Drastic decadal decline of the seagrass Cymodocea nodosa at Gran Canaria (eastern Atlantic): interactions with the green algae Caulerpa prolifera

1  Drastic decadal decline of the seagrass Cymodocea nodosa at Gran Canaria (eastern Atlantic): interactions with the green algae Caulerpa prolifera Fernando Tuya*, Harue Hernandez-Zerpa, Fernando Espino, Ricardo J. Haroun Centro en Biodiversidad y Gestión Ambiental, Marine Sciences Faculty, Universidad de Las Palmas de Gran Canaria, Las Palmas, Canary Islands, Spain *Corresponding author. Tel.: +34 928457456, Fax: +34 928452900. E-mail adress: [email protected] (F. Tuya) 2  Abstract The shoot density, leaf length and biomass of the seagrass Cymodocea nodosa (Ucria) Ascherson were found to severely decline in the last 17 years in the oceanic island of Gran Canaria (central Eastern Atlantic). Five seagrass meadows were sampled in summer and winter of 1994-1995 and in winter and summer 2011. The decrease in C. nodosa correlated with a 3-fold increase in the biomass of the green rhizophytic algae Caulerpa prolifera (Forsskål) J.V. Lamoroux over the same time period, although this increase varied notably among meadows. We also documented a negative correlation between the biomass of C. nodosa and C. prolifera at the island-scale, sampling 16 meadows in 2011. Experimental evidence demonstrated that C. prolifera can cause significant negative impacts on C. nodosa: plots with total (100%) removals of C. prolifera had ca. 2.5 more shoots and 3.5 times more biomass of C. nodosa, after 8 months, compared to plots with 50% removals and untouched control plots. Interference by C. prolifera appears to partially explain the decay in the abundance of C. nodosa populations in Gran Canaria. This study, however, did not identify potential underlying processes and/or environmental alterations that may have facilitated the disappearance of C. nodosa. Keywords: seagrass, decline, competition, decadal trends, Caulerpa, Canary Islands. 3  1. Introduction On shallow subtidal soft bottoms, seagrasses are the main foundation species from tropical to temperate oceans. The mechanisms behind their influence on community structure are multifaceted, but key ecological functions include modifying local environmental conditions and the provision of food and habitat for a wide range of organisms (Constanza et al., 1997). Conservation of these valuable habitats is therefore important, particularly since seagrass meadows are declining worldwide, mainly in areas of intense human activities (Duarte et al., 2008; Hughes et al., 2009; Waycott et al., 2009). There is, however, no information on seagrass distribution and abundance patterns from most coasts of the world over the last decades, and so many losses are unknown (Duarte et al., 2008). Cymodocea nodosa (Ucria) Ascherson is a seagrass distributed across the Mediterranean Sea and the adjacent eastern Atlantic coasts, including the Macaronesian oceanic archipelagos of Madeira and the Canaries (Alberto et al., 2006; Mascaró et al., 2009). Meadows constituted by C. nodosa are the dominant vegetated communities in shallow soft substrates throughout the Canaries (Pavón-Salas et al., 2000; Barberá et al., 2005), where they provide food and shelter for diverse invertebrate and fish assemblages (Tuya et al., 2001; Tuya et al., 2006; Espino et al., 2011). These meadows are generally located along the eastern and southern coasts of the islands, sheltered from the dominant swells from the north and north-west. C. nodosa forms extensive, but often fragmented, subtidal meadows (Reyes et al., 1995a; Pavón-Salas et al., 2000; Espino et al., 2003; Barberá et al., 2005). In this region, C. nodosa shows a clear seasonal pattern, with a summer peak in shoot density and biomass (Reyes et al., 1995a, 1995b; Tuya et al., 2006), similar to what has been observed in the Mediterrean (e.g. Terrados and Ros, 1993). 4  In the Mediterranean, widespread decline of seagrass meadow has often resulted in the replacement by green algae such as Caulerpa (Ceccherelli and Cinelli, 1997; Lloret et al., 2005), though the time scale can affect interactions between C. nodosa and Caulerpa. For example, on the long-term, effects of Caulerpa on C. nodosa may not be so severe (Ceccherelli and Sechi, 2002). The ecological mechanisms behind these shifts in seagrass abundance (cover) vary between studies. For example, human-induced increases in nutrient loading and suspended sediments in the water column, involving a reduction in water transparency, can locally facilitate the replacement of C. nodosa by Caulerpa prolifera (Forsskål) J.V. Lamoroux (Lloret et al., 2005; Morris et al., 2009). In other circumstances, competition for nutrients in the sediment has been pointed out as the main ecological mechanism explaining the regression of C. nodosa and the concurrent expansion of Caulerpa beds (Ceccherelli and Cinelli, 1997). In this study, we (i) compared the shoot density, leaf length and biomass of the seagrass C. nodosa in 5 seagrass meadows at the oceanic island of Gran Canaria (Canary Islands, eastern Atlantic) between 1994-1995 and 2011. Since we detected a sharp decrease in the abundance (shoot density and biomass) of C. nodosa that was partially matched with an increase in the biomass of the green, rhizophytic, native algae C. prolifera, we additionally (ii) tested whether the biomass of C. nodosa and C. prolifera are currently negatively correlated across the island, and in a manipulative field experiment (iii) tested if C. prolifera can have a negative effect on C. nodosa, i.e. whether removal of C. prolifera would increase the abundance of C. nodosa. 2. Materials and methods 2.1. Historical comparison 5  Five seagrass meadows (Table 1) were selected across the entire distribution area of C. nodosa in Gran Canaria. Each meadow was between 0.5-10 km apart from the adjacent studied meadow to encompass a range of conditions across the island. Each meadow was sampled in 4 occasions (Table 1), including a winter and summer season in 1994-1995, and winter and summer of 2011. On each sampling time, three cores (20 cm of inner diameter) were pushed into the sediment by a SCUBA diver. All material was then transferred to labelled bags and frozen (-20ºC) until being processed in the lab. For each sample, we counted the number of shoots (shoot density), as well as measured the length of 30 randomly selected leaves. The biomass was separated into leaves and rhizomes and subsequently oven-dried (24h at 70ºC) to obtain dry-weight biomass measurements. The dry biomass of all accompanying macroalgae, mainly the green algae Caulerpa prolifera, was also obtained. All measurements were standardized to m-2 to facilitate comparisons with other studies, and followed standardized procedures (Bortone, 2000). Temporal differences between years (hereafter mid-1990s vs. 2011), seasons (winter vs. summer) and sites (=meadows) for all demographic descriptors were tested by a 3-way permutation-based ANOVA, including the factors: ‛Year’ (fixed factor), ‛Season’ (fixed factor and orthogonal to ‛Year’) and ‛Site’ (random factor orthogonal to both ‛Year’ and ‛Site’). Pairwise comparisons using permutations (Anderson, 2001) resolved differences between years for each site (significant ‛Year x Site’ interactions). Permutational Analysis of Variance uses permutations to calculate Pvalues. This was preferable because the data were over-dispersed and contained many zeros. In this sense, the Cochran’s test was used to check for homogeneity of variances of each variable. However, no transformation rendered homogeneous variances for the biomass of C. nodosa and C. prolifera (Cochran’s test, p < 0.05, for all type of transformations). The ANOVAs were then carried out on untransformed data, as it is 6  robust to heterogeneity of variances for large balanced experiments (Underwood, 1997). To avoid an increase in a type I error rate, α values were then established at a conservative value of 0.01 (Underwood, 1997). Similarly, the significance of pairwise comparisons was fixed at the α = 0.01 level. The test statistic (pseudo-F) is a multivariate analogue of the univariate Fisher’s F ratio, and in the univariate context the two are identical when using Euclidean distance as the dissimilarity measure (Anderson, 2001).   2.2. Interaction between C. nodosa and C. prolifera: comparative field analysis We sampled another 11 meadows (for a total of 16 meadows) in winter and summer of 2011 along the entire perimeter of Gran Canaria (Table 1), following the same criteria outlined previously. All these seagrass meadows were included in the shallow-water marine qualitative seagrass cartography of the island produced in 2002 and 2003 (Espino et al., 2003). A linear regression model tested whether the total biomass of C. nodosa and C. prolifera were significantly correlated at the island scale, i.e. including all 16 meadows, separately for winter and summer 2011, since both C. nodosa and C. prolifera show larger biomasses in summer than in winter in the Canary Islands (Reyes, 1993).  2.3 Interaction between C. nodosa and C. prolifera: experimental approach We set up twelve 2 x 2 m plots on a mixed C. nodosa and C. prolifera meadow (‛Gando Castillo’, Table 1); adjacent plots were 2 m apart. Four plots were randomly assigned to each of 3 treatments: total (100%) removal of C. prolifera (ca. 140 ± 21 g 7  DW m-2), partial removal (50%, ca.70 ± 9 g DW m-2) and no removal (0%, procedural control) of C. prolifera. Removal was performed by SCUBA divers that carefully handpicked stolons and blades of C. prolifera. An analogue disturbance, through flipping fins, but without actually removing C. prolifera, was conducted in control plots to avoid confounding results with manipulation artifacts. The experiment started on early March 2011 and lasted for 8 months; plots were visited every 5-6 weeks to maintain experimental treatments. At the end of the experimental period, 2 cores (20 cm of inner diameter) were collected, as previously described, from the center of each experimental plot, to avoid edge effects. All material was transferred to labelled bags and frozen (- 20ºC) until processed in the lab, following the same routines outlined before. Differences in shoot density, leaf length and total biomass were tested with 1-way ANOVAs, including the factor ‛Treatment’ (as fixed) with 3 levels: 100, 50 and 0%). Pairwise comparisons resolved differences among treatments. ANOVAs were carried out on untransformed data, since all variables showed homogeneous variances among groups (Cochran’s test, p > 0.05). 3. Results 3.1. Historical comparison Shoot density, leaf length and total biomass of C. nodosa suffered a reduction from the mid-1990s to 2011 (Table 2) that resulted in significant differences between years (Table 3, 3-way ANOVA: ‛Year’, p < 0.05 in all cases). This reduction over time, however, varied in magnitude, but not in direction, from site to site (Table 3, 3-way ANOVA: ‛Year x Site’, p < 0.01 in all cases): pairwise comparisons showed a significant decrease, from the mid-1990s to 2011, for all descriptors, except leaf length, 8  at the 5 sites. At one site (‛Arinaga’, Table 2), the entire meadow had disappeared. Over the same time period, the total biomass of C. prolifera overall increased ca. 3-fold, although this trend varied notably among sites (Table 2) and resulted in an inconsistency in the differences between years (Table 3, 3-way ANOVA: ‛Year x Site’, p < 0.01). Post-hoc pairwise comparisons showed that only two meadows experienced a significant increase in the biomass of C. prolifera. 3.2. Interaction between C. nodosa and C. prolifera In the island-scale comparison, biomass of C. nodosa and C. prolifera correlated negatively (Fig. 1) in the growing season (summer), but not in winter. At the end of the experimental period, shoot density and total biomass of C. nodosa experienced a significant (Table 4, 1-way ANOVA: ‛Treatment’, p < 0.0001) increase in total C. prolifera removal plots (100%) relative to plots where C. prolifera was partially removed (50%) or left untouched (0%) (Figs. 2a and 2b, respectively). However, the leaf length of C. nodosa did not differ significantly between treatments (Fig. 2c, Table 4, 1-way ANOVA: ‛Treatment’, p > 0.05). 4. Discussion We observed a severe decline in the abundance, i.e. shoot density and biomass, of the seagrass Cymodocea nodosa at Gran Canaria between the mid-1990s and 2011, which is consistent with other studies reporting declines in seagrass abundance over the last decades (Hall et al., 1999; Hemminga and Duarte, 2000; Hughes et al., 2002; Waycott et al., 2009). Similarly, a recent study has also documented an overall decrease 9  in the cover of C. nodosa meadows at Gran Canaria in the last 3 decades (MartínezSamper, 2011). The exact reasons for the loss of C. nodosa in Gran Canaria are not clear, but we here discuss possible patterns. Four of the five studied sites are far away from any welldefined, visible, sources of human-induced disturbance, such as sewage and brine outlets, fish farms and ports. The only exception is ‛Arinaga’, which could have been seriously affected by the construction of an industrial port, only 500 m away, 10 years ago (Martínez-Samper, 2011). This is the only site with a complete disappearance of C. nodosa and C. prolifera during the study period (Table 2). Therefore, it could well be that processes occurring at the island-scale, rather than punctual stressors have caused the seagrass deterioration at Gran Canaria. Interactions with the green algae C. prolifera may partially contribute to the reduction in the abundance of C. nodosa, as we have empirically demonstrated here using both observational and experimental evidence. Caulerpa prolifera is a nitrophilic alga with clonal modular morphology (Collado-Vides, 2002). Contrary to many seaweeds, rhizoids of C. prolifera can take up nutrients from the sediment porewater, which may cover the total N requirement (Williams, 1984; Chisholm et al., 1996). In the Mediterranean, C. prolifera has been shown to overgrow the rhizomes of Posidonia oceanica and C. nodosa (Ceccherelli et al., 2000); therefore, species interactions in these systems may occur both in the below and above ground compartments, interfering respectively with nutrient acquisition and light availability. In the Mediterranean, the growth, as well as the germination, of C. nodosa has been shown to be nutrient limited (Pérez et al., 1991; Terrados and Ros, 1993; Balestri et al., 2010). This is likely pertinent for the Canary Islands, since adult leaves are considerably N-limited in the field (e.g. ~ 1% in N content, F. Tuya, unpublished data). Hence, it is plausible that C. 16  Malta, E.J., Ferreira, D.G., Vergara J.J., Pérez-Lloréns J.L. 2005. Nitrogen load and irradiance affect morphology, photosynthesis and growth of Caulerpa prolifera (Bryopsidales : Chlorophyta). Mar. Ecol. Prog. Ser. 298, 101-114. Martínez-Samper, J., 2011. Análisis espacio-temporal de las praderas de Cymodocea nodosa (Ucria) Ascherson en la isla de Gran Canaria. Master Thesis, Universidad de Las Palmas de G.C. Mascaró, O., Oliva, S., Pérez, M., Romero, J., 2009. Spatial variability in ecological attributes of the seagrass Cymodocea nodosa. Bot. Mar. 52, 5-429. Morris, E.P., Peralta, G., Benavente, J., Freitas, R., Rodrigues, A.M., Quintino, V., Álvarez, O., Valcárcel-Pérez, N. Vergara, J.J., Hernández, I., Pérez-Lloréns, J.L., 2009. Caulerpa prolifera stable isotope ratios reveal anthropogenic nutrients within a tidal lagoon. Mar. Ecol. Prog. Ser. 390, 117–128. Pavón-Salas, N., Herrera, R., Hernández-Guerra, A., Haroun, R., 2000. Distributional pattern of seagrasses in the Canary Islands (Central-East Atlantic Ocean). J. Coastal Res. 16, 329-335. Pérez, M., Romero, J., Duarte, C., Sand-Jensen, K., 1991. Phosphorus limitation of Cymodocea nodosa growth. Mar. Biol. 109, 129-133. 17  Pesando, D., Leme´e, R., Ferrua, C., Amade, P., Girard, J.P., 1996. Effects of caulerpenyne, the major toxin from Caulerpa taxifolia on mechanisms related to sea urchin egg cleavage. Aquat. Toxicol. 35, 139–155. Portillo, E., 2008. Arribazones de algas y plantas marinas en Gran Canaria: características, gestión y posibles usos. Instituto Tecnológico de Canarias, Las Palmas. Raniello, R., Mollo, E., Lorenti, M., Gavagnin, M., Buia, M.C., 2009. Phytotoxic activity of caulerpenyne from the Mediterranean invasive variety of Caulerpa racemosa: a potential allelochemicals. Biol. Invas. 9: 361–368. Reyes, J., 1993. Estudio de las praderas marinas de Cymodocea nodosa Cymodoceaceae, Magnoliophyta) y su comunidad de epifitos, en El Médano (Tenerife, Islas Canarias). PhD Thesis, Universidad de La Laguna, Tenerife. Reyes, J., Sansón, M., Afonso-Carrillo, J., 1995a. Distribution and reproductive phenology of the seagrass Cymodocea nodosa (Ucria) Ascherson in the Canary Islands. Aquat. Bot., 50: 171-180. Reyes, J., Sansón, M., Afonso-Carrillo, J., 1995b. Leaf phenology, growth and production of the seagrass Cymodocea nodosa at El Médano (south of Tenerife, Canary Islands). Bot. Mar. 38, 457-465. 18  Terrados, J., Ros, J.D., 1993. Limitación por nutrientes del crecimiento de Cymodocea nodosa (Ucria) Ascherson en sedimentos carbonatados en el Mar Menor, Murcia, SE de España. Boletín del Instituto Español de Oceanografía 11, 9-14. Tuya, F., Pérez, J., Medina, L., Luque, A., 2001. Variaciones estacionales de la macrofauna invertebrada de tres praderas marina de Cymodocea nodosa en Gran Canaria (Centro-Este del océano Atlántico). Cien. Mar. 27, 223-234. Tuya, F., Martín, J.A., Luque, A., 2006. Seasonal cycle of a Cymodocea nodosa seagrass meadow and of the associated ichthyofauna at Playa Dorada (Lanzarote, Canary Islands, eastern Atlantic). Cien. Mar. 32, 695-704. Underwood, A.J., 1997. Experiments in Ecology: their logical design and interpretation using Analysis of Variance. Cambridge University Press, Cambridge. Waycott, M., Duarte, C.M., Carruthers, T.J.B., Orth, R.J., Dennison, C.W, Olyarnik, S., Calladine, A., Fourqurean, J.W., Heck, Jr., K.L., Hughes, A.R., Kendrick, G.A., Kenworthy, W.J., Short, F.T., Willians, S.L., 2009. Accelerating loss of seagrasses across the globe threatens coastal ecosystems. Proc. Natl. Acad. Sci. USA 106, 1237712381. Williams, S.L., 1984. Uptake of sediment ammonium and translocation in a marine green macroalga Caulerpa cupressoides. Limnol. Oceanogr. 29, 374–379. 19   Table 1. Geographical description of sites (= seagrass meadows, from north to south) sampled for Cymodocea nodosa and Caulerpa prolifera across the island of Gran Canaria; dates of sampling are included. Sites in bold are those considered for the historical comparison (sampled in both the mid-1990s and 2011), while the rest were only sampled in 2011. Mid-1990s 2011 Site UTM X UTM Y Depth (m) Winter Summer Winter Summer Gando 463434 3089688 4 Jan-95 May-95 Feb Aug Gando-piscina 464252 3089224 15 Feb Aug Gando Castillo 463105 3089320 10 Feb Aug Gando-boya 463760 3089133 19 Feb Aug Roque de Arinaga 462589 3081567 14 Jan-95 Aug-94 Feb Sep Risco Verde 462095 3081237 10 Mar-95 Jul-94 Feb Sep Arinaga 460946 3081019 5 Jan-95 Jul-94 Feb Sep Arinaga-muelle viejo 461108 3080464 9 Feb Sep Arinaga-2 460863 3080742 7 Feb Sep Pasito Blanco Playa 439077 3069016 8 Feb Aug Pasito Blanco 439109 3068979 8 Dec-94 Aug-95 Feb Aug Pasito Blanco-fuera 439142 3068942 9 Feb Aug Centro-Com 439175 3068905 10 Feb Aug Meloneras-1 439224 3068923 5 Feb Aug Meloneras-2 439386 3068553 11 Feb Aug Faro Maspalomas 439550 3068460 12 Feb Aug 20  Table 2. Shoot density (shoots m-2), leaf length (cm), total biomass (g DW m-2) of Cymodocea nodosa and total biomass (g DW m-2) of Caulerpa prolifera at each site on each sampling occasion. Values are means ± SD (n=3). Risco Verde Roque Arinaga Arinaga Gando Pasito Blanco Shoot density Mid-1990s Winter 1866 ± 76 777 ± 581 1586 ± 441 1980 ± 299 629 ± 268 Summer 1732 ± 120 1294 ± 101 3078 ± 1349 1650 ± 424 856 ± 263 2011 Winter 76 ± 19 164 ± 18 0 28 ± 21 57 ± 55 Summer 66 ± 60 108 ± 36 0 42 ± 7 81 ± 33 Leaf length Mid-1990s Winter 12.5 ± 2.9 17 ± 2.4 19.6 ± 3.9 12.1 ± 1.7 23 ± 4.3 Summer 24.1 ± 1.2 28.6 ± 1.5 18.6 ± 2.1 17.6 ± 2.9 21.2 ± 2.2 2011 Winter 12.1 ± 1.4 8.4 ± 1.1 - 3.6 ± 1.9 12.1 ± 3 Summer 24.1 ± 2.9 18.6 ± 2.7 - 6.4 ± 1.1 13.7 ± 1 C. nodosa biomass Mid-1990s Winter 262.5 ± 0.5 334 ± 255 404.2 ± 79.1 333.3 ± 54.3 128.7 ± 20 Summer 439 ± 78 942 ± 2 612.5 ± 335 366.7 ± 130.2 93.8 ± 84.2 2011 Winter 19.9 ± 1.9 56 ± 12 0 5.6 ± 5.7 17.2 ± 18 Summer 25.1 ± 12.6 45 ± 17.1 0 29.4 ± 27.8 54.5 ± 39.9 C. prolifera biomass Mid-1990s Winter 0 34.4 ± 28.5 18 ± 5.3 10.3 ± 17.9 19.8 ± 33.1 Summer 5 ± 5 108.5 ± 77.5 80.83 ± 91.1 18.6 ± 18.2 74.7 ± 54.8 2011 Winter 0 22.4 ± 20 0 0,7 ± 0.8 0,5 ± 0,9 Summer 4.8 ± 8.3 24.8 ± 28.9 0 0.5 ± 0.7 0 21  Table 3. Results of 3-way ANOVAs testing for differences in shoot density, leaf length and total biomass of Cymodocea nodosa and total biomass of Caulerpa prolifera between years (fixed factor, mid-1990s vs. 2011), seasons (fixed factor, winter vs. summer) and among sites (random factor). Source of variation df Shoot density Leaf length C. nodosa total biomass C. prolifera total biomass MS F P MS F P MS F P MS F P Year 1 2983467.34 23.57 0.0072 1388.88 10.38 0.0318 2013538.98 17.17 0.0132 4509.74 1.18 0.3568 Season 1 457111.28 1.21 0.3446 399.76 3.99 0.1166 164361.28 4.04 0.1158 325.77 0.06 0.8128 Site 4 1107301.13 7.88 0.0004 248.16 46.30 0.0002 119448.70 10.84 0.0002 1270.34 1.00 0.4062 Year x Season 1 486351.33 1.25 0.3304 0.33 0.09 0.7782 131414.64 2.45 0.1952 12780.93 22.39 0.0874 Year x Site 4 1399306.62 9.96 0.0002 133.74 24.96 0.0002 117234.94 10.64 0.0002 3807.91 2.99 0.0426 Season x Site 4 377483 2.69 0.0410 100.16 18.69 0.0002 40676.78 3.69 0.0118 5335.82 4.18 0.0128 Year x Season x Site 4 389980.17 2.78 0.0444 3.65 0.68 0.6 53626.35 4.87 0.0026 570.83 0.45 0.718 Residual 40 140423.20 5.36 11015.32 1275.14 22  Table 4. Results of 1-way ANOVAs testing for differences in shoot density, leaf length and total biomass of Cymodocea nodosa between experimental treatments (fixed factor) at the end of the experimental period. Source of variation df Shoot density Leaf length Total biomass MS F P MS F P MS F P Treatment 2 1030.16 20.75 0.0000 9.56 0.83 0.4489 26.62 17.22 0.0000 Residual 21 49.64 11.49 1.54 23  Figure legends Figure 1. Relationship between the total biomass of Cymodocea nodosa and Caulerpa prolifera at 16 sites throughout the island of Gran Canaria on winter and summer 2011. Linear regression models tested the significance of this relation separately for winter and summer 2011. Only the linear regression for summer is shown, as the winter regression was non-significant (p= 0.08). Figure 2. Differences in (a) shoot density, (b) total biomass, and (c) leaf lenght of Cymodocea nodosa between plots where complete (100%), partial (50%) and no (0%, controls) removals of Caulerpa prolifera were carried out. Different letters above bars denote significant differences. 24  Contributions of each author: Conceived and designed the study: FT, FE, RJH. Financially managed the study: FE, RJH. Performed the study in the field: FT, HH, FE. Performed the study in the lab: FT, HH. Analyzed the data: FT. Wrote the paper: FT. Commented on the paper: HH, FE, RJH. All authors have approved the final version of article.