In the framework of the ConAfrica project, weekly sampling was carried out from January 2005 to December 2006 (n=76 samples) during daylight hours aboard the R/V Solana II. This study focuses on station 1 (28°04.00'N 15°21.62'W), located over the 100 m isobath in an area of weak flow associated with the stagnation point upstream of the island (Moyano and Hernández-León 2011, Landeira et al. 2013, Fig. 1). Zooplankton were collected using a Bongo net of 40 cm mouth diameter fitted with nets of 200 µm mesh size. The net was towed obliquely from 90 m to the surface at 2 knots (⁓1 m s^–1) and a wire speed of 0.5 m s^–1, and it was attempted to maintain a wire angle of approximately 45°. Flow rate was estimated using calibrated flowmeters (General Oceanics), and the mean volume of water filtered was 27.84 m³. For taxonomic identification, one of the samples was immediately fixed in a 4% solution of formalin in seawater. Prior to zooplankton sampling, vertical profiles of temperature, salinity and fluorescence were recorded using a CTD SBE25 (Sea-Bird Electronics, Inc., Bellevue, WA, USA). Phytoplankton chlorophyll a was derived from water samples taken in the mixed layer at a depth of 15 m with a Niskin bottle.

In the laboratory, decapod larvae were sorted and quantified. For holoplanktonic pelagic shrimps which are planktonic during their whole lives (e.g. Lucifer typus), only larval stages were analysed. Following Anger (2001) and Martin et al. (2014), the transition from planktonic to benthic stages such as the decapodite stage in Caridea or megalopae in Brachyura were considered as last larval stages, whereas postlarval stages were not considered in this study. Larvae were identified using the guides given by dos Santos and Lindley (2001) and dos Santos and González-Gordillo (2004), and using the specific taxonomic descriptions recommended in the checklist of González-Gordillo et al. (2001). General taxonomical nomenclature follows De Grave et al. (2009). The catches were standardized by number of decapod larvae per 100 m³.

The Shannon–Wiener diversity index (H'),

(where S is the number of species and p_i is the proportion of individuals in species i), was used to analyse changes in temporal and spatial diversity in the decapod larvae community. The parametric statistical method (Student t test, p>0.05) was used to evaluate interannual differences in larval abundance and diversity (previously tested for homogeneity of variances using Levene’s test). Multivariate analysis was used to identify larval assemblages with distinct community structure. Only species present in both 2005 and 2006 were used in the analysis in order to eliminate the effect of rare and multispecies groups (e.g. Sergestidae spp., Pagurus spp.). A total of 49 species were left from the initial 105.

In a first step, to examine temporal differences between months, within years and between years, an analysis of similarities (two-way nested ANOSIM) was performed on the species resemblance matrix using the log-transformed abundance data from all the samples in a Bray-Curtis similarity matrix. After this analysis, using the same Bray-Curtis similarity matrix, a non-metric multidimensional scaling (MDS) was performed to obtain a graphical ordination of the samples (Clarke and Warwick 2001). It was assumed that stress coefficients greater than 0.15 indicate a good representation of data (Clarke and Gorley 2006).

Due to the absence of any temporal structure in the decapod larvae community (see Results), forced by the elevated number of taxa and the weak seasonality of decapod larvae in subtropical waters (Reyns and Sponaugle 1999), a month-averaged larval abundance data was used in a second step of the analysis (removing the factor year). In this second step, in order to divide the 12 months into distinct periods of larval assemblages, a cluster analysis and similarity profile routine (SIMPROF; p<0.01; 999 permutations) were performed using the log-transformed abundance data in a Bray-Curtis similarity matrix. SIMPROF is a permutation test that objectively determines whether any significant group structure exists within a set of samples (Clarke and Gorley 2006).

After this analysis, using the same Bray-Curtis similarity matrix, a non-metric MDS was performed. The significant results of the SIMPROF test were entered into the MDS plot to assess the level of agreement between the two techniques. Complementarily, the RELATE procedure, employing Spearman rank correlation coefficients (p), was used to determine whether the series of sequential points for mean monthly samples on MDS ordination plots approximated a circle and, if so, the extent to which the distribution of those points was correlated with a true circle (Clarke and Warwick 2001). The groups of months detected were used as factors to test significant differences in temporal larval assemblages of decapod species using a one-way similarity analysis (ANOSIM).

A similarity percentages (SIMPER) test was then used to determine which species contributed most to characterization of each period of distinct larval assemblage (Clarke and Warwick 2001, Clarke and Gorley 2006). The non-parametric Spearman rank correlation was used to explore the relationship between sea surface temperature, salinity and chlorophyll a, and decapod larvae variables (total abundance and diversity). Finally, an MDS plot was represented by superimposing bubbles of increasing size related to significantly correlated variables in order to visualize the link with larval assemblages. Statistical analyses were carried out using PRIMER v.6.1 and SPSS 15.0.

Temperature showed a consistent seasonal trend in which the heating due to strong insolation led to maximum values in the mixed layer at 20-30 m (22.9°C in 2005 and 24.1°C in 2006) from August to October. The cooling of the water column started in November and finished around March-April, showing minimum values of 17.7°C in 2015 and 18.3°C in 2016 (Fig. 2). Salinity did not show a seasonal pattern and ranged from 36.54 to 36.97 (Fig. 2). The temporal distribution of chlorophyll a was negatively correlated with temperature (Spearman rank correlation: r=–0.73, p<0.01) and showed the typical seasonality of the Canary Island waters (Table 1, Fig. 2). The quasi-permanent thermocline, which promotes oligotrophic conditions during most of the year and limits phytoplankton production in the Canary Islands (Arístegui et al. 2001), led to standing stocks of chlorophyll a lower than 0.2 mg m^–3 during the summer (Fig. 2). This situation changed in winter, when the temperature droped below 19°C and the cooling of the surface eroded the thermocline, promoting the development of a najor phytoplankton bloom in February-March, followed by another smaller peak around April known as the late winter bloom. Interannual differences were observed. The highest chlorophyll a values (0.9 mg m^–3) occurred during the bloom of 2005, with a mean standing stock of 0.7±0.16 mg m^–3. In 2006 the peak occurred during the same period but was weaker, with a mean standing stock of 0.36±0.05 mg m^–3 and maximum values of 0.47 mg m^–3. The temporal evolution of temperature suggested different durations of the mixing period. In January 2006 the temperature was still above 19°C, preventing deep convection. In May the heating of the surface waters was already visible, especially in 2006, when the temperature was above 20°C at the beginning of the month. This situation led to a shorter mixing period in 2006 that promoted a lesser phytoplankton bloom.

A total of 6967 larvae belonging to 105 different taxa were identified during the two-year study. Gathered in different suborders, the Pleocyemata and Dendrobranchiata were represented by 85 and 20 taxa, respectively. Within the Pleocyemata, the infraorders with the highest number of taxa were Brachyura (36 taxa) and Caridea (31 taxa). Average species diversity was relatively high (2.40±0.59), with the highest values (>3) in June-August of both years and the lowest (0.60) in January 2005 (Fig. 3). A significant positive correlation with temperature (Spearman rank correlation: r=0.33, p<0.01) was also evidenced this tendency (Table 1). Differences in diversity between years were also significant (Student t test, df=74, t=–2.982, p=0.004), with 2005 showing a lower mean value (2.19±0.86) than 2006 (2.57±0.44).

In terms of relative abundance, the infraorders Anomura and Caridea accounted for 33.4% and 32.8% of the total decapod larvae catches, respectively. Other less abundant taxonomic groups were Brachyura (17.5%), Dendrobranchiata (17.5%), Axiidea and Gebiidea (6%). Achelata and Stenopodidea did not show abundance greater than 5%, while Polychelida, Astacidea, and Glypheidea (this infraorder has not been recorded in the Canary Islands) were not observed in the samples (Table 2). The remaining infraorders are not present in the Canary Islands. The most abundant families within Anomura were Galatheidae (12.2%), Diogenidae (9.2%), and Paguridae (11.8%) due to the main contribution of Galathea intermedia, Calcinus tubularis and Pagurus spp. (Table 2). In Caridea, the families Apheidae, Hippolytidae and Processidae had abundances of around 7% of the total sample, with Processa nouveli and Latreutes fucorum as the most abundant species. The families Brachyura, Majidae, Xanthidae and Grapsidae accounted for around 3% of total abundance. The average abundances of Upogebia spp. (3.8%) and Pandalina brevirostris (3.1%) were also noteworthy. Regarding adult habitat, larvae of benthic species (86.1%) were more abundant than larvae of pelagic species (13.9%) (Table 2). Larvae in the first stage of development, zoea I, accounted for 37.8±13.4% of the total decapod larvae, whereas larvae in the last stage of development always accounted for less than 5% of the total larvae.

Total abundances ranged from a minimum value of 30.2 larvae/100 m³ recorded in February 2005 to the maximum value of 2925 larvae/100 m³ in 12 August 2006. The mean abundance in 2005 (281.02±86.35 larvae/100 m³) was significantly lower (Student t test, df=74, t=–2.07, p=0.021) than in 2006 (478.14±116.13 larvae/100 m³). Decapod larvae were present in the plankton all year round, making it difficult to observe, a priori, any seasonal/interannual patterns (Fig. 3). A two-way ANOSIM test revealed a slight significant difference among months (Global R=0.372, p=0.03) but not among years (Global R=0.133, p=0.4). This result was supported by the MDS plot, which showed a poor spatial ordination of the samples based on their decapod larvae composition (Fig. 4).

However, when the factor year was removed and month-averaged larval abundance data were used, the multivariate statistical analysis revealed seasonality in decapod larvae community (Fig. 5). Two significantly different groups of months (p<0.001) were distinguished using SIMPROF at a 58% level of similarity. An MDS plot (2D stress, 0.14) with a superimposed significant cluster shows the separation of the two distinct larval assemblages. A “warm” cluster includes May-September, while the other “cold” cluster includes November-April, and October appears as a transition between the two seasons (Fig. 5). In this ordination, the months tend to undergo a clockwise cyclical spatial distribution (Fig. 5), and RELATE confirmed that the cyclicity was consistent with that of a circle (p=0.001), with a rank correlation coefficient of 0.485. Moreover, the ANOSIM routine revealed that cold and warm larval assemblages were significantly different (Global R=0.489, p=0.002). The average abundance of decapod larvae, temperature, and chlorophyll a are superimposed as proportional bubbles over the MDS plot to visualize the relationship of these variables with the assemblages (Fig. 5), showing that during the warm period the larval abundance is also higher, in agreement with the positive correlation (Spearman rank correlation: r=0.43, p<0.01) between these two variables (Table 1). Conversely, as mention above, chlorophyll a was higher during colder months and therefore negatively correlated (Spearman rank correlation: r=–0.27, p<0.05) with decapod larva abundance.

This larval assemblage was difficult to visualize in the temporal distribution plot of abundant species, since they were collected in almost every single sampling event, indicating year-round spawning (Figs 6, 7). However, the SIMPER routine revealed that changes in composition and/or abundance were characteristic of the warm and cold assemblages (Table 3). Warm assemblages were found in species that spawn in summer, such as Calcinus tubularis larvae primarily collected in summer (54.1±23.8 larvae/100 m³) and Nanocassiope melanodactyla that was abundant during this period (19.4±11.2 larvae/100 m³) (Fig. 6). In the case of the pelagic species Deosergestes henseni, the larvae were collected exclusively in summer and showed a peak around June (19.9±8.8 larvae/100 m³), and Parasergestes vigilax was especially abundant during the summer and autumn of 2006 (Fig. 6). Pachygrapus spp. larvae showed a clear peak (May-October) in 2006 that was not evident in 2005. Other species, such as Alpheus glaber, Clibanarius aequabilis, Percnon gibbesi and Lysmata seticaudata, helped typify the warm assemblage but with lower (<4) similarity percentages (Fig. 6, Table 3). The cold assemblage was characterized by species that exhibited winter-autumn peaks, but their presence in the plankton was not always restricted to cold months. This is the case of Latreutes fucorum (peaks in late summer), and of Eualus occultus, Pandalina brevirostris, and Philocheras bispinosus (peaks in spring-summer) (Fig. 7). Other species that had elevated larval abundance throughout the year (e.g. Processa nouveli, Galathea intermedia, or Lucifer typus) contributed strongly to the cold assemblage but were also important for the warm assemblage (Table 3).

We report, for the first time in the Canary Islands waters, the presence of larvae of the species Pandalina brevirostris, Processa edulis, Necallianasa truncata, Parapenaeus longirostris, Crangon crangon and Nematopagurus longicornis. All larval stages of P. brevirostris were found and, as mentioned above, it is an abundant species (mean abundance of 12.21 larvae/100 m³) characterizing the cold assemblage (Fig. 6, Table 3). Processa edulis (zoea I-III stages) was caught in March, May, June-September and November, when it peaked with a mean abundance of 10.18 larvae/100 m³ (Table 2). Larvae of mud shrimp N. truncata (zoea I-III stages) were frequently collected in the plankton, especially in summer, when they achieved significant concentrations in September (mean abundance of 44.46 larvae/100 m³). N. longicornis zoeae III and zoeae IV were observed in March (maximum value of 13.85 larvae/100 m³), July and December 2005 and in January and May 2006. P. longirostris larvae in protozoea III stage of development were observed sporadically in June (3.19 larvae/100 m³) and July (2.61 larvae/100 m³) 2005, and in February and December 2006 (around 7 larvae/100 m³). Only one larva (zoea IV) of C. crangon was caught in January 2006.

This description of the entire larval community of Gran Canaria provides valuable information about the composition and potential spawning season for the most abundant species, which have been largely under-studied. Broadly, the decapod larvae community in Gran Canaria is characterized by contracted larval hatching periods, significant contributions from pelagic species, and weaker seasonality in comparison with higher latitudes, but has two distinct larval assemblages.

The year-round occurrence of decapod larvae in the plankton indicates that crustacean decapods can reproduce throughout the year. The temporal distribution of decapod larvae suggests a contracted larval hatching period for many species, but with abundance peaks associated with seasonality. This characteristic pattern observed in the subtropical waters of Gran Canaria contrasts with what happens at higher latitudes of the NE Atlantic. For example, in Svalbard, in the Arctic Ocean, decapod larvae occurred in the plankton during short periods from May to August, with a peak in mid-June (Stübner et al. 2016). In the North Sea, the presence of a marked seasonality characterized by the virtual absence of larvae from December to February and by the presence of two main abundance peaks in spring and summer and extending into autumn is well documented (Highfield et al. 2010, Pan et al. 2011). This larval seasonality is also visible, but the presence of larvae in the plankton is year-round at lower latitudes of the Mediterranean Sea (Bourdillon-Casanova 1960, Fusté 1982), the Atlantic coasts of Portugal (dos Santos 1999) and southern Spain (González-Gordillo and Rodríguez 2003). In temperate regions, increases in water temperature trigger breeding events in which reproduction intensifies. However, in the tropics and subtropics the higher and less variable temperatures favour continuous reproduction (Bauer 1992). For instance, xanthid larvae are found in the water column all year round at the Canary Islands, with settlement peaks in spring and autumn (García-Sanz et al. 2014), whereas in the southern Iberian Peninsula, xanthid larvae are only present in the plankton in spring (González-Gordillo and Rodríguez 2003), or from April to September (Paula 1987), with a restricted settlement period in July-October (Flores et al. 2002). The less variable annual temperature regime in the Canary Islands (18-24°C, Barton et al. 1998) leads to a protracted breeding season coupled with multiple peaks of zoeae I, which may indicate non-synchronous release and the presence of multiple cohorts.

In addition, in coincidence with the results of Moyano and Hernández-León (2011) for fish larvae, we identified two distinct seasonal larval assemblages, “cold” and “warm”, in which temperature seemed to play a key role in structuring the temporal variability of the decapod larvae community. The cold assemblage (November-April) occurred during the mixing period, when low temperatures and surface cooling erode the thermocline, allowing diffusion of nutrients that promotes elevated production of phytoplankton and zooplankton (Arístegui et al. 2001, Moyano and Hernández-León 2011). Species that were more abundant in winter-spring, such as P. nouveli, L. typus, Philocheras sculptus and G. intermedia, were typical of the cold assemblage. Little is known about their reproductive strategies, although results from settlement experiments conducted in seagrass meadows and macroalgal beds off Gran Canaria (García-Sanz et al. 2014) have provided valuable information that is in agreement with that observed in the present study, suggesting pelagic/benthic coupling. Thus, Galathea spp. (including G. intermedia) and Pagurus spp. had two settlement periods, but the main peak (April-May) is related to the highest larval abundance observed in the plankton in winter. During the warm assemblage period (May-September), the surface heating stratifies the water column. This leads to reformation of the main thermocline, which then limits vertical nutrient fluxes and phytoplankton production (Arístegui et al. 2001, Moyano and Hernández-León 2011). High abundances of N. melanodactyla, D. henseni, C. tubularis and C. aequabilis are characteristic of this assemblage. The settlement of C. tubularis and grapsid crabs (including Pachygrapsus spp.) occurred from October to December (García-Sanz et al. 2014), just after the summer plankton peak. Larval abundance in summer of the pelagic shrimp D. henseni is also characteristic of this assemblage. This species is one of the most abundant of the mesopelagic community in Canary Island waters (Ariza et at. 2015), although its larval stages have also been collected in September (FAX99 cruise, Landeira et al. 2017), and October-November (CANARIAS 9110 cruise, Landeira et al. 2009), suggesting a longer spawning season than that observed in the present study.

The contribution of pelagic shrimps to the larval community is significantly higher than that reported from other Atlantic regions (e.g. from Portugal by dos Santos 1999 and from southern Spain by González-Gordillo and Rodríguez 2003), indicating considerable oceanic influence on the coastal region of the Canary Islands (Arístegui et al. 2001). It seems that this is a common pattern, since relatively high larval densities of pelagic species over the island shelf have been reported in other archipelagos, such as the Balearic Islands in the western Mediterranean (Torres et al. 2014) and Saint Paul’s Rock in the equatorial Atlantic (Brandão et al. 2012).

Several deep-water shrimp pandalids, particularly Plesionika narval (Fabricius, 1787) but also P. edwardsii (Brandt, 1851) and Heterocarpus ensifer A. Milne-Edwards, 1881, are targeted by local small-scale fisheries operating with bottom traps at some island localities (González 1995). Ovigerous females occur year-round but a mass spawning peak occurs in the Canary Islands in spring and summer (Tuset et al. 2009, González et al. 2016a, Triay-Portella et al. 2017). In the present study, larvae of Heterocarpus spp. and Plesionika spp. were included in Pandalidae spp., because it is still impossible to distinguish these two genera using morphological characters in view of their remarkable similarity. Despite this limitation, larvae of Pandalidae spp. were present in the plankton throughout the year and peaked in September, supporting the spawning period reported.

According to González (2016), intertidal crab species such as Plagusia depressa, Xantho spp., Grapsus adscensionis, Pachygrapsus spp. and Percnon gibbesi are intensively harvested, mainly by hand, to be used for human consumption and/or as bait by small-scale fisheries. Harvesting pressure is frequently high and, in some regions, there are clear symptoms of overexploitation. This author recommended an urgent study of basic biological parameters such as population status for the implementation of conservation measures (González 2016). The presence of these larvae in the plankton can be interpreted as an indicator of the breeding season of these species. Abundance of P. depressa and Pachygrapsus spp. peaked during summer, whereas P. gibbesi and G. adscensionis seem to have two peaks, in spring and summer, respectively.

Equally notable in our results is the absence of larvae from taxa that are common in our sampling region as adults. This was the case of the shrimp Palaemon elegans Rathke, 1837 and the crab Eriphia verrucosa (Forskål, 1775), which inhabit intertidal ponds and shallow water rocky shores (González 1995, d’Udekem d’Acoz 1999). The fact that we did not find these taxa is difficult to explain with the available data.

Plankton surveys have proved useful for the assessment of diversity in marine ecosystems. This brings us the possibility of sampling, at the same time, larvae of both pelagic and benthic species inhabiting shallow, deep-water and/or cryptic habitats, which are sometimes difficult to collect as adults. For example, similar samplings have facilitated the detection of non-indigenous species such as a new unreported pinnotherid crab in the Gulf of Cádiz (Marco-Herrero et al. 2017). In the western Mediterranean Sea, the study of plankton samples has also led to the detection of larvae of the invasive shrimp Palaemon macrodactylus Rathbun, 1902 (Torres et al. 2012), but it also has provided valuable information on deep-water species of interest for conservation and fishing exploitation, such as Aristeus antennatus (Risso, 1816), Parapenaeus longirostris and Scyllarides latus (Latreille, 1803) (Torres et al. 2013).

In the present study, we identified larvae of species that have not yet been reported from the Canary Islands. For example, the high abundance and temporally stable occurrence of P. brevirostris, P. edulis, and N. truncata larvae suggest that these species may not only be present in the study region, but also relatively common. If the presence of adult populations is confirmed, the geographical distribution of these four species would extend to the south (d’Udekem D’Acoz 1999).

We also report the occurrence of larval forms of unknown crustacean adults. This is the case of Stenopidae sp1 that were collected in November 2006. In the Canary Islands, Stenopus spinosus was the only member of Stenopodidea until the recent record of Spongiocaris koehleri (Caullery, 1896) observed by González et al. (2016b). However, S. koehleri larvae cannot be found in the plankton since they have direct development (Kemp 1910). This finding suggests that another stenopodid species may be present in the Canary Islands. Another example is the larval form of Callianasidae SL16, which was relatively abundant from May to August. This larval form was described by dos Santos (1999) from plankton specimens collected south of Portugal in July. It has also been observed in the NW African upwelling, where it was transported towards the ocean during a strong filament event in August-September 1999 (Landeira et al. 2017). The only callianassid species reported for the Canary Islands are Pestarella tyrrhena (Petagna, 1792) (González and Quiles 2003) and Pestarella candida (Moro et al. 2014), which are also present off NW Africa and Portugal (d’Udekem D’Acoz 1999). However, their larval morphology is clearly different, since Callianasidae SL16 has spines on the entire ventral margin of carapace, and in the other two species the spines are present only on the anteroventral margin (for details see the original descriptions listed in González-Gordillo et al. 2001). Therefore, it is reasonable to think that another callianasid species is present in the study area.

The hermit crab N. longicornis is distributed at a depth of 70-800 m along the NE Atlantic coast, including Madeira, the Azores, Morocco, and Cape Verde Islands (d’Udekem d’Acoz 1999, d’Udekem d’Acoz and Wirtz 2002). The Canary Islands is the only archipelago in the area without records, so it seems that González (1995) was right when he suggested that this species could also be present in the Canary Islands. Crangon crangon is widely distributed along the NE Atlantic coasts from the White Sea and Iceland down to Morocco, in the Baltic Sea, the Mediterranean Sea, and also the Black Sea (d’Udekem d’Acoz 1999). If this species was present in the Canary Islands, it would constitute the southernmost population. However, in our opinion it is unlikely that find adult populations will be found there since there is no evidence for its occurrence in other northern Macaronesian archipelagos, such as the Selvagen Islands, Madeira and the Azores (d’Udekem d’Acoz 1999). Moreover, C. crangon is an epibenthic shrimp that preferably inhabits soft-sediment in estuarine habitats (Holthuis 1980) that do not exist in the Canary Islands. It is possible that this larva reached the Gran Canaria coast transported by upwelling filaments from Africa, as has been suggested by Landeira et al. (2017), but with the available data it is not possible to support this hypothesis.

P. longirostris is one of the most important commercial shrimps found in the Mediterranean Sea and the NE Atlantic, from Galicia to Angola (d’Udekem d’Acoz 1999, Holthuis 1980). P. longirostris larvae have already been observed offshore, south of the Canary Islands in association with an African upwelling filament but, until now, these larvae have never been sampled over the island shelf (Landeira et al. 2009). The presence of extensive adult populations is unlikely but the recent observations of single specimens of Penaeus kerathurus (Forskål, 1775) by different scuba divers in muddy-sandy habitats of Lanzarote, Gran Canaria and Tenerife islands (Moro et al. 2014) make the occurrence of P. longirostris more probable in the Canary Islands.

This study provides accurate information about the composition and assemblage of decapod larvae in the Canary Islands region. Furthermore, the data collected during this study provide a baseline for future comparisons with respect to fishery pressure and climate variability. Despite the less evident seasonality displayed by most of the species examined, we identify two distinct temporal assemblages of decapod larvae for the subtropical waters of the Canary Islands. Despite significant inter-year differences in diversity and larval abundance, there were no evident changes in the larval assemblages. The detection of such changes in the community requires a multiyear dataset obtained in long-term monitoring programmes (Lindley et al. 2010). This may be important, because the temporal distribution of spawning of different species is likely to vary under climate change.

Since 1997 the sea surface temperature has undergone a warming trend of 0.25°C decade^–1 in the Canary Islands region (Vélez-Belchí et al. 2015), but little is known about subsequent alterations in breeding success and distribution of crustacean species. However, there is already clear evidence of a tropicalization in the Canary Islands ecosystems. For instance, many of the new records of fish (Brito et al. 2005), decapod crustaceans (González et al. 2017), other invertebrates (Brito 2008) and algae (Afonso-Carrillo et al. 2006, 2007) are tropical species. Also, the warming during recent decades has enabled species with tropical affinities to spread quickly across the archipelago (e.g. ephemeral benthic algae, Sangil et al. 2012), or to increase their populations significantly due to more favourable and longer recruitment events (e.g. echinoid Diadema africana Rodríguez, Hernández and Clemente, 2010, Hernández et al. 2010). The successful establishment of the tropical hydrocoral Millepora sp. (Clemente et al. 2011) is also noteworthy, as it suggests profound future changes in benthic communities if warming trends continue. We consider the establishment of long-term plankton sampling programmes in tropical-subtropical regions of the NE Atlantic to track these interannual alterations in phenology and non-native crustacean species occurrence, which are already visible in temperate regions (Kirby and Beaugrand 2009, Lindley et al. 2010), to be of paramount importance.

To detect northward distribution shifts of tropical species in relation to current and future conditions in the subtropical and temperate Atlantic, more accurate larval descriptions are needed. According to González (2016), at least 23 brachyuran benthic species occurring in the Canary Islands have their northern limit of distribution at this archipelago. However, from these species only the larval morphology of Stenorhynchus lanceolatus (which actually occurs also in Madeira) and Microcassiope minor (Dana, 1852) (which is present also in the Azores islands) are known, and are described by Paula and Cartaxana (1991) and Clark et al. (2004), respectively. We therefore encourage the scientific community to continue to improve the taxonomic description of the larval morphology of decapod crustaceans following the current standards, especially of tropical species that are presently less studied.