**4. Discussion**

Our understanding of the long-lasting effects of GCC on long-lived populations of marine benthic communities is quite limited. Generally, simulations try to fill this gap by devising possible scenarios of affected population trends [40,47] among others. An analysis of a dataset from 1998 to 2013 allowed us to follow the main trends of the community under study and the locally dominant gorgonian population before and after two dramatic mass mortality events occurring throughout the northwest Mediterranean during late summer in both 1999 and 2003.

The area where the temperate gorgonian *Paramuricea clavata* population examined here dwells is characterized by a sciaphilic epibenthic community settled on a vertical cliff. The entire area is affected by high sedimentation rates due to the outflow of the Magra river [38]. The high concentration of seston typical of this area is likely a factor favoring passive suspension feeders [32]. Cnidarians, and the gorgonian *P. clavata* in particular, are the dominant group, while the algal cover is poor, limited by the high turbidity of the area.

The *P. clavata* population of Tinetto Islet has been divided into 18 annual age/size classes, according to the mean annual colony growth rate [20]. The distribution of size classes observed in 1998 in the crowded, pre-mortality population with abundant larger (older) colonies could be driven, by the effect of the increasing density of larger/older colonies on recruitment rate, progressively affecting the population over some years, leading to a structure characterized by a low abundance of recruits and smaller/younger classes. After the 1999 and 2003 mortalities, only a few of the largest sized gorgonians survived, and the structure of the population drastically shifted towards the smaller/younger classes. In 2007, colony density significantly increased and remained stable subsequently. As mortality disproportionally affected the larger/older colonies and fecundity is strictly correlated with colony size [21], a drastic reduction in the larger classes could have decreased the population's reproductive output. However, the abundant egg and planulae production of this species could have attenuated the effects of the loss of larger colonies on population resilience [21].

Ten years after the last *P. clavata* anomalous mortality event (2003) a twofold increase was found in the substrate covered by TSH, indicating that a clear-cut change in the whole community occurred. Concerning the *P. clavata* population, both % cover and density of adult colonies were also reduced, albeit to different degrees, as they were respectively one third and half those recorded in 1998. Recruitment, on the other hand, was four and a half times higher than in the pre-mortality population. The difference between the reduction in adult colony cover and the density found could be due to the different size/age structure of the post-mortality population, in which the number of larger/older colonies fell disproportionally [20], thereby affecting colony cover more than density.

Several papers have reported that algal turf coverage and high sedimentation rates negatively affect recruitment and reproduction in benthic communities [48,49] among others. For example, in the gorgonian *Eunicella singularis*, algal turf seems to negatively affect the process of settlement by decreasing the larva's access to suitable substrate [50]. Contrary to this general finding, *P. clavata* recruitment density increased since 2007, despite a significant rise in TSH % cover between 1998 and 2004 [45], and remained substantially unchanged in the following years. However, the lack of any correlation between the TSH and *P. clavata* recruitment or adult colony density likely indicates the absence of any linear effect of TSH cover on the gorgonian population examined. This finding seems to be confirmed by the similarity in TSH % cover between the two sites sampled which, however, differ in *P. clavata* adult and recruit densities. As no correlation between these parameters and TSH cover was observed, other factors likely caused the difference found between the sites. According to Padròn et al. [34], who examined the genetic structure and connectivity of *P. clavata* populations across our sampling area and six other sites in the eastern Ligurian Sea, the locally dominant currents may have fostered a higher larval input in our more external site (Tinetto Shoal), thereby partly driving the dynamics of this population. Indeed, the input of larvae released in other areas may play an important role in the recovery of

benthic sessile populations affected by anomalous mortality events. In *P. clavata* the larval stage can span between 6 and 26 days under experimental conditions [45,51], and planulae are able to travel distances of up to tens of kilometers. Recent genetic analyses in the Ligurian Sea suggest that tight connectivity among *P. clavata* populations affected by anomalous mortality can avoid their collapse [34], while highly frequent mortality events may lead isolated octocoral populations even to extinction [12]. Thus, the strategic position of the population under study and the connectivity with the other neighboring populations could be a key factor in fast recovery after disturbing events [51].

There is some hint towards a small-scale spatial constancy of recruitment over time; this is based on two findings: (1) a significant positive linear correlation between the density of recruits and that of the cohort of small/young adult colonies measured in the same plot; (2) some linear trend of recruit density recorded in the same plot in two consecutive years. Both these findings suggest that the plots more intensively recruiting in one year will also recruit more in following years. Thus, some plots should likely, present particular features that make them more suitable for *P. clavata* settlement over time. It can be speculated that micro-scale hydrodynamics, local substrate texture, and/or chemical attractors produced by the local species associations could foster more abundant recruitment in some plots.

The lack of any relation between the density of recruits and that of larger/older colonies suggests some independence of recruitment from the density of these colonies, as well as the lack of any effect of their high fecundity on local recruitment [21]. Although these findings may appear to be in contrast with the initial relation between recruits and small/young adults, the selectively higher mortality affecting the larger/older classes may, over time, lead to an alteration of such relation.

The highly dense canopy made mainly by older colonies in the pre-mortality, overcrowded population may have had some "shadow effect" to avoid larval settlement in their surroundings by limiting the substrate suitable for recruits. After the 1999 and 2003 mortalities, reduction of the canopy by about 2/3 could have driven a clear-cut increase in recruitment. As the increase in recruitment was delayed by four years since the mortality events and co-occurred with the detachment of dead colonies from the cliff in 2007 [20], the shadow effect caused by the dead 'scaffold' of larger colonies could have been maintained until they detached. Only after the dead colonies fell off the cliff and their shadow effect disappeared did recruitment start to rise, remaining at similar values in the following years, as well.

*Paramuricea clavata* has been considered a species with low population turnover and recovery rates [52,53]. Moreover, according to a demographic model aimed at projecting the trends of this population over time by means of simulations based on an ad hoc Leslie-Lewis transition matrix [46], wide fluctuations in density that stabilize only after several decades have been described [13]. However, in the present study, we observed, in the real population, faster density stabilization than predicted by said model.

After 14 and 10 years since the two mortality events, the population of *P. clavata* is still alive and has partially recovered, although with severely reduced density and canopy. Moreover, the population structure, dominated in recent years by recruits, is quite different from the pre-mortality one, dominated by larger/older colonies. As these features have been maintained for 7 years, the population seems to have reached a new equilibrium point.

All the findings presented herein indicate that the effects of drastic mortality events on a long-lived gorgonian population cannot be limited to the narrow definitions of "extinction or recovery", but that a stricken population may reach a new equilibrium. The main concern regards the stability of such a new equilibrium reached at remarkably lower densities: repeated mortality events and undamped oscillations may even devolve the local population to extinction.
