ecology (Tett & Wilson 2000;
Marine Ecology 36 (2015) 1155–1170ª 2014 Blackwell Verlag GmbH 1155
Marine Ecology. ISSN 0173-9565
Strom 2008). Plankton communities show multiple traits able to influence biogeochemical cycles with cascade effects. Many microbes can shift their metabolism between autotrophy and heterotrophy and effectively feed on other microbes (Sherr & Sherr 2007; Jeong et al. 2010; Schmoker et al. 2013), planktonic animals can graze selectively and show niche partitioning (Katechakis 2004; Katechakis &
Stibor 2004; Olson et al. 2006) and unexpectedly complex trophic cascades can emerge as a consequence of changes in community structure (Stibor et al. 2004; Caron &
Hutchins 2012).
Depicting the structure of communities and the links amongst their components encompasses two complemen-tary analytical steps: (i) the inference of biological links from empirical data (by describing, comparing and corre-lating the abundances of species or functional groups) and (ii) the construction and analysis of ecological net-works based on pairwise co-variations (Bl€uthgen et al.
2008; Vermaat et al. 2009). In this context, ecological networks are structured graphs consisting of species as nodes and biological links as edges; they represent a use-ful conceptual tool to schematize community structure, e.g. in terms of food-web relationships or the presence and relative importance of keystone species (Beckerman et al. 2006; Jordan 2009). Such an approach has been applied successfully to marine food-webs, revealing eco-logically consistent structures in different systems (Dunne et al. 2004; de Santana et al. 2013). As for plankton, net-work approaches have revealed non-random community organization and suggested crucial biological mechanisms, such as symbiosis, parasitism, competition and predation amongst unicellular organisms (Steele et al. 2011).
The present study aimed to (i) infer the structure of marine plankton communities, including both unicellular and multicellular organisms, in terms of statistical and presumably trophic links based on field data of species composition and abundance and (ii) relate short-term temporal changes in community structure to differences in the hosting environment. We applied an ecological network approach to plankton data collected weekly at the long-term ecological research station MareChiara (LTER-MC) in the inner Gulf of Naples (GoN, Tyrrhe-nian Sea, western Mediterranean). This well-studied sys-tem shows strong seasonality and resilience in the succession of plankton communities (Zingone et al. 1995, 2009; Modigh 2001; Modigh & Castaldo 2002; Mazzocchi et al. 2011, 2012) with strong regularity in the species’ life histories (D’Alelio et al. 2010). Herein, we describe the characteristics of two alternative modes of organization in a plankton community (including phyto-, microzoo- and mesozooplankton) in this coastal area. We call these modes the ‘green’ and ‘blue’ phases, occurring in lower and higher salinity water-masses, respectively, in the
sur-face water-layer at LTER-MC during summer. Based on the observation of community properties during the green and blue phases, we present two scenarios of food-web structure that switch through time in the summer plankton of the GoN.
Material and Methods Study site
The sampling station LTER-MC (40°48.50 N, 14°150E) is located in the GoN, two nautical miles off the coastline over the 75-m isobath. Sampling at LTER-MC has been conducted since 1984, except for a major interruption from 1991 to 1994. The sampling frequency was fort-nightly until 1990 and has been weekly since 1995.
The GoN is a relatively deep and wide embayment (average depth = 170 m, area = ~870 km2) that is rela-tively open to the offshore Tyrrhenian Sea waters. Land runoff from a very densely populated region influences the water typologies in the GoN; yet, in contrast to other coastal sites, riverine inputs are limited and intermittent, and salinity rarely goes below 37.5 (Ribera d’Alcala et al.
2004; Iermano et al. 2012, 2013). The close proximity of oligotrophic offshore waters to the coastline results in the co-existence of two subsystems: a relatively eutrophic coastal zone and an oligotrophic area similar to the off-shore Tyrrhenian waters. The position and width of the boundary between the two subsystems are variable over the seasons and the exchange between the subsystems at times can be enhanced by local circulation. Noticeably, in the coastal GoN, summer is not a period of low phyto-plankton biomass because of nutrients coming from land.
By contrast with the predictions from Margalef’s Mandala (Wyatt 2012), the community is dominated by diatoms despite stratification (Zingone et al. 1990).
Data collection
Conductivity, temperature, and fluorescence profiles were obtained with a SBE911 mounted on a Rosette sampler equipped with Niskin bottles (12 l). Chlorophyll a (chl a) concentrations was determined at 0.5, 2-, 5-, 10-, 20-, 40- and 60-m depths, whereas salinity and nutrient concentrations were determined at 0.5, 2-, 5-, 10-, 20-, 30-, 40-, 50-, 60- and 70-m depths. Ammonium (NH4), nitrate (NO3), nitrite (N02), phosphate (PO4) and silicate (SiO4) concentrations were determined with a TECHNICON II autoanalyzer up to 2005 and, starting from 2006, using a FlowSys Systea Autoanalyzer, accord-ing to Hansen & Grasshoff (1983), modified as described in Ribera d’Alcala et al. (2004). Chl a s was determined with a spectrofluorometer (Holm-Hansen et al. 1965;
1156 Marine Ecology 36 (2015) 1155–1170ª 2014 Blackwell Verlag GmbH
The green–blue plankton swing D’Alelio, Mazzocchi, Montresor, Sarno, Zingone, Di Capua, Franze, Margiotta, Saggiomo & Ribera d’Alcala
Neveux & Panouse 1987). Phytoplankton and microzoo-plankton samples were collected from the 0.5-m Niskin bottle and fixed with neutralized formaldehyde (0.8–1.6%
final concentration) and acid Lugol’s iodine (2% final concentration), respectively. Mesozooplankton samples were collected from 50-m depth to the surface using a Nansen net (113-cm mouth diameter, 200-lm mesh size) and fixed with formaldehyde (2–4% final concentration).
Phytoplankton, micro- and mesozooplankton were counted according to standard procedures as reported by Ribera d’Alcala et al. (2004).
Data analysis
We analysed the physical and ecological variability at LTER-MC at a weekly time scale from the end of June to the end of August for 8 years (2002–2009). Salinity and chl a values in the 0–2-m layer were used to characterize the superficial water-masses as coastal ‘green’ or offshore
‘blue’. Green and blue phases were identified in each summer season and the plankton community associated with each phase was characterized. Within each of the very large phyto-, microzoo- and mesozooplankton data sets, taxa were kept as such or assembled in homoge-neous groups according to different criteria (Table 1).
The most abundant species were kept separated. Less abundant taxa were aggregated based on taxonomy (e.g.
congeneric species). Less common species were aggregated according to size (e.g. dinoflagellates and ciliates smaller or larger than 15lm) or trophic level (e.g. carnivorous mesozooplankton, mixotrophic ciliates), according to current knowledge. This aggregation allowed zero values to be limited in the time-series data sets. After grouping, the whole plankton community was represented by 32 elements (Table 1).
Cluster, correlation and principal component analyses were carried out with the open-source software PAST (http://palaeo-electronica.org/2001_1/past/issue1_01.htm).
Hierarchical clustering for environmental and plank-tonic community data were conducted using the unweighted pair group method with arithmetic mean (UPMGA) algorithm and according to Euclidean dis-tance and Bray–Curtis similarity metrics, respectively (Legendre & Legendre 2012). Spearman correlations amongst plankton community elements (Table 1) were carried out for green and blue phases (see Results) including a minimum number of five samples, i.e.
across the longer time periods that had a coherent match between the environmental and planktonic com-munity data clustering.
All positive and negative correlations with r >0.7 or
<0.7 were considered. The largest fraction of P-values (relating to almost 400 correlations) was within the 0.05
threshold, a lower number (almost 200) was within the 0.10 threshold, a limited fraction (almost 60) exceeded 0.10. A histogram showing the frequency of P-values from all correlations considered in the present study is shown in Supporting Information Fig. S1. A non-strin-gent statistical limit was imposed upon our analyses because the high P-values of some correlations with high coefficients may have arisen because of the limited num-ber of samples analysed (numnum-ber of samples in continu-ous time-series ranging from five to nine). In order to visualize in a single elaboration all of the possible links between community elements, correlations detected in green (light and dark green) and blue (light and dark blue) periods in the different years were compiled into distinct matrices, namely: green positive, green negative, blue positive and blue negative. Correlation networks were built and analysed with the open-source software yED 3.11.1 (yWorks GmbH, http://www.yworks.com).
Network connectance was estimated as the number of links/node2 (Beckerman et al. 2006). The relative central-ity of network nodes was estimated in the frame of the yED software.
To corroborate patterns detected with these first exploratory analyses, further networks were built includ-ing only those links detected in at least two summer sea-sons and having ecological significance. The structure of plankton communities during the blue and green phases was reconstructed based on correlation networks amongst taxa. The ecological significance of correlations amongst taxa was assessed based on available knowledge from pre-vious studies, as described in the Results and Discussion section. Special emphasis was given to potential trophic links at different levels of the trophic-web, such as amongst protists, between the latter and zooplankton, and between carnivores and non-carnivorous mesozoo-plankton.
Results and Discussion
The main hindrance to the various approaches to com-munity studies of marine plankton is how to cope with highly diverse microbial communities floating in unstable and non-conservative environments. Like other coastal embayments, the GoN lies at the boundary between off-shore and coastal waters with different levels of produc-tivity. To reduce the degrees of freedom, we focused our analysis on the summer season, when the upper layer (0–
2 m) is permanently decoupled from the subsurface layer because of thermal stratification. This allowed us to focus on phytoplankton assemblages confined to the surface layer, and to consider the 0–50 m integrated samples for more mobile organisms, e.g. mesozooplankton. The phys-ical–ecological oscillations at LTER-MC during summer –
Marine Ecology 36 (2015) 1155–1170ª 2014 Blackwell Verlag GmbH 1157
D’Alelio, Mazzocchi, Montresor, Sarno, Zingone, Di Capua, Franze, Margiotta, Saggiomo & Ribera d’Alcala The green–blue plankton swing
herein, the end of June–end of August – were studied for 8 years (2002–2009) at a weekly time scale. Community composition during the green and blue phases, defined as described above, were tagged and characterized.
Environmental variability
In the summer seasons under investigation, vertical pro-files of salinity and chl a showed the largest variations
Table 1. Codes and descriptions for plankton taxa (species or supra-specific groups) considered in the present study.
taxon code
trophic
role plankton categories
taxa, genera, species or life stages included
1 A, M, H Small flagellates (cell size <10lm) mainly UIT plus Ollicola sp., Pyramimonas spp.
2 A Chaetoceros Chaetoceros socialis, Chaetoceros simplex, Chaetoceros throndsenii, Chaetoceros tenuissimus, Chaetoceros spp.
3 A Leptocylindrus Leptocylindrus danicus, Leptocylindrus aporus, Leptocylindrus spp.
4 A Skeletonema Skeletonema pseudocostatum, Skeletonema menzelii
5 A Small diatoms (cell size <10lm) Bacteriastrum sp., Cyclotella spp., Minidiscu ssp., Minutocellus sp., Thalassiosira spp., UIT
6 A Pennate diatoms (cell size >10lm) Thalassionema spp., UIT
7 A Pseudo-nitzschia Pseudo-nitzschia delicatissima, Pseudo-nitzschia galaxiae, Pseudo-nitzschia pseudodelicatissima, Pseudo-nitzschia spp.
8 H Large dinoflagellates (cell size >15lm) mainly UIT (both thecate and naked dinoflagellates) plus Dinophysis sp., Gymnodinium spp., Gyrodinium spp., Lessardia sp., Oxyphysis sp., Oxytoxum spp., Palaeophalacroma sp., Pronoctiluca sp., Pyrocystis sp., Protoperidinium spp., Torodinium sp.
9 M, H Small dinoflagellates (cell size <15lm) UIT (both thecate and naked dinoflagellates)
10 A Centric diatoms (cell size >10lm) Cerataulina sp., Dactyliosolen spp., Eucampia sp., Guinardia sp., Hemiaulus sp., Lauderia sp., Lioloma spp., Lithodesmium sp., Odontella sp., Proboscia sp., Rhizosolenia spp., UIT 11 A Coccolithophores Mainly UIT plus Acanthoica sp., Calciopappus sp., Calciosolenia sp., Calyptrosphaera spp.,
Ceratolithus sp., Helicosphaera sp., Holococcolithophora sp., Homozygosphaera sp., Emiliania sp.
12 A, M Rare flagellates and dinoflagellates
(cell size >15lm) UIT plus rare autotrophic flagellates and dinoflagellates (e.g. Alexandrium spp., Ceratium spp., Gonyaulax spp., Karenia spp., Prorocentrum spp., Scrippsiella spp.
13 H Nanociliates (cell size <25lm) UIT
14 H Heterotrophic ciliates (cell size >25lm) Strombidium spp., Strobilidium sp. (Ciliata) 15 M Mixotrophic ciliates Laboea strobila, Tontonia sp. (Ciliata)
16 H Prostomatids UIT (Ciliata Prostomatida) 17 A Mesodinium rubrum Mesodinium rubrum (Ciliata)
18 H Tintinnids Eutintinnus spp., Helicostomella sp., Nolaclusila sp., Proplectella sp., Salpingella spp., Tintinnopsis spp., Undella spp., UIT (Ciliata Tintinnida)
19 H, S Penilia avirostris Penilia avirostris (Cladocera)
20 H, S Calanoid juveniles Mainly juvenile stages of Clausocalanus spp. and Paracalanus parvus 21 H, S Cladocerans Evadne spp., Pseudevadne tergestina (Cladocera)
22 H, S Paracalanus parvus Adult stages of Paracalanus parvus (Copepoda Calanoida) 23 H. S Appendicularians UIT (Tunicata Appendicularia)
24 H, S Acartia clausi Adult stages of Acartia clausi (Copepoda Calanoida) 25 H, S Temora stylifera Adult stages of Temora stylifera (Copepoda Calanoida) 26 H, S Centropages typicus Adult stages of Centropages typicus (Copepoda Calanoida)
27 H, O Oithona spp. Different life stages of Oithona atlantica, O. decipiens, O. longispina, O. nana, O. setigera, O. similis (Copepoda Cyclopoida)
28 H, O Meroplankton Larval stages of Anellida Polychaeta, Crustacea Maxillopoda, Echinodermata, Mollusca 29 H, S Thaliaceans Salps and doliolids (Tunicata Thaliacea)
30 H, S Other calanoids Different life stages of rare calanoids (Copepoda Calanoida) (e.g. Acartia spp., Calocalanus spp., Centropages spp., Clausocalanus spp., Ctenocalanus vanus, Paracalanus spp.)
31 H, S, D Detritivores Corycaeus spp., Farranula rostrata and Oncaeidae (Copepoda Cyclopoida) plus Euterpina acutifrons (Copepoda Harpacticoida)
32 H, C Carnivores UIT of Chaetognata, Mollusca Pteropoda, Cnidaria Siphonophora plus Candacia spp. and Pleuromamma spp. (Copepoda Calanoida)
A = autotrophic; C = carnivorous; D = detritivorous (feeding on detritus, e.g. faecal pellets and aggregates); H = heterotrophic; M = mixotrophic;
O = omnivorous (feeding on both microbes and animals); S = suspension feeders (feeding on microbes, both autotrophic and heterotrophic);
UIT = unidentified taxa.
From 1 to 18 are protists, from 19 to 32 are metazoans.
1158 Marine Ecology 36 (2015) 1155–1170ª 2014 Blackwell Verlag GmbH
The green–blue plankton swing D’Alelio, Mazzocchi, Montresor, Sarno, Zingone, Di Capua, Franze, Margiotta, Saggiomo & Ribera d’Alcala
within the upper 10-m layer and rather similar values below 10 m, with a steep halocline often present between 2 and 10 m, and peak chl a values in the 0.5–2-m layer (not shown). Salinity and chl a also underwent the largest oscillations over time in the upper 2-m layer and showed a significant negative correlation (r =0.63; n = 87;
P < 0.001). The increment of chl a values at lower salin-ity levels was indicative of the presence of waters directly affected by coastal runoff and, therefore, richer in plank-ton. Thus, the relative shifts of surface salinity and chl a represented a good proxy for waters showing the influ-ence of more coastal-green, or more offshore-blue, condi-tions at LTER-MC. A more detailed discrimination of water types based on chl a and salinity values was per-formed through a cluster analysis. Two main groups were detected, both including two subgroups. The four groups, or phases, namely, light and dark green and light and dark blue phases, showed different average values and combinations of chl a and salinity, as shown in Fig. 1a (see also Fig. S2).
Under typical summer conditions, when stable high pressure fields prevail in the Mediterranean Sea, breezes alternating in direction on a diurnal scale are the domi-nant local forcing in the GoN (Uttieri et al. 2011). The surface current field makes a complete clockwise (anti-cyclonic) rotation over 24 h, with relatively strong coast-ward-orientated currents under the action of sea breeze, in contrast to relatively weaker offshore-moving waters in the presence of a land breeze. Under these dynamic con-ditions, the exchange between the coastal area and the open Tyrrhenian Sea is hampered (Uttieri et al. 2011).
Accordingly, the surface waters in the GoN during sum-mer are characterized by oscillating dynamics, with (i) alternation between phases reflecting the coastal influence to a different extent and (ii) green phases lasting between 2 and 7 weeks.
Ecological variability
Plankton taxa occurring in the summer communities of the years 2002–2009 and their abundance are listed in Table 1 and Supporting Information Table S1, respec-tively. A cluster analysis based on Bray–Curtis similarity amongst all plankton samples identified two main clusters (Fig. 1b). Further subgroups were present, especially in the left-hand cluster (Fig. 1b). A large number of plank-ton samples in the latter group corresponded to the green (both light and dark green) clusters based on chl a and salinity, whereas the majority of samples in the right-hand group corresponded to the blue chl a and salinity clusters (Fig. 1a and b). A high number of plankton sam-ples corresponding to light blue conditions were scattered across the plankton clusters. Nonetheless, green and blue
plankton samples were mutually segregated. Such congru-ence between the environmental and planktonic data was evident in six out of 8 years, but not in 2003 and 2008 (Figs 1c–h, S3 and S4).
The close match between environmental and plank-tonic clustering and the trophic regime was also reflected by significant differences in the abundance and relative percentage of the dominant taxa between the green and blue phases (Table 2, Fig. 2a–e). The most evident result was the relative shift between small-sized phytoplankton, dominating the blue phase, and relatively large diatoms, prevailing in the green phase (Fig. 2a). From the blue to the green phase, the average density of small flagellates (code 1 in Table 2) only showed a threefold increase, whereas the most abundant diatom genera Chaetoceros and Leptocylindrus (codes 2 and 3) underwent 15- and eightfold increases in cell density, respectively, and Skel-etonema and Pseudo-nitzschia (codes 4 and 7) increased sixfold (Table 2). As for ciliates, the obligate heterotro-phic prostomatids and tintinnids (codes 16 and 18) underwent eight- and sixfold increases in the green phase, respectively (Table 2). Against these significant shifts in protist communities, meso zooplankton showed less marked variations between the green and the blue phases (Table 2, Fig. 2d,e).
Mesozooplankton resilience in the GoN has previously been highlighted at different spatial (Ianora et al. 1985)
Mesozooplankton resilience in the GoN has previously been highlighted at different spatial (Ianora et al. 1985)