Mediterranean nekton traits: distribution, relationships and significance for marine ecology monitoring and management

Biological traits are increasingly used in order to study aspects of ecology as they are related to the organisms’ fitness. Here we analyze a dataset of 23 traits regarding the life cycle, distribution, ecology and behavior of 235 nektonic species of the Mediterranean Sea in order to evaluate the distribution of traits, identify rare ones, detect relationships between trait pairs and identify species functional groups. Trait relationships were tested using correlation and non-linear regression for continuous traits, parametric and non-parametric inference tests for pairs of continuous-categorical traits and cooccurrence testing for categorical traits. The findings have significant implications concerning the potential effects of climate change (e.g., through the relationships of the trait of optimal temperature), fisheries or habitat loss (from the relationships of traits related to tolerance ranges). Furthermore, some unexpected relationships are documented, like the inversely proportional relationship between longevity and age at maturity as a percentage of life span. Associations between functional traits show affinities derived from phylogenetic constraints or life strategies; however, relationships among functional and ecological traits can indicate the potential environmental filtering that acts on functional traits. In total, 18 functional groups were identified by Hill-Smith ordination and hierarchical clustering and were characterized by their dominant traits. For the assessment of the results, we first evaluate the importance of each trait at the level of population, community, ecosystem and landscape and then propose the traits that should be monitored for the regulation and resilience of ecosystem functioning and the management of the marine ecosystems.

Trait information is often found in printed sources (e.g. fish identification keys) and 142 terms are not standardized. Thus, no systematic review could be performed. Instead, the 143 information was collected from books, review articles or journal research articles by searching 144 using the species name and adding the name of the trait or relevant terms (e.g. "life span" instead 145 of "longevity", "reproduction" instead of "spawning period"). Regarding reference sources, we 146 preferred to use information from peer-reviewed publications or books over grey literature. As 147 the objective was to collect data on Mediterranean species, when there was a unique reference 148 about a species it was used regardless of the source, however, if there was information coming 149 from locations outside and inside the Mediterranean (e.g. from two different papers), preference 150 was given to the latter. If there were multiple sources of information from different areas within 151 the Mediterranean (e.g. from various papers), we chose the references from the central 152 Mediterranean (Ionian Sea). Regarding habitats, again we preferred information coming from 153 marine habitats (e.g. rather than brackish waters or lagoons) if a species is distributed in many 154 habitat types. In the case of species that can be found over different substrates and we had 155 different publications with variable trait values we tried to focus on the most common habitat. 156 The various traits comprise different types of variables: continuous (e.g. size or 157 maximum lifespan), range (e.g. depth) and categorical (e.g. spawning habitat: pelagic or 158 benthic). Concerning categorical traits, each species was assigned to one trait category/modality 159 per trait. The definition of these 23 traits and their modalities is provided in Supplementary Table 160 S1. The information on traits as continuous variables and the information on traits categories per 161 species as well as the bibliographic reference for the documentation of each trait per species can 162 be found at: 163 https://figshare.com/articles/Koutsidi_Moukas_Tzanatos_23_biological_traits_of_235_species/1 164 1347406 165 For the detection of patterns in their distribution, the range type traits (fecundity, optimal 166 depth, optimal temperature, trophic level) were expressed as such, while for statistical tests and 167 the detection of relationships between trait pairs they were used as the average between the 168 minimum and the maximum (thus used as continuous type traits). Continuous and range-type 169 traits were investigated for outlier detection using the Grubbs test (Grubbs, 1950). As this test is 170 parametric, it was performed after the log-transformation of the traits: longevity and maximum 171 length, while age at maturity was transformed using the square root to achieve normality. As 172 average fecundity and average depth could not be described by a normal distribution with any 173 possible transformation the Grubbs test was not used on these variables. Regarding the 174 identification of rare traits, the distribution of trait values (for continuous variable-type trait) or 175 the frequencies of trait categories (for categorical variable-type trait) allowed the identification of 176 rare traits as these shared by less than 5 % of species. 177 To identify potential relationships the continuous traits longevity, fecundity, maximum 178 length and depth were transformed using the natural logarithm. Each of the total of 23 traits was 179 examined for the existence of a potential relationship with all other ones, depending on their 180 types. Regarding pairs of continuous traits, Pearson correlation between all pairs of continuous 181 traits was used. As carrying out a test for multiple hypotheses increases the probability of a rare 182 event, the likelihood of incorrectly rejecting the null hypothesis (Type I error) increases; hence in 183 the results the Bonferroni correction was incorporated. Correlation has the advantage of 184 investigating relationships without assuming causality, however it can only detect linear 185 relationships. In cases where the residuals indicated a non-linear pattern, a polynomial 186 regression was additionally used to investigate the existence of non-linear relationships and, in 187 cases of better fit, these relationships are presented instead.

188
For the detection of relationships between continuous and categorical traits, the t-test or 189 Analysis of Variance (ANOVA) was used, depending on the number of categories of the 190 categorical trait. As these tests are parametric, the Shapiro-Wilk's test was used to test for 191 normality. Fischer's exact test (for traits including only two trait categories) or Levene's test (for 192 traits with more than two trait categories) were used to examine the homogeneity of variances. 193 Regarding normality, only longevity and maximum length (their logarithm) were found to follow 194 a normal distribution. As in general, both ANOVA and t-test are considered as robust inferential 195 tests (Zar 1999), if the tests for homogeneity of variance did not reject the null hypothesis the 196 parametric tests were used. In the cases where the parametric prerequisites were not fulfilled, the 197 Mann-Whitney or Kruskal-Wallis tests were used instead (Zar, 1999). Similarly, we incorporated 198 the Bonferroni correction here. 199 264 (61%) and solitary behavior (57%) were dominant trait categories among the species examined. 265 Flat (9%) and long body shape (7%), tropical (9%) and cosmopolitan geographic distribution 266 (3%), autumn (3%), winter (7%) and all year spawning (6%), hard substrate seabed type (9%), 267 ambusher mobility (6%), ambushing predation feeding behaviour (7%) and herbivorous (2%) 268 and zooplankton diet (9%) were found to be shared by below 10% of the total species.

269
The relationships between continuous traits indicated eight statistically significant 270 correlations, incorporating the Bonferroni correction (Table 3, Figure 3). Longevity had most 271 correlations, with three positive ones (long-living organisms are larger, have high trophic level 272 and dwell deeper) and one negative (long-living organisms have lower age at maturity -as a 273 percentage of life span). Fecundity was found to increase with optimal temperature ( Figure 3J), 274 while maximum length increased with both trophic level and depth ( Figure 3B, E). Depth and 275 trophic level were also positively correlated ( Figure 3H). Finally, in three cases, non-linear 276 relationships provided better residual fit than linear ones ( Figure 3C, F, I): fecundity was higher 277 in species of low and high longevity (R 2 =0.286, p<0.001), fecundity had the highest values in 278 intermediate depths (R 2 =0.055, p=0.004) and optimal temperature was higher in species of 279 intermediate depths (R 2 =0.046, p=0.008), in the last two cases the correlation model explaining a 280 low percentage of variance.

281
The relationships between continuous and categorical traits indicated 20 cases where 282 there are significant statistical differences in the value of a continuous trait between the different 283 modalities incorporating the Bonferroni correction (Table 4). The main findings are summarized 284 in Table 5 (but see Supplementary Figure S1 for pairwise comparisons between trait categories). 285 Longevity was highest in flat-shaped species, in ambushing and active predators, piscivorous 286 species and pelagic spawners. Fecundity was higher in atractoid and pelagic species; however, if 287 seabed type is also taken into account apart from open sea species hard substrate ones had higher 288 values. Maximum length had significant variation across six categorical traits, with the most 289 striking being the highest values in pelagics, pelagic spawners and non-migratory species. 290 Regarding depth, eurybathic species and benthic spawners were found to occur deeper, herbivore 291 diet/grazing behavior and euryhaline species were found shallower. The highest trophic level 292 was naturally found in piscivorous species, ambushing predators (and mobility type) and 293 eurybathic species. Optimal temperature was found to be higher in species of high and medium 294 mobility, stenothermal species and species of tropical distribution.

295
The trait co-occurrence analysis documented 170 (17.4 %) positive, 183 (18.8 %) 296 negative, and 622 (63.8 %) random modality associations (Figure 4). The modalities with the 297 highest number of positive co-occurrences are associated with the pelagic (e.g. free exposure 298 with 13 positive co-occurrences) or the benthic way of living (e.g. benthic spawning habitat with 299 14 positive co-occurrences, flat body shape and benthic habitat type with 13). The modalities 300 with the highest number of negative co-occurrences are associated mostly with the pelagic way 301 of living (atractoid body shape, sociability schools, seabed type water column and pelagic habitat 302 all had 15-17 negative co-occurrences). Additionally, deep body shape and solitary sociability 303 had relatively many positive and negative co-occurrences. Relatively rare trait categories (e.g. 304 tropical distribution, autumn spawning) had a small number of co-occurrences. At the scale of 305 entire traits, body shape, depth range, mobility and exposure had the highest cumulative positive 306 co-occurrences of all their modalities, while body shape, mobility, habitat type and exposure had 307 the highest cumulative negative cooccurrences. The lowest number of cumulative co-occurrences 308 (both negative and positive) was found for the traits: hermaphroditism, diet, seasonal migrations, 309 seabed type and optimal temperature. The Hill-Smith ordination resulted into many axes, each explaining a relatively small 311 variance percentage (in Figure 5, the first two axes presented explain 19% of cumulative 312 variance). The dendrogram resulting from the hierarchical clustering of the species coordinates 313 (Figure 6), if cut at a high value of dissimilarity (e.g. 70 %, not shown in the figure) distinguishes 314 three major groups of pelagic, benthopelagic and benthic species and the associated traits. At 315 lower dissimilarities (42 %), six main groups are extracted. As these groups include in some 316 cases a wide range of species with very different functional roles (e.g. group A includes pelagic 317 species spanning from swordfish and tuna to anchovy and sardine) it was deemed necessary to 318 determine grouping in lower dissimilarity levels (13.5 % as shown in the figure) resulting in the 319 definition of 18, more homogeneous internally, functional groups. The traits characterizing the 320 six major and the 18 minor functional groups are presented in Supplementary Table S2. From 321 both Figures 5 and 6 and the supplementary table, it is evident that while the coarse distinction 322 (six groups: A-F) can indicate the major groupings of nektonic organisms, the grouping at a 323 higher level of similarity can highlight major functional components of the nekton within the 324 ecosystem, like small pelagic species (Group 1) or herbivorous fish (Group 3). The nekton functional groups identified here can be a useful tool to study the ecology of the 328 Mediterranean Sea, both in analyses using empirical data and in simulation models that utilize 329 functional groups to operate (e.g. Ecopath - Pauly et al., 2000). Despite the fact that, from the 330 clustering of Figure 5, the initial choice would be to divide the dendrogram either in the three 331 major clades (pelagic, benthopelagic and benthic groups) or in the six groups (A-F) identified at 332 42% dissimilarity level, it is more informative and reasonable to use a lower dissimilarity that 333 leads to the distinction of groups with different actual functional roles (e.g. 1: small pelagics, 3: 334 vegetation grazers) or with higher homogeneity in traits (like the division the pelagic group A 335 that on average had large size into the species with small size in group 1 and those with larger 336 size in group 2). As nekton can be generally expected to occupy a variety of ecological niches 337 because of the variety in species traits like size, habitat use and diet, it is not surprising to have 338 more functional groups than those found in zooplankton (e.g. Benedetti et al., 2018). Further 339 steps examining only traits related to resource use could shed light into potential inter-specific 340 competition relationships and niche overlap (M. Koutsidi unpublished data). . This is also relevant to the concept of keystoneness (a keystone species being a 346 species whose importance for its community is disproportionally high in comparison to its 347 abundance - Bond, 2001). Here we document the rarity (even below 5% of the species total) of 348 autumn spawning (but also winter and all year spawning) as well as that of herbivory. Herbivory 349 is anyway considered a crucial aspect of ecosystem functioning as alterations in herbivory may and Northeast Pacific continental shelf seas (in that work the importance of piscivory is much 359 lower, however the dominant modality is that of generalist feeding, which was not used in the 360 dataset of the present work). It is true that herbivory is also carried out by other, benthic, biota 361 (e.g. echinoderms); however, they have different traits (e.g. mobility) that may change this 362 function. Successful seasonal spawning, like autumn and winter spawning -and also the success 363 of the recruitment that follows it-may be prone to various environmental factors, also possibly 364 affected by anthropogenic effects like fisheries (that are characteristically seasonal in the 365 Mediterranean) or climate change that may decrease the duration of the window-period suitable 366 for spawning ( Table 1). The above indicate the clear need for a holistic assessment evaluation of 367 traits including all biotic elements of the ecosystem (plankton, nekton and benthos).

368
The rarity of other traits like hard seabed type preference may be related to the relatively 369 small extent of this habitat type in the marine environment. Similarly, the low occurrence of long 370 body shape may be related to the scarcity in characteristics of the habitat (e.g. structurally 371 complex habitat for long body shape which here was indeed found to cooccur with cryptic 372 exposure). However, the similarly rare flat body shape cooccurs with soft seabed preference 373 which is a relatively common trait, indicating that trait relationships may be less straightforward. 374 Even though the body-shape trait categories in Beukhof et al. (2019) for North Atlantic and 375 Northeast Pacific are not exactly similar, the long and eel-like body shapes are overall more 376 common there, while the deep body shape tends to be more common in the Mediterranean.

377
It is important to note that trait rarity should not only be evaluated at the species level, but 378 also weighted with species abundance or biomass to indicate the actual "abundance" of traits in 379 the ecosystem (Violle et al., 2017). E.g. the planktivorous diet trait category may be rare if 380 evaluated using the number of species but very abundant as the small pelagic or benthopelagic 381 species that possess it have very high abundances. Still, the fact that it is shared by only a 382 handful of species may be a risk for ecosystem functioning, especially taking into account the 383 fact that these species are known to have interannual abundance fluctuations. Even more so, 384 some of these species like the European anchovy Engraulis encrasicholus and the European 385 pilchard Sardina pilchardus are under intense fishing pressure and have been shown to be 386 affected by climate (Tzanatos et al., 2013).  In the present work, longevity and age at maturity are negatively correlated, in ostensible 397 disagreement with previous studies documenting a positive relationship (Froese & Binohlan, 398 2000; Jarić & Gavcić, 2012). Contrary to these works (where age at maturity is expressed in 399 years), here age at maturity was examined as a percentage of the species lifespan. Indeed, if age 400 at maturity is expressed in years in our dataset, there is a positive correlation with longevity 401 (r=0.72, p<0.001); however here we had intended to determine how early or late a species 402 matures regarding its life duration. Therefore, species with a short life span tend to mature 403 relatively late in their lifetime. This can be interpreted, if we take into account that even a short-404 lived species needs to have completed an amount of growth to reach a minimum size and 405 biomass for reproduction (Beverton, 1963). 406 The positive relationship between size and trophic level found in this study has also been 407 documented in other works (e.g. Romanuk  The co-occurrence analysis indicates some positive and negative associations between 416 pairs of trait modalities. The main characteristics of small pelagic fish (e.g. Sardina pilchardus, 417 Engraulis encrasicolus) have the highest number of positive and negative associations with the 418 other modalities. The traits of the family Sparidae, such as deep body shape, hermaphroditism 419 (e.g. Sparus aurata) and grazing feeding (e.g Sarpa salpa), were found to have a relatively high 420 number of positive associations with other modalities. The detection of relationships between 421 traits is important not only as a way to explore the relationships of characteristics shaping life, 422 but also because it could be useful to predict the possible effects of anthropogenic pressures on 423 these traits. For example, climate change can be expected to favour thermophilic species, thus 424 traits related to high optimum temperatures (high fecundity, deep body shape, high mobility) 425 may be favoured as well. Koutsidi   The current work has some findings differing from those of Koutsidi et al. (2016), as e.g. 439 the associations of depth with fecundity and optimal temperatures documented here are not 440 reported there. This can be a result of the inclusion of many more species here, but also because 441 of the treatment of traits according to the variable type here and not as categories in all cases. 442 Many rare traits identified like ambushing feeding behavior, flat and long body shape, autumn 443 spawning, cosmopolitan and tropical distribution and low trophic level and fecundity are 444 common in both works while others like long lifespan and distribution in deep water are novel 445 here and again indicate that continuous traits are better analyzed as such.
Naturally (and as shown also here) traits are related. This is not only with regard to life 447 strategies shaped by evolutionary processes (e.g. larger species having longer life duration too), 448 but also as they may be relevant (e.g. diet and trophic level). Still, apart from the trait affinity, 449 and despite the fact that there is the tendency to try to include only functional traits (i.e. more 450 relevant to life cycle and resource use), different traits may still convey different information and 451 still vary (e.g. as shown here piscivorous fish tend to have higher trophic level, but may still span 452 a range of trophic level values depending on their prey). Furthermore, even relevant traits may 453 incorporate information with different significance for ecosystem functioning or resilience (see 454 e.g. the significance of traits regarding tolerance range for variables like temperature and depth 455 in comparison to the optimal values of these factors in Table 5). Thus, depending on the research 456 question, some, even relevant traits can be useful for the evaluation of findings.

457
Here, we have not limited our analyses to the 13 functional traits (Violle et a., 2007) of 458 our dataset, but also include ten traits that would be rather characterized as ecological 459 (Beauchard et al., 2017). The determination of relationships between functional traits can 460 indicate affinities derived from phylogenetic constraints or indicate life strategies as they have 461 been shaped by evolution. Relationships between ecological traits can indicate the major aspects 462 of organismal distribution in time and space, while relationships among functional traits and 463 ecological traits can indicate the potential environmental filtering that acts on functional traits. It 464 should be noted that, apart from traits that are not related with many others (like age at maturity 465 and fecundity), most traits were found to have significant relationships with both functional and 466 ecological other traits independent of themselves being of a functional or ecological significance, 467 thus showing some level of environmental filtering on functional aspects; however these 468 relationships should also be examined weighted by abundance or biomass. Anthropogenic 469 stressors may also affect differently these two trait types, as e.g. fisheries act on both functional 470 (e.g. life-cycle, behaviour) and ecological (e.g. habitat) aspects of a population, while the effect 471 of climate is primarily ecological (e.g. temperature optimum and range), but both stressors can 472 have indirect effects on other, related, traits.

473
In this work, we have assigned each species to a single modality per trait. However, it is 474 true that traits can vary across individuals or populations (Violle et el., 2017) and in some cases a 475 species could be assigned to have more than one modality in a quantitative way (e.g. a species 476 spawning from December to April is here assigned to spawn in winter, while alternatively it 477 could be assigned as spawning in 100% of winter months and 33% of spring) using fuzzy 478 coding. Such an approach should be evaluated in future works not only regarding the correct 479 assignment of information to modalities, but also to account for species plasticity thus rendering 480 the analyses more realistic (Chevenet et al. 1994). In the same context, while we used one value 481 as representative of a species, trait values may vary across the entire species distribution or the 482 region examined (e.g. there may be differences between the western and eastern Mediterranean, 483 especially in some continuous traits). Additionally, while in many cases, trait datasets are 484 assembled only regarding the mature stages of a species, it is true that juveniles can possess 485 different trait values (e.g. diet). Furthermore, the major part of abundance/biomass of a 486 population may belong to the juvenile stages. An interesting approach would be to enrich trait 487 datasets with distinct trait values between juveniles and adults. Data from monitoring programs 488 regarding population structure (e.g. through the length distribution that is typical in fisheries 489 monitoring) could be used along with community composition for a more realistic depiction of 490 actual trait space occupied. This would be a very valuable expansion of trait datasets and their 491 usage; however, information on juvenile stages might be hard to obtain, especially for non-492 commercial species, whose biology is in some cases not fully documented. Finally, basic 493 biological research (especially for the largely unknown deep-sea fish) and assembly of even 494 more extensive trait datasets in terms of species, at least with a focus to the traits that are more 495 representative of functional diversity would help fill the gaps of the trait database, especially 496 with the aim to focus on understudied Mediterranean ecosystems.

497
The question as to which traits constitute more fundamental information for ecosystem 498 functioning still pertains. As ecosystem functioning is related to the transfer of energy and 499 material and the regulation and maintenance of ecological processes (Naeem et al. 1999; 500 (Table 1). In plants, it has been shown that, in cases where response traits are also effect traits, 506 there can be loss of ecosystem function (Suding et al., 2008). Moving from the scale of nekton to 507 the scale of the entire ecosystem, i.e. incorporating biotic components like plankton and benthos, 508 as functional traits like maximum length and diet/trophic level are key for energy transfer they 509 should be prioritized to be taken into consideration. Pecuchet et al. (2019), investigating 510 ecosystem-wide functional reorganization in the Baltic Sea by examining multi-trophic 511 communities indicate that diet or type of feeding traits are in many cases relevant for many of the 512 groups examined; however, they underline that different traits are involved and demonstrate 513 diverse dynamics among areas. Litchman et al. (2013) refer to the significance of body size as a 514 trait that is related to many others that are worth monitoring for ecosystem studies (like growth 515 rate, stoichiometric requirements, grazing rate and trophic niche breadth), but underline that the 516 choice on which traits to monitor ultimately relies on the questions asked. In this regard, 517 ecological traits regarding spatial and temporal distribution (habitat and seabed type, 518 environmental variable ranges and optima) or occurrence of critical life cycle events like 519 spawning (spawning period and habitat) could also be useful in supplementing this information 520 to describe the spatio-temporal cooccurrence of the different elements for a holistic evaluation of 521 the marine ecosystem.

522
With the global ocean being under a multitude of anthropogenic effects (e.g. Crain et al., 523 2009), it is crucial to identify traits that are significant for monitoring human induced alterations 524 in the structure and dynamics of the marine ecosystem. These traits are not only important as 525 descriptors of the marine community (see e.g. the "mean temperature of the catch" -Cheung, 526 Watson & Pauly, 2013) useful in monitoring, but should also be maintained to some minimum 527 levels, to avoid function loss or the creation of too many empty ecological niches that could 528 more easily be colonized, e.g. by alien species (Givan et al., 2017). Thus, regarding climate 529 change, traits like optimal temperature and temperature range are significant, as communities 530 with a diversity of thermal affinities and narrow ranges of thermal tolerance are more sensitive to 531 climate change (Burrows et al., in press); yet all range-type traits and also distribution seem to 532 have important implications for ecosystem dynamics and resilience. Regarding fisheries effects, 533 size (also because of the various significant relationships it has with other traits and its 534 implications for management) is a crucial trait. Other core biological traits like longevity, 535 fecundity and age at maturity are also important for fisheries management, but also some 536 behavioural traits relevant for the interaction of nekton with fishing gear may bear some 537 importance. Lavorel & Garnier (2002) and Violle et al. (2007) suggest that traits whose attributes  Table 1 we also 542 document that, in some cases, a response trait (e.g. optimal temperature), having been affected 543 by change in environmental conditions may in turn act as an effect trait altering subsequently the 544 dynamics or composition of the community. This is especially important regarding the direct and  Relationships between biological traits have long been studied to investigate how evolution has 553 shaped life into form and function and how characteristics are combined into life strategies. 554 However, the study of nektonic trait distribution and combinations presented here can be useful 555 to elucidate trait interactions significant for indirect alterations of ecosystem functioning, 556 especially today, when the marine environment is under a multitude of anthropogenic stressors 557 that can act on specific traits. The documentation of rare traits (like winter and autumn spawning, 558 herbivory, very low or high size or fecundity, hard substrate type and ambushing predation) in 559 species together with the appraisal of the significance of traits at various scales indicates aspects 560 of crucial importance that need to be preserved. Under a more synthetic scope, nektonic 561 functional groups are broadly determined around major aspects of habitat use (pelagic, 562 benthopelagic or benthic), but can be distinguished in more detail showing affinities among and 563 between functional and ecological traits that can be used in the future to understand nektonic 564 communities and model ecosystems. In any case, the documentation of a multitude of 565 relationships between functional and ecological traits found here indicates how the environment, 566 through the delimitation of species distribution in space and time depending on their traits, can 567 filter functional traits, while the validation of functional trait associations hints at functional 568 interdependencies determined by evolution. The findings documented here highlight the traits 569 that should be evaluated and monitored in the future both at the level of nekton or in combination 570 with other major ecosystem components for the assessment of ecosystem functioning and those 571 that should be maintained to ensure ecosystem resilience.  Higher longevity renders individual more important both as prey and as a predator as more instances of predation Longevity is related with natural mortality and thus with energy transfer in the ecosystem 4 Longevity and age at maturity are related with the ability to recover from anthropogenic disturbance 27, 50 May indicate population stability over time and potential of the various life stages to disperse 7 2 Age-at-maturity (30% of lifespan) Early maturity may increase resilience in unfavourable environmental conditions 2 NR Ecosystem characteristics (e.g. productivity) may enhance or delay maturation Longevity and age at maturity are related with the ability to recover from anthropogenic disturbance 50 Associated with cessation of growth 25 Early maturation may increase resilience in high exploitation rates. Maturity significant for fisheries management (measures planned to ensure population part achieves sexual maturity) 3 Fecundity (5-10 eggs) If low should ensure offspring survivalpopulation fitness, as energy allocated to survival of offspring or fecundity (r/K-selection strategy) 47 High fecundity means higher abundance of young "defenceless" stages (eggs, larvae) that are possible prey for other populations, but higher inter-specific competition later on 2 As it provides easy-to-capture and rich in energy prey (compared to adult prey) may influence energy flow rates Together with mortality until recruitment may affect stock size which is very relevant for fisheries 24 4 Hermaphroditism (gonochoristic) Sexual maturity of the second (in succession) sex must be achieved through survival to guarantee successful spawning and recruitment NR NR As both sex ratio and gear selectivity change with size, exploitation of one size part of the population may affect sex ratio and possibly reproductive success. Related to individual biomass, food web position, abundance, metabolic rates, and dispersal 7 As in the marine ecosystem there is an "eat what is smaller" pattern, there has to be some variability in species sizes to support a community 17,30 Related to energy flow in the ecosystem (because of association with trophic level/diet) and resulting food webs 16,23 Relevant for fisheries (with regard to body shape) for selectivity 24 6 Body form (flat) Related to position in the water column/habitat, diet/potential prey, activity 21,59 Because of association with habitat, specific communities may have higher frequencies of some body forms Not relevant Related to the way fishing gear may affect selectivity (together with size) 24 7 Optimal depth (0-50 m) Physical factor determining potential species habitat 7 Depth is a major factor shaping marine communities 46,57 Depth may affect productivity and energy flow as e.g. below the euphotic zone the lack of primary production modifies trophic links.  Manuscript to be reviewed  Figure 1 A. Distribution of continuous traits of the nekton species examined. B. Ranges and/or means of range-type traits. Species are ranked according to the mean of the range.
Note: In the bibliography, sometimes fecundity is provided as the maximum number of offspring with no indication of the minimum. In these cases it is here denoted not as a range, but with the same symbol as the mean.