Spatiotemporal and Environmental Dynamics of Abundances and Diversity of Larval Fish in Artificial Reef Edge Habitats of Kitros, Pieria (Northern Aegean Sea, Eastern Mediterranean)

Abstract

Spatiotemporal and physiochemical influences on the abundances and diversity of ichthyoplankton were assessed in the Thermaikos Gulf and edge habitats surrounding the Pieria (Northern Greece) artificial reef complex. The collection of data was performed in edge habitats in the marine protected area near the artificial reef offshore of Kitros. Sampling trips occurred in each season of the spring, summer, and autumn in each year from 2015–2017. This artificial reef complex lies at a distance offshore of 11.5 km from Kitros and the delta of River Aliakmonas. A bongo net sampler was used to sample ichthyoplankton over a network of 16 sampling stations. Seventy species of larval fish were identified. The greatest measures of ichthyoplankton diversity were found during warmer seasons. Non-metric multidimensional scaling showed that seasons play a significant role in species assemblages, as months during the spring, summer, and autumn from different years clustered together. Ten groups of larvae were identified to the family or genus level, such as Arnoglossus spp., Callionymus spp., Crenilabrus spp., Gobius spp., Liza spp., Scorpaena spp., Solea spp., and Spicara spp. Overall, the species with the greatest abundance was the larvae of the European anchovy (Engraulis encrasicolus). The second genera (in order of highest abundance) were Gobius spp. followed by relatively abundant (but lesser numbers of) larvae representing the families Callionymidae, Centracanthidae, and Bothidae. The largest total abundances were found in July 2015 and September 2016. The biodiversity index indicated that measures of diversity were greater in July 2015, which was the only sampling performed in the middle of the summer, and indicated higher diversity in autumn 2015. The aim of this study was to present the assemblages of fish larvae in the marine protected area surrounding the artificial reef of Kitros Pierias as the result of a 3-year sampling program (2015–2017). The novelty of this study is that it is the only study of Ichthyoplankton ever performed in Greece with seasonal repetitions and densely located stations around a coastal marine protected area.

1. Introduction

Studying ichthyoplankton dynamics facilitates our understanding of the ecology and evolution of fish faunas [1]. Fish represent an important link between smaller planktonic and larger nektonic organisms. Abundances of pelagic fishes can be elucidated by monitoring respective ichthyoplankton, which can be much simpler than monitoring adult populations [2].

Spatiotemporal dynamics and environmental cues influencing the spawning of fishes are generally believed to have evolved, such that meroplanktonic early life stages emerge into environmental regimes suitable for survival [3]. Thus, biophysiological characteristics making up such regimes in spawning habitats and sites of larval emergence and development are crucial to understanding as these areas facilitate connecting the life history of spawning adults and offspring and can influence subsequent abundances of larval and adult stocks [4]. Favorable conditions for the survival of ichthyoplankton include an abundance of food, low predator densities, circulation patterns promoting retention within as well as promoting transport out of nursery areas and spawning grounds [5]. Ichthyoplankton of pelagic spawning species, spawned fish eggs and hatched larvae can be increasingly dispersed in over vast marine areas over time, and therefore samples of ichthyoplankton comprised mainly of a single species of abundant eggs and larvae can help to elucidate particularly important spawning areas [6]. Long-term monitoring systems based upon fishery-independent biological sampling are essential to clarify the dynamics of lower trophic level organisms that may influence and interact with fish eggs and larvae, and essential to the ascertainment of fish spawning habitats. Ultimately, such information can help estimate the survival rate of fish larvae and juveniles, which is important as this eventually determines the recruitment success of adults [4].

Such assessments of the influences of spatiotemporal and environmental dynamics upon larval fish abundances have been carried out in the Central Pacific [7], Indo-Pacific [8], Antarctic [9], South Atlantic [10], Western North Atlantic [11], and North Pacific oceans [12]. Likewise, studies have been done in the northwestern and central Mediterranean Sea [13,14,15,16,17]; where results indicated that late spring to early summer is a transition period for pre- and post-spawning and that many ichthyoplankton are at maximum abundance [13,14,18,19]. However, in some parts of the Mediterranean Sea, there is a lack of such information. Detailed studies are needed to understand the process of successful recruitment.

The Gulf of Thermaikos is a region of the Aegean Sea, which harbors a marine ecosystem of high biophysiological complexity and diverse biota. Major rivers including the Axios, Aliakmon, Loudias, and Gallikos contribute to inputs into the Gulf of Thermaikos [20]. However, few if any spatiotemporal, environmental, and multispecies ichthyoplankton investigations have been completed in the Gulf of Thermaikos [21]. The aim of this study was to investigate the spatiotemporal and environmental parameter influences on ichthyoplankton assemblages in edge habitats of the Kitros artificial reefs complex in the Pieria region of the Gulf of Thermaikos. The hypothesis that ichthyoplankton species would be caught in accordance with the seasonality of their spawning periods was tested. Moreover, the levels of association between ichthyoplankton abundance and/or diversity with selected environmental parameters were examined.

2. Materials and Methods

2.1. Sample Sites, Hydrographic Features

Thermaikos is a micro-tidal gulf in the western part of the North Aegean Sea, with depths ranging between 30 and 200 m. It is the most important gulf of Northern Greece since it hosts Thessaloniki, the second largest Greek city (1,500,000 population), a great urban and commercial center. The main anthropogenic activities in the region include several industrial plants, mainly to the west, touristic and maritime transport activities, fishing, agriculture, as well as aquaculture farming to the eastern and western parts of the gulf. Thermaikos Gulf receives freshwater inputs from four rivers (Axios, Aliakmonas. Loudias, and Gallikos). The first two are considered the most important, forming multi-channel deltaic systems [22,23]. Axios is heavily polluted, transporting high nutrient loads through its catchment into Thermaikos Gulf [24,25].

The circulation pattern includes the more saline water entrance from the eastern part, following a north-western direction, and the exit of lighter water in the form of river plumes with a southward direction, along the western coastline [26]. Thermaikos Gulf is considered a mesotrophic system; however, extreme eutrophication events often occur under persistent southerly winds [23]. Finally, as [27] denote, an overall ecosystem degradation is observed, with fish catch and biomass decreasing, as a result of overfishing and environmental factors. Nine surveys were conducted during 3 seasons (spring, summer, and autumn) for 3 years (2015, 2016, 2017). The samplings took place in June, July, and October 2015; April, July, and September 2016; and April, May, and September 2017. Winter sampling was omitted as sample sites receive brackish water influx with low temperatures at this time and ichthyoplankton are expected to be scarce. Ichthyoplankton and ecological survey data were collected each year at 16 field sites in the outer region of the Thermaikos Gulf in the Aegean Sea and offshore of the coastal zone of Kitros, in the Pieria region (Greece). Field sites were located in edge habitats along the boundary of a previously installed artificial reef complex (Figure 1).

2.2. Environmental Parameters

During the above surveys, the environmental parameters were recorded at the surface (from 0–5 m depth) and the water column (>5 m depth to bottom). The average values for the surface and non-surface categories were used for subsequent analyses. Seawater’s temperature (nearest 0.001 °C), and salinity (practical salinity unit, PSU, nearest 0.001), were measured using a Seabird SBE-19plus CTD. The temperature gradient and salinity gradient of the seawater were calculated by subtracting the surface value of each mentioned parameter from the bottom value. Environmental gradients are used to reduce the spatial scale required for observation and experimentation. A Niskin Bottle sampler (1 L) was used to acquire samples for chlorophyll-a (mg·L⁻¹ chl-a) from five depths (surface, 5, 10, 20, and bottom). Concentrations of chl-a (nearest 0.01 mg·L⁻¹) were determined following trichromatic spectroscopy guidelines and methods outlined in [28] using a HITACHI U-2001 spectrophotometer. The recorded spatiotemporal and environmental parameters were used in further analyses to determine their influences on ichthyoplankton distributions and abundances.

2.3. Ichthyoplankton Collection

A paired bongo net sampler was used to collect ichthyoplankton. Each bongo net mouth had a 60 cm diameter frame (each mouth = 0.28 m²) and was fitted with 250 μm mesh conical nets. All plankton was transferred into respective individual glass sample jars for fixation into a mixture of 70% pure alcohol, 29% deionized H₂O, and 1% glycerin. Glycerin was added to the mixture in order to make alcohol evaporate at a much slower rate. At the Fisheries Research Institute of Kavala, individual samples were processed and ichthyoplankton were identified using existent taxonomic keys and literature [29,30], to the lowest possible taxonomic level.

2.4. Statistical Analyses—Biological Indices

The number of fish larvae collected at each station was standardized to a unit of sea surface area (10 m²) to provide measures of abundance for each sampling year. Further, the sampling Julian day of the year (JDOY) was converted to the month of the year and/or season, and the monthly average abundances of the larvae of each species were calculated as the mean of the standardized number of larvae at all stations for each individual species. Using these metrics, the diversity indexes were calculated. These diversity indexes chosen for this analysis were the Margalef index, the diversity index of Shannon–Wiener, and Simpson’s diversity index. The indexes were calculated in an attempt to draw conclusions regarding species assemblages according to the existing bibliography [31].

2.5. Statistical Analyses—Generalized Linear Modeling

Using SPSS software, spatiotemporal and environmental influences on the abundances of ichthyoplankton groups of species were examined. The initial factors that were used were sample site, year of sampling, water temperature gradient (°C), water temperature (°C), salinity (PSU), salinity gradient (PSU), Julian Day of the Year, surface salinity (PSU), and surface water temperature (°C). Except for sample site and year used as fixed effect categorical independent variables, all other variables were used as continuous covariates.

To address and reduce co-linearity among predictors, [32] was followed. Accordingly, the variance inflation factor (VIF) was examined with factors among predictors. Sequentially dropped was dropped with the highest VIF, then recalculated the VIF, and repeated this process until all VIFs ≤ 5.0—a moderately conservative cutoff chosen based upon [33] cited within [32]. Ultimately, five covariates of the original 12 were removed for GLMs in the VIF-based process.

The remaining predictors with acceptable VIF scores included sample year, Julian, day of the year, sample site, water temperature, salinity, water temperature gradient, and salinity gradient. These predictors were used in all subsequent models wherein the subjects were the individual fish abundances for each sampling station. Moreover, 4 of 144 sampling stations were excluded as they lacked a majority of measures for predictor variables, and 140 sampling stations were used for analyses.

Relatively high zero counts across sampling stations for subjects in the data set lead to the selection of only the most important abundances of fish (of 75 available species/groups) for GLMs. Thus, only species and groups that were in both the top 20 most abundant and the top 20 most frequently encountered were used for further quantitative assessment. Three species/groups that were in the top 20 by abundance were not in the top 20 by frequency. Thus, 17 species that overlapped in such a way were subsequently selected for further modeling.

2.6. Statistical Analyses—Species-Based Non-Metric Multi-Dimensional Scaling

After transforming mean species abundances to the fourth root, the non-metric multidimensional scaling was used (NMDS; [34,35]) in PC-ORD [36] to produce a species-based visual representation of the dissimilarity between ichthyoplankton species abundances based upon the influences of the 14 spatiotemporal and environmental parameters for each individual sample station (N = 140 sample stations). NMDS was applied to calculate the dissimilarity matrix. All 75 species or species classifications were used.

Finally, the K-means cluster analysis procedure in SPSS was applied to species coordinates results from the NMDS analyses to determine which species clustered together. Based on the number of taxa used in the NMDS (75), 12 clusters were selected as the optimal number of clusters. Single species clusters indicated some degree of uniqueness across the measured predictors and, clusters containing multiple species suggested some common environmental or spatiotemporal factor(s) were important in influencing grouped species abundances [36,37].

2.7. Statistical Analyses—Seasonal-Based Non-Metric Multi-Dimensional Scaling

Similarity Matrices based on the Bray–Curtis measure were generated and NMDS was used again to graphically display a plot of relationships between sampled months based on the relative abundance of each taxonomic group [34,35]. The analysis of similarities (ANOSIM) identified whether differences in assemblage groupings in sampling periods in the NMDS ordinations were significant. Similarity percentage analysis (SIMPER) was used to determine the dominant taxa in each seasonal grouping. An analysis using the abundances of the most frequently occurring species was later performed, both between stations as well as between seasons. Diversity markers were then determined, and the principal component analysis was used to facilitate the statistical identification of assemblages between species and stations. Finally, the interrelationship between the physicochemical parameters and the ichthyoplankton abundance was performed using the Spearman rank correlation.

3. Results

3.1. Environmental Parameters

Mean measurements and standard deviations for environmental parameters recorded during each sampling station (N = 140) are presented in a summarized format in

Table 1. Average values for each sampling season and yearly average values of the environmental parameters measured at field sites in the outer Thermaikos Gulf in the Aegean Sea and offshore of the coastal zone of Kitros, in the Pieria region of Greece. During surveys (N = 140) environmental parameters were recorded at the surface (from 0–5 m depth) and in the water column (>5 m depth to bottom) and used average values, then averaged these across all the sample sites for each category for the presented values. Units: temperature = °C; chl-a = μg·L⁻¹; salinity = PSU; Std. Dev. = standard deviation.

Year	Julian Day of the Year	Statistic	Surface chl-a	chl-a	Surface Salinity	Salinity	Salinity Gradient	Surface Temperature	Temp.	Temperature Gradient
2015	176	Mean	5.375	1.557	33.134	36.102	−4.561	23.527	19.03	8.092
		Std. Dev.	2.491	0.352	1.239	0.489	1.058	0.203	0.562	0.48
	195	Mean	4.452	1.888	33.53	36.002	−4.178	25.947	20.68	10.146
		Std. Dev.	2.281	0.667	0.89	0.605	0.664	0.48	1.419	1.054
	303	Mean	3.492	2.157	32.546	35.913	−5.117	17.938	19.75	−3.041
		Std. Dev.	0.521	1.013	1.308	0.641	1.066	0.573	0.343	0.495
	Yearly	Mean	4.354	1.896	33.064	35.997	−4.624	22.375	19.89	4.79
		Std. Dev.	2.017	0.776	1.203	0.581	1.001	3.556	1.127	6.088
2016	98	Mean	6.623	2.5	29.66	37.222	−7.993	17.543	14.69	3.286
		Std. Dev.	3.462	1.013	1.44	0.36	2.406	0.222	0.155	0.228
	206	Mean	6.51	2.491	30.963	36.33	−6.941	20.235	18.88	2.816
		Std. Dev.	1.104	0.855	1.935	0.948	1.864	0.257	0.292	0.452
	210	Mean	4.962	2.797	34.19	36.2	−3.686	22.137	22.7	0.174
		Std. Dev.	1.976	1.192	1.261	0.648	1.008	0.235	0.215	0.919
	Yearly	Mean	6.032	2.596	31.604	36.584	−6.207	19.972	18.76	2.092
		Std. Dev.	2.459	1.018	2.463	0.82	2.591	1.919	3.316	1.506
2017	116	Mean	1.521	1.227	36.17	37.706	−2.166	15.685	14.69	1.73
		Std. Dev.	0.85	0.347	1.505	0.731	1.368	0.395	0.222	0.385
	144	Mean	0.944	0.451	35.304	36.389	−1.872	20.542	19.28	2.2
		Std. Dev.	0.468	0.191	0.77	0.264	0.724	0.325	0.21	0.503
	272	Mean	0.558	0.623	36.721	37.29	−1.523	22.804	22.89	1.462
		Std. Dev.	0.326	0.428	0.335	0.221	0.386	0.457	0.191	0.709
	Yearly	Mean	1.008	0.767	36.065	37.128	−1.854	19.677	18.95	1.797
		Std. Dev.	0.703	0.471	1.138	0.719	0.94	3.026	3.398	0.619

. Yearly salinity profiles are shown in Figure 2.

Water temperatures ranged from 13.8 (bottom layer during April 2017) to 27.2 °C (surface layer during July 2015). Maximum water temperatures were measured during the summer and early autumn period on the surface (June–September; JDOYs 176, 210), and reduced temperatures were measured in April and May (JDOYs 98, 144). Thermal stratification was found during the summer months (June and July; JDOYs 176, 195), in April 2016 (JDOY 98), and in September (JDOYs 210) at deeper layers (19–25 m).

In the northern sample sites, surface salinity was relatively low during the period of October, April, and May (26.9–29.9; Table 1; JDOYs 303, 116, 144). The proportion of brackish water was more pronounced in the northern part of the artificial reef, whereas freshwater presence was weakened in the south (Figure 1).

Regarding chl-a, but for a few observations, maximum concentrations were recorded in the surface layers of the northern sample sites. The highest chl-a values were measured in June (9.0 μg·L⁻¹; JDOY 176) and July (7.7 μg·L⁻¹; JDOY 195) of 2015, as well as during April, May, and September of 2016 (10.4, 7.7 and 7.2 μg·L⁻¹, respectively; Table 1; JDOYs, 98, 206, 210). Moreover, chl-a concentrations decreased to the south and at depths below the surface. An exception was during September 2017 (JDOY 272) where the highest concentration of chl-a was found in the deeper portions of the water column (2.3 μg·L⁻¹).

3.2. Ichthyoplankton Sampling

A total of 4142 individual larvae were collected during the three sampling periods, of which 4054 larvae were identified while the remaining 88 individuals were unassigned. Overall, fish larvae from 75 taxa representing 70 species in 28 families were noted (

Table 2. List of taxa collected at field sites in the outer region of the Thermaikos Gulf in the Aegean Sea offshore of the coastal zone of Kitros, in the Pieria region of Greece. Numbers show the mean abundance (N, larvae/10 m²).

Family	Species	15-Jun	15-Jul	15-Oct	16-Apr	16-May	16-Sep	17-Apr	17-May	17-Sep
Atherinidae	Atherina boyeri	0	0	0	0	1	0	0	136	0
Blenniidae	Blennius gattorugine	0	0	0	0	0	0	0	23	0
	Blennius ocellaris	0	0	0	0	1	0	0	0	0
	Blennius pilicornis	0	6	0	0	0	0	0	0	0
	Parablennius gattorugine	0	0	0	0	4	11	15	10	10
	Parablennius pilicornis	0	0	0	7	0	0	0	0	0
Bothidae	Arnoglossus spp.	0	0	0	0	1	0	0	0	0
	Arnoglossus thori	0	15	7	11	0	10	22	0	8
Callanthiidae	Callanthias ruber	0	0	0	0	0	0	8	0	0
Callyonimidae	Callionymus lyra	15	18	0	12	1	0	20	20	11
	Callionymus maculatus	34	28	0	12	0	0	0	0	0
	Callionymus spp.	15	21	0	27	0	0	0	0	0
Carangidae	Trachurus mediterraneus	0	20	0	0	0	0	0	0	0
	Trachurus trachurus	0	0	0	0	0	4	0	0	0
Centracanthidae	Spicara maena	0	0	0	0	1	0	0	0	0
	Spicara smaris	0	0	0	0	6	0	59	24	6
	Spicara spp.	0	0	0	0	5	0	0	0	0
	Centracanthus cirrus	0	0	19	0	0	0	0	0	6
	Spicara flexuosa	0	0	0	14	1	0	51	0	0
Cepolidae	Cepola macrophthalma	0	21	6	0	0	7	0	0	9
Clupeidae	Sardina pilchardus	0	0	6	16	0	0	0	0	0
	Sardinella aurita	0	50	0	12	1	0	0	0	0
Cynoglossidae	Symphurus nigrescens	0	0	0	0	0	6	0	0	7
Echeneidae	Remora remora	0	7	0	0	0	0	0	0	0
Engraulidae	Engraulis encrasicholus	152	248	52	57	16	261	48	104	324
Gadidae		62	23	0	0	0	0	0	0	0
	Micromesistius poutassou	0	19	0	0	0	0	0	0	0
	Trisopterus minutus capelanus	0	8	0	0	0	0	7	0	0
Gobiesocidae	Lepadogaster candolii	0	0	0	0	0	0	0	0	5
	Lepadogaster lepadogaster	0	0	0	19	0	5	0	0	0
	Lepadogaster purpurea	0	0	0	0	0	0	8	0	0
Gobiidae	Aphia minuta	0	0	0	9	16	9	0	0	0
	Deltentosteus quadrimaculatus	0	0	0	0	0	0	7	0	0
	Gobius niger	0	0	0	0	0	23	0	0	0
	Gobius spp. II	32	42	10	13	0	18	145	96	31
Labridae	Crenilabrus melops	0	0	0	0	0	0	7	0	0
	Crenilabrus spp.	0	18	0	0	3	0	0	0	0
	Symphodus cinereus	0	0	0	0	1	0	7	16	0
Mugilidae	Chelon labrosus	0	0	0	0	0	0	0	0	5
	Liza aurata	0	0	0	0	0	0	0	2	5
	Liza saliens	0	0	0	0	0	11	0	0	6
	Liza spp.	0	0	14	0	0	0	0	0	0
	Mugil cephalus	0	0	0	0	0	0	0	0	18
Mugilidae		0	17	0	0	0	0	0	0	13
Mullidae	Mullus surmuletus	0	0	0	0	0	0	0	0	6
Myctophidae	Ceratoscopelus maderensis	0	0	13	0	0	0	0	0	0
	Hygophum benoiti	0	0	12	0	0	0	0	0	0
	Notoscopelus spp.	0	0	0	0	0	0	0	2	0
Ophidiidae	Parophidion vassali	0	0	0	0	0	106	0	0	0
Pomacentridae	Chromis chromis	0	8	0	0	0	0	0	0	0
Scombridae	Scomber japonicus	0	16	0	0	0	6	0	0	13
	Scorpaena porcus	0	0	0	0	0	5	0	0	12
	Scorpaena spp.	0	0	0	0	0	13	0	0	7
Serranidae	Anthias anthias	0	0	0	0	0	20	0	0	9
	Serranus cabrilla	0	0	0	0	0	20	0	0	18
	Serranus hepatus	63	7	0	0	2	10	0	7	5
	Serranus scriba	0	0	0	0	0	5	0	0	0
Soleidae	Buglossidium luteum	0	0	0	8	0	6	0	0	0
	Solea lascaris	0	0	0	0	0	0	0	0	7
	Solea spp.	0	0	0	0	0	0	8	0	0
Sparidae	Boops boops	0	0	9	0	0	0	0	0	0
	Diplodus annularis	0	0	0	10	0	0	7	0	6
	Diplodus vulgaris	0	0	0	0	0	5	0	0	0
	Pagellus acarne	0	0	0	0	0	16	0	0	17
	Pagellus bogaraveo	0	0	0	0	0	22	0	0	0
	Pagrus pagrus	30	12	0	0	0	9	0	0	9
	Sparus aurata	0	0	0	0	0	0	0	0	13
Sparidae		39	15	0	0	0	0	0	0	9
Sygnathidae	Syngnathus acus	0	0	7	0	0	0	0	7	0
Trachinidae	Trachinus draco	0	0	0	0	0	7	0	0	6
	Total abundance	442	619	155	227	60	615	419	447	601

).

Ten groups of larvae were identified to the family or genus level only including: Arnoglossus spp., Callionymus spp., Crenilabrus spp., Gobius spp., Liza spp., Scorpaena spp., Solea spp., and Spicara spp. Overall the species with the greatest abundance was the larvae of the European anchovy (Engraulis encrasicolus). The second most abundant genera in order were Gobius spp. followed by a relatively abundant but lesser numbers of larvae representing the families Callionymidae, Centracanthidae, and Bothidae.

3.3. Ichthyoplankton Diversity Indices

The Margalef index (

Table 3. Diversity index assessments for ichthyoplankton sampling at field sites in the outer Thermaikos Gulf of the Aegean Sea offshore of the coastal zone of Kitros, in the Pieria region of Greece. The number of species (S) is given, and results are presented for the abundance index Margalef (d), the Brillouin index, Fisher’s statistical parameter a, Pielou’s normality index (J), the diversity index of Shannon-Wiener (H´), and the Simpson’s diversity index (1–Lambda).

Sample Month, Year	S	d	J’	Brillouin	Fisher	H’(Loge)	1-Lambda’
June 2015	9	1.31	0.88	1.89	1.6	1.93	0.82
July 2015	21	3.11	0.78	2.30	4.2	2.38	0.82
October 2015	11	1.98	0.88	2.01	2.7	2.12	0.84
April 2016	15	2.57	0.92	2.37	3.57	2.50	0.90
May 2016	15	3.42	0.79	1.84	6.42	2.13	0.84
September 2016	24	3.60	0.68	2.08	5.02	2.16	0.77
April 2017	15	1.37	0.88	1.89	1.6	1.93	0.82
May 2017	12	1.17	0.78	1.50	2.2	1.38	0.82
September 2017	30	2.98	0.88	1.71	3.7	2.12	0.84

) indicated that measures of diversity were greater in July 2015 (JDOY 195), which was the only sampling performed in the middle of the summer, as well indicated higher diversity in autumn 2015 (JDOY 303). The Shannon and Simpson 1-Lambda indices indicated that April and May 2017 (JDOYs 116, 144) surveys had the most balanced number of individuals per species, with respect to the autumn 2017 period (JDOY 272) (Table 3). This imbalance was caused by an exceptionally large number of Engraulis encrasicolus larvae caught during autumn, 2017 (JDOY 272). The Pielou and Fisher indices indicated that there were differences between samples with respect to the distributions of evenness of abundances between species with decreases in abundances as the sampling progressed towards autumn. The Pielou and Fisher index results were also influenced by relatively high amounts of anchovy larvae that were observed. In general, autumn sampling caught the most diverse sets of species, but with extreme variation in individual samples and with a dominant presence of anchovy and Gobius spp. sampling during spring months gave the most even number of species (Table 3).

3.4. Generalized Linear Models

Generalized linear modeling results for the 24 most abundant ichthyoplankton species and taxa are presented in

Table 4. Generalized linear modeling results for the 24 most abundant species and taxa of ichthyoplankton sampled at field sites in the outer Thermaikos Gulf of the Aegean Sea offshore of the coastal zone of Kitros, in the Pieria region of Greece. SIG = significance values (p ≤ 0.050 are significant and in bold).

	Intercept		Year		Site		Surface Salinity (PSU)		Surface Temp (°C)		chl-a ug/L		Surface chl-a ug/L		Julian Day of the Year		Salinity Gradient (PSU)		Salinity (PSU)		Temp Gradient (°C)		Temp (°C)
Deg. freedom	1		2		15		1		1		1		1		1		1		1		1		1
	W C S	S I G	W C S	S I G	W C S	S I G	W C S	S I G	W C S	S I G	W C S	S I G	W C S	S I G	W C S	S I G	W C S	S I G	W C S	S I G	W C S	S I G	W C S	S I G
Anthias anthias	0.00	0.957	0.65	0.721	17.98	0.264	7.05	0.008	6.28	0.012	2.34	0.126	0.32	0.570	1.46	0.227	6.69	0.010	8.46	0.004	4.09	0.043	6.35	0.012
Aphia minuta	1.24	0.265	16.27	0.000	19.95	0.174	1.03	0.309	1.45	0.228	0.70	0.402	0.21	0.650	7.17	0.007	0.00	0.960	5.42	0.020	0.53	0.467	4.86	0.027
Arnoglossus thori	0.75	0.385	0.42	0.809	28.39	0.019	1.78	0.182	1.74	0.187	0.23	0.632	0.04	0.833	1.47	0.225	0.08	0.776	1.16	0.282	0.00	0.986	1.29	0.256
Atherina boyeri	1.63	0.201	3.48	0.176	20.19	0.165	0.05	0.815	0.00	0.987	1.42	0.233	0.02	0.897	2.80	0.094	0.16	0.690	1.01	0.314	0.80	0.372	1.50	0.221
Callionymus lyra	1.40	0.237	1.52	0.468	16.85	0.328	0.34	0.558	0.24	0.626	0.01	0.937	2.25	0.134	2.41	0.121	0.18	0.671	1.03	0.309	0.03	0.861	1.02	0.312
Callionymus maculatus	6.98	0.008	0.64	0.726	22.47	0.096	14.55	0.000	12.78	0.000	0.16	0.691	10.23	0.001	0.43	0.511	3.83	0.050	18.24	0.000	0.91	0.341	16.45	0.000
Callionymus spp.	0.32	0.571	2.85	0.241	13.23	0.585	13.80	0.000	13.01	0.000	0.84	0.359	0.00	0.949	0.23	0.632	2.33	0.127	6.05	0.014	1.15	0.283	0.32	0.571
Engraulis encrasicholus	0.11	0.739	9.66	0.008	47.48	0.000	0.32	0.572	0.88	0.349	0.07	0.789	6.35	0.012	1.16	0.281	1.49	0.223	1.62	0.203	2.76	0.096	2.04	0.153
Gobius niger	1.53	0.216	26.42	0.000	21.66	0.117	2.96	0.085	0.94	0.333	0.56	0.453	0.13	0.715	4.86	0.027	1.84	0.175	0.00	0.972	0.46	0.500	0.23	0.632
Gobius spp. II	1.97	0.160	7.62	0.022	17.47	0.291	0.89	0.345	0.55	0.457	1.01	0.314	1.21	0.271	4.01	0.045	0.26	0.610	5.07	0.024	0.00	0.985	4.20	0.040
Lepadogaster lepadogaster	0.24	0.622	2.22	0.330	16.75	0.334	0.98	0.322	1.23	0.267	0.07	0.792	0.05	0.818	0.07	0.794	0.53	0.468	2.47	0.116	0.59	0.444	2.36	0.124
Liza saliens	2.13	0.144	0.66	0.718	15.30	0.430	1.48	0.223	2.09	0.148	1.34	0.248	0.13	0.713	0.10	0.750	66.68	0.000	0.36	0.546	72.65	0.000	0.08	0.783
Mugilidae	0.00	0.969	3.47	0.177	21.01	0.136	17.90	0.000	15.87	0.000	0.01	0.924	2.73	0.098	0.07	0.797	2.47	0.116	9.71	0.002	0.68	0.410	7.77	0.005
Pagellus acarne	0.46	0.500	1.24	0.539	18.26	0.249	7.92	0.005	8.76	0.003	0.04	0.839	0.99	0.319	1.86	0.172	2.75	0.098	14.08	0.000	2.73	0.098	12.43	0.000
Pagellus bogaraveo	0.31	0.579	18.43	0.000	16.20	0.369	0.02	0.893	0.01	0.904	2.60	0.107	0.35	0.556	10.95	0.001	0.02	0.895	1.09	0.298	0.00	0.958	0.39	0.532
Pagrus pagrus	0.00	0.990	8.32	0.016	19.54	0.190	3.84	0.050	3.82	0.051	3.20	0.074	1.15	0.283	0.67	0.414	2.42	0.120	4.87	0.027	2.06	0.152	3.85	0.050
Parablennius gattorugine	8.68	0.003	4.86	0.088	22.82	0.088	0.02	0.880	0.04	0.840	0.41	0.521	2.36	0.125	11.43	0.001	0.30	0.582	4.41	0.036	0.21	0.649	5.16	0.023
Parophidion vassali	0.31	0.579	3.28	0.194	18.97	0.215	1.72	0.190	0.45	0.501	1.70	0.192	0.82	0.364	0.11	0.735	1.02	0.312	0.53	0.467	0.10	0.754	0.02	0.891
Sardinella aurita	0.00	0.949	0.55	0.761	22.06	0.106	0.66	0.417	0.24	0.621	4.33	0.037	5.44	0.020	0.43	0.513	1.09	0.297	0.27	0.606	2.43	0.119	0.05	0.817
Scomber japonicus	0.11	0.739	4.18	0.124	28.51	0.019	0.59	0.441	1.66	0.198	0.35	0.557	6.88	0.009	8.33	0.004	3.14	0.077	1.89	0.169	5.72	0.017	3.03	0.082
Serranus hepatus	1.61	0.204	2.95	0.229	19.31	0.200	3.77	0.052	4.08	0.043	2.36	0.125	0.74	0.391	0.03	0.858	0.18	0.674	0.99	0.319	0.05	0.818	0.84	0.360
Sparidae	2.18	0.140	11.25	0.004	26.25	0.035	0.95	0.330	1.34	0.246	0.66	0.416	4.87	0.027	0.02	0.880	1.05	0.306	0.02	0.902	0.86	0.354	0.03	0.866
Spicara smaris	2.63	0.105	6.97	0.031	25.29	0.046	0.06	0.812	0.22	0.638	3.80	0.051	0.20	0.651	7.73	0.005	0.17	0.678	3.54	0.060	0.60	0.439	4.05	0.044
Spicara spp.	0.00	0.969	7.37	0.025	20.13	0.167	0.46	0.499	0.68	0.411	1.60	0.207	0.20	0.654	3.00	0.083	0.46	0.500	1.20	0.273	0.06	0.812	1.04	0.308

. Of the 75 total species and taxa, 17 had significance (p ≤ 0.050) for the sample year, 11 had significance for the sample site, 11 had significance for surface salinity, 13 had significance for surface temperature, 2 had significance for chl-a, 8 had significance for surface chl-a, 11 had significance the JDOY, 7 had significance for the salinity gradient, 18 had significance for salinity, 7 had significance for the temperature gradient, and 20 had significance for temperature.

3.5. Species-Based Non-Metric Multidimensional Scaling

The most influential variables in the species-based NMDS upon species distributions within the ordination space appeared to have been surface salinity, surface temperature, JDOY, salinity gradient, and water temperature (Table 4) as the majority of species and clusters correspondingly fell on the positive side of axis 1. With regards to axis 2, the distributions of species in the ordination space seemed relatively evenly influenced by the corresponding significant predictor variables as somewhat equal numbers of species and taxa and clusters were distributed in both the negative and positive directions from the zero-point origin of this axis. Of the predictor variables used in the species-based NMDS (Figure 3) only sample site and water temperature gradient did not have any significant correlations with at least one axis in the NMDS (Table 4, Figure 3).

3.6. Season-Based Non-Metric Multidimensional Scaling

The season-based non-metric multidimensional scaling (Figure 4) showed the clusters of the nine seasons of sampling where May 2016, May 2017, April 2016, and April 2017 JDOY’s form the cluster of spring; September 2016, September 2017, and October 2015 form the cluster of autumn; and June 2015 and July 2015 form the cluster of summer.

The Margalef index (Table 3) shows that the diversity was larger in July 2015, which was the only sampling performed in the middle of the summer, as well as the autumn sampling of 2015, since during that time ichthyoplankton includes both summer and autumn species that have been reproduced, such as the species Pagellus acarne. The Shannon and Simpson 1-Lambda index shows that April and May 2017 surveys had the most balanced number of individuals per species, in relation to the 9th sampling that has the most species. This imbalance is due to the exceptionally large number of anchovy larvae, caught during the autumn sampling of 2017. The Pielou and Fisher indexes, which measure whether the abundance is evenly distributed between species and the differences between samples, show a decrease as the samplings progress toward autumn, which was also caused by the huge amounts of anchovy larvae that created this imbalance. In general, the autumn sampling caught the most species but with extremely varying samples, with the dominant presence of the anchovy and Gobius species. The spring samplings gave the most even number of species.

4. Discussion

Several captured species are of high and moderate commercial value, such as the common sea bream, Pagrus pagrus, the stripped red mullet Mullus surmuletus, and the annular sea bream, Diplodus annularis. Other species captured represent major proportions of the commercialized marine fisheries in the Aegean Sea and included the European anchovy (Engraulis encrasicolus), the chub mackerel (Scomber japonicus), and the Mediterranean horse mackerel (Trachurus mediterraneus). Larvae of three mesopelagic species belonging to the family of Myctophidae were caught in small abundances (Ceratoscopelus maderensis, Hygophum benoiti, and Notoscopelus spp.).

Though the Mediterranean Sea is mostly oligotrophic, it nonetheless presents environmental characteristics that enhance the fertility and recruitment of fishes at certain times of the year or in connection with relatively localized habitats and hydrographic structures [38]. In the region of Thermaikos in the Aegean Sea, seasonal variability in environmental parameters affects water column stratification and fluctuations in continental-derived river run-off, hence, their importance. During spring and headed into summer, a seasonal thermocline forms as the daily photoperiod becomes extended and the solar radiation is increasing. Concurrently, the fresh-water input increases, not only because of high rain in autumn but also because of high flow rates in the rivers due to the ice melting from the mountains of the region.

Axios (annual average discharge 118 m³/s) and Aliakmonas rivers (31.9 m³/s; [39], which flow to the north, are the primary source of fresh water in the study area. These waters arrive from the north and move parallel to the west coastline outside the Thermaikos Gulf. According to [25], nutrient loadings from Axios to the coastal zone are significant but do not appear to be too much in excess in relation to silica loadings (negative ICEP-N and ICEP-P indicators for coastal zone eutrophication potential; Billen and Garnier, 2007). Nutrient fluxes combined with short residence time and high flush rates reduce eutrophication risks in the coastal zone. Aliakmonas nutrient loads are by far less than Axios [39].

Despite the limited size of the investigated area, a pattern of freshwater intrusion in the northern part is recorded in every survey. This low saline water body is confined up to 5 m and progressively mixed with the ambient water to the south. Thus, there is a distinct separation of the northern part from the rest of the region regarding salinity (Figure 2). As [40] refer, fish spawning grounds are associated with freshwater nutrient-rich influx areas. Axios Aliakomanas and two other small continental water bodies (Gallikos and Loudias rivers) are fulfilling that role for the investigated area. This is probably the reason for salinity significance in the distribution of some larvae species.

It is also known that temperature and climate change affect the diversity and geographical distribution of fish larvae. The temperature is significant in the recruitment, reproduction, growth, and behavior of fish species [41], and affects food quality and availability [42,43]. During the last decades, the temperature increased by 1.1 °C at the surface and around 0.7 °C at 80 m depth [44]. These increases produced spatial shifting for many fish species since the temperature ranges permitted their migration and survival. For example, the geographical area of the round sardinella (Sardinella aurita) on the Mediterranean Spanish coast expanded to the north [45], and many exotic species enter the Mediterranean through Suez Canal and Gibraltar [46,47,48]. Especially, the expansion of alien species in the Eastern Mediterranean due to sea temperature rise is a major concern for the scientific community since the mid-’50s [49,50].

Furthermore, temperature is one of the main factors that affect the spawning season of thermophilic fish species. Prolongation of the spawning season has been detected in the Black Sea, where anchovy ichthyoplankton, usually collected from June to September, extended their presence from May to October in the period 2011–2016, and the larvae of Sprattus sprattus have been found all year-round instead of the winter period in the same region [51]. Another example of spawning season alteration of round sardinella between different areas due to sea temperature is the limited duration in the northern part of the West Mediterranean (July–September; [52]), compared to the southern African coast, which lasts for several months [53] Finally, [40] demonstrated that temperature, as well as depth and chl-a values, are significant parameters in the spawning ground selection for small pelagic species. Considering the above, it can be declared that the study area has all the elements that characterize a spawning ground, while the high temperature together with the nutrient and chl-a availability are the parameters that explain the high abundance of anchovy throughout the entire year.

The high anchovy abundance was observed in previous studies in other areas of the North Aegean [54] The same pattern was observed in the Pagasitikos Gulf, where the high concentrations of larval abundance in early autumn were also attributed to the high prevailing temperature, as well as to the seasonal peak in phyto- and zoo- plankton abundance [55]. The highest number of taxa recorded in September has also been previously reported for the Pagasitikos Gulf and the entire Aegean Sea [55,56].

The generalized linear modeling results for the 24 most abundant ichthyoplankton species and taxa (Table 4) showed significant values for salinity (PSU), chl-a ug/L, and temperature (factors that influence larval growth). The anchovy, which was the most abundant species of this research, showed a significant value related to surface chl-a ug/L. Since fish production partly results from fish growth [57], the chlorophyll concentration is a good indicator of the phytoplankton primary production [58]. Therefore, the abundance of anchovy may be explained by the seasonal chlorophyll concentration of the area.

Apart from the dominance of anchovy in the sampling area, species composition was affected by seasonality (Figure 4). Samples from different years but obtained in the same seasons show the greatest similarities in abundances and diversities of ichthyoplankton. However, all fishes in the sample region do not reproduce with the same intensity in exactly the same months each year. Namely, the periods of largest abundances of particular species are dependent upon complex dynamics of the maturity cycles of the species that are affected by the hydrographical conditions of the area. It is known that an early spring season results in an early beginning of spawning in fishes and therefore the assemblages of ichthyoplankton are affected accordingly. Spring spawners recognize the rise of water column temperature as a spawning signal during the first appearance of a thermocline need that water reaches a certain temperature value to start spawning. Likewise, a prolonged summer season with high temperatures lasting well into autumn often leads to a delayed start of spawning for autumn spawners. In these species, the start of spawning depends on the initial seasonal-based reductions in seawater temperature.

The spawning seasons do not start identically each year but can differ by a few weeks. This may explain the small differences in samples that were theoretically collected in the same seasons. In the ichthyoplankton sampling of Kitros, the greatest diversity was found in July 2015, when the sampling survey was performed later than usual, likely as many species whose eggs had hatched and larvae formed. Spring samples, according to diversity indices, showed more balance without the dominance of any single species. On the contrary, autumnal samples were greatly differentiated, since anchovy larvae were very abundant causing an imbalance in sample composition. In the spring, the anchovy has not yet entered the main spawning phase, and taking into account the time needed for egg hatching, this species is found in the samples late in the spring and during the summer period. Mediterranean anchovy has a short life expectancy which is 3–4 years; [59], and matures with the completion of the first year of life. It can be considered a “birth-pulse” species and its reproduction is extended from May to October [60]. The peak of abundance is in July with a spawning peak taking place at the end of the spring, coinciding with the time when continental run-off in the sur- face layer reaches its maximum extent. Therefore, high egg and larval abundances for the anchovy are often associated with areas of freshwater input [60]. By the end of the spring, the resources upon which anchovy larvae survive are limited due to the environmental conditions of that season and as enrichment from the surface is reduced.

Planktonic anchovy larvae that develop in the area of Kitros are carried along by the currents formed by the river runoffs along the western part of Thermaikos Gulf. Despite the fact that the larvae are carried away from their spawning areas, the environment remains ideal for their growth as continental run-off induces high productivity and provides trophic advantages. Anchovy is apparently well adapted and may take better advantage of this factor compared to other fish species, as anchovy has the ability to dwell in habitats with varying salinity levels. In contrast, for larvae of other species, their potential trophic advantages by dwelling in those more productive waters are offset by the drawback of having to survive at lower-than-usual salinity levels. For example, ref. [13] found that the anchovy was the only species present in areas with extremely low salinity values. Similar displacements from the area of spawning have been reported in clupeoid species in other geographical areas e.g., [61]. However, since the anchovy is a pelagic species, advection does not significantly affect its survival. In contrast, transport away from the spawning areas could be a serious predicament for the future recruitment of larvae of inshore benthic species, whose adults are bottom-dwellers [62].

Apart from July 2015, most species were found during autumn sampling, since this is the period when both summer and spring species are present, coexisting with autumn species already spawning, such as Pagellus acarne. Most Mediterranean fishes are spring and summer spawners [63]. Thus, high taxonomic diversity and larval densities are observed in the spring/summer surveys in estuarine and coastal waters of the Mediterranean Sea [13,60,61].

To conclude, the marine protected area of Kitros has all the elements that characterize a spawning ground. High temperatures together with nutrient and chl-a availability explain the high abundance of anchovy. The nine samplings showed that seasons play a significant role in ichthyoplankton assemblages as species composition was characterized by seasonality.