Inputs of seabird guano alter microbial growth, community composition and the phytoplankton–bacterial interactions in a coastal system
Abstract
Seabird guano enters coastal waters providing bioavailable substrates for microbial plankton, but their role in marine ecosystem functioning remains poorly understood. Two concentrations of the water soluble fraction (WSF) of gull guano were added to different natural microbial communities collected in surface waters from the Ría de Vigo (NW Spain) in spring, summer, and winter. Samples were incubated with or without antibiotics (to block bacterial activity) to test whether gull guano stimulated phytoplankton and bacterial growth, caused changes in taxonomic composition, and altered phytoplankton–bacteria interactions. Alteromonadales, Sphingobacteriales, Verrucomicrobia and diatoms were generally stimulated by guano. Chlorophyll a (Chl a) concentration and bacterial abundance significantly increased after additions independently of the initial ambient nutrient concentrations. Our study demonstrates, for the first time, that the addition of guano altered the phytoplankton–bacteria interaction index from neutral (i.e. phytoplankton growth was not affected by bacterial activity) to positive (i.e. phytoplankton growth was stimulated by bacterial activity) in the low-nutrient environment occurring in spring. In contrast, when environmental nutrient concentrations were high, the interaction index changed from positive to neutral after guano additions, suggesting the presence of some secondary metabolite in the guano that is needed for phytoplankton growth, which would otherwise be supplied by bacteria.
INTRODUCTION
Seabirds are considered one of the main connectors between marine and terrestrial ecosystems, especially in areas close to breeding colonies (De La Peña-Lastra et al., 2021; Riddick et al., 2012). During the breeding season, the great concentration of individuals in specific areas generates an accumulation of eggs, bones, pellets, feathers, and excrements (De La Peña-Lastra et al., 2019). For centuries, the excrement of seabirds, commonly called guano, has been used as a natural fertilizer in agriculture due to its chemical properties (Szpak et al., 2012). Guano composition depends on the diet of the seabird, so it varies depending on species and habitats (Hahn et al., 2007). It is mainly composed of large amounts of bioavailable macronutrients (such as N and P) and trace metals (Fe, Zn, Cu, As, etc.) (Smith & Johnson, 1995). The total amount of excreted N and total P supplied by global seabird colonies is estimated to be 591 Gg N y−1 and 99 Gg P y−1, respectively, of which 72.5 Gg N y−1 and 21.8 Gg P y−1 may be bioavailable (Otero et al., 2018).
Guano bioavailable compounds can be accumulated over long periods of time and eventually mobilized to coastal waters, especially during rainy periods (Kazama, 2019). These extra nutrient inputs, mostly accumulate in the photic layer surrounding the seabird colonies (Petkuviene et al., 2019) and can produce or accelerate the so-called ‘ornitheutrophication’ process (Otero et al., 2018), thereby increasing the biomass of primary producers, especially phytoplankton, present in coastal areas around the colonies (Kazama, 2019; Methratta, 2004). Nutrient enrichment from seabird guano can also generate changes in the structure and composition of the phytoplankton community (Shatova et al., 2017).
Marine phytoplankton growth may be frequently limited by N, while bacteria appear to be commonly limited by P in coastal and estuarine waters (Bristow et al., 2017; Moore et al., 2013). Due to the nutritional requirements of phytoplankton and bacteria, the input of inorganic nutrients (N and P) may cause increases in phytoplankton and/or bacterial biomass in coastal regions (Caron et al., 2000; Mills et al., 2008). On the other hand, several studies carried out in micro- or mesocosms have also observed positive responses of phytoplankton communities to additions of organic nutrients, alone or in combination with inorganic nutrients, that could be associated with the concomitant stimulation of heterotrophic bacteria (Davidson et al., 2007; Martínez-García et al., 2010). Several recent studies suggest that phytoplankton may be co-limited by other compounds, such as vitamins or other bacterial secondary metabolites (Joglar et al., 2020, 2021; Prieto et al., 2015; Sañudo-Wilhelmy et al., 2014).
Phytoplankton–bacteria interactions play a key role in the regulation of microbial communities, that is, the bacterial activity could have an impact on primary productivity in coastal and oceanic waters (Joglar et al., 2020; Mayali, 2018). Both negative (e.g. competition or parasitism) and positive (e.g. mutualism or commensalism) interactions between these organisms can eventually determine how coastal microbial communities respond to nutrient inputs (Prieto et al., 2015). In addition, the magnitude and nature of such interactions can change both seasonally or locally due to changing light, temperature, nutrient availability or depending on the microbial community composition (Cirri & Pohnert, 2019). Yet, studies focused on the impact of nutrients associated with seabird guano on phytoplankton–bacteria interactions, have not been conducted so far.
Given the important role of microbial communities in coastal systems and the large impact of seabird guano on nutrient cycles, we aimed at investigating the response of the microbial plankton community, including phytoplankton and bacteria, to nutrient enrichment by seabird excrements. We additionally intended to assess the role of phytoplankton–bacteria interactions in the response of primary producers to these experimental nutrient inputs.
We followed a relatively simple approach to quantify the role of phytoplankton–bacteria interactions on the response of phytoplankton to guano inputs. We selectively blocked bacterial activity using a previously tested combination of antibiotics (Prieto et al., 2015). Then, the impact of bacteria on phytoplankton, the phytoplankton–bacteria interaction index, was calculated by dividing the phytoplankton-related response variable (Chl a concentration) in the presence of an active bacterial community (without antibiotic) by the corresponding variable after blocking bacteria (with antibiotic).
Our hypothesis is that nutrient enrichment from seabird guano causes changes in the growth rate and community composition of microbial plankton communities and that those alterations will modify the intensity and nature of phytoplankton–bacteria interactions and, consequently, the response of primary producers. This hypothesis was tested using microcosm experiments conducted under contrasting hydrographic conditions in three different seasons (spring, autumn, and winter), evaluating the response of different natural microbial plankton communities collected in surface waters near a large gull colony in the Cíes Islands (NW Spain).
EXPERIMENTAL PROCEDURES
Study area
The study area is the Ría de Vigo in the northwest of the Iberian Peninsula. This is a submerged unglaciated river valley invaded by the ocean affected by seasonal upwelling and downwelling events (García-Moreiras et al., 2018). During the spring and summer, when northerly winds are dominant, upwelling of nutrient-rich water generates high primary production (Fraga, 1981). Downwelling events occur in autumn and winter when southerly winds prevail. In this season, vertical mixing occurs, and primary production is lower than during summer (Nogueira et al., 1997). The mild stratification of the water column in early spring generates important phytoplankton blooms (Tilstone et al., 1994). The average annual precipitation of the studied area is ~1690 mm, mostly occurring during the autumn and winter seasons (Ninyerola et al., 2005).
The Cíes Islands (Illas Atlánticas National Park) (42°14′ N, 8°54′ W) are located at the mouth of the Ría de Vigo and host a high density of yellow-legged gull (L. michahellis) breeding colonies (Otero et al., 2015). These seabirds release a large amount of N (32.7 ± 12.7 mg kg−1) and P (16 ± 7.9 mg kg−1) into colony soils, mainly through excrements and pellets (De La Peña-Lastra et al., 2019; Otero & Mouriño, 2002). In addition, seabird excrements have large amounts of Zn, Cu and As (Otero et al., 2018; Rial et al., 2016).
Preparation of the seabird guano stock
Gull faeces were collected from rocky areas in the large breeding colony at the nearby of Illas Atlánticas National Park. Faeces were kept at −20°C until the preparation of the stock. Seawater was filtered through GF/F filters, inoculated with a culture of Thalassiosira rotula and incubated in a culture chamber at room temperature with a natural light cycle for 3 days, to force dissolved nutrient exhaustion. Thereafter, seawater was filtered (through GF/F) again to remove the microalgae. Then, 60 g (dry weight) of gull guano was dissolved into 2 L of the nutrient-depleted seawater. After the dissolution of the faeces in seawater, the extract was sequentially filtered through a 100 μm mesh, and through 20 and 2 μm polycarbonate filters, to remove the particulate material resulting from the faeces. Immediately, 1 L of the resulting extract was filtered sequentially through GF/C and GF/F filters. A sample was taken to analyse the concentration of organic and inorganic nutrients from the seabird guano extract stock. Finally, the stock was kept at −20°C until the experiments were carried out.
Estimation of the nutrient content of the experimental treatments
We estimated the maximum input of dissolved inorganic nutrients to the surrounding waters associated with seabird excrements as a prerequisite to design the experimental guano addition treatments used in our experiments.
Most of the gulls breeding on Cíes Islands build their nests on the western cliffs of the islands (Pérez et al., 2006), and a large number of their excrements accumulate in this area (De La Peña-Lastra et al., 2019; Otero et al., 2018). Thus, we expected large nutrient input from gull faeces into surface waters surrounding especially the western face of the Cíes Islands.
To estimate the maximum N input derived from guano into this area, we first estimated the perimeter of Cíes islands using Google Earth Pro Software. It has been also assumed that the faeces accumulated on the rocks during a 20-day period without rainfall and that upon precipitation, the N compounds in the guano would get dissolved within the first metre of the water column and across 100 m from the coastline through runoff.
M is the breeders mass (g bird−1), FEC is the energy content of the food (kJ g −1), Aeff is the assimilation efficiency of ingested food (kJ [energy obtained] kJ−1 [energy in the food]), FNC is the N content of the food (gN g−1 wet mass), tbreeding is the breeding time (days), and Ftc is the proportion of time spent at the colony during the breeding season (%).
The number of seabird individuals present in the colony of the Cíes Islands was obtained from the Mar de Aves website (https://mardeaves.org/). Generally, the clutches of the studied population are composed of three eggs, but hatching date, size and composition are highly variable. Usually, the last-laid egg is more vulnerable than the other two and more prone to die due to starvation (Pérez et al., 2006). We assumed that each pair of gulls has two developing chicks and that each chick generates half as much faeces as an adult. Then, the contribution of N excreted by the colony in 20 days was calculated from the annual estimate. The volume of water where the excrements dissolve was calculated by multiplying a strip 100 m away from the coast and 1 m deep by the perimeter.
As a result of these calculations, it was estimated that the maximum increase in excreted inorganic N concentration associated with gull excrements accumulated for 20 days in waters surrounding the Cíes Islands would be 5 μM.
Experimental design
To evaluate the microbial responses to guano derived from L. michahellis and the impacts of phytoplankton–bacteria interactions on the phytoplankton response, we carried out the following experimental design. We collected water at a 3-m depth with a 5 L acid cleaned Niskin bottle from the east of the Cíes Islands, more specifically from Playa de Rodas (42°12′53″ N, 8°53′49″ W), in three different seasons with the aim of sampling different microbial communities. The seasons were spring (March 2019), summer (June 2019), and winter (November 2018). Water was filtered in situ using a 200 μm mesh to remove mesozooplankton and stored in acid-washed polycarbonate bottles.
The water collected in the Cíes Islands was used for the guano addition experiments. Half of the experimental units were treated with antibiotics. Therefore, the initial conditions were identical for samples treated or not treated with antibiotics except for the negligible amount of antibiotics added to this treatment. To characterize the initial conditions of the experiments, samples were taken for the measurement of Chl a concentration, bacterial abundance, dissolved inorganic and organic nutrients concentrations and microbial community composition.
The triplicate treatments included the addition of guano extract at two concentrations (1× and 5×) and one unamended control. The chemical composition of guano and the added concentration of nutrients in treatment 1× is detailed in Figure 1. The added concentration of nutrients in the 5× treatment was 5-fold that in 1× treatment.
The experiments included two identical sets of guano treatments, one with antibiotics, to block bacterial activity, and one without antibiotics (Figure 1). For the treatment with antibiotics, a previously tested commercial mixture of three antibiotics including: penicillin (ca. 5000 units mL−1), neomycin (ca. 6000 units mL−1), and streptomycin (ca. 3600 units mL−1) (SIGMA, P4083) was used as described by Prieto et al. (2015) with a final concentration of 12,150 units L−1. These mixture of antibiotics have a bacteriostatic effect while phytoplankton growth and photosynthetic efficiency remain unaffected (Mulholland et al., 2011; Prieto et al., 2015). Antibiotics could inhibit prokaryote phytoplankton (i.e cyanobacteria); however, the abundance of this group was negligible in time zero samples (shown in Figure 6). We are also aware that a negative interaction index could result in the samples treated with antibiotics due to an eventual bacterial lysis and further stimulation of phytoplankton growth sustained by the incorporation of released DOM. However, the effect of a potential bacterial breakage on phytoplankton growth is expected to be negligible due to the low amount of DOM which would be released given the measured bacterial abundance (on average 1.2 × 106 cells mL−1, which considering an average of 20 fg C per bacterium and Redfield ratios, would imply ca. 2 μM of C, ca. 0.125 μM of N, and 0.02 μM of P) and because most of the released DOM would be not readily assimilable by potentially mixotrophic phytoplankton. Moreover, we did not observe a systematic increase of heterotrophic or mixotrophic eukaryotes in the samples treated with antibiotics (Figure 6G–I). Finally, if the eventual lysis of bacteria associated with the antibiotic treatment released available resources or nutrients for phytoplankton growth, then a negative interaction index would be mostly expected in March, when ambient nutrient concentration was the lowest, and not in June, when concentrations of inorganic nutrients and organic matter were higher than in March. Thus, the combined analysis of natural microbial communities treated or not treated with antibiotics allowed for the assessment of the existence and the magnitude of the effect of the phytoplankton–bacteria interaction on the microbial response to nutrient inputs from guano.
These addition experiments in combination with antibiotic and non-antibiotic treatments were carried out in sterile, non-toxic, 2 L UV-transparent Whirl-pack® bags that were placed in mesocosms with a continuous water intake from the sea. The experiments were incubated for 72 h under natural temperature and irradiance to simulate the ocean in situ conditions.
Triplicate samples were taken daily for Chl a and bacterial abundance estimations. We took samples for inorganic nutrients at the end of the experiment from the experimental units treated and non-treated with antibiotics (Figures 2 and S1), as phytoplankton, main consumer of inorganic nutrients, was actively growing in the two sets of experimental units. In contrast, and due to budgetary constraints, samples for dissolved organic nitrogen/dissolved organic carbon (DON/DOC) analyses were collected only from the experimental units without antibiotics to help in the interpretation of the bacterial community response to guano additions, as bacteria are the main consumers of organic nutrients. In contrast, bacterial activity was blocked in the treatment with antibiotics and, therefore, no measurable organic matter consumption would be expected.
Duplicates were performed for inorganic nutrients and one replicate for dissolved organic nutrients. The water from the triplicates of each treatment was pooled and 2 L was filtered through a 0.2 μm pore size Sterivex filter unit and immediately frozen in liquid nitrogen for DNA sequencing. The DNA samples were conserved at −80°C until extraction.
Inorganic and organic nutrients concentrations
Concentrations of inorganic nutrients (ammonium, nitrate, nitrite, and phosphate) were determined in 10 mL of water samples. Aliquots were taken in acid clean polyethylene bottles using plastic gloves and frozen at −20°C. These samples were analysed with an Alliance Futura segmented flow analyser by standard colorimetric methods (Grasshoff et al., 1999) except for ammonium that was measured according to the fluorometric method of Kérouel and Aminot (1997).
Aliquots for DOC and DON were filtered through 0.2 μm filters (Pall, Supor membrane Disc Filter) and collected into pre-combusted (450°C, 20 h) glass bottles and frozen at −20°C. Samples were analysed with a Shimadzu TOC-V total organic C analyser in line with a Shimadzu TNM-1 total N measurement unit. Dissolved organic N (DON) was calculated by subtracting nitrite + nitrate + ammonium from total dissolved nitrogen (TDN).
Bacterial abundance
Bacterial abundance was determined in 10 mL of water samples fixed with formaldehyde (2% final concentration) for 2 h. All samples were filtered through 0.2 μm polycarbonate filters supported by 0.45 μm nitrocellulose filters to ensure a homogeneous distribution of the cells. Prokaryote cells retained in the filter were stained with a DAPI-mix (5.5 parts of Citifluor [Citifluor], 1 part of Vectashield [Vector Laboratories] and 0.5 parts of phosphate-buffered saline solution with 4′,6-diamidino-2-phenylindole [DAPI, final concentration 1 mg mL−1]). Finally, bacterial counts were performed with epifluorescence microscope (Leica) with a 100 W Hg lamp and appropriate filter sets for DAPI. We are aware that some archaea may be also included in the counts; nevertheless, as the contribution of archaea in these samples appeared to be very low (see Figure 6), we will refer to bacterial abundance throughout the manuscript.
Chlorophyll a concentration
We measured the Chl a concentration as a phytoplankton biomass proxy. For that purpose, 100 mL of water samples was taken from the bags and filtered using 0.2 μm polycarbonate filters. The filters were immediately preserved at −20°C. The pigments were extracted with 90% acetone and were kept at 4°C overnight. Chl a fluorescence was measured by the non-acidification technique (Welschmeyer, 1994) using TD-700 Turner Designs fluorometer calibrated with pure Chl a standard solution. We used the absorption coefficient as 87.7 at 663 nm (Lorenzen & Newton Downs, 1986).
Microbial community composition
Microbial community DNA retained in 0.2 μm Sterivex was extracted with Power Water isolation kit (Qiagen) following the manufacturer's instructions. DNA sample concentration was quantified using a Qubit® 2.0 fluorometer and the Qubit dsDNA HS Assay Kit (Thermo Fischer Scientific Inc., Massachusetts, USA). Prokaryotic community composition was assessed by sequencing the hypervariable V4–V5 region from 16S rRNA gene using the following universal primers: 515F-Y and 926R (Parada et al., 2016). The region V4 from the 18S rRNA gene of eukaryotic was amplified using the following primers: TAReuk454FWD1 and TAReukREV3 (Stoeck et al., 2010). These regions were sequenced using Illumina MiSeq platform and a minimum of 2 × 250 bp paired-end reads at the Fasteris Laboratory (Geneva, Switzerland).
Sequence reads obtained by Illumina MiSeq were processed following a published pipeline (Logares, 2017). Raw reads were corrected using BayesHammer following Nikolenko et al. (2013) and Schirmer et al. (2015) methods. The corrected paired-end reads were fused with PEAR (Zhang et al., 2014). Subsequently, errors were checked for sequences over 200pband VSEARCH- 2.14.1 was used for dereplication (Rognes et al., 2016). We obtained OTUs (Operational Taxonomic Units) reads abundances by mapping back OTUs with 99% similarity. The chimera check and removal were performed using the database SILVA (Quast et al., 2013). BLAST (Altschul et al., 1990) was used for taxonomic assignment of 16S OTU reads using representative sequences against the SILVA database and for taxonomic assignment of 18S OTU reads. Three databases were used: the PR2 database (Guillou et al., 2013) and two marine protist databases (available at https://github.com/ramalok) from the BioMarKs project (Massana et al., 2015) and from collection of Sanger sequences from molecular surveys (Pernice et al., 2013). To complete the taxonomy assignment, OTUs with alignments of less than 200 bp, with coverage <60% and similarity <90% and e-values >0.00001 were discarded. Metazoan, Charophyta and nucleomorph OTUs were discarded prior to eukaryotic community composition analysis, while chloroplasts, mitochondria, or eukaryote OTUs were removed prior to bacterial community composition analysis. Assigned taxonomy and OTU tables were merged using R. The Vegan package in R was then used for subsampling the OTU read table at the lowest number of reads of 16S (7023 reads) and 18S (7805 reads) rRNA genes. The sequence data reported in this article have been deposited in NCBI GenBank under the accession number PRJNA821653.
Statistical analyses
Chl a and bacterial abundance data (24, 48 and 72 h) were analysed using a linear modelling technique. We used a linear mixed-effect model (LMM) on Chl a and a generalized mixed linear model (GLMM) on bacterial abundance (generalized Poisson distribution), using lme4 (Bates et al., 2015) and glmmTMB (Brooks et al., 2017) packages in R, respectively. These models included season, guano addition treatment, antibiotic treatment and time as fixed factors, microbial community as covariate and microcosm bag identity as a random factor. All two-way interactions were included in the models but excluded when non-significant (Engqvist, 2005). Significance was assessed using a type III Wald chi-squared test using the car R package (Fox & Weisberg, 2019).
The normal distribution of data was proved by a Kolmogorov–Smirnov and Shapiro Wilks test. Non-normal data were log transformed to obtain normality. The homogeneity of variance was tested using Levene test.
To estimate the effect of the different guano inputs on phytoplankton and bacterial growth, we used a post hoc a paired t-test to assess the effect of the treatment (1× and 5× compared to the corresponding control) at each sampling time (24, 48 and 72 h) of each experiment. Additionally, we calculated a response ratio (RR) by dividing time-averaged Chl a concentration or bacterial abundance of each treatment by the respective control. A value equal to 1 implies no response, a value, <1 implies a negative response and a value >1 implies a positive stimulation effect to guano input (growth). A paired t-test was used to compare the mean Chl a concentration or bacterial abundance in the control (pooling the data from all sampling times of the control treatment) against a mean Chl a concentration or bacterial abundance in the addition treatments (pooling the data from all sampling times of the additions treatments) to detect significant responses.
To estimate the strength and nature of the impact of bacteria on phytoplankton response to the different treatments, we calculated a phytoplankton–bacteria interaction index by dividing time-averaged Chl a concentration in the treatment without antibiotic (i.e. with an active bacterial community) by the corresponding mean Chl a concentration in the treatment with antibiotic (i.e. blocking bacterial community). A value equal to 1 implies a neutral interaction, a value <1 implies a negative impact of bacteria on phytoplankton growth, and a value >1 indicates a positive impact of bacteria on phytoplankton growth. A paired t-test was used to compare the mean Chl a concentration in the non-antibiotic treatment (pooling the data from all sampling times of non-antibiotic treatment) against a mean Chl a concentration in the antibiotic treatments (pooling the data from all sampling times of antibiotic treatments) to detect significant responses.
We additionally studied changes in the microbial structure and community composition by a principal component analysis (PCA) and their relation to the interaction index. For the PCA (Aitchison, 1983; Kucera & Malmgren, 1998), Phyloseq package was used to import the OTU data and filtered counts by 5% prevalence of samples (Callahan et al., 2016). Counts of the retained OTUs were agglomerated at order level, obtaining 31 taxa distributed in 7 bacterial phyla. Zeros were removed by using Bayesian-multiplicative treatment in zCompositions (cmultRepl function) (Palarea-Albaladejo & Martín-Fernández, 2015) and the centered log-ratio was performed using the PCAtools package (Blighe & Lun, 2019).
In order to identify which factors significantly explain the variability in the interaction index a linear model (LM) was used. In this model, we included the first three PCs (i.e. bacterial community) and the guano addition treatment as fixed effects and the interaction index as a dependent variable. Finally, we divide the variance of the interaction index variance explained by the PCs (i.e. bacterial community) and guano addition treatment using the variance partitioning function in the Vegan package (varpart function) (Oksanen et al., 2018).
For each variable, we considered significant differences when p < 0.05. The p value was standardized as in the study by Good (1982) to overcome the low number of replicates.
RESULTS
Guano derived from the high density of individuals in the Larus michahellis colony of the Cíes Islands can be mobilized to the coastal ocean affecting the functioning of microbial communities. To explore the response of microbial community to guano inputs, we carried out L. michahellis guano addition microcosm experiments with natural microbial communities from seawater samples collected near the Cíes Islands during three different seasons (March, June, and November). The water soluble fraction (WSF) of guano was prepared by stirring dried faeces with nutrient-depleted, 0.7 μm-filtered seawater. The experiments consisted in the addition of two realistic concentrations of the obtained WSF (1×, representing an input of ca. 1 μM of N; and 5×, representing an input of ca. 5 μM of N) with or without an antibiotic treatment (to block bacterial activity) to natural microbial plankton communities that were incubated during 72 h (Figure 1).
Initial conditions
Initial nitrate, ammonium and phosphate concentrations were different in the three seasons. Nitrate concentrations varied over one order of magnitude between March (0.8 ± 0.06 μM) and June (2.17 ± 0.19 μM) and two orders of magnitude between March and November (9.52 ± 0.26 μM) (Figure 2A–C). Ammonium and phosphate concentrations were higher in November (4.05 ± 0.09 and 0.69 ± 0.0003 μM, respectively) than in March (0.85 and 0.08 ± 0.01 μM, respectively) or June (0.47 ± 0.06 and 0.19 ± 0.007 μM, respectively) (Figure 2A–C). The highest initial DOC concentration was measured in June (83.3 μM) and the lowest in March (67.3 μM) (Figure 3A–C). Initial DON concentrations were 8 μM (March), 6.66 μM (June) and 17.51 μM (November) (Figure 3A–C). The highest bacterial abundances were found in March and the lowest in June (Figure 4A–C). In March and June, the prokaryotic community was dominated by Flavobacteriales and, to a lesser extent, by Rhodobacterales and Verrucomicrobia (only in March) (Figure 6A,C). Rhodobacterales was the dominant prokaryotic group in November, followed by Flavobacteriales and SAR11 (Figure 6C). Additionally, the lowest initial Chl a concentration was measured in November (0.67 ± 0.08 μg L−1) compared to the concentrations determined in March (3.77 ± 0.22 μg L−1) and June (3.32 ± 0.19 μg L−1) (Figure 4D–F). In March, the eukaryotic community was dominated by Dinophyceae, followed by Bacillariophyta and, to a lesser extent, by marine alveolate (MALV_I) and Rhizaria (Figure 6D). Bacillariophyta and Dinophyceae, followed by marine stramenopiles (MAST), MALV_I and Rhizaria dominated in June (Figure 6E). In addition, Mamiellophyceae, Bacillariophyta, Cryptophyceae, Dinophyceae and, to lesser extent, Ciliophora dominated in November (Figure 6F).
Bacterial and phytoplankton response to guano inputs
Nutrient concentration
In the control treatment, ammonium, nitrate, and phosphate were almost exhausted by the end of the experiments in March and June (Figure 2A,B). In November, ammonium was almost depleted, and phosphate was only slightly consumed at the end of the experiment. However, nitrate concentration in the control was similar to that in initial conditions (Figure 2C). DOC accumulated in the control in the three experiments (Figure 3A–C). In March and November, DON concentration decreased in the control treatment (Figure 3A,C), but accumulation of DON was observed in June (Figure 3B).
The guano addition treatments gave rise to increases in ammonium concentration. In 1× treatment, the addition of guano generated a 0.35 μM increase in ammonium concentration, while the 5× treatment, augmented ammonium concentration by one order of magnitude. Additionally, phosphate concentration increased very slightly in all guano addition treatments. The concentration of nitrate in the addition treatments remained as in the control treatment.
Nitrate and phosphate concentration decreased in addition treatments at the end of the incubation time in March, as in the control. Ammonium concentration also decreased in the control in the 1× addition treatment but in the 5× treatment, this concentration was 1.3 μM at the end of the March experiment (Figure 2A). The final DOC concentration was slightly higher in the 1× than in the 5× treatment in March (Figure 3A), whereas the final DON concentration was similar in 1× and 5× addition treatments.
In the guano addition treatments, phosphate and nitrate were almost exhausted at the end of the experiment in June (Figure 2B). Ammonium final concentration was similar in both addition treatments. Initial and final DOC concentrations were similar in 1× treatment but increased at the end of the experiment in the 5× treatment. Likewise, there was an increase in DON concentration in the 5× treatment (Figure 3B).
In November, the final concentrations of ammonium, nitrate, and phosphate were similar in the guano addition treatments. Furthermore, ammonium and phosphate consumption were similar in the different addition treatments, while increasing nitrate concentration was observed in 1× and 5× treatments (Figure 2C). DOC and DON final concentrations were similar in the addition treatments and both were slightly consumed at the end of the incubation time (Figure 3C).
Inorganic nutrient concentrations in the treatments with antibiotics were very similar in March and June and lower in November than in those without antibiotics (Figure S1).
Heterotrophic responses to guano additions
Bacterial abundance was significantly affected by the significantly interaction between guano addition treatment and time (Table 1). Thus, bacterial responses to guano inputs were different according to the time of the experiment (Figures 4A–C and 5A–C). In the experiment performed in March, bacterial abundance increased after 24 h of incubation and then decreased until the end of the experiment in the control treatment (Figure 4A). In the 1x guano treatment, a significant increment in bacterial abundance was observed at 24 h (t-test, p < 0.01) compared with the control treatment, while no significant changes were observed in the 5× treatment. Then, bacterial abundance decreased until the end of the experiment (Figure 4A). The time-averaged bacterial abundance response was significantly positive in 1× treatment (t-test, p < 0.05) (Figure 5A). No response was observed in the time-averaged bacterial abundance in 5× treatment (Figure 5A).
| Bacterial abundance | Chisq | df | p value | |
|---|---|---|---|---|
| Addition treatment | 6.50 | 2 | 0.038 | |
| Time | 5.49 | 2 | 0.064 | |
| Season | 26.34 | 2 | <0.001 | |
| Addition treatment × Time | 9.51 | 4 | 0.049 |
| Chl a | Source of variation | |||
|---|---|---|---|---|
| Addition treatment | 76.6 | 2 | <0.001 | |
| Time | 119.4 | 2 | <0.001 | |
| Season | 1044.5 | 2 | <0.001 | |
| Antibiotic treatment | 0.78 | 1 | 0.37 | |
| Addition treatment × Time | 53.8 | 4 | <0.001 | |
| Season × Time | 603.6 | 4 | <0.001 | |
| Season × Antibiotic treatment | 10.6 | 2 | 0.004 |
- The significant values are in bold.
In June, bacterial abundance increased after 24 h of incubation and decreased slightly until the end of the experiment in the control treatment. In the 1× treatment, bacterial abundance was significantly higher than in the control, at 24 (t-test, p < 0.01) and 48 h (t-test, p < 0.001) of incubation, then a decrease was observed (Figure 4B). Additionally, a significantly positive response was observed in time-averaged bacterial abundance response in the 1× treatment (t-test, p < 0.001) (Figure 5B). In the 5× treatment, a significant increase in bacterial abundance compared to the control was observed at 48 h of incubation (Figure 4B). However, no response was found in time-averaged bacterial abundance in the 5× treatment (Figure 5B).
In November, bacterial abundance in the control increased up to 48 h and then decreased until the end of the experiment. No significant differences were observed compared with the control in the 1× guano treatment (Figure 4C). Also, no response was observed in time-averaged bacterial abundance in 1× treatment (Figure 5C). In the 5× treatment, higher bacterial abundance compared with the control was observed at 24 h (t-test, p < 0.05) (Figure 4C). However, a non-significant response was observed in time-averaged bacterial abundance in 5× treatment (Figure 5C).
Autotrophic responses to guano additions
Chl a was significantly affected by the interactions between time and guano addition treatment (p > 0.001) and season (p > 0.001) (Table 1). Thus, three different patterns of phytoplankton responses were found in the experiments depending on the season (Figures 4D–F and 5D–F). In March, Chl a concentration increased 2-fold after 24 h of incubation in the control treatment and then progressively decreased until the end of the experiment. The 1× guano treatment did not cause significant differences in the Chl a concentration compared with the control treatment throughout the incubation in March, while a significant increase was observed in the 5× treatment in comparison with the control treatment at 48 (t-test, p < 0.01) and 72 h (t-test, p < 0.05) of incubation (Figure 4D). In addition, a statistically significant effect was observed in the time-averaged response of Chl a to the 5× guano addition (t-test, p < 0.05) (Figure 5D).
In June, Chl a concentration slightly increased after 24 h of incubation and then decreased until the end of the experiment in the control treatment. The same pattern was observed in the guano addition treatments (Figure 4E). In general, no significant differences were observed in Chl a concentration compared with the control treatment throughout the incubation time in the 1× guano treatment (Figure 4E). Consistently, no response was observed in the time-averaged Chl a in this treatment (Figure 5E). In the 5× treatment, Chl a significantly increased throughout the incubation (t-test, p < 0.05) (Figure 4B), and the time-averaged response of Chl a was significantly positive (t-test, p < 0.01) (Figure 5E).
In November, Chl a concentration in the control increased slightly throughout the incubation. A significant increase, compared with the control, was observed in Chl a concentration at the end of the experiment in both 1× and 5× treatments (t-test, p < 0.05) (Figure 4E). The time-averaged Chl a response was significantly positive in 1× (t-test, p < 0.05) and 5× (t-test, p < 0.001) addition treatments (Figure 5E).
Changes in prokaryotic community composition
In March, the final prokaryotic community was mainly dominated by Flavobacteriales and, to a lesser extent, by Verrucomicrobia and Rhodobacterales in the control treatment at the end of the incubation time. The 1× guano addition caused an increase in the relative abundance of Verrucomicrobia. Interestingly, the 5× guano addition drastically changed the composition of the prokaryotic community, that was mostly dominated (ca 70%) by Alteromonadales (Figure 6A).
In the June experiment, Flavobacteriales were the dominant order in all the treatments at the end of the incubations. However, the relative abundance of Sphingobacteriales showed an increase in the 5× guano treatment compared with the control treatments (Figure 6B).
Finally, the relative abundance of the different prokaryotic groups showed differences among all treatments in November. Flavobacteriales were relatively more abundant in the control compared with the 1× and 5× guano treatments. Rhodobacterales also showed an important decrease associated with guano additions. In contrast, the addition of guano (1× and 5×), caused an increase in the proportion of Alteromonadales, Vibrionales, Sphingobacteriales, Planctomycetes and Verrucomicrobia compared with the control. Interestingly, these patterns were more evident in the treatment with a greater addition of guano (5×) (Figure 6C).
Changes in the eukaryotic community composition
Phytoplankton taxa clearly dominated the final eukaryote community (>70%) in all treatments and experiments (Figure 6D–F). At the end of the March experiment, the eukaryotic community in the control was dominated by Dinophyceae and Bacillariophyta (Figure 6D). The addition of 1× guano caused an increase in the relative abundance of Pyramonadales compared with the control, while the 5× guano treatment generated a great increase in the relative abundance of Bacillariophyta and, to a lesser extent, of MAST, compared with the control (Figure 6D).
In the June experiment, the relative abundance of the different eukaryotic groups showed important differences among all treatments (Figure 6E). Bacillariophyta and Dinophyceae, and to a lesser extent MAST, dominated in the control treatment. In the 1× guano treatment, the proportion of Bacillariophyta showed an important increase and the relative abundance of Dinophyceae decreased compared with the control. In the 5× guano treatment the relative abundance of Bacillariophyta and MAST showed an increase compared with the control.
The proportion of Bacillariophyta considerably increased in the final control (C) compared with the initial control (T0) in November (Figure 6F). The guano addition treatments did not noticeably change the proportion of Bacillariophyta. However, Dinophyceae abundance showed a slight decrease in the guano treatments compared with the control.
Phytoplankton–bacteria interaction response to guano inputs
Chl a was not significantly affected by the addition of antibiotics (p = 0.37). However, the interaction between season and the antibiotic treatment on Chl a was significant (p > 0.001) (Table 1). Thus, although antibiotics did not directly affect phytoplankton, its growth appears to be affected by presence or absence of an active bacterial community, and the impact of bacteria differed among seasons. The interactions between phytoplankton and bacteria are expected to change seasonally, as different microbial communities are associated with different environmental (e.g. nutrient) conditions.
The magnitude and sign of the phytoplankton–bacteria interaction changed in response to guano inputs (Figure 7A–C). In March, the interaction index in the control and in 1x treatment was neutral (Figure 7A), shifting to positive in the 5× guano treatment (t-test, p < 0.01) (Figure 7A). Overall, the interaction indices were significantly negative in all treatments in June (Figure 7B) (t-test, p < 0.01). In November, the interaction index was significantly positive in the control (t-test, p < 0.01), shifting to neutral in the 1× and 5× guano treatments (Figure 7C).
To explore the impact that bacteria exert on phytoplankton community composition, we compared the phytoplankton communities evolving in the treatments without antibiotics (Figure 6D–F) with those developing in the treatments with antibiotics (Figure 6G–I). In the March experiment, the proportion of Bacillariophyta at the end of the experiment was higher in the control treated with antibiotics than in the control without antibiotics (Figure 6D,G). The relative abundance of Bacillariophyta and Dinophyceae were very similar in 1× treatment with and without antibiotics. In contrast, the proportion of Bacillariophyta after the 5× guano addition was lower with than without antibiotics (Figure 6D,G).
In June, the relative abundance of Bacillariophyta was lower in the final control without antibiotics than in the final control treated with antibiotics (Figure 6E,H). Additionally, the proportions of MAST and Dinophyceae were higher in control without antibiotics compared to antibiotic-treated control. The relative abundance of the different taxonomic groups was very similar in 1× guano addition treated or not with antibiotics. In contrast, the relative abundance of Dinophyceae was higher in 5× treatment treated with antibiotics than the 5× treatment without antibiotics. Moreover, Bacillariophyta and MAST proportions were higher in the 5× treatment without antibiotics than in the same addition treatment with antibiotics.
Finally, the relative abundance of Bacillariophyta at the end of the November experiment was higher in the control without antibiotics compared to the control treated with antibiotics (Figure 6F,I). The relative abundance of MAST and Rhizaria and MALV_I was higher in the final control with antibiotics than in final control without antibiotics. On the other hand, the proportions of different taxonomic groups were very similar in both addition treatments treated or not with antibiotics.
A principal component analysis based on the prokaryote community composition at the end of the experiments showed that the first two principal components (PC) explained 82.9% of the variation. The first principal component (PC1), explaining 50.7% of the variation, clearly separated March and June samples from those of November, while PC2, explaining 32.2% of the variation separated March samples with a neutral interaction index (control and 1× treatment) from the 5× treatment sample, where a positive impact of bacteria on phytoplankton was observed (Figure 8A). Additionally, PC3 explained 8.9% of the variation and separated June samples from March samples, and within March samples, those with a neutral interaction index (control and 1× treatment) from that with a positive interaction index (5× treatment) (Figure 8B). The results of a linear model performed with these data showed that PC1 (p = 0.02) and the guano addition treatment (p = 0.04) and to lesser extent by PC3 (p = 0.06), significantly explained the variability observed in the interaction index (Table S1). A variance partitioning analysis was carried out to assess the contribution of guano additions, PC1, PC2, and PC3 to the variance of the interaction index, showing that PC1, PC3, and the guano addition explained, respectively, 43%, 30%, and 21% of variation in the interaction index, highlighting the role of the prokaryotic community composition on the phytoplankton–bacteria interaction index (Figure 8C).
DISCUSSION
The guano collected in the area and used in our experiments was mainly composed of ammonium, phosphate, DOC, and DON (Figure 1). This result is consistent with other studies carried out in this area concluding that L. michahellis guano is mostly composed of ammonium, uric acid, proteins, and phosphate (Alba-González et al., 2022; Otero et al., 2015; Rial et al., 2016). Uric acid (Wright, 1995) can be mineralized into ammonium (Wainright et al., 1998); therefore, inputs of guano can further increase the availability of ammonium in the receiving seawater. In addition, seabird guano can have a high concentration of essential micronutrients (Fe, Mn, Zn, and Co) and other metals (Cu, Cd, Zn, and Pb) (Alba-González et al., 2022; Otero et al., 2018; Rial et al., 2016). Some of these nutrients and micronutrients can eventually stimulate the growth and activity of microplanktonic communities (Bristow et al., 2017; Teira et al., 2016).
In March, when nutrient concentration was low, addition of guano generated a significant increase in bacterial abundance and Chl a concentration. The positive responses of bacteria to the 1× addition treatment may partially explain the higher consumption of ammonium and DON in this treatment, compared with the control (Figures 2A, 3A and 5A). In addition, the significantly positive response observed in the autotrophs (Figure 5D) was accompanied by a high consumption of ammonium, phosphate, and nitrate (Figure 2A). These results suggest a nutritional limitation of the bacterial and autotrophic community that was alleviated by the entry of inorganic nutrients and organic substances in the guano. Previous studies on the impact of guano enrichment showed a positive effect on phytoplankton growth in different marine ecosystems (Lorrain et al., 2017; Shatova et al., 2016), including a recent one in the Ría de Vigo (Alba-González et al., 2022). Additionally, preceding studies carried out in the study area have concluded that under low nutrient ambient conditions, nutrient additions (controlled or natural) have the capacity to stimulate bacterial and phytoplankton growth (Martínez-García et al., 2010; Prieto et al., 2015; Teira et al., 2016; Teixeira et al., 2018).
Remarkably, in the experiments with the lowest initial nutrient concentration, the magnitude of the bacterial response was higher in the low-addition treatment (1× treatment). This could possibly be due to the increase in the concentration of some toxic elements that could inhibit bacterial growth. Excrements of L. michahellis on the Cíes Islands present high concentrations of trace metals including micronutrients (Zn and Cu) or toxic elements (Cd, Pb, and Cr) (Alba-González et al., 2022; De La Peña-Lastra et al., 2019). High concentrations of these metals may have an inhibitory effect on the activity of natural bacterial communities (Sunda & Gillespie, 1979). Another possible explanation for these responses may be that guano additions could be stimulating bacterivores such as the MAST group, whose relative abundance was higher in the 5× than in the 1× guano addition in the March and June experiments. (Figure 6D,E). Some MAST groups have been described as bacterial grazers playing an important role in the microbial food webs (Jürgens & Massana, 2008).
Addition of a large amount of guano (5× treatment) largely stimulated bacteria belonging to the order Alteromonadales in March (Figure 6A), coinciding with the large development and subsequent decay of a phytoplankton community dominated by Bacillariophyta (diatoms) (Figures 4D and 6D). Also, a seabird guano addition experiment conducted in Antarctic waters reported enhanced diatom biomass associated with this input (Shatova et al., 2017). Importantly, Alteromonadales have been described as copiotrophic bacteria, with high growth potential when the concentration of organic nutrients is high and are frequently associated with phytoplankton bloom decay phases (Taylor & Cunliffe, 2017; Zhang et al., 2015).
Prokaryotic community structure, and to a lesser extent guano treatments, played a significant role in determining the variability in the phytoplankton–bacteria interaction index (Figure 8C). The shift from a neutral to a positive phytoplankton–bacteria interaction index in March was associated with a drastic change in the prokaryotic community, suggesting that the guano inputs may lead to changes in bacterial composition that may be crucial for the modulation of this interaction (Figures 6A and 8A). The relative abundance of Alteromonadales coincided with the increase in the relative abundance of Bacillariophyta observed in the 5× guano treatment (Figures 6D and 7A), which suggests that these taxonomic groups are likely to be involved in this positive interaction. Furthermore, Bacillariophyta considerably reduced its abundance when bacterial activity was blocked, possibly indicating that these diatoms might require some secondary metabolite derived from bacterial metabolism (Figure 6G). Some studies reported that diatoms are strongly and recurrently associated with specific bacterial taxonomic groups such as Alteromonas (Sapp et al., 2007; Tada et al., 2017). It is also known that some diatoms are B-vitamin auxotrophs and can fulfil this requirement from mutualistic interactions with bacteria that are involved in vitamin biosynthetic pathways, such as bacteria of the order Alteromonadales (Amin et al., 2012; Sañudo-Wilhelmy et al., 2014). In this context, Alteromonadales showed a high expression of genes related to B12 synthesis in shelf waters off the Ría de Vigo during the winter, suggesting that this taxon can be an important producing source of this vitamin in the sampling area (Joglar et al., 2021). Our results suggest that guano inputs could favour mutualistic interactions between B12-auxotrothic diatoms and B12-producing Alteromonadales.
The high relative abundance of Flavobacteriales in the control and addition treatments in June suggests that guano enrichment does not alter the bacterial community composition in this experiment (Figures 6B and 8A). Some studies in this area have reported a high abundance of Flavobacteriales in spring (Gutiérrez-Barral et al., 2021; Teira et al., 2011). The negative phytoplankton–bacteria interactions observed in this experiment, particularly in the guano treatments, suggest that the relatively low ambient nutrient concentrations measured in this period may imply competition between microbial taxa for some compound not present in the guano. The fact that phosphate consumption was considerably higher when bacteria were active (Figure 2B) than when bacterial activity was blocked (Figure S1B), suggests a strong competition for this nutrient during this experiment. The fact that guano faeces are poor in phosphate relative to nitrogen forms could explain why the impact of bacteria becomes even more negative in the 1× and 5× treatments.
Even though guano additions did not significantly impact bacterial abundance in November (Figures 4C and 5C), progressive increases in the relative abundance of Alteromonadales, Sphingobacteriales, Planctomycetes and Verrucomicrobia were observed with increasing additions of guano (Figure 6C). The increase in the relative abundance of these groups in the addition treatments could be related to either the guano inputs or the concomitant phytoplankton growth. Alteromonadales, Sphingobacteriales and Verrucomicrobia have been previously seen to respond to nutrient additions in mesocosms experiments (Allers et al., 2007; Joglar et al., 2021; Park et al., 2020).
The relatively low Chl a increase measured in the control during the high-nutrient conditions sampled in November (Figure 4F), suggests that phytoplankton could be limited by light or other compounds apart from macronutrients. However, it is unlikely that phytoplankton growth was light-limited during the incubation, as the samples were exposed to surface irradiance levels during the incubations. On the other hand, guano additions gave rise to a more than 2-fold Chl a increase compared with the control (Figure 4F), even though the consumption of inorganic nutrients, mostly ammonium, was only slightly higher in the 5× guano treatment than in the control (Figure 2C). These results suggest that the phytoplankton community could be limited by some specific compounds, other than inorganic nutrients, present in the guano. It has been widely documented that many eukaryotic phytoplankton species can eventually limited by growth factors totally or partially derived from bacterial activity (Amin et al., 2012; Sañudo-Wilhelmy et al., 2014). The consumption of inorganic nutrients in the November experiment was lower in the samples treated with antibiotics than in those without antibiotics, indicating that either bacteria were also consuming inorganic nutrients or bacterial activity favoured phytoplankton growth. The interaction index was positive and high in the control but became neutral in the guano treatments (Figure 7C), while prokaryote community composition remained similar in the three treatments (Figure 8A,B), which is coherent with the hypothesis that a positive interaction between phytoplankton and bacteria mediated by the supply of bacterial-derived compounds, which could have been provided by the guano, occurred in this period. We suggest that guano could directly supply the same or a similar beneficial compound to that derived from bacterial activity. If this were the case, phytoplankton would not depend on bacteria to obtain this compound, as it would be provided by guano supply, and phytoplankton growth would be similar in the presence or absence of active bacteria, yielding an interaction index close to 1 in the guano treatment (Figure 7C).
Additionally, the final eukaryotic community in the control in November was dominated by Bacillariophyta (Figure 6C,F), suggesting that several bacterial taxa and Bacillariophyta taxa could be involved in the positive phytoplankton–bacteria interaction in the absence of guano, which is consistent with the frequently reported benefits provided by bacterial associations with diatoms (Amin et al., 2012). We observed that when bacterial activity was blocked with antibiotics, the proportion of Bacillariophyta in the control treatment drastically dropped (Figure 7I), which support the idea of diatoms as potential beneficiaries of the bacteria–phytoplankton interaction in this particular situation. As noted earlier, the addition of guano could neutralize this positive interaction due to the possible presence in the guano of some secondary metabolite or factor necessary for phytoplankton growth. Fe, hormones, or extracellular polysaccharide substances (EPSs) could be key compounds involved in this bacteria–diatom interaction (Rinta-Kanto et al., 2012; Teplitski & Rajamani, 2011). More research is needed to identify the nature of this interaction between phytoplankton and bacteria and to fully characterize the organic compounds present in seabird guano. Therefore, further studies are necessary to better understand the impact of seabirds on the dynamics of the microbial plankton compartment in other coastal areas.
CONCLUSIONS
Overall, L. michahellis faeces may introduce large amounts of potentially limiting inorganic and organic nutrients in nearby surface coastal waters that can stimulate the growth of microbial plankton. The continuous supply of this guano could not only generate positive responses in the bacterial and phytoplankton communities but also modify the composition of microplankton communities. The results of this study demonstrate, for the first time, that seabird guano enrichment might alter the magnitude and sign of phytoplankton–bacteria interactions, likely by introducing new nutrients or essential organic compounds into the receiving seawater. Seabird colonies can thus modulate the functioning of microbial communities, especially at the local level, subsequently affecting higher trophic levels.
AUTHOR CONTRIBUTIONS
Maider Justel-Díez: Formal analysis (lead); investigation (equal); methodology (equal); software (lead); visualization (equal); writing – original draft (lead); writing – review and editing (lead). Erick Delgadillo-Nuño: Investigation (supporting); software (supporting); supervision (supporting). Alberto Gutiérrez-Barral: Formal analysis (supporting); software (supporting); supervision (supporting). Paula García-Otero: Investigation (supporting). Isaac Alonso-Barciela: Investigation (supporting). Pablo Pereira-Villanueva: Investigation (supporting). Xosé Antón Álvarez-Salgado: Investigation (equal); supervision (equal). Alberto Velando: Formal analysis (equal); methodology (equal); supervision (equal). Eva Teira: Conceptualization (lead); funding acquisition (lead); project administration (lead); resources (lead); writing – review and editing (lead). Emilio Fernández: Conceptualization (lead); funding acquisition (lead); investigation (lead); methodology (lead); resources (lead); supervision (lead); writing – review and editing (lead).
ACKNOWLEDGEMENTS
This research was part of the INTERES project (CTM2017-83362-R) by the Spanish Ministry of Economy and Competitiveness. Funding for open access charge: Universidade de Vigo/CISUG. The authors thank Ciencias Mariñas de Toralla Station (ECIMAT) for the technical support during the experiments and María José Pazó and Vanesa Vieitez for inorganic nutrient and dissolved organic matter determinations at CSIC Instituto de Investigaciones Marinas. The authors thank Espe Broullón for the gull illustration. Maider Justel-Díez was supported by Xunta de Galicia fellowship and Erick Delgadillo-Nuño by an FPI fellowship from the Spanish Ministry of Economy and Competitiveness.
CONFLICT OF INTEREST STATEMENT
The authors declare no conflict of interest.
Open Research
DATA AVAILABILITY STATEMENT
The data deposited in NCBI under the accession number PRJNA821653.
