To detail distributional limits of intertidal macroalgae species in northern Portugal and neighboring regions - Galicia and central Portugal - we selected a list of 34 species to be surveyed with high spatial resolution (52 locations over 750 km). We then applied a survey method (ad-libitum timed search with in-situ determination of abundance on a semi-logarithmic scale - SACFOR) appropriate to detect rare species, from our list of selected species (such as cold-water species disappearing from a region or warm-water species recently arriving), in highly variable environments (such as the intertidal), see description in 2.3.

Species were selected based on previous studies (Veiga et al., 2014; Pereira et al., 2021a) that identified macroalgae with range limits in the area (either rear- or poleward-edge) or that reached neighboring regions and thus could be in the process of migrating into the area facilitated by ongoing environmental changes and/or anthropogenic disturbance (e.g., NIS, Supplementary Table 1 and Pereira et al., 2022). Importantly, Lima and colleagues have previously described that while warm-water species in this region had been moving northwards, cold-water species had been moving on both directions, with no clear trend (Lima et al., 2007). Hence, to better understand the direction of change and to compare our results with those previous findings, we classified species as having “cold-water”, “warm-water” or “neutral” affinity. The classification was based on how their Species Temperature Index (STI) compared with temperatures in the region - if it was lower, higher, or in-between the range of average temperatures in surveyed sites, respectively. The STI is the median of temperatures recorded in the area where the species is known to occur (Burrows et al., 2020), here calculated for the North Atlantic only. Then, to compare between regions and between historical and present-day surveys, we computed the average STI of all surveyed species, per site. For details, see Pereira et al. (2022).

Sites surveyed included nearly all wave-exposed rocky shores (38) in our region of interest – northern Portugal. In addition, four sites in central Portugal and 11 in Galicia (north-west Spain) were used to determine the uniqueness of northern Portugal relative to its neighboring regions, in a total of 52 sites ( Figure 1 ; Supplementary Table 2 ). Twenty-three of the locations had been surveyed in the past (2001-2005) by the same research group (Pereira et al., 2021a; Pereira et al., 2021b; Pereira et al., 2021c, unpublished data). Sampling occurred between October 2020 and August 2021. To account for seasonality in species presence and abundance, some sites (6) were visited twice at different seasons. Furthermore, we sampled 18 additional sites with artificial hard substrate to finely track the ongoing early-stage invasion of the kelp Undaria pinnatifida (already reported in 2 locations in Portugal at the beginning of our surveys, Veiga et al., 2014). This subset of sites where all located within large stretches of sandy coast, and some were in proximity of marinas and harbors, Undaria’s likely entry points (Pereira et al., 2022).

The quadrat sampling methodology is a staple method in ecology, which owing to its inherent objectivity and precision in the assessment of the percentage cover of species, is widely used to quantify the occurrence and abundance of sessile marine species (Boaventura et al., 2002; Bertocci et al., 2012; Franklin et al., 2013). In addition, it can be easily integrated with automated methodologies (e.g., quadrats can be photographed and pictures automatically processed, Bravo et al., 2021). It does not, however, efficiently detect rare species or species inside tide pools or in otherwise hard to reach surfaces (such as rock overhangs, slippery slopes, or areas exposed to the surf during the surveys). In turn, ad-libitum timed search for selected species, quantifying and recording their abundance in a simple, semi-logarithmic scale (i.e., the SACFOR scale, Southward et al., 1995; Burrows et al., 2008) is, at least in theory, a more appropriate method to pinpoint the distribution limits of species, where it is more likely that populations will occur with low densities or even cryptically in specific microhabitats.

All locations were surveyed during low tides by a two-person team. The region has a semi-diurnal, meso-tidal regime, the largest tidal range is 3.5–4 m during spring tides. Each site was surveyed for 60 minutes. Whenever possible due to geographical proximity, the team surveyed two locations during the same low tide. At each site, the team carefully searched for each of the selected species and recorded the abundance based on the SACFOR scale (Hiscock, 1981; Burrows et al., 2008). This scale grades abundances as Superabundant (more than 90% cover), Abundant (60-90% cover), Common (widespread on the shore, 30-59%), Frequent (patches apparent, up to 30% cover), Occasional (3 to 20 individuals scattered) and Rare (only 1 or 2 individuals). Absent species are recorded as Not Found. In the case of locations surveyed more than once during this study, the higher SACFOR abundance was kept in all analyses.

The identification of the selected species was based on Chapman and Goudey (1983); Campbell et al. (1994); Molenaar et al. (1996); Stuart et al. (1999); Faes & Viejo (2003); Aziza et al. (2008); Araújo et al. (2009; 2011); Bárbara (2009; 2013); Vieira et al. (2010); Edwards et al. (2012); Bunker et al. (2017); Poza et al. (2017); Benita et al. (2018). For further details on the species surveyed, locations visited, and specific methodology refer to Pereira et al. (2022).

Data analysis was divided into (i) the analysis of present-day data (years 2020 and 2021) from all 52 locations, and (ii) a comparative analysis between present-day data and historical data (recorded between 2001 and 2005) from the 22 locations for which historical data was available ( Supplementary Table 2 ). Data analyses were performed in R (R Core Team, 2020).

Table 1 summarizes the information on the species that increased, decreased, or maintained their average abundance between historical and present-day surveys. Among those species that increased their distribution, 4 were warm-water species, 6 were cold-water species and 3 were neutral species. In addition, 4 species were NIS. Among the species that decreased their distribution, 3 were warm-water, 2 were neutral species, and 10 were cold-water species. In other words, when compared to past data and considering only species that changed their distribution and abundance within the study area, more cold-water species decreased than increased (10 to 6 out of 18) while more warm-water species increased than decreased (4 to 3 out of 10).

Considering macroalgae composition and abundance, our results show that the regions of Galicia, northern Portugal and central Portugal are all significantly different from each other ( Figure 3 ). The distinction between central Portugal and other regions was expected, since it is geographically isolated by a long stretch of sand and is less affected than the other regions by the seasonal upwelling of cold and nutrient-rich waters (Fraga, 1981; Fiuza, 1983; Lima et al., 2007; Seabra et al., 2015). Previous studies on the distribution of intertidal species along the Portuguese coast had also classified northern and central Portugal as distinct biogeographic regions (Van den Hoek and Donze, 1967; Boaventura et al., 2002; Lima et al., 2007).

Remarkably, this study also suggests that Galicia and northern Portugal are significantly different in their seaweed composition and abundance despite being geographically very close (they are part of the same continuum of rocky coast) and having (at least apparently, and based on remote sensing data) similar sea surface temperature profiles (Banzon et al., 2016). The underlying reason for these differences is not evident as the two regions have rarely been surveyed and compared in a standardized way until now. One hypothesis is that macroalgae communities in Galicia and northern Portugal are distinct due to small-scale temperature variability which is not easily detected from remotely sensed data due to its insufficient resolution (Meneghesso et al., 2020). In reality, it would be surprising if the contrasting topography and geomorphology contexts associated with each of the regions would not create dissimilar environmental conditions, not only in relation to water and atmospheric temperatures, but also regarding water salinity, wave exposure, tidal regime, current speed, nutrient availability, or magnitude of upwelling, all of which have the potential to influence local biodiversity. Specifically, the coast of Galicia is characterized by the occurrence of Rias (large coastal inlets formed by the partial submergence of unglaciated river valleys), while the coast of Northern Portugal is linear, with a north to south orientation, and absent of sheltered sites.

Another hypothesis is that the differences detected could be related with the different anthropogenic pressures exerted in these regions. The aquaculture production of seaweed and bivalves (Minchin and Nunn, 2014; James and Shears, 2016) and the heavy traffic of leisure and fishing vessels in the Rias might represent vectors for biological invasions (Blanco et al., 2021). Several non-indigenous species (NIS) have higher local abundances and are distributed more widespread in Galicia than in northern Portugal (e.g., Undaria pinnatifida – Supplementary Figure 1 and Asparagopsis armata – Supplementary Figure 6 ).

Interestingly, Galicia and central Portugal have a higher average species temperature index than northern Portugal. Additionally, all indicator species that significantly represent both Galicia and central Portugal are considered warm-water species in Northern Portugal (Asparagopsis armata and Gelidium corneum), thus supporting the hypothesis that temperature may play an important role in the patterns here described. Our results suggest that Northern Portugal is an important thermal refugia for several cold-water species (e.g., Laminaria hyperborea, Ascophyllum nodosum or Fucus serratus) which, considering the surveyed locations, were only found there. Nevertheless, for some species such as Himanthalia elongata or Halidrys siliquosa, the potential thermal refugia of this region seems to be eroding. It will soon be possible to test this hypothesis through the analysis of high-resolution intertidal temperature profiles currently being collected by autonomous temperature loggers (Lima and Wethey, 2012) at the same locations where the biodiversity surveys were carried out.

There is high variability in species composition and abundance in northern Portugal, as revealed by the lack of significant representative species ( Supplementary Table 5 ) and large dispersion among sites within the region ( Figure 3 ). Notwithstanding the fact that the sampling effort in this region was much higher, the within-region biodiversity variability suggests that even a seemingly linear and homogeneous coastline holds significant high-frequency spatial variability. One explanation for this pattern is that biodiversity variability is simply mirroring the micro-climate variability created by subtle, yet biologically important differences in solar radiation exposure, salinity, topography, wave exposure and hydrodynamics (Seabra et al., 2011; Lima et al., 2016). Combinations of these different environmental conditions create a complex mosaic of micro-habitats, even among neighboring locations, which might explain the high biodiversity variability (Helmuth et al., 2006; Potter et al., 2013). Thus, the northern Portuguese coast is a good case study to explore how the effects of small-scale environmental variability interact with large-scale patterns, namely global changes in sea temperature.

In conclusion, our analyses indicate that the regions of Galicia and central Portugal, although geographically separated, share a regionally warm-water species pool. In turn, Northern Portugal is, at least from the standpoint of the average Species Temperature Index and from the distribution of cold-water species, the coldest region, highlighting its potential importance as a climate change refugia that may be sustaining the equatorial range edge of several macroalgae. This calls for protection of the rocky intertidal in this region since climate change refugia are increasingly considered in conservation priorities (Morelli et al., 2020).

Macroalgae occurrence and abundance have changed considerably in the 15 years that mediate our historical (2000-2005) and present-day surveys (2020-2021). Currently, macroalgae communities seem to be more homogeneous, although the effect is small and not significant ( Figure 4 ). This homogenization, also evident in subtidal macroalgae communities along the coast of Portugal (de Azevedo, 2019), could lead to impacts in ecosystem functioning, productivity and ecosystem services (Clavel et al., 2011).

In addition, we recorded a decrease in presence in abundance and/or distributional range of 15 species, of which 10 were cold-water species ( Table 1 ), and among these 2 were significantly associated with historical surveys (section 3.2.2). Species with warm-water affinity exhibited a mixed response, with some increasing (4), 3 maintaining their distribution and abundance and 3 showing declines in present-day data. Importantly, species whose abundance increased the most were non-indigenous macroalgae ( Table 1 ), of which 2 neutral ones were significantly associated with present-day data. This contrasts with the results from Lima et al. (2007), that observed a clear trend of northward migration of warm-water species and mixed responses of cold-water species (some were retreating northwards while others were advancing southwards). Given that this work comprises two datapoints in time separated by 15 years, and the reduced geographical area surveyed when compared with Lima et al. (2007), we cannot confirm that a shift in the overall direction of change in communities is occurring. Still, our results are concerning especially given the accentuated decrease in some cold-water (foundation) species and the parallel increase of some warm-water (NIS) macroalgae, both with the potential to disproportionately impact the entire ecosystem.

Several macroalgae for which we report decreases in presence and abundance are canopy-forming brown algae that play an important role in coastal biodiversity as foundation species (sensu Dayton, 1972), providing food, shelter, and habitat to other species of flora and fauna, as well as protection from desiccation, temperature extremes, and wave action (Smale et al., 2013; Seitz et al., 2014; Bulleri et al., 2018). Similar declines have been reported in other temperate regions of the Atlantic, with impacts on the structure and functioning of coastal ecosystems (Smale, 2020, Yesson et al., 2015). This is the case for Himanthalia elongata (Creed, 1995) which, among all the species here surveyed, and in agreement with previous studies (Piñeiro-Corbeira et al., 2016; Casado-Amezúa et al., 2019; Barrientos et al., 2020), showed the greatest populational and geographic declines in Galicia and Portugal (it was present in less 7 sites and showed an average 1.13 reduction in SACFOR score). Currently, it can only be found in two locations south of the NW corner of the Iberian Peninsula – one healthy population in Praia Central (ID – 34) and a very small population in Mindelo (ID – 28) with just a few scattered individuals ( Supplementary Figure 9 ). Similarly, a noticeable decrease of the foundation species Laminaria hyperborea was also evident (a it occurred in less 5 sites and an average 0.74 reduction in SACFOR score, Supplementary Figure 11 ). This kelp forms marine forests and sustains a diverse community along the north-east Atlantic coast (Christie et al., 2003; Norderhaug et al., 2021). In parallel, the warm-water counterpart L. ochroleuca increased slightly in abundance, though not enough to replace L. hyperborea. It is important to note that even if successful in replacing L. hyperborea, L. ochroleuca supports a significantly less diverse community and has lower biomass (Teagle and Smale, 2018).

In addition, the decrease of Laminaria hyperborea could have contributed to the decrease of Palmaria palmata ( Supplementary Figure 10 ), as this red algae occurs often as an epiphyte of L. hyperborea towards the rear edge of its distribution (Whittick, 1983). If confirmed, this would be a prime example of how changes in the distribution of one species impact another, having cascading effects in the community. The decrease in these canopy-forming algae could lead to negative ecological and economic outcomes since they simultaneously are at the basis of the food-web and are cultivated and harvested with applications in several industries (Plaza et al., 2008). Moreover, populations at the rear-edge are often relict populations, holding unique genetic diversity that might be key for a species’ persistence under current global change trends (Provan and Maggs, 2012).

Contrastingly, some species highlighted in this study as indicator species have recently increased their distributional range and abundance in the region. The NIS warm-water macroalgae Grateloupia turuturu ( Supplementary Figure 5 ) and the NIS, neutral species Sargassum muticum ( Supplementary Figure 4 ) have both greatly increased their distribution in the last 15 years. S. muticum exhibits extremely high growth rates during spring and is a strong competitor – limiting the distribution and replacing native species, which once established can have a drastic impact in ecosystem flora and fauna composition (Critchley et al., 1986; Boudouresque et al., 1995; Stæhr et al., 2000; Sanchez and Fernandez, 2005). Additionally, the kelp Undaria pinnatifida, an non-indigenous cold-water species which was never recorded during the surveys of 2000-2005 and since then had only been recorded in Portugal in Buarcos (ID – 48) and within a marina in Viana do Castelo (Báez et al., 2010; Veiga et al., 2014), was now observed to be spreading rather quickly. It is currently present in four more sites, two with artificial substrate (a breakwater in Cabedelo and the northern breakwater at Barra), and two with natural rocky substrate (24 – Carvalhido and 26 – Forte São João). This species has a year-round presence in non-native areas, where it competes with native species with seasonal reproductive, growth and senescence stages, resulting in a successful establishment and spread in cold to temperate systems (Epstein and Smale, 2017). It is considered one of the top-100 worst invaders (Lowe et al., 2000), causing severe ecological damages in some regions, however with rather neutral and even benign effects in others (South et al., 2017), making it very difficult to predict ecosystem-wide effects of its current spreading in north-western Iberia.

The dramatic changes in the distribution of some non-indigenous species are cause for concern and call for a close and continuous monitoring of these macroalgae in north-western Iberia. Considering that their increase in abundance led to significant ecological disturbance in other regions (Stæhr et al., 2000; Casas et al., 2004; Sanchez and Fernandez, 2005; Williams and Smith, 2007; Silva et al., 2021), it is likely that they will too significantly and negatively impact ecosystem functioning in north-western Iberia (Mineur et al., 2015). Moreover, NIS such as G. turuturu in northern Portugal (Araújo et al., 2011) could also be taking the role of passenger of ecological change rather than driver, benefiting from the loss of algae canopies and relying on the disruption of native assemblages to spread (Mulas and Bertocci, 2016), in which case, these species can be important indicators of ecosystem health.

Temperature is a fundamental driver of change in marine systems, and metrics based on species thermal affinities have successfully been used to track community-wide changes over large geographical areas (see, for example, Burrows et al., 2019). Community Temperature Index (CTI) integrates in a single index four distinct processes – increase/decrease in the abundance of warm-affinity species (tropicalization and detropicalization, respectively), and decrease/increase of cold-affinity species (deborealization and borealization, respectively). This complexity is further increased by a variety of factors related to environmental conditions, human impacts and to the community structure itself (Mclean et al., 2021). Still, deborealization has been described as an important component frequently leading to increasing CTI (Mclean et al., 2021). In this study, however, and despite the decrease in abundance of several cold-water species over the last 15 years, the average Species Temperature Index (STI) did not increase significantly across the study area. This may be related to the fact that we only analyzed a restricted subset of species among all possible, and thus we may not have had enough resolution to find significant differences. Indeed, it has been shown for coastal communities of NW Atlantic that the utility of CTI analyses may be limited when a reduced number of species is used, especially if individual species have non-linear responses to temperature (Flanagan et al., 2019). In addition, it may happen that the macroalgae community as a whole is less responsive to changes in temperature, which would explain the marginal increase in the average STI in this work. In fact, a different study in the NE Atlantic including substantially more species showed that CTI is increasing as sea surface temperature increases, though the CTI of animals is responding more readily to changes in temperature than that of macroalgae (Burrows et al., 2020).

We used a fine-scale survey strategy that allowed us to obtain distribution and abundance data at an average resolution of 3.4 ± 5.9 km within the region of highest interest (northern Portugal), a value higher than what is typically provided in similar studies (Boaventura et al., 2002; Araújo et al., 2009). Such high-resolution distribution data proved instrumental, particularly in circumstances where certain species were found to have disappeared from historical sites only to be recorded in other locations just a few hundreds of meters away (e.g., Pelvetia canaliculata, which disappeared from Cabo do Mundo (ID-39) but was found 2.2 kms away at site Memória (ID-38)). This means that the most common long-term monitoring strategy, which is to target all the effort towards revisiting the same sites repeatedly (Mieszkowska et al., 2021) may yield oversimplified datasets, and be insufficient to accurately pinpoint the distribution limits of certain species or to detect and therefore track fast changes. Moreover, previous studies have already demonstrated that species distribution models based on fine-scale surveys result in different patterns from those fed with data obtained from coarser surveys (Trivedi et al., 2008; Gastón and García-Viñas, 2010), yielding qualitatively distinct predictions with important consequences for the conservation and management of biodiversity (Franklin et al., 2013).

By adopting the ad-libitum timed search for selected species and recording their abundance in a simple, semi-logarithmic scale (i. e., the SACFOR scale, Southward et al., 1995; Burrows et al., 2008), we successfully mitigate the drawbacks associated with the commonly-used quadrat sampling method. In particular, while the combined ad-libitum timed search with SACFOR recording can be regarded as a less objective or accurate method which often requires expert knowledge (and thus is harder to train for, Mieszkowska et al., 2006), our study confirms that it can efficiently detect rare species or species in hard-to-reach microhabitats, unlike the quadrat method. Furthermore, our data strongly suggests that the use of intensive, fine-scale surveys was essential to accurately assess the distribution of the studied species, which in turn allowed us to formally test hypotheses on effects of warming on species abundances at or near their range edges. The presence and abundance of some macroalgae species varies with the season. Although we visited some sites more than once and found similar results between the two sampling events, we do expect that for some species seasonal variation in abundance may be greater than long-term change, and thus suggest that future studies strive to survey consecutive seasons along the year. This clearly highlights the importance of taking into consideration the overall goals of any assessment of intertidal biodiversity when choosing the spatial coverage required and the sampling method to be used.

In conclusion, the contrasts between historical and present-day biodiversity and biogeography patterns suggest that a homogenization of the macroalgae communities may be underway across the north-western Iberia. We found that the average community dissimilarity seems to have been decreasing over the last two decades, both within and among regions. Some cold-water species are becoming rarer while warm-water counterparts are found over larger spans of coast. Another factor of homogenization is the increasing prevalence of some NIS, which already dominate some microhabitats and/or sites. Hence, a higher effort in conservation and restoration is vital to protect the health of these coastal ecosystems and the communities that depend on them (García Molinos et al., 2016). Although this work did not formally explore the role ongoing changes in temperature may be playing in these alterations, both the net decline of cold-water species and the net increase in warm-water species, as well as the higher prevalence of NIS is congruent with the effects expected from an increase in temperature driven by climate change. These effects could soon be exacerbated by any decrease in the magnitude of the upwelling in the region, which mitigates further warming (Seabra et al., 2019; Sousa et al., 2020). Further studies that explicitly analyze the joint change in biodiversity and temperature are urgently needed to test some of the hypotheses here discussed.