Professional Documents
Culture Documents
Aquatic Botany
journal homepage: www.elsevier.com/locate/aquabot
A R T I C LE I N FO A B S T R A C T
Keywords: Assessing factors affecting growth and mortality of mangrove seedlings, the critical transition between propagule
Mortality dispersal and recruitment to the sapling state, is crucial to understand and predict the fate of mangrove forests in
Mangrove recruitment our changing climate. This work aimed to understand seedling’s responses to contrasting temperature and
El Niño-Southern Oscillation precipitation regimes and analyze consequences for the persistence of these ecosystems under expected climate
Estuaries
change scenarios. Growth rate and mortality of seedlings were monitored monthly in an array of mangrove
Biogeographic Chocó
forests located along a tidal and salinity gradient in the central Pacific coast of Colombia, one of the rainiest
places on Earth. Seedlings were monitored from January to December 2016, coinciding with a major El Niño
(EN) episode and the abrupt transition to La Niña (LN) conditions. Seedling growth rates were generally low and
observed spatial patterns generally mirrored interspecific differences in the tolerance to stress induced by
salinity and inundation levels. Seedlings in basin forests showed higher growth rates than those in riverine and
fringe forests. Mortality was not different among species and unexpectedly low, considering the rates reported in
the literature and the supposedly stressful conditions associated with the EN-LN cycle. According to our analyses
the magnitude of local anomalies in air temperature and precipitation throughout the EN-LN cycle can be
stronger than those expected locally for 2071–2100 in relation to global climate change. Current and expected
shifts in precipitation regimes seem the main macroclimatic drivers of ecological changes in mangrove forests
thriving in the Pacific coast of Colombia.
⁎
Corresponding author.
E-mail addresses: jose.m.riascos@correounivalle.edu.co (J.M. Riascos), jaime.cantera@correounivalle.edu.co (J.R. Cantera), juan.blanco@udea.edu.co (J.F. Blanco-Libreros).
https://doi.org/10.1016/j.aquabot.2018.03.002
Received 28 September 2017; Received in revised form 22 January 2018; Accepted 16 March 2018
Available online 27 March 2018
0304-3770/ © 2018 Elsevier B.V. All rights reserved.
J.M. Riascos et al. Aquatic Botany 147 (2018) 34–42
climate anomalies associated with ENSO affect regional rainfall, river 2.2. Sampling design
discharge and sediment transport to the ocean through teleconnections
(Vörösmarty et al., 1996; Restrepo and Kjerfve, 2000; Riascos, 2006, Five ecological types of mangrove forests (sensu Twilley et al., 1998)
Blanco, 2009). Such hydrological variability is probably the most im- distributed along a salinity gradient through the estuarine system were
portant forcing function for tropical coastal marine ecosystems, parti- selected for this study: Basin, Riverine I and II, Fringe I and II (see
cularly estuarine ecosystems, in Latin America, particularly along the Supplementary material 1 for details). These forests were monitored
Pacific coast (Kjerfve et al., 2001; Riascos et al., 2008). Therefore, cli- monthly between December 2015 and December 2016. Three re-
matic anomalies associated with El Niño and La Niña phases of ENSO plicated places were randomly chosen within each mangrove type, with
seem to be important, particularly for reproduction, dispersion and at least 200 m distance among them (Fig. 1). In each place two 78.5 m2
recruitment in mangroves located in the Tropical Eastern Pacific (TEP; plots were established: a low-level water plot that was called “external”
Ecuador, Colombia, Panamá and Costa Rica). and a high-level water plot called “internal”, so 30 plots were mon-
Despite the past warnings on the disappearance of mangroves due to itored. Therefore, three sources of variability were considered in the
growing anthropogenic pressures combined with climate change (Duke sampling design: forest type, position in the intertidal and time, thus
et al., 2007), Alongi (2015) predicted that climate change will generally conforming a 5 × 2 × 12 sampling design. There was, however, an
have mild effects, with some negative effects for mangrove forest along important source of variability that could not be accounted for by the
arid coasts, subsiding river deltas and small islands. He based his pre- sampling design: the mangrove species composition, which differed
diction using the most recent global climatological forecasts of sea among forest types, positions and plots. Therefore, spatial and temporal
surface temperature, oceanic pH, CO2 concentrations, sea level rise and variability was analyzed separately for each species (Mora oleifera,
precipitation and salinity by the Intergovernmental Panel on Climate Pterocarpus officinalis, Rhizophora spp., Pelliciera rhizophorae and Avi-
Change toward the end of the century. However, little is known about cennia germinans).
the effects of sub-decadal variability associated with ENSO, which can In each plot, between five and ten mangrove seedlings were ran-
be stronger than the expected long-term changes forecasted by the IPCC domly selected, identified and tagged with a consecutive number using
for tropical regions, given the strong influence on freshwater runoff. adhesive plastic tape. Growth rate and mortality of mangrove seedlings
Recently, Banerjee et al. (2016) reported on the decline in the above- was monthly monitored. Mean growth rate (mm day−1) was estimated
ground biomass of a mangrove species in India related to decadal-scale by averaging differences in shoot length (from the top of the hypocotyl
salinity alterations. to the top of the plant) from the previous month in each plot. Any
In 2015–2016 the TEP experienced one of the strongest El Niño on tagged seedlings which died were replaced by haphazardly selecting
record; it was popularly depicted as the “Super El Niño” (Zhai et al., and tagging individuals from those remaining in the plot. There are not
2016; Ren et al., 2017). Globally, precipitation patterns changed, sea large gaps in the canopy of these forest that may influence growth and
surface temperature exceeded historical records for 12 consecutive mortality. However, structure, age and development do vary among
months and sea levels were driven ∼7 millimeters higher as warming forest. Therefore, the Holdridge Complexity Index (HCI) was used as an
ocean water expanded. estimator of spatial structure and development of the mangrove forest.
Here we describe the dynamics of growth and mortality of man- The index was computed as:
grove seedlings from Colombia’s Pacific pluvial and meso-tidal forests,
sdbh
considered among the most luxuriant forests in the world (West, 1956), HCI =
1000
during the latest El Niño warming episode. The aims of this study were
(i) to analyze spatial-temporal differences in growth and mortality of where (s) is the number of mangrove species, (d) the stand density, (b)
mangrove species thriving in the study area, (ii) to estimate anomalies the basal area, and (h) mean height in each plot (see Blanco et al., 2001
in precipitation and temperature in Bahía Málaga, their relationship for details and a discussion of this index). The ratio of sapling density
with El Niño and La Niña episodes and compare the magnitude of these (RSD) i.e., the density of Trees 2.5–10 cm diameter at breast height with
anomalies with the expected variations under climate change scenarios respect to the total tree density was estimated as an indicator of forest
by 2071–2100; and (iii) to discuss the relevance of these findings for the age or successional stage (Blanco et al., 2001). See supplementary
long-term persistence of Colombia’s Pacific mangrove forest. material 1 for estimations of these indices.
35
J.M. Riascos et al. Aquatic Botany 147 (2018) 34–42
Fig. 1. Map of the study area in Bahía Málaga, Pacific coast of Colombia. The map shows the different mangrove forest types (see Appendix A) monitored and the
meteorological stations La Misión and Malaguita. Two plots (internal and external, see materials and methods) were monitored in each of the 15 sampling points. The
distribution of mangroves was plotted from the global distribution of mangrove forest derived from earth observation satellite imagery version 1.3 (Giri et al., 2011).
36
J.M. Riascos et al. Aquatic Botany 147 (2018) 34–42
scenarios were built from the outputs of the Coupled Model conducted. Tukey HSD test were applied to assess differences between
Intercomparison Project Phase 5 produced by the World Climate species. As in the previous section, a multiple logistic regression was
Research Programme. IDEAM provided multi-model and multi-scenario used to determine if type of forest, position and time (categorical fac-
temperature and precipitation projections: 16 global climatic models tors) and HCI and RSD (continuous predictors) are good predictors of
were selected and assembled, based on their relevance in a regional seedlings mortality. The model used the binomial distribution for the
context and projections were averaged for the distinct Representative response variable and Logit as the link function. The model was tested
Concentration Pathways describing different 21st century pathways of using the likelihood ratio test, as mentioned before. Changes in tem-
anthropogenic greenhouse emission scenarios (RCP2.6, RCP4.5, poral patters were analyzed in light of local fluctuations of temperature
RCP6.0, RCP8.5). and precipitation during January-December 2016.
Moreover, data on the Oceanic Niño Index (three month running
mean of ERSST.v4 SST anomalies in the Niño 3.4 region) were gathered 2.6. Local vs. regional ENSO anomalies
from the Climate Prediction Center of the National Oceanic and
Atmospheric Administration, United States to describe anomalies re- To calculate local anomalies in temperature and precipitation
lated to the last El Niño-La Niña episodes. Finally, in situ data during El Niño-La Niña, we followed a similar approach to that used by
(1969–2016) on monthly precipitation (mm) and monthly mean tem- the Climate Prediction Center (NOAA, United States) to calculate the
perature (°C) were gathered from IDEAM for two meteorological sta- Ocean Niño Index1: we estimated the three-month running means of
tions located close to the study area (Fig. 1). For temperature, data for temperature/precipitation “anomalies”, defined as the difference be-
2016 were incomplete. Therefore, temperature data based on satellite tween the observed mean temperature/precipitation and the corre-
observations were obtained from the NASA Goddard Institute for Space sponding monthly mean of a reference 10-year base period updated
Studies (Global Modeling and Assimilation Office, 2015) and a com- every 5 years. Thereafter, we used a simple regression model (least
parison and correlation of in situ data versus satellite-based data was square fit) to test for the relationship (teleconnection) between tem-
used to estimate missing temperature data for 2016 (see Supplementary perature anomalies described by the ONI and local anomalies observed
material 2). in the study area. The significance of the relationship (the regression
slope) was tested by an analysis of variance.
2.5. Data analyses
3. Results
2.5.1. Between species comparisons
An analysis of covariance (ANCOVA) was used to assess the effect of 3.1. Between-species comparisons of growth and mortality
a categorical factor (mangrove species) on mean growth rate, control-
ling for the effects of two continuous covariates (HCI and RSD). Species The ANCOVA analysis showed that there were significant differ-
was treated as fixed factor. Mean growth rate was Log-transformed to ences in growth rate of seedlings among mangrove species (Table 1).
meet assumptions of the model concerning distribution or errors. Tukey Post hoc tests indicated that M. oleifera and P. officinalis grew sig-
HSD test were conducted to assess specific between species differences. nificantly faster than Rhizophora spp, P. rhizophorae and A. germinans
A multiple logistic regression running under the Generalized Linear (Fig. 2). On the other hand, the spatial structure and development of the
Model approach implemented in “R” (R Development Core Team, 2008) forest and its demographic condition, as depicted by the HCI and RSD
was used to determine if a categorical factor (species) and two con- respectively, were not significant as predictors of seedling growth rate
tinuous independent variables (HCI and RSD) are good predictors of (Table 1).
seedling mortality. The model used the binomial distribution for the Mortality of mangrove seedlings was low, Only 51 seedlings died
binary response variable (death: 1: alive: 0) and the Logit as the link during the year. The mean mortality was 2.7% (4.1 seedlings per
function. To test the model, we first fitted the full model involving all month). The Wald test showed that, overall, species was not significant
the predictors and based on those results we fit a reduced model. Wald (χ2(4) = 6.1; p = 0.19) as a predictor of the observed differences in
tests were used to assess the significance of categorical factors. There- mortality. Among continuous predictors, only HCI was significant
after, we tested the significance of the final model using the likelihood (Table 2). Therefore, data were fitted to a reduced model that only
ratio test. included this predictor. This suggest that in the array of mangrove
forest evaluated here the mortality of seedlings was higher in spatially
2.5.2. Species-specific spatial and temporal patterns in growth and mortality complex and developed forests.
To assess spatial and temporal differences in seedlinǵs growth rate,
an ANCOVA analysis was performed, using three categorical (fixed) 3.2. Species-specific spatial and temporal patterns
factors (type of forest, position in the intertidal and time). All the
higher-order interactive effects of factors were included in the model 3.2.1. Growth rate
when possible (i.e. some species did not occur in all type of forest/ There were significant differences in the growth rate of Rhizophora
positions/times). The HCI and RSD were included in the model as spp. with space and time (Table 3). Growth rate was significantly faster
covariates or continuous regressors. When a species occurred only in in Basin mangrove forest than in Riverine II and Fringe I forest
one forest type, covariates were not included and standard ANOVA was (Fig. 3A). On the other hand, growth rate was low in January, peaked in
February and decreased thereafter, with a small peak in July (Fig. 3B).
Table 1 Neither the interaction term between type of forest and time nor the
Results of the ANCOVA analysis on the effects of categorical and continuous covariates had a significant effect on the growth rate of Rhizophora sp.
predictors on the mean growth rate of mangrove seedlings. Significant factors Avicennia germinans seedlings were restricted to external plots in Fringe
(α = 0.05) are printed in bold.
II forest. There were significant temporal differences in growth rate
Effect SS DF MS F P (Table 3): seedlings grew about one order of magnitude faster in March
in comparison with the rate observed between May and November
Intercept 0.757 1 0.757 49.318 < 0.0001
Species 3.267 4 0.817 53.221 < 0.0001
(Fig. 4A). P. rhizophorae occurred only in internal plots of Fringe I and
HCI 0.006 1 0.006 0.412 0.5211 Fringe II mangrove forest. They grew significantly faster (almost twice)
RSD 0.017 1 0.017 1.090 0.2969
Error 6.829 445 0.015
1
http://www.cpc.noaa.gov/products/analysis_monitoring/ensostuff/ensoyears.shtml.
37
J.M. Riascos et al. Aquatic Botany 147 (2018) 34–42
Table 3
Results of the ANCOVA/ANOVA assessing the effects of categorical and con-
tinuous factors on the growth rate of seedlings of distinct mangrove species.
Significant factors (α = 0.05) are printed in bold.
Species/effect SS DF MS F P
Rhizophora spp.
Intercept 0.027 1 0.027 4.755 0.0302
Month 0.183 11 0.017 2.888 0.0014
Forest type 0.061 4 0.015 2.658 0.0336
Month × Forest type 0.266 44 0.006 1.051 0.3939
HCI 0.000 1 0.000 0.009 0.9247
RSD 0.020 1 0.020 3.481 0.0633
Error 1.346 234 0.006
Mora oleifera
Intercept 4.458 1 4.458 63.051 0.0000
Position 0.018 1 0.018 0.248 0.6290
Month 1.047 11 0.095 1.346 0.3237
Position × Month 0.496 11 0.045 0.638 0.7640
Error 0.707 10 0.071
Pelliciera rhizophorae
Intercept 0.365 1 0.365 62.299 < 0.0001
Forest type 0.028 1 0.028 4.803 0.0332
Fig. 2. Variability of mean growth rate of seedlings of different mangrove Month 0.038 11 0.003 0.590 0.8277
species (January to December 2016). Levels of the factors not connected with Forest type × Month 0.019 11 0.002 0.295 0.9837
the same lettering are significantly different after Tukey HSD tests (p < 0.05). Error 0.287 49 0.006
Avicennia germinans
Intercept 0.332 1 0.332 56.627 < 0.0001
Table 2 Month 0.299 11 0.027 4.630 0.0011
Coefficient estimates for the parameters of full and reduced logistic regression Error 0.129 22 0.006
models on the effects of species and continuous predictors on the mortality of
Pterocarpus officinalis
mangrove seedlings. AIC: Akaike information criterion. Significant factors
Intercept 3.988 1 3.988 184.766 0.0000
(α = 0.05) are printed in bold. Month 0.624 11 0.057 2.627 0.0117
Effect Estimate SE z Pr(> |z|) Error 0.928 43 0.022
Full model
Intercept −4.387 1.257 −3.489 0.0001
M. oleifera 1.390 1.166 1.192 0.2333
P. rhizophorae −0.461 1.180 −0.391 0.6957
Rhizophora spp. 0.421 1.043 0.404 0.6863
P. officinalis 0.382 1.173 0.326 0.7445
Holdridge Complexity Index 0.255 0.122 2.088 0.0367
Ratio of Staples Density 0.346 0.749 0.462 0.6441
AIC 459.160
Reduced model
Intercept −3.863 0.194 −19.931 < 0.0001
Holdridge Complexity Index 0.311 0.092 3.377 0.0007
AIC 455.320
3.2.2. Mortality
As there were no species-specific differences in mortality, the lo-
Fig. 3. Variability in the growth rate of Rhizophora spp. seedlings (A) among
gistic regression model was run again including forest type, position on forest types; (B) among months. Levels of the factors not connected with the
the intertidal and month along with HCI and RSD for all species. Results same lettering are significantly different after Tukey HSD tests (p < 0.05).
of this modeling showed that there were no significant spatial or tem-
poral effects on the observed mortality (Table 4). In contrast, HCI
showed again a significant effect on mortality, confirming our previous
analysis. This reduced model (only HCI as an explanatory variable)
38
J.M. Riascos et al. Aquatic Botany 147 (2018) 34–42
Fig. 4. Variability in the growth rate of mangrove seedlings. (A) Temporal variability in the growth rate of Avicennia germinans; (B) Differences in the growth rate of
Pelliciera rhizophora between forest types. C. Variability in the growth rate of Pterocarpus officinalis among months. Levels of the factors not connected with the same
lettering are significantly different after Tukey HSD tests (p < 0.05).
Table 4
Coefficient estimates for the parameters of full and reduced logistic regression
models on the effects of categorical and continuous predictors on the mortality
of mangrove seedlings. AIC: Akaike information criterion. Significant factors
(α = 0.05) are printed in bold.
Effect Estimate Std. Error z value Pr(> |z|)
Full model
Intercept −3.702 0.763 −4.850 < 0.0001
Riverine I −0.189 0.586 −0.323 0.7465
Riverine II 0.008 0.445 0.020 0.9843
Fringe I −0.490 0.513 −0.955 0.3396
Fringe II −1.309 0.768 −1.703 0.0885
Interior −0.165 0.322 −0.514 0.6071
January 0.231 0.575 0.402 0.6874
February −0.021 0.594 −0.036 0.9711
March −0.434 0.660 −0.658 0.5107
May −0.406 0.660 −0.614 0.5390
June −0.150 0.622 −0.242 0.8091
July −1.066 0.828 −1.287 0.1982
August −1.758 1.089 −1.614 0.1065
September −0.404 0.660 −0.613 0.5401
October −1.762 1.089 −1.617 0.1058
November −0.402 0.660 −0.610 0.5421
December 0.192 0.574 0.334 0.7380
HCI 0.317 0.131 2.420 0.0155
RSD 0.898 1.070 0.840 0.4009
AIC 469.050
Reduced model
Intercept −3.863 0.194 −19.931 < 0.0001
HCI 0.311 0.092 3.337 0.0007
AIC 455.300
39
J.M. Riascos et al. Aquatic Botany 147 (2018) 34–42
anomalies in the study area during El Niño and La Niña episodes Ball et al., 1997; Koch, 1997; López-Hoffman et al., 2007), differences
(Fig. 6A and B). During strong episodes (Fig. 6C), monthly mean tem- in structure and demographic stage of the forests included in this study
perature in the study area can depart 1.55 °C above and 0.66 °C below were not as strong as to influence seedling growth.
mean historical temperature during El Niño and La Niña, respectively. Mortality of seedlings, which did not differ among species, was low.
In turn, precipitation can decrease up to 52% and increase up to 29% in The accumulated number of dead seedlings (all species) reached ∼35%
December in comparison with the historical mean during El Niño and during a year. This was unexpected, considering the particular en-
La Niña episodes, respectively (Fig. 6D). Therefore, the magnitude of vironmental conditions during 2016 and the available field data on
local anomalies observed during El Niño-La Niña cycle (∼2.2 °C in annual rates of mortality of conspecifics elsewhere (Supplementary
temperature and ∼80% in precipitation) can be stronger than those material 3). In contrast with growth, mortality of seedlings was sig-
expected for 2071–2100 as affected by global climate change. nificantly affected by the HCI; mortality was higher in more developed
and complex forests. This is in line with several studies showing an
4. Discussion influence of the canopy structure and light incidence on seedlings
mortality (e.g., Rabinowitz, 1978a; Sousa et al., 2003; López-Hoffman
In the context of expected effects of climate change on estuarine et al., 2007). Moreover, the fact that HCI affected seedlings mortality
systems the interest has focused on sea-level rise. It will likely have the but not growth suggests that these processes are not necessarily cou-
largest impact on coastal wetlands because inundation and salinity re- pled, as observed by López-Hoffman et al. (2007) for A. germinans. The
gimes are key drivers of ecological and physiological processes in these lack of differences in mortality found in this study contrasts with
systems (Koch and Snedaker, 1997; Mitsch and Gosselink, 2007; Gilman findings by Rabinowitz (1978b) in Panama, Ellison and Farnsworth
et al., 2008; Alongi, 2015; Armitage et al., 2015). Our findings on (1993) in Belize and Davis (1940) in Florida. They found that seedlings
spatial and temporal structure of growth and mortality of seedlings of Rhizophora survive for more than a year, while those of Avicennia,
largely mirror species-specific tolerances and responses to these para- Laguncularia, and Pelliciera show lower survival rates and cohorts
meters (Ball et al., 1997; Krauss et al., 2008) and the distribution of usually die within 6 months.
different forest types along the salinity and tidal inundation gradient. Seedlings of R. mangle were by far the more abundant and occurred
While fringe forests in Málaga Bay are exposed to frequent tidal in- in all forest types and intertidal positions. As such, spatial patterns were
undation and salinities ranging between 19 and 28 ups, basin forests clearer and growth rate was higher in forest with less exposure to tidal
close to small rivers are rarely inundated and salinity ranges between flooding and salinity stress (Basin and Riverine I). This is in line with
1.3 and 10 (Cantera et al., 1998; Lazarus and Cantera, 2007) previous findings on decreased stomatal conductance and chlorophyll
The faster growth rate observed for seedlings of M. oleifera and P. fluorescence, measures of photosynthetic activity, with increasing
officinalis in comparison with those of Rhizophora spp, P. rhizophorae salinity stress (Biber 2006). Seedlings of A. germinans in contrast were
and A. germinans seems to be related with the spatial distribution of restricted to external plots Fringe II forests, probably reflecting the
these species. M. oleifera and P. officinalis are restricted to basin man- preference for relatively higher salinity observed in Avicenniaceae in
grove forests, thus being less exposed to tidal flooding with brackish comparison with Rhizophoraceae (Downton, 1982; Hutchings and
water and wave action. Life history traits in these forests seem to favor Saenger, 1987; Lovelock and Feller, 2003). Occurring in internal
seedlings having thin, fast-growing stems. A compilation of growth (landward) plots, seedlings of P. rhizophorae grew faster in Fringe I than
rates of conspecific seedlings from reported literature (Supplementary in Fringe II forests. As the main difference between these forests was the
material 3) seems to support this view; growth rate of seedlings gen- RSD, describing the forest age or successional stage (see Appendix A), it
erally decreases from species occupying seaward edges (e.g. Rhizophora, seems that growing conditions for seedlings of P. rhizophorae may be
A. germinans) to species thriving in more elevated areas or landward more demanding in forests with a higher density of sapling.
edges of mangrove forests (e.g. M. oleifera, P. officinalis). While light The growth rate of M. oleifera did not show spatial or temporal
levels and canopy structure strongly influence seedling growth (e.g., differences. This result should be taken with caution, because it could
40
J.M. Riascos et al. Aquatic Botany 147 (2018) 34–42
be related with its restricted distribution (Basin forests) and low thriving in one of the rainiest places on Earth.
abundance in our dataset. Although P. officinalis has been considered a
plant representing the transition between mangrove and terrestrial Author contributions
forests (Francis and Lowe, 2000; Medina et al., 2007), it was abundant
only in external (seaward) plots in Basin forests, among seedlings of M. JMR, JFB, JRC conceived and designed the study. JMR carried out
oleifera and Rhizophora spp. and its growth rate displayed strong tem- the samplings, analyzed data and wrote drafts. JRC provided logistic
poral differences. support. JFB, JRC reviewed drafts and provided insights. All authors
Giving the limited temporal scope of our data, it is difficult to revised the final draft of the current manuscript.
conclusively assess growth and mortality responses of seedlings to
changes in air temperature and precipitation. From the species showing Funding
significant temporal differences in growth rate (Figs. 3 and 4), only P.
officinalis showed a significant correlation with these environmental Funding for this research was provided by Universidad del Valle.
parameters. The beginning of our study coincided with the inflexion This research was supported by a scholarship to JMR by the
point of strongest thermal anomalies associated to the El Niño episode Departamento Administrativo de Ciencia, Tecnología e Innovación
2015–2016, one of the strongest on records. By August 2016, conditions (Colciencias) − Programa “Tiempo de Volver”.
had turned to La Niña episode, which lasted until the end of the year.
These shifting large-scale oceanographic conditions were reflected in Conflict of interest
anomalously high temperature for most of the year and a deficit of
precipitation during the first half of the year, which turned in excess in None.
the second half. Hence, sublethal or lethal effects may be expected,
particularly considering high mortality rates reported in literature. As Acknowledgements
this was not the case, we could infer that the seedling stage of man-
groves in this region can cope with local anomalous conditions asso- We thank Juan Carlos Mejia for providing maps and plotted man-
ciated with ENSO extremes. Therefore, a larger dataset involving ENSO- grove coverage from satellite images processing. Thanks to the Consejo
neutral periods is needed to objectively assess responses of growth rate comunitario de Comunidad Negra de La Plata Bahia Málaga for their
and mortality. It is noteworthy that no new seedlings recruited in the hospitality and for allowing access to their territory. Thanks to
plots being monitored in this study. Willington Aguirre, Diego Morales, Charlotte Hopfe for their help in
Our results show that, regarding temperature and precipitation, the field work.
magnitude of local anomalies related to the ENSO cycle can be stronger
than those expected for the last part of the century in relation to global Appendix A. Supplementary data
climate change. As ENSO has been changing in strength and frequency
and this trend could continue under global warming (Timmermann Supplementary data associated with this article can be found, in the
et al., 1999; Yeh et al., 2009; Wang et al., 2017) it brings up the online version, at https://doi.org/10.1016/j.aquabot.2018.03.002.
question: will mangrove ecosystems be resilient or resistant to these
mixed short-term and long-term climatic anomalies? Despite their References
limited spatial and temporal extent, our results tend to support the
assertion that mangrove forests will either experience little change in Alongi, D.M., 2015. The impact of climate change on mangrove forests. Curr. Clim.
areas where precipitation is forecasted to increase (Alongi, 2015), at Change Rep. 1, 30–39. http://dx.doi.org/10.1007/s40641-015-0002-x.
Armitage, A.R., Highfield, W.E., Brody, S.E., Louchouarn, P., 2015. The contribution of
least in hyper-humid mangrove forests. Regarding temperature and mangrove expansion to salt marsh loss on the Texas Gulf Coast. PLoS One 10,
precipitation, this prediction is based on the following facts or as- e0125404. http://dx.doi.org/10.1371/journal.pone.0125404.
sumptions: (i) for most mangrove species, rates of leaf photosynthesis Ball, M.C., Sobrado, M.A., 2002. Ecophysiology of mangroves: challenges in linking
physiological processes with patterns in forest structure. In: Press, M.C., Scholes, J.D.,
peak at about 30 °C and leaf CO2 assimilation decline at 30–35 °C (Ball Bark, M.G. (Eds.), Physiological Plant Ecology. Blackwell, Oxford, pp. 331–346.
and Sobrado, 2002); temperature increases alone are likely to result in Ball, M.C., Cochrane, M.J., Rawson, H.M., 1997. Growth and water use of the mangroves
faster growth, reproduction, photosynthesis, and respiration, changes in Rhizophora apiculata and R: stylosa in response to salinity and humidity under am-
bient and elevated concentrations of atmospheric CO2. Plant Cell Environ. 20,
community composition, diversity and an expansion of latitudinal 1158–1166. http://dx.doi.org/10.1046/j.1365-3040.1997.d01-144.x.
limits (Tittensor et al., 2010) and (iii) precipitation is largely seen as a Banerjee, K., Gatti, R.C., Mitra, A., 2016. Climate change-induced salinity variation im-
co-factor. Our historical dataset (1970–2016) on air temperature in the pacts on a stenoecious mangrove species in the Indian Sundarbans. Ambio 46,
492–499. http://dx.doi.org/10.1007/s13280-016-0839-9.
study area, shows that the annual temperature averages 25.6 °C, with
Biber, P.D., 2006. Measuring the effects of salinity stress in the red mangrove, Rhizophora
little seasonal variation (25.1–26.1 °C) and that the highest monthly mangle L. Afr. J. Agric. Res. 1, 1–4.
mean on records was 28 °C (registered in March 1998, during El Niño Blanco, J.F., Bejarano, A.C., Lasso, J., Cantera, J.R., 2001. A new look at computation of
1997–1998). This suggests that temperature conditions are still in the the complexity index in mangroves: do disturbed forests have clues to analyze canopy
height patchiness? Wetl. Ecol. Manage. 9, 91–101. http://dx.doi.org/10.1023/
range that are supposedly favorable for mangrove growth, reproduc- A:1011115220126.
tion, photosynthesis and respiration. However, this seems a gross gen- Blanco, J.F., 2009. The hydroclimatology of Gorgona Island: seasonal and ENSO-related
eralization made from scant evidence. According to Osland et al. patters. Actual. Biol. 31, 111–121.
Cantera, J., Neira, R., Ricaurte, C., 1998. Bioerosión en la costa Pacífica colombiana: un
(2016), the vulnerability assessments of mangrove forest to climate estudio de la biodiversidad, la ecología y el impacto de los animales destructores de
change is too focused on their ability to adjust to sea-level rise, but our acantilados rocosos sobre el hombre. Fondo FEN, Colombia (89 pp).
knowledge on temperature and precipitation regimes as macroclimatic Cantera, J.R., Londoño-Cruz, E., Mejía-Ladino, L., Herrera-Orozco, L., Satizabal, C., Uribe-
Castañeda, N., 2013. Environmental issues of a marine protected area in a tectonic
drivers of structure and function of mangroves is scant, and lags behind estuary in the tropical eastern pacific: uramba (Malaga bay Colombia): context.
that of terrestrial forests. Osland et al. (2017) identified 14 range limit biodiversity. threats and challenges. J. Water Resour. Protect. 5 (11), 1037–1047.
regions where temperature and precipitation regimes can cause abrupt http://dx.doi.org/10.4236/jwarp.2013.511109.
Cerón-Souza, I., Rivera-Ocasio, E., Medina, E., Jiménez, J.A., McMillan, W.O.,
ecological transitions; they found that in the context of climate change,
Bermingham, E., 2010. Hybridization and introgression in new world red mangroves,
changes in rainfall patterns are most likely to drive changes in structure Rhizophora (Rhizophoraceae). Am. J. Bot. 97, 945–957. http://dx.doi.org/10.3732/
and function of mangrove forest in western South America. Therefore, ajb.0900172.
Clarke, P.J., 2004. Effects of experimental canopy gaps on mangrove recruitment: lack of
the results of this study provides baseline data to understand how
habitat partitioning may explain stand dominance. J. Ecol. 92, 203–213. http://dx.
present and projected changes in temperature and precipitation regimes doi.org/10.1111/j.0022-0477.2004.00861.x.
are likely to affect the most critical life stage of mangrove forest Davis, J.H., 1940. The ecology and geologic role of mangroves in Florida. Carnegie Inst.
41
J.M. Riascos et al. Aquatic Botany 147 (2018) 34–42
42