Introduction

Over the past 2 centuries human activities have increased the concentration of CO₂ in the atmosphere by ~400 petagrams (1 Pg = 10¹⁵ g), that is, by ~40% (^{Zeebe 2012}). Atmospheric CO₂ content has changed from 280 μatm before the industrial revolution to 407 μatm in recent times (^{NOAA 2016}), and could reach 850 μatm in 2100 (^{Hoegh-Guldberg et al. 2014}). Oceans absorb about 2 Pg C yr^-1 of the CO₂ emitted through human activities (^{Takahashi et al. 2009}), but the absorption rate is not high enough to prevent the increase of CO₂ in the atmosphere. Since CO₂ is a greenhouse gas, global mean temperature has risen by 0.85 °C and an increase of ~4 °C relative to the global mean estimated from 1980 to 1999 has been reported for 2100 (^{IPCC 2014}).

On average, oceans act as a net CO₂ sink but the direction and magnitude of the exchange of CO₂ with the atmosphere varies considerably over space and time. ^{Hales et al. (2012)} characterized the region of the Pacific Ocean between 22°N and 50°N (30 to 370 km offshore) as a source of atmospheric CO₂ of 0.60 mol C m^-2 yr^-1, a value 3 times higher than that estimated by ^{Chavez et al. (2007)} for the same area. ^{De La Cruz-Orozco et al. (2010)} studied air-sea CO₂ exchange off the Baja California Peninsula from October 2004 to October 2005, and on average it acted as a source of CO2, with an annual value of 0.40 mol m^-2. The average annual CO₂ partial pressure (pCO₂) data collected from 1993 to 2001 by ships of opportunity off Baja California revealed that the area is a source of CO₂, but that it acts as a sink in winter and spring and is a source in summer and autumn (^{Hernández-Ayón et al. 2010}).

In the California Current, corrosive water (pH less than 7.80) has been detected near the surface, a condition that was not expected to occur for some 50 years (^{Feely et al. 2008}, ^{Alin et al. 2012}). Though some of the corrosive characteristics of these waters are the result of respiration processes in subsurface waters, the accumulation of atmospheric CO₂ continues, modifying the natural dynamics of the CO₂ cycle in coastal upwelling zones (^{Feely et al. 2008}).

Large-scale events such as El Niño/Southern Oscillation (ENSO) affect the air-sea flux of carbon (FCO₂). During the 1997-1998 El Niño event, in the central California upwelling system (to the north of our study area), the FCO₂ fluctuated between -0.30 and -0.70 mol C m^-2 yr^-1, with inputs of 1.50 to 2.20 mol C m^-2 yr^-1 under 1999 La Niña conditions (^{Friederich et al. 2002}). Off the tropical Pacific coast of Mexico (to the south of our study area), the FCO₂ ranged from -0.40 to 2.50 mmol C m^-2 d^-1 during the 2009 El Niño event and from -4.40 to 3.30 mmol C m^-2 d^-1 during the 2010 La Niña event (^{Franco et al. 2014}). In the southern region of the California Current off the Baja California Peninsula, our area of interest, we would expect fluxes from the ocean to the atmosphere under El Niño conditions and from the atmosphere to the ocean under La Niña conditions.

Analysis of high-frequency (3-h) pCO₂ data obtained by sensors on an autonomous buoy moored off the northern coast of Baja California revealed that this coastal zone is a net source of CO₂ (~1.00 mol C mm^-2 yr^-1) to the atmosphere (^{Reimer et al. 2013b}, ^{Muñoz-Anderson et al. 2015}). ^{Reimer et al. (2013b)} concluded that 91% of the positive FCO₂ (source) was contributed during coastal upwelling events, and ^{Muñoz-Anderson et al. (2015)} found that the system acts as a slight sink (0.06 mol C m^-2 yr^-1) during non-upwelling periods.

Few studies have been published on the seasonal variability of air-sea CO₂ exchange in the southern region of the California Current off Baja California. In this paper we present an analysis of pCO₂ and FCO₂ calculated with in situ data from 22 oceanographic cruises. This database allows us to affirm that the results are robust for the study area and contribute significantly to the knowledge of the spatial and seasonal variability of pCO₂ and FCO₂ in this region. The data also indicate that the temporal variations and intrinsic controls are subject to dynamic spatial variations and should be examined individually in each spatial domain.

Materials and methods

The Pacific Ocean off the Baja California Peninsula presents high spatial and temporal variability caused mainly by the physical dynamics derived from changes in the intensity of flow of the California Current (^{Durazo and Baumgartner 2002}); coastal upwelling events (^{Pérez-Brunius et al. 2007}); the presence of fronts, meanders, and eddies (^{Barocio-León et al. 2007}); and seasonal and interannual variability (^{Aguirre-Hernández et al. 2004}, ^{Gaxiola-Castro et al. 2008}). Because of its seasonal variability, the study area comprising the southern region of the California Current has been divided into northern and southern portions at latitude 28°N (^{Durazo 2015}).

In the southern region of the California Current, the IMECOCAL (Spanish acronym for Mexican Research of the California Current) program undertakes oceanographic cruises 4 times a year (winter, spring, summer, and autumn) off the Baja California Peninsula, between ~30 and ~220 km offshore. The sampling grid consists of 90 stations, separated by ~37 km, arranged in lines perpendicular to the coast, with a distance of ~70 km between hydrographic lines (Fig. 1a). In this paper we analyze the hydrographic and surface pCO₂ data obtained during 22 cruises carried out between 2004 and 2011 (Fig. 1b).

Figure 1 (a) Zonification of the study area and sampling sites of the IMECOCAL program. Clear circles and squares indicate the stations in the northern oceanic zone and southern oceanic zone, respectively. Black circles and squares indicate the stations in the northern coastal zone and southern coastal zone, respectively. (b) Distribution of the number of hydrographic stations occupied during the IMECOCAL cruises conducted from 2004 to 2011. 

During the oceanographic cruises, casts were made with a Sea-Bird Electronics 911 plus CTD (calibrated by the manufacturer), coupled to a rosette fitted with 5-L Niskin bottles. The CTD contained sensors to measure temperature, salinity, and dissolved oxygen. Water samples were taken with the Niskin bottles at discrete depths (0, 10, 20, 50, 100, 150, and 200 m) to analyze dissolved oxygen and phyto-plankton chlorophyll a.

The samples for the dissolved oxygen analysis were collected in 125-mL glass bottles and were analyzed on board by the micro-Winkler method (^{Helm et al. 2009}). The linear correlation coefficient between the discrete dissolved oxygen values obtained with the micro-Winkler method and those derived by the CTD was 0.96 (n = 346, P < 0.05). Based on the CTD oxygen data, the percentage of oxygen saturation (%OS) was estimated with the ^{Weiss (1970)} equations.

The water aliquots for the chlorophyll a analysis were collected in 1-L dark plastic bottles, and the concentration was determined using Turner Designs 10-AU-05 and Trilogy (model 7200-000) fluorometers by the method proposed by ^{Holm-Hansen et al. (1965)} and modified by ^{Venrick and Hayward (1984)}. Further details of the collection methods and analysis used by the IMECOCAL program can be found at http://imecocal.cicese.mx.

Sea surface pCO₂ (pCO_2sea) was analyzed using an automated continuous flow LI-COR 6262 system that measures the mole fraction of CO₂ in dry air (xCO₂). Seawater was pumped from an intake in the ship's hull (below the water-line, ~2.50 m depth). It flowed to a thermo-salinometer and then to the LI-COR 6262 analyzer, both located in the dry laboratory. The accuracy of the LI-COR 6262 analyzer for the pCO_2sea measurements is ±1 |iatm and it autocalibrated every 2 h against commercial standard CO₂ obtained from the National Institute of Standards and Technology and the National Oceanic Atmospheric Administration Climate Monitoring Diagnostic Laboratory (^{Friederich et al. 2002}).

Further details of how the pCO₂ data were obtained and processed can be found in ^{Reimer et al. (2013a)}. Since the pCO_2sea data were obtained continuously during navegation, it was necessary to obtain an average during the time (~40 min) that the boat remained at each sampling station in order to associate them with the discrete temperature, salinity, %OS, and chlorophyll a data.

The sea-air CO₂ exchange was calculated using the following equation: FCO₂ = K_w x K₀ x (ΔpCO₂), where K_w is the CO₂ transfer coefficient as a function of wind speed (^{Wanninkhof 2014}), K₀ is the CO₂ solubility coefficient as a function of temperature and salinity (^{Weiss 1974}), and ΔpCO₂ is the difference between the pCO₂ in seawater (pCO_2sea) and in the atmosphere (pCO_2air). By convention, negative FCO₂ values indicate fluxes from the atmosphere to the ocean and positive values indicate fluxes from the ocean to the atmosphere. Atmospheric CO₂ is well mixed and its partial pressure varies within a very narrow range relative to its mean value (^{Sarmiento and Gruber 2006}); hence, the daily pCO_2air data were obtained from http://scrippsco2.ucsd.edu/data/atmospheric_co2. These data correspond to measurements taken at Mauna Loa Observatory, Hawaii, from 2004 to 2011. Based on this information, pCO_2air averages were obtained for the period close to each oceanographic campaign and used to calculate an average pCO_2air value for each of the 22 cruises analyzed. Monthly ocean surface wind speed data (spatial resolution of 0.25° x 0.25°) were obtained from http://podaac.jpl.nasa.gov/DATACATALOG/ccmpinfo.html; they are the product of cross calibration of several multiplatforms (^{Atlas et al. 2011}).

The seasonal climatology of temperature, salinity, chlorophyll, %OS, pCO₂, and FCO₂ was calculated for the period under study (5 winters, 6 springs, 5 summers, and 6 autumns). Since the sampling grid was not always entirely covered, the averages were calculated with the number of data available for each cruise using an arithmetic mean. To minimize measurement errors, the values that fell outside the range of 3 standard deviations were eliminated. To determine the response of pCO₂ and FCO₂ to the ENSO cycle in the region, we analyzed the effects during its warm phase (El Niño) in autumn 2004 and during its cold phase (La Niña) in spring 2011, both events identified according to the Multivariate ENSO Index (http://www.cpc.ncep.noaa.gov/products/analysismonitoring/ensostuff/ensoyears.shtml).

Results

The seasonal climatology (2004-2011) of sea surface temperature showed a latitudinal gradient, with higher values in the southern portion (below line 120) and lower values in the northern portion (above line 120) of the IMECOCAL grid (Fig. 2). Surface temperature showed 2 different conditions: lower values in winter (16.50 °C; Fig. 2a) and spring (16.30 °C; Fig. 2b), and higher values in summer (19.90 °C; Fig. 2c) and autumn (20.20 °C; Fig. 2d). A longitudinal pattern was persistent in the northern portion, with lower values close to shore and higher values in the open sea. A latitudinal pattern predominated in the southern portion, with isotherms perpendicular to the coast, particularly in summer and autumn (Fig. 2c, d).

Figure 2 Seasonal climatology (2004-2011) of sea surface temperature (SST, °C) in (a) winter, (b) spring, (c) summer, and (d) autumn. 

The seasonal climatology of sea surface salinity showed a latitudinal gradient, with isohalines perpendicular to the coast (Fig. 3a, c). A longitudinal pattern was observed only in spring in the northern coastal zone (NCZ), from Punta Eugenia to Ensenada (Fig. 3b), with isohalines parallel to the coast and extending ~50 km offshore. The mean salinity values were higher in autumn (33.65; Fig. 3d) and winter (33.62; Fig. 3a), and lower in spring (33.53; Fig. 3b) and summer (33.57; Fig. 3c).

Figure 3 Seasonal climatology (2004-2011) of sea surface salinity in (a) winter, (b) spring, (c) summer, and (d) autumn. 

Chlorophyll and %OS showed an inshore-offshore gradient, with higher values close to shore and lower values towards the open sea. Mean chlorophyll concentration was higher in spring (0.88 mg m^-3; Fig. 4b) and the isolines of 0.50 mg m^-3 extended ~100 km offshore. A plume of chlorophyll values exceeding 0.50 mg m^-3 that extends from Punta Eugenia to ~140 km offshore (Fig. 4a, c) can be observed in all the seasons. The relative chlorophyll maximum values occurred in the NCZ. During winter and spring, chlorophyll concentrations ranged from 0.50 to 1.50 mg m^-3 in the southern coastal zone (SCZ). The longitudinal pattern of %0S distribution was more pronounced in summer (Fig. 5c), with isolines of 100%0S parallel to the peninsula between the coast and ~100 km offshore, particularly in the NCZ. Saturation values below 100% predominated in winter (97%0S; Fig. 5a), indicating a predominance of the thermodynamic effect. In spring (Fig. 5b), the area off Ensenada had saturation values above 110%, with isolines parallel to the peninsula between the coast and ~100 km offshore. In summer, the >100%OS conditions persisted off Ensenada, but the isolines parallel to the peninsula covered an area between the coast and ~30 km offshore.

Figure 4 Seasonal climatology (2004-2011) of sea surface chlorophyll a (mg m ³⁾ in (a) winter, (b) spring, (c) summer, and (d) autumn. 

Figure 5 Seasonal climatology (2004-2011) of sea surface dissolved oxygen saturation (%0S) in (a) winter, (b) spring, (c) summer, and (d) autumn. 

In general, the seasonal climatology of pCO_2sea showed a latitudinal pattern, with higher values in the southern portion and lower in the northern portion of the study area (Fig. 6); these values are associated with spatial changes in sea surface temperature. The 380 μatm isoline perpendicular to the coast was situated at ~29°N in autumn (Fig. 6d) and moved to ~27°N in winter (Fig. 6a). In spring (Fig. 6b), the coastal area to the south of Ensenada stands out, with pCO_2sea values below 350 μatm and isolines parallel to the peninsula between the coast and ~80 km offshore. In summer (Fig. 6c), the area covered by the isoline with values below 350 μatm decreased (from the coast to ~40 km offshore), and values higher than 400 μatm predominated in the oceanic fringe at 29°N and to the south of the study area. The coastal zone to the north of Punta Eugenia consistently had low pCO₂ values (372 μatm), indicating that it acted as a possible CO₂ sink during the seasonal cycle. The pCO₂ climatology revealed a temporal pattern similar to that of temperature: lower values in winter (378 μatm; Fig. 6a) and spring (388 μatm; Fig. 6b) and higher in summer (410 μatm; Fig. 6c) and autumn (395 μatm; Fig. 6d).

Figure 6 Seasonal climatology (2004-2011) of sea surface partial pressure of CO₂ (pCO₂, μatm) in (a) winter, (b) spring, (c) summer, and (d) autumn. 

Air-sea FCO₂ showed a latitudinal distribution pattern, with higher values in the southern portion and lower in the northern portion (Fig. 7), similar to the distribution of pCO₂ (Fig. 6). In winter (Fig. 7a), the coverage of the area with negative values extended to ~26°N in the northern oceanic zone (N0Z) and to 28°N in the NCZ. Further south, the values were predominantly positive, on average 0.85 mmol m^-2 d^-1. In spring (Fig. 7b), the NCZ stands out, with FCO₂ values above -6.00 mmol m^-2 d^-1 and isolines parallel to the coast and extending ~80 km offshore. An area with positive FCO₂ values (1.25 mmol m^-2 d^-1) extended from ~28.5°N in the N0Z towards Punta Eugenia and covered practically all the southern portion of the study area. In summer (Fig. 7c), the coverage of the negative flows decreased in the NCZ (extending ~40 km offshore) and the positive flows predominated. A fringe with positive FCO₂ values between 1.20 and 1.80 mmol m^-2 d^-1 occurred in the NOZ (31°N) and extended southwards covering the southern oceanic zone (S0Z) and the SCZ, with values reaching 6.00 mmol m^-2 d^-1. In autumn (Fig. 7d), the area to the north of 28°N showed negative fluxes of up to -3.00 mmol m^-2 d^-1 (Fig. 7d), but to the south of this latitude, the values were positive, of up to 3.50 mmol m^-2 d^-1. The NCZ consistently had negative FCO₂ values, indicating that it acts as a CO₂ sink during the seasonal cycle. The FCO₂ climatology revealed a similar pattern to that of pCO₂: negative fluxes in winter (-0.26 mmol m^-2 d^-1; Fig. 7a) and spring (-0.37 mmol m^-2 d^-1; Fig. 7b) and positive in summer (2.33 mmol m^-2 d^-1; Fig. 7c) and autumn (0.92 mmol m^-2 d^-1; Fig. 7d).

Figure 7 Seasonal climatology (2004-2011) of air-sea CO₂ flux (FCO₂, mmol m ² d ¹⁾ in (a) winter, (b) spring, (c) summer, and (d) autumn. 

In the 2004-2011 time series for the northern portion of the study area, mean sea surface temperature ranged from 14.5 to 20.0 °C, the lowest values occurring in winter and spring and the highest in summer and autumn (Fig. 8a). A decreasing trend (by 0.10 °C) occurred after the summer of 2007. Chlorophyll concentration was higher in spring. There was a period of low chlorophyll concentration (0.42 mg m^-3) between 2004 and mid-2007 (Fig. 8b), after which it increased. The average pCO₂ tended to decrease after 2007, with values below the atmospheric value as of 2009 (Fig. 8c). This same tendency was observed for FCO₂, with negative values as of 2009 (Fig. 8d). In the spring of 2011 the northern portion had the maximum CO₂ sink value of all the time series (-6.00 mmol C m^-2 d^-1).

Figure 8 Time series (2004-2011) of average (a) temperature (°C), (b) chlorophyll (Chl, mg m ³⁾, (c) partial pressure of CO₂ (pCO₂, μatm), and (d) air-sea CO₂ flux (FCO₂, mmol m^-2 d^-1) in the northern portion of the study area (above IMECOCAL line 120). The bars indicate the standard error. The continuous line in (c) indicates the atmospheric pCO₂ values. The dashed line in (d) indicates the equilibrium between the atmosphere and ocean. 

In the southern portion of the study area, mean sea surface temperature ranged from 15.80 to 23.50 °C, the lowest values occurring in winter and spring and the highest in summer and autumn (Fig. 9a). Mean chlorophyll concentration (0.48 mg m^-3) was slightly lower than in the northern portion (0.62 mg m^-3) and showed an increasing trend from the summer of 2006 (Fig. 9b) to the end of the study period. In general, pCO₂ was higher than the atmospheric value (Fig. 9c), except in 2011, when it was lower, and a decreasing trend occurred from 2007 until the end of the study period. The southern portion acted mainly as a source of CO₂ to the atmosphere, with an annual average of 1.56 mmol m^-2 d^-1 (Fig. 9d) and values close to equilibrium in winter, except in 2011 when it acted as a sink.

Figure 9 Time series (2004-2011) of average (a) temperature (°C), (b) chlorophyll (Chl, mg m ³⁾, (c) partial pressure of CO₂ (pCO₂, μatm), and (d) air-sea CO₂ flux (FCO₂, mmol m^-2 d^-1) in the southern portion of the study area (IMECOCAL line 120 and below). The bars indicate the standard error. The continuous line in (c) indicates the atmospheric pCO₂ values. The dashed line in (d) indicates the equilibrium between the atmosphere and ocean. 

During the study period (2004-2011), the southern region of the California Current was under the influence of El Niño (2004, 2007, 2009-2010) and La Niña (2008, 2009, 2011) events (^{Bjorkstedt et al. 2012}) (http://www.cpc.ncep.noaa.gov/products/analysis_monitoring/ensostuff/ensoyears.shtml). For all the region, the average pCO₂ during the 2004 El Niño (Fig. 10a) was 44 μatm higher than during the 2011 La Niña (Fig. 10b). During the 2004 El Niño, the spatial distribution of pCO₂ showed a latitudinal pattern, with higher values (414 μatm) in the southern portion and lower (399 μatm) in the northern portion; in the latter, a longitudinal pattern was observed, with lower pCO₂ values (394 μatm) in the coastal zone. During the 2011 La Niña, the average pCO₂ was 360 μatm, lower than the value (393 μatm) obtained for pCO_2air. During the 2004 El Niño, the ocean off Baja California acted as a source of CO₂ to the atmosphere (on average, 2.00 mmol m^-2 d^-1; Fig. 10c), whereas during the 2011 La Niña it acted as a sink (on average, 5.30 mmol m^-2 d^-1; Fig. 10d). Note the marked latitudinal gradient in pCO₂ and FCO₂ under La Niña conditions, with low (high) values of 355 μatm (-6.40 mmol m^-2 d^-1) in the northern portion and high (low) values of 366 μatm (-4.00 mmol m^-2 d^-1) in the southern portion.

Figure 10 Spatial distribution of (a, b) the partial pressure of CO₂ (pCO₂, μatm) and (c, d) air-sea CO₂ flux (FCO₂, mmol m ² d^-1) under El Niño conditions (autumn 2004; a, c) and La Niña conditions (spring 2011; b, d). 

Discussion