Introduction

Species distributions are shifting globally in response to climate change, with large variability in responses among taxa and regions (Lourenço et al. 2016). Therefore, describing and reporting changes in species ranges are necessary for the successful assessment of the impact of contemporary climate variability on species distributions (Johnson et al. 2011). Among the drivers of coastal species, abundances and distributions that are likely to undergo dramatic change is upwelling. Upwelling has important effects on the distribution of coastal marine organisms (Fenberg et al. 2015; Reddin et al. 2015; Cefalì et al. 2016) through several mechanisms. These include the enhancement of primary production by bringing cold, nutrient-rich water to the surface, an influence on local/regional recruitment of larvae through the advection of nearshore waters (e.g. Barshis et al. 2011; Moyano et al. 2014; Fenberg et al. 2015), and providing cooler areas that allow species persistence (e.g. Hu and Guillemin 2016; Lima et al. 2006; Lima et al. 2007; Lourenço et al. 2016).

Upwelling is spatially and temporally heterogeneous while its strength is site-specific (e.g. Wang et al. 2015; Sousa et al. 2017), so that its effects on intertidal assemblages differ and can be community-specific, for instance by causing cascading effects on the composition of the intertidal biota (Nielsen and Navarrete 2004; Guerry and Menge 2017). Consequently, variations in the frequency and intensity of upwelling can determine the functional and trophic structure of intertidal communities (Bosman et al. 1987; Broitman et al. 2001; Blanchette et al. 2009; Reddin et al. 2015 but see Kelaher and Castilla 2005; Puccinelli et al. 2016a; Puccinelli et al. 2016b).

The Benguela (Bustamante and Branch 1996a), Humboldt and California Current (e.g. Broitman et al. 2001; Blanchette et al. 2008, Blanchette et al. 2009) upwelling systems are major upwelling systems that have long been the focus of studies aimed at describing and comparing the biological and environmental structure of intertidal rocky shore communities. In contrast, the Canary Current upwelling system, another major eastern boundary upwelling system, remains largely unexplored. Geographic data on current species distributions along northern African shores are so limited and outdated (e.g. Fischer-Piette 1957; Fischer-Piette and Prenant 1957) that the Census of Marine Life (http://comlmaps.org/mcintyre) describes this coastline as a major biodiversity gap requiring improved taxonomy and an improved understanding of the scales of temporal and spatial variability in nearshore habitats (Stuart-Smith et al. 2015). At the same time, analysis of the vulnerability of the world’s shallow marine fauna based on their thermal preferences indicates that this region is among the most sensitive to long-term climate warming. Despite numerous studies of the oceanography and upwelling dynamics of the system (e.g. Marcello et al. 2011; Benazzouz et al. 2014; Cropper et al. 2014; Sousa et al. 2017) and focussed studies on groups of species (e.g. algae, Benhissoune et al. 2001, Benhissoune et al. 2002b, Benhissoune et al. 2002a, Benhissoune et al. 2003), we are aware of no large scale investigation linking environmental conditions and biological gradients among intertidal communities along this coast. Furthermore, the region includes an important biogeographic transition. The Iberian and north African shores are influenced by the Canary Current upwelling system and represent a biogeographic transition where warm- and cold-water species reach their northern and southern distributional limits, respectively (e.g. Smale et al. 2013; Neiva et al. 2015; Assis et al. 2017). Not only is the region strongly affected by ongoing climate change (Belkin 2009; Lima and Wethey 2012), with marked shifts in the ranges of ecosystem-structuring species in response to warming conditions (e.g. the macroalga Fucus vesiculosus, Nicastro et al. 2013), but reports of new species from Morocco (e.g. Hassoun et al. 2014; Belattmania et al. 2017) suggest that it may harbour higher levels of biodiversity than recognized.

Here, we describe patterns of diversity and abundance of intertidal rocky shore species along c. 2000 km of the Atlantic coast of Morocco and Western Sahara and identify upwelling-based drivers that influence to these patterns. Specifically, we: (a) create a baseline for future studies investigating climate-driven shifts in the distribution of intertidal rocky shore species along the north African Atlantic coast and (b) assess and relate biological (intertidal community) and environmental (upwelling) structure along the Canary Current system.

Material and methods

Study region

Qualitative and quantitative field surveys were conducted at 12 intertidal rocky shore sites in the Canary Current upwelling system (CCS) along the Atlantic shores of Morocco and Western Sahara (Marcello et al. 2011; Benazzouz et al. 2014; Table S1 supplementary material). Sites were sampled between September 2013 and October 2014. Sites were roughly equidistantly distributed along the region and selected based on similarity in wave exposure, habitat type, topography and proximity to upwelling cells. Due to the inaccessibility of Nouifed (24°54′30.29″N; 14°49′45.36″W) during the second survey, the closest accessible rocky shore, Hassi El Kraa (24°41′06.18″N; 14°54′08.87″W), was selected as its replicate (approx. distance 22 km; Table S1 in supplementary material; Fig. 1).

Fig. 1
figure 1

Study site and species richness. Sites sampled and species richness at 12 sites along the coasts of Morocco and Western Sahara, categorized by functional group

Full size image

The CCS comprises multiple upwelling cells that vary in timing and intensity (Marcello et al. 2011; Benazzouz et al. 2014). Three main centres of upwelling can be detected through low sea surface temperatures (SST) or upwelling indices along this stretch of coast: the first at 31–32°N (north of Cap Ghir), a second at 26.5–28°N (south of Cap Juby) and a third at 21–25°N (north of Cap Blanc; Marcello et al. 2011; Benazzouz et al. 2014). Adjacent surrounding areas are intermittently affected by upwelling, the effects of which decrease as the distance from the upwelling centres increases (Marcello et al. 2011; Benazzouz et al. 2014). Northern Morocco (33–36°N) shows low seasonality and weak upwelling indices, central and south Morocco and northern Western Sahara (26–33°N) show the strongest seasonality and the highest upwelling indices (peak during late summer, August–September; (Marcello et al. 2011; Benazzouz et al. 2014), while central and southern Western Sahara (21–26°N) show high upwelling indices and very little seasonality (Benazzouz et al. 2014).

Environmental variables

Site-specific monthly sea surface temperature (SST) data with a 4-km resolution were retrieved from the Moderate Resolution Imaging Spectroradiometer-Aqua (MODIS-Aqua) dataset available from the National Aeronautics Space Administration (NASA) Goddard Earth Sciences (GES) Data and Information Services Center (DISC) for the period from January 2010 to December 2014 using Giovanni, a web-based application developed by the NASA GES DISC. An area of 25 km2 situated 5 km offshore of each sampling site was selected to investigate annual minimum and maximum SST and SST variation (maximum SST–minimum SST). Annual minimum and maximum SST were obtained by selecting the lowest and highest monthly SST over each year, respectively. Annual minimum and maximum SST and SST variation values were then averaged over the 5-year period considered.

The wind-based upwelling index cross-shore Ekman transport (CSET) was used to estimate upwelling intensity during the period 2010–2014 following (Krug et al. 2017). Daily sea surface wind fields (speed and direction) at a spatial resolution of 0.25o were obtained from the Blended Sea Winds dataset (National Climatic Data Centre—National Oceanic and Atmospheric Administration, NCFC-NOAA, http://www.ncdc.noaa.gov/oa/rsad/air-sea/seawinds.html). The Blended Sea Winds dataset combines multiple scatterometers standardized across platforms, resulting in high-quality temporal and spatial coverage of ocean wind vectors (Zhang 2006). CSET values were estimated for 13 locations along the meridionally oriented Atlantic coast of northern Africa (Table S1 in supplementary material). For each location, CSET values represented the average of a 0.75 × 0.75o box centred on the target location.

The zonal component of the Ekman transport (CSETx), induced by the meridional component of wind-stress (τy), was used as an upwelling index for each station. CSETx (m3 s−1 km−1 coastline) was calculated following Bakun (1973), as modified by Alvarez et al. (2011):

$$ {\mathrm{CSET}}_x=\frac{\ 1000\ {\mathrm{T}}_{\mathrm{y}}\ }{\uprho_{\mathrm{w}}\ \mathrm{f}} = \frac{\rho_{a\ {C}_d}\ }{\uprho_{\mathrm{w}}\ \mathrm{f}\ }\ {\left({W}_x^2+{W}_y^2\right)}^{1/2}\ 1000\ {W}_{\begin{array}{c}y\\ {}\ \end{array}} $$

where W represents wind velocity (m s−1), ρw is seawater density (1025 kg m−3), ρa is air density (1.22 kg m−3), Cd is the drag coefficient (1.4 × 10−3) and f is the Coriolis parameter, estimated as 2 Ω sin(θ) where Ω and θ represent the vertical component of the Earth’s angular velocity and local latitude, respectively. Negative CSETx values indicate upwelling-favourable periods with offshore Ekman transport and conversely, positive values indicate downwelling-favourable periods and onshore Ekman transport.

Daily CSET data were first monthly averaged to reduce the influence of daily anomalies. Over each year, the lowest CSET monthly average was identified to obtain the annual minimum. The overall minimum (UImin) and mean (UImean) upwelling indices for the 5-year period (2010–2014) were estimated for each site and used in statistical analyses. The UImin was obtained by selecting the lowest upwelling index of the 5-year period, while the overall UImean was obtained by averaging the five minimum annual CSET (i.e. one value per year). Annual probability of upwelling (UIp in %) was calculated as the average of the monthly frequency of upwelling favourable days (CSET values < 0). The overall UIp was the average over the 5-year period.

Biological sampling design

A point-intercept sampling method was used to quantify relative abundance (% cover) of sessile invertebrates, macrophytes and lichen species at each site (adapted from Blanchette et al. 2008). A representative shore section was designated at each site and a measuring tape was laid out from the upper edge of the highest intertidal barnacle zone, perpendicular to the shore, to the lowest level of the low tide. Vertical point-intercept transects (n = 2 each site) were divided into 50 equidistant points. Intervals between points were adjusted at each site and depended on the width of the shore. The five species under each point, including layering and epibionts, or closest to the point directly attached to the substratum were recorded. Tide pools and gulleys were not sampled to avoid a misrepresentation of the intertidal height. If an intercept point fell on a tide pool or gulley, the closest horizontal non-tide pool/inundated area was sampled instead.

When species identification was not possible in the field, specimens were collected for identification in the laboratory. Algae and lichen specimens were preserved in KEW solution (40% ethanol (70%), 40% seawater, 10% glycerine and 10% formaldehyde (4%)) and sessile invertebrates were preserved in 96% ethanol. A random search of 15 min at each site was performed to include species that did not comprise one of the five taxa at each point but which were present along the transect. Species were identified and accounted for in the overall qualitative description of the site’s community composition, but that were not considered in the statistical analyses.

The abundances of mobile species were determined using 30 × 30 cm quadrats placed along the transect following (Engle 2008). Specifically, three quadrats were placed haphazardly on the substratum in each of the low, mid and high shore and the macroinvertebrate target taxa (limpets, gastropods and pulmonate species) > 5 mm found within the quadrat were identified and counted. The abundance of littorinids (mostly < 5 mm) was only determined in the high zone and this species were sub-sampled in a 7.5 × 10 cm section of the quadrat due to their high densities. When species identification was not possible in the field, specimens were collected and preserved in 96% ethanol for further morphological or genetic identification in the laboratory. Again, tide pools and gulleys were not sampled to avoid a misrepresentation of the intertidal height.

Main sources for identification were: Sansón and Carrillo 1999, Gómez-Garreta 2002, Brodie et al. 2007, Cabioc’h et al. 2006, Rodríguez Prieto et al. 2013, Fish and Fish 2011, Preston-Mafham 2010, taxonomic notes and references from Algaebase (http://www.algaebase.org/).

Historical data on the distributions of algae in northern Africa described in the literature and in Algaebase (http://www.algaebase.org/) were used as a baseline for the distributional patterns of the species identified in this work. Published literature was screened up until January 2017 using Google Scholar and the ISI Web of Knowledge by using the names of the species identified in the present study in combination with the following keywords: Morocco, Maroc, Western Sahara, Spanish Sahara. New local records depicted novel descriptions of a species at a site, despite its confirmed presence in the country. A new southern limit recorded a species farther south than its previous historical limit. A new record was defined as the first record of a species from Moroccan or Western Saharan shores.

Data analyses

Site-specific species richness was estimated by summing the total number of taxa identified at each site from both transect and quadrat surveys. Species were categorized into functional groups based on their feeding guilds (macrophytes, filter-feeders, herbivores and lichens (as in Blanchette et al. 2009).

To examine similarity of spatial patterns in the biological and environmental data along the study area and to understand if the composition of intertidal communities was influenced by upwelling-related variables, the multivariate methods of Clarke (Clarke 1993) in PRIMER 6.1.3 (Plymouth Routines in Multivariate Ecological Research) software package were used. Abundances of sessile and sedentary species on transects were calculated by determining the total percentage (%) of species presence detected along the 50 points of each transect. Abundances of mobile species in quadrats were determined by estimating species density in each quadrat. Taxon abundance was averaged across sampling replicates (transects or quadrats) for each site. The data matrix of taxon abundances was fourth-root transformed to reduce the contribution of very abundant species and increase that of rare species (as in Blanchette et al. 2009). A biological similarity matrix was constructed using the Bray–Curtis similarity coefficient and cluster analysis was performed using a hierarchical method with group-average linking. Environmental data were normalized after fourth-root transformation and a similarity matrix was constructed using Euclidean distance. A SIMPROF test was run for the biological and environmental dendrograms separately using 9999 permutations to indicate group structure at a significance level of 5%.

The SIMPER routine was performed to identify the taxa of each group that were most responsible for the differences among groupings, with a cut-off of 25% contribution. Sites were assigned to groups defined a priori based on SIMPROF analyses of the biological dendrogram.

Two-dimensional, non-metric multidimensional scaling (nMDS) was performed on the environmental variables to examine regional segregation among sites (Kruskal and Wish 1978).

The RELATE routine was used in PRIMER to match the environmental resemblance matrices with the resemblance matrices of taxon richness, abundance of functional groups, taxon abundance based on transects (TAT) and based on quadrats (TAQ) separately, running 9999 permutations under the Spearman rank correlation method at a significance level of 5%.

Distance-based linear models (DistLM) were carried out to determine the contribution of the environmental variables to the variability in community composition. DistLM analyses were performed through a dissimilarity matrix, using the ‘all specified’ selection procedure under the Akaike Information Criterion (AIC), performing 9999 permutations for taxon richness, abundance of functional groups, TAT, TAQ, and presence/absence for the taxa identified in transects, separately. The environmental variables were analysed individually (marginal tests) and a sequential test was employed to evaluate the cumulative effect of each variable once the previous variable(s) had been accounted for.

Results

Environmental data

The environmental variables were averaged over the 5-year period (2010–2014) for each site (Table 1) and analysed to detect geographical clustering of sites. The minimum (UImin) and mean (UImean) values of wind-based upwelling indices ranged from − 1723.51 m3 s−1 km−1 coastline (at Imsouane, site S6) to − 418.99 m3 s−1 km−1 coastline (at Rabat, site S2) and − 1543.06 m3 s−1 km−1 coastline (at Imsouane) and − 312.55 m3 s−1 km−1 coastline (at Rabat, site S2), respectively. Favourable wind-conditions for upwelling phenomena occurred between 68.9% (at Rabat) and 95.73% (at Dakhla, site S12) of the time over the 5-year period. Maximum and minimum sea surface temperature (SSTmax and SSTmin) ranged between 20.41 °C (at Essaouira, site S5) and 23.33 °C (at Rabat) and between 16.08 °C (at Imsouane) and 17.12 °C (at Tarfaya, site S9), respectively. SST variation (SSTv) ranged from 4.05 °C (at SNouifed/Hassi El Kraa, site 11) to 6.94 °C (at Rabat).

Table 1 Summary of the environmental variables analysed at each sampling site
Full size table

Hierarchical cluster analysis based on the six environmental variables revealed significant geographical structure (Fig. 2). The SIMPROF test identified three significant groups (E1, E2 and E3). Group E1 contained Imsouane (site S6) only. Group E2 comprised sites Larache (site S1), Rabat (site S2), Sidi Bouzid (site S3) and El Beddouza (site S4; the four northernmost sites). Group E3 comprised sites Essaouira (site S5) and Mirleft (site S7), El Ouatia (site 8), Tarfaya (site S9), Boujdour (site S10), Nouifed/Hassi EL Kraa (site S11) and Dakhla (site S12; central and southern sites).

Fig. 2
figure 2

Dendrogram based on environmental variables. Euclidean distance dendrogram of the similarity of sites based on environmental variables. Labelled nodes depict significant clustering (p < 0.05)

Full size image

Biological sampling

A total of 186 taxa (26 Ochrophyta, 107 Rhodophyta, 26 Chlorophyta, 2 Ascomycota, 3 Cnidaria, 16 Mollusca, 1 Annelida, 4 Arthropoda, 1 Chordata) inhabiting the intertidal shores of Atlantic Morocco and Western Sahara were identified (Table 2). A considerably greater number of algal taxa were identified in comparison with lichens or animals. Algae constituted 85.5% of the identified taxa, while animals and lichens constituted 13.4% and 1.1%, respectively. The surveys reported 376 new local records of algae; new overall southern limits were detected for 25 algal species and nine algal species were recorded for the first time from the study area. Overall, distribution novelties or changes were described for 89% (141 species) of the algal taxa identified.

Table 2 List of all species recorded along intertidal Atlantic Moroccan and Western Sahara shores
Full size table

Taxon richness varied along the study area, but did not follow a clear latitudinal gradient (Fig. 1). However, the four southernmost sites showed a trend of decreasing taxon richness towards the south. Average taxon richness across all sites was 46, but some sites had particularly low (29 taxa, Mirleft and El Ouatia) or high (62 taxa, Tarfaya) richness. Macrophyte, filter-feeder, herbivore and lichen taxa ranged between 15 and 49, 4–8, 6–10 and 0–2 taxa per site, respectively.

Hierarchical cluster analyses based on taxon abundance from the transect (TAT) matrix revealed significant geographic structure that was not related to latitude, while no significant structure was detected for taxon abundance from quadrats (TAQ; Fig. 3). The SIMPROF test performed on TAT data identified three significantly different clusters (groups B1, B2 and B3). Group B1 included only Imsouane. Group B2 comprised sites Larache, Rabat, Mirleft and El Ouatia at a Bray-Curtis similarity of around 64%. Groups B1 and B2 showed significant similarity of around 55%. Group B3 comprised sites Sidi Bouzid, El Beddouza and Essaouira and Tarfaya, Boujdour, Nouifed/Hassi El Kraa and Dakhla at a similarity level of 53%.

Fig. 3
figure 3

Dendrograms based on community composition. Bray-Curtis similarity dendrograms based on the community composition of a taxon abundance from transect data and b taxon abundance from quadrat data. Labelled nodes depict significant clustering (p < 0.05). B1–B3 refers to the clustering groups

Full size image

The SIMPER results showed that a total of 19 algae, two animal taxa and one lichen contributed the most to the dissimilarities among the three groups (B1B2, B1B3 and B2B3, Table 2). While seven taxa contributed the most to differences between groups B1 and B2 (dissimilarity of 45.04%), 11 and 12 taxa contributed the most to differences between groups B1 and B3 (dissimilarity of 51.01%) and B2 and B3 (dissimilarity of 50.82%), respectively (Table 3). SIMPER contributions based on TAT per site highlighted dissimilar distributional arrangements and changes in species abundance among and within groups (Fig. 4). Species that most contributed to dissimilarities among groups were sessile species, namely the algae species Bifurcaria bifurcata, Ulva clathrata, Osmundea pinnatifida, Padina pavonica, Codium adhaerens, Jania rubens and Fucus guiryi. Abundance of B. bifurcata gradually increased from group B3 (Sidi Bouzid) to its maximum abundance in group B1 (Imsouane; Fig. 4) and was absent south of Imsouane. Ulva clathrata was exclusive to Imsouane (group B1), where it was the most abundant, and Boujdour (group B3). Osmundea pinnatifida was present at all sites of group B3, but absent from the remaining two groups. Padina pavonica, C. adhaerens and J. rubens were relatively abundant in group B1 (Imsouane), but absent or extremely rare in the other two groups. Finally, F. guiryi was abundant at most sites of group B3, with a gradual increase of abundance from Sidi Bouzid to Essaouira, but it was absent from groups B1 and B2.

Table 3 Results of Simper analysis
Full size table
Fig. 4
figure 4

SIMPER results. Geographic pattern of distribution and abundance for the species that contributed most to the dissimilarity between groups B1B2, B1B3 and B2B3. Sites belonging to groups B1, B2 and B3 are represented in light grey, grey and dark grey, respectively. Species abundance is represented by total percentage (%) of presence detected along the 50 points of each transect, averaging the abundance of each of the two replicates at each site. Study sites are as in Table 1

Full size image

Abundances of macrophytes, filter feeders and lichens differed among the study sites, but did not follow a latitudinal gradient (Fig. 5). Macrophytes were the only functional group present at all sites. Lichens were the least abundant. The highest abundance for filter feeders was observed at Imsouane, Mirleft and El Ouatia.

Fig. 5
figure 5

Cover abundance. Abundance by cover of sessile or sedentary species at 12 sites along the coasts of Morocco and Western Sahara, for functional groups that contributed most to the dissimilarity between groups

Full size image

Biological-environmental comparison

The RELATE routine did not identify significant similarity between the environmental variables and taxon richness (Rho = − 0.163, p = 0.086), abundance of functional groups (Rho = − 0.079, p = 0.262), TAT (Rho = 0.121, p = 0.795), TAQ (Rho = 0.022, p = 0.584) or presence/absence data from taxa identified on transects (Rho = 0.098, p = 0.762). However, both biological (transect based) and environmental hierarchical cluster analyses showed similar clustering structure. Site S6 (Imsouane) was depicted as an outlier group in both cluster analyses. Additionally, both cluster analyses grouped sites Larache and Rabat in B2/E2 and Essaouira and Tarfaya, Boujdour, Nouifed/Hassi El Kraa and Dakhla in B3/E3.

DistLM mostly retrieved non-significant contributions of the environmental variables to the variability of taxon richness, abundance of functional groups, TAT, TAQ and presence/absence data from taxa identified in transects (most marginal and sequential tests on environmental variables p > 0.05, Table S2 in supplementary material). Although marginal tests on the individual environmental variables explaining TAQ were all non-significant (all p > 0.05, Table S2 in supplementary material), sequential tests showed that UImin, UImean, UIp and SSTmax together explained 47% of the variation (p = 0.0426, Table S2 in supplementary material). Additionally, a DistLM analysis retrieved a significant contribution of UIp (marginal tests, p = 0.0357) to the variability of presence/absence data from taxa identified in transects, explaining ~ 16% of the variation (Table S2 in supplementary material).

Discussion

The results of this study suggest that strong upwelling conditions influence community structure of intertidal benthic biota of Atlantic Moroccan and Western Sahara shores. Here, we further discuss the intertidal biodiversity of the region and the importance of upwelling as a thermal buffer in the context of climate change.

Upwelling influence on intertidal benthic communities

The distribution and abundance of coastal marine species are strongly influenced by large-scale oceanographic processes (Bosman et al. 1987; Broitman et al. 2001; Blanchette et al. 2008).The environmental analysis structured our study area into three groups: an outlier location (site Imsouane, group E1, western Morocco), a northern region (Larache, Rabat, Sidi Bouzid and El Beddouza, group E2, northern Morocco) and a southern region (Essaouira and Mirleft, El Ouatia, Tarfaya, Boujdour, Nouifed/Hassi El Kraa and Dakhla, group E3, southern Morocco and Western Sahara). Imsouane was most likely identified as an independent group due to the combined effects of the strongest upwelling indices (UI), some of the lowest maximum sea surface temperature (SSTmax) and the lowest minimum SST (SSTmin), highlighting a location characterized by the coldest water conditions and most intense upwelling of the entire study area. In sharp contrast, E2 combined the weakest UI with the lowest probability of upwelling (UIp) and the greatest SST variation, which matches previous studies describing northern Morocco as a region characterized by weak upwelling indices (Marcello et al. 2011; Benazzouz et al. 2014; Cropper et al. 2014). In agreement with the described intense upwelling events and the lack of seasonality along southern Morocco and Western Sahara due to conditions that are favourable to year-round upwelling (Marcello et al. 2011; Benazzouz et al. 2014), E3 revealed strong upwelling indices and the greatest probability of upwelling.

Because they can interrupt the general pattern of warmer SST towards the equator, oceanographic features such as fronts, currents or upwelling cells cause latitudinal thermal discontinuities that influence the distribution of intertidal assemblages (e.g. Blanchette et al. 2008; Ling et al. 2009). Latitudinal discontinuities associated with upwelling are key elements in explaining the community structure of benthic intertidal biota (e.g. Humboldt Current system: Broitman et al. 2001; Tapia et al. 2014; California Current system: Blanchette et al. 2008). For example, upwelling conditions were proposed as major drivers of biogeographic variation of the intertidal fauna and flora in South Africa (Bustamante and Branch 1996b; Xavier et al. 2007). In this study, we demonstrate that strong upwelling-related conditions (i.e. the strongest upwelling indices coupled to the lowest SST) drive changes in species abundance and community composition, as suggested by DistLM analyses of species abundances in quadrats and of presence/absence data, and by the agreement between environmental and biological dendrograms for Imsouane. This influence seems to be exclusive to the site where environmental variables were particularly different from the surrounding locations. Specifically, the strongest upwelling indices and the lowest SST displayed by Imsouane most likely drive an effect on the abundance of the intertidal biota, primarily algae, at this location. In fact, the community patterns described largely reflect variations in the relative abundances of taxa rather than changes in species composition. This pattern has been previously described by Blanchette et al. (2008) from the intertidal shores of the California Current system. In the present study, many of the species that contributed most to differences between groups did not show large scale presence/absence patterns, but rather had striking differences in abundances, with particular expression where upwelling was stronger. For example, Bifurcaria bifurcata, Padina pavonica and Ulva chlathrata were relatively abundant in group B1/E1 (i.e. Imsouane) but extremely rare in the other groups. Bifurcaria bifurcata is a warm temperate species distributed from the British Isles to Morocco, on moderately exposed rocky shores in the mid/low intertidal and in rock-pools (e.g. Boaventura et al. 2002; Cires Rodríguez and Cuesta Moliner 2009; Neiva et al. 2015). Group B1/E1 most likely provides optimum conditions as a thermal refugium for the persistence of this brown alga. For example, cover and abundance of B. bifurcata were particularly high at Imsouane, where upwelling indices were the greatest, while minimum and maximum SST were the lowest. This is in line with recent evidence highlighting the role of upwelling cells as contemporary refugia for marine species in a context of warming climate (Riegl and Piller 2003; Hu and Guillemin 2016; Lourenço et al. 2016), by delivering cold upwelled waters that counter the effect of warming SST, allowing the long-term persistence of species and relatively high within-species genetic diversity (Lourenço et al. 2016).

Upwelled waters also enhance algal growth as a result of increased nutrient supply (Bosman et al. 1987; Ormond and Banaimoon 1994). Ulva chlathrata, an opportunistic foliose algae characterized by fast growth (Gaspar et al. 2017), reached its greatest relative abundance at Imsouane. This suggests a bottom-up effect of increased nutrient supply through upwelling (Bustamante et al. 1995; Head et al. 1996). Importantly, as our biological dataset was dominated by algae, the patterns observed in dendrogram analyses may largely reflect the influence of nutrients. While the key role of nutrient-rich upwelling waters in structuring assemblage composition across the Canary Current upwelling system (CCS) has been recently demonstrated for pelagic communities (Anabalón et al. 2014), the drivers of community structure of intertidal benthic biota along Moroccan and Western Sahara shores still warrant further investigation.

Importantly, local and meso scale features, not directly linked to larger environmental gradients, can be key drivers of intertidal community composition and abundance (Hawkins et al. 1992; Helmuth et al. 2006b; Raffaelli and Hawkins 2012). At small spatial scales, topographic and hydrodynamic features such as shore elevation and wave exposure play an important role in the trophic structure and diversity in rocky intertidal habitats (Blanchette et al. 2008; Nicastro et al. 2010; Zardi et al. 2006a; Waters et al. 2014). For example, experimental manipulation of hydrodynamics at cm-scales significantly laters mussel growth rates (McQuaid and Mostert 2010), while small scale thermal heterogeneity can exceed large-scale variability (Helmuth et al. 2006a). At meso scales both physical factors and biological interactions affect the biotic landscape of intertidal rocky shores. For example, coastal topography and habitat continuity are critical in shaping species richness gradients, boundaries and genetic structure (Fenberg and Rivadeneira 2019; Nicastro et al. 2008). In addition, periodic phenomena such as sand inundation have large effects on species richness, composition and competitive interactions (e.g. Zardi et al. 2008; Zardi et al. 2006b). Pollution has also been identified as a determinant of alterations of macrofaunal intertidal communities with significant repercussions for the functioning of ecosystems (e.g. Leopardas et al. 2016; Sabri et al. 2017), however, except after acute oils spills, its influence on more exposed rocky shores is limited compared to other anthropogenic stressors (Thompson et al. 2002). In the case of intertidal benthic animals with planktonic larvae, the processes of larval supply (influenced by coastal geomorphology and nearshore hydrodynamic features) and settlement from the water column into the benthos are key to population regulation (e.g. Poloczanska et al. 2008; Porri et al. 2007).

Intertidal biodiversity along the Atlantic shores of Morocco and Western Sahara in the context of climate change

In spite of the multiple expeditions performed within the CCS since the 19th century (reviewed in Ramos et al. 2015), benthic communities in the northwest African region are among the least known globally (Decker et al. 2003; Brito et al. 2014; Ramos et al. 2015), with most studies being limited to the Atlantic coast of northwest Morocco or Mediterranean Moroccan shores, and little sampling effort in southern Morocco or the Western Sahara (Franchimont and Saadaoui 2001).

While we omitted many small taxa such as crustaceans, our data clarify the ranges of 141 algal species, particularly along the southern region, and add nine novel records for Morocco and Western Sahara, highlighting the gaps in our knowledge of the composition and distribution of intertidal algal species in northern Africa (Franchimont and Saadaoui 2001; Ramos et al. 2015). These gaps also extend to macroinvertebrates from Moroccan shores (Franchimont and Saadaoui 2001). For example, the sandworm Sabellaria alveolata has been described in multiple studies as being distributed from Scotland to southern Morocco (e.g. Mieszkowska et al. 2006; Dubois et al. 2007; Plicanti et al. 2016). While its distribution is well described from the northern part of the range (e.g. Dubois et al. 2007; Mieszkowska et al. 2013; Firth et al. 2015), few studies refer to specific locations along the southern range (Ocaña et al. 2005; Rouhi et al. 2007; Muir et al. 2016). Here, we showed that S. alveolata extends towards the Western Sahara, at least as far south as Dakhla (site S12), approximately 24oS.

Importantly, the region is experiencing variable, severe and rapid climatic change, particularly in terms of warming (Lima and Wethey 2012). Climate change is expected to further alter species richness and community composition worldwide (Molinos et al. 2016; Woodworth-Jefcoats et al. 2016). The CCS is a temperate zone bordered by the Mediterranean Sea, the shores of tropical West Africa and the cool temperate northeastern Atlantic (Spalding et al. 2007), representing a biogeographical transition where several warm and cold water species meet and reach their northern or southern range limits (e.g. Lima et al. 2007; Lourenço et al. 2012; Nicastro et al. 2013; Neiva et al. 2015). In particular, Moroccan and Western Saharan shores have experienced warming of sea surface temperature of 0.02–0.30 and − 0.02–0.29 °C decade−1 over the last 30 years, respectively (Lima and Wethey 2012), and are characterized by distributional shifts linked to climatic changes (Nicastro et al. 2013; Lourenço et al. 2016). Recent studies have focused on understanding how upwelling intensity is changing and will change worldwide as a consequence of climate change (McGregor et al. 2007; Bakun et al. 2010; Wang et al. 2015; Sousa et al. 2017). Regardless of the expected increase in upwelling intensity in northern Africa (Wang et al. 2015 but see also Sousa et al. 2017), which could mitigate the negative effects of climate change, warming is expected to continue to increase (Collins et al. 2013; IPCC 2014), threatening intertidal species that already live close to their thermal tolerance limits. This study provides a baseline for studies investigating how intertidal benthic communities shift in a globally important upwelling systems.