The marine coastal environment is threatened both locally and globally in different ways. In Europe the coastal habitats have been progressively degraded over the last few decades (e.g. Benedetti-Cecchi et al. 2001, Lotze et al. 2006, Claudet and Fraschetti 2010). Future projections are not encouraging: The Mediterranean Action Plan predicts that the urban population of the coastal Mediterranean could reach 176 million by 2015 plus 350 million tourists yearly. Furthermore, an increase in the number of introduced species (Boudouresque et al. 2005) and climate-induced stressors are contributing to changes in the Mediterranean biodiversity (Fraschetti et al. 2011), emphasizing the need to foster the surveillance of coastal systems.

Maintaining diversity in coastal systems is essential for the sustainability of communities, ecosystem functioning and services (Pickaver 2009, Costanza et al. 2014). Changes in coastal ecosystems can be one-off events and have dramatic effects (i.e. artificial modifications of the coast or severe storms), or rather progressive and relatively slow (at the scale of human life), and are difficult to notice through snapshot studies. Moreover, the negative impact on them may be difficult to document (Støttrup 2009). One of the aims of the European Marine Directive is to ensure that biodiversity is maintained, that is, kept in line with the natural state appropriate to the area in question. However, baseline shifts can affect reference sites as much as impact sites (Støttrup 2009), masking variations and distorting perceptions.

In the Mediterranean, littoral fish communities are threatened by numerous stressors, including habitat loss, global warming, changes in the continental water discharges, artificialization of the coastlines, introduction of alien species and fishing pressure (e.g. Guidetti et al. 2002, Claudet et al. 2006). Fish communities are a key component of the aquatic ecosystem (Holmlund and Hammer 1999) as fish provide fundamental services for ecosystem functioning and resilience (Schindler et al. 1997, Vanni 2002, Myers et al. 2007). Moreover, numerous studies support the use of fish assemblages as biological indicators for marine coastal waters (e.g. Sano 2000, Seytre and Francour 2008, Azzurro et al. 2011). Nevertheless, studying and monitoring fish assemblages is as complex as the diversity of groups represented. Littoral fish assemblages include groups of different niches, so different complementary sampling methods are required to capture all the components (Elliott et al. 2002). It is widely recognized that no single assessment technique can provide unbiased, qualitative or quantitative estimates of fish assemblages (Sale and Douglas 1981, Ackerman and Bellwood 2000). This problem was highlighted decades ago (Harmelin-Vivien et al. 1985) but it is still an open issue and it is recommended to apply different sampling methods simultaneously to the same study site (Connell et al. 1998, Willis et al. 2000, Cappo et al. 2004).

Studies of fish populations in shallow littoral waters usually rely on underwater visual censuses (e.g. Harmelin-Vivien et al. 2008, Bussotti et al. 2015, Prato et al. 2017). Video techniques have been increasingly used (Mallet and Pelletier 2014) and are recognized mainly as an additional technique to underwater visual censuses, with stationary (Francour 1999), diver-operated (Tessier et al. 2013) or roving cameras (Tessier and Chabanet 2006), or stationary baited underwater video (Gledhill et al. 1996, Willis and Babcock 2000, Stobart et al. 2007). Fishing techniques are also used as a complementary method in littoral fish research surveys (Franco et al. 2012). The appropriateness of a methodological approach depends on its capacity to fulfil the purposes of the study. However, accurate standardized measurements of fish species richness and community structure are essential for monitoring the progress towards biodiversity targets (Hutchings and Baum 2005), as well as for conservation actions (MacNeil et al. 2008). Underwater visual census methods cannot assume equal detectability across all species, and it is recognized that this methodology under-represents a large number of cryptobenthic fish species (e.g. Smith 1988, Ackerman and Bellwood 2000, Kovačić et al. 2012) because more than 90% of them may go undetected by underwater visual censuses (Willis 2001). In fact, cryptic fish are particularly difficult to accurately survey. They are under-sampled components of fish communities and their ecological roles have generally been ignored (Smith-Vaniz et al. 2006). The use of destructive techniques such as ichthyocides and anaesthetics for cryptic species is controversial but also justified for obtaining reasonably complete inventories of reef fish (Smith-Vaniz et al. 2006, Glavičić et al. 2016, Thiriet et al. 2016).

In addition to research surveys, fisheries-dependent data are also used for fish assemblage monitoring. For all their flaws and lack of scientific methodology, fisheries-dependent data have some advantages: they generally provide large data sets with wide temporal and spatial coverage and they are cost-effective. Furthermore, other cost-effective sources of information are providing good findings in different fields (http://www.observadoresdelmar.es, Fairclough et al. 2014) jointly with citizen science (e.g. www.seawatchers.org, redmap.org.au), which is currently expanding worldwide. Valuable data gathered by public agencies or private entities also have informative potential that is worthy of being analysed. This is the case of recreational fishing tournaments in certain regions, which have extensive spatial coverage and/or temporal continuity, and have been shown to be valuable for characterizing and monitoring littoral fish assemblages (Lincoln Smith 1989, Coll et al. 2004, Gordoa 2009).

Another potential source of information is underwater photo contests. The technological developments in underwater photography have led to the expansion of a new sport activity: free-diving underwater photography. A recent event in underwater fish photography contests rewards the number of species photographed. Participants are given scores according to the number of species photographed, and they therefore search in every possible habitat to obtain the largest possible number of species. Consequently, the species richness resulting from these tournaments is expected to be higher than when traditional visual survey techniques are used. In Spain, and particularly in Catalonia, these tournaments began in the 1980s and have now become regular. The objective of this study was to assess species richness and classify fish communities based on the underwater free-diving fish photography contests held in Catalonia at four different locations from 2008 to 2015, including a no-take marine protected area (Medes Islands marine reserve).

In total, 16307 photographs corresponding to 29 different contests were analysed. The mean number of participants per contest was 17 and the average number of photographs per participant was 33. The number of photographs per participant has increased over recent years at two locations, Palamós and the Medes Islands, with no effect on the total number of species recorded (Table 1).

A total of 88 taxa belonging to 26 families were photographed (Table 2). The actual species richness was higher, as the genera Atherina, Sphyraena, Trachurus and Trachinus and the family Mugilidae were not identified to the species level. The families with the highest number of species were Sparidae with 17 species and Labridae with 15 species, followed by Blenniidae and Gobiidae with 14 and 10 species, respectively. According to the International Union for Conservation of Nature (IUCN) Red List categories and criteria, 2 of these species are not yet evaluated, 1 is data-deficient to assess its status, 80 are on the least-concern list, 3 are considered vulnerable, 1 is near threatened endangered and 1 is already endangered. It is worth mentioning that the species identified until now as Gobius bucchichi in this region, should be considered to be the recently described G. incognitus (Kovačić and Šanda 2016). The lists of species observed at the Medes Islands are 7 species that do not appear in the previous available fish inventory in this MPA (Dufour et al. 2007). Species richness and Margalef diversity indices showed significant differences among locations (Fig. 2). The multiple comparison Tukey test showed that only Mataró has a significantly lower species richness (Table 1), mainly due to the lower number of species of blenniids and gobids (Table 2).

Cluster analysis based on the species catch rates by location and year yielded four distinctive clusters (Fig. 3) corresponding to each specific location. Regardless of the year, each location was grouped in the same cluster, with one exception, the Medes Islands in 2015. The silhouette analysis (mean silhouette width=0.46) revealed that all contests had positive silhouette widths, which is indicative of correct classification within groups. The two-dimensional ordination of the 29 contests (Fig. 4) yielded a moderate level of ordination (stress=0.14), reflecting cluster separation through a gradual continuum of change. A gradient from north to south was observed among locations near the shoreline (Medes, Palamós and L’Ametlla). Consistently, the results of the cluster analysis showed that the most dissimilar group was represented by the contests held in Mataró.

The first cluster, grouping all the samples from Mataró, had a low number of gobiid and bleniid species, scarcity of mugilids and the absence of Atherinidae and European seabass (Dicentrarchus labrax), which mainly marked the difference of this group. An additional difference was the high values of species such as Scorpaena notata, Thalassoma pavo, Boops boops and Gobius geniporus.

The second cluster grouped all the contests of L’Ametlla de Mar, the southernmost location (the last contest held at the Medes Islands was also part of this group). The main differences compared with other groups were the high abundance of the genera Pagellus, Symphodus cinereus and Parablennius incognitus together with the scarcity of Ctenolabrus rupestris, Microlipophrys canevae, Symphodus melanocercus and Diplodus cervinus. In this group the relative abundance of European seabass was also high, and only comparable with the values observed at the Medes Islands.

The third and fourth cluster comprised the contests held at the northernmost location corresponding to a non-protected and protected area respectively with the exception of the most recent year (see above) for the latter group. The main features of this group were the low relative abundance of gobiids, the maximum observed abundance of some species of labrids (Symphodus melanocercus, Symphodus viridis and Labrus merula) and the scarcity of S. ocellatus. This group was also characterized by high abundance of commercial species (e.g. Epinephelus marginatus, Conger conger and Phycis phycis), including the red scorpion fish (S. scrofa), but it also had the lowest abundance of the other three species of the same genus (S. notata, S. porcus and S. maderensis).

Although the clusters did not show any temporal variability within locations, except for the last contest held in Medes Islands, some shifts can be depicted. Some temporal changes were observed in the contest distances within the Medes Islands cluster, in which the 2013 and 2014 contests were clearly separated from the earlier ones. Similarly, differences were observed within the Palamós (cluster C) contests: those from 2012 onwards were distant from the previous ones, while in L’Ametlla the temporal pattern was more gradual.

The post hoc test showed significantly higher catch rates at the Medes Islands MPA with respect to the other locations (Table 2). These higher rates were mostly observed in species of commercial interest such as P. phycis and E. marginatus, but the list can be extended to Conger conger, D. labrax, S. umbra, S. scrofa, D. cervinus and D. sargus. In addition, the MPA also showed significantly higher catch rates for two species of labrids that are of negligible commercial interest but are highly targeted by spear fishers (L. merula and L. viridis). Furthermore, the relative abundance of S. cinereus decreased from southern to northern locations, while S. melanocercus displayed the opposite pattern.

Overall, the obtained outcomes show that free-diving fish photography contests could be a very useful tool for analysing fish assemblages and species richness in the littoral system. In general, fish classification results grouped each location separately, as expected from their differences in latitude, depth range and habitat. In addition, the already known effect of the Medes Islands MPA (García-Rubies and Zabala 1990, García-Rubies et al. 2013) was evidenced by the higher abundance of species of high commercial values.

The high number of species recorded in this study exceeds by far those reported in any published study on littoral communities carried out in the Mediterranean (e.g. Consoli et al. 2016), and is only comparable to but still higher than the number of species reported in the marine reserve of Ustica Island (La Mesa and Vacchi 1999). It is worth mentioning that the high number of species reported in the study of Ustica Island could be due to the combined sampling methodology used (strip transects by SCUBA, SCUBA diving tracts and snorkelling), which minimized the constraints inherent in survey methods, particularly for cryptic species (Brock 1982, Willis 2001, MacNeil et al. 2008). This indicates the effectiveness of the analysed information source for measuring fish species richness with a single data source.

In the Mediterranean, the littoral fish community is under various anthropogenic threats (Bianchi and Morri 2000, Claudet and Fraschetti 2010, Costello et al. 2010), but systematic monitoring is limited to a small number of commercial species or to protected areas (e.g. Coll et al. 2012, García-Rubies et al. 2013). Although overall littoral fish monitoring would be unrealistic to finance, many commercial groups, such as sparids, are not subject to assessment, so they are not under any systematic monitoring programmes, and this lack of information is hard to justify. The species recorded in this study that have a known status are mainly on the least-concern list: only one is endangered (E. marginatus) and one is considered vulnerable (L. viridis).

Among the results shown in this study, of particular relevance is the capacity of free-diving photography contests to count and assess cryptic species, which can be missed by underwater visual censuses (Ackerman and Bellwood 2000, Willis 2001, Smith-Vaniz et al. 2006). A total number of 32 cryptic species were photographed during these contests, a figure that is also higher than in any previous studies targeting this group based either on visual censuses (e.g. Macpherson 1994, La Mesa et al. 2006, Bussotti et al. 2015) or destructive techniques (Patzner 1999, Kovačić et al. 2012, Glavičić et al. 2016). Although they contribute greatly to the species richness of littoral systems, small cryptic species are perceived as minnows (small non-game or non-commercial fish species) due to their small size and cryptic behaviour. Consequently, they are generally overlooked or underestimated (Ackerman and Bellwood 2000, Willis 2001), and it is difficult to gain support for monitoring and conserving them (Sheldon 1988). Therefore, an important part of the strictly substratum-dependent littoral fish community is usually ignored, preventing any changes from being detected both in the community and in the microhabitat of the substratum.

Overall, the most abundant species were C. julis, S. roissali, S. tinca, S. cabrilla, D. vulgaris and C. chromis. In numerous studies these species have been reported as the most common in the rocky reef habitats of the Mediterranean littoral system (Bell 1983, García-Rubies and Zabala 1990, Claudet et al. 2006). However, the most abundant species differed between locations. The most abundant species in Mataró was T. delaissi. The preferred habitat of this species, flat, rocky and sheltered from wave action of the open sea (La Mesa et al. 2004), match the Mataró habitat location, a small, relatively flat, natural rock over 150 m in length. Further characteristics of this location, offshore at a minimum depth of 6 m that minimizes wave action and light, are also considered to be preferred by T. delaisi (Zander and Heymer 1970). Mataró’s minimum depth also explained the absence of species typical of shallow waters, such as many species of blenniids (e.g. A. sphynx, L. trigloides and M. canevae), whose highest diversity (Kotrschal 1988, Macpherson 1994) is restricted to the narrow depth zone from 0 to 1 m (Illich and Kotrschal 1990) and the negative correlation between depth and blenniid diversity was recently proved (Tiralongo et al. 2016). This absence was also detected for some other species characteristic of shallow waters (C. argentatus and Z. zebrus). Another interesting observation in Mataró was the coexistence with high relative abundances of the four species of scorpenids (S. maderensis, S. notata, S. porcus and S. scrofa). On this basis, we could reject the mutual species displacements by interspecific competition for habitat or food resources that have been hypothesized for this group (La Mesa et al. 2004). However, based on the findings in the MPA, we cannot entirely reject this hypothesis. In the MPA, while S. scrofa was highly abundant, the other species of scorpaenids showed values generally below those observed in the non-protected locations. S. scrofa is bigger than the other species (Petrakis and Stergiou 1995, Ordines and Massuti 2009, La Mesa et al. 2010) and under the protection effect the species can get older and bigger (e.g. Halpern and Warner 2003) and easier to see.

The results on species richness and diversity index showed no positive effects on the MPA, as has been proposed in other studies (Guilhaumon et al. 2015). The results are indicative of the positive effect of the MPA on species of commercial interest, but not on all species, because responses to protection can be highly variable among fish taxa (Claudet et al. 2006). The positive effect of the MPA in contrast with the abundance observed in the closest comparable non-protected location (Palamós) was unquestionable for species of high commercial value. A total of eight species showed higher and significant values within the MPA: C. conger, P. phycis, D. labrax, S. umbra, S. scrofa, E. marginatus and D. sargus. The recovery of the dusky grouper in MPAs has been reported in numerous studies (e.g. Bell 1983, García-Rubies and Zabala 1990, Sahyoun et al. 2013) as has that of white seabream (Sahyoun et al. 2013). The positive effect in this particular MPA, the Medes Islands, has already been observed for most of the above-mentioned species (i.e. D. labrax, E. marginatus, D. cervinus, S. aurata and S. umbra) (García-Rubies et al. 2013). In addition, two species of labrids (L. merula and L. viridis) of no commercial interest but of recreational spear fishing interest were also favoured by the protection. L. viridis is considered vulnerable and its protection is recommended. The lack of information on fish sizes prevents us from analysing the well-known positive effect of MPA on this key indicator (Lester et al. 2009).

A different argument is that the response to protection can also be negative for certain taxa, and changes in fish assemblages in MPAs could occur because predation pressure is expected to be higher (Francour 1994, Ashworth and Ormond 2005). A potential negative effect on small cryptic species has been suggested (Willis and Anderson 2003, Claudet et al. 2006), but our results confirm that only two species were significantly higher at the unprotected site (P. gattorugine and T. tripteronotus). However, not all the differences in species abundance can be attributed to protection measures or to the type of data, because we also found a latitudinal gradient of variability that was unrelated to these factors. S. melanocercus, which prefers rocky areas and seagrass beds, decreases from north to south, while S. cinereus and S. ocellatus, which generally inhabits seagrass beds and sometimes soft bottoms and estuarine lagoons, increases from north to south. This partitioning is consistent with the previous categorization in the Catalan littoral system based on shore-based fishing tournament reports (Gordoa 2009, Boada et al 2017).

The type of information used in this study has several weaknesses, such as a potential underestimation of the most frequent species because the number of photographs per species and participant is limited to one, and the possibility that the same fish could be photographed by different participants. However, despite the potential weaknesses of photographic free-diving contest data, the results of this study show their effectiveness for evaluating species richness, in particular of cryptobenthic species, and for analysing littoral fish communities. Furthermore, it has also proved its worth as a tool for updating fish inventories. We can summarize that this type of contest enhances competition between participants to obtain maximum fish diversity. Additional positive aspects of this information source are that it is cost-effective, non-destructive, a potential observatory, an platform for interaction between scientists and free divers, and an alternative for spear fishers in MPAs. We conclude that the monitoring of photographic free-diving contests could be a complementary information source to scientific monitoring.