Amazon River plume habitats shape planktonic cnidarian assemblages in the Western Atlantic

The impact of the Amazon River freshwater plume on planktonic cnidarians over neritic and oceanic provinces is unknown. To provide further knowledge we took advantage of an oceanographic cruise performed in October 2012 in the Western Atlantic off the North Brazilian coast (8°N, 51°W—3.5°S, 37°W). A complex and dynamic system was observed, with strong currents and eddies dispersing the plume over a large area. Our results show that the Amazon River shapes marine habitats with a thin highly productive surface layer compressed by a deeper oxygen minimum zone both over the shelf and in the open ocean. We hypothesized that such habitat structure is particularly advantageous to planktonic cnidarians, which have low metabolic rates, being able to survive in hypoxic zones, resulting in high species richness and abundance. Over the shelf, distinctions were sharp and the area under the influence of the plume presented a diverse assemblage occurring in large abundance, while outside the plume, the hydromedusa Liriope tetraphylla was dominant and occurred almost alone. Divergences in the oceanic province were less pronounced, but still expressive being mostly related to the abundance of dominant species. We concluded that Amazon River plume is a paramount physical feature that profoundly affects the dynamics of the mesoscale habitat structure in the Western Equatorial Atlantic Ocean and that such habitat structure is responsible for shaping planktonic cnidarian assemblages both in neritic and oceanic provinces.

The author(s) received no specific funding for this work. (SIMPER) analysis was performed in order to identify key species and their contribution to 149 similarity within the groups defined in the cluster analysis. 150 To identify associations between representative planktonic cnidarian taxa (species 151 occurring in more than 21 stations and species with high abundance in few stations) and the 152 physical environment, a constrained ordination analysis was performed. Detrended Canonical 153 Correspondence Analysis (DCCA) revealed a small length of environmental gradients (<3), 154 indicating that a linear method was appropriate, and thus Redundancy Analysis (RDA) was 155 selected [57]. Mesoscale physical processes were included as dummy categorical explanatory 156 variables (neritic and oceanic habitats, presence of ARP, predominant current, presence of 157 cyclonic and anticyclonic eddies). 120 µm mesh zooplankton biomass (considered as food 158 availability), maximum value of fluorescence (as a proxy of biological productivity), maximum 159 value of dissolved oxygen and surface temperature and salinity were included as continuous 160 explanatory variables. Monte Carlo test was used to test the significance of the first and all 161 canonical axes together [57].

Spatial distribution patterns 238
Total medusa abundance was higher and more variable over the continental shelf, 239 ranging from 1.4 to 1710 ind. 100 m -3 (891.7±1161.3 in average). In this province, while medusa 240 species richness was higher in the stations influenced by the ARP, high abundances occurred in 241 stations both inside and outside of the ARP. In oceanic waters, highest medusa abundance 242 occurred at stations located in the area influenced by the ARP and species richness was similar 243 both inside and outside the influence of ARP (Fig. 4). Abundances in neritic stations also tended to be lower, but very high abundance (up to 3381.3 269 ind. 100 m -3 ) occurred at station 9 (Fig. 4). influence of the ARP (Fig. 6, Table 2). Also widespread, but with lower densities, Diphyes dispar 274 and N. bijuga were more abundant in stations under the influence of the ARP or near its 275 boundaries both in the neritic (significant for both species) and oceanic habitats (significant for 276 N. bijuga, Fig. 6, Table 2). All other abundant siphonophores occurred exclusively at oceanic 277 stations and the neritic station 45, located near the shelf break (Fig. 6)

Assemblage structure 292
The cluster analysis depicted three main groups with low resemblance to each other 293 (Fig. 7). Group A, with 39.7% similarity, included the neritic stations without influence of the 294 ARP, with the exception of the station 45. The group was represented mainly by L. tetraphylla in 295 high abundance ( Table 2). The two neritic stations under influence of the ARP belonged to Group 296 hyalinum, M. kochii and D. dispar (Table 3)  Group C, that occurred mainly in oceanic stations, encompassed three subgroups and 306 two outliers. Stations 45 and 10 were located near the shelf break and considered outliers.
Subgroup C1 clustered stations 34 to 38, located in the southeastern portion of the study area 308 (Fig. 7). It was mainly represented by C. appendiculata, B. bassensis and A. tetragona and 309 differed from other oceanic groups by the low occurrences of E. mitra and S. chuni ( Fig. 6; Table  310 2). The similarity within the group was 69.2%. Subgroup C2 included the remaining oceanic 311 stations outside the influence of the ARP. With an average similarity of 69.5%, C. appendiculata, 312 E. mitra and B. bassensis were the main representative of this subgroup ( Fig.7; Table 3). Except 313 for station 17 and 22 placed in subgroup C2, all oceanic stations under the influence of the ARP 314 and stations 25 and 26 located near its limit were included in subgroup C3 ( Fig. 7

Responses to mesoscale processes and environmental gradients 319
The first two axes of the RDA explained 54.9% of planktonic cnidarian species variance 320 (Table 4). Monte Carlo test showed that the first (F-ratio = 18.3, P-value = 0.002) and all canonical 321 axes together (F-ratio = 6.3, P-value = 0.002) were significant. Axis 1 (37.8% of variance) was 322 mainly related to the oceanic/neritic gradient. Zooplankton biomass and cold-core cyclonic 323 eddies were positively related to this axis. The second axis (17% of variance) was negatively 324 related to surface salinity and cyclonic eddies, and positively related to the ARP, fluorescence 325 and zooplankton biomass. Axes 3 and 4 explained together less than 10% of species variance 326 and were not considered (Fig. 8, Table 4).  Most oceanic species were closely related to the left portion of the first axis with few 347 relation with axis 2, which represented the ARP and surface salinity gradient, reflecting their 348 wide distribution over the oceanic province and low effect of the ARP on their distribution. Other oceanic species, such as D. bojani, E. mitra and R. velatum correlated with both the negative 350 portion of axis 1 and the positive portion of axis 2, indicating their higher abundance in the 351 oceanic environment under influence of the ARP (Fig. 8). ,43] in two stations (8 and 9) to the west of the study domain (Fig. 2c). This is the zone where 368 the NBC is retroflected to the north (Fig 2a), spreading the ARP over the entire length of the 369 shelf [58]. In addition to the freshwater input lowering salinity, the ARP discharges a massive 370 amount of nutrients, organic matter and sediments [24], boosting the primary production 371 (evidenced by the fluorescence; Figs. 2g, 3d). Consequently, higher trophic levels, including 372 planktonic cnidarians benefit from the higher food availability derived from bottom-up 373 processes.
Over the continental shelf, the ARP was restricted to the first 8 m of the water column. 375 Below lied an oxygen minimum layer (Figs. 3 a, e), likely in consequence of the high rates of 376 organic matter that sink and fuel microbial respiration [59]. These features compressed the bulk 377 of productivity in the ARP, both vertically and horizontally (Fig. 9). Although with the oblique 378 trawls performed in our study, we could not infer the exact vertical position of the huge amount 379 of cnidarians observed in the area of influence of the ARP and its relation with the oxygen 380 minimum layer, such three-dimensional habitat configuration seems particularly beneficial for 381 cnidarians. While cnidarians usually are not restricted by oxygen concentration due their low 382 metabolism, low oxygen levels compress the suitable habitat for fish and other predators with 383 higher metabolic demand to the thin surface layer [32,60-62]. In addition, in this thin productive 384 layer the water is turbid due to the sediment runoff and suspended particulate matter.