An analysis of the variability in δ13C in macroalgae from the Gulf of California: indicative of carbon concentration mechanisms and isotope discrimination during carbon assimilation

The isotopic composition of carbon in macroalgae (δ13C) is highly variable, and its prediction is complex concerning terrestrial plants. The determinants of δ13C macroalgal variations were analyzed in a large stock of specimens that vary in taxa and morphology and were collected in shallow marine habitats in the Gulf of California (GC) with distinctive environmental conditions. A large δ13C variability (−34.6 ‰ to −2.2 ‰) was observed. Life-forms (taxonomy 57 %, morphology and structural organization 34 %) explain the variability related to carbon use physiology. Environmental conditions influenced the δ13C macroalgal values but did not change the physiology, which is most likely inherently species-specific. Values of δ13C were used as indicators of the presence or absence of carbon concentrating mechanisms (CCMs) and as integrative values of the isotope discrimination during carbon assimilation in the life cycle macroalgae. Based on δ13C signals, macroalgae were classified in three strategies relative to the capacity of CCM: (1) HCO−3 uptake (δ13C>−10 ‰), (2) using a mix of CO2 and HCO−3 uptake (−10< δ13C>−30 ‰), and (3) CO2 diffusive entry (δ13C<−30 ‰). Most species showed a δ13C that indicates a CCM using a mix of CO2 and HCO−3 uptake. HCO − 3 uptake is also widespread among GC macroalgae, with many Ochrophyta species. Few species belonging to Rhodophyta relied on CO2 diffusive entry exclusively, while calcifying macroalgae species using HCO−3 included only Amphiroa and Jania. The isotopic signature evidenced the activity of CCM, but it was inconclusive about the preferential uptake of HCO−3 and CO2 in photosynthesis and the CCM type expressed in macroalgae. In the study of carbon use strategies, diverse, species-specific, and complementary techniques to the isotopic tools are required.

Abstract. The isotopic composition of carbon in macroalgae (δ 13 C) is highly variable, and its prediction is complex concerning terrestrial plants. The determinants of δ 13 C macroalgal variations were analyzed in a large stock of specimens that vary in taxa and morphology and were collected in shallow marine habitats in the Gulf of California (GC) with distinctive environmental conditions. A large δ 13 C variability (−34.6 ‰ to −2.2 ‰) was observed. Life-forms (taxonomy 57 %, morphology and structural organization 34 %) explain the variability related to carbon use physiology. Environmental conditions influenced the δ 13 C macroalgal values but did not change the physiology, which is most likely inherently species-specific. Values of δ 13 C were used as indicators of the presence or absence of carbon concentrating mechanisms (CCMs) and as integrative values of the isotope discrimination during carbon assimilation in the life cycle macroalgae. Based on δ 13 C signals, macroalgae were classified in three strategies relative to the capacity of CCM: (1) HCO − 3 uptake (δ 13 C > −10 ‰), (2) using a mix of CO 2 and HCO − 3 uptake (−10 < δ 13 C > −30 ‰), and (3) CO 2 diffusive entry (δ 13 C < −30 ‰). Most species showed a δ 13 C that indicates a CCM using a mix of CO 2 and HCO − 3 uptake. HCO − 3 uptake is also widespread among GC macroalgae, with many Ochrophyta species. Few species belonging to Rhodophyta relied on CO 2 diffusive entry exclusively, while calcifying macroalgae species using HCO − 3 included only Amphiroa and Jania. The isotopic signature evidenced the activity of CCM, but it was inconclusive about the preferential uptake of HCO − 3 and CO 2 in photosynthesis and the CCM type expressed in macroalgae. In the study of carbon use strategies, diverse, species-specific, and complementary techniques to the isotopic tools are required.
In marine environments, where pH ∼ 8.1±1, the diffusion rate of CO 2 in seawater is low. Thus, HCO − 3 accounts for 98 % of the total dissolved inorganic carbon (DIC), resulting in a high HCO − 3 : CO 2 ratio (150 : 1) (Sand-Jensen and Gordon, 1984). Low CO 2 concentrations in seawater, which limit macroalgae growth, are compensated for by carbon concentrating mechanisms (CCMs) that increase the internal inorganic carbon concentration near the site of RuBisCo activity . Therefore, the absorption of HCO − 3 by most macroalgae is the primary source of inorganic carbon for photosynthesis, but some species depend exclusively on the use of dissolved CO 2 that enters cells by diffusion (Beardall and Giordano, 2002;Giordano et al., 2005;Maberly et al., 1992;Raven et al., 2002a, b). Hence, macroalgal species with productivity limited by lacking CCMs (having low plasticity for inorganic carbon uptake) seems to be restricted to subtidal habitats and composed mainly of red macroalgae (but without a morphological patron apparent) Kübler and Dudgeon, 2015). The rest of the macroalgae with CCM occupies the intertidal to the deep subtidal zone.
The δ 13 C macroalgal values indicate the carbon source used (CO 2 or HCO − 3 ) in photosynthesis and allow the presence or absence of CCMs to be inferred Maberly et al., 1992;Raven et al., 2002a). However, the isotopic signature may be inconclusive for determining the carbon source's preference . Also, the δ 13 C signal in the algal thallus can be used to indicate the physiological state of photosynthetic metabolism (Kim et al., 2014;Kübler and Dungeon, 2015). For example, δ 13 C variability depends, in part, on the life-forms' taxonomy, morphology, and structural organization (Lovelock et al., 2020;Marconi et al., 2011;Mercado et al., 2009;. δ 13 C is also modulated by the interaction with environmental conditions (e.g., light, DIC, and nutrients) (Carvalho et al., 2010a, b;Cornelisen et al., 2007;Dudley et al., 2010;Mackey et al., 2015;Rautenberger et al., 2015;. In this study, our objective was to investigate the contributions of life-forms, the changes in the habitat features, and environmental conditions to the δ 13 C macroalgal variability in communities in the Gulf of California (GC). We collected a large stock of macroalgae specimens of a diversity of species characterized by various morphological and physiological properties to reach our objective. Besides high diversity, in terms of life-forms, we selected various shallow marine habitats along a latitudinal gradient in the GC or the sample collection, characterized by unique and changing environmental factors. The GC features abundant and diverse macroalgae populations, acclimated and adapted to diverse habitats with environmental conditions determining the light, DIC, and nutrient availability. The δ 13 C signal from the thallus of macroalgae was used as indicative of the presence or absence of CCMs and as integrative values of the isotope discrimination during carbon assimilation and respiration along the life cycle of macroalgae in macroalgal communities in the GC as a function of taxa and environmental factors (Díaz-Pulido et al., 2016;Hepburn et al., 2011;Maberly et al., 1992;Raven et al., 2002a). Because the GC is a subtropical zone with high irradiance and specimens were collected in the intertidal and shallow subtidal zone, we expect to find a high proportion of species with active uptake HCO − 3 (δ 13 C > −10 ‰). A third objective was to explore any geographical pattern in the δ 13 C macroalgae along and between the GC bioregions. Previous studies have indicated changes in the δ 13 C signal with latitude, mainly related to the light and temperature (Hofmann and Heesch, 2018;Lovelock et al., 2020;Marconi et al., 2011;Mercado et al., 2009;Stepien, 2015). Macroalgae as biomonitors constitute an efficient tool in monitoring programs in large geographical regions (Balata et al., 2011) and for environmental impact assessments (Ochoa-Izaguirre and Soto-Jiménez, 2015).
The Gulf of California is different in the north and the south, related to a wide range of physicochemical factors. The surface currents seasonally change direction and flow to the southeast with maximum intensity during the winter and to the northwest in summer (Roden, 1958). The northern part is very shallow (< 200 m deep on average), divided into the upper gulf, northern gulf, and Midriff Islands regions (Roden, 1958;Roden and Groves, 1959). The surrounding deserts largely influence this region (Norris, 2010), which shows marked seasonal changes in coastal surface seawater temperatures Martínez-Díaz-de-León et al., 2006). Tidal currents induce a significant cyclonic circulation through June to September and anticyclonic from November to April (Bray, 1988;Carrillo and Palacios-Hernández, 2002;Martínez-Díaz-de-León, 2001;Velasco-Fuentes and Marinone, 1999). The southern part consists of a series of basins whose depths increase southwards (Fig. 1). The intertidal macroalgae in the southern region are subject to desiccation, mostly during summer. The water column's physicochemical characteristics are highly influenced by the contrasting climatic seasons in the GC: the dry season (nominally from November to May) and the rainy season (from June to October). Annual precipitation (1080 mm yr −1 ) and evaporation (56 mm yr −1 ) rates registered during the past 40 years were 881 ± 365 mm yr −1 and 53 ± 7 mm yr −1 , respectively (CNA, 2012).
In the GC around 669 macroalgae species exist, including 116 endemic species (Espinoza-Avalos, 1993;Norris, 1975;Pedroche and Sentíes, 2003). Many endemic species currently have a wide distribution along the Pacific Ocean coast but with GC origin (Aguilar-Rosas et al., 2014;Dreckman, 2002). Based on oceanographic characteristics (Ro-den and Groves, 1959) and in the endemic species distribution (Aguilar-Rosas and Aguilar-Rosas, 1993;Avalos, 1993), the GC can be classified into three phycofloristic zones: (1) the first zone located from the imaginary line connecting San Francisquito Bay, B.C. (Baja California), to Guaymas, Sonora, with 51 endemic species; (2) the second zone with an imaginary line from La Paz Bay (B.C.S.; Baja California Sur) to Topolobampo (Sinaloa) with 41 endemic species; (3) the third zone is located with an imaginary line from Cabo San Lucas (B.C.S.) to Cabo Corrientes (Jalisco) with 10 endemic species. Besides, 14 endemic species are distributed throughout the GC (Espinoza-Ávalos, 1993). The macroalgal communities are subject to the changing environmental conditions in the diverse habitats in the GC that delimit their zonation, which tolerates a series of anatomical and physiological adaptations to water movement, temperature, sun exposure, light intensities, low pCO 2 , and desiccation (Espinoza-Avalos, 1993).

Macroalgae sampling
In this study, the GC coastline (21-30 • N latitude) was divided into six coastal sectors based on the three phycofloristic zones along peninsular and continental GC coastlines (Fig. 1a). In each coastal sector, selected ecosystems and representative habitats were sampled based on macroalgae communities' presence and habitat characterization. Habitats were classified by substrate type (e.g., sandy-rock, rocky shore), hydrodynamic (slow to faster water flows), protection level (exposed or protected sites), and immersion level (intertidal or subtidal) (Fig. 1b).
Based on the local environmental factors, four to five macroalgae specimens of the most representative species were gathered by hand (free diving) during low tide. A total of 809 composite samples were collected from marine habitats along both GC coastlines. The percentages of specimens collected for the substrate type were 28 % sandy-rock and 72 % rocky shores based on the habitat features. In the hydrodynamic, 30 % of the specimens were collected in habitats with slow to moderate and 70 % with moderate to fast water movement. Regarding the protection level, 57 % were exposed specimens, and 43 % were protected. Finally, 56 % were intertidal and 44 % subtidal macroalgae organisms concerning the emersion level. About half of the protected specimens were collected in isolated rock pools, which was noted.
In four to five sites of each habitat, we measured in situ the salinity, temperature, and pH by using a calibrated multiparameter sonde (Y.S.I. 6600V) and the habitat characteristics mentioned above noted. Besides, composite water samples were collected for a complementary analysis of nutrients, alkalinity (and their chemical components), and δ 13 C DIC (data not included). Briefly, the representative habitats were classified by pH levels of > 9.0 "alkalinized", 7.9-8.2 "typical", and < 7.9 "acidified". Based on colder (< 20 • C), typical (20-25 • C), and warmer (> 25 • C) temperatures, 72 % of the specimens were collected at typical, 22 % at alkalinized, and 6 % at acidified pH values. Regarding the temperature, about 55 % of the specimens were collected at typical, 31 % at warmer, and 14 % at colder seawaters. Regarding salinity, most of the ecosystems showed typical values for seawater (35.4 ± 0.91 PSU, from 34.5 to 36.1 PSU). In this study, the collection surveys were conducted during spring (March-April) and dry season (nominally from November to May) from 2008 to 2014. Only in a few selected ecosystems located at C1, C2, and C3 sectors was one sampling survey conducted at the end of the rainy season (nominally from June to October in 2014). Thus, these ecosystems were possible to include habitat with a salinity range varying from estuarine (23.5 ± 3.0 PSU) to hypersaline (42.7 ± 7.0 PSU) values. These habitats were mainly isolated rock pools, and only a few were sites near tidal channels receiving freshwater discharges. About 95 % of the specimens were collected at typical seawater salinity (34-36 PSU) and only 1.5 % and 3.5 % in estuarine (< 30 PSU) and hypersaline (> 37 PSU) environments, respectively. Detailed information on the selected shallow marine ecosystems, habitat characterization, and environmental conditions is summarized in the inserted table in Fig. 1.

Macroalgae processing and analysis of the isotopic composition of carbon
The collected material was washed in situ with surface seawater to remove the visible epiphytic organisms, sediments, sand, and debris and then thoroughly rinsed with Milli-Q water. The composite samples were double-packed in a plastic bag, labeled with the locality's name and collection date, placed in an ice cooler to be kept to 4 • C, and immediately transported to the laboratory UAS-Facimar in Mazatlán. In the field, sample aliquots were also preserved in 4 % v/v formaldehyde solution for taxonomic identification to the genus or species level (when possible). The following GC macroalgal flora identification manuals were consulted (Abbot and Hollenberg, 1976;Dawson, 1944Dawson, , 1954Dawson, , 1956Dawson, , 1961Dawson, , 1962Dawson, , 1963Norris, 2010;Ochoa-Izaguirre et al., 2007;Gardner, 1920, 1924). In the laboratory, macroalgae samples were immediately frozen at −30 • C until analysis. Then, samples were freezedried at −38 • C and 40 mm Hg for 3 d, upon which they were ground to a fine powder and exposed to HCl vapor for 4 h (acid-fuming) to remove carbonates and dried at 60 • C for 6 h (Harris et al., 2001). Aliquots of ∼ 5 mg were encapsulated in tin cups (5 × 9 mm) and stored in sample trays until analysis. Macroalgae samples were sent to the Sta-ble Isotope Facility (SIF) at the University of California at Davis, CA, USA. Natural 13 C relative abundance relative to 12 C in samples was determined with mass spectrometry, using a Carlo Erba elemental analyzer attached to a Finnigan Delta S mass spectrometer equipped with a Europa Scientific stable isotope analyzer (ANCA-NT 20-20) and a liquidsolid preparation unit (PDZ, Europa, Crewz, UK). Isotope ratios of the samples were calculated using the equation δ (‰) = [(R sample /R standard − 1) × 1000], where R = 13 C/ 12 C. The R standard is relative to the international V-PDB (Vienna PeeDee Belemnite) standard. During the isotopic analysis, the SIF lab used different certified reference materials (e.g., IAEA-600, USGS-40, USGS-41, USGS-42, USGS-43, USGS-61, USGS-64, and USGS-65) for the analytical control quality. The analytical uncertainties reported for the SIF lab were 0.2 ‰ for δ 13 C (https://stableisotopefacility. ucdavis.edu/carbon-and-nitrogen-solids, last access: 18 January 2021). We also included triplicate aliquots of several specimens of the same species and condition, collected from one patch or attached to the same substrate, to assess the method error by sampling and processing procedures. The methodological uncertainties were < 0.4 ‰.
Macroalgae were grouped according to their morphofunctional characteristics proposed initially by Littler and Littler (1980) and modified by Balata et al. (2011). Most of the macroalgae species showed a limited distribution along the GC coastlines. Few cosmopolites' species included Colpomenia tuberculata, Sargassum sinicola, Padina durvillei, and Ulva lactuca. Also, not all morphofunctional groups and taxa were present in every site during each sampling survey, and the sample size in each group varied for taxa, location, and time.
A basic statistical analysis of δ 13 C values in different macroalgae groups was applied to distribute and calculate the arithmetic mean, standard deviation, and minimum and maximum. Because not all macroalgal species were present in sufficient numbers at different collection habitats, several macroalgal groups were not considered for statistical analysis. We compared taxa and morphofunctional groups collected in the same habitat (within-subjects factor) by multivariate analysis of variance. When differences were noted, a Tukey-Kramer HSD (honestly significant difference) test was performed. Besides, variations in δ 13 C macroalgae in specimens of the same morphofunction and taxon collected in different habitats were also investigated with a Kruskal-Wallis test.
The relationships between δ 13 C with the inherent macroalgae properties (taxon and morphology), biogeographical collection zone (GC coastline and coastal sector), habitat features (substrate, hydrodynamic, protection, and emersion level), and environmental conditions (temperature, pH, and salinity) were examined through simple and multiple linear regression analyses. Excepting temperature, pH, and salinity, most of the independent variables are categorical independent variables. Simple linear regression analyses were performed to establish the relationships between δ 13 C macroalgae with each environmental parameter analyzed as possible driving factors (e.g., temperature, salinity, and pH). Multiple linear regression analyses were conducted to evaluate the combined effects of those independent variables (macroalgae properties, biogeographical collection zone, habitat features, and environmental conditions) on the δ 13 C macroalgae. In the multivariable regression model, the dependent variable, δ 13 C macroalgal values, is described as a linear function of the independent variables X i , as follows: where a is regression constant (it is the value of intercept, and its value is zero), b 1 , b 2 , and b n are regression coefficients for each independent variable X i . From each one of the fitted regression models, we extracted the estimated regression coefficients for each of the predictor variables: e.g., Bayesian information criterion (BIC), Akaike information criterion (AIC), root-mean-square error (RMSE), Mallow's Cp criterion, F ratio test, the p value for the test (prob > F ), coefficients of determination (R 2 ), and the adjusted R 2 statistics (Stroup et al., 2018). All regression coefficients were used as indicators of the quality of the regression (Burnham and Anderson, 2002;Draper and Smith, 1998). The Kolmogorov-Smirnov normality test was applied for all variables, and all were normally distributed. Most of the δ 13 C values in each group showed a normal distribution. For all statistical tests, a probability P < 0.05 was used to determine statistical significance. The statistical analysis of the results was using JMP 14.0 software (SAS Institute Inc.).
Aggrupation of δ 13 C values based on morphofunctional features is displayed in Fig. 4. The most representative groups in the phylum Chlorophyta varied from −15.8 ± 0.3 ‰ for C-Tubular to −12.4 ± 0.5 ‰ for C-Thallus erect. The phylum Ochrophyta includes O-Thick leathery with the lowest mean (−14.8 ± 0.3 ‰) and O-Hollow with a spherical or subspherical shape with the highest values (−9.2 ± 0.3 ‰). The lowest and highest δ 13 C values for Rhodophyta were observed for R-flattened macrophytes (−24.0 ± 9.6 ‰) and R-Larger-sized articulated coralline (−7.9 ± 0.8 ‰), respectively. Significant differences were observed among groups, which were ordered as follows: R-Flattened macrophytes < R-Blade-like < C-Tubular < O-Tick leathery and R-Larger-sized corticated < C-Blade-like and C-Filamentous uniseriate < C-Thallus erect and O-Compressed with branch < O-Hollow with spherical < R-Larger-sized articulated coralline.
Significant differences were observed among the genera related to the pH level in seawater (Fig. 7b). Under typical pH seawater, Amphiroa and Colpomenia were 1 ‰-2 ‰ more negatives than in alkaline waters, while Ulva and Spyridia were 3 ‰-5 ‰ less negative than in acidic waters. Amphiroa and Colpomenia were not collected in acidic water, and neither was Spyridia in alkaline waters to compare. Another genus also showed extremes values between alkaline (Tacanoosca −7.6 ± 1.0 ‰) and acidic waters (Schizymenia −32.9 ± 2.0 ‰). The following order was observed in the genera collected at the three pH ranges: alkaline > typical > acidic. Significant differences were observed for genera Ahnfeltiopsis, Caulerpa, Gymnogongrus, Padina, and Ulva, with higher values in alkaline than in acidic waters. Values of δ 13 C for specimens of the same genus collected in typical pH waters are mostly overlapped between alkaline and acidic seawaters. Non-significant differences in δ 13 C values were observed for Grateloupia, Hypnea, and Polysiphonia concerning pH-type waters.
We analyzed the carbon uptake strategies on macroalgal assemblages as a function of environmental factors like temperature, pH, and salinity (Fig. 8). The temperature and salinity non-significantly explained the δ 13 C macroalgal variability. A poor but significant correlation was observed be- Figure 5. Proportion of species using different DIC sources according to their carbon uptake strategies: HCO − 3 only users (CO 2 concentrating mechanism active), users of both sources (HCO − 3 and CO 2 ), and CO 2 only users (non-CO 2 concentrating mechanism active) on the coast along the GC. tween δ 13 C and pH (R 2 = 0.04) ( Table 4). The proportion of specimens with a strategy of only HCO − 3 use was different between environmental factors and taxa (previously described). For example, Ochrophyta showed the highest proportion (35 %) in colder temperatures, in pH alkaline (31 %), and in a typical salinity regimen (27 %). Chlorophyta was enhanced to 30 % in acid pH, and Rhodophyta recorded 21 % in normal seawater. The opposite strategy (only use of dissolved CO 2 ) was observed only in Rhodophyta. The high-est percentage was observed in the estuarine salinity regimen (10 %).

Variation latitudinal of δ 13 C macroalgae
The δ 13 C macroalgal variation in the GC biogeography was evaluated by linear regression analysis between δ 13 C values along the 9 • latitude of both GC coastlines. A nonsignificant latitudinal trend was observed for datasets, but for the three phyla's most representative genera, δ 13 C values correlated with latitude ( Fig. 9). In Chlorophyta, with the higher genera number, δ 13 C values increased with latitude, with low but significant correlation. Contrarily, in Ochrophyta and Rhodophyta specimens, the δ 13 C values decreased non-significantly with latitude.

Analyses of δ 13 C macroalgal variability
The δ 13 C macroalgal variability was analyzed as a function of the life-form and environmental factors. Firstly, simple linear regression analyses were performed to evaluate the de-  Table 4. Summary of the estimated regression coefficients for each simple linear regression analysis and of the constant of fitted regression models. Estimated regression coefficients include degree of freedom for the error (DFE), root-mean-square error (RMSE), coefficient of determination (R 2 ) and the adjusted R 2 statistics, Mallow's Cp criterion (Cp), Akaike information criterion (AIC), Bayesian information criterion (BIC) minimum, F ratio test, and p value for the test (prob > F ). Model information includes value of the constant a (δ 13 C, ‰), standard error (SE), t ratio, and prob > |t| (values * are significant). pendent variable's prediction power (δ 13 C macroalgal variable) as a function of several independent variables controlling the main macroalgae photosynthesis drivers (light, DIC, and inorganic nutrients). Regression coefficients were estimated for each fitted regression model, which are used as indicators of the quality of the regression (Burnham and Anderson, 2002;Draper and Smith, 1998) as was described in Methods; however, the description of our results focused on the coefficients of determination (R 2 and adjusted R 2 ). The coefficient R 2 describes the relationship between the independent variables X i with the dependent variable Y (δ 13 C macroalgal values). R 2 is interpreted as the percent of contribution to the δ 13 C variability. In comparison, the adjusted R 2 statistics compensate for possible confounding effects between variables.
Results of the analysis of the relationships between δ 13 C with each independent variable are summarized in Table 4. Phyla explain only 8 % variability regarding the inher-12 R. Velázquez-Ochoa et al.: An analysis of the variability in δ 13 C in macroalgae Figure 8. Proportion of species using different DIC sources according to their carbon assimilation strategies: HCO − 3 only users (CO 2 concentrating mechanism active), users of both sources (HCO − 3 and CO 2 ), and CO 2 only users (non-CO 2 concentrating mechanism active) as a function of (a) pH ranges, (b) temperature ranges, and (c) salinity ranges. ent macroalgae properties, the morphofunctional properties 35 %, genera 46 %, and species 57 %.
The biogeographical collection zone, featured by coastline (continental versus peninsular) and coastal sectors (C1-C3 and P1-P3), explained a maximum of 5 % variability. Only the emersion level (6 %) contributed to the δ 13 C variability related to the habitat features. The contribution of the seawater's environmental conditions was marginal for pH (4 %) and negligible for temperature and salinity. A marginal reduction in the percentage of contribution was observed for phyla (1 %) and morphofunctional properties (1 %), but it was significant for genera (5 %) and species (10 %).
Multiple regression analyses were also performed to interpret the complex relationships among δ 13 C macroalgae, considering the life-forms (morphofunctional properties and taxa by genus) and their responses to environmental parameters. Results for the fitted regression models performed for morphofunctional groups (Table 5) and genera (Table 6) evidenced that the effect of the coastal sector and pH ranges on the δ 13 C macroalgae increased the contribution by 9 %-10 % for each one. The emersion level increased by 5 %-6 %, the contribution with respect to the individual effect of morphofunctional group and genus, and the temperature and pH by 1 % and 3 %, respectively, while salinity decreased by 1 %-2 %. The combined effect of the biogeographical collection zone (e.g., coastline sector) and morphofunctional group (Table 5) and genus (Table 7) increased in 11 %-12 %.
Considering the combined effect of the coastline sector + habitat features for morphofunctional group or genus (Table 7), the full model showed R 2 values of 0.60 and 0.71. In contrast, coastline sector + environmental conditions + morphofunctional group or genus the R 2 increased to 0.62 and 0.72, respectively. The interactive explanations of environmental factors increased the explanation percentage of δ 13 C variability; however, these contributions were significantly lower than those explained by life-forms, such as the morphofunctional properties and taxa by genus and species.

Preliminary estimations of 13 C macroalgae
Concurrent analysis of surface seawater for alkalinity, proportions of the chemical species of DIC (CO 2 , HCO − 3 , and CO 2− 3 ), and δ 13 C DIC evidenced that δ 13 C DIC in GC seawater averages 1.4±0.4 ‰ (−1 ‰ to 4.9 ‰) (Fig. S1). In our preliminary data, the δ 13 C-DIC seawater slightly (in 0.5 ‰) decreased during the rainy season in those zones influenced by river discharges along the continental coastline. Non-significant differences were observed among coastal sectors. The δ 13 C-DIC values in GC seawater are comparable to the averages 1.4 ‰-1.6 ‰ reported for the surface seawaters in the eastern North Pacific in the 1970s-2000s (Hinger et al., 2010;Quay et al., 2003;Santos et al., 2011).
Based on the subtraction of δ 13 C macroalgae to δ 13 C-DIC seawater, the integrative discrimination factor against 13 C averaged 16.0 ± 3.1 ‰, 16.8 ± 4.3 ‰, and 14.0 ± 3.8 ‰ for phyla Chlorophyta, Rhodophyta, and Ochrophyta, re- Figure 10. Trends in the δ 13 C macroalgae in specimens for morphofunctional groups by taxa along the coastline of the Gulf of California as a function of latitudinal gradient. Table 5. Summary of the estimated regression coefficients for each multivariate linear regression analysis and of their constant of fitted regression models performed in individuals binned by genus. Estimated regression coefficients include degree of freedom for the error (DFE), root-mean-square error (RMSE), coefficient of determination (R 2 ) and the adjusted R 2 statistics, Mallow's Cp criterion (Cp), Akaike information criterion (AIC), Bayesian information criterion (BIC) minimum, F ratio test, and p value for the test (prob > F ). Model information includes value of the constant a (δ 13 C, ‰), standard error (SE), t ratio, and prob > |t| (values * are significant). spectively. Five groups were identified as a function of the 13 C values: one for Chlorophyta ( 13 C = 16.0 ± 3.1 ‰), two for Rhodophyta (16.6 ± 3.8 ‰ and 34.6 ± 1 ‰), and two for Ochrophyta (9.1 ± 1.7 ‰ and 15.7 ± 2.7 ‰) (Fig. S2). Values of 13 C were comparable to δ 13 C of the thallus of macroalgae. Thus, δ 13 C macroalgae reflect mainly the discrimination during carbon assimilation. Like δ 13 C macroalgae, the 13 C values were subject to considerable variation.

Explaining the δ 13 C macroalgal variability
A high variability in the δ 13 C values was revealed in the large inventory of macroalgae collected along the GC coastline. A linear regression analysis of the effects of life-forms revealed that the δ 13 C variability in the macroalgal community is mainly explained by taxonomic (genus 46 %, species 57 %) and morphofunctional group (35 %). This result is consistent with the report of Lovelock et al. (2020), which found that 66 % of δ 13 C variability was explained by taxonomy. Even so, the variability associated with each genus is not the same and can be classified in three groups: (1) high variability (e.g., Schizymenia = ±19.1 ‰), moderate variability (e.g., Hydroclathrus = ±7.3 ‰; Amphiroa = ±6.8 ‰), and low variability (e.g., Gracilaria = ±0.89; Spyridia = ±1.46 ‰). The observed δ 13 C variability in this study is comparable with those reported in the literature, compiled in Table S4.
Most authors studying the isotopic composition of C in macroalgae have reported the high isotopic variability, which has been attributable to the taxon-specific photosynthetic DIC acquisition properties (Díaz-Pulido et al., 2016;Lovelock et al., 2020;Marconi et al., 2011;Mercado et al., 2009;Raven et al., 2002a;Stepien, 2015). Our study observed that the intrinsic characteristics of each morphofunctional group of macroalgae (e.g., thallus structure, growth form, branching pattern, and taxonomic affinities) also influence the δ 13 C macroalgal signals. The thallus thickness influences the diffusion boundary layer on the surface of the macroalgae, where they carry out the absorption of essential ions and dissolved gases (Hurd, 2000;Sanford and Crawford, 2000). Thus, morphology can modulate the photosynthesis rates. Table 7. Summary of the estimated regression coefficients for each multivariate linear regression analysis and of their constant of fitted regression models performed in individuals binned in coastline sector, habitat features, environmental conditions, and physiological state separately by morphofunctional group and genus. Estimated regression coefficients include degree of freedom for the error (DFE), rootmean-square error (RMSE), coefficient of determination (R 2 ) and the adjusted R 2 statistics, Mallow's Cp criterion (Cp), Akaike information criterion (AIC), Bayesian information criterion (BIC) minimum, F ratio test, and p value for the test (prob > F ). Model information includes value of the constant a (δ 13 C, ‰), standard error (SE), t ratio, and prob > |t| (values * are significant).  However, a non-biological or ecological explanation of the δ 13 C variability, and therefore carbon use physiology, can be given in terms of morphology.
Many species that recorded high δ 13 C values (and low 13 C values) were fleshy macroalgae that are characterized to be bloom-forming macroalgae belonging to genera Ulva, Gracilaria, Cladophora, Spyridia, and Sargassum Table 9. Constant of fitted regression model explaining the δ 13 C variability by genus. Model information includes value of the constant a (δ 13 C, ‰), standard error (SE), t ratio and prob > |t|. Only genera with significant effects are enlisted.  (Páez-Osuna et al., 2013;Valiela et al., 2018). It is not surprising that species with high photosynthetic activity and high relative growth rates (Hiraoka et al., 2020) have high carbon demand that results in lower isotopic discrimination against 13 C (Carvalho et al., 2010a, b;Cornelisen, et al., 2007;Kübler and Dungeon, 2015;Rautenberger et al., 2015). Bloom-forming macroalgae (e.g., Ulva, Gracilaria, Sargassum) have been remarked as facultative species capable of switching from C 3 to C 4 pathway (Valiela et al., 2018). C 4 pathway reduces photorespiration, the antagonist process of RuBisCo, enhancing the DIC assimilation in 25 %-40 % and increasing the δ 13 C values (Bauwe et al., 2010;Ehleringer et al., 1991;Zabaleta et al., 2012). C4 pathway has more energy investment in CCMs than in RuBisCo protein content than C3 pathway (Young et al., 2016). Also, the reports of C4 or C4-like pathway features in algae have increased in the last years (Doubnerová and Ryšslavá, 2011;Roberts et al., 2007;Xu et al., 2012Xu et al., , 2013. For example, high activity of key enzymes of C4 metabolisms, such as pyruvate orthophosphate dikinase (PPDK), phosphoenolpyruvate carboxylase (PEPC), and phosphoenolpyruvate carboxykinase (PCK), has been described in many algae species. But the establishment of a true C4 pathway in marine algae is not clear since the massive changes in gene expression patterns seem to be incomplete, and it is suggested that many marine algae have high plasticity to use a combination of CCM to overcome DIC limitations (Doubnerová and Ryšlavá, 2011;Roberts et al., 2007;Xu et al., 2012Xu et al., , 2013. A stepwise model of the path from C3 to C4 photosynthesis is explained by Gowik and Westhoff (2011). More research is required on this topic considering the increasing frequency, intensity, and extension of bloom-forming macroalgae events worldwide (Teichberg et al., 2010;Valiela et al., 2018) and in México (Ochoa-Izaguirre et al., 2007;Ochoa-Izaguirre and Soto-Jiménez, 2015;Páez-Osuna et al., 2017). Changes in the habitat features and environmental conditions, such as light intensity and DIC availability, influencing the growth rate and photosynthetic intensity, have a strong influence on δ 13 C signal (Carvalho et al., 2007(Carvalho et al., , 2009aCarvalho and Eyre, 2011;Mackey et al., 2015;Rautenberger et al., 2015;Stepien, 2015). The light intensity is the external factor with more influence on the 13 C macroalgae due to the regulation of carbon assimilation intensity (Carvalho et al., 2009a, b;Cooper and DeNiro, 1989;Grice et al., 1996). Experimental studies found the light levels to be a critical factor affecting the δ 13 C values. For example, under saturating light conditions, Ulva switched from a carbon uptake of HCO − 3 and CO 2 to increased HCO − 3 use (Rautenberger et al., 2015). Furthermore, field studies have shown that species growing in low-light habitats like deep subtidal zones tend to have more negative δ 13 C values than those in higherlight environments Díaz-Pulido et al., 2016;Hepburn et al., 2011;Marconi et al., 2011;Mercado et al., 2009;Stepien, 2015). In this study, intertidal specimens recorded less negative values than subtidal in most macroalgae genera. However, our study did not record the vertical effect in the δ 13 C signal related to the light limitation because only shallow habitats (non-light limited) were studied.
The effect of other environmental factors, such as salinity and pH, on δ 13 C macroalgal signals was evaluated. Regarding salinity, the influence of freshwater discharge by rivers and groundwater decreases the δ 13 C signal, which could be explained by the reduction in the salinity regimen that follows a decrease in δ 13 C DIC in water (Hinger et al., 2010;Santos et al., 2011). In our study, a non-significant correlation between δ 13 C macroalgae and salinity was observed.
Based on pH, differences in δ 13 C were found only for a few genera (e.g., Amphiroa, Colpomenia, Ulva, Spyridia), with an increasing trend in the δ 13 C values with pH increase, such as was reported by Maberly et al. (1992) and Raven et al. (2002b). Similar results were reported for Cornwall et al. (2017) in the field study, with the differential response of Table 10. Constants of fitted regression model explaining the δ 13 C variability by species. Model information includes value of the constant a (δ 13 C, ‰), standard error (SE), t ratio, and prob > |t|. Only genera with significant effects are enlisted. the δ 13 C signals to pH among 19 species, in which only four species were sensitive to pH changes. A very weak but significant positive linear regression was observed between δ 13 C and pH. Also, a decreasing trend in the δ 13 C was recorded in the following order: alkaline > typical > acidic. According to Stepien (2015), the result of meta-analyses between pH drift experiments and δ 13 C thresholds was positive only for Rhodophyta and Ochrophyta but not for Chlorophyta. About 86% of the Stepien metadata met the theoretical CCM assignation based on both parameters, with exceptions for species with δ 13 C < −30 ‰ that have been capable of raising pH to > 9. A strong association between pH compensation point and δ 13 C was reported by Iñiguez et al. (2019) in three taxa of polar macroalgae. Environmental conditions may influence the δ 13 C macroalgal values but not change the carbon use physiology in the macroalgae, which is most likely inherently species-specific.

Using δ 13 C macroalgae to indicate the presence of an active CCM
In our study, the δ 13 C macroalgal signals were used to evidence the presence of an active CCM. This tool was first used in macroalgal shallow communities of the GC. Most macroalgae species displayed δ 13 C values that exhibit active CCMs. Then, macroalgae were classified into three strategies for DIC uptake, in agreement with the Maberly et al. (1992) and Raven et al. (2002a) thresholds: (1) CCM-only by active uptake HCO − 3 (δ 13 C > −10 ‰), (2) CCM active uptake HCO − 3 and diffusive uptake CO 2 (δ 13 C < −11 ‰ to −30 ‰), and (3) non-CCM CO 2 by diffusion only (δ 13 C < −30 ‰).
plemented by simultaneous measurements of O 2 and CO 2 produced and consumed, respectively, using MIMS. For example, photosynthetic O 2 production in a certain macroalgal species with an active CCM preference (e.g., CO 2 ) is about 10 times higher than a non-active CCM (Burlacot et al., 2020).
Based on the δ 13 C values, it is possible to assume that at least one basal CCM is active. However, it is not possible to discern what type of CCM is expressed in the organisms (e.g., direct HCO − 3 uptake by the anion-exchange protein -AE; Drechsler and Beer, 1991;Drechsler et al., 1993) or types of mitochondrial carbonic anhydrase (e.g., internal and external) that enhance the fixation of DIC by recycling mitochondrial CO 2 (Bowes, 1969;Sand-Jensen et al., 2020;Zabaleta et al., 2012). Also, the co-existence of different CCMs has been described for the same species (Axelsson et al., 1999;Xu et al., 2012), and it has even been described that different CCMs can operate simultaneously, generating different DIC contributions to RuBisCo internal pool (Rautenberger et al., 2015). The variety of CCMs and their combinations could contribute to the high δ 13 C variability for the same species. In our field study, it is impossible to explain the variations in δ 13 C or 13 C macroalgae relative to CCM or CA activity types. Controlled experiments, like those conducted by Carvalho and collaborators (e.g., Carvalho et al., 2009aCarvalho et al., , b, 2010a, are required to obtain this knowledge.

Variability in δ 13 C macroalgae between the GC bioregions
Changes in the δ 13 C signal with latitude, mainly related to the light and temperature, have been reported in the literature (Hofmann and Heesch, 2018;Lovelock et al., 2020;Marconi et al., 2011;Mercado et al., 2009;Stepien, 2015). For example, a negative correlation between latitude and δ 13 C macroalgae was described by Stepien (2015). The authors concluded that the δ 13 C signal increased by 0.09 ‰ for each latitude degree from the Equator. Hofmann and Heesch (2018) showed a robust decreasing latitudinal effect in δ 13 C signals (R 2 = 0.43δ 13 C total and 0.13, for δ 13 C organic-tissue , p = 0.001) for rhodolite and macroalgae from coral reefs in Australia. In both cases, the latitude range is higher than what we tested (30 to 80 • and from 10 to 45 • , respectively). These differences on a large scale tend to be associated with a temperature effect (Stepien, 2015) and their effect on CO 2 solubility in seawater (Zeebe and Wolf-Gladrow, 2001). However, in our study, no geographical pattern in the δ 13 C macroalgae was observed. Our linear regression analyses for latitudes showed a low but significant correlation for the dataset classified by morphofunctional group and genus -negative in the cases of Rhodophyta and Ochrophyta groups and positive for Chlorophyta.
Light is not limited along the GC latitudes. Most of the shallow habitats occupied by macroalgal communities in the GC were high-light environments. In agreement with the literature, the surface seawater temperature across the GC varies by only 1 • C annual mean (Escalante et al., 2013;Robles-Tamayo, 2018). However, larger temperature variations of 5-10 • C were recorded in the coastal waters across the GC bioregions in both climatic seasons. The combined effect of the coastline sector, habitat feature, and environmental condition for morphofunctional group or genus explained 60 %-62 % and 71 %-72 % of the δ 13 C variability, respectively. Our analysis of variability for the bestrepresented morphological groups (e.g., R-Filamentous uniseriate and pluriseriate with erect thallus and C-Tubular) and genera (e.g., Colpomenia, Padina, Polysiphonia, and Gracilaria) revealed that certain life-forms are better monitors explaining the variability in δ 13 C macroalgae (and 13 C values) than others. The δ 13 C variability in morphological groups refers to change within a specific carbon use strategy but not change in the carbon use physiology that is inherently species-specific. The biological or ecological relevance of the δ 13 C variability as a function of the morphology, in terms of the efficiency in the use of DIC and the isotope discrimination during carbon assimilation and respiration, must be investigated in species of the same genus but which are morphologically different or between the same morphological structures belonging to a different taxon.
The proportion of specimens with different carbon uptake strategies also showed regional variations. For example, the facultative uptake of HCO − 3 and CO 2 was dominant in the macroalgal shallow communities in the GC (60 % to 90 % of specimens). Exceptions were observed for Ochrophyta in the P1 (68 %) and C1 (37 %) regions, where the strategy using only HCO − 3 dominated, while the strategy based on the use of only CO 2 was observed in the peninsular coast in P2 and P3 for Rhodophyta with 2 %-3.3 %. Finally, the coastal sector C2 showed more negative δ 13 C values in macroalgae specimens of the same genus compared to the peninsular coastline (P1-P3). Small but detectable changes were observed in the phylum distribution based on environmental conditions. For example, Ochrophyta showed the highest proportion (35 %) in colder temperature, in pH alkaline (31 %), and in typical salinity regimen (27 %), while Chlorophyta enhanced to 30 % in acid pH, and Rhodophyta recorded 21 % in normal seawater. The opposite strategy (only use of dissolved CO 2 ) was observed only in Rhodophyta. The highest percentage was observed in the estuarine salinity regimen (10 %). Again, more research is required to obtain valuable information on the physiological and environmental status of macroalgae.

Conclusions
In conclusion, we observed high δ 13 C macroalgal variability in macroalgae communities in the Gulf of California, such as reported in other worldwide marine ecosystems. The lifeform is the principal cause of δ 13 C macroalgal variability, which explains up to 57 %. Changes in habitat characteristics and environmental conditions also influence the δ 13 C macroalgal variability within a specific carbon use strategy. Considering the combined effect of the life-form, coastline sector, and environmental conditions, the full model explains up to 72 % (genus) of the variability. The effects of the coastal sector, pH ranges, and emersion level were significant, while for salinity and temperature they were negligible.
Most macroalgae inhabiting in GC displayed the presence of CO 2 concentrating mechanisms to uptake HCO − 3 for photosynthesis, and 84 % of the total analyzed specimens were able to use both HCO − 3 and/or CO 2 employing active uptake plus passive diffusion (strategy 2: −10 < δ 13 C > −30 ‰). Specimens belonging to 58 species of 170 total species showed carbon uptake strategy 1 that uses only HCO − 3 . A higher proportion of CCM species (HCO − 3 users) was expected because we focused on intertidal and shallow subtidal habitats featured by high light intensities. Only three non-calcifying species (Schizymenia pacifica, Halymenia sp., Gigartina sp.) belonging to Rhodophyta (3 %) were CO 2exclusive users (strategy 3: δ 13 C < −30 ‰). The low percentage of CO 2 dependents versus 40 %-90 % reported for temperate regions could be related to the shallow habitat sampled in our surveys (< 2 m depth low tide). The calcifying macroalgae genera Amphiroa and Jania using HCO − 3 (high δ 13 C values) were present in the macroalgal communities in the GC. Because of the ongoing ocean acidification, these calcifying organisms constitute excellent ecological sentinels in the GC.
Finally, diverse authors have reported significant correlations between δ 13 C signal and latitude, mainly related to the light and temperature. However, in our study's latitude range (21-31 • N), the linear regression analyses showed a low correlation for the δ 13 C macroalgal dataset classified by morphofunctional group and genus, which was negative for Rhodophyta and Ochrophyta and positive for Chlorophyta. Non-clear δ 13 C macroalgal patterns occur along the GC latitudes. However, detectable changes were observed in the δ 13 C macroalgae and the proportion of specimens with different carbon uptake strategies among coastal sectors. For example, the facultative uptake of HCO − 3 and CO 2 was dominant in the macroalgal shallow communities in the GC (60 % to 90 % of specimens), but in the P1 (68 %) and C1 (37 %) the use of only HCO − 3 was the dominant strategy. Our research is the first approximation to understand the δ 13 C macroalgal variability in one of the most diverse marine ecosystems in the world, the Gulf of California. We did not pretend to resolve the intricate processes controlling the variations in δ 13 C or 13 C macroalgae during carbon assimilation and respiration and determine the isolated influence of each environmental factor. Despite the large dataset and corresponding statistical analyses, our study faces limitations due to research design and because no research on δ 13 C macroalgal analysis was developed previously in the GC. The primary deficiency is the lack of pH drift experiments to discriminate δ 13 C signal variations in the carbon uptake strategies to determine preferential DIC uptake of macroalgae (CO 2 or HCO − 3 ). The second limitation concerns the lack of controlled experiments to discern what type of CCM is expressed in macroalgae (e.g., direct HCO − 3 uptake by the anion-exchange protein AE, types of mitochondrial AC, or the co-existence of different CCMs). Also, more research is required to assess the biological or ecological relevance of the δ 13 C variability as a function of the morphology (e.g., DIC uptake efficiency and isotope discrimination during carbon assimilation and respiration). Future studies assessing the ability of macroalgae to use CO 2 and/or HCO − 3 can be assessed by pH drift experiments and MIMS in the cosmopolites' species and within genus with differences in the δ 13 C values between species (e.g., Ulva and Sargassum). Finally, controlled experiments in laboratory and mesocosm type combined with field studies are required to elucidate what type of CCM is expressed in macroalgae. Even so, the δ 13 C macroalgae were a good indicator to infer the presence or absence of CCMs, to identify the macroalgae lineages that could be in a competitive advantage based on their carbon uptake strategy, and to identify their geographical distribution along with GC. Under the current climate change conditions and their effects as ocean acidification progresses and the bloom-forming macroalgae events increase in Mexico and worldwide, the analysis of δ 13 C macroalgae constitutes an excellent tool to help to predict the prevalence and shift of species in macroalgal communities which are focused on carbon metabolism. However, to obtain the maximum benefit from isotopic tools in the carbon-use strategy study, diverse and species-specific, it is necessary to use them in combination with other techniques referred to herein.
Author contributions. RVO participated in the collection, processing, and analysis of the samples as a part of his master's degree thesis. MJOI also participated in sample collections and identified macroalgae specimens. MFSJ coordinated the research, was the graduate thesis director, and prepared the manuscript with contributions from all co-authors.
Competing interests. The contact author has declared that neither they nor their co-authors have any competing interests.

22
R. Velázquez-Ochoa et al.: An analysis of the variability in δ 13 C in macroalgae Disclaimer. Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Acknowledgements. The authors would like to thank Humberto Bojórquez-Leyva, Yovani Montaño-Ley, and Arcelia Cruz-López for their invaluable field and laboratory work assistance. Thanks to Sarahí Soto-Morales for the English revision. UNAM-PAPIIT IN206409 and IN208613 provided financial support, and UNAM-PASPA supported MF Soto-Jimenez for a sabbatical year. Thanks to CONACYT for a graduate fellowship to Roberto Velázquez-Ochoa.
Financial support. This research has been supported by the Dirección General de Asuntos del Personal Académico, Universidad Nacional Autónoma de México (grant nos. PAPIIT IN206409 and PA-PIIT IN208613).
Review statement. This paper was edited by Aninda Mazumdar and reviewed by Matheus C. Carvalho, Michael Roleda, and one anonymous referee.