Spatial and Temporal Variability of Posidonia oceanica Monitoring Indicators, Valencian Community, Spain

: This article presents the results from the long-term Water Framework Directive monitoring of Posidonia oceanica in the Valencian Community, Spain. A total of six areas where Posidonia occurs were studied. Locations were characterized by a di ﬀ erent ecological condition: degraded Posidonia meadows; meadows in good conditions; meadows close to the best reference conditions. The spatiotemporal variability of di ﬀ erent indicators were sampled from 2005 to 2017. Lower quality meadows are characterized by a lower density and coverage of Posidonia , a higher proportion of dead Posidonia , a higher proportion of plagiotropic rhizomes, smaller shoots, and higher epiphyte biomass, as well as the presence of foliar necrosis marks. For all study sites, the temporal trends indicated a decrease in the quality of the meadows from 2005 to 2011 and a recovery from 2012 to 2017. Some variability observed in the indicators such as P. oceanica cover in patchy meadows, could not be explained by changes in the meadow health.


Introduction
Seagrasses constitute fundamental ecosystems lying on soft coastal bottoms worldwide, by providing for key ecological functions (e.g., habitat, food, nutrient cycling) and ecosystem services that contribute to human welfare (e.g., global carbon sequestration, shoreline protection, enhanced fisheries [1,2]).
In the Mediterranean Sea, Posidonia oceanica is an endemic seagrass that forms dense meadows in coastal areas. These meadows represent one of the most productive and valuable key ecosystems in the Mediterranean but they are also very sensitive to changes in the quality of the environment [3,4].
The Water Framework Directive (WFD) has instituted an ambitious strategy with the aim of reaching a good ecological status for all surface water bodies within the European Union (EC, 2000). Monitoring in the WFD aims at detecting how eventual changes in the quality of the water are reflected in ecosystem status. To this aim, the WFD establishes the monitoring and protection of surface water bodies based on biological, hydromorphological, and physico-chemical quality elements. Biological quality elements (BQEs) can include different bioindicators, including seagrasses. For instance, in the Mediterranean Sea, Posidonia oceanica has been used as the main seagrass indicator [3,[16][17][18]. Some indicators have been more often used in research and monitoring programs such as the shoot density, the P. oceanica meadow and dead-matte cover, the meadow limits, epiphytic coverage, leaf biometry, shoot balance, and total non-structural carbohydrates [16][17][18][19].
In this work, we present the results of a long-term WFD monitoring of P. oceanica in the Valencian Community of Spain. The objectives of this paper were to study temporal and spatial fluctuation of the P. oceanica indicators used in this monitoring; investigating possible sources of this variation. We also evaluated the ecological status of these meadows over time. A relevant aspect for seagrass management is the rate of the seagrass decline. While some authors warn that a global process of P. oceanica decline is currently ongoing [5,6], others argue that the perceived decline is due to an accumulation of local impacts [13]. We also discuss the relevance of local versus global drivers of seagrass meadow quality.

Study Area and Sampling Design
This study was conducted along the coast of the Valencian Community (Spain), in the western Mediterranean, during September-October from 2005 to 2017, in the context of the EU Water Framework Directive. Sampling was done by scuba divers in 6 locations ( Figure 1) encompassing a depth range of 14-17 m, as meadows at these depths are usually not affected by natural alterations, e.g., those caused by waves or storms [20]. The locations were selected based on existing knowledge of the ecological status of their respective Posidonia meadows [17]. We selected two degraded localities (WFD status poor-moderate), namely Benicassim and Alicante; two localities where meadows present good conditions (WFD status good), namely Altea and Oropesa; two localities with meadows close to the best reference conditions (WFD status high) in the municipalities of Calpe and Santa Pola. At each locality, three sampling sites each at a distance of hundreds of meters from each other were randomly selected to prevent spatial pseudo-replication [21]. Logistical constraints prevented us to carry out sampling in Oropesa in 2005, in all the localities during 2011 and 2012, and to obtain plagiotropic rhizome proportion and necrosis data in 2005.

Posidonia Indicators
We studied Posidonia indicators of ecological condition related with the meadow community (such as the herbivore pressure and existing epiphyte biomass), the population (dead-matte cover, meadow cover, shoot density and proportion of plagiotropic rhizomes), and characteristics of individual plants (rhizome baring/burial, shoot foliar surface, and foliar necrosis).

Structural Indicators
At each sampling site, three 40 × 40 cm quadrats were randomly set to measure the shoot density, percentage of plagiotropic rhizomes, and rhizome baring. The living and dead Posidonia cover were estimated as the proportion of living and dead patches, respectively, in three replicate 20 m transects (for a more detailed methodology see [17]). In addition, the conservation index (CI) was derived from the ratio of the Posidonia meadow cover divided by the total meadow cover (calculated as the Posidonia cover plus the dead-matte cover) [22].

Shoot Morphology and Epiphytic Biomass
Five P. oceanica shoots were randomly collected by hand at each of the tree sampling sites per locality and sampling time and brought to the laboratory in a cooler for further analysis. In the laboratory, we quantified the number of leaves per shoot, as well as the length and width of all leaves. The foliar shoot surface was calculated as the sum of all individual leaf areas per shoot. Necrosis and herbivore occurrence were also estimated by calculating the percentage of leaves inside a shoot that presented these kind of marks. Macroscopic epiphytes were removed using a razor blade, and the epiphytes and leaves were subsequently oven-dried to obtain the epiphyte biomass (i.e., dry weight, DW, of epiphytes per cm 2 of leaf) and shoot biomass.

Data Analysis
An analysis of variance (ANOVA; [23]) was performed in order to estimate whether the mean values of the various indicators varied significantly among locations, sites and sampling times. Previously, data normality was verified using the Kolmogorov-Smirnov test of normality and the homogeneity of variance was checked with a Cochran test [23]. If necessary, data were sqrt or log transformed. Minimum significance level was established at p < 0.05, but if the ANOVA assumptions were not accomplished, alpha was changed to 0.01. When analysis of variance identified a significant difference, the post-hoc Tukey test was applied to identify specific treatment differences. All statistical analyses were implemented in the open source software R3.6.1 (www.r-project.org).
We also used GAMs (Generalised Additive Models) models to observe fluctuations of Posidonia indicators through time in different locations, in order to estimate trends during years where data was lacking. In our case, GAMs allowed the treatment of our data consisting of temporal series irregularly spaced in time (there is a lack of sampling during 2012-2013), and that do not follow a linear trend [24,25]. Models were fitted using the mgcv package [26] for R [27].

Spatial Variation of Posidonia Indicators
All structural indicators showed significant differences of means in all of the studied spatial scales (locality and site; Table 1). Average shoot density ranged between 88 and 310 shoots/m 2 in the six studied localities (Figure 2a). Significant differences were detected among localities, with higher values coinciding with high ecological status quality (Santa Pola and Calpe). Significant differences were also detected between sampling sites (Table 1). Some outlier values were observed in Alicante, Altea, and Calpe. Table 1. Summary of the results from the ANOVA applied to different Posidonia indicators for the six locations (* p-value < 0.05; ** p-value < 0.01; *** p-value < 0.001; when p-valor is not significant, its real value is shown).
Dead-matte cover significantly differed among localities following the opposite pattern (Figure 2d), while conservation index (CI) showed a similar trend (Figure 2f).
Shoot morphological indicators, except herbivore pressure, showed significant differences for at least one of the spatial scales (locality and site; Table 1). Shoot foliar surface (Figure 3a), foliar biomass (Figure 3h) and maximum leaf length (Figure 3d) were significantly higher for high status localities (Santa !Pola and Calpe) and one of the good status localities (Altea)-with average values of 149-156 cm 2 /shoot, 0.69-0.73 g DW/shoot, and 52.0-57.5 cm, respectively-and lowest in localities with a poor status (Benicassim and Alicante). Significant differences in this indicator were also detected at site scale (Table 1). Shoots from Altea meadow presented a significantly higher average number of leaves (5.9 leaves/shoot) and a maximum leaf width (1.05 cm) than the rest of the meadows (Figure 3b). The lowest values for these indicators were found for localities with a poor status (4.7-4.8 leaves/shoot and 0.91-0.95 cm, respectively). Foliar necrosis (Figure 3c) and leaf epiphyte biomass (Figure 3e) were significantly higher in shoots from the Alicante meadow (23% and 1.84 mg/cm 2 ), decreasing with the ecological status of the meadow, with lowest values found in the Santa Pola locality (3% and 0.68 mg/cm 2 ). The percentage of herbivore marks on shoots did not showed significant differences among studied localities (Figure 3g).

Temporal Trends of Posidonia Indicators
Significant differences of shoot density were detected with time (Table 1), although the temporal trend of this indicator was similar for all the studied localities (Figure 4a). A slight reduction from 2005 to 2011 was observed, followed by a reversed trend towards the initial density values in 2017. Localities with lower quality status showed lower values throughout the years. Posidonia and dead matte cover also showed significant differences with time (Table 1), but temporal trends were different between localities (Figure 4b). High quality status localities showed stable values of Posidonia cover (75-98%), while in localities with a good quality status this percentage fluctuated (57-98%). Dead matte cover values were steady at high and good quality meadows, showing a small increase in Altea in 2008 (Figure 4d). Posidonia and dead matte cover values also fluctuated (3-40 and 60-97%, respectively) in one of the poor quality status locality (Alicante). The temporal pattern of the Conservation Index was similar to the trends observed for P. oceanica cover (Figure 4f). The percentage of plagiotropic rhizomes significantly varied with time (Table 1), but not in the same way for all the localities (Figure 4c). Its average values slightly fluctuated in high and good quality status meadows (0-24%) when compared with the poor ones (9-68%). In one of the poor conservation status meadows (Alicante) an important increment was observed for this parameter from 2005 to 2012, followed by a reduction to initial values in 2017. Significant differences of rhizome baring were also detected with time (Table 1). Meadows with a poor status of conservation showed stable and lower values for this variable (0.7-3.5 cm), while the rest of the meadows showed higher and fluctuating values (3.5-12.4 cm; Figure 4e). Regarding the morphological variables, all of them presented significant differences in their means throughout the studied temporal scale (Table 1). Temporal trends for average shoot foliar surface, shoot biomass, and maximum leaf length were similar (Figure 5a,h). In most of the high and good status meadows, these indicators showed higher average values during the first years of study, a decline during 2011-2014, and a subsequent recovery. In poor status meadows, these descriptors remained steady (Alicante) or showed a slight increment (Benicassim). The number of leaves per shoot showed a similar trend for all the studied localities (Figure 5b)

Discussion
In this study, we analyzed the variation of different indicators of a seagrass (Posidonia oceanica) associated to spatial and temporal scales. Spatial variability is mainly associated with meadow quality. High quality meadows are characterized by a high density, covering, and Conservation Index, low dead matte and plagiotropic rhizomes proportion, bigger shoots, and low foliar necrosis marks. However, some variability observed in the studied descriptors could not be explained by changes related to the meadow health. This is the case of the number of leaves per shoot and maximum leaf width, which do not correlate with the meadow status. This is also the case for the temporal variability of meadow cover in the northernmost stations (Oropesa, Benicasim). Therefore, this variability has to be explained by the high heterogeneity of these meadows with a patchiness of rocky bottoms and Posidonia. In fact, this type of fluctuation at small or medium spatial scales may not always be explained by anthropic influence [27]. In this case, the utilization of a sampling design with an adequate spatial replication is necessary, together with the use of less variable indicators as it is the Conservation Index [28].
We also detected spatial variability at a small scale (sites) for all of the structural indicators and for some of the morphological ones (leaf epiphyte biomass, shoot biomass, maximum leaf width, and length). This is in agreement with other studies, which have observed a high variability at small or medium spatial scales for Posidonia meadow cover and shoot density [29,30]. In order to detect and to be able to analyze this local variability, we recommend using a nested sampling design with spatial factors such as sites or zones.
A similar pattern of temporal evolution was observed in all the studied localities for some indicators (such as shoot density, number of leaves per shoot, foliar necrosis, and herbivore pressure), regardless of their environmental quality. It has been noted that indicators related with low meadow quality (such as proportion of plagiotropic rhizomes, foliar necrosis, or epiphyte biomass) increase for most of the stations until 2011-2014 and decrease until 2017. This similar pattern may indicate that not only local factors [13] are drivers of meadow quality, but importantly, some global factor may also be acting [6]. One of these drivers may be temperature, as some extreme warming events have been detected in the Mediterranean Sea from 2003-2008, causing mortality events and significant changes in Posidonia oceanica meadows during these and subsequent years [31]. Recovery of Posidonia morphological indicators can be fast, and related with the lack of effects of this kind of events, but also with the management measures taken, such as the Water Framework Directive, which aims to achieve an improvement of water quality. Recovery detected in Posidonia density may be explained also in this way, as its decay was not very strong in most of the localities. Moreover, no significant change in general meadows quality was observed between 2005 and 2017 which may confirm the reversion of seagrass decline that has been described by de los Santos et al. [15].
From this study, we may also infer the importance of selecting an adequate quantity and type of indicator to determine the ecological status of a meadow. Some descriptors can be easy to estimate, such as the meadow cover. However, as we have seen, they may also need to incorporate other information related with bottom structure or heterogeneity. Monitoring indicators must be specifically related with the driver or stressor we want to evaluate [32]. If general indicators are used to indicate a change, the cause of such change cannot be determined. For this reason, using specific indicators is highly recommendable [32]. Another option is to incorporate a suite of different indicators in order to provide strong evidence of an environmental change [3,17,18]. We do not recommend using few descriptors because it may lead to erroneous assumptions, while using many of indicators may be time and economically costly. A compromise between both situations should therefore be aimed for. For example, the present study selected metrics that implied a rapid processing, which reduces economic and time costs. A more complex analysis was discarded since it could be expensive and time-consuming and, sometimes, it may lead to analytical errors [33].
Moreover, we want to highlight the importance of this kind of study, in which a long time series of data is analyzed in order to estimate temporal trends. This is particularly necessary when studying slow-growing species, such as Posidonia oceanica, which are highly sensitive to perturbations but with a very slow recovery rate [34].

Conclusions
In this study, Posidonia oceanica meadows with a low conservation status were characterized by reduced density and covering values, elevated dead-matte covering, high proportion of plagiotropic rhizomes and small shoots with elevated epiphyte biomass and foliar necrosis marks. Spatial and temporal variability has been observed which may be related with natural variability but also with global factors such as temperature or water quality. No significant change in general meadows quality was observed between 2005 and 2017, which may indicate that there is no general decline of Posidonia meadows currently ongoing in the Mediterranean. Long monitoring series with an adequate spatial replication are necessary to detect trends on meadow ecological status.