Marine ecosystems are subject to escalating pressure from the cumulative impact of multiple anthropogenic stressors (e.g., pollution, eutrophication, ocean acidification, and fishing), causing biodiversity loss, habitat degradation, and stock declines (Halpern et al., 2008, 2015; Coll et al., 2012; Micheli et al., 2013a). Fishing activities, in particular those employing non-selective gear such as bottom trawling and drift nets, are considered one of the most important anthropogenic sources of marine ecosystem decline, causing both direct (crushes and buries marine animals) and indirect (sediment removal, alteration of water-column fluxes, reduction of the original complexity of fishing grounds) impacts on marine populations and habitats (Watling and Norse, 1998; Jackson et al., 2001; Puig et al., 2012). These impacts are evident in the Mediterranean Sea, which combines a long history of exploitation with a high level of social, economic, and political complexity that present major challenges for effective marine management and conservation (Coll et al., 2012; Micheli et al., 2013a,b). To date, 85% of assessed Mediterranean stocks are overfished (Colloca et al., 2013), and current fisheries management is considered inadequate (Fouzai et al., 2012). The management strategies adopted in the Mediterranean basin largely take a single species approach instead of an ecosystem-base management approach (de Juan et al., 2012). Most regulations are aimed at reducing fishing effort and fishing capacity, and/or at implementing technical measures such as the regulation of mesh size, the establishment of a minimum landing size, and temporal, mostly seasonal fishing closures (de Juan et al., 2012; Fouzai et al., 2012; Colloca et al., 2013). However, scientific advice is rarely used to implement the spatial and temporal fisheries management strategies needed to achieve sustainable yields and to preserve the ecological role of the exploited species and their habitats (Colloca et al., 2013).

A major shift in management focus has occurred over the last 10 years through an increased awareness of the fundamental role played by habitat in fished stocks conservation and recovery, which has, in turn, led to the key concepts of Vulnerable Marine Ecosystems (VMEs) and Essential Fish Habitats (EFHs) (UNGA, 2006; FAO, 2009). VMEs and EFHs include both water column and sea bottom areas that support the productivity of commercial species and that are vulnerable to human activities, in particular to bottom trawling (Rosenberg et al., 2000; FAO, 2009). VMEs and EFHs include spawning, nursery and feeding grounds, together with foundation species (Dayton, 1972), i.e., “a single species that defines much of the structure of a community by creating locally stable conditions for other species, and by modulating and stabilizing fundamental ecosystem processes.” This is the role played for examples, by the animal forests, in particular anthozoans (Cerrano et al., 2010; Valisano et al., 2016) whose functional and structural role is receiving increasing attention (Rossi et al., 2017). Numerous initiatives have been developed in order to map the presence and distribution of VMEs and EFHs, and to provide useful tools to help managers and decision makers in the selection of priority areas and in the definition of management plans to ensure the long-term conservation and sustainable use of marine resources (Stecf, 2006; OSPAR Commission, 2010; Rogers and Gianni, 2010; Rengstorf et al., 2013).

Unfortunately, the lack of historical information limits our ability to reconstruct habitat distribution and trends, and assess the current status of VMEs and EFHs. Most of the studies of changes through time have focused on decline of exploited fish populations or top predators (Barausse et al., 2011; Ferretti et al., 2013; Mazzoldi et al., 2014). More limited historical information is available for non-target species, such as benthic invertebrates caught as bycatch. Thus, reconstructing past distribution and abundances of benthic habitats and species is challenging. Such baselines and trends, however, are critical for assessing the current status of EFHs and VMEs and establishing reference targets for their recovery (Engelhard et al., 2016). Over the last decades, “Local Ecological Knowledge” (LEK) has emerged as an alternative approach to collecting information on species presence or abundances when historical data are lacking (Huntington, 2000; Anadón et al., 2009). However, up to now, the use of LEK in the Mediterranean Sea has been limited to collecting information and describing trends in fish diversity and abundances (Azzurro et al., 2011), and discarding of commercially important fish species in the bottom trawl fishery (Damalas et al., 2015a,b). Here, we apply LEK to examine the temporal change of habitat-forming invertebrates in the Adriatic Sea.

The Adriatic Sea is one of the most productive regions of the Mediterranean Sea, hosting a variety of endemic species, and important nursery, spawning, and foraging grounds (Coll et al., 2010; de Juan and Lleonart, 2010; Colloca et al., 2015). Humans have exploited the Adriatic Sea since the prehistoric era (Lotze et al., 2011). This long history of human use, together with global environmental changes (Conversi et al., 2010; Zenetos et al., 2011; Giani et al., 2012) have greatly altered the Adriatic marine environment and ecosystems (Coll et al., 2007, 2009, 2010; Lotze et al., 2011), and ranked the basin as one of the most threatened regions of the Mediterranean Sea (Micheli et al., 2013b). The description and distribution of Adriatic benthic communities have been studied from ancient time both, on a larger scale (Vatova, 1949; Gamulin-Brida, 1974) and a local scale (Paolucci, 1923; Scaccini, 1967; Scaccini and Piccinetti, 1969; Fedra et al., 1976; Crema et al., 1991) with an exhaustive description of its biocoenosis and biodiversity of megabenthic species. Several studies, most of which conducted in the northern Adriatic Sea, have described negative trends and chronic effects of commercial species and benthic communities due to trawling activities (Hall-Spencer et al., 1999; Jukic-Peladic et al., 2001; Pranovi et al., 2001, 2005; Morello et al., 2005; Romanelli et al., 2009). More than 90% of Adriatic marine resources are depleted and the current management of fisheries is inadequate (Lotze et al., 2011; Fouzai et al., 2012). Mean discard rate in Adriatic bottom trawl fisheries ranges between 20 and 67% of total catches, higher than the Mediterranean average (Tsagarakis et al., 2013; FAO, 2016), with a rate that varies according to fishing intensity.

Little is known about temporal variation in the abundance of megabenthic species, foundation species, VMEs and EFHs in the Adriatic Sea. In the northern Adriatic, studies have revealed a shift from benthic communities characterized by the presence of filter-feeding epifaunal organisms forming complex 3D habitat (such as sponges, sea pens, ascidians, holothurians, and large bryozoans) to a community dominated by infaunal and scavengers species (Raicevich et al., 2004; Lotze et al., 2011). This information is not available for other Adriatic sectors. In this study, we used LEK to describe changes in the abundance of habitat-forming megabenthos, and highlight species and ecosystem at risk.

We conducted a total of 44 interviews (to 25 fishers from Ancona, 12 from San Benedetto, and 7 from Civitanova Marche). The age of interviewees ranged from 42 to 82 years, with 80% of them older than 50 (around 64% of fishermen were 50–60 years old; fishermen between 40–50 years and between 60–80 years were 18% of interviewees). Only 20% of fishers gave a detailed description of the otter trawl gears they use, the others only stated they use otter trawl gear.

The results of the ordinal regression mixed models indicate an overall reduction of the analyzed species over time (p-values for time ranges from <0.001–0.04; Figures 1B–G; Table 3). Of all the independent variables used, time was significant for all species (p-values <0.001–0.04; Table 3), longitude was significant in P. jacobaeus, Neopycnodonte cochlear, and Holothuria spp, (p-values <0.001–0.005; Table 3). The residuals of the regression between latitude and longitude was significant in Amathia semiconvoluta (p-values < 0.002; Table 3). Distance from the coast was significant in P. jacobaeus, A. semiconvoluta, and Holothuria spp. (p-values from <0.001 to 0.33; Table 3), while the interaction between time and distance from the coast was significant in A. semiconvoluta, P. jacobaeus, and Holothuria spp. (p-values <0.001–0.007; Table 3). Moreover, in each model, the random effect was significant (p-values always < 0.001) indicating that individual fishers added a non-negligible level of subjectivity in their perception of the changes in abundance of the selected megabenthic species.

Declining abundances were reported for the sponge Geodia cydonium (Figure 1B), while the abundances of the sea pen Funiculina quadrangularis remained relatively stable from the 1980s up to now, except for very few sub-areas where the species showed a slight decline (Figure 1C). The bryozoan A. semiconvoluta is the only species that, based on the fishers' perception, had an increasing trend in the last 40 years (Figure 1D). Species belonging to the phylum Mollusca, in particular the scallop P. jacobaeus, the fan mussel Atrina fragilis and the deepsea oyster N. cochlear, have declined from very abundant to rare in the study area from the 1980s to the present time (Figures 1E–G). The abundance of the holothurians also shows declining trends, even though Holothuria spp. are perceived by fishers as still common in most of the central Adriatic sea-bottoms (Figure 1H).

Our results did not reveal significant spatial patterns in the trends of species. These trends were similar throughout the study area and no significant differences were apparent between coastal and offshore areas.

Fishers recognized all the species listed in our survey allowing us to map several of them in the central Adriatic fishing grounds (Figure 1; Supplementary Figure 1). In particular, the sponge G. cydonium was recognized by 84% of interviewed fishers, the bryozoan A. semiconvoluta by 65%, the sea pen F. quadrangularis by 74%, the bivalves P. jacobaeus, A. fragilis, and N. cochlear by 77, 98, and 84% respectively, and Holothuria spp. were recognized by 100% of the interviewed fishers (Table 2; Figure 1). The sponge G. cydonium was reported mainly in offshore sub-areas (Figure 1B). The bryozoan A. semiconvoluta has been found in several sub-areas of the central Adriatic, usually as a rare occurrence, but with increasing abundances moving toward offshore sub-areas (Figure 1D). P. jacobaeus was reported mainly in the northern sectors and in sub-areas no deeper than 70 m, while A. fragilis and N. cochlear were collected also in deeper sub-areas (Figures 1E–G). Holothurians were reported in almost all of the analyzed sea bottoms of the central Adriatic Sea (Figure 1H). Although all fishers provided us with answers about the abundances and the geographical location of common invertebrates (such as molluscs bivalves or holothurians), only a fraction of them detected the presence in the bycatch of anthozoan species, which occurrences are less frequent and distribution more patchy in the Adriatic sea-bottoms (Table 2; Supplementary Figure 1).

The impact of towed gears on benthic communities has been extensively studied in many exploited demersal ecosystems of the world (Dayton et al., 1995; Collie et al., 1997; Jennings and Kaiser, 1998; Hall-Spencer et al., 1999; Thrush and Dayton, 2002). These destructive practices have contributed to the decline of habitat-forming species, VMEs and EFHs worldwide. This study reveals that using fishers' LEK can provide a useful tool to describe long-term trends of both target and non-target species. Other studies have recently demonstrated this methods is useful to detect patterns in exploited Mediterranean fish populations (Fortibuoni et al., 2010; Azzurro et al., 2011). Here we highlighted its utility also for a broader range of bycatch species across different marine taxa, including those important to structure benthic habitats. Species such as sponges, bivalves, and holothurians that historically were reported as common in the soft-bottom communities of the central Adriatic Sea (Scaccini, 1967; Scaccini and Piccinetti, 1969; Gamulin-Brida, 1974) were perceived to decline in the last 40 years, especially in the decade 1980–1990 (this study). In contrast, the bryozoan A. semiconvoluta increased its distribution and abundance in some areas during the surveyed period, revealing that fishers can easily detect the increase of megabenthic species, particularly those that may affect fishing activities. Fishing grounds where A. semiconvoluta is present in high abundance, in fact, are usually not trawled because colonies of this bryozoan can clog trawl nets' meshes (Grati et al., 2013; Salvalaggio et al., 2014).

Our study reveals that LEK may also provide a reliable and alternative source of information to study the spatial distribution of the benthic invertebrates. Clear spatial patterns in the distribution of the selected species in the Adriatic fishing grounds were apparent. In the Adriatic Sea, P. jacobaues lives in sandy bottoms shallower than 70 m (Piccinetti et al., 1986), and this aspect was confirmed in our interviews. Moreover, our analysis showed that P. jacobaeus is found by fishers also in southern areas respect those previously described, even if in the same bathymetric range were the bivalves used to live. Higher numbers and biomass of Holothuria tubulosa and Holothuria forskali were found between 20 and 100 m depth, unevenly distributed (Šimunović, 1997; Šimunović et al., 2000). Our analysis revealed that the presence of Holothuria spp. goes from 20 m depth down to the deeper sub-areas of the central Adriatic Sea. Thus, we can suppose that environmental factors, such as depth, may be considered directly related to the distribution of Adriatic benthic invertebrates. LEK was also useful to detect rare and spatial restricted species such as the cnidarians Madrepora oculata, and Lophelia pertusa, which were only previously recorded in death assemblages in the Pomo/Jabuka Pit (Angeletti et al., 2014; UNEP/MAP-RAC/SPA, 2015). Although a small number of fishers gave us answers in relation to these scleractinan species, probably because a small fraction of the interviewees trawled in the Pomo pit area, our maps overlap with the known species distributions (Supplementary Figure 1).

LEK can be an instrumental management tool to reconstruct historical information, such as changes in fish community structure following commercial exploitation and climatic change, or to detect rare species, and species invasions (Berkes et al., 2000; Drew, 2005; Azzurro et al., 2011). In addition, LEK can be used to describe changes in fishing methods and strategies (Damalas et al., 2015a,b), leading in some cases, to approaches of adaptive and qualitative management strategies of marine resources and ecosystems (Berkes et al., 2000). Here we aimed to demonstrate LEK's utility and potential applications as an information tool to characterize the structural changes and alteration of benthic invertebrate assemblages, often unmonitored in conventional fisheries management. Fishers' perceptions may represent in some cases the only option to reconstruct historical baselines for habitats status and to map potential VMEs. Thus, LEK may represent an additional tool to help driving actions needed to reach the ecological targets of “Good Environmental Status” (GES). In fact, the maintenance of benthic biodiversity, sea-floor integrity, and a good status of benthic ecosystems through the protection and restoration of benthic sensitive species and habitats are among the targets of the 11 descriptors of GES of the European Marine Strategy Framework Directive (MSFD-EU, 2008). Moreover, LEK may contribute to the Habitat Directive (92/43 CEE) through the identification of priority habitats present in the central Adriatic Sea, such as biogenic-carbonate reefs or oyster reefs, representing rich and fragile biotopes affected by the high pressure of destructive fishing (Conti et al., 2002; Beck et al., 2011; Taviani et al., 2015). Thus, LEK could provide important information for defining areas to be protected from trawling, providing maps of hotspots of biodiversity, priority habitats and areas with presence of VMEs and EFHs, promoting the development of an efficient and sustainable management of the Adriatic fishing as aimed by the Common Fisheries Policy (CFP) of the European Union (EU).

Despite the potential of LEK for describing temporal changes and spatial distribution of benthic invertebrate species, some limitations of this approach emerged from our analysis. In particular, for some of the selected species (see Table 2) information about their presence in the bycatch is limited. This could be related to the fact that fishers do not pay particular attention to species that are not commercially important, or that these species are not so abundant to be commonly observed, or do not affect fishing activities. The interviewed fishers also trawled different fishing grounds with different bottom characteristics and species associations. Thus, the description and identification of the selected megabenthic species, and the likelihood they are observed by fishers, could be related to the natural distribution of the benthic species and to the characteristics of the Adriatic fishing grounds. Moreover, the difference in the number of fishers' responses for the sub-areas identified in our study could be related to the port of origin. In particular, the northern and southern analyzed sectors might be trawled only by a subset of the interviewed fishers, depending on the geographic location of their port of origin. Thus, the number of observations for these sub-areas is smaller compared to the central sub-areas because of their greater distance from the different ports of the Marche region. It was not possible to control for these aspects in this study, but future analyses should address these issues. Finally, our models suggested that there was a significant variability in the response of the individual fishers (random intercept in our model). This aspect needs to be considered when analyzing results from interview surveys to obtain unbiased parameter estimates for other fixed effects. The variability among fisher may be due to a variable perception of abundance among individuals due to experience, recollection ability, and any other factor capable of biasing the index of abundance being modeled (Grant and Berkes, 2007). In the absence of specific information to control for these biasing factors, it is reasonable to assume that each fisher influenced the variability of the responses in a random fashion according to a normal distribution with mean 0 and standard deviation to be estimated from the data.

The widespread perceived decline of benthic species playing important ecological roles (Table 1) in the central Adriatic may have altered the Adriatic marine ecosystem functioning over the past decades. Changes in benthic invertebrates we described here are congruent with patterns of decline described by other authors in the northern Adriatic through use of standard sampling methodologies such as dredges and trawl surveys (Scardi et al., 1999; Raicevich et al., 2004). These studies reported a net reduction of the ratio discards/commercial species, with a decline or disappearance of large filter feeding organisms (e.g., the sponge Geodia) documented from 1980s to 2000s (Raicevich et al., 2004) together with a general decreasing trend of the diversity of macrobenthic assemblages (Scardi et al., 1999). In other ocean sectors, the declines in benthic invertebrates triggered entire regime shifts (Kaiser et al., 2000; Jackson, 2001) and we expect that similar consequences may have occurred also for the Adriatic Sea. Detecting the occurrence of these ecological changes is of paramount importance for future studies.

Several factors may have driven the declines of megabenthos species living on soft bottoms of the Adriatic basin. Declines of sponges and other benthic invertebrates, for example, has been associated with anoxic events (Fedra et al., 1976) in the northern Adriatic basin. Climate change, such as temperature anomalies, caused mass mortalities events in the central basin (Di Camillo et al., 2013; Di Camillo and Cerrano, 2015; Kružić and Popijač, 2015), and direct and indirect impacts of human activities, such as fishing, have reduced the biodiversity and the complexities of the Adriatic benthic communities (Raicevich et al., 2004; Pronzato and Manconi, 2008; Lotze et al., 2011). While we cannot exclude the influence of multiple factors in driving the decline of megabenthic species described here, our analysis supports the hypothesis that intense trawling in the Adriatic Sea over the past decades may have been a major factor determining the alteration of the Adriatic soft bottom communities. In 1980s, Italian Adriatic regions reached the maximum number of fishing vessels together with the complete development of highly damaging fisheries introduced in the 1960s (Froglia, 2000; AdriaMed, 2004; Romanelli et al., 2009). In the 2000s the total number of fishing vessels decreased (AdriaMed, 2004), however, new technologies such as GPS systems have been introduced, improving the exploitation of new fishing ground (Fortibuoni et al., 2017) and the total fishing pressure on Adriatic seabed bottoms is currently considered unsustainable. Because the LEK data we collected in our study to detect fishers' perceptions is mainly qualitative, our models did not detect clear patterns moving from coastal to offshore areas. However, distance from the coast is one of the most important variables affecting our regression models, for example for P. jacobaeus. In particular, our model suggests that at increasing distance from the coast, higher classes of abundance are more likely (Table 3). This relation with the proximity to the coastline is characteristic of a community being exploited, such as coastal communities that typically are exposed to a heavier and more prolonged history of exploitation than those offshore. Automatic Identification System and Vessel Monitoring System analysis clearly revealed that trawling fishing effort is higher in coastal areas with respect to offshore areas in the central Adriatic Sea (Santelli et al., 2017). However, chronic and intensive effects of bottom trawling fishing, with habitat degradation are well-known (Pusceddu et al., 2014), and the long-term exploitation of the Adriatic basin could have homogenized and simplified Adriatic soft bottoms habitats and species composition even in offshore areas. In particular, habitats formed by slow growing and long-lived specimens such as sea pens, hydroids, or corals, have a high vulnerability to fishing and even reduced fishing effort may cause considerable damage to these species, preventing their recovery (Troffe et al., 2005; Greathead et al., 2014). Moreover, the impacts of trawl fishing gears on the seabed differ depending on the sediment compositions and on bottom trawl target species (Pranovi et al., 2001, 2005; Eigaard et al., 2016). Gear characteristics (e.g., changes in number, the size of meshes in the cod end net, modification of the design of the doors, and other parts of the trawl net) also possibly affected the level and the type of damage by trawling gear on megabenthos. Our study did not consider different gear types, thus the pattern described by fishers is only relevant to a specific type of fishing gear. All the interviewed fishers were otter trawler and used a fishing gear that is generally standard across our focal area (that is an Italian otter trawl as specified in Fiorentini et al., 1999). However, because the interviewed fishers in most cases did not give us the specific characteristics of their fishing gears (e.g., detailed size of trawl net and numbers of used gears per haul), it was not possible to confirm and clearly relate the fishing effort and fishing gear characteristics with the observed megafauna trends. More detailed analysis and new interviews are needed to fill these gaps and to explore the most adequate restoration measures (Bastari et al., 2016) that need to be urgently adopted.

Historical studies are fundamental for understanding long-term changes in marine ecosystems. LEK surveys provide an opportunity to fill this knowledge gaps as we demonstrated here by focusing on historical changes of benthic invertebrates species in the exploited Adriatic Sea. These approaches provide an opportunity to reconstruct reference points for benthic communities and may help management in setting recovery target for ecosystem structure and even function at local and regional scale. Therefore, extending these studies on a broader geographic scale is a promising approach for drawing historical baselines and inform marine management.

Ethics approval was not required for this study as per institutional guidelines and Italian law and regulations. In compliance with the aforementioned guidelines, laws and regulations, oral informed consent was obtained from all research participants. All potential interviewees were provided of the purpose of the study and of the usage of collected data before obtaining their consent. Their answers were anonymized and it is not possible to link the statements back to individual subjects.

AB gave substantial contributions to the conception and design of the work, analysis of data, and drafting the work, and final approval of the version to be published. JB gave substantial contributions to the acquisition and analysis of data. FF gave substantial contributions to the work conception, analysis and interpretation of data, and revising the work critically for important intellectual content, and final approval of the version to be published. FM and CC conceive the work and gave substantial contributions to its design, revising the work critically for important intellectual content, and the writing of the article.