A New Characterization of the Upper Waters of the central Gulf of México 1 based on Water Mass Hydrographic and Biogeochemical Characteristics

1Facultad de Ciencias Marinas, Universidad Autónoma de Baja California, Transpeninsular 11 Tijuana-Ensenada, no. 3917, Fraccionamiento Playitas, CP 22860. Ensenada, Baja 12 California, México. 13 2Instituto Tecnológico de Guaymas/ Tec. Nacional de México, Guaymas, Sonora, México. 14 3LEGOS, CNRS/IRD/UPS/CNES UMR 5566, 18 av. Ed Belin, 31401 Toulouse Cedex 9, 15 France 16 4Insitituto Geofisico del Perú. Lima, Perú. 17 5Instituto de Investigaciones Oceanológicas, Universidad Autónoma de Baja California, 18 Transpeninsular Tijuana-Ensenada, no. 3917, Fraccionamiento Playitas, CP 22860. 19 Ensenada, Baja California, México. 20 6Scripps Institution of Oceanography, University of California, San Diego, 9500 Gilman 21 Drive, La Jolla, California 92093, USA. 22 7Departamento de Oceanografía Biológica, Centro de Investigación Científica y de 23 Educación Superior de Ensenada (CICESE), Baja California, Carretera Ensenada-Tijuana 24 No. 3918, Zona Playitas, 22860 Ensenada, Baja California, México. 25


Introduction.
Circulation in the central Gulf of Mexico's (GoM) is dominated by the Loop Current (LC) and its associated eddies.Anticyclonic Loop Current Eddies (LCE)  200 -300 km in diameter separate from the LC every 4 to 18 months (Sturges and Leben, 2000;Hall and Leben, 2016).Another feature associated with the LC is the separation of relatively smaller cyclonic and anticyclonic eddies throughout the basin, which interact in an apparently turbulent manner (Schmitz, 2005;Hamilton, 2007a).These eddies extend vertically from a few hundred to about a thousand meters and appreciably influence the surface dynamics by modifying the circulation of the GoM (Morey et al., 2003a).The position of the LC within the gulf is variable, and the level of intrusion into the northeastern GoM varies temporally and spatially (Bunge et al., 2002;Delgado J. A. et al., 2019).
Near the surface, the spatio-temporal variability in temperature, salinity and dissolved oxygen (DO) reflect the LC, LCE and other eddy dynamics, freshwater inputs from river discharge, and seasonal processes such as heat fluxes, evaporation and wind stress that influence the depth of the mixed layer (Morey et al., 2003b;Müller-Karger et al., 2015;Portela et al., 2018;Damien et al., 2018).A major source of variability in the northern GoM is the Mississippi River flow, which has been shown to influence areas hundreds of kilometers from its discharge zone (Morey et al., 2003a) over the first 50 m of the water column (Jochens & DiMarco, 2008;Portela et al., 2018).Together, the aforementioned mechanisms influence water mass characteristics in approximately the first 250 m (or more) of the water column.For example, upon entering the GoM, the Caribbean Surface Water (CSW) affects salinity, temperature, and density with values of 34.5 to 36.6;T ≥ 25 ºC, and https://doi.org/10.5194/bg-2019-340Preprint.Discussion started: 23 September 2019 c Author(s) 2019.CC BY 4.0 License.σθ ≤ 24.5 kg•m-3 (Carrillo et al., 2016).Below the CSW, North Atlantic Subtropical Underwater (NASUW, hereinafter referred to as SUW) can be identified by a salinity between 36.5 to 36.9 at  100 to 150 m (Herrig, 2010;Hamilton et al., 2018).The Gulf Common Water (GCW) is distinguished by the relatively homogeneous vertical distribution of its thermohaline properties, with salinity ranging from 36.3 to 36.49(Elliott, 1982;Merrell and Morrison, 1981).Underneath the SUW and GCW, Tropical Atlantic Central Water (TACW) is found between 300 and 600 m, and is characterized by a DO minimum of 2.3 ml•L-1, T from 7.9 to 20 °C, S from 34.9 to 36.6, and θ from 26.25 to 27.2 kg•m-3 (Vidal et al., 1994;Gallegos, 1996;Carrillo et al., 2016;Portela et al., 2018).The main sources of variability in the physical and chemical properties of the surface to approx.250 m (above 26 kg•m-3) can be related to changes in the relative proportions of water masses.
There have been limited surveys of the hydrographic characteristics of the central GoM and Yucatan Channel within Mexico's Exclusive Economic Zone (including the Campeche Basin (CB)) based on in situ data, and of those, several have been limited to relatively small regions: For example, Morrison et al. (1983) studied the distribution of physical-chemical properties of the water masses (GCW, TACW, Antarctic Intermediate Water (AAIW) and the mixture of Caribbean Intermediate Water (CIW) and North Atlantic Deep Water (NADW) and the NADW) in the northwestern GoM during winter.Similarly, Vidal et al. (1994) also investigated the spatial distribution of thermohaline properties and DO of the GCW, SUW, TACW, AAIW, as well as the mixture between CIW and NADW, and NADW in the western region of the GoM during winter and spring.Among these efforts, Rivas et al. (2005) studied the area of the Yucatan Channel, they found five different water masses (SUW, 18º SSW, TACW, AAIW y NADW).Finally, Hamilton et al. (2018) performed an analysis with highresolution data from the deeper waters (SUW, AAIW y NADW) of the western and eastern in the GoM, with results that were consistent with the findings of the previous authors.
Obviously, different water masses may be present depending on the region of the GoM that is being studied.
In particular, the above authors focused on the role of the dominant LCE's on the hydrographic characteristics of the central and western GoM (Fig. 1a).However, their proposed classification did not include near-surface waters; for example, lower salinities (likely due to river inputs) were not included.Excluding water masses with lower salinities in the classification scheme limits the inferences that can be made regarding source waters.This points to the necessity of generating a more detailed classification system in the surface layers above the 26 kg•m-3 isopycnals, which includes the full range of thermohaline properties of water masses.When DO concentrations are added to the Θ-SA diagram as a third axis, it can be observed that DO shows a high variability (> 200 μmol•kg-1) upwards of the 26 kg•m-3 isopycnal (Fig. 1b).This change in DO is a result of biogeochemical processes via, photosynthesis, respiration, and exchange with the atmosphere, which also lead to changes in dissolved inorganic carbon (DIC) and nutrients.
From a biogeochemical perspective, the surface waters of the deep GoM are considered oligotrophic as they are relatively isolated from the more eutrophic waters of the coast and continental shelves (Heileman and Rabalais, 2009;Damien et al., 2018;but  productivity in subsurface waters maybe two to three times higher (El-Sayed, 1972;Biggs and Ressler, 2001).Dynamic features such as mesoscale processes, river inputs, the extent of the seasonal LC incursion, and wind stress can greatly alter the distribution of chemical properties in the GoM (Linacre et al., 2015;Damien et al., 2018).Overall, the effect that water masses have on the seasonal extension of the mixed layer is not well understood, though its deepening and shallowing play an important role in the rates of primary production (Damien et al., 2018).
In this work propose a classification for water masses lighter than 26 kg•m-3 that more precisely defines the ranges of thermohaline circulation and DO of the CSW and GCW, thereby providing a better basis for understanding the processes associated with water mass formation, distribution, and biogeochemistry in surface waters of the central and western regions of the GoM.Our purpose is to provide a better tool for studying the drivers that modulate water mass distribution and its formation in surface waters, as well as the links between water masses and their biogeochemical properties.The reclassification includes an adjustment of the thermohaline range of CSW and the GCW.In this work also propose the formal recognition of Freshwater Influenced Surface Water (FISW) that is characterized by riverine influence.Finally, examine the role of CSW in the biogeochemistry of the GoM by comparing the seasonal variations in Tθ and S in our in situ water to the climatological database CARS 2009.

1. Data collection
Five oceanographic cruises covering the central region of Mexico's Exclusive Economic Zone were carried out in November 2010, July 2011, February-March 2013, August-September 2015, and July 2016 (XIXIMI-01 through XIXIMI-05, respectively) on board the R/V Justo Sierra (Fig. 1c).During these campaigns, a minimum of 30 and maximum of 51 stations per cruise were occupied, and a total of 235 hydrographic casts were performed to characterize the vertical distribution of potential temperature (Tθ), salinity (S), potential density (σθ), and DO.An SBE 911plus CTD was used; the instrument and sensors were serviced and calibrated regularly.
In addition to CTD casts, water samples were collected for measurements of Dissolved Inorganic Carbon (DIC), nutrients, and DO analyses in 10 or 20 L Niskin bottles at 12 set depths between the surface and bottom.The protocols and best practices established by Dickson et al. (2007) were followed for DIC sample collection.For the collection of nutrient samples, 50 ml of seawater were filtered through Whatman GF/F filters previously calcinated at 450 °C for 2 hours, transferred to centrifuge tubes and frozen.Each sample was transported frozen to the laboratory for later analysis.During each cruise, seawater was also routinely sampled for DO (evaluated by the Winkler method) measurements and to calibrate the CTD data.Additionally, the apparent oxygen utilization (AOU) was calculated from DO, T, and S using TEOS-2010 equations.AOU is defined as the deviation of the measured dissolved oxygen from a DO concentration in equilibrium with the atmosphere (Benson and Krause, 1984).When calculating the AOU the DO is corrected for temperature.This allowed us to determine if DO concentrations were in equilibrium with oxygen in the atmosphere.

1. Identification of water masses
An analysis of Tθ-S diagrams was carried out for the five cruises; Tθ and S were converted to conservative temperature (Θ) and absolute salinity (SA) as described by McDougall and Barker (2011).For water mass identification, in this work first used the limits described by Vidal et al. (1994), Morrison et al. (1983) and Nowlin et al. (2001) and the recent classification proposed by Portela et al. (2018), as shown in figure 1a.

Seasonal variation
Two of the five cruises took place during the late fall and winter (2010 and 2013), and three during summer (2011, 2015, and 2016).Since sampling in winter and summer covered approximately the same region of the GoM (Fig. 1c and 3), in this work could perform a separate seasonal analysis of hydrographic and geochemical characteristics for densities lower than 26 kg•m-3 in the Θ-SA diagrams using the Portela et al. (2018) classifications (Fig. 2).DO was incorporated into the diagrams to evaluate the role of seasonality on its vertical distribution in relation to water masses.It was noted that the depth of the 26 kg•m-3 isopycnal varied by more than 100 m regardless on the time of year (Fig. 3 and Supplementary Fig. 2).

3. Tθ-S patterns above 26 kg•m-3
Four patterns were visually identified in the Tθ-S diagrams by focusing on the most distinctive characteristics for densities less than 26 kg•m-3 (Fig. 4).The four distinct Tθ-S patterns (indicated by parallelograms) shown in Table 1 and figure 4 had the following characteristics: • The blue Tθ-S pattern was characterized by a subsurface salinity maximum and lower concentrations of DO associated with the Subtropical Underwater (SUW) (Fig. 2b    and 4).
• The pink Tθ-S pattern was characterized by shallow fresh waters (low than 36; see Table 1) that are likely associated with river inputs and their offshore transport.In this study, this water mass is referred to as Freshwater Influenced Surface Water (FISW) (Fig. 2b and 4).
• The green Tθ-S pattern was observed during summer cruises and was characterized by a wide range of temperatures (23.7 to 27.5ºC; see Table 1) and salinity, and a subsurface DO maximum ( 232 μmol•kg-1) at a density of approximately 24.5 kg•m-3.This pattern is heavily influenced by the CSW.
• The red Tθ-S pattern was observed during winter and had a narrow salinity range (36.4 to 36.6; see Table 1), indicating the limited influence of the CSW coupled with seasonally lower temperatures (22.9 to 23.2 ºC; see Table 1).This so so-called Gulf Common Water (GCW) is closer to the surface during winter.
Finally, in this work carried out a reclassification of the range limits for the water masses lighter than 26 kg•m-3.This reclassification was done using a Matlab program that separated and binned the data based on the four Tθ-S patterns previously described (Table 1): these classification established by Vidal et al. (1994), Morrison et al. (1983) and Nowlin et al. (2001).A final readjustment was done based using Tθ-S patterns analysis of the existing thermohaline ranges (σθ, Tθ, S; Table 1) and the DO concentration of the water masses that were observed in the Tθ-S diagrams.An extended description of the code with the criteria for classification is provided in Appendix A.

4 Analysis of the vertical variability of σθ, Tθ and DO in surface waters
Sections of the vertical distribution of σθ, Tθ and DO were made for each cruise (2010( , 2011( , 2013( , 2015( and 2016, Fig. 5a-j and 6) , Fig. 5a-j and 6) to examine differences in the density, temperature and DO to arising from different oceanographic conditions (Fig. 5 and 6).

4. Analysis of chemical variables
To determine the concentration of DIC, coulometric methods were used following the methodology described by Johnson et al. (1987).Reference materials were provided by the laboratory of Dr. A. Dickson of Scripps Institution of Oceanography.The accuracy obtained with respect to the reference material was ± 2 μmol•kg-1 with a precision of ± 1.5 μmol•kg-1.
To quantify the concentrations of combined nitrite and nitrate (NO2-+ NO3-, hereafter, nitrate) present in the samples from the winter 2010 and 2013 cruises, a Skalar SAN Plus autoanalyzer was used.The reference material MOOS-2 was obtained from the National Resource Council Canada.The analytical precision was better than 5% for nitrite and nitrate combined.For the quantification of the summer 2015 cruise, samples were analyzed with an AA3-HR SEAL nutrient analyzer according to the GO-SHIP Repeat Hydrography Manual Japan) as reference materials (see description in Aoyama and Hydes, 2010).Precision is expressed as a coefficient of variation (CV) and was 0.2% for nitrate.
In order to explore possible relationships between water masses and their nitrate and DIC content, Tθ-S vs. nitrate for late fall-winter of 2010 and 2013, and summer of 2015, (respectively) were plotted and Tθ-S vs. DIC diagrams for late fall-winter and summer 2011, 2015 and 2016 (respectively) were also plotted.This allowed for a seasonal comparison.

5. Absolute Dynamic Topography (ADT) maps
Absolute Dynamic Topography (ADT) maps were generated to infer the seasonal influence of the CSW during the different cruises as Delgado et al. (2019) suggest.The images are products of the AVISO + database (Archiving, Validation, and Interpretation of Satellite Oceanographic data) available on the website https://www.aviso.altimetry.fr/en/data.The ADT maps only considered the time in which sampling was carried out for each cruise.In this work, present the surface dynamics based on these ADT maps, particularly from our winter (Feb-Mar) 2013 and summer (Aug-Sep) 2015 cruises (Fig. 10b and 10e).

6. Climatological data analysis
An analysis of the temperature and salinity data from the climatological database CARS 2009 (CSIRO Atlas of Regional Seas; http://www.marine.csiro.au/~dunn/cars2009)was performed to contrast climatological averages between in situ data from winter (February) Finally, in this work developed the new reclassification of the water masses based on the characteristic of the thermohaline and biogeochemical variables at densities lower than 26 kg•m-3 for each identified water mass.

Results
Potential temperature and salinity showed spatial and temporal variability at densities < 26 kg•m-3 during the five sampling campaigns included in this study (Fig. 1b).The four patterns that in this work considered relevant for the designation of new thermohaline ranges for water masses above the isopycnal of 26 kg•m-3, namely CSW, SUW, GCW, and the FISW, are described in the following section.It is noticeable that during the winter of 2013, when CSW was absent, the 24 kg•m-3 isopycnal and the 27 ºC isotherm were not observed (Fig. 5b, and 5g).In contrast, water with these characteristics was present during the summer when CSW entered the GoM through the LC (Fig. 5c,5d,5e,5h,5i,and 5j).Therefore, the summer characteristics of density and temperature represent the water of Caribbean origin.

2. Subsurface maximum DO and its association with GCW
In addition to the low density/high temperature waters typical of the CSW, in this work also noted the presence of a summer DO subsurface maximum.Figure 6  In this work found that during summer, AOU tends towards negative values (from 2 to -26.5 μmol•kg-1; see Supplementary Fig. 1c-e), above atmospheric equilibrium and supersaturated in waters above densities of  24 kg•m-3.In contrast, in late autumn and winter, AOU values in the GCW were positive at the same depths and ranged from 9 to 90 μmol•kg-1 due to the vertical transport of subsurface water (Supplementary Fig. 1a-b).This suggests that DO and AOU profiles can be used as criteria with which to separate the CSW from the GCW.

Description of water masses identification using the new classification.
To readjust the thermohaline ranges corresponding to CSW and GCW, oxygen was used as a tracer to separate these two water masses.It is important to note that the thermohaline ranges associated with the SUW were not modified because this water mass is only detected inside the LC.The thermohaline and chemical characteristics of each water mass are described in the following sections.

1. Subsurface Underwater (SUW).
Figure 7 shows the Tθ and S data above the isopycnal of 26 kg•m-3 as well as the new limits of salinity and temperature of surface waters (see Table 2).Figure 7a, shows typical oceanographic characteristics of water from the Caribbean, including the horseshoe structure present in Tθ-S diagrams that describe the SUW (Fig. 2 and 4).The principal thermohaline characteristic of the SUW is the presence of a salinity maximum ( 36.9)paired with a relative oxygen minimum ( 137 μmol•kg-1) located between 150 and 250 m (Fig. 7a).In this work found that SUW typically occurs in summer between 100 to 250 m and transports low oxygen water into the GoM (Table 2).above the 24 kg•m-3 isopycnal that includes the full range of thermohaline properties needs to be better defined.The T and S ranges in this work propose for this water mass are: temperatures between 27 and 32 ºC, salinities between 36 and 36.8, and a DO concentration range of 180 to 220 μmol•kg-1 (Table 2).The presence of CSW can be observed from relatively high salinities (up to  36.8)accompanied by relatively high surface temperatures of approx.30 ºC (Fig. 7a, 7b, and 7d).

Gulf Common Water (GCW)
The surface water between the 24 and 26 kg•m-3 isopycnals also needs to be defined by including the subsurface DO maximum as the upper limit of GCW.In this work propose new range limits for GCW to be temperatures between 20 to 27 ºC, salinities between 36.3 to 36.6 and DO between 112 to 232 μmol•kg-1 (Fig. 7c, Table 2).Brunt-Väisälä frequency analysis confirms the late fall data from 2010 and winters 2013 indicated vertical mixing in the first 200 m of the water column induced by season "Nortes" (not show figure).

4. Freshwater Influenced Surface Water (FISW)
The presence of the FISW was observed in summer.FISW was detected in the interior region of the CB and was distributed along the 25 ºN transects during 2010, 2011, 2015, and 2016 campaigns (Fig. 7d).This coincided with periods of high precipitation prior to and during the campaigns (https://smn.conagua.gob.mx/es/climatologia/temperaturas-y-lluvias/resumenesmensuales-de-temperaturas-y-lluvias).Based on the aforementioned thermohaline characteristics and the distribution of this water mass, the following limits were established: temperature between 24 to 30 ºC, salinity between 33 and 36, and DO concentrations between 180 and 220 μmol•kg-1 (Fig. 7d; Table 2).The input of freshwater resulted in a lowering of surface salinity in the first 20 m below approx.36 (Fig. 7d).The temperature range was from 24 ºC in late fall of 2010 to 30 ºC during summer (2011, 2015, and 2016).
As mentioned, the CSW was only detected during the summer oceanographic campaign.This water mass was characterized by low concentrations of nitrate from 0 to 0.48 μM in the first 90 m of the water column (Fig. 8c and 9f; Table 2).Similarly, DIC in this water mass was lower than 2090 μmol•kg-1 (Fig. 8d; Table 2).
The GCW contained relatively high concentrations of nitrate during late fall and winter, approx. 2 μM near 75 m.The highest concentrations of nitrate above 200 m,  8.4 μM was detected during this season, and it was observed within the lower limit of the GCW and the upper limit of the TACW (Fig. 8a; Table 2).In summer, the highest nitrate concentrations of  1.5 μM were found near 100 m, reaching values of approx.9.4 μM near the lower limit of the GCW at  210 m (Fig. 8c; Table 2).In the GCW, the vertical distribution of DIC mimicked the nitrate profiles.During late fall and winter, DIC concentrations higher than 2080 μmol•kg-1 were found below 50 m and reached maximum values of 2172 μmol•kg-1 near the bottom depth of this water mass (Fig. 8b; Table 2).
During summer at 50 m (σθ = 24.6 kg m-3), DIC values slightly lower than 2075 μmol•kg-1 were observed to increase with depth to  2169 μmol•kg-1 at  210 m (Fig. 8d).The deepening of the nutricline and carbocline observed during summer was associated with the transport of oligotrophic waters by CSW into the GoM, with low values of nitrate < 1 μM near the surface (Fig. 8c and 8d; Table 2).
Finally, the chemical composition of FISW depended to a large extent on the seasonality of precipitation, fluvial inputs, and mesoscale dynamics.Stations of low salinity and low nitrate concentrations ranging from 0.02 to 1.27 μM, and DIC ranging from 2005 to 2062 μmol•kg-1 in the first 50 m of the water column were sampled in winter (Fig. 8a, and 8b; Table 2).In contrast, during summer the concentrations of nitrate and DIC were slightly lower and ranged between 0.08 to 0.34 μM, and 1968 to 2053 μmol•kg-1, respectively (Fig. 8c, and 8d; Table 2).

Discussion.
A recent detailed analysis in the central and western GoM by Portela et al. (2018) of water masses from glider data, 14 cruises and Argo floats within the GoM, indicated the presence of seven water masses.While this is an improvement, there are still some problems in the classification and understanding of waters upwards of the 26 kg•m-3.In this work maintain that it is necessary to have a better understanding of how the GoM's water masses are formed to attain a classification that gives insight into 1) the dynamics of the water masses in the gulf, and 2) the physical mechanisms affecting biogeochemical processes, and 3) the resulting effects within biological processes.Upwards of the 26 kg•m-3 isopycnal, biogeochemical variables, such as oxygen, nitrate, and DIC concentrations exhibit large changes in concentration (≈ 200 μmol•kg-3, 0 and 9 μM, and 160 μmol•kg-3, respectively) that reflect the dynamic and variable characteristics of surface waters.These variations are caused by mixing and advection, processes that are important to be identified and understood.
For this reason, it was important to reclassify the shallower water masses of the GoM by including DO as a key tracer.

1. Reclassification of CSW and CGW using T, S and dissolved oxygen
In this work found a noticeable presence of CSW associated with the incursion of the LC during spring-summer as described by Delgado et al. (2019); this water mass was absent in late autumn and winter.Recently, the spring-summer incursion of the LC that transports CSW into the GoM has been confirmed, with a maximum presence in summer and a minimum in winter (Delgado et al., 2019).In this work emphasize that the extended "pulsing" by the LC and the Yucatán Current into the GoM explains the presence of CSW.
In this work attribute this absence of the CSW to the weakening of the LC.
In this work agree that the CSW increases its salt content above the 24 kg•m-3 isopycnal from about 36 at its entry into the GoM in the Yucatan Channel to about 36.8 due to LCE's and coastal upwelling (Wüst, 1964;Hernández-Guerra and Joyce, 2000;Carrillo et al., 2016).Also, evaporation likely contributes to the increase in salinity, caused by an increase in surface temperature during the summer when CSW is found within the GoM (Fig. 7a, and    7b).Previous studies have reported that the increasing stratification during the summer (mixed layer depth < 40 m) isolates the surface layer of the water column, which results in an increase in salinity due to intense evaporation (Zavala-Hidalgo et al., 2014).
Recently, Portela et al. (2018) redefined the T-S limits of the CSW within the GoM, renaming it a remnant of the Caribbean Surface Water (CSWra).They indicated that the distribution of "CSWra" is restricted to depths of 50 and 150 m.However, from the surface to 50 m they attributed to the influence of river discharge (Fig. 1a and 9a).In this work consider that the top 50 m should be included in an analysis that leads to the range of values used for the classification of this water mass.By not including the full range of salinity values, the actual volume of the CSW within the GoM would be underestimated, affect hydrography budgets and, potentially, estimates of productivity.Additionally, in the classification proposed by Portela et al. (2018) the overlap in the thermohaline ranges of the CSW and GCW was overlooked (see figure 2 of Portela et al., 2018).
In this work, solved the overlap problem based on the fact that the CSW is closely linked to the LC by the Yucatán Current input to the CB.In this work suggest that the overlap in the characteristics of the CSW and GCW that was not addressed by the Portela et al. (2018) classification can be addressed by considering the subsurface DO maximum.Our analysis revealed the existence of a subsurface DO maximum, which allowed us to separate the upper limit of the GCW from the bottom of the CSW.However, in this work suggest that the mechanism by which do behaves conservatively is as follows: during autumn-winter when the incursion of LC is minimal, the GCW is distributed at the surface.Intense winds are known as "Nortes" occur during this period and intense mixing takes place in surface waters of the GCW, resulting in the homogenization of all properties.The oxygen concentration measured during the winter months was approx.220 μmol•kg-1 (Fig. 6a-b and 7c).During spring-summer, the LC advects the warm, oligotrophic waters of the CSW into the interior of the GoM on top of the GCW.This water has a lower DO concentration than that found in the surface waters of GCW in winter, which is caused by temperature-related differences in solubility (Benson and Krause, 1984).The warm water of the CSW induces stratification that limits the exchange of oxygen with the underlying GCW (Fig. 6a-c, 7a, and 7c).The boundary between both water masses is therefore indicated by the maximum subsurface DO concentration (Fig. 6).In this work estimate that the DO concentration difference is approx.50 μmol•kg-1 (180 to 230 μmol•kg-1 see figure 6), and can this difference can be explained by differences in solubility, ruling out that the DO maximum is associated with photosynthesis.This is supported by a depth difference between the peak of maximum fluorescence and the maximum subsurface DO, maximum fluorescence occurs below of the subsurface DO maximum (Supplementary Fig. 3).Also, during summer cruises, the AOU in the CSW tends towards negative values (Supplementary Fig. 1c-e), these are usually found above densities of approx.24 kg•m-3 (Fig. 5), from a greater exchange with the atmosphere.In contrast, during late autumn and winter, the AOU presented positive values due to more respiration within the GCW (Supplementary Fig. 1a-b).The surface presence of the GCW in the autumn and winter is caused by: 1) the absence of CSW due to the retraction of the LC, and 2) the strong winds that result in a well-defined and deep (100 m) mixed layer.This last observation was previously pointed out by Nowlin and MCLellan (1967), Elliott (1979Elliott ( ,1982)), Vidal et al. (1994), andPortela et al. (2018).It has been suggested that the formation of the GCW originates from the erosion of the SUW (Vidal et al., 1992(Vidal et al., , 1994;;Portela et al., 2018).However, our results suggest GCW formation originates from the mixture of the remains of CSW and SUW within the GoM when the LC is retracted.During fall and winter, the remnant of these water masses in the interior of the gulf is mixed with TACW to form GCW.

1. 2. On the formation of GCW
During winter, when the CSW is absent, the TACW was also shallower than in summer.The proximity of the TACW to the GCW facilitates the vertical exchange of chemical properties towards the surface.Convective mixing leads to low DO concentrations of the TACW to be reflected in the GCW, as well as causing an observable increase in nitrate and DIC concentrations (Fig. 6).Furthermore, observations by satellite of the GoM found maximum concentrations of chlorophyll in winter (Pasqueron et al., 2017).This is in agreement with Damien et al. (2018), who found a winter chlorophyll concentration increase explained by the amount of nutrient injected into the euphotic layer by the dynamic of the winter mixedlayer.

1. 3. Freshwater Influenced Surface Water (FISW)
The presence of FISW reported in this study during the summers in the central region (24º-25ºN, 95.6º-88ºW) is likely due to river inflows, precipitation and offshore transport.In the central region of the GoM, relatively low salinities were measured that can only be explained by the contribution of freshwater from rivers or precipitation.For example, in the central stations located along 25 ºN, salinities of approximately 33.1 were detected in the first 20 m of the water column, which would lead to the formation of FISW (Fig. 1c).These freshwater inputs were also reported by Portela et al. (2018), who detected the influence of low salinity waters (33 g•kg-1) within the first 50 m in the central gulf.These low salinities have been attributed to the influence of freshwater inflow from rivers to the continental shelf in the northern GoM and transport to the central gulf by anticyclonic eddies; hence, low surface salinities can be found hundreds of kilometers from the river source (Morey et al., 2003a;Morey et al., 2003b;Jochens & DiMarco, 2008, Brokaw et al., 2019).
In the northern GoM, the Mississippi and Atchafalaya rivers flow into the GoM.Their outflow is generally transported westward along the Louisiana shelf during the summer months (Cochrane and Kelly, 1986;Ohlmann and Niiler, 2005;Smith and Jacobs, 2005) in response to predominant winds from the north and east (Wang et al., 1998).Besides, it has been reported that these rivers have their highest inflow during the spring/summer (Morey et al., 2003a).
In the southern GoM the Tonalá, Coatzacoalcos, and Usumacinta rivers flow into the region bordering CB.It has been reported that the propagation of low salinity filaments can be caused by local circulation resulting in a salinity gradient from coast to ocean (Vidal et al., 1994).In this work also observed the FISW as part of a salinity gradient of 35.4 to 36.3 that extended from the edge of the shelf toward the ocean, particularly during the winter.Also, a decrease in offshore salinity was attributed in the coastal region of the CB to freshwater input by Vidal et al. (1994); FISW was also detected at stations closer to the coastal region of the CB in the three summers oceanographic camping.It may also be noted that this type of water was observed in the semi-permanent cyclonic eddy reported by Nowlin (1972) and Pérez-Brunius et al. (2013), which could contribute to the transport of the FISW in the Campeche region during both summer and winter.
Concerning the biogeochemical role of the FISW in the surface waters within the GoM, the following questions remain: 1) what is its influence of the FISW in the first 20 m? and 2) what is its influence in the central GoM?
These questions highlight the need to carry out studies of biogeochemical processes at smaller scales to determine their role within the GoM.Undoubtedly, it is also important to carry out studies at the river mouths to determine the flow of nutrients and organic matter to the gulf.

1. The surface water masses modulate the depth of the nutricline
One of the biological implications of the presence of CSW is that it is oligotrophic reaching down to 90 m in spite.This can be seen in figure 10, wherein this work compares the vertical distribution of nitrate and density with ADT maps for summer 2015 (when mesoscale eddies were abundant) and winter 2013 (when the number and spatial extent of eddies were smaller).
During summer, a near-surface incursion of low-density water associated with the CSW was observed (white line Fig. 10a).This incursion brought water with oligotrophic characteristics to depths shallower than 70 m (nitrate from 0 to 0.48 μM; DIC  1978 μmol•kg-1, Fig. 8).
Nitrogen fixation process uses to be present on this oligotrophic surface North Atlantic Ocean waters (Montoya et al., 2002).The horizontal distribution of the concentration of nitrate and DIC was reduced by stratification following the entrance of the LC that transport the CSW into the interior of the gulf.In winter, the absence of the CSW is accompanied by a wellmixed density distribution in the first 200 m as the GCW predominates (Fig. 10d).Higher nitrate (0.02 to 13.7 μM) and carbon (> 2036 μmol•kg-1) concentrations were observed near the surface above depths 75 m.
Therefore, the alternating absence or presence of the CSW is related to the nutricline depth; in summer when CSW overlies the GCW, the nutricline is deepest (Fig. 10).In winter, when the GCW predominates and the TACW is shallower, deeper and well-defined, the nutricline is found closer to the surface.The importance of this redefinition of the water masses contributes to a better understanding of their role in the dynamics of nutrients (and carbon).
Finally, an analysis was carried out using the CARS2009 database (CSIRO Atlas of Regional Seas) in order to evaluate the temporal changes of the CSW and the GCW. Figure 11 contrasts climatological averages between winter (February) and summer (July).The Tθ-S diagram, as well as the vertical sections, show that CSW is only evident during the summer while during the winter only the GCW is detected from the surface to approximately 200 m deep.This supports our suggestion that the seasonal extension and retraction of the LC favors the formation of the subsurface maximum of DO during the summer and disappears in winter.and the CARS2009 climatological data sets affirm that the DO subsurface maximum can be used to define the upper limit of the GCW.During the summer months, with the entry of LC and dissipation by eddies, the presence of CSW dominates in the first 100 m, potentially having an impact on the primary productivity of the GoM, as indicated throughout this work and by other authors (Nowlin & McLellan, 1967;Tanahara, 2004;Schmitz, 2005;Delgado et al., 2019),

Conclusions.
A re-classification of the water masses above the 26 kg•m-3 isopycnal was carried out resulting in a modification of the present thermohaline ranges defining the CSW and GCW water masses.For the re-classification of the CSW and the GCW, DO concentrations were a key indicator of water mass limits.In addition, another water mass, the FISW, formed by the influence of the freshwater inputs, was included in the new classification.
CSW was detected only during the summer with a vertical spatial domain encompassing the first 90 m and featured warm waters, high salinities, non-detectable nitrate concentration, and negative values of the AOU.It was also found that the lower limit of this water mass is delimited by a maximum subsurface DO.The presence of this subsurface maximum was found only in the summer and separates the CSW from the GCW.Likewise, the presence and absence of CSW was found to modulate the depth of the nutricline and likely influences primary productivity.
In winter, the replacement of the CSW by the GCW affected the biogeochemical composition of surface water, specifically with an increase in nitrate concentrations, positive values of AOU and a decrease in surface temperatures.The TACW lies below the GCW and is closer https://doi.org/10.5194/bg-2019-340Preprint.Discussion started: 23 September 2019 c Author(s) 2019.CC BY 4.0 License.
https://doi.org/10.5194/bg-2019-340Preprint.Discussion started: 23 September 2019 c Author(s) 2019.CC BY 4.0 License.and summer (July).Diagrams and vertical sections reflecting 50 years of monthly July and February Tθ and S data were plotted to identify the seasonal presence or absence of CSW.
Tθ and σθ in presence or absence of CSW Vertical sections of seasonal changes in potential density and potential temperature occurring in the first  250 m (above 26 kg•m-3) of the study area are shown in figure 5.In general, the relatively low temperatures (T 24 ºC with ΔT < 5 ºC over densities < 26 kg•m-3; Fig. 5a, and 5b) indicate the absence of CSW in late autumn and winter and show a more mixed column in the first 100 m.Additionally, the density was, on average  24.5 kg•m-3 (with Δ σθ < 1 kg m-3; Fig. 5f and 5g).These characteristics are associated with the near-surface presence of GCW.During the summer, evidence of CSW with a temperature of  31ºC was https://doi.org/10.5194/bg-2019-340Preprint.Discussion started: 23 September 2019 c Author(s) 2019.CC BY 4.0 License.observed with ΔT ± 6 ºC (Fig.5h, 5i, and 5j).On the other hand, density fluctuated from σθ = 22 to 24 kg•m-3 (Fig.5c, 5d, and 5e).
displays transects of vertical sections of DO for the five cruises carried out during summer 2011, 2015, and 2016 a DO subsurface maximum of  210 to 232 μmol•kg-1 is shown to exist (Fig. 6c, 6d, 6e, 7b, and 7c).This pattern was observed consistently in the three summer cruises.The DO maximum was located between the isopycnals of 24 kg•m-3 and 25 kg•m-3 and can be considered a boundary between CSW and GCW.In contrast, with the absence of CSW during late fall (November 2010), the DO subsurface maximum was no longer clearly observable.During winter (February/March 2013), vertical mixing homogenized the DO in the first 200 m to concentrations of 200 to 220 μmol•kg-1 (Fig. 6b).

Figure 11
Figure11shows that during the presence of the CSW cause a deepening of the nutricline during the summer to  150 m in contrast to winter when the nutricline rises to100 m.

Figure 1 :
FIGURE CAPTIONS: Figure 1: (a) Distribution of the water masses using the classification system proposed by Portela et al. (2018) using conservative temperature (Θ) vs absolute salinity [SA g•kg-1], water masses as: Caribbean Surface Water remnant (CSWra), North Atlantic Subtropical Underwater (NASUW), Gulf Common Water (GCW), Tropical Atlantic Central Water (TACW), TACWna (nucleus), Antarctic Intermediate Water (AAIW) and North Atlantic Deep Water (NADW).(b) Θ-SA vs dissolved oxygen [DO, μmol•kg-1] diagram showing upwards of the isopycnal of the 26 kg•m-3 using the Portela et al. (2018) classification.The data from the five cruises from 2010 to 2016 were used to generate the Θ-S diagrams.(c) The coverage area for the stations analyzed (transect delimited in black lines) in the GoM from 2010 to 2016.

Figure 3 :Figure 4 :
Figure 3: A comparison of winter (a) and summer (b) conditions of the variability of the depth of 26 kg•m-3 density field in the GoM (in situ hydrographic data collected in February/March 2013 and August/September 2015, respectively)

Figure 5 :
Figure 5: The vertical distribution [250 m] of potential density [kg•m-3] and potential temperature [ºC] are shown for the late fall of 2010 (a and f), winter of 2013 (b and g) and summers of 2011 (c and h), 2015 (d and i), and 2016 (e and j).The location of the transect is shown in figure 1c.

Figure 6 :
Figure 6: The vertical distribution [250 m] of dissolved oxygen [μmol•kg-1] are shown for the late fall of 2010 (a), winter of 2103 (b) and summers of 2011 (c), 2015 (d), and 2016 (e).The white contours indicate the lower limit of CSW [24 kg•m-3] and GCW [26 kg•m-3] in all sections.The location of the transect is shown in figure 1c.

Figure 7 :
Figure 7: This figure shows the new classification of the water masses with the adjustments to the thermohaline range limits based on the distribution that the four patterns in Figure 2c.(a) The Tθ-S vs DO [μmol•kg-1] diagram shows the profiles with SUW characteristics.(b) Tθ-S vs DO [μmol•kg-1] diagram presents characteristics particular with SCW (c) Tθ-S vs DO [μmol•kg-1] diagram associated with GCW.(d) Tθ-S vs DO [μmol•kg-1] diagram associated to the water mass called Freshwater Influenced Surface Water (FISW).

Figure 9 :
Figure 9: Comparison of the classification system proposed by (a) Portela et al. (2018) and (b) this study.Shows the Θ-SA vs. DO [μmol•kg-1] diagram showing upwards of the isopycnal of the 26 kg•m-3 using the reclassification proposed in this work.The names of the water masses used in this work are: Caribbean Surface Water (CSW), Subtropical Underwater (SUW), Gulf Common Water (GCW), and the Freshwater Influenced Surface Water (FISW).

Figure 10 :
Figure 10: The vertical distribution [250 m] of potential density [kg•m-3] is shown for the summer of 2015 (a) and winter of 2013 (d).The white contours indicate the lower limit of CSW [24 kg•m-3; (a)] and of GCW [26 kg•m-3; (d)] in both sections.The ADT maps show the trajectory of the summer (b) and winter (e) sections of each cruise.The nitrate profiles [μM] (c=summer; f=winter) only include the stations that are found within the trajectory traced in the ADT maps for each cruise.The blue color points indicate the stations that are found outside of the areas influenced by anticyclonic rings while the red color points denote stations located in the area of influence of the anticyclonic gyres.

Figure 11 :
Figure 11: Θ-SA vs. DO [μmol•kg-1] annual diagrams from February (a) and July (b) showing the reclassification proposed in this work.Data derived from the CARS-2009 database.Annual vertical sections (-95.5 to -86.5 ºW, 25 ºN; the section shown in figure 1c from the station C to D) of oxygen [μmol•kg-1] concentration and nitrate [μM] for February (c, and d), and July (e, and f) derived from the CARS-2009 database

Figure 10 .
Figure 10.The vertical distribution [250 m] of potential density [kg•m -3 ] is shown for the summer of 2015 (a) and winter of 2013 (d).The white contours indicate the lower limit of CSW [24 kg•m -3 ; (a)] and of GCW [26 kg•m -3 ; (d)] in both sections.The ADT maps show the trajectory of the summer (b) and winter (e) sections of each cruise.The nitrate profiles [μM] (c=summer; f=winter) only include the stations that are found within the trajectory traced in the ADT maps for each cruise.The blue color points indicate the stations that are found outside of the areas influenced by anticyclonic rings while the red color points denote stations located in the area of influence of the anticyclonic gyres.