Water Quality Anomalies following the 2017 Hurricanes in Southwestern Puerto Rico: Absorption of Colored Detrital and Dissolved Material

: Absorption of colored dissolved organic matter or detrital gelbsto ﬀ (aCDOM / ADG) and light attenuation coe ﬃ cient (K d 490) parameters were studied at La Parguera Natural Reserve in southwestern Puerto Rico, before and following Hurricanes Irma (6–7 September) and Mar í a (20–21 September) in 2017. Water quality assessments involving Sentinel 3A ocean color products and ﬁeld sample data was performed. The estimated mean of ADG in surface waters was calculated at > 0.1 m − 1 with a median of 0.05 m − 1 and aCDOM443 ranged from 0.0023 to 0.1121 m − 1 in ﬁeld samples (n = 21) in 2017. Mean ADG443 values increased from July to August at 0.167 to 0.353 m − 1 in September–October over Turrumote reef (LP6) with a maximum value of 0.683 m − 1 . Values above 0.13 m − 1 persisted at o ﬀ shore waters o ﬀ Gu á nica Bay and over coral reef areas at La Parguera for over four months. The ADG443 product presented values above the median and the second standard deviation of 0.0428 m − 1 from September to October 2017 and from water sample measurement on 19 October 2017. Mean K d 490 values increased from 0.16 m − 1 before hurricanes to 0.28 right after Hurricane Irma. The value remained high, at 0.34 m − 1 , until October 2017, a month after Hurricane Mar í a. Analysis of the Sentinel (S3) OLCI products showed a signiﬁcant positive correlation (r s = 0.71, p = 0.0005) between K d 490_M07 and ADG_443, indicating the inﬂuence of ADG on light attenuation. These signiﬁcant short-term changes could have ecological impacts on benthic habitats highly dependent on light penetration, such as coral reefs, in southwestern Puerto Rico.


Introduction
Hurricane María was recorded as the third costliest hurricane in USA history [1]. It is considered the most damaging atmospheric event to have impacted the island in the past 90 years. Hydrological data availability during the study period was limited; nevertheless, estimates suggest that the 24 h-rainfall intensity exceeded 100-250 year values [2]. Severe flooding affected most of the island, and river discharges were at record levels. Hurricane Irma brought maximum inundation levels of 30.48 to 60.96 cm above ground level along Puerto Rico (PR) coastal areas with an estimated storm surge at Remote Sens. 2020, 12,3596 3 of 15

Study Site
The study area includes the region from Guánica Bay (GB) to La Parguera Natural Reserve (LPNR) in southwestern Puerto Rico. LPNR is located about 8 km west of GB and is known for its highly developed coral reefs and extensive seagrass habitats. The average annual water temperature in LPNR is 26.5-30 • C, and the salinity fluctuates from 31 to 36 PS [7].
Coral reefs habitats are shown within the contour lines representing the study area as a region of interest (ROI) (Figure 1). These are delineated using live coral cover classification [39], while the perpendicular lines represent the limits of La Parguera Natural Reserve. The ROI was considered for statistical data analysis. Seven out of the thirteen stations are presented here. Stations GUA4, GUA5, LP12, and LP13 were located offshore and along the insular platform. While sites LP6, LP8, and LP10 were closer to the coast (Figure 1).

Study Site
The study area includes the region from Guánica Bay (GB) to La Parguera Natural Reserve (LPNR) in southwestern Puerto Rico. LPNR is located about 8 km west of GB and is known for its highly developed coral reefs and extensive seagrass habitats. The average annual water temperature in LPNR is 26.5-30 °C, and the salinity fluctuates from 31 to 36 PS [7].
Coral reefs habitats are shown within the contour lines representing the study area as a region of interest (ROI) (Figure 1). These are delineated using live coral cover classification [39], while the perpendicular lines represent the limits of La Parguera Natural Reserve. The ROI was considered for statistical data analysis. Seven out of the thirteen stations are presented here. Stations GUA4, GUA5, LP12, and LP13 were located offshore and along the insular platform. While sites LP6, LP8, and LP10 were closer to the coast (Figure 1).

Satellite Data
The Sentinel 3A (S3A) Ocean and Land Colour Instrument (OLCI) is a push-broom imaging spectrometer with 21 spectral bands in the range of 400-1020 nm [40]. It was launched in February 2016, followed by S3B, launched in March 2017. Their products have a full-spatial resolution of 300 meters and include water-leaving reflectance in 16 bands, algal pigment concentrations [41] and neural network algorithms [42], total suspended matter concentration (TSM), diffuse attenuation coefficient (Kd490_M07) Morel method [41], and absorption of colored detrital and dissolved organic matter (ADG_443_NN) GSM method [43,44]. The temporal resolution for OLCI is daily with an optimum orbit above the study area every two days.

Satellite Data
The Sentinel 3A (S3A) Ocean and Land Colour Instrument (OLCI) is a push-broom imaging spectrometer with 21 spectral bands in the range of 400-1020 nm [40]. It was launched in February 2016, followed by S3B, launched in March 2017. Their products have a full-spatial resolution of 300 meters and include water-leaving reflectance in 16 bands, algal pigment concentrations [41] and neural network algorithms [42], total suspended matter concentration (TSM), diffuse attenuation coefficient (K d 490_M07) Morel method [41], and absorption of colored detrital and dissolved organic matter Remote Sens. 2020, 12, 3596 4 of 15 (ADG_443_NN) GSM method [43,44]. The temporal resolution for OLCI is daily with an optimum orbit above the study area every two days.
We obtained the data from the EUMETSAT Copernicus data system. S3A/OLCI data were extracted from the pixel coinciding with our field water monitoring stations and pixels over the ROI (Figure 1). The Sentinel Application Platform (SNAP) tools, developed by the European Space Agency (ESA) for satellite product analysis, were used for obtaining OLCI Level 2 data products. Only Sentinel 3-A data were used in the study.
Water quality products (ADG and K d 490) were extracted from Sentinel 3 OLCI imagery dating from July to December 2017. A subset of three images (out of 20) was evaluated considering the region of interest (ROI). It included the following dates: 22 July, 11 September, and 8 October 2017. This subset was selected to reduce uncertainty due to the following factors: negative reflectance values from bands one to six, sunglint effect, cloud, or land adjacency effect, or products fail/flags. Imagery is visualized with a median (7 × 7) pixel value. The complete set of images was divided into five time-frames, summarizing four images in each period. The time frames included one period previous to the hurricane events, one period immediately after the event and three additional periods after the event to identify the long-term effect on light attenuation. Only the LP6 site data were used in the time frame to avoid negative values and other sensor issues previously mentioned.

Water Quality Measurements-Field and Laboratory Analysis
Water samples and optical data were collected in 2017. The number of sampling stations varied based on the sea state, environmental parameters, and imagery availability. Sampling was conducted monthly at three to 13 stations in southwestern Puerto Rico ( Figure 1). The locations were selected based on depth, bottom type, and habitat in relation to coral reefs. Water samples were obtained from the first meter depth and analyzed in the laboratory for CDOM absorption (aCDOM).

aCDOM
Duplicate samples were collected at each station using gloves, avoiding any contamination with organic matter. They were stored in previously cleaned 250 mL amber glass bottles and transferred to 140 mL bottles after filtration. Sterile membrane filters (0.2 µm pore) were employed (Pall©). The filtration system was rinsed beforehand and between each filtration with a 50 mL portion of sample water and was then discarded [45]. Spectrophotometric analysis was carried out using a Shimadzu 1800-UV diode array instrument. Samples were analyzed in 10 cm path length quartz cells at 0.5 nm intervals over a wavelength range from 250 nm to 800 nm. Milli-Q water absorbance was subtracted from the sample data, and subsequently, the value at 700 nm was subtracted from the entire spectrum [46]. The absorbance values were converted to absorption coefficients, a (λ, m −1 ), and absorption coefficients at 443 nm (aCDOM 443 m −1 ) were reported as quantitative aCDOM. The absorption coefficients an (m −1 ) were calculated using the following equation: where A (L) is the absorbance at a wavelength, and l is the optical path length of the cell in meters.

Satlantic HyperPro
The Satlantic profiling spectroradiometer measures in-water downwelling plane irradiance (Ed) and upwelling radiance (Lu) with 256 spectral bands for a full spectral range of 305-1100 nm [47]. A surface Ed radiometer measures downwelling irradiance above the water surface and is used to normalize the in-water data for fluctuations in the incident light field from passing clouds. The instrument derives spectral water column attenuation coefficients, including the K d 490 following Aurin and Petzold (1981) in the manufacturer manual [48]: K( 490 ) = 0.0833 (Lu( 443 )/Lu ( 550 )) −1.491 + 0.022 (2)

Statistical Analysis
The SNAP© Sentinel toolbox, pixel extraction, and histogram tools were used to obtain satellite data statistics. S3A data were divided into 5-time frames (July-September, September-October, October-November, November-December, and December 2017) to evaluate the mean and median differences over time. A Spearman correlation was applied to ADG443_NN satellite data, aCDOM field data and K d 490 for field and satellite data to understand the influence of ADG/aCDOM on light attenuation. The analysis was employed using Origin Pro 2016© software.

Satellite Data (ADG443_NN) and In Situ Data (aCDOM443)
Results were based on in situ data for a year (2017) and the last six months (2017) data retrieved from satellite sensor S3A. Before the hurricane events, oligotrophic stations located at the shelf edge showed ADG443_NN values below 0.05 m −1 ; while values below 0.1 m −1 were associated with insular shelf sites. The value of the ADG443_NN was above the median of 0.0435 m −1 (prior to the events, over the ROI) for the entire sampling period. On September 11, four days after the first event, amounts above 0.1 m −1 were detected at GUA5, LP6, and LP8 for one month ( Figure 2). The highest ADG values at offshore waters were detected on Sep 11, after the first hurricane (H. Irma) which was considered less severe because its eye did not make landfall. On the other hand, the effects of H. María on the values were evident on Oct 8th satellite data in most of our study area. It had a similar or lower effect on ADG (GUA5, LP12, and LP13) values at outer shelf waters but the sensor detected higher values at inner shelf waters.

Statistical Analysis
The SNAP© Sentinel toolbox, pixel extraction, and histogram tools were used to obtain satellite data statistics. S3A data were divided into 5-time frames (July-September, September-October, October-November, November-December, and December 2017) to evaluate the mean and median differences over time. A Spearman correlation was applied to ADG443_NN satellite data, aCDOM field data and Kd490 for field and satellite data to understand the influence of ADG/aCDOM on light attenuation. The analysis was employed using Origin Pro 2016© software.

Satellite Data (ADG443_NN) and in situ Data (aCDOM443)
Results were based on in situ data for a year (2017) and the last six months (2017) data retrieved from satellite sensor S3A. Before the hurricane events, oligotrophic stations located at the shelf edge showed ADG443_NN values below 0.05 m −1 ; while values below 0.1 m −1 were associated with insular shelf sites. The value of the ADG443_NN was above the median of 0.0435 m −1 (prior to the events, over the ROI) for the entire sampling period. On September 11, four days after the first event, amounts above 0.1 m −1 were detected at GUA5, LP6, and LP8 for one month ( Figure 2). The highest ADG values at offshore waters were detected on Sep 11, after the first hurricane (H. Irma) which was considered less severe because its eye did not make landfall. On the other hand, the effects of H. María on the values were evident on Oct 8th satellite data in most of our study area. It had a similar or lower effect on ADG (GUA5, LP12, and LP13) values at outer shelf waters but the sensor detected higher values at inner shelf waters.  Figure 3B). Station LP6 is located to the southwest of Guánica Bay. The extreme values belong to the sample size and should not be treated as outliers even though a Grubbs' outlier test detect these as such. We can consider them as extreme values as a result of the events.  Figure 3B). Station LP6 is located to the southwest of Guánica Bay. The extreme values belong to the sample size and should not be treated as outliers even though a Grubbs' outlier test detect these as such. We can consider them as extreme values as a result of the events.
Remote Sens. 2020, 12, x FOR PEER REVIEW 6 of 15  S3A data were divided into five-time frames (July-September, September-October, October-November, November-December, and December 2017) to evaluate the mean and median differences ( Table 1). The mean for ADG_443_NN was doubled in the second period from 0.1675 (pre hurricanes) to 0.3536 m −1 (September-October). The maximum value of 0.6834 m −1 was detected in the same period. The values extracted from S3A started in July with values above the maximum of field data for 2017. Values above 0.13 m −1 persisted until December, four months after the events. Moreover, the median showed the same tendency (> 0.1 m −1 ) over four months. River discharges and coastal drainage persist several weeks after the events. No major events took place after September, which may indicate we are seeing the long term effect of the hurricanes in coastal water biogeochemistry.
Satellite imagery show the absorption of dissolved organic matter over time. Figure 4 shows the S3A ADG443_NN product prior to (July 22), and following (September 11 and Oct. 8) the passage of hurricanes Irma and María over Puerto Rico. Contour lines represent coral reefs as the region of interest (ROI). The ROI was considered for graphs and statistics on Figure 4 and Table 2. The high values of ADG443_NN in Figure 4 correspond to pixels that cover mainly shallow areas and emergent reefs. However, the analysis only considered the extracted values in submerged areas. The histogram S3A data were divided into five-time frames (July-September, September-October, October-November, November-December, and December 2017) to evaluate the mean and median differences ( Table 1). The mean for ADG_443_NN was doubled in the second period from 0.1675 (pre hurricanes) to 0.3536 m −1 (September-October). The maximum value of 0.6834 m −1 was detected in the same period. The values extracted from S3A started in July with values above the maximum of field data for 2017. Values above 0.13 m −1 persisted until December, four months after the events. Moreover, the median showed the same tendency (>0.1 m −1 ) over four months. River discharges and coastal drainage persist several weeks after the events. No major events took place after September, which may indicate we are seeing the long term effect of the hurricanes in coastal water biogeochemistry. Satellite imagery show the absorption of dissolved organic matter over time. Figure 4 shows the S3A ADG443_NN product prior to (July 22), and following (September 11 and Oct. 8) the passage of hurricanes Irma and María over Puerto Rico. Contour lines represent coral reefs as the region of interest (ROI). The ROI was considered for graphs and statistics on Figure 4 and Table 2. The high values of ADG443_NN in Figure 4 correspond to pixels that cover mainly shallow areas and emergent reefs. However, the analysis only considered the extracted values in submerged areas. The histogram for July 22 shows around 46 pixels lower than 0.05 m −1 and more than 95 % of pixels with values < 0.1 m −1 . The maximum value was 1.0 m −1 ( Table 2). After the first hurricane event (Irma), an increase in the ADG443_NN values from Guánica Bay to La Parguera was observed (Figure 4) as expected after an event of such magnitude. Approximately, 12% of pixels in the selected area were considered in Table 2. The histogram shows an increment of pixels with values in the range of 0.1 to 0.5 m −1 (Figure 4) and shows pixels with up to 4.5 m −1 . Table 2 shows the increase in the median ADG443_NN value over time from 0.04 to 0.08 m −1 .
Remote Sens. 2020, 12, x FOR PEER REVIEW 7 of 15 the ADG443_NN values from Guánica Bay to La Parguera was observed ( Figure 4) as expected after an event of such magnitude. Approximately, 12% of pixels in the selected area were considered in Table 2. The histogram shows an increment of pixels with values in the range of 0.1 to 0.5 m −1 ( Figure  4) and shows pixels with up to 4.5 m −1 . Table 2 shows the increase in the median ADG443_NN value over time from 0.04 to 0.08 m −1 .    To visualize the effect on light attenuation, we chose sampling station (LP6), located between Guánica Bay and LPNR. It is near the coastline but, far enough to be outside the influence of land pixels. Taking a look on satellite data of this site, a spike value was observed on 7 October 2017, for both parameters ADG443_NN and K d 490, with high values on 18 August, 23 October, and 16 December 2017 ( Figure 5). These values were concurrent with two heavy rain periods during the last six months of the year 2017. The image from October 8 showed the impact on water quality parameters three weeks after the events. All ADG443_NN values were over 0.04 m −1 for the entire sampling period. To visualize the effect on light attenuation, we chose sampling station (LP6), located between Guánica Bay and LPNR. It is near the coastline but, far enough to be outside the influence of land pixels. Taking a look on satellite data of this site, a spike value was observed on October 7, 2017, for both parameters ADG443_NN and Kd490, with high values on August 18, October 23, and December 16, 2017 ( Figure 5). These values were concurrent with two heavy rain periods during the last six months of the year 2017. The image from October 8 showed the impact on water quality parameters three weeks after the events. All ADG443_NN values were over 0.04 m −1 for the entire sampling period.

Kd490 and Correlation with ADG443/aCDOM443
Values for diffuse attenuation coefficient (Kd490) share the ADG443_NN tendencies (  (Table 3). The highest mean value of 0.34 m −1 was observed in the period of Sep 11 to Oct 8; during that period, a maximum of 0.48 m −1 was detected. The maximum value of Kd490 derived from field data in 2017 was 0.33 m −1 on October 19, 2017, four weeks after the last hurricane ( Figure 8A,B).
The attenuation coefficient showed a slight variation in outer shelf waters with a greater impact in inner shelf, specifically in LP6, alias Turrumote II (Figure 7). The cumulative effect of biogeochemical processes in production and degradation of organic matter is shown by this increment on October values. Kd490 values reach to the normal between October and November (Table 3).

K d 490 and Correlation with ADG443/aCDOM443
Values for diffuse attenuation coefficient (K d 490) share the ADG443_NN tendencies (  (Table 3). The highest mean value of 0.34 m −1 was observed in the period of Sep 11 to Oct 8; during that period, a maximum of 0.48 m −1 was detected. The maximum value of K d 490 derived from field data in 2017 was 0.33 m −1 on 19 October 2017, four weeks after the last hurricane ( Figure 8A,B).
The attenuation coefficient showed a slight variation in outer shelf waters with a greater impact in inner shelf, specifically in LP6, alias Turrumote II (Figure 7). The cumulative effect of biogeochemical processes in production and degradation of organic matter is shown by this increment on October values. K d 490 values reach to the normal between October and November (Table 3). after the last hurricane ( Figure 8A,B).
The attenuation coefficient showed a slight variation in outer shelf waters with a greater impact in inner shelf, specifically in LP6, alias Turrumote II (Figure 7). The cumulative effect of biogeochemical processes in production and degradation of organic matter is shown by this increment on October values. Kd490 values reach to the normal between October and November (Table 3).   The field data for 2017 (N = 21) show a high correlation between Kd490 and aCDOM443 absorption coefficients (rs = 0.79, p = 0.0003) and a lower but similar correlation between S3 OLCI products, (rs = 0.71, p = 0.0005). It cannot be interpreted as a sensor validation.

Discussion
Caribbean Sea water is mostly oligotrophic with a high light penetration in the water column, although it is seasonally influenced by the Orinoco and Amazon rivers from South America, seasonally [49][50][51]. Light penetration changes after hurricane events can affect seagrasses [11] and  The field data for 2017 (N = 21) show a high correlation between K d 490 and aCDOM443 absorption coefficients (r s = 0.79, p = 0.0003) and a lower but similar correlation between S3 OLCI products, (r s = 0.71, p = 0.0005). It cannot be interpreted as a sensor validation.

Discussion
Caribbean Sea water is mostly oligotrophic with a high light penetration in the water column, although it is seasonally influenced by the Orinoco and Amazon rivers from South America, seasonally [49][50][51]. Light penetration changes after hurricane events can affect seagrasses [11] and other light-dependent organisms like corals. Previous research shows that coral photo-physiology is altered by light availability [25,52]. García-Sais and collaborators (2017) studied K d 490 and Chl-a trends over individual coral reefs in Puerto Rico using L2 and L3 imagery from SeaWiFS and MODIS Aqua satellite data [10]. A recent publication based on VIIRS data described the tendencies of K d 490 and Chl-a parameters on water quality around PR using a value of 0.1 m −1 for K d 490 and 0.45 µg/L for Chl-a as a threshold value for coastal waters [20]. Despite the detrimental effects documented by several authors [26,[53][54][55], an intermittent high turbidity over coral reefs can be photo-protective [10]. García-Sais and collaborators (2017) observed a negative correlation between K d 490 and the percent of coral cover which can be interpreted as a positive light shadow effect during sea surface temperature anomalies [10]. The severity of the damages can be highly influenced by the prevalence of adverse conditions during and after the events.
In 2017, the duration of the abnormal values (above 0.05 m −1 ) lasted four months as can be seen in Figure 5. Higher values are not necessarily coincident with the events but, rather these were detected two to three weeks later in October. The high number of landslides (>40,000) combined with runoff after hurricanes Irma and María over the Island were unprecedented [56] and washed sediments reached nearshore waters [19]. Miller (2019) documented elevated turbidity values nearshore until February 2018 related to inland hydrological disturbances caused by the hurricanes [19]. Gilbes and collaborators (2001) documented changes in Chl-a due to hurricane Georges up to two and a half weeks after the event [18]. Recently, Hernández and collaborators (2020) documented high K d 490 and Chl-a values from July to December 2017 all around Puerto Rico using VIIRS data; reporting Chl-a values above 0.45 µg/L in August and November 2017 [20]. These authors reported anomalous attenuation coefficient values for July 2017 (0.06 m −1 ) being persistently high until December. Chlorophyll-a is a parameter highly correlated with K d 490 on ocean color data [18,20]. It is important to highlight the oligotrophic water conditions on this study area, being influenced by Guánica Bay dynamics. The values considered in this area can be compared to coral reefs or benthic areas with low influence of rivers.
The aCDOM443 values above 0.05 m −1 are not typical for coral reef waters in Puerto Rico. CDOM values with means < 0.043 m −1 are the most common values over coral reefs and seagrass beds in the studied area. Otherwise, the values closer to 0.02 m −1 are found in offshore waters. An absorption coefficient higher than 0.1 m −1 is frequently found on coastal embayments like Guánica Bay or the Bioluminescent Bay surrounded by mangroves [57]. Anomalies like the ones measured in this study lasted for the entire study period.
In terms of attenuation coefficient (K d 490), values above 0.2 m −1 corresponded to coastal embayments while values from 0.1 to 0.2 were observed at shallow coral reef or seagrasses areas close to the coast (<1 mile) or closer to the coral cays [20]. The lower values (<0.1 m −1 ) were found at midand outer-shelf coral reef stations. A mean K d 490 value of 0.056 m −1 was documented (for 10 y data) in a coral reef site at Guánica by satellite data [10]. Their values are lower than the values reported here from July to December 2017 using OLCI data. A variant of K d 490 parameter, K d PAR, was measured in situ before and after hurricane events in St. John Island recording the lowest level of light in coral reef in the Caribbean after a hurricane event [58,59]. Certainly, these events had an unprecedent effect on light attenuation over sensitive benthic communities.
These data show the influence of ADG443_NN/aCDOM in light attenuation. However, in estuarine areas, a significant correlation between K d 490 and Chl-a was documented after a hurricane event [13,20]. The high anomalous values of ADG and K d 490 can be related to the unprecedented runoff produced by defoliation and landslides [19,56] followed by biogeochemical oceanographic processes over the coastal waters.

Conclusions
As expected from episodic events of this magnitude, significant water quality parameter changes occurred in southwestern Puerto Rico. Sentinel 3A OLCI data was used to extract information on ADG and K d 490 values. These data were compared with in situ data trends and correlated between them. The amount of data acquired during the study period (before N = 5, N = 16 after hurricanes) duplicates the quantity of data obtained from the field (N = 3 after the hurricane) in 2017. Although cloud cover in tropical islands can be high, remote sensing is an accessible and useful tool for short and long-term water quality studies.
Increasing values of satellite-derived water quality parameters were detected with S3 OLCI and field data in southwestern Puerto Rico. The anomalies were observed during the 9/11-10/8 period as expected and extended until December. The ADG values increased throughout all the coral reef zones. The estimated ADG mean in this zone was > 0.1 m −1 with a median of 0.05 m −1 . The mean values of K d 490 increased from 0.16 m −1 before the hurricanes to 0.28 m −1 shortly after Hurricane Irma, and 0.34 m −1 in October 2017, a month after Hurricane María.
Satellite data are useful for water quality assessment in PR coastal waters with a judicious understanding of their uncertainties and limitations. On the other hand, we cannot conclude the performance of the sensor measurement on ADG443_NN or K d 490 products since we do not have enough in situ data from July to December 2017. Our results represent a pioneering effort in the establishment of tendencies for water quality studies in Puerto Rico. Usually, government agencies' data are a single snapshot influencing the mean values that can be misinterpreted for the establishment of patterns on water quality. These gaps can be addressed with satellite data, as we showed throughout the manuscript. Remote sensing tools can help understand coastal and benthic habitat changes and biogeochemical processes in waters surrounding oceanic islands, especially after extreme weather events. Previous studies have mainly documented the importance of chlorophyll on light attenuation, but this study highlights the importance of detrital and gelbstoff matter on light attenuation coefficient. This is not only done as a historical perspective of the consequences of these events but as an analysis that could be integrated into future efforts aimed at describing the consequences of such events in benthic communities changes on the long run.