Estuarine, Coastal and Shelf Science 237 (2020) 106660

Contents lists available at ScienceDirect

Estuarine, Coastal and Shelf Science

journal homepage: http://www.elsevier.com/locate/ecss

The role of multiple stressors in a dwarf red () dieback

R.E. Rossi a,1,*, S.K. Archer a,b, C. Giri c, C.A. Layman a a Department of Applied Ecology, North Carolina State University, Raleigh, NC, 27695, USA b Louisiana Universities Marine Consortium, Chauvin, LA, 70344, USA c US Environmental Protection Agency, RTP, NC, 27713, USA

ARTICLE INFO ABSTRACT

Keywords: are habitat-forming foundation species that provide the framework for entire coastal communities. Drought Unfortunately, mangrove forests are declining globally, often due to multiple stressors. We report a recent Disease mangrove dieback on Abaco Island, The Bahamas, that appears to be the result of multiple stressors. First, we Herbivory investigated the role of hurricane and drought events in relation to green vegetation in the dieback area, as Hurricane measured using Normalized Difference Vegetation Index (NDVI). We used historical Landsat imagery of the Multiple stressors dieback region from 1989 to 2013 to identify changes in green vegetation and to determine when the dieback likely began. We determined that the dieback began around 2008, but found little evidence of a relationship between NDVI and hurricane or drought events, particularly in relation to the estimated start of the dieback. Preliminary observations in the dieback area suggested grazing by leaf-chewing organisms and disease were both present at high levels. We surveyed individual mangrove trees to determine the intensity of leaf-chewing her­ bivory and foliar disease on mangroves in the dieback area. We found leaf-chewing herbivory ranged from 0 to 79%, with a mean of 29% of leaves chewed per tree and disease incidence ranged from 0% to 97% with a mean of 47%. Finally, we conducted a simulated grazing experiment to test if experimental opening mimicking grazing could facilitate disease infection in mangroves . Experimental opening of plant tissues showed that grazing can increase disease. Our results suggest that drought and hurricanes did not initiate this dieback, but that herbivory likely facilitated the spread of disease thereby contributing to the dieback.

1. Introduction Friess and Webb, 2014). Understanding how multiple stressors impact coastal ecosystems is Coastal ecosystems are often composed of habitat-forming founda­ imperative for on-the-ground ecosystem-based management. Disease tion species that provide the framework for entire coastal communities presence and severity is correlated with other stressors, such as nutrient (Dayton, 1972; Bruno et al., 2003; Ellison et al., 2005). These ecosys­ enrichment and thermal stress on coral reefs (Bruno et al. 2003, 2007; tems provide a range of services, such as carbon sequestration and Ban et al., 2014). The combination of biotic stressors (a snail grazer and nursery habitat for many organisms including commercially important fungal pathogen) and drought interacted to drive salt marsh dieback in species (Costanza et al., 1997; UNEP, 2006; Barbier et al., 2011). Un­ the Southeastern U.S (Silliman and Bertness, 2002; Silliman and Newell, fortunately, coastal ecosystems are in decline: coral reefs are bleaching, 2003; Silliman et al., 2005). In Northeastern U.S salt marshes, eutro­ salt marshes are eroding and collapsing, and mangrove forests are dying phication and herbivore grazing has led to marsh diebacks (Bertness off (Hughes et al., 2003; Halpern et al., 2008; Deegan et al., 2012; et al., 2008). By understanding the links between, and consequences of, Lovelock et al., 2017). These ecosystems regularly experience multiple multiple stressors, managers can proactively mitigate against or actively stressors that have driven extensive loss. Additionally, human activities, manage stressors. such as aquaculture, development, and overexploitation can exacerbate Both abiotic and biotic stressors have been linked with mangrove these stressors (Alongi, 2002; Halpern et al., 2008; Barbier et al., 2011; diebacks. Relative sea-level rise, drought, and extreme salinities are

* Corresponding author. Department of Applied Ecology, North Carolina State University, Raleigh, NC, 27695, USA. E-mail address: [email protected] (R.E. Rossi). 1 Present Address: Oak Ridge Institute for Science and Education, U.S. Environmental Protection Agency, Gulf Breeze, FL, USA. https://doi.org/10.1016/j.ecss.2020.106660 Received 1 April 2019; Received in revised form 19 August 2019; Accepted 21 February 2020 Available online 25 February 2020 0272-7714/© 2020 Elsevier Ltd. All rights reserved. R.E. Rossi et al. Estuarine, Coastal and Shelf Science 237 (2020) 106660 physical factors linked to mangrove death around the world (Gilman between 44 and 67% of R. mangle canopies in Belize (Feller and Mathis, et al., 2008; Duke et al., 2017; Lovelock et al., 2017). In Australia, two 1997; Feller, 2002). Mangrove canopies are also extensively grazed by recent mangrove diebacks were associated with drought (Duke et al., the Mangrove Crab, Aratus pisonii (Erickson et al., 2003). Root-boring 2017; Lovelock et al., 2017). Long-term data from Mangrove Bay and stem-boring beetles can cause mortality of mangrove seedlings demonstrated that low sea levels resulted in increases in soil saliniza­ (Sousa et al., 2003; Alongi, 2009) and crab species, such as the Spotted tion, which contributed to canopy loss in Rhizophora stylosa, Avicennia Mangrove Crab (Goniopsis cruentata) and Mangrove Land Crab (Ucides marina, and Ceriops australis (Lovelock et al., 2017). Storm events also cordatus), consume mangrove propagules (Lee, 1998). Disease has been contribute to mangrove loss through canopy loss, sulfide soil toxicity, reported in mangroves in many places (Osorio et al., 2016). For and changes in soil elevation via several mechanisms (e.g., erosion, example, a fungus, Cytopsora rhizophorae, caused stem dieback and deposition, collapse) (Cahoon et al., 2003; Gilman et al., 2008; Asbridge resulted in mangrove loss in Puerto Rico, and an oomycete, Phytophthora et al., 2018; Servino et al., 2018). Often, the recovery of mangroves from sp., infected roots of mangroves in Australia resulting in death (Pegg storm events depends on species’ ability to resprout from remaining et al., 1980; Wier et al., 2000). tissue, which may be impacted by the presence of other abiotic or biotic Ultimately, disentangling abiotic and biotic stressors in mangrove stressors (Ward et al., 2016; Asbridge et al., 2018). In other cases, biotic ecosystems is necessary to identify the suite of stressors causing factors have been found to be equally or more important, including mangrove loss. Here we investigate the roles of two abiotic stressors and herbivory and disease. He and Silliman (2016) summarize examples of two biotic stressors and their potential association with a dieback of the effect of herbivores on survival and reproduction of mangroves. For dwarf red mangroves (R. mangle) in The Bahamas. We used hurricane example, wood-boring, stem-girdling, and leaf-feeding destroyed and drought data to identify when each of these events may have

Fig. 1. Map of dieback (in the Marls) and experiment sites on Abaco Island, The Bahamas, with an inset of the dieback area. The herbivory and disease survey was conducted at the Marls site and the simulated grazing experiment was conducted at all other sites: CA, HC, SC, and TB.

2 R.E. Rossi et al. Estuarine, Coastal and Shelf Science 237 (2020) 106660

impacted the dieback area. Using Landsat imagery and Normalized mangrove vegetation since it is often used an indicator of mangrove Vegetation Index (NDVI), we measured green mangrove vegetation in health (Chellamani et al., 2014; Giri et al., 2015; Flores-Cardenas� et al., the dieback area from 1989 to 2013 to determine if these stressors 2018; Wang et al., 2019). Landsat 5 and 7 annual NDVI composites from affected this region prior to the reported dieback. We hypothesized that 1989 to 2013 were acquired from Google Earth Engine (https://eart hurricane occurrence and drought severity would correlate with hengine.google.com/). Images were previously transformed to NDVI reduced NDVI (or greenness) and would signal the onset of the reported by the U.S Geological Survey (USGS) following Chander et al., 2009. mangrove dieback. Our preliminary surveys in the area suggested her­ Years with severe cloud cover were removed from the dataset resulting bivory and disease were present at high densities on remaining live in the exclusion of 1990 and 2003. We used coordinates from handheld mangroves leaves. As a result, we investigated the role of herbivory, GPS units (Garmin etrex 20) to create an outline of the dieback area as a disease, and their interaction using surveys and a manipulative field shapefile using ArcMap 10.3.1. This shapefile was checked against a experiment. We hypothesized that herbivory and disease would be basemap of the Caribbean produced by the USGS and high-resolution greater in the dieback area and that biotic stressors contribute to imagery from WorldView-2 (http://www.satimagingcorp.com/s dieback. atellite-sensors/worldview-2/). Annual NDVI composites from each year were clipped to the shapefile in ArcMap 10.3.1. We classified 2. Materials and methods vegetation cover into four categories based on NDVI: barren substrate (À 0.5-0.1), very sparse vegetation (0.1–0.2), sparse vegetation 2.1. Site description (0.2–0.5), and dense vegetation (0.5–0.7). Area of each category was calculated for every year in the dieback region using R (version 3.4.1). In 2011, a 1.23 km2 mangrove dieback was reported on Abaco Is­ We performed spot checks on the ground to confirmthe classificationof land, The Bahamas, by local fishermen. The dieback (26.51232000, each vegetation category within the dieback region. Based on this, we À 77.25843600) is located on the west side of the island in an area determined that vegetation cover values categorized as sparse and dense commonly referred to as The Marls (Fig. 1). This area is a complex vegetation (0.2–0.5, 0.5–0.7, respectively) were considered live dwarf mosaic of dwarf red mangroves (R. mangle), seagrass (primarily turtle mangrove. grass, Thalassia testudinum), sponges, and sand flats.Most of the area is Hurricane paths that directly passed over Abaco in 1989–2013 were isolated from direct human impacts, as it is extremely shallow and can acquired from the National Oceanic and Atmospheric Administration only be accessed with flats boats by those knowledgeable of the (NOAA) Digital Coasts website (https://coast.noaa.gov/hurricanes/) seascape. At the time of reporting, there were no consistent scientific (Table 1). Drought information was acquired from Herrera and Ault surveys of the mangroves in the Marls making the exact start of the (2017). Publicly available drought maps based on the self-calibrating dieback unclear. It is possible that dieback wasn’t reported until it Palmer Drought Severity Index (scPDSI) in the Caribbean (http://ecrl. reached a certain size as anglers and guides, who fishalong the margins eas.cornell.edu/node/56) were viewed and the average scPDSI value of the mangroves, may not have noticed the initial signs of dieback. In was estimated visually for Abaco Island for each month from 1989 to 2013 our research team visited the site and noted that the area was 2013. An average scPDSI value was then computed for each year and dominated by standing, dead, dwarf (~1m tall) R. mangle with no leaves used to classify droughts into four categories: none (>-1), mild (À 1 to but did have an area of dwarf R. mangle with green leaves still intact. We À 1.9), average (À 1.9 to À 2.9), and severe (<À 3) (Table 1). continued to visit the site 2–3 times/year between 2014 and 2017 when We constructed a time series of live mangrove vegetation cover and we began more in depth monitoring and surveys. During this time, we of maximum annual NDVI from 1989 to 2013 in R (version 3.4.1). saw no evidence of resprouting from any mangroves that did not have leaves the previous visit. Surveys in the area in 2014 revealed porewater salinities ranged Table 1 from 47 to 51 (n ¼ 48) and surface water temperatures ranged from 26 Storm tracks that crossed Abaco, The Bahamas, and mean scPDSI values between � to 42 C. Porewater salinities in the dieback area were similar to sur­ 1989 and 2013.Strength categories reflectthe strength of the storm as it passed over Abaco. Storm data was adapted from NOAA historical hurricane track rounding areas of live mangroves on Abaco (26–50, n ¼ 30) (Rossi, database (https://coast.noaa.gov/hurricanes/) and mean scPDSI values were unpublished data). All porewater samples were collected using Rhizon adapated from Herrera and Ault, 2017. samplers at a ~10 cm depth and salinity was measured with handheld refractometers using the Practical Salinity Scale (Seeberg-Elverfeldt Year Mean scPDSI Range Drought Storm Name Strength et al. 2005). Additional surveys suggested that chewing grazing scars 1989 À 2.50 À 3, À 1.5 Average None NA and disease symptoms (lesions) were present on remaining live 1990 À 3.42 À 4.5, À 1.5 Severe None NA 1991 À 2.21 À 3, 0 Average None NA mangrove leaves. Preliminary surveys for herbivorous crabs were con­ 1992 À 0.29 À 2.5, 1 None None NA ducted and no A. pisonii were documented. Sticky traps were deployed 1993 0.59 À 1.5, 2.5 None None NA throughout the dieback area to determine potential herbivores. 1994 0.67 À 0.5, 1.5 None None NA The Robust Bush Cricket (Tafilasca eleuthera) and Mangrove 1995 2.58 1, 4 None None NA ( pigmalion) caterpillars were found, both of which are known 1996 1.92 0.5, 3.5 None None NA 1997 1.33 -.05, 3 None None NA consumers of green R. mangle leaves (Onuf et al., 1977; Otte and 1998 2.00 À 2.5, 5 None None NA Perez-Gelabert, 2009). Isolations performed on live, green leaves adja­ 1999 À 1.42 À 3, 1 Mild Floyd H4 cent to the dieback suggest that an ascomycete fungus, Pestalotiopsis sp., 2000 À 0.33 À 1.5, 1 None None NA is responsible for the disease symptoms (Rossi, 2018). 2001 0.42 À 1.5, 3 None None NA 2002 3.00 1, 4 None None NA 2003 NA NA NA None NA 2.2. NDVI, hurricanes, and drought 2004 À 0.86 À 3, 2 None Jeanne H3 2005 À 1.42 À 2.5, 3 Mild None NA To investigate potential abiotic stressors, we quantified patterns of 2006 À 0.21 À 1, 0.5 None None NA À NDVI in the context of drought and hurricane occurrence from 1989 to 2007 0.21 2.5, 1.5 None None NA 2008 À 0.50 À 2, 1 None None NA 2013 to 1) determine if there were any long term correlations with NDVI 2009 À 0.58 À 3, 1.5 None None NA and 2) determine when the dieback likely began. Both drought and 2010 0.63 0, 2.5 None None NA hurricanes may have latent or lagging effects on ecosystems, as such we 2011 À 0.33 À 3, 2.5 None Irene H2 felt it was important to understand the history of drought and hurricanes 2012 2.17 0, 3.5 None Sandy H1 2013 2.96 1.5, 4.5 None None NA in this area over an extended period. We used NDVI as a proxy for green

3 R.E. Rossi et al. Estuarine, Coastal and Shelf Science 237 (2020) 106660

Maximum annual NDVI was chosen to ensure minor cloud cover did not leaf. The other leaf was marked as the control using a permanent marker affect the overall values and reflectsthe maximum NDVI value over the and flaggingtape. Leaf pairs were monitored for lesion development and entire dieback area. We then superimposed hurricane events and senescence over a 28-day period. If leaves senesced prior to the end of drought events on the time series to determine if a pattern between the 28-day period, they were removed and photographed for digital either event and changes in mangrove cover and maximum NDVI were analysis. All other leaves were collected at the end of the study period evident. Data were analyzed in R using a one-way ANOVA with type III and photographed for digital analysis of lesion density and grazing scars sums of squares (due to unbalanced sample sizes) to determine if there using ImageJ (Schneider et al., 2012). Data were analyzed using a were significantdifferences between maximum NDVI and live mangrove split-plot ANOVA with leaf pair as the block, site as a whole-plot factor, cover during hurricane or drought events (Fox and Weisberg, 2011). and type (cut vs uncut) as the split-plot factor. All analyses were completed in R (version 3.3.2 R Core Team (2018)). 2.3. Herbivory and disease surveys 3. Results Individual R. mangle trees (n ¼ 48) in the dieback area were tagged in May 2014 and monitored each summer (May–August) and winter 3.1. NDVI, hurricanes, and drought (December–January) until January 2016. Seasonal sampling was con­ ducted as there are significantdifferences between average temperature Maximum annual NDVI ranged 0.10 to 0.69 (Fig. 2) in the dieback and rainfall between the summer and winter seasons (weatherspark. between 1989 and 2013 and was correlated with live mangrove cover 2 com, Gelaro et al., 2017). If abiotic conditions interact with herbivory (t21 ¼ 5.48, p¼<0.001, R ¼ 0.59, Fig A4). Prior to 2008, mangrove and disease to impact mangrove health we hypothesized that we may see vegetation covered 0.67–0.82 km2 in the dieback region and barren area seasonal differences in measures of disease severity and mangrove covered ~0.2 km2 out of a total area of 1.2 km2 (Fig. 3). Following 2008, health. The total number of leaves, incidence and severity of putative there is a drastic drop in maximum annual NDVI through 2010 (Fig. 3). disease phenotypes, and grazing severity (number of leaves with grazed Mangrove vegetation declined from 0.62 km2 in 2008 to 0.10 km2 in edges) were recorded every six months for each tree (Fig A1). 2010 (Fig. 3). Maximum annual NDVI increased following the lowest Combining all observations from the herbivory and disease surveys, values in 2010, however, live mangrove vegetation coverage dropped to linear regression was used to determine if there was evidence of a ~0 km2 in 2011–2012 and increased slightly to 0.09 km2 in 2013 relationship between grazing pressure, disease, and the health of the (Fig. 3). During this period, four named storms hit Abaco ranging in mangroves. Specifically, the relationships between disease incidence strength from a category 1 to a category 4 hurricane (Table 1). Mangrove and proportion grazed, the number of dead branches and disease inci­ vegetation cover and maximum NDVI were not significantly different dence, and the number of damaged shoots and disease incidence were from non-hurricane years (F2,20 ¼ 2.81, p ¼ 0.08 (cover), F2,20, p ¼ 0.9 determined using individual linear regressions. The number of dead (max NDVI), Table A1). branches was logged transformed and the number of damaged shoots There were fivemild to average drought events on Abaco during this was cube root transformed to meet assumptions of normality and het­ period: 1989, 1990, 1991, 1999, and 2005. Annual scPDSI values ranged erogeneity of the residuals. After visual inspection of the data, there was from À 3.42 to 2.96 on Abaco during 1989–2013 (Table 1). In 2008 and evidence that the relationship between disease incidence and the pro­ 2009 the mean scPDSI values were À 0.50 and À 0.58, respectively portion of grazed leaves differed seasonally. As a result, this relationship (Table 1). Slightly positive (0.63) scPDSI values began in 2010 and was examined independently in data collected in the summer and winter increased to 2.16 and 2.96 in 2012 and 2013, respectively (Table 1). sampling periods. Live mangrove cover did not drop below 0.3 km2 and maximum NDVI was not lower than 0.37 during drought years. Mangrove vegetation 2.4. Simulated grazing experiment cover and maximum NDVI were not significantly different from non- drought years (F2,20 ¼ 2.81, p ¼ 0.08 (cover), F2,20, p ¼ 0.9 (max In June–July 2015, we conducted an experiment to determine if NDVI), Table A1). grazing resulted in increased frequency of putative disease phenotypes, We hypothesized that hurricane occurrence and drought severity specifically lesions. Four different mangrove creek sites were used for would correlate with reduced NDVI (or greenness) and would signal the the simulated grazing experiment: Camp Abaco (CA), Hills Creek (HC), onset of the mangrove dieback. Based on changes in NDVI, we suggest Snake Cay (SC), and Twisted Bridge (TB) (Fig. 1). Dwarf mangroves that the dieback began some time in 2008 when the area of live (predominately R. mangle) fringe these tidal creeks with primary benthic mangrove cover decreased, and was exacerbated 2010 to 2012 despite habitats being submerged seagrass (primarily T. testudinum), benthic increases in maximum NDVI. The last documented hurricane prior to macroalgae (e.g., Halimeda discoidea, Caulerpa spp., Batophora oerstedii, 2008 was in 2004, which suggests that a storm event was not directly Acetebularia spp.) and sand flats. responsible for the initial stage of this dieback. However, hurricanes did Prior to establishing the experiment, disease incidence surveys were occur in 2011 and 2012 that could have expanded the damage that conducted at each site. Disease incidence surveys consisted of sampling began in 2008. We also found that there was no drought event after all mangrove trees along a 50m transect and recording disease incidence 2005, suggesting that drought did not play a direct role in the dieback. and severity, DBH, number of shoots missing, grazing incidence, number Furthermore, we found no evidence that maximum annual NDVI or live of leaves, presence/absence of canker disease, number of shoots, pres­ mangrove coverage was statistically different from non-hurricane or ence/absence of snails, presence/absence of leaf galls, and presence of non-drought years. Together, this suggests that hurricane and drought dead matter (number of dead prop roots and branches) (Layman, 2016a, events were not major drivers of this dieback. 2016b). These sites varied in their disease incidence and severity, with TB having the highest disease incidence, followed by CA, HC, and SC 3.2. Herbivory and disease survey (Fig A2). At each site, leaf pairs that showed no signs of damage (i.e., no dis­ A wide range of herbivory was observed (0%–79% leaves grazed, ease and no grazing) were selected (n ¼ 150 pairs/site); each pair was mean ¼ 29%). Likewise, a wide range of disease damage was observed from the same leaf shoot. One leaf from the pair was cut, simulating (0%–97% leaves diseased, mean ¼ 47%). In the summer, there was no grazing, using crafting scissors. Cuts were performed along the leaf significant relationship between disease severity and grazing (t94 ¼ margins to mimic grazing scars noted in the dieback region (Fig A3). 1.57, p ¼ 0.12, R2 ¼ 0.02), however, in the winter there was a strong 2 2 Cuts were approximately 4 cm . Scissors were sterilized in a 10% bleach positive relationship (t45 ¼ 7.28, p < 0.001, R ¼ 0.53) with higher solution for 30s, rinsed with water, and dried by shaking between each grazing levels resulting in higher levels of disease severity (Fig. 4).

4 R.E. Rossi et al. Estuarine, Coastal and Shelf Science 237 (2020) 106660

Fig. 2. Landsat 7 images of the dieback region from 2008 to 2013 (when peak dieback occurred) with NDVI classification.Note that the color scheme reflectsbarren substrate (dark brown) to dense vegetation (pale yellow). (For interpretation of the references to color in this figurelegend, the reader is referred to the Web version of this article.)

Disease severity was positively correlated with both the number of dead 4. Discussion 2 branches (t94 ¼ 3.59, p < 0.001, R ¼ 0.11) and the number of damaged 2 shoots (t94 ¼ 4.24, p < 0.0001, R ¼ 0.15) in the entire dieback area. Mangroves are important habitat-forming foundation species that provide the framework for entire coastal communities in addition to 3.3. Simulated grazing experiment providing numerous ecosystem services (Dayton, 1972; Costanza et al., 1997; Bruno et al., 2003; Ellison et al., 2005; Barbier et al., 2011; There was no interaction between site and type (cut vs uncut) for Barbier, 2016). Mangroves regularly experience multiple stressors, lesion density and leaf area covered by lesions (F3,594 ¼ 0.10, p ¼ 0.96, which have driven extensive loss (Alongi, 2002; Halpern et al., 2008; Table 2). Yet, sites differed in leaf area covered by lesions (F3,594 ¼ 3.62, Barbier et al., 2011; Friess and Webb, 2014). Understanding these in­ p ¼ 0.01, Table 2) and by type (F1,594 ¼ 5.55, p ¼ 0.02, Table 2) with dividual drivers of loss, and interactions between them, is critical to simulated grazed leaves having a significantlyhigher proportion of leaf improving mangrove management and combating mangrove loss in area covered in lesions than control leaves (Fig. 5). There was not a general. We found that hurricanes and drought did not correlate with significant difference in the total number of lesions between simulated reduced NDVI and likely did not initiate the dieback. We also found grazed and control leaves (F1,595 ¼ 2.86, p ¼ 0.09); however, there was a evidence that herbivory likely facilitated the spread of disease contrib­ trend toward cut leaves having more lesions at two sites, CA and TB, uting to the dieback at our study site. which have greater disease severity than SC and HC (Fig A5). Although there were no hurricane events between 2008 and 2010 when NDVI (2010) and mangrove cover (2008) began to decline in our study, there is evidence that storm activity can have secondary effects

5 R.E. Rossi et al. Estuarine, Coastal and Shelf Science 237 (2020) 106660

Fig. 3. Top panel: Time series of annual mangrove cover from the dieback region between 1989 and 2013. The dashed lines indicate years hurricanes hit Abaco and the black lines indicate drought years based on Table 1. Bottom panel: Time series of maximum annual NDVI from the dieback region between 1989 and 2013. The maximum value is the greatest NDVI value from the entire dieback area. The dashed lines indicate years hurricanes hit Abaco and the black lines indicate drought years based on Table 1.

Table 2 Split-plot ANOVA table for herbivory survey and simulated grazing experiment.

Degrees of Freedom F-value p-value

Numerator Denominator

Herbivory survey Location 2 13 0.66 0.53 Treatment 2 26 0.96 0.40 Location*Treatment 4 26 0.86 0.50 Simulated grazing experiment Site 3 594 3.62 0.013 Type 1 594 5.55 0.02 Site*Type 3 594 0.10 0.96

Fig. 4. Linear regression of the proportion of diseased leaves by proportion of grazed leaves from the herbivory and disease survey in the dieback region. Data points that are open circles are from summer (R2 ¼ 0.02) and those that are soild red are from winter (R2 ¼ 0.53). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) that might not be visible for months to years (Ward et al., 2016; Asbridge et al., 2018). The most recent hurricane prior to the initiation of the dieback was four years before (2004). From 2004 to 2008, both maximum NDVI and live mangrove area increased in the dieback area suggesting that mangroves in this area were not likely impacted nega­ tively by the 2004 hurricane. Hurricanes impact soil elevation and hy­ Fig. 5. Disease severity across sites from simulated grazing experiment. Dark drology that could contribute to dieback immediately or months to years gray represents grazed and light gray represents control leaves. Error bars are after an event (Gilman et al., 2008; Asbridge et al., 2018). Although we standard deviations.

6 R.E. Rossi et al. Estuarine, Coastal and Shelf Science 237 (2020) 106660 did not measure this in our study, it is possible that these effects hurricanes (Sippo et al., 2018). Less work has documented herbivores contributed to the dieback, particularly in 2011 and 2012, after the and disease in diebacks, however, there are exceptions (Pegg et al., dieback likely began, when hurricanes occurred. The closest drought 1980; Wier et al., 2000; Reef et al., 2012; Osorio et al., 2016). Since event to the dieback was in 2005, suggesting that it had no immediate mangrove dieback events are likely underreported, potentially skewing effect on the mangroves. However, drought events, changes in relative our understanding of the main causes of diebacks, documenting indi­ sea levels, and changes in precipitation patterns, all of which increase vidual dieback events is extremely important. We have shown that salinities, may result in increased salt in mangrove tissues. Increased salt hurricane and drought events are not major drivers of this dieback, loads decrease water availability and lead to reductions in productivity suggesting other stressors are relevant. We suggest that herbivory and (Field, 1995). Red mangrove (R. mangle) is considered to have relatively disease contributed to the dieback and that herbivory facilitates disease high salinity tolerance, suggesting soil salinization may not be an infection. Whether hurricane and drought events facilitated the spread important factor (Reef and Lovelock, 2015). We investigated a dieback of herbivores and/or disease is unclear. Future studies should incorpo­ of dwarf R. mangle, that may have different physiological limits and rate multiple factors, both abiotic and biotic, to understand how these salinity tolerance. Dwarf R. mangle could have higher salinity tolerances factors may interact to affect mangrove ecosystems. than tall-form R. mangle, which might explain why we saw no evidence of a relationship between NDVI and drought events. Acknowledgements We also hypothesized that herbivory and disease would be greater in the dieback area and that biotic stressors would contribute to dieback We would like to thank A. Karaczynski, A. Todd, J. Albury, O. White, only after initiation by abiotic stressors. We did find high incidence of K. Perkinson, and A. Perkinson for assistance with fieldwork.We thank disease and chewing grazing damage in the dieback area suggesting that C. Ring for assistance with laboratory work. We thank R. Dunn, R. Irwin, both likely had a role in the dieback. Surveys in the mangrove dieback S.F. Jones, J.B. Ristaino, and B. Silliman and 2 anonymous reviewers for area show that grazing is strongly correlated with disease in this system. comments that improved the manuscript. This work was funded by the This relationship was stronger in winter months suggesting some po­ National Science Foundation (DGE-0946818, OCE-1541637, and OCE- tential seasonality, but this should be investigated further before strong 1405198). The project described in this publication was also supported conclusions can be drawn. Further, disease severity was positively by Cooperative Agreement No. G10AC00624 from the United States correlated with indicators of distress in mangroves, specifically Geological Survey. Its contents are solely the responsibility of the au­ damaged shoots and dead branches. Additionally, our simulated grazing thors and do not necessarily represent the views of the Southeast Climate experiment suggests that chewing herbivory facilitates disease infection Science Center or the USGS. This manuscript is submitted for publication of mangrove leaves. We found that disease severity was significantly with the understanding that the United States Government is authorized higher in experimentally opened leaves, regardless of the initial disease to reproduce and distribute reprints for Governmental purposes. The severity found at each site. We recognize that simulated grazing in our authors declare they have no conflict of interest. experiment does not incorporate secretions from insect mouthparts; however, our work does show that experimental opening of plant tissue Appendix A. Supplementary data does facilitate disease infection. Together, this suggests that chewing grazers could contribute to mangrove loss through facilitation of plant Supplementary data to this article can be found online at https://doi. disease and an overall reduction in available photosynthetic tissue. org/10.1016/j.ecss.2020.106660. Grazing and mechanical damage are linked to disease and ultimately die-offs in other coastal systems. For example, grazing by a butterflyfish, Chaetodon capistratus, increased the rate of infection of black-band dis­ References ease on corals such as Montastrea faveolata (Aeby and Santavy, 2006). Aeby S., G., Santavy L., D., 2006. Factors affecting susceptibility of the coral Montastraea Simulated grazing on the red macroalga D. pulchra facilitated disease faveolata to black-band disease. Mar. Ecol. Prog. Ser. 318, 103–110. infection by a bacterial pathogen, Nautella sp., that lead to bleaching of Agrios, G.N., 2005. Plant Pathology, 5 edition. Elsevier Academic Press, MA. ’ the alga (Campbell et al., 2014). Grazing by the green turtle, Chelonia Alongi, D.M., 2002. Present state and future of the world s mangrove forests. Environ. Conserv. 29, 331–349. mydas, was found to increase the likelihood of the seagrass Alongi, D.M., 2009. The Energetics of Mangrove Forests. Springer Science þ Business T. testudiunum being infected with Labyrinthula (Bowles and Bell, 2004). Media B.V. In U.S South Atlantic salt marshes, snail grazers injure plant tissue Asbridge, E., Lucas, R., Rogers, K., Accad, A., 2018. The extent of mangrove change and potential for recovery following severe Tropical Cyclone Yasi, Hinchinbrook Island, enabling the growth of Fusarium fungi that kill plant tissue that snails Queensland, Australia. Ecol. Evol. 8, 10416–10434. subsequently consume (Silliman and Zieman, 2001; Silliman and New­ Ban, S.S., Graham, N.A.J., Connolly, S.R., 2014. Evidence for multiple stressor ell, 2003). This combination of grazing and fungal infection led to interactions and effects on coral reefs. Global Change Biol. 20, 681–697. Barbier, E.B., 2016. The protective service of mangrove ecosystems: A review of greater loss of salt marsh grass than that of any single stressor alone valuation methods. Mar. Pollut. Bull. 109 (2), 676–681. https://doi.org/10.1016/j. (Silliman et al., 2005). Both grazing and disease have direct implications marpolbul.2016.01.033. on photosynthesis (e.g., lesions reduce area for photosynthesis to occur) Barbier, E.B., Hacker, S.D., Kennedy, C., Koch, E.W., Stier, A.C., Silliman, B.R., 2011. The value of estuarine and coastal ecosystem services. Ecol. Monogr. 81, 169–193. (Roloff et al., 2004; Agrios, 2005). The more severe the damage, the Bertness, M.D., Crain, C., Holdredge, C., Sala, N., 2008. Eutrophication and consumer greater the potential reduction in photosynthesis. For mangroves in control of new England salt marsh primary productivity. Conserv. Biol. : J. Soc. particular, increased severity of foliar pathogens or herbivory could Conserv. Biol. 22, 131–139. have major implications as mangrove leaves are fairly long lived; mean Bowles, J.W., Bell, S.S., 2004. Simulated herbivory and the dynamics of disease in Thalassia testudinum. Mar. Ecol. Prog. Ser. 283, 127–132. leaf life span is 16 months (Reef et al., 2010) and the leaves have a thick, Bruno, J.F., Petes, L.E., Drew Harvell, C., Hettinger, A., 2003. Nutrient enrichment can waxy cuticle both of which require high resource investment (Reich increase the severity of coral diseases. Ecol. Lett. 6, 1056–1061. et al., 1997). Ultimately, both grazing and disease lead to reductions in Bruno, J.F., Selig, E.R., Casey, K.S., Page, C.A., Willis, B.L., Harvell, C.D., Sweatman, H., Melendy, A.M., 2007. Thermal stress and coral cover as drivers of coral disease photosynthetic tissue and can lead to dieback events in a variety of outbreaks. PLoS Biol. 5, e124. systems (Lugo, 1998; Silliman et al., 2005; Campbell et al., 2014, Lewis Cahoon, D.R., Hensel, P., Rybczyk, J., McKee, K.L., Proffitt,C.E., Perez, B.C., 2003. Mass et al., 2016). tree mortality leads to mangrove peat collapse at Bay Islands, Honduras after Hurricane Mitch. J. Ecol. 91, 1093–1105. Campbell, A.H., Verg�es, A., Steinberg, P.D., 2014. Demographic consequences of disease 5. Conclusions in a habitat-forming seaweed and impacts on interactions between natural enemies. Ecology 95, 142–152. Chander, G., Markham L., B., Helder L., D., 2009. Summary of current radiometric Approximately 70% of published mangrove dieback events due to calibration coefficients for Landsat MSS, TM, ETMþ, and EO-1 ALI sensors. Remote natural causes occurred as a result of an extreme weather event such as Sensing of Environment 113, 893–903.

7 R.E. Rossi et al. Estuarine, Coastal and Shelf Science 237 (2020) 106660

Chellamani, P., Singh, C.P., Panigrahy, S., 2014. Assessment of the health status of Indian Oceanography Data Management Office (BCO-DMO). https://www.bco-dmo.org/ mangrove ecosystems using multi temporal remote sensing data. Trop. Ecol. 55, dataset/653909/data. 245–253. Lee, S., 1998. Ecological role of grapsid crabs in mangrove ecosystems: a review. Mar. Costanza, R., d’Arge, R., de Groot, R., Farber, S., Grasso, M., Hannon, B., Limburg, K., Freshw. Res. 49, 335–343. Naeem, S., O’Neill, R.V., Paruelo, J., Raskin, R.G., Sutton, P., van den Belt, M., 1997. Lewis III, R.R., Milbrandt, E.C., Brown, B., Krauss, K.W., Rovai, A.S., Beever III, J.W., The value of the world’s ecosystem services and natural capital. Nature 387, 253. Flynn, L.L., 2016. Stress in mangrove forests: early detection and preemptive Dayton, P., 1972. Toward an understanding of community resilience and the potential rehabilitation are essential for future successful worldwide mangrove forest effects of enrichments to the benthos at McMurdo Sound, Antartica. In: BC, P. (Ed.), management. Mar. Pollut. Bull. 109, 764–771. Proceedings of the Colloquium on Conservation Problems in Antartica. Allen Press, Lovelock, C.E., Feller, I.C., Reef, R., Hickey, S., Ball, M.C., 2017. Mangrove dieback Lawrence, KS. during fluctuating sea levels. Sci. Rep. 7, 1680. Deegan, L.A., Johnson, D.S., Warren, R.S., Peterson, B.J., Fleeger, J.W., Fagherazzi, S., Lugo, A.E., 1998. Mangrove forests: a tough system to invade but an easy one to Wollheim, W.M., 2012. Coastal eutrophication as a driver of salt marsh loss. Nature rehabilitate. Mar. Pollut. Bull. 37, 427–430. 490, 388. Onuf, C.P., Teal, J.M., Valiela, I., 1977. Interactions of nutrients, plant growth and Duke, N.C., Kovacs, J.M., Griffiths, A.D., Preece, L., Hill, D.J.E., van Oosterzee, P., herbivory in a mangrove ecosystem. Ecology 58, 514–526. Mackenzie, J., Morning, H.S., Burrows, D., 2017. Large-scale dieback of mangroves Osorio, J.A., Wingfield,M.J., Roux, J., 2016. A review of factors associated with decline in Australia’s Gulf of Carpentaria: a severe ecosystem response, coincidental with an and death of mangroves, with particular reference to fungal pathogens. South Afr. J. unusually extreme weather event. Mar. Freshw. Res. 68, 1816–1829. Bot. 103, 295–301. https://doi.org/10.1016/j.sajb.2014.08.010. Ellison, A.M., Bank, M.S., Clinton, B.D., Colburn, E.A., Elliott, K., Ford, C.R., Foster, D.R., Otte, D., Perez-Gelabert, D., 2009. Caribbean Crickets. Orthopterists’ Society. Kloeppel, B.D., Knoepp, J.D., Lovett, G.M., Mohan, J., Orwig, D.A., Rodenhouse, N. Pegg, K., Gillespie, N., Forsberg, L., 1980. Phytophthora sp. associated with mangrove L., Sobczak, W.V., Stinson, K.A., Stone, J.K., Swan, C.M., Thompson, J., Von death in Central coastal Queensland. Australas. Plant Pathol. 9, 6–7. Holle, B., Webster, J.R., 2005. Loss of foundation species: consequences for the Reef, R., Lovelock, C.E., 2015. Regulation of water balance in mangroves. Ann. Bot. 115, structure and dynamics of forested ecosystems. Front. Ecol. Environ. 3, 479–486. 385–395. Erickson, A.A., Saltis, M., Bell, S.S., Dawes, C.J., 2003. Herbivore feeding preferences as Reef, R., Feller, I.C., Lovelock, C.E., 2010. Nutrition of mangroves. Tree Physiol. 30, measured by leaf damage and stomatal ingestion: a mangrove crab example. J. Exp. 1148–1160. Mar. Biol. Ecol. 289, 123–138. Reef, R., Ball, M.C., Lovelock, C.E., 2012. The impact of a locust plague on mangroves of Feller, I.C., 2002. The role of herbivory by wood-boring insects in mangrove ecosystems the arid Western Australia coast. J. Trop. Ecol. 28, 307–311. in Belize. Oikos 97, 167–176. Reich, P.B., Walters, M.B., Ellsworth, D.S., 1997. From tropics to tundra: global Feller, I.C., Mathis, W.N., 1997. Primary herbivory by wood-boring insects along an convergence in plant functioning. Proc. Natl. Acad. Sci. Unit. States Am. 94, architectural gradient of Rhizophora mangle. Biotropica 29, 440–451. 13730–13734. Field, C.D., 1995. Impact of expected climate change on mangroves. Hydrobiologia 295, Roloff, I., Scherm, H., van Iersel, M.W., 2004. Photosynthesis of blueberry leaves as 75–81. affected by Septoria leaf spot and abiotic leaf damage. Plant Dis. 88, 397–401. Flores-Cardenas,� F., Millan-Aguilar,� O., Díaz-Lara, L., Rodríguez-Arredondo, L., Hurtado- Rossi, R.E., 2018. The Role of Multiple Stressors in a Mangrove Die-Off: A Case Study in � Oliva, M.A., Manzano-Sarabia, M., 2018. Trends in the normalized difference the Bahamas Archipelago. North Carolina State University. vegetation index for mangrove areas in Northwestern Mexico. J. Coast Res. 34, Servino, R.N., Gomes, L.E.d.O., Bernardino, A.F., 2018. Extreme weather impacts on 877–882. tropical mangrove forests in the Eastern Brazil Marine Ecoregion. Sci. Total Environ. Fox, J., Weisberg, S., 2011. An {R} Companion to Applied Regression, second ed. Sage, 628–629, 233–240. Thousand Oaks, CA. Silliman, B.R., Bertness, M.D., 2002. A trophic cascade regulates salt marsh primary Friess, D.A., Webb, E.L., 2014. Variability in mangrove change estimates and production. Proc. Natl. Acad. Sci. Unit. States Am. 99, 10500–10505. implications for the assessment of ecosystem service provision. Global Ecol. Silliman, B.R., Newell, S.Y., 2003. Fungal farming in a snail. Proc. Natl. Acad. Sci. U. S. A. Biogeogr. 23, 715–725. 100, 15643–15648. Gelaro, R., McCarty, W., Suarez,� M.J., Todling, R., Molod, A., Takacs, L., Randles, C.A., Silliman, B.R., Zieman, J.C., 2001. Top-down control of Spartina alternifloraproduction Darmenov, A., Bosilovich, M.G., Reichle, R., Wargan, K., Coy, L., Cullather, R., by periwinkle grazing in a Virginia salt marsh. Ecology 82, 2830–2845. Draper, C., Akella, S., Buchard, V., Conaty, A., Silva, A.M.d., Gu, W., Kim, G.-K., Silliman, B.R., Van de Koppel, J., Bertness, M.D., Stanton, L.E., Mendelssohn, I.A., 2005. Koster, R., Lucchesi, R., Merkova, D., Nielsen, J.E., Partyka, G., Pawson, S., Drought, snails, and large-scale die-off of southern US salt marshes. Science 310, Putman, W., Rienecker, M., Schubert, S.D., Sienkiewicz, M., Zhao, B., 2017. The 1803–1806. modern-era retrospective analysis for research and applications, version 2 (MERRA- Sippo, J.Z., Lovelock, C.E., Santos, I.R., Sanders, C.J., Maher, D.T., 2018. Mangrove 2). J. Clim. 30, 5419–5454. mortality in a changing climate: an overview. Estuarine. Coast. Shelf Sci. 215, Gilman, E.L., Ellison, J., Duke, N.C., Field, C., 2008. Threats to mangroves from climate 241–249. change and adaptation options: a review. Aquat. Bot. 89, 237–250. Schneider A., C., Rasband S., W., Eliceiri W., K., 2012. NIH Image to ImageJ: 25 years of Giri, C., Long, J., Abbas, S., Murali, R.M., Qamer, F.M., Pengra, B., Thau, D., 2015. image analysis. Nature Methods 9, 671–675. Distribution and dynamics of mangrove forests of South Asia. J. Environ. Manag. Seeberg-Elverfeldt, J., Schlüter, M., Feseker, T., Kolling,€ M., 2005. Rhizon sampling of 148, 101–111. porewaters near the sediment-water interface of aquatic systems. Limnol Oceanogr. Halpern, B.S., Walbridge, S., Selkoe, K.A., Kappel, C.V., Micheli, F., D’Agrosa, C., Methods 3, 361–371. Bruno, J.F., Casey, K.S., Ebert, C., Fox, H.E., Fujita, R., Heinemann, D., Lenihan, H.S., Sousa, W.P., Kennedy, P.G., Mitchell, B.J., 2003. Propagule size and predispersal damage Madin, E.M.P., Perry, M.T., Selig, E.R., Spalding, M., Steneck, R., Watson, R., 2008. by insects affect establishment and early growth of mangrove seedlings. Oecologia A global map of human impact on marine ecosystems. Science 319, 948–952. 135, 564–575. He, Q., Silliman, B.R., 2016. Consumer control as a common driver of coastal vegetation UNEP, U. E. P., 2006. Marine and Coastal Ecosystem and Human Well-Being: A Synthesis worldwide. Ecol. Monogr. 86, 278–294. Report Based on the Finding of the Millennium Ecosystem Assessment. UNEP, Herrera, D., Ault, T., 2017. Insights from a new high-resolution drought atlas for the Nairobi, Kenya. Caribbean spanning 1950–2016. J. Clim. 30, 7801–7825. Wang, L., Jia, M., Yin, D., Tian, J., 2019. A review of remote sensing for mangrove Hughes, T.P., Baird, A.H., Bellwood, D.R., Card, M., Connolly, S.R., Folke, C., forests: 1956–2018. Rem. Sens. Environ. 231, 111223. Grosberg, R., Hoegh-Guldberg, O., Jackson, J.B., Kleypas, J., 2003. Climate change, Ward, R.D., Friess, D.A., Day, R.H., MacKenzie, R.A., 2016. Impacts of climate change on human impacts, and the resilience of coral reefs. Science 301, 929–933. mangrove ecosystems: a region by region overview. Ecosys. Health Sustain. 2, Layman, C.A., 2016. Data from an herbivore exclusion experiment conducted on Abaco e01211. Island in the Bahamas from 2014-2016 (Mangrove Die-off project). Biological and weathersparkcom. Weather Spark. Cedar Lake Ventures, In. Chemical Oceanography Data Management Office (BCO-DMO). https://www.bc Wier, A.M., Tattar, T.A., Klekowski, E.J., 2000. Disease of red mangrove (Rhizophora o-dmo.org/dataset/664990/data. mangle) in Southwest Puerto Rico caused by Cytospora rhizophorae. Biotropica 32, Layman, C.A., 2016. Disease incidence survey and site descriptions on Abaco Island, 299–306. Bahamas in 2015-2016 (Mangrove Die-off project). Biological and Chemical

8