Life stage influences the resistance and resilience of black mangrove forests to winter climate extremes

In subtropical coastal wetlands on multiple continents, climate change-induced reductions in the frequency and intensity of freezing temperatures are expected to lead to the expansion of woody plants (i.e., mangrove forests) at the expense of tidal grasslands (i.e., salt marshes). Since some ecosystem goods and services would be affected by mangrove range expansion, there is a need to better understand mangrove sensitivity to freezing temperatures as well as the implications of changing winter climate extremes for mangrove-salt marsh interactions. In this study, we investigated the following questions: (1) how does plant life stage (i.e., ontogeny) influence the resistance and resilience of black mangrove (Avicennia germinans) forests to freezing temperatures; and (2) how might differential life stage responses to freeze events affect the rate of mangrove expansion and salt marsh displacement due to climate change? To address these questions, we quantified freeze damage and recovery for different life stages (seedling, short tree, and tall tree) following extreme winter air temperature events that occurred near the northern range limit of A. germinans in North America. We found that life stage affects black mangrove forest resistance and resilience to winter climate extremes in a nonlinear fashion. Resistance to winter climate extremes was high for tall A. germinans trees and seedlings, but lowest for short trees. Resilience was highest for tall A. germinans trees. These results suggest the presence of positive feedbacks and indicate that climate-change induced decreases in the frequency and intensity of extreme minimum air temperatures could lead to a nonlinear increase in mangrove forest resistance and resilience. This feedback could accelerate future mangrove expansion and salt marsh loss at rates beyond what would be predicted from climate change alone. In general terms, our study highlights the importance of accounting for differential life stage responses and positive feedbacks when evaluating the ecological effects of changes in the frequency and magnitude of climate extremes.


INTRODUCTION
There is a need to better understand and prepare for the ecological and conservation implications of future changes in climate extremes (Jentsch et al. 2007, Glick et al. 2011, IPCC 2013, Stein et al. 2014).Potential ecosystem responses to climate extremes are diverse, ranging from abrupt state changes that produce major alterations in ecosystem structure and function to minor events with negligible ecological implications (Smith 2011).Research that improves our understanding of the resistance and resilience of ecosystems to climate extremes plays an especially important role in helping future-focused environmental managers better gauge and plan for potential climate change impacts (Lloret et al. 2012, Hoover et al. 2014, Ruppert et al. 2015).In this study, we examine the life-stage dependent effects of winter climate extremes upon the resistance and resilience of an important coastal wetland foundation species.We define resistance as the capacity to withstand change (i.e., not be perturbed or affected by a climate extreme) and resilience as the capacity to recover after perturbation (i.e., recover after being affected by a climate extreme; sensu Pimm 1984, Tilman and Downing 1994, Hoover et al. 2014).
Across the globe, minimum temperature extremes greatly influence the distribution and productivity of plant species and ecosystems (Holdridge 1967, Sakai and Weiser 1973, Woodward 1987).In tidal saline wetlands, changing climatic conditions, in the form of a decrease in the frequency and intensity of severe freeze events, are expected to lead to the poleward migration of woody plants (i.e., mangrove trees and shrubs) at the expense of graminoid and succulent plants (i.e., salt marsh plants; Osland et al. 2013, Saintilan et al. 2014, Alongi 2015, Cavanaugh et al. 2015).In coastal wetlands located near the transition between subtropical and warm temperate climatic zones, severe freeze events greatly affect the abundance and productivity of mangrove trees relative to salt marsh graminoid and succulent plants (West 1977, Woodroffe and Grindrod 1991, McKee et al. 2012).Low winter air temperature extremes can damage or kill mangrove trees (Fig. 1; Davis 1940, Sherrod and McMillan 1985, Stuart et al. 2007, Ross et al. 2009), which can lead to salt marsh plant dominance.However, small increases in extreme minimum winter air temperatures can lead to landscape-scale changes in ecosystem structure and function as mangrove forests replace salt marshes (Osland et al. 2013, Cavanaugh et al. 2014;Osland et al., in press).From a functional perspective, mangrove and salt marsh plants are both considered foundation species because they create habitat, modulate ecosystem functions, and support entire ecological communities (sensu Dayton 1972, Bruno and Bertness 2001, Ellison et al. 2005, Angelini et al. 2011).Despite many similarities, there are several functional and structural differences between mangrove and salt marsh ecosystems (Day et al. 2013).Given the potential for poleward mangrove range expansion in response to climate change and the expected concomitant changes in the provision of some goods and services, there is a need to better understand the resistance and resilience of mangrove forests to severe freeze events.
Ontogeny influences resistance and resilience in most natural systems.As individuals, populations, or ecosystems become older, more complex, and/or more developed, their ability to resist and recover from extreme events often changes (Odum 1969, Pimm 1984, Sousa 1984, Tilman and Downing 1994, Boege and Marquis 2005, Gabler and Siemann 2012).There are some examples of plants becoming more resistant and resilient to severe freeze events as they develop and mature (Sakai andLarcher 1987, Larcher 2003).For example, older citrus trees (Citrus spp.) are often more resistant to severe freeze events than younger trees and transplants (Yelenosky 1996), mature Ohio buckeye trees (Aesculus glabra Willd.) are often more frost tolerant than saplings (Augspurger 2013), and adult chaparral shrubs are more freeze-tolerant than seedlings (Boorse et al. 1998).Along the Gulf of Mexico coast, the black mangrove (Avicennia germinans (L.) L.) is the mangrove species that is most resistant and resilient to severe freeze events (West 1977, Sherrod andMcMillan 1985).Because extreme winter events are rare by definition, field-based studies of mangrove sensitivity to freezing are logistically challenging and dependent upon the occurrence of fortuitous events and/or opportunistic field experiments.To our knowledge, there has not been a study that has quantified the effects of ontogenetic differences in A. germinans tree resistance and resilience to severe freeze events (but see Pickens and Hester 2011 for a valuable comparison of early life stages: propagules vs. seedlings).Although it is possible that there would be no effect of life stage upon resistance and resilience (i.e., the relationship illustrated in Fig. 2A), our hypothesis was that A. germinans resistance and resilience would in-crease as plants develop and grow; in other words, taller and more developed A. germinans individuals would be more resistant and resilient to winter climate extremes than smaller individuals (i.e., the relationship illustrated in Fig. 2B).A positive ontogenetic effect upon resistance and resilience in mangrove forests (e.g., the relationships illustrated in Fig. 2B or C) would imply the presence of a positive feedback loop that may v www.esajournals.orgaccelerate the rate of mangrove growth and expansion in a future with a lower frequency and intensity of severe freeze events.Such information is important because it will improve future-focused models as well as climate change adaptation planning and restoration efforts.
Here we investigated the effects of ontogeny upon the resistance and resilience of A. germinans individuals to winter climate extremes.We specifically investigated the following question: how does life stage influence the resistance and resilience of A. germinans individuals to extreme low winter air temperatures?To address this question and test our hypothesis, we quantified freeze damage (i.e., resistance) and recovery (i.e., resilience) for different vegetation strata (i.e., seedling, short tree, and tall tree) following extreme winter air temperatures that occurred in January 2014 near the poleward range limit of A. germinans in the northern Gulf of Mexico.

Study area
This research was conducted within a 5-km stretch of tidal saline wetlands located between Port Fourchon and Leeville, Louisiana (USA; 29810 0 0 00 N and 90814 0 0 00 W).This location is in the Mississippi River deltaic plain and the soils are classified within the Bellpass and Scatlake soil series (Terric Medisaprists and Typic Hydraquents, respectively).The study area contains a matrix of salt marshes and freeze-affected mangrove forests.Climate-mediated marsh-mangrove interactions in Louisiana are dynamic and have attracted the interest of many ecologists studying the implications of mangrove expansion into salt marsh (e.g., West 1977, Patterson and Mendelssohn 1991, McKee and Rooth 2008).The salt marshes within this area are dominated primarily by smooth cordgrass (Spartina alterniflora Loisel.).To our knowledge, A. germinans is the only species of mangrove found in the study area and throughout the state of Louisiana.Although A. germinans has been present in parts of Louisiana for at least the last century (Penfound and Hathaway 1938, O'Neil 1949, West 1977) and probably much longer (Woodroffe and Grindrod 1991, Saintilan and Rogers 2015), the abundance, structural complexity, and spatial coverage of mangrove forests in the Port Fourchon area over the past century have likely expanded and contracted in response to the duration, frequency, and intensity of extreme winter events (West 1977, Osland et al. 2013).For the last several decades (i.e., since the last major freeze event in 1989), the spatial coverage of A. germinans has increased within Port Fourchon (Perry andMendelssohn 2009, Giri et al. 2011) and other parts of the northern Gulf of Mexico (Stevens et al. 2006, Montagna et al. 2011) at the expense of salt marsh vegetation.Although severe freeze events in Louisiana can lead to mangrove mortality, moderate freeze events can cause aboveground damage that is typically followed by vigorous resprouting from the stem and/or base.Damaging freeze events in Louisiana are frequent enough that A. germinans individuals are typically multi-stemmed, shorter, and more shrub-like than their tropical counterparts (Osland et al. 2014a).Most A. germinans individuals in the Port Fourchon area are

Temperatures during the winter of 2013-2014
Across much of the eastern United States, air temperatures during the winter of 2013-2014 were below average.In coastal Louisiana, two extreme winter events (6-8 and 28-30 January 2014) resulted in minimum air temperatures recorded at nearby weather stations that ranged from À3.38 to À4.48 and À1.18 to À3.98C, respectively (Table 1).For each of these two events, the number of consecutive hours below 08C recorded at these weather stations ranged from 12 to 18 and 9 to 17 hours, respectively (Table 1).Although these temperatures were not cold enough in the Port Fourchon area to cause extensive mangrove mortality or damage similar to the freezes of the 1980s, temperatures were low enough in some areas to cause minor damage and some mortality worthy of investigation (see photos in Fig. 1).
In the two months following these 2014 freeze events, R. H. Day and/or M. J. Osland made four separate trips to the study area to plan this study and observe the vegetation damage due to the freeze events.For A. germinans, leaf damage due to freezing occurs quickly and can be easily identified in the days following a freeze event.Freeze-damaged A. germinans leaves turn brown (Fig. 1A) and the color of these leaves contrasts sharply with any live leaves that may be present.In moderate cases, only a few leaves may be damaged on each individual, and freeze-damage observations require close examination (e.g., within ;1 m of an individual); however, in more extreme cases, all of the leaves on an individual may already be brown on the day following a freeze event and the color contrast can be seen from a larger distance (e.g., ;100 m).With the passage of time (i.e., ;weeks to months depend-ing upon conditions), freeze-damaged leaves fall off of the tree, leaving branches and stems that remain leafless (Fig. 1B and C) until the growing season when resprouting occurs and new leaves are produced.Where tidal inundation is minimal, piles of these freeze-damaged leaves can be observed at the base of the tree.After a moderate freeze, many of the leafless branches and stems will still be visible from a distance at the end of the first growing season.However, by the end of the second post-freeze growing season, the signs of freeze damage are less visible and may require close examination (R. H. Day, personal observation).Because other physiological stressors can also result in leafless mangroves (e.g., drought, high salinity, excessive flooding), field observations made within the first several days and weeks following a freeze event can help to determine that the cause of damage was due to freezing temperatures.Initial scouting trips leading to this study were made on 8 January, 16 January, 23 January, and 7 March 2014 (i.e., approximately 1, 9, 16, and 59 days following the first 2014 freeze event, respectively).During these trips, R. H. Day and/or M. J. Osland made observations of mangrove freeze damage in coastal wetlands present in the area between Port Fourchon and Golden Meadow, Louisiana.During these trips, direct field observations were made of the life-stage dependent freeze effects quantified in this study.Those initial post-freeze field observations along with similar observations made following a freeze event that occurred in the study area in 2010 (R. H. Day, unpublished data), led to the development and implementation of this study.Pre-and post-freeze observations are an important component of this study because they allow the authors to attribute the damage quantified in June 2014 to the freeze events that occurred in January 2014, and also to recognize remnant dead wood still present from the 2010 freeze.

Study design
Because the effects of freeze damage were spatially heterogeneous, we quantified resistance and resilience for multiple vegetation strata in areas with and without extensive freeze damage.Our study design included two freeze damage treatments: (1) a ''Damaged'' treatment consisting of sites with visible freeze damage in the form of mangroves with brown leaves and/or leafless branches; and (2) a ''Not Damaged'' treatment consisting of sites with little to no visible freeze damage (i.e., a very small number of brown leaves and/or leafless branches).The study included a total of six sites: three Damaged sites and three Not Damaged sites.Each of the three Damaged sites was paired with a Not Damaged site using a block design.The distances between paired sites within blocks were 1.6, 1.4, and 3.5 km.At each site, we established one 50m 2 circular plot (radius ¼ 4 m).Within each 50m 2 plot, we established a series of randomlylocated nested subplots that included three 2-m 2 subplots (i.e., 1 3 2 m subplots) and three 0.25-m 2 subplots (i.e., 0.5 3 0.5 m subplots) for strataspecific vegetation measurements.

Vegetation
All vegetation measurements were conducted in June 2014.Vegetation measurements were made within three strata: (1) a tall mangrove stratum that consisted of individuals with a height that was greater than or equal to 140 cm; (2) a short mangrove stratum that consisted of individuals with a height that was greater than or equal to 50 cm but less than 140 cm; and (3) a mangrove seedling stratum that consisted of individuals with a height of less than 50 cm.The seedling stratum included a mixture of seedlings that became established prior to and after the 2014 freezes.Strata-specific measurements were conducted in density-appropriate plot sizes.The tall mangrove stratum measurements were conducted across the entire 50-m 2 plot.Within each 50-m 2 plot, the short mangrove and the mangrove seedling strata measurements were conducted within the three 2-m 2 subplots and the three 0.25-m 2 subplots, respectively.
For each individual in each of the three strata, we recorded the following information: (1) the percentage of the plant that was damaged by freeze; (2) the height of the plant prior to the freeze (i.e., the height of the tallest stem or branch tip regardless of damage); (3) the height of the plant after the freeze (i.e., the height of the tallest non-freeze damaged stem or branch tip); (4) the longest crown diameter of the individual prior to the freeze (CD1); (5) the longest secondary crown diameter of the individual prior to the freeze in a direction perpendicular to CD1 (CD2); (6 ) whether the individual resprouted from the stem; (7) whether the individual resprouted from the base; (8) whether there was indication (i.e., presence of standing dead stems less than 1 m in height) that the plant had suffered damage due to a freeze event in 2010 (9-11 January 2010).At these sites, damage from that 2010 freeze event was still present in the form of standing dead stems that were less than 1 m in height.A gradual decay and loss of damaged aboveground material following the 2010 freeze event has been documented (R. H. Day, personal observation and unpublished data).
Within the 2-m 2 subplots, we also estimated the percent coverage of salt marsh species (i.e., graminoid and succulent plants).We calculated A. germinans biomass via allometric equations that utilize plant volume (i.e., a combination of crown area and plant height; Osland et al. 2014a).Freeze resistance metrics that were determined include freeze damage (i.e., the percentage of the plant damaged), freeze-induced height reduction (i.e., the difference between the pre-freeze and post-freeze heights), and the freeze-induced percent height reduction (i.e., the percentage of the pre-freeze height that was reduced due to the freeze).Freeze resilience was quantified as the percent of individuals resprouting from the base or stem after freeze damage.

Soil and elevation
In order to account for potential edaphic or hydrologic effects, we measured the elevation and a basic suite of soil properties present at each site.We measured elevation because we expect that tidal inundation may affect near-soil surface temperatures during freeze events.Elevation was measured at each 50-m 2 plot centroid using a high-precision Global Navigation Satellite System (GNSS; Trimble R8 and TSC3, Trimble, Sunnyvale, California, USA) and the Continuously Operating Reference Stations (CORS) contained within Louisiana State University's v www.esajournals.orgGULFNet real-time network (RTN).
Soil samples were collected to 15-cm depth using a custom-made stainless steel split-coring cylinder (diameter: 4.7 cm).At each site, one soil sample was collected within a 1-m buffer of each of the three 2-m 2 subplots (i.e., a total of three samples were collected per site).Samples were stored in a cooler with ice while in the field and at 48C while in the laboratory.Soil bulk density was determined as a simple dry weight to volume ratio (Blake and Hartge 1986).Soil organic matter was determined via loss on ignition (Karam 1993).

Data analyses
For the site-level data analyses, the subplotlevel data were converted to site-level data (i.e., site-level means were calculated from the subplot-level data).Initial models indicated that the block effect was not significant and was therefore not included as a variable in subsequent models.To determine the effects of vegetation strata, freeze damage treatment, and their interaction upon mangrove resistance and resilience, we used analysis of variance (ANOVA) models that included vegetation strata, freeze damage treatment, and their interaction as independent variables.Response variables included site-level mean percent freeze damage, percent height reduction, and actual height reduction.For those individuals that were most damaged by the 2014 freeze events (i.e., all of the data from the tall and short strata within the Damaged treatment), we also evaluated ANOVA models that included vegetation strata, resprout location, and their interaction as independent variables and the percent resprouting individuals as the dependent variable.To illustrate the height-dependent freeze damage at the Damaged sites, we combined all of the individual-based data within the Damaged sites and calculated the proportion of individuals within 10% freeze damage increments for each the following 50-cm increment plant height categories: 0-50 cm, 50-100 cm, 100-150 cm, 150-200 cm, 200-250 cm, and 250-300 cm.Post hoc mean comparisons were conducted using Tukey's Studentized Range (HSD) tests.A Student's t-test was used to compare the percentage of tall stratum individuals with visible damage remaining from the winter of 2010 within the freeze damage treat-ments.Soil and elevation differences between the freeze damage treatments were evaluated using Student's t-tests, and for the freeze resistancefocused response variables (i.e., site-level mean percent freeze damage, percent height reduction, actual height reduction), we evaluated analysis of covariance (ANCOVA) models that included the freeze damage treatment as a factor and individual soil or elevation properties (i.e., soil bulk density, soil organic matter, or elevation) as a covariate.The data met the assumptions for ANOVA and ANCOVA.All data analyses were conducted in R (http://cran.r-project.org;Version 3.0.1).

Resistance to winter climate extremes
For the freeze resistance metrics (i.e., percent freeze damage, percent height reduction, and actual height reduction), there was a significant interaction between the freeze damage treatments and vegetation strata (Table 2; Figs.3A-C and 4).Freeze resistance metrics were affected by vegetation strata within the Damaged treatment, but not in the Not Damaged treatment (Figs. 3A-C and 4).Within the Damaged treatment, percent freeze damage and percent height reduction were highest in the short mangrove stratum, lower in the tall mangrove stratum, and close to zero in the mangrove seedling stratum (Figs.3A-B and 4).Within the Damaged treatment, actual height reduction was highest in the short and tall mangrove strata and close to zero in the mangrove seedling stratum (Fig. 3C).Within the Not Damaged treatment, percent freeze damage, percent height reduction, and actual height reduction were close to zero and not different for all three strata (Fig. 3A-C).

Resilience to winter climate extremes
For the percentage of individuals that resprouted after damage from the 2014 freeze events (i.e., for those individuals in the tall and short strata within the Damaged treatment), there was a significant interaction between vegetation strata and resprouting location (F 1,8 ¼ 22.1, P , 0.01; Fig. 5).For both the tall and short strata, resprouting percentage was higher from the stem than the base (Fig. 5).The percentage of individuals resprouting from the stem was 100 % for both the tall and short strata (Fig. 5).Basal resprouting, however, was higher in the tall mangrove stratum than in the short mangrove stratum (Fig. 5).
Freeze-damage from 2010 was still visible at some of the Damaged and Not Damaged sites (Fig. 6).The percent of individuals within the Tall Stratum that still had visible damage from the 2010 freeze event was higher at the Damaged sites than the Not Damaged sites (P ¼ 0.046; Fig. 6).

Soil and elevation
The ANCOVA models indicated that elevation and soil properties did not significantly affect any of the freeze resistance-focused response variables (i.e., site-level mean percent freeze damage, percent height reduction, actual height reduction).There was no difference in soil bulk density, soil organic matter, or elevation between the freeze damage treatments (P ¼ 0.63, P ¼ 0.75, P ¼ 0.40, respectively).The mean 6 SE soil bulk density and soil organic matter for all six sites were 0.66 6 0.06 g/cm 3 and 12.90% 6 1.32%, respectively.The mean 6 SE elevation for all six sites was 0.29 6 0.01 m NAVD88.The similarity in elevation at our sites implies that the sites were inundated at the same level during the freeze events.

DISCUSSION
Our analyses demonstrate that ontogeny (i.e., plant life stage) influences black mangrove resistance and resilience to winter climate extremes.Following freeze events in January 2014, we quantified A. germinans freeze damage and recovery as indicators of resistance and resilience, respectively.Our original hypothesis was that there would be a positive relationship between mangrove life stage development and the ability to withstand and recover from freezing temper-atures (i.e., the relationship illustrated in Fig. 2B).There are many examples from other ecosystems of mature plant individuals being more freeze tolerant than juveniles (e.g., Sakai and Larcher 1987, Yelenosky 1996, Boorse et al. 1998, Augspurger 2013).In this study, we expected that tall mangrove trees would be more resistant to winter climate extremes than short trees and seedlings.Our data indicate that tall A. germinans trees are indeed more freeze-tolerant than short trees.However, we found that freeze damage to seedlings was lower than for tall and short trees (Figs. 3 and 4).These findings indicate that the relationship between life stage and A. germinans resistance is best illustrated by the pattern in Fig. 2C.
The extremely low occurrence of freeze damage (,1%) among all mangrove seedlings provides a strong indication that seedlings which established prior to the freeze remained essentially undamaged (Fig. 3A-C).Multiple authors have noted cases where mangrove seedlings and propagules survived freeze events that had resulted in the mortality or damage of more mature individuals (Lugo and Patterson-Zucca 1977, Sherrod and McMillan 1985, Olmsted et al. 1993, Ross et al. 2009, Pickens and Hester 2011).Protection of seedlings, propagules and low mangrove individuals (i.e., ,50 cm height) may be due to the presence of a buffering layer of warmer and more humid air near the soil surface.Ross et al. (2009) noted that leaf mortality following a freeze event was lowest near the soil surface.Field observations imply that seedling protection and survival under cold temperatures may be higher beneath and near vegetation (i.e., a mangrove canopy layer or a tall salt marsh layer such as that produced by S. alterniflora; Lugo and Patterson-Zucca 1977, Sherrod and McMillan 1985, Olmsted et al. 1993, Ross et al. 2009, Pickens and Hester 2011, Guo et al. 2013), which can minimize nocturnal radiative cooling   et al. 2013).However, vertically-explicit air temperature and vegetation measurements are needed to confirm the presence of these feedbacks in mangrove forests.Experiments focused on the effect of vegetation density upon mangrove freeze survival and facilitation would be valuable.
Physiological differences in water transport and vulnerability to xylem cavitation could also be responsible for the life-stage dependent results observed in this study.From a physiological perspective, woody plants in tidal saline wetlands have much in common with those in terrestrial dryland ecosystems.In both ecosystems, freshwater limitation and freeze events can interact to result in bubble formation the xylem (i.e., xylem cavitation; xylem embolism), which can alter water transport and lead to plant damage and/or mortality (Pockman andSperry 1997, Stuart et al. 2007).In dryland terrestrial  v www.esajournals.orgecosystems, several studies have noted life-stage dependent vulnerability to cavitation (Sperry and Saliendra 1994, Mencuccini and Comstock 1997, but see Linton et al. 1998, Medeiros andPockman 2010).Similar physiology-focused studies conducted in black mangrove forests would help identify and differentiate between the physiological and microclimatic processes responsible for the life-stage dependent findings observed in this study.
In addition to the significant relationship between ontogeny and resistance, our results indicate that there is a positive relationship between ontogeny and recovery from freeze damage (i.e., resilience).Although recovery was high for both short and tall trees, recovery via basal resprouting was highest for the tall trees (Fig. 5).In comparison to short trees, tall trees likely have greater densities of meristem reserve buds and also have greater access to resources needed for resprout growth (e.g., greater access to carbohydrate reserves, freshwater, and nutrients).To our knowledge, this is the first study to quantify an effect of ontogeny upon the resilience and resistance of A. germinans trees.In an experiment focused on the temperature tolerance of early life stages, Pickens and Hester (2011) found no difference in the survival of A. germinans propagules and seedlings.Following a low-temperature event in south Florida, Ross et al. (2009) found a species-specific effect of mangrove height upon the survival of Rhizophora mangle L. and Laguncularia racemosa (L.) C.F. Gaertn.(the red and white mangrove, respectively).Whereas shorter R. mangle individuals had a higher probability of survival than taller individuals, the probability of L. racemosa survival increased with height (Ross et al. 2009).Higher mortality of tall R. mangle individuals may be explained by the lack of meristem reserve buds in older individuals of this species (Tomlinson 1986), which results in an inability to resprout following freeze damage (Olmsted et al. 1993).In contrast, L. racemosa and A. germinans both have high densities of meristem reserve buds, which can result in vigorous basal and stem resprouting following moderate freeze events (Fig. 5; Tomlinson 1986).
Aboveground A. germinans biomass at the study sites ranged from 9.6 to 19.8 Mg/ha (mean 6 SE: 12.1 6 1.6 Mg/ha).At our study area and other mangrove-marsh ecotones in the northern Gulf of Mexico, A. germinans individuals are typically freeze-stunted and shrub-like due in part to repeated cycles of freeze-induced aboveground damage followed by vigorous basal and/ or stem resprouting.Prior to the freeze of 2014, many (mean 6 SE: 60% 6 9%) of the tall tree individuals at our Damaged sites had produced new sprouts after losing a portion of their aboveground biomass due to a freeze in 2010 (Fig. 6).Some of these tall individuals in 2014 may have been in the short strata in 2010 and  In forests across the globe, microclimatic gradients greatly influence landscape-level patterns in ecosystem structure and function (Chen et al. 1999).For mangrove forests in the northern Gulf of Mexico and near their latitudinal limit, the literature is replete with field observations of spatially heterogeneous effects of freeze damage due to various interacting factors (e.g., landscape position relative to wind, water and vegetation; topography; tidal inundation history; genetic variability; positive feedbacks; e.g., Lugo and Patterson-Zucca 1977, Olmsted et al. 1993, Ross et al. 2009).These observations, in combination with the large differences in the extent of freeze damage between the two freeze damage treatments in this study, despite similarity of measured elevation and edaphic variables, highlight the complexity of making landscape-scale spatial predictions of mangrove forest damage and recovery in response to winter climate extremes.
In recent years, the identification of temperature-based thresholds that would lead to mangrove forest damage and/or mortality has become a productive and interesting research arena (e.g., Pickens and Hester 2011, Quisthoudt et al. 2012, Osland et al. 2013, Cavanaugh et al. 2014).At the landscape-scale, the minimum temperature thresholds that separate mangrove forests from salt marsh ecosystems are comparatively abrupt and more dramatic than those observed in many terrestrial systems (Osland et al. 2013, Cavanaugh et al. 2014).Targeted field and greenhouse-based experiments are needed to refine our understanding of mangrove freeze damage and mortality thresholds.There is a need for the identification of species-specific thresholds (Cavanaugh et al. 2015;Lovelock et al., in press) and for a better understanding of the physiological effects of interactions between low temperatures and other stressors including freshwater limitation and high salinity (Osland et al. 2014b).Our research has not directly quantified the relationships between specific temperature regimes and mangrove life stage-dependent damage and recovery, but rather the response of different mangrove life stages to environmental conditions associated with low temperatures.There is a need to identify life stage-specific thresholds because the temperature-based thresholds that would cause mortality or damage to seedlings, juveniles, and mature mangrove trees are likely different.Our results also highlight the need to account for the effects of vegetation-microclimatic feedbacks that would influence resistance and resilience.A more nuanced understanding of ontogeny-based thresholds and positive feedbacks would improve predictions of subtropical coastal wetland ecological responses to changing climatic conditions.
The positive relationship between ontogeny and resistance and resilience to winter climate extremes suggests that a decrease in the frequency and intensity of extreme minimum air temperatures could lead to a nonlinear increase in mangrove forest resistance and resilience, which would likely affect the rate of future mangrove expansion in response to climate change.In Fig. 7A, we present a conceptual model that illustrates a potential climate changeinduced positive feedback loop.In the model, a decrease in the frequency and intensity of winter air temperature extremes would lead to an increase in mangrove forest growth, structural development, and reproduction, which would lead to an increase in the resistance and resilience of mangrove forests to subsequent winter climate extremes (Fig. 7A).The increases in resistance and resilience would then lead to additional mangrove forest development, which would increase resistance and resilience (Fig. 7A).Due to this positive feedback loop, we expect that the relationship between changing winter climate extremes and mangrove forest resistance and resilience is probably nonlinear (see the relationship illustrated in Fig. 7B), and that small decreases in the frequency and intensity of winter climate extremes could increase mangrove forest resistance and resilience, which could accelerate the rate of mangrove expansion.
Although our data provide initial support for the presence of a positive feedback loop that would affect the rate of mangrove expansion into salt marsh, additional studies are needed to confirm and better quantify the processes responsible for the proposed positive feedback.In particular, more research is needed to test and quantify the linkages between life stage-dependent resistance, resilience, and ecological processes that are directly related to mangrove forest expansion (e.g., propagule production, dispersal distances, and recruitment).Although moderate freeze events like those that occurred in 2010 and 2014 do not result in widespread A. germinans mortality, we expect that freeze-induced reductions in aboveground biomass and height likely reduce: (1) the resources available for reproduction; and (2) the distances that propagules disperse from their source plants.Propagule dispersal distances are expected to be larger for taller and wider A. germinans individuals compared to short and compact individuals.Hence, within these poleward mangrove-marsh ecotones, we expect that there is a positive relationship between the absence of winter temperature extremes and propagule production rates and dispersal distances.Though logical, additional data is needed to confirm that these relationships are present.Additional studies are also needed to determine whether the life-stage dependent results observed here are also present during more extreme freeze events.
In addition to the climate-change focused implications, our results also have important implications for the restoration of coastal wetlands in the northern Gulf of Mexico.Avicennia germinans individuals are occasionally used to restore coastal wetland habitats.Our results indicate that planted A. germinans individuals are likely most vulnerable to freeze-damage in the short-tree stage (i.e., between 50-150 cm height) and that taller trees and seedlings are less vulnerable to freeze damage.Therefore, targeting a diversity of life stages during restoration in freeze-prone areas may minimize the potential for vegetation loss due to freeze damage.Also, salt marsh vegetation and/or densely planted mangroves could potentially be used to create a buffer layer near the soil surface, which could decrease radiative cooling and facilitate the establishment of mangrove transplants.Experiments are needed to test the validity of these restoration-focused hypotheses and recommendations.v www.esajournals.orgEcologists and natural resource managers are increasingly challenged to better understand, predict, and plan for the ecological effects of future changes in the frequency and intensity of climate extremes (Glick et al. 2011, Stein et al. 2014).Ecological responses are greatly influenced by stabilizing processes that affect the capacity of an ecosystem to withstand and recover from extreme climatic conditions (Smith 2011, Lloret et al. 2012, Hoover et al. 2014).Our results emphasize the importance of considering ontogeny (i.e., plant life stage) in predictions of ecological resistance and resilience to climate extremes.We found that taller mangrove trees/ shrubs were more resistant and resilient than short trees/shrubs.These findings could indicate the presence of a positive feedback loop between winter climate extremes and mangrove forest resistance and resilience.A future decrease in the frequency and intensity of extreme winter minimum temperatures could lead to an increase in mangrove forest resistance and resilience and accelerate the rate of poleward mangrove expansion at the expense of some salt marshes.

Fig. 1 .
Fig. 1.Photos of: (A-C) freeze damage to black mangrove (Avicennia germinans) individuals near their poleward limit in the northern Gulf of Mexico; and (D) a healthy individual without freeze damage.

Fig. 2 .
Fig. 2. Three potential relationships between vegetation life stage and the resistance and resilience of black mangrove (A.germinans) forests to winter climate extremes: (A) no effect of life stage upon resistance and resilience; (B) a positive relationship between life stage and resistance and resilience; and (C) a nonlinear relationship where resistance and resilience is highest during the seedling and tall tree life stages.

(
D'Odorico et al. 2013).Our data along with these observations indicate that a positive vegetation-microclimate feedback (sensu D'Odorico et al. 2013) may affect mangrove resistance and resilience to winter climate extremes.Similar vegetation-microclimate feedbacks have been observed in other cold-prone woodland-to-grassland transition zones across the globe (D'Odorico

Fig. 3 .
Fig. 3. Black mangrove (A.germinans) resistance to winter climate extremes.Freeze damage (A), percent height reduction (B), and actual height reduction (C) within freeze damage treatments and across vegetation strata.Different letters denote significant differences.For A, the vertical axis indicates freeze damage as the percentage of the plant that was damaged.

Fig. 6 .
Fig.6.Black mangrove (A.germinans) resilience to winter climate extremes.The percentage of tall tree stratum individuals that had visible damage remaining from the 2010 freeze event.These individuals resprouted and recovered from damage that resulted in a large loss of aboveground biomass.Different letters denote significant differences.

Fig. 7 .
Fig. 7.A hypothetical positive feedback loop that would be induced by a decrease in the intensity, duration and/or frequency of winter climate extremes (A), and the hypothetical effects of changes in winter climate extremes upon the resistance and resilience of black mangrove forests (B).

Table 1 .
Minimum air temperature and consecutive hours below freezing determined from weather station measurements recorded near the study area for 6-8 January and 28-30 January 2014.

Table 2 .
Effect of freeze damage treatment, vegetation strata, and their interaction upon freeze resistance metrics (i.e., percent freeze damage, percent height reduction, and actual height reduction).