Introduction
Physical factors such as wind, marine currents and temperature have a major influence on species physiology and community processes (Kordas et al., Reference Kordas, Harley and O'Connor2011). Winds drive horizontal movement of the water, creating surface currents and vertical movements of upwelling and downwelling (Ottersen et al., Reference Ottersen, Nils, Hurrell, Stenseth, Ottersen, Hurrell and Belgrano2014). These movements are important in the dispersion of many species (Frith et al., Reference Frith, Leis and Goldman1986; Treml et al., Reference Treml, Halpin, Urban and Pratson2008), determining benthic community structure (Foster, Reference Foster2001) and macroalgal nutrient uptake (Neushul et al., Reference Neushul, Benson, Harger and Charters1992; Ho & Carpenter, Reference Ho and Carpenter2017). Moreover, changes in marine temperature affect reproduction, growth and survival of invertebrates (Levin & Creed, Reference Levin and Creed1986; Randall & Szmant, Reference Randall and Szmant2009), fishes (Pörtner et al., Reference Pörtner, Berdal, Blust, Brix, Colosimo, De Wachter, Giuliani, Johansen, Fischer, Knust, Lannig, Naevdal, Nedenes, Nyhammer, Sartoris, Serendero, Sirabella, Thorkildsen and Zakhartsev2001; Herbing, Reference Herbing2002), fleshy algae (Fong & Zedler, Reference Fong and Zedler1993; Andrews et al., Reference Andrews, Bennett and Wernberg2014) and rhodoliths (Horta et al., Reference Horta, Riul, Amado Filho, Gurgel, Berchez, Nunes, Scherner, Pereira, Lotufo, Peres, Sissini, Bastos, Rosa, Munoz, Martins, Gouvêa, Carvalho, Bergstrom, Schubert, Bahia, Rodrigues, Rörig, Barufi and Figueiredo2016). Rhodoliths are structures composed of red coralline algae that have an important ecological function, providing refuges for a wide diversity of flora and fauna (Bahia et al., Reference Bahia, Abrantes, Brasileiro, Pereira-Filho and Amado-Filho2010; Pascelli et al., Reference Pascelli, Riul, Riosmena-Rodríguez, Scherner, Nunes, Hall-Spencer, Oliveira and Horta2013; Amado-Filho et al., Reference Amado-Filho, Bahia, Pereira-Filho, Longo, Riosmena-Rodríguez, Nelson and Aguirre2017). Their morphology, physiology and biomass are influenced both by abiotic factors, such as hydrodynamics, temperature and light (Figueiredo et al., Reference Figueiredo, Kain and Norton2000; Steller et al., Reference Steller, Hernandez-Ayon, Riosmena-Rodriguez and Cabello-Pasini2007; Sañé et al., Reference Sañé, Chiocci, Basso and Martorelli2016), and by biotic factors, such as the presence of associated animals and macroalgae (Guillou et al., Reference Guillou, Grall and Connan2002; Legrand et al., Reference Legrand, Riera, Lutier, Coudret, Grall and Martin2017).
Free-living rhodoliths, also known as maerl, require environmental factors such as water motion (e.g. waves and currents) and bioturbation for their dispersion and rotation on the seafloor (Steller et al., Reference Steller, Riosmena-Rodríguez, Foster and Roberts2003). If water turbulence is low, the rhodolith bed does not develop because it gets covered by sedimentation and fouling (Foster, Reference Foster2001). If water motion is too high, the thalli become fragmented (Foster, Reference Foster2001; Hinojosa-Arango et al., Reference Hinojosa-Arango, Maggs and Johnson2009). In addition, these environmental drivers also play a key role in their growth and morphogenesis (Steller & Foster, Reference Steller and Foster1995; Pascelli et al., Reference Pascelli, Riul, Riosmena-Rodríguez, Scherner, Nunes, Hall-Spencer, Oliveira and Horta2013). Water motion (e.g. tidal currents and storms) may fragment rhodoliths, reducing their structural complexity and thus, habitat heterogeneity (McConnico et al., Reference McConnico, Carmona, Morales and Rodríguez2017). However, periodic rotation is necessary as it allows light to reach all sides of the thalli (Hinojosa-Arango et al., Reference Hinojosa-Arango, Maggs and Johnson2009).
In addition to waves and currents, seasonal environmental factors such as changes in seawater temperature (Kamenos & Law, Reference Kamenos and Law2010) and irradiance (Burdett et al., Reference Burdett, Keddie, MacArthur, McDowall, McLeich, Spielvogel, Hatton and Kamenos2014) can affect rhodolith beds due to their direct influence on photosynthesis and calcification. These physiological processes are intimately linked. During photosynthesis, consumption of CO2 increases the surrounding seawater pH and carbonate saturation state, favouring calcification (Johansen, Reference Johansen1981). In this way, environmental factors that affect photosynthesis may also affect calcification. Available information suggests that crustose coralline algae require relatively low irradiance levels for photosynthesis (Figueiredo et al., Reference Figueiredo, Kain and Norton2000), but long summer days increase photosynthesis and growth (Teichert & Freiwald, Reference Teichert and Freiwald2014). When seawater temperature increases due to seasonal fluctuations (i.e. 10–16°C in the Bay of Brest and 10–30°C in the Gulf of California), the photosynthesis also increases, which enhances the growth rate and biomass of free-living coralline algae (Martin et al., Reference Martin, Castets and Clavier2006; Steller et al., Reference Steller, Hernandez-Ayon, Riosmena-Rodriguez and Cabello-Pasini2007).
Changes in rhodolith density, biomass and morphology can influence the local diversity of flora and fauna since they provide microhabitats to many organisms (Foster et al., Reference Foster, Amado-Filho, Kamenos, Riosmena-Rodriguez and Steller2013; Neill et al., Reference Neill, Nelson, D'Archino, Leduc and Farr2015). Rhodoliths in the north-western Gulf of Mexico harbour microalgal life history stages residing within cells of Lithothamnion sp. (Krayesky-Self et al., Reference Krayesky-Self, Schmidt, Phung, Henry, Sauvage, Camacho and Fredericq2017; Fredericq et al., Reference Fredericq, Krayesky-Self, Sauvage, Richards, Kittle, Arakaki and Schmidt2018). These cells may act as marine biodiversity hotspots that function as seedbanks, i.e. temporary reservoirs for life history stages of ecologically important eukaryotic microalgae and macroalgae, or as refugia for ecosystem resilience (Krayesky-Self et al., Reference Krayesky-Self, Schmidt, Phung, Henry, Sauvage, Camacho and Fredericq2017; Fredericq et al., Reference Fredericq, Krayesky-Self, Sauvage, Richards, Kittle, Arakaki and Schmidt2018). Hence, it is necessary to investigate environmental factors that influence both rhodolith morphology and density in order to assess changes in local community structure.
Associated animals and macroalgae that have seasonal life cycles or traits can respond more quickly to environmental conditions than rhodoliths (Martin et al., Reference Martin, Clavier, Chauvaud and Thouzeau2007). Movement of rhodoliths represents a potential disturbance event to the community associated with the rhodolith bed. Algal biomass and motile fauna abundances are lower under more windy conditions, while the abundance of sessile species tends to increase with wind speed (Hinojosa-Arango et al., Reference Hinojosa-Arango, Maggs and Johnson2009). During turnover events, attached macroalgae and animals could be abraded or buried, whereas motile fauna that take shelter in the bed could be impacted by the abrasion of rhodoliths. Such disturbances can prevent the development of a stable community (Maughan & Barnes, Reference Maughan and Barnes2000; Hinojosa-Arango et al., Reference Hinojosa-Arango, Maggs and Johnson2009). In addition, changes in rhodolith density promoted by burial/exposure or displacement will directly influence habitat availability for associated flora and fauna. Therefore, when these rhodolith beds have higher densities, this trait will be associated with higher abundances of associated biodiversity (Pascelli et al., Reference Pascelli, Riul, Riosmena-Rodríguez, Scherner, Nunes, Hall-Spencer, Oliveira and Horta2013).
The presence of associated organisms determines the interspecific relations that influence the whole community structure of the rhodolith bed, since they can behave as partners or pests (Scherner et al., Reference Scherner, Riul, Bastos, Bouzon, Pagliosa, Blankensteyn, Oliveira and Horta2010). Besides trophic relation, flora and fauna can change micro-niche chemical conditions, which are factors that play a key role in ecophysiological processes (Stachowicz & Hay, Reference Stachowicz and Hay1996; Semesi et al., Reference Semesi, Beer and Björk2009; Legrand et al., Reference Legrand, Riera, Lutier, Coudret, Grall and Martin2017). Regarding macroalgae, their presence can either be positive or negative for rhodoliths. Turf algae benefit associated coralline algae since they elevate the pH of the surrounding medium, thus facilitating calcification (Short et al., Reference Short, Pedersen and Kendrick2015). Under high irradiances, the epiphytes can protect the crustose coralline algae from photoinhibition (Figueiredo et al., Reference Figueiredo, Kain and Norton2000). However, associated macroalgae can also shade rhodoliths and limit the availability of light and other environmental resources, such as CO2 and nutrients (Wahl, Reference Wahl2008), which cause a decrease in photosynthesis (Dodds, Reference Dodds1991). Furthermore, animals, such as polychaetes, crabs and molluscs, can help maintain rhodolith health by controlling epiphyte growth (Scherner et al., Reference Scherner, Riul, Bastos, Bouzon, Pagliosa, Blankensteyn, Oliveira and Horta2010; Legrand et al., Reference Legrand, Riera, Lutier, Coudret, Grall and Martin2017). For example, coralline algae are relatively rare in shallow areas with low rates of herbivory. These organisms rely on herbivory or low light levels to avoid being overgrown by competitively superior fleshy algae (Stachowicz & Hay, Reference Stachowicz and Hay1996). Some crabs clean the surface of coralline algae by consuming a wide array of epiphytic macroalgae that commonly co-occur with their host (Stachowicz & Hay, Reference Stachowicz and Hay1996). These macroalgae include chemically defended species of the genera Halimeda, Dictyota and Laurencia, which are usually avoided by herbivorous fishes (Stachowicz & Hay, Reference Stachowicz and Hay1996). Thus, small grazers such as amphipods, gastropods, molluscs and polychaetes remove grazer-susceptible epiphytes, thus allowing for the growth of less competitive, grazer-resistant species such as coralline red algae (Lubchenco, Reference Lubchenco1983; Hay et al., Reference Hay, Renaud and Fenical1988; Steneck et al., Reference Steneck, Hacker and Dethier1991; Stachowicz & Whitlatch, Reference Stachowicz and Whitlatch2005; Scherner et al., Reference Scherner, Riul, Bastos, Bouzon, Pagliosa, Blankensteyn, Oliveira and Horta2010).
Since the structure and resilience of rhodolith beds are affected by environmental and biological factors, the aim of this study was to characterize the communities associated with rhodolith beds over two years and analyse the abiotic factors that could cause changes in rhodolith physiology and morphology. The biomass of rhodoliths and associated species were analysed during this period. In situ experiments were conducted in order to measure the primary production and calcification of rhodolith communities, focusing on the dominant species. In this experiment, we investigate how photosynthesis and calcification rates of the dominant rhodolith-forming species Lithothamnion crispatum Hauck respond to the presence of the dominant fleshy alga Padina gymnospora (Kützing) Sonder. Our hypotheses are: (1) the biomass of rhodoliths and associated communities vary according to environmental changes during the study period; (2) the growth of P. gymnospora, which is a result of changes in environmental factors and trophic relations, alters the physiological responses of L. crispatum.
Materials and methods
Study site
The experiment was conducted in a rhodolith bed at Rancho Norte (27°17′S 48°22′W), Brazil, which is part of the 176 km2 Marine Protected Area (MPA) called Arvoredo Marine Biological Reserve (Rebio Arvoredo), created in March 1990. The rhodolith bed extends over a sandy bottom on the north-western shore of Arvoredo Island, covering an area of ~100,000 m2. The isobaths are between 7 m and 20 m. Average annual seawater temperature is around 22°C. Since Rebio Arvoredo is only 10 km away from the coastal zone, depending on winds and currents, the continental runoff from urban and industrial contaminants of the metropolis of Florianópolis can reach the area (Freire et al., Reference Freire, Varela, Fonseca, Menezes, Fest, Obata, Gorri, Franco, Machado, Barros, Molessari, Madureira, Coelho, Carvalho, Pereira, Segal, Freire, Lindner, Krajewski and Soldateli2017). The area is influenced by the Brazilian Current (BC), which carries warm and salty tropical water from the low latitudes, and by an intense seasonal mixture of coastal, shelf and open ocean water masses from the Malvinas Current, which carries cooler and less saline water derived from the Antarctic Circumpolar Current (Matano et al., Reference Matano, Palma and Piola2010; Orselli et al., Reference Orselli, Kerr, Ito, Tavano, Mendes and Garcia2018). These two opposing currents converge to form the Brazil–Malvinas Confluence Zone (Matano et al., Reference Matano, Palma and Piola2010). The South Atlantic Central Water (SACW) formed in this region is transported to the south of Brazil by the BC under the tropical water (Freire et al., Reference Freire, Varela, Fonseca, Menezes, Fest, Obata, Gorri, Franco, Machado, Barros, Molessari, Madureira, Coelho, Carvalho, Pereira, Segal, Freire, Lindner, Krajewski and Soldateli2017). Water coming from the Brazil–Malvinas Confluence also mixes with the low salinity plume from Rio de La Plata, the Patos-Mirim Lagoon and other local sources of continental runoff (Möller et al., Reference Möller, Piola, Freitas and Campos2008; Strub et al., Reference Strub, James, Combes, Matano, Piola, Palma, Saraceno, Guerrero, Fenco and Ruiz-Etcheverry2015). The influence of this plume on the region is seasonal, where in the winter, the south-westerly winds force the plume to move to lower latitudes (28°S), and in the summer, north-easterly winds lead the plume poleward (Möller et al., Reference Möller, Piola, Freitas and Campos2008). This seasonality results in a dynamic and complex environment (Eichler et al., Reference Eichler, Sen Gupta, Eichler, Braga and Campos2008; Paquette et al., Reference Paquette, Bonetti, Bitencourt and Bonetti2016).
The rhodolith bed of Rancho Norte represents the southernmost limit of this habitat in the western Atlantic (Gherardi, Reference Gherardi2004; Pascelli et al., Reference Pascelli, Riul, Riosmena-Rodríguez, Scherner, Nunes, Hall-Spencer, Oliveira and Horta2013). This bed provides ecosystem services, such as habitat for epiphytic algae (Horta et al., Reference Horta, Salles, Bouzon, Scherner, Cabral and Bouzon2008), refuge and food sources for a faunal community formed by zoanthids (Zoanthus sp., Anthozoa, Hexacorallia), ascidians, polychaetes, crabs, bivalves (Rocha et al., Reference Rocha, Metri and Omuro2006; Scherner et al., Reference Scherner, Riul, Bastos, Bouzon, Pagliosa, Blankensteyn, Oliveira and Horta2010), ophiuroids, bryozoans, sponges and starfishes (Gherardi, Reference Gherardi2004).
Environmental conditions
To investigate the variation of monthly seawater temperature between 2014 and 2016, data loggers (HOBO® Data Logger UA-002) were installed at 10 m depth, in the rhodolith bed. These data loggers were periodically changed to avoid biofouling and the temperature was recorded at an interval of 20 min. The average of a month was considered in analyses. Measurements of photosynthetically active radiation (PAR; μmol s−1 m−2) were performed at 10 m depth where incubations took place using a LI-COR LI-1400 coupled to a hemispherical sensor.
The average wind data from six months prior to summer and winter sampling efforts were considered in the analyses of the influence of predominant wind direction (i.e. N, NE, E, SE, S, SW, W, NW (%)) and speed (m s−1) on the rhodolith bed. This interval enables investigation of the history of wind changes before each collection, since the rhodolith bed's response to environmental change could be delayed. Hourly data of wind direction and speed were obtained from the online database of the National Institute of Meteorology – INMET (INMET, accessed online August 2017).
Community structure
For community structure analysis, sampling was conducted using quadrats (25 × 25 cm) that were placed randomly over the Rancho Norte rhodolith bed. All the organisms in each square were stored in plastic bags and transported to the laboratory. Quadrats were collected during the summer, in February 2015 (N = 9) and February 2016 (N = 14), late spring, in November 2016 (N = 9), and in winter, in September 2015 (N = 6) and June 2016 (N = 5). Rhodolith and epiphyte macroalgae were separated from animals and sorted by species (the identification following Woelkerling, Reference Woelkerling1988; Littler & Littler, Reference Littler and Littler2000; Sissini et al., Reference Sissini, Oliveira, Gabrielson, Robinson, Okolodkov, Riosmena-Rodríguez and Horta2014). For rhodoliths composed of more than one species, the species covering the majority of the surface (more than 50%) was considered. Macroinvertebrates were separated (by phylum) and weighed (in g, fresh weight). Epiphytes and rhodoliths were separated by species, dried at 60°C and then weighed (precision of ±0.001 g). To detect which rhodolith species could more easily be transported by currents, we weighed 50 unbroken samples from each species.
Since epiphytes only occurred during one summer (February/2015), only the herbivores from summer/late spring samples were analysed to compare the differences in abiotic factors that could have favoured such occurrences. However, to analyse the seasonal variations in biomass of rhodoliths, all sampling efforts were considered.
In this study, underwater visual censuses were applied to collect and quantify fish population density data (UVC: 20 × 2 m strip transects = 40 m2). The procedure required a diver to swim along a transect 1 m above the substrate. While unrolling a measuring tape, the diver recorded all fish by species and calculated their size by placing them into 5 cm categories (Floeter et al., Reference Floeter, Krohling, Gasparini, Ferreira and Zalmon2007). All sampling campaigns were conducted in the morning, by the same diver, during summers.
The biomasses of fishes were calculated using the following equation (1) (published weight-length relationships):
$$W = ax{\rm T}{\rm L}^b$$in which W is the total wet weight in grams, a and b are species-specific parameters of the relationship, and TL is the total length in cm (Anderson et al., Reference Anderson, Bonaldo, Barneche, Hackradt, Félix-Hackradt, García-Charton and Floeter2014; Froese & Pauly, Reference Froese and Pauly2016).
Experimental design
Primary production, respiration and calcification were measured on the rhodolith species L. crispatum, and the epiphyte P. gymnospora alone and associated with L. crispatum. In situ physiological experiment was conducted during one summer (3–4 February 2015) at a depth of 10 m. During this period, we conducted ~2 h incubations inside closed chambers and analysed changes in dissolved oxygen concentrations (DO) and total alkalinity (TA). The following combinations were incubated (N = 5 at daylight and N = 3 at night): (1) – two specimens of L. crispatum alone (64.89 ± 20.20 g), (2) – two specimens of P. gymnospora alone (2.67 ± 0.79 g) and (3) – one specimen of L. crispatum and one specimen of P. gymnospora together (66.79 ± 30.68 g). These different combinations were placed inside transparent nylon chambers, which did not allow for gas exchange and did not influence the light quality. All chambers were filled with ~2 l of bottom seawater and sealed with a holder made of PVC tube. They were tied and suspended with a rope, just above the bottom, subjecting them to gentle movement from the current. This enabled circulation inside the chambers, thus reducing the formation of large diffusion boundary layers around the organisms and providing a homogeneous distribution of nutrients (Hurd, Reference Hurd2000). All combinations were incubated once at each of three different natural irradiances (Time 1, 2 and 3) and one time at night (Time 4), always using different samples (during ~2 h). PAR (μmol s−1 m−2) was determined next to the chambers using a LI-COR hemispherical sensor throughout each incubation. Averaged PAR values (3 points) were considered for each incubation period for the productivity analyses. After the incubation, the chambers were brought to the surface to measure the parameters used to calculate productivity and calcification. The volume of seawater was measured in each chamber with a graduated beaker.
Productivity and respiration
For dissolved oxygen concentration analyses, five samples of 12 ml of seawater were taken from each chamber at the beginning and at the end of the incubation. The mean value per chamber was used in the calculations. The dissolved oxygen was measured by the Winkler method, modified by Labasque et al. (Reference Labasque, Chaumery, Aminot and Kergoat2004). This method enables a repeatability of 0.45% (30 samples) and a reproducibility of 0.73% near 250 μmol kg−1 of O2. Although we incubated five chambers per combination at daylight, we only used four in the calculation of primary production after detecting outlier values probably caused by bubbles produced in the sample collection. In the laboratory, the organisms were dried for 48 h at 60°C and then weighed. The DO values were used to calculate the Net production (NP), Respiration (R) and Gross Production (GP) rates (in μmol O2 cm−2 h−1) according to Noisette et al. (Reference Noisette, Duong, Six, Davoult and Martin2013) and the following equations (2) to (4):
$${\rm NP} = \displaystyle{{\Delta {\rm O}_2\lpar {{\rm \mu mol}\;{\rm l}^{{-}1}} \rpar \lpar {{\rm light}} \rpar xV} \over {\Delta txA}}$$
$$R = \displaystyle{{\Delta {\rm O}_2\lpar {{\rm \mu mol}\;{\rm l}^{{-}1}} \rpar \lpar {{\rm dark}} \rpar xV} \over {\Delta txA}}$$
$${\rm GP} = {\rm NP}-R$$where: ΔO2 = Difference between final and initial oxygen (μmol l−1), A = surface area of algae (cm2), V = chamber volume (L) and Δt = incubation time (h). Noisette et al. (Reference Noisette, Duong, Six, Davoult and Martin2013) considered the dry weight in the calculation. However, the surface area of L. crispatum and P. gymnospora was considered separately instead of the dry weight in order to make the physiological results of both more comparable.
Calcification
To analyse the calcification rates of the organisms, we measured changes in seawater total alkalinity within chambers during the incubations. At the beginning and at the end of the incubations, two samples of seawater were stored in 40 ml vials. Samples were poisoned with mercuric chloride (0.02%) and the analyses were conducted according to the open-cell protocol described by Dickson et al. (Reference Dickson, Sabine and Christian2007) at the LEOC laboratory in the Institute of Oceanography at the Federal University of Rio Grande (FURG). Regular analyses of Certified Reference Material (CRM) Batch 149, obtained from the Scripps Institution of Oceanography, were carried out for quality control purposes (Dickson et al., Reference Dickson, Afghan and Anderson2003). The accuracy of the total alkalinity measurements was set using the CRM by applying a correction factor to the measured values that was based on the nominal CRM values. The analytical precision of the total alkalinity measurements was investigated daily through replicate analyses of a single sample and was determined to be ± 1.0 μmol kg–1. The pH (pH AT-315 Alfakit, resolution 0.01 and precision ±0.01%) was measured at the beginning and end of each incubation. Before use, the pH meter was calibrated to buffer solutions (pH 4.00, 7.00 and 10.00 – NBC scale) according to commercial protocols. Light and dark calcification rates were estimated using the alkalinity anomaly technique (Smith & Key, Reference Smith and Key1975), where for each mole of CaCO3 precipitated, total alkalinity (At) decreases by two equivalents: Ca2+ + 2HCO3− → CaCO3 + H2O + CO2(aq) (Johansen, Reference Johansen1981; Wolf-Gladrow et al., Reference Wolf-Gladrow, Zeebe, Klaas, Kortzinger and Dickson2007). The difference between initial and final alkalinity of the incubation (ΔAt) was considered to calculate calcification rates (G, μmol CaCO3 cm−2 h−1), following equations (5) and (6) below (Noisette et al., Reference Noisette, Duong, Six, Davoult and Martin2013):
$$G_{{\rm biomass}} = \displaystyle{{-\lpar {\Delta At} \rpar xV} \over {2x\Delta tx{\rm DW}}}$$
$$G_{{\rm area}} = \displaystyle{{-\lpar {\Delta At} \rpar xV} \over {2x\Delta txA}}$$For subsequent analyses, calcification normalized to biomass (G biomass) was used to quantify rhodolith production of g CaCO3 on the day of the experiment, and calcification normalized to surface area (G area) was used to compare the metabolic rates of the rhodoliths and P. gymnospora occurring alone and together.
Algae surface area
The surface area of Padina gymnospora was calculated from photos that were analysed using Image J software and expressed in cm2. The samples were placed individually under a white background with a scale and then photographed at a distance of 50 cm. A methodology modified from Hoegh-Guldberg (Reference Hoegh-Guldberg1988), was used for the rhodolith surface area estimation. Dried L. crispatum samples (N = 19) from the Rancho Norte rhodolith bed, with varied sizes, were weighed and then coated in a commercial blank dye (composed of resin and water). The first coat sealed the surface, reducing the porosity. After 20 min, they received another layer and were re-weighed before and after the second coat. The conversion from weight increase to the surface area was done by doing a calibration with four expanded polystyrene cubes of known area (13.6–61.45 cm2). The relationship between the weight of the second dye coat and the surface area of the cubes was used to calculate the three-dimensional surface area of 19 dried samples (y = 183.04x–0.458, r 2 = 0.9953). Then, the regression relationship between the surface area and the weight of the samples was used to calculate the surface area of rhodoliths used in the experiment (y = 2.5108x + 9.9264, r 2 = 0.9337).
Rhodolith growth
The growth rate of rhodoliths was determined in situ. Random rhodoliths (N = 30) from the Arvoredo bed at 10 m depth were collected in May 2014 and taken to the laboratory and stained with an aerated 0.025% (w/v) alizarin red seawater solution for 24 h (Blake & Maggs, Reference Blake and Maggs2003). The specimens were tagged with nylon lines and plastic beads. Afterwards, they were taken back to the field and were recollected in November 2015. To estimate the length of the growth layer, the rhodoliths were cross-sectioned radially using a circular rock saw (Caragnano et al., Reference Caragnano, Basso and Rodondi2016) and then visualized under a light microscope (Leica S8AP0) (Supplementary Figure S1). Some rhodoliths with undetectable alizarin marks were left out. From each sample (N = 11), at least 20 measurements were taken. The measurements were equally distributed around the circumference.
The amount of CaCO3 fixed in the rhodolith bed (g m−2 year−1) was obtained by considering the growth lengths, the radii, and the dry weights of the rhodoliths of each quadrat from the biomass sampling, as described in Amado-Filho et al. (Reference Amado-Filho, Moura, Bastos, Salgado, Sumida, Guth, Francini-Filho, Pereira-Filho, Abrantes, Brasileiro, Bahia, Leal, Kaufman, Kleypas, Farina and Thompson2012a). We made one modification; the weight of the rhodoliths was used instead of their volume. The rhodoliths were weighed individually and the rhodolith radii were measured according to the shortest, intermediate and largest radii. The proportion of the total rhodolith weight (Wr) relative to the growth layer weight (W l) was calculated from the ratio of the growth length (estimated to be 0.319 ± 0.226 mm year−1, based on measurements of rhodoliths that were left in the field) to the mean radii (R r), as described in equation (7):
$$W_l = W_rx\displaystyle{{0.319} \over {R_r}}$$The amount of CaCO3 produced by the rhodolith bed per m2 (0.0625 m2 per quadrat) in g m−2 year−1 (CaCO3pr) considered the sum of the W l of all the rhodoliths of each quadrat (SWl), according to the equation (8):
$${\rm CaC}{\rm O}_{3pr} = \displaystyle{{{\rm S}{\rm W}_l} \over {0.0625}}$$Using the calcification data measured in situ (G biomass) and normalized to weight, we calculated diel calcification (DG), expressed as μmol CaCO3 g−1 DW day−1 (considering 13:11 h Light: Dark cycle), according to equation (9):
$${\rm DG} = G_{{\rm light}} \times 13 + G_{{\rm dark}} \times 11$$G light takes the three daylight incubations into account. The DG value (converted from μmol to g) was used to calculate the CaCO3 production of L. crispatum (R CaCO3pr) of each quadrat individually, expressed in g CaCO3 day−1 according to equation (10):
$$R\;{\rm CaC}{\rm O}_{3pr} = \displaystyle{{{\rm DG} \times W_r} \over {0.0625}}$$The average R CaCO3pr of the individual quadrats (Total R CaCO3pr) was used to calculate the CaCO3 production of L. crispatum per area during in situ incubations.
Data analyses
The photosynthetic, respiratory and calcification rates of the macroalgae were analysed using parametric statistics. After evaluating the normality (Shapiro–Wilk test) and homoscedasticity (Levene's test) of the data, a two-way ANOVA was performed to test the significant differences of photosynthesis and calcification data under different light levels and the effects of individual vs combined species. For the respiration data, one-way ANOVAs were performed using incubation data at dark. The Newman–Keuls post hoc test was applied when significant differences were observed (P < 0.05). ANOVA assumptions were not met for the total biomass of rhodoliths and fauna, so these data were analysed using a Kruskal–Wallis test followed by pairwise multiple comparisons. The correlation between the total biomass of rhodoliths and the mean values of wind direction/velocity from six months prior to each collection was analysed with a Spearman correlation and was considered significant when P < 0.05. The analyses were performed in Statistica 13.0.
Linear Models (LM) and subsequent Analyses of Variance (ANOVA) were used to test the effect of season on the biomass of rhodolith species and the effect of time on the fish density inside the Arvoredo MPA. Total densities and species biomass were used as dependent variables, whereas time (years) was used as a factor (Underwood, Reference Underwood1981; Chatfield, Reference Chatfield1989; Snedecor & Cochran, Reference Snedecor and Cochran1989). When significant differences were found, the Tukey HSD post-hoc test was used to verify sources of variation. Assumptions of normality and homoscedasticity were assessed with Kolmogorov–Smirnov/Lilliefors and Bartlett's tests (Underwood, Reference Underwood1981; Snedecor & Cochran, Reference Snedecor and Cochran1989; Zar, Reference Zar1999). Analyses were run in the R environment package ‘Agricolae’ (de Mendiburu, Reference de Mendiburu2013).
Results
Environmental conditions
During the six months before each of the summer/late spring collections, the north winds were predominant. The highest frequency was during the semester before February 2015 (33.19%), followed by February 2016 (32.42%) and November 2016 (25.89%) (Figure 1). However, the speed of the wind six months before February 2015 was the lowest among the summer collections (Figure 2). Regarding the winter collections, the north wind was higher during the semester before September 2015 (30.61%), whereas in the months before July 2016, the south-east wind prevailed (23.82%).
Fig. 1. Frequency (%) of wind direction in the six months prior to each month of collection. Data: INMET (Brazilian Meteorology Institute – Florianópolis/São José/A806 Station).
Fig. 2. Mean wind speed (m s−1) of six months prior to each month of collection. Data: INMET (Brazilian Meteorology Institute – Florianópolis/São José/A806 Station).
The values of monthly seawater temperature from May 2014 to December 2016 varied between 16.6°C in winter and 26.1°C in summer (Figure 3). The seawater temperature at the rhodolith bed during the physiological experiment varied between 25.6 and 26.9°C. The mean PAR values measured at each incubation time were, in order of execution, 35 μmol photons m−2 s−1 at Time 1, 472 (± 20.64) μmol photons m−2 s−1 at Time 2, 119 (±0.68) μmol photons m−2 s−1 at Time 3 and 0 μmol photons m−2 s−1 at Time 4. Salinity varied between 33 and 34.
Fig. 3. Variation of mean (±SE) monthly seawater temperature at 10 m depth from May 2014 to December 2016 at the Marine Protected Area (MPA) of Arvoredo, Brazil. The black arrow indicates the month of in situ physiological experiment.
Community structure
Four species of rhodoliths were found: Lithothamnion crispatum, Lithophyllum atlanticum Vieira-Pinto, M.C.Oliveira & P.A.Horta, Mesophyllum erubescens (Foslie) Me.Lemoine and Lithophyllum margaritae (Hariot) Heydrich (Figure 4). The total biomass of rhodoliths was significantly different between seasons (H: 24.51; P < 0.01, Supplementary Table S1, for all statistical results), with the highest value in winter, in September 2015, and the lowest in February 2016 and November 2016 (Figure 5). Lithothamnion crispatum was the most abundant species in the summer (F: 3.32; P < 0.01) at all collections, followed by L. atlanticum (Figure 6). However, the biomass of L. atlanticum increased in the winter (F: 3.32; P < 0.01). Lithophyllum margaritae only occurred in low abundance in February 2015 and July 2016. Lithophyllum atlanticum and M. erubescens specimens had similar weights and were significantly heavier than L. crispatum (Figure 7). The growth rates of rhodoliths from Arvoredo were 0.035–0.566 mm year−1. The mean annual rate of net CaCO3 production was estimated to be 105.97 (±49.26) g m−2 year−1 and ranged from 44.90 to 193.94 g m−2 year−1.
Fig. 4. Morphology of rhodoliths collected at Arvoredo (MPA), Brazil. Letters A–B: Lithothamnion crispatum; C–D: Lithophyllum atlanticum; E–F: Mesophyllum erubescens; G–H: Lithophyllum margaritae. Scale bar: 1 cm.
Fig. 5. Total biomass of rhodolith in g m−2 in summer (black bars, February 2015 (N = 9) and 2016 (N = 15)), late spring (grey bar, November 2016 (N = 14)) and winter (striped bars, September 2015 (N = 6) and July 2016 (N = 5)) at Arvoredo (MPA), Brazil. Letters indicate results of Kruskal–Wallis multiple comparisons (P < 0.01).
Fig. 6. Mean biomass (±SE) of rhodolith species in summer/late spring months (black and grey bars) and winter (white bars) at Arvoredo (MPA), Brazil. Letters indicate the results of Tukey test.
Fig. 7. Mean weight (±SE) per individual (N = 50) of three rhodolith species from Arvoredo MPA, Brazil. Letters indicate results of Kruskal–Wallis multiple comparisons (P < 0.01).
Spearman test results indicate negative correlations between total rhodolith biomass and wind speed from all directions except east (Table 1). Biomass of L. crispatum was negatively correlated with wind speed from all directions and with the frequency of wind from south quadrants (SE, S and SW). However, it was positively correlated with winds from the north (N and NE). There were no significant correlations between L. atlanticum, M. erubescens and L. margaritae and wind speed for the majority of response variables (P > 0.05).
Table 1. Summary of Spearman correlation results (r) between wind direction and speed and the biomass of rhodoliths. Bold numbers indicate significant values (P < 0.05)
In relation to associated organisms, epiphytic macroalgae occurred only in January 2015, with Padina gymnospora presenting major cover with 36.44 (±14.42) g m−2, followed by Amphiroa fragilissima (Linnaeus) J.V.Lamouroux with 3.90 (±1.54) g m−2 and Canistrocarpus cervicornis (Kützing) De Paula & De Clerck with 1.57 (±0.83) g m−2 (Supplementary Table S2). Invertebrates increased from February 2015 to November 2016 (Figure 8). However, only the biomass of Annelida (H = 10.22; P < 0.01) and Mollusca (H = 10.21; P < 0.01) presented significant differences.
Fig. 8. Mean biomass (±SE) of Annelida, Mollusca and Arthropoda in gFW/gDW in February 2015 (N = 9), February 2016 (N = 3) and November 2016 (N = 3) at Arvoredo MPA, Brazil. Letters indicate results of Kruskal–Wallis multiple comparisons (P < 0.01).
Grouper (macrocarnivores) density varied significantly throughout the survey. Densities increased from 2015 to 2016 (P < 0.05, Tukey HSD). No significant difference was found on their biomass (P > 0.05, Tukey HSD) (Figure 9).
Fig. 9. Arvoredo MPA densities and biomass of Carnivores (A) and Invertebrate feeders (B).
Grunts (macroinvertebrate feeders) densities showed significant differences considering the factor time (P < 0.05, Tukey HSD). Otherwise, biomass did not show significant differences (P > 0.05, Tukey HSD). Considering both population descriptors (density and biomass), populations of grunts show a modest decrease throughout the survey (from 2014 to 2016) (Figure 9).
Productivity and respiration
The mean rates (±SD) of gross production (GP) in L. crispatum varied between 0.112 (±0.056) and 0.554 (±0.149) μmol O2 cm−2 h−1, while in P. gymnospora they ranged from 0.078 (±0.033) to 0.324 (±0.017) μmol O2 cm−2 h−1 (Figure 10). Significant differences were detected in GP and respiration between species and times of incubation (F: 9.57; P < 0.01, Supplementary Table S1). Lithothamnion crispatum alone had the highest values of GP and respiration at all light levels. However, L. crispatum and P. gymnospora together (L + P) had the lowest values of GP, reaching only 0.214 (±0.077) μmol O2 cm−2 h−1.
Fig. 10. Mean (±SE) gross production (N = 4) and respiration (N = 3) of L. crispatum, P. gymnospora and P. gymnospora and L. crispatum together (L + P) in summer at Arvoredo MPA, Brazil. Shades of grey indicates mean PAR values. Letters indicate results of Newman–Keuls post-hoc test (P < 0.01).
Calcification
All light net calcification rates were positive and all dark net calcification rates were negative in all treatments (Figure 11). Significant differences in light and dark net calcification rates were detected in all species and at all times of incubation (Supplementary Table S1). Light net calcification rates of rhodoliths alone were the highest values at all light levels, reaching 0.293 (±0.014) μmol CaCO3 cm−2 h−1, while P. gymnospora reached only 0.063 (±0.014) μmol CaCO3 cm−2 h−1 (Figure 11). The net calcification rates of L. crispatum and P. gymnospora together were similar to the values for P. gymnospora alone. Net dissolution in the dark was significantly higher in L. crispatum, reaching 0.071 (±0.005) μmol CaCO3 cm−2 h−1. The diel net CaCO3 precipitation of L. crispatum was 8.7 ± 2.9 g CaCO3 m−2 d−1.
Fig. 11. Mean (±SE) light (N = 5) and dark (N = 3) calcification of L. crispatum, P. gymnospora and P. gymnospora and L. crispatum together (P + L) in summer at Arvoredo MPA, Brazil. Shades of grey indicates mean PAR values. Letters indicate results of Newman post-hoc test (P < 0.01).
Discussion
While variations in hydrodynamic conditions seem to cause the change in seasonal abundance and composition of rhodoliths in the Arvoredo rhodolith bed, the genesis of associated benthic community variability seems to be more complex. Integrated analysis indicates changes in the abundance of macroalgae and animals which may have an impact on community trophic structure. Trophic cascade effects caused a top-down control, structuring a benthic community with a reduced abundance of fleshy macroalgae, as was observed in different reef systems (Littler et al., Reference Littler, Littler and Taylor1995; Stachowicz & Hay, Reference Stachowicz and Hay1996; Scherner et al., Reference Scherner, Riul, Bastos, Bouzon, Pagliosa, Blankensteyn, Oliveira and Horta2010). Suppression of fleshy algae canopy cover seems to increase primary production potential, which contradicts the functional group theory (Steneck & Dethier, Reference Steneck and Dethier1994).
Community structure
The rhodolith bed at Arvoredo represents the southernmost limit of rhodolith bed distribution in the tropical South Atlantic. Lithothamnion crispatum, the most abundant rhodolith species in summer, is frequently cited for Brazilian beds (Amado-Filho et al., Reference Amado-Filho, Pereira-Filho Guilherme, Bahia, Abrantes, Veras and Matheus2012b, Reference Amado-Filho, Bahia, Pereira-Filho, Longo, Riosmena-Rodríguez, Nelson and Aguirre2017; Pascelli et al., Reference Pascelli, Riul, Riosmena-Rodríguez, Scherner, Nunes, Hall-Spencer, Oliveira and Horta2013; Cavalcanti et al., Reference Cavalcanti, Gregoracci, Dos Santos, Silveira, Meirelles, Longo, Gotoh, Nakamura, Iida, Sawabe, Rezende, Francini-Filho, Moura, Amado-Filho and Thompson2014). Its associated flora was reduced to zero after 2015 and had a lower abundance compared with other tropical formations (i.e. Amado-Filho et al., Reference Amado-Filho, Maneveldt, Manso, Rosa Marins, Pacheco and Guimarães2007; Riul et al., Reference Riul, Lacouth, Pagliosa, Christoffersen and Horta2009; Bahia et al., Reference Bahia, Abrantes, Brasileiro, Pereira-Filho and Amado-Filho2010). The annual production of the Arvoredo bed (106 g m−2 year−1) is lower than reported values of tropical beds, such as in Abrolhos (1000 g m−2 year−1: Amado-Filho et al., Reference Amado-Filho, Moura, Bastos, Salgado, Sumida, Guth, Francini-Filho, Pereira-Filho, Abrantes, Brasileiro, Bahia, Leal, Kaufman, Kleypas, Farina and Thompson2012a). However, it is similar to the neighbour bed in Deserta Island, for Lithophyllum sp. (55–136.3 g m−2 year−1) (Gherardi, Reference Gherardi2004), and close to reported values for temperate beds (490 g m–2 year–1 in Martin et al., Reference Martin, Clavier, Chauvaud and Thouzeau2007; 200 g m–2 year–1 in Teichert & Freiwald, Reference Teichert and Freiwald2014).
Rhodolith species composition
Total biomass of rhodoliths was higher in winter months. However, rhodoliths grow slowly and do not respond as rapidly to environmental factors as faster growing macroalgae (Wilson et al., Reference Wilson, Blake, Berges and Maggs2004; Francini-Filho et al., Reference Francini-Filho, Coni, Meirelles, Amado-Filho, Thompson, Pereira-Filho, Bastos, Abrantes, Ferreira, Gibran, Güth, Sumida, Oliveira, Kaufman, Minte-Vera and Moura2013). Despite their slow growth (Basso, Reference Basso2012), there is no evidence that rhodoliths lose a significant amount of biomass annually. Therefore, major changes in their abundance must be related to local dynamics and rhodolith displacement. To analyse the influence of environmental factors on their development, it is necessary to consider data from the previous season. Thus, the mean values of wind direction and speed during the six-month window prior to each collection were used. The correlation analyses between these data and total rhodolith biomass indicate a negative correlation with wind speed, which suggests that stronger currents generated by winds could cause scattering of most rhodoliths. The same results were also found in L. crispatum, which is lighter, more fragile and more branched than the other three species. Pascelli et al. (Reference Pascelli, Riul, Riosmena-Rodríguez, Scherner, Nunes, Hall-Spencer, Oliveira and Horta2013) performed sampling in the summer and winter at Rebio Arvoredo and reported a reduction in rhodolith diameter in winter, suggesting that storms in this season could have caused breakage. Here, the history of wind course before the period of collections revealed that wind changes are not restricted to seasons and even so, rhodolith biomass is responsive.
In relation to wind direction, L. crispatum was positively correlated with north quadrants and negatively correlated with south quadrants, whereas L. atlanticum had the opposite response. When the frequency of winds from the north quadrant is higher, the lighter, more fragile and branched rhodoliths (i.e. L. crispatum) tend to accumulate at the sampling site in the north portion of Arvoredo Island. In contrast, L. crispatum had the highest biomass in February 2015, and the previous months had a lower wind speed in all quadrants. Plausible mechanisms of seasonal rhodolith abundance changes may include the displacement of lighter rhodoliths in the north portion of the island, or the selective burial of heavier and sleeker morphotypes, L. atlanticum and M. erubescens. When the south wind returns, lighter rhodoliths were scattered and heavier rhodoliths unearthed. Even if rhodolith composition is correlated with wind direction and speed, other environmental factors such as bottom currents and depth could influence it as well (Sañé et al., Reference Sañé, Chiocci, Basso and Martorelli2016). Due to this, we suggest future monitoring of other environmental changes alongside community composition observation.
Associated fauna
The faunal biomass had lower values for Mollusca in February 2015 when the mean wind speed of previous months was lower and local currents consequently decreased. These results oppose those of previous studies, which found that higher current velocity disturbs the settlement of motile animals (Maughan & Barnes, Reference Maughan and Barnes2000; Hinojosa-Arango et al., Reference Hinojosa-Arango, Maggs and Johnson2009). This suggests that other abiotic or biotic factors could be related to the decrease in faunal biomass. Reef fish are key players that directly and indirectly influence the dynamic balance of marine food webs (Dunne et al., Reference Dunne, Williams and Martinez2004; Bellwood et al., Reference Bellwood, Goatley and Bellwood2017). Their biology (e.g. feeding behaviours, functional roles/trophic affinities) can influence variations in biomass and density levels of populations in the trophic web of rocky and coral reef systems (Dunne et al., Reference Dunne, Williams and Martinez2004; Bellwood et al., Reference Bellwood, Goatley and Bellwood2017). The significant increase in invertebrate community biomass detected after 2015 raised two hypotheses regarding the probable influence of reef fish. (1) Top-down control (Worm & Myers, Reference Worm and Myers2003; Begon et al., Reference Begon, Townsend, John, Colin and John2006; Baum & Worm, Reference Baum and Worm2009): population increase mediated by predators in the studied area over time (2015–2016). (2) Top-down control associated with the ‘fear effect’ (Worm & Myers, Reference Worm and Myers2003; Côté et al., Reference Côté, Darling, Malpica-Cruz, Smith, Green, Curtis-Quick and Layman2014; Heupel et al., Reference Heupel, Knip, Simpfendorfer and Dulvy2014): both are mediated by the presence of a recently detected alien predator inhabiting the rhodolith beds in the studied area (more details below).
Considering the first hypothesis, the modest increase in predator populations in the Arvoredo MPA during the past three years may have reduced densities and biomass of fish populations of subjacent levels in the trophic web, such as grunts (i.e. Haemulidae). These reef fish feed mainly on macroinvertebrates (e.g. annelids, molluscs and crustaceans) and their population decrease may have caused an increase in their prey's densities and biomass (‘prey release effect’) (Friedlander & DeMartini, Reference Friedlander and DeMartini2002; Heithaus et al., Reference Heithaus, Frid, Wirsing and Worm2008). However, populations of grunts in the Arvoredo MPA are still much higher than groupers, which leads us to assume that predator–prey interactions may not have been the indirect cause of the increase in invertebrate populations. Moreover, considering fish biogeography, fish populations inhabiting thresholds of distribution, such as the Arvoredo MPA (Anderson et al., Reference Anderson, Carvalho-Filho, Morais, Nunes, Quimbayo and Floeter2015; Anderson, Reference Anderson2017), tend to have more pronounced fluctuations in their populations caused by stochastic mechanisms (e.g. oscillations in temperature during harsh winters) (Almada & Faria, Reference Almada and Faria2004; Anderson, Reference Anderson2017).
In regards to the second hypothesis, the presence of an alien species may cause a dramatic effect in local populations over a short period of time (Anderson, Reference Anderson2017; Andradi-Brown et al., Reference Andradi-Brown, Vermeij, Slattery, Lesser, Bejarano, Appeldoorn, Goodbody-Gringley, Chequer, Pitt and Eddy2017). In the summer of 2015, the black-spotted snake eel Quassiremus ascensionis (Studer, 1889) was first recorded for the Southern Atlantic, inhabiting the rhodolith bed of Rancho Norte, Arvoredo MPA (Anderson et al., Reference Anderson, Carvalho-Filho, Morais, Nunes, Quimbayo and Floeter2015). In 2016, Q. ascensionis was detected in all parts of the Arvoredo MPA (Arvoredo Is., Deserta Is. and Galé Is.). This species is a cryptic predator (e.g. it preys with its body buried in the substrate and only the head showing), preying mainly on small fish and invertebrates (Froese & Pauly, Reference Froese and Pauly2016). The presence of a new predator may have influenced fluctuations in populations of small serranids, blennies and gobies (e.g. Serranus baldwinni, Diplectrum radiale, Parablennius marmoreus, Parablennius pilicornis, Hyplerochilus fissicornis, Coryphopterus glaucofraenum) and thus, a subsequent increase in density and biomass of their prey (annelids, molluscs and crustaceans).
Further investigations and long-term monitoring are necessary to understand the mechanism involved in population dynamics of organisms inhabiting the rhodolith beds of southern Brazil. Beyond trophic dynamics, temperature changes can also influence faunal abundance. In summer 2015, the mean seawater temperature registered at Rebio Arvoredo was the highest it had been in three years (Figure 3), whereas the coolest was in winter 2016. This could be related to the El Niño in 2015/2016, resulting in substantial environmental changes. This event could cause altered rain regimes, winds and marine currents, which could consequently influence temperature (Freire et al., Reference Freire, Varela, Fonseca, Menezes, Fest, Obata, Gorri, Franco, Machado, Barros, Molessari, Madureira, Coelho, Carvalho, Pereira, Segal, Freire, Lindner, Krajewski and Soldateli2017). In 2015, the mean summer temperature was 26°C and the winter had few cold fronts when compared with the other years (Freire et al., Reference Freire, Varela, Fonseca, Menezes, Fest, Obata, Gorri, Franco, Machado, Barros, Molessari, Madureira, Coelho, Carvalho, Pereira, Segal, Freire, Lindner, Krajewski and Soldateli2017). This value is much higher than the mean seawater temperature (22°C) of this region (Gherardi, Reference Gherardi2004; Pascelli et al., Reference Pascelli, Riul, Riosmena-Rodríguez, Scherner, Nunes, Hall-Spencer, Oliveira and Horta2013). Climate changes and temperature increase can cause a reduction in survivorship of invertebrates and fishes (Vinagre et al., Reference Vinagre, Mendonça, Cereja, Abreu-Afonso, Dias, Mizrahi and Flores2018). The survival, body size and larval production of the polychaeta, Streblospio benedicti Webster, are higher in winter–spring than summer–autumn conditions (Levin & Creed, Reference Levin and Creed1986). Marine heatwaves have also caused a decrease in abundance of the large gastropod, Lunella off the west coast of Australia (Smale et al., Reference Smale, Wernberg and Vanderklift2017). McConnico et al. (Reference McConnico, Carmona, Morales and Rodríguez2017) reported an increase in molluscs in the summer following an increase in macroalgal biomass in a rhodolith bed off Baja California, Mexico. Here, the opposite occurred. The variable responses of faunal behaviour to temperature and wind indicate that experiments involving the faunal community and larger replication are necessary. This will improve our knowledge of the environmental conditions that influence the local faunal biomass and explain the observed temporal changes. Further investigations on the dynamics of species composition by phylum in addition to biomass would contribute additional important information about community function.
Associated macroalgae
The occurrence of epiphytes only in February 2015 could indicate that higher temperature and lower wind speed, and thus, lower hydrodynamics, favoured fleshy algal growth, particularly Padina gymnospora. This species can grow in tropical temperatures similar to the ones that we recorded in the summer (Scherner et al., Reference Scherner, Pereira, Duarte, Horta, Castro, Barufi and Pereira2016). Lower hydrodynamics have been associated with an increase in abundance and richness of biomass of maerl-associated algal species (Hily et al., Reference Hily, Potin and Floc'h1992; Hinojosa-Arango et al., Reference Hinojosa-Arango, Maggs and Johnson2009). However, although smaller in quantity, previous data of total epiphyte biomass from February 2002 (11.41 g × m−2 Horta, 2002, unpublished data) and February and July of 2008 (0.9 and 0.16 g × m−2, Pascelli, Reference Pascelli2009) show that the occurrence of epiphytes before February 2015 was not so rare (Supplementary Figure S2).
Coralline algae can adopt strategies to reduce the growth rate and recruitment potential of fleshy macroalgae (Vermeij et al., Reference Vermeij, Dailer and Smith2011), aside from producing allelopathic substances (Gross, Reference Gross2003; Kim et al., Reference Kim, Choi, Kang, Cho, Jin, Chun and Hong2004). Suzuki et al. (Reference Suzuki, Takabayashi, Kawaguchi and Matsunaga1998) suggested that an allelopathic non-polar substance produced by Lithophyllum spp. destroys zoospores of the brown alga Laminaria religiosa Miyabe, which contributes to the predominance of this crustose coralline alga in the coastal region of the Northern Japan Sea. However, the abrupt decrease after 2015 indicates a combination of climate events referred to above and an increase in invertebrate biomass that possibly played roles in the control of epiphytes.
The association with macroinvertebrates can benefit rhodoliths because some of these animals feed on epiphytes (Legrand et al., Reference Legrand, Riera, Lutier, Coudret, Grall and Martin2017). The limpet Patella longicosta Lamarck has a mutualistic relationship with the crustose alga Ralfsia verrucosa (Areschoug) Areschoug that prevents the overgrowth of Ulva sp. and other grazers (McQuaid & Froneman, Reference McQuaid and Froneman1993). Mesograzers such as polychaetes and gastropods have herbivorous members and have been related to biomass control at the Rebio Arvoredo bed (Scherner et al., Reference Scherner, Riul, Bastos, Bouzon, Pagliosa, Blankensteyn, Oliveira and Horta2010). Snails can be advantageous for macrophytes, reducing the density of bacteria and epiphytic algae that may be potentially deleterious to the host due to shading of the thallus and increased competition for resources (Underwood et al., Reference Underwood, Thomas and Baker1992; Stachowicz & Whitlatch, Reference Stachowicz and Whitlatch2005). Some animals can even feed on rhodoliths without causing mortality. The chiton Choneplax lata Guilding grazes on the coralline alga Porolithon pachydermum (Foslie) Foslie and stimulates new meristematic activity, and also removes sporelings of competitive epiphytes (Littler et al., Reference Littler, Littler and Taylor1995).
Community production/consumption balance
In this study, the association of rhodoliths with herbivorous invertebrates probably benefits the rhodoliths. Experimental results showed that the interaction between Padina gymnospora and L. crispatum caused a decrease in photosynthetic and calcification rates when compared with the treatments where these organisms were incubated alone. However, photosynthesis of L. crispatum alone had the highest value. A comparative study showed that red algae experience increased photosynthetic rates compared with green and brown algae at depths down to 10 m, which indicates that they are better adapted chromatically to photosynthesize in this range (Dring, Reference Dring1981). This could be related to the fact that red coralline algae are found in a wide range of depths and irradiances, from shallow tropical coral reefs (PAR >1500 μmol photons m−2 s−1) (Burdett et al., Reference Burdett, Keddie, MacArthur, McDowall, McLeich, Spielvogel, Hatton and Kamenos2014) to great depths (>200 m PAR = 0.0015 μmol photons m−2 s−1) (Littler et al., Reference Littler, Littler, Blair and Norris1986). Light and dark calcification results display a pattern similar to the one observed for gross photosynthesis and respiration.
Epibiosis is usually considered harmful to the host alga because of the competition for light and nutrients, which results in a decrease in growth and reproduction (Amsler, Reference Amsler2008). Competition among algae for one or more limited resources may be direct (interference competition) or indirect, through the depletion of a resource (exploitative competition), and may occur within or between algal species (Reiskind et al., Reference Reiskind, Beer and Bowes1989). Padina gymnospora may shade the rhodolith and thus decrease the light availability for photosynthesis. In earlier studies, the effects of light reduction generated by sedimentation caused a decrease in L. crispatum abundance (Riul et al., Reference Riul, Targino, Farias, Visscher and Horta2008). Moreover, numerous epiphytic organisms generally colonize rhodolith beds (Grall et al., Reference Grall, Le Loc'h, Guyonnet and Riera2006; Peña et al., Reference Peña, Bárbara, Grall, Maggs and Hall-Spencer2014). Epiphytes modify the interface between the hosting seaweed and the external environment, creating greater heterogeneity at the seaweed surface (Wahl, Reference Wahl2008). This may create a physical barrier for light absorption (Drake et al., Reference Drake, Dobbs and Zimmerman2003) and carbon uptake (Sand-Jensen, Reference Sand-Jensen1977). Light absorption by epiphyte pigments lowered the photosynthetic rate of Cladophora glomerata (Linnaeus) Kützing when irradiance was below 200–500 μmol photons m−2 s−1 (Dodds, Reference Dodds1991). Rohde et al. (Reference Rohde, Hiebenthal, Wahl, Karez and Bischof2008) reported that epiphytes modified the surface of the Fucus thallus, creating a barrier for nutrient uptake and gas exchange.
The presence of P. gymnospora not only caused a decrease in photosynthesis, but also in calcification. This suggests that the shade caused by P. gymnospora damaged rhodolith calcification. The calcification of the alga Hydrolithon reinboldii was 29% greater in high (650 μmol quanta m−2 s−1) than in low (150 μmol quanta m−2 s−1) irradiance (Comeau et al., Reference Comeau, Carpenter and Edmunds2014). The net dissolution that occurred at night confirmed the importance of light in the calcification process of L. crispatum. Approximately 50% of carbon fixed in photosynthesis is lost in dark respiration (Borowitzka & Larkum, Reference Borowitzka and Larkum1987). Calcification and inorganic carbon uptake predominately occur in the light (Cornwall et al., Reference Cornwall, Comeau and McCulloch2017). Although calcification is correlated to photosynthesis, there are also metabolically controlled photosynthesis-independent ion pumps and channels with the efflux of Ca2+ and H+ outside the cells in the dark and influx under light (Hofmann et al., Reference Hofmann, Koch and de Beer2016). The proton pumps in light are important to pH regulation at the thalli surface and calcification under ocean acidification (McNicholl et al., Reference McNicholl, Koch and Hofmann2019).
The experiment showed the negative effect of overgrowth of P. gymnospora on the rhodolith surface, which coincided with a decrease in invertebrates. In the following years, an opposite pattern was observed. Epiphytes disappeared, the biomass of invertebrates increased, but invertebrate feeders slightly decreased. These effects on the rhodolith bed community are probably the result of environmental changes that occurred during the years of this study. Changes in temperature and wind regimes were detected, which not only can cause altered food webs, but also changes in rhodolith composition and biomass. The variation in characteristics of the community during these two years and the environmental changes make us aware of how climate change events can affect ecosystems.
Conclusion
We conclude that environmental factors, such as wind and temperature, as well as biological factors, can cause differences in community structure over time. As a consequence of these changes, the biomass of macroinvertebrate feeders decreases, which increases the biomass of their prey. As a result, these organisms (in the major part herbivores) can feed on rhodoliths, which could possibly be one of the causes of the disappearance of macroalgae after 2015. The growth of P. gymnospora on L. crispatum detected in summer 2015 caused a decrease in photosynthetic and calcification rates of the rhodolith. These results reinforce the importance of the association with herbivores for the control of epiphytic organisms on the rhodolith surface. More long-term experiments are necessary to improve the knowledge about the role of environmental changes on rhodolith bed communities and the response of associated sessile and motile organisms. Finally, this study suggests that a healthy trophic structure with abundant top predators can indirectly increase rhodolith fitness, and potentially their resilience to climate change and local threats.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0025315420000284.
Acknowledgements
This study represents a contribution to the activities of the Rede de Monitoramento de Habitats Bentônicos Costeiros (ReBentos) and the Brazilian Ocean Acidification Network (BrOA). We thank the Instituto Chico Mendes de Conservação da Biodiversidade (ICMBio) for their support during fieldwork and permits for sampling, Maare (Environmental Monitoring of Arvoredo Marine Biological Reserve) for the temperature data and Laboratory of Lamination of the Department of Geology of the UFSC for cutting the rhodolith samples. We also acknowledge Iole B.M. Orselli for performing part of the chemical analyses of total alkalinity and Ellie Bergstrom for the tips throughout the text.
Financial support
We thank the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq grant no. 407365/2013-3 to PA Horta) and the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) for funding this study. RK acknowledges a CNPq researcher grant no. 302604/2015-4. We also thank FAPES (Fundação de Amparo à Pesquisa e Inovação do Espírito Santo, Brazil)/CAPES No 10/2018 – PROFIX program for A.B.A. post-doctoral scholarship.
