Massive decline of Cystoseira abies-marina forests in Gran Canaria Island ( Canary Islands , eastern Atlantic )

Grupo en Biodiversidad y Conservación, IU-ECOAQUA, Universidad de Las Palmas de Gran Canaria, 35017, Las Palmas de G.C., Canary Islands, Spain. (JV) (Corresponding author) E-mail: joanvahe79@gmail.com. ORCID-iD: http://orcid.org/0000-0003-2062-4999 (MAV-R) E-mail: mviera@dbio.ulpgc.es. ORCID-iD: http://orcid.org/0000-0002-6423-619X (FE) E-mail: fesprod@gobiernodecanarias.org. ORCID-iD: http://orcid.org/0000-0003-3988-7297 (RH) E-mail: ricardo.haroun@ulpgc.es. ORCID-iD: http://orcid.org/0000-0003-1348-692X (FT) E-mail: ftuya@yahoo.es. ORCID-iD: http://orcid.org/0000-0001-8316-5887


INTRODUCTION
Coastal ecosystems are suffering severe impacts worldwide due to excessive human pressure.Habitat destruction, pollution, eutrophication, species introduction, overfishing and global warming, which often act synergistically, are affecting species, ecosystems and their ability to provide ecosystem services (Halpern et al. 2008).For example, the Canary Islands are a "hot spot" of marine biodiversity in the North Atlantic (Sansón et al. 2001), which is threatened by human impacts, e.g.pollution, overfishing, occupation of the coast and progressive tropicalization (Riera et al. 2015).
Along rocky shores of temperate and subtropical areas, large canopy-forming brown algae, in particular kelps (Laminariales, Phaeophyceae, Ochrophyta) and fucoids (Fucales, Phaeophyceae, Ochrophyta), are the dominant species in pristine environments (Schiel and Foster 2006).These large perennial macroalgae are considered as "engineering species" (Jones et al. 1994), because their three-dimensional structure dramatically alters the physical, chemical and biological environment.These forests provide shelter, food, habitat and nurseries for a multiplicity of species (Cheminée et al. 2013).The decline of kelps and fucoids is a global phenomenon due, directly or indirectly, to humanmediated activities (Wernberg et al. 2011, Lamela-Silvarrey et al. 2012, Franco et al. 2015).Some species have even been driven to regional and local extinction (Thibaut et al. 2005, Franco et al. 2015, Thibaut et al. 2016a).The loss of these well-structured and diverse ecosystems facilitates the appearance of less complex habitats, such as filamentous algal turfs, ephemeral seaweed assemblages and barren grounds dominated by encrusting algae and sea urchins (Benedetti-Cecchi et al. 2001, Ling et al. 2015).
The genus Cystoseira C. Agardh (Fucales, Phaeophyta) is distributed in temperate and subtropical coasts around the world, although 80% of the species live in the Mediterranean Sea (Oliveras and Gómez 1989).In the Mediterranean and the adjacent Atlantic Ocean, species of the genus Cystoseira are the main group of habitat-forming macroalgae, from the littoral to the lower limit of the euphotic zone (Giaccone et al. 1994, García-Fernández andBárbara 2016).Losses of Cystoseira forests have been reported all around the Mediterranean and attributed to habitat destruction, eutrophication and overgrazing by herbivores (Thibaut et al. 2005, Iveša et al. 2016, Blanfuné et al. 2016a).Due to their high sensitivity to anthropogenic impacts, several species of Cystoseira indicate high quality waters and facilitate the implementation of the EU Water Framework Directive (2000/06/EC) (Ballesteros et al. 2007, Blanfuné et al. 2016b, 2017).All the Mediterranean species of the genus Cystoseira, except C. compressa, have been protected under the Annex II of the Barcelona Convention (2010).Five species, Cystoseira amentacea, Cystoseira mediterranea, Cystoseira sedoides, Cystoseira spinosa and Cystoseira zosteroids, are protected under the Berne Convention (Annex I, 1979).In addition, all Mediterranean Cystoseira species are under surveillance by international organiza-tions, such as IUCN, RAP/ASP and MedPan (Thibaut et al. 2014).All species of Cystoseira are "habitatforming" and are therefore considered EU habitats of interest (Micheli et al. 2013).
The brown alga Cystoseira abies-marina (S.G. Gemelin) C. Agardh has been considered the most abundant fucoid species on rocky shores of the Canarian Archipelago (Wildpret et al. 1987, Tuya andHaroun 2006), and its populations typically form extensive stands in both the eulittoral and shallow sublittoral, mainly on rocky wave-exposed zones (Wildpret et al. 1987, Medina andHaroun 1993).C. abies-marina is a caespitose plant with large numbers of erect branches, up to 50 cm long.Similar to other species in the genera Cystoseira, this species undergoes an annual thallus loss at the end of summer, when a high proportion of the fronds break down at the base.The holdfasts overwinter and regrow the next year.Therefore, although individuals are perennial, the thalli are annual (Buonomo et al. 2017).However, the plant never goes through a total rest phase: during unfavourable months, branches from different seasons coexist (González-Rodríguez and Afonso-Carrillo 1990).This alga spreads through both vegetative propagation and sexual reproduction (Medina 1997).Similar to other species of the genus, thalli are negatively buoyant and propagules normally settle at <20-40 cm from the source population (Mangialajo et al. 2012), which gives the species a lowdispersal ability (Bulleri et al. 2002).This is one of the most productive macroalgae in the Canary Islands (Johnston 1969), and at the end of summer, after the maximum reproductive peak, it is possible to find a large amount of wrack on beaches from nearby forests (Portillo-Hahnefeld 2008).
In the last few decades, Cystoseira abies-marina forests have declined significantly at certain points of the Canaries (Medina andHaroun 1993, Rodríguez et al. 2008).In order to analyse the long-term patterns in the distribution and extension of C. abies-marina along the entire coastal perimeter of the island of Gran Canaria, we collected all available data to reconstruct historical distributions.The aims were: (i) to provide an up-to-date assessment of the current distribution and extent of C. abies-marina, (ii) to facilitate a comparison with historical data, including populations from certain sites, and (iii) to evaluate the influence of local anthropogenic pressures, as drivers of regression.

Study area
The island of Gran Canaria (28°51'N, 15°36'W) is located 200 km off the northwest African coast, in the middle of the Canary Islands (east Atlantic) (Fig. 1).The island has a circular shape of 256 km of coastal perimeter.Abrupt cliffs mostly dominate the north and west sides of the island, with coastal platforms and beaches predominating in the east and south.Although 76.02% of the coastal perimeter is rocky (Ramírez et al. 2008), rocky reefs only account for 17% of the shallow-water bottoms (up to 50 m).Gran Canaria is situated at the centre of a west-east oceanographic gradient along the Canarian archipelago, because of the varying proximity from the upwelling of the African coast (Tuya et al. 2006).The waters are typically oligotrophic and the surface temperature varies between 18°C in March and 24°C in October.

Mapping historical and current distribution: GIS analysis
Changes in the distribution (km of coastal perimeter) and extent (occupied area in ha) of Cystoseira abiesmarina over time were analysed with the open-source GIS (gvSIG) and Sextante tools, using a 1:2500 scale and a WGS-84/UTM Zone 28N coordinate system.
Historical records concerning Cystoseira abiesmarina distribution in the Canary Islands are scarce (Table 1).To reconstruct long-term patterns of change, we used unpublished reports from the late 1980s and 2000s, and oral scientific contributions.However, we did not take into account herbarium vouchers.The first map dates back to 1985, when Wildpret et al. (1987) defined and mapped 15 types of vegetation between 0 and 10 m depth: 12 correspond to stands of macroalgae, two to seagrass meadows and one to a mixed community of seagrass and algae.Additionally, they mapped three ecosystems devoid of vegetation.We digitalized six of these types of vegetation, in which C. abies-marina was the principal floral component (Supplementary material Table S1, Fig. S1A).We used complementary sources to enlarge this map from the 1980s, focusing mainly on the eastern side of the island.Information provided by scientists and technicians, which was contrasted with historical orthophotos (Vuelos históricos: 1989-1991 Costas, Instituto Geográfico Nacional),  supplied additional populations to those provided by Wildpret et al. (1987) (Supplementary material Table S1, Fig. S1B).
The first digitalized map of Cystoseira abies-marina was undertaken in 2008 by Rodríguez et al. (2008), who mapped the distribution of C. abies-marina according to three levels of abundance: as continuous belts, as discontinuous belts and as isolated individuals (Fig 2B).
Field surveys were carried out between 2015 and 2016, during the maximum development of Cystoseira abies-marina (spring to autumn).The entire coast of Gran Canaria was explored on foot or by boat, and the shallow subtidal by snorkelling.Locally, populations were categorized, following Rodríguez et al. (2008), as C1, rare scattered patches; C2, abundant patches; and C3, continuous belts.All the C. abies-marina populations were geo-localized and recorded on A4 format aerial photographs from the IGN (Instituto Geográfico Nacional, 1:2500 scale).Rodríguez et al. (2008) studied seven populations (Fig. 1), providing the average coverage and belt width of Cystoseira abies-marina forests.In 2016, we repeated the study in the same locations, carrying out three transects (ca. 10 m apart), which covered the entire eulittoral and the shallow subtidal.Along each transect, the coverage (n=3) of C. abies-marina was obtained with a square (50×50 cm), divided into 25 sub-squares of 10×10 cm; the belt width was measured with a transect.We tested for differences in average coverage and belt width between 2008 and 2016 using a Wilcoxon test (i.e.all populations were pooled).

Human pressures as drivers of change
We assessed the Human Activities and Pressures Index (HAPI) (Blanfuné et al. 2017) on the coast of Gran Canaria.Five water bodies (WD) surround the island, according to the Water Framework Directive (WFD, 2000/60/EC).We divided these WD into 28 coastal sectors (grid cells of 5×5 km, Fig. 1) to identify more precisely the relationship between levels of anthropogenic pressures and the decline of Cystoseira abies-marina forests.For this study, we adapted the information available for the Canary Islands, following the method of Blanfuné et al. (2017).
The HAPI index has three metrics to estimate both continental and marine pressures.For continental pressures (urban, industrial and agricultural areas), the three metrics were expressed as the percentage of land area covered (data from Corine Land Cover 2012, available at centrodedescargas.cnig.es)within each coastal sector.For marine pressures, we estimated (i) the level of artificialization of the coast, expressed as the percentage of the artificialized coastline, (ii) fish farms, expressed as the percentage of rocky coastline potentially impacted (within a 500 m radius), and (iii) sewage outfalls, expressed as the percentage of rocky coastline potentially impacted (within a 500 m radius).This information was provided by the on-line GIS of the Canary Islands Autonomous Government (www.idecanarias.es).For each sector, we calculated the change in the extent of C. abies-marina between 1980s (i.e.Wildpret et al. 1987) and 2016 (i.e. this study).We applied a linear regression to test whether varying levels of human pressures explained the magnitude of changes in surface area over time at the island scale.Additionally, the HAPI index was calculated for each of the seven populations under study; we calculated the level of human pressures using a 500 m radius circular buffer from the centre of each population (Fig. 1), following a similar approach to that of Tuya et al. (2014).

Distribution and extent
During the 1980s (Fig. 2A), Cystoseira abies-marina dominated the rocky coasts of Gran Canaria, along 120.5 km of coastal perimeter, covering 928 ha.It was abundant on most rocky substrates and the populations were mainly composed of continuous belts (Fig. 2A).Subtidal populations reached up to 9 m depth in many places of the north coast; in the east and southeast coast, stands reached up to 20 m depth in some places.The species was absent from the south and southwestern coast, mainly due to a lack of suitable hard substrates.At the start of the 21st century (Fig. 2B), populations were clearly fragmented, occupying 52.2 km of the coast (19.45% of the coastline) and covering 12.6 ha; this corresponds to a regression of 98.64% in 20 years.Populations rarely get down into the subtidal, except in a few locations in the north and east, where populations go down to 8-10 m. Between 2014 and 2016, C. abies-marina was present along 37.8 km (14.08% of the coastline) and occupied an area of only 7.4 ha.Populations forming continuous belts have practically disappeared (0.3 ha).Fragmented populations are becoming more prominent and sublittoral populations have totally disappeared.As a result, ca.99% of the area formerly covered by C. abies-marina has been lost in a few decades (Fig. 3).

Comparison of populations: 2008 vs 2016
Overall, the seven Cystoseira abies-marina populations studied in 2016 have suffered a significant decline relative to 2008, in terms of coverage (V=231, P=0.00006, Fig. 4A) and belt width (V=231, P=0.0000001, Fig. 4B).In 2008, all populations had high cover (>50%) and formed continuous belts; in some localities, belts extended to the subtidal down to 8-10 m depth.

Human pressures as drivers of regression
Values of the HAPI index were calculated for the 28 coastal sectors and 7 populations of Gran Canaria Island (Tables 2 and 3, Supplementary material Table S3).There was no significant effect of varying

DISCUSSION
Changes in the distribution and extent of Cystoseira abies-marina on the island of Gran Canaria over the last few decades are evident and dramatic.In the late 1980s, C. abies-marina occupied 928 ha (12.84% of the rocky bottoms) and now it only covers 7.4 ha (0.1%).Existing C. abies-marina populations have been reduced to narrow belts in the lower eulittoral, i.e. as scattered patches with underdeveloped branches.Our results are in agreement with those found for other Cystoseira species from the Mediterranean Sea (Thibaut et al. 2005, Mangialajo et al. 2008, Sales et al. 2011), for other fucoids from the Canary Islands (Rod-ríguez et al. 2008, Riera et al. 2015) and, in general, for habitat-forming brown algae worldwide (Wahl et al. 2015).Our data show a similar trend to that observed for C. brachyccarpa var.brachycarpa, a species having the same depth range and ecological function as C. abies-marina, including a massive decline from the sublittoral to a narrow fringe immediately below the surface (Thibaut et al. 2015(Thibaut et al. , 2016b)).
It is plausible that the area occupied by Cystoseira abies-marina in the 1980s is not entirely accurate, because of the lack of technical procedures to accurately trace communities at this time.The map of Wildpret et al. (1987) only reached 10 m depth, so they may even have underestimated the area occupied by C. abiesmarina.Even assuming these inaccuracies, the regression of C. abies-marina is acute, in particular because all sublittoral populations have been lost.
In our study, we found no direct relationship between local levels of anthropogenic pressures and the magnitude of local regression; the decay has been dramatic from almost pristine environments to highly altered coasts.This result contrasts with the disappearance of some Cystoseira species only from highly artificialized areas (harbours, marinas, piers, etc.) and waters severely polluted in the Mediterranean (Thibaut et al. 2014, 2015, Iveša et al. 2016).However, a similar decline has been observed in the pristine environments of the National Park of Port-Cros in France (Thibaut et al. 2016b).In a similar study, the temporal decay in the vitality of the seagrass Cymodocea nodosa in Gran Canaria was connected with an increase in local anthropogenic impacts (Tuya et al. 2014).
The decline of C. abies-marina in Gran Canaria has occurred in a period of pronounced urban and tourism development and, therefore, of many local impacts (Tuya et al. 2014, Ferrer-Valero et al. 2017).Today, the population, urbanization and infrastructure are heavily concentrated on the coast of the island, particularly in the northeast, east and south (Fig. 1).Gran Canaria currently has 847830 inhabitants and a very high population density (543 inhabitants km -2 ) (ISTAC 2015); 87% of the population is located along the littoral perimeter, giving a coastal population density of 3142 ind km -1 .In addition, about 2 million tourists visit the island every year (e.g.1805058 tourists in 2015, ISTAC 2015)).This has led to the occupation and degradation of most coastal areas (Ferrer-Valero et al. 2017).Importantly, however, populations of C. abies-marina in poorly impacted areas have also suffered significant regressions.Hence, it remains elusive to unravel the reasons for the loss of C. abies-marina.
The possible causes of the decline may be multiple and cumulative, as happens around the world (Thibaut et al. 2005, Wahl et al. 2015, Franco et al. 2015).Potentially, both local and global stressors are interacting to explain the severe regression of Cystoseira abies-marina in Gran Canaria, as is the case with the disappearance of other fucoids from the study region (Riera at al. 2015).In the Canary Islands, sea surface temperature has increased about 1°C in recent decades (Lima and Wethey 2012, Riera . 2015).Global warming is a key factor in the ongoing decline of fucoids and their displacement to colder waters (Wernberg et al. 2011).There is recent regional evidence of the adverse effect of warming on species of both brown and red macroalgae (Sansón et al. 2013).The decrease in the size of thalli of these seaweeds, and in their reproductive success (Zhang et al. 2009), have also been correlated with the warming of the Canarian waters (Sansón et al. 2013, Riera et al. 2015).Furthermore, Cystoseira are low-dispersal species whose propagules do not have a planktonic stage, and reproductive drifting thalli in floating rafts are the main mechanism of connectivity between populations (Susini et al. 2007).Therefore, if connectivity is limited, the subsequent smaller population gene pools and sizes render populations more vulnerable to threats (Buonomo et al. 2017).
The regime shift from marine forests to barren grounds devoid of erect macroalgae is generally linked to overexploitation of predatory fishes (Ling et al. 2015, Thibaut et al. 2015, 2016a).In the Canary Islands, the long-spined sea urchin Diadema africana controls the transition from rocky bottoms dominated by erect macroalgae to barren grounds (Tuya et al. 2004, Sangil et al. 2014).This sea urchin may consume thalli of C. abies-marina at rates of 1-2 mg of algae per day and individual (Tuya et al. 2001).In addition, it is plausible that the sea urchin Paracentrotus lividus and herbivorous fishes (Sparisoma cretense, Sarpa salpa and Diplodus spp.) can contribute to the consumption of C. abies-marina, as in the Mediterranean for other Cystoseira spp.(Verges et al. 2009).
This study highlights the urgent need to monitor remaining Cystoseira abies-marina populations of the Canary Islands, and compare this data with other Macaronesian islands.It is also necessary to promote urgent actions to conserve current populations, including restoration programmes.Currently, C. abies-marina is regionally protected within the framework of the Canary Islands Catalogue of Protected Species (Law 4/2010, of 4 June 2010).This highlights the uselessness of legislation if it is not enforced.Furthermore, in the last update of this catalogue, the species lost the category of "vulnerable": it now belongs to a recently created category called "species of interest for the Canarian ecosystems", which only protects the species within zones of the Natura 2000 network.Our results clearly do not support this legislative change.
Cystoseira abies-marina has not yet been assessed for the IUCN Red List, i.e. it is "Not Evaluated" (IUCN 2017), nor is species included in the Catalogue of Life (Roskov et al. 2016).We are aware that C. abies-marina is undergoing a very important decline throughout the Canaries (Rodríguez et al. 2008) and also on Madeira and the Azores (Ballesteros pers. com.), but more data are needed to verify the magnitude of this decline.Nevertheless, with current data and evidence of the regional decline, we propose that C. abies-marina should be classified as "Critically Endangered" under the IUCN criteria CR A2ac.There are only very few algal species in the world whose conservation status has been properly assessed (Blanfuné et al. 2016a), due to lack of historical data.Information provided here could be used as a basis for improving the evaluation of the conservation status of C. abies-marina, an ecologically important species.

Fig. 1 .
Fig. 1. -Map of Gran Canaria Island, including the 28 sectors (grids of 5×5 km) encompassing the entire coastal perimeter.The location of the seven analysed populations is also shown, with the circular area (500 m radius) where the HAPI index was calculated.

Fig. 3 .
Fig. 3. -Temporal changes in the extent of Cystoseira abies-marina, in terms of the surface in hectares (A) and length in kilometres (B) of the coastline occupied at different times.

Table 1
. -Cartographic sources on the distribution of Cystoseira abies-marina in Gran Canaria Island.

Table 2 .
-Values of the HAPI index, area covered and rate of temporal change of Cystoseira abies-marina in each of the 28 sectors along the coastal perimeter of Gran Canaria.

Table 3 .
-Values of the HAPI index, coverage and rate of temporal change for each of the seven studied Cystoseira abies-marina populations.