Fuel beds variation of coastal tropical freshwater forested wetlands in three disturbance regimes at La Encrucijada, Biosphere Reserve, Mexico

Background: Tropical freshwater forested wetlands in coastal regions are rapidly disappearing, one of the causes is forest res. This is caused by high accumulation of fuel beds that can vary in origin and type. Therefore, the objective of this study was to characterize the fuel beds in tropical freshwater forested wetlands with three different level of disturbance at El Castaño, La Encrucijada Biosphere Reserve. Methods: Seventeen sampling units were used to described the strates of forest fuel beds (canopy, sub-canopy and understory) in both the vertical and horizontal stratum. Quantity and quality of dead (fallen woody material, surface litterfall and fermented litterfall) fuels were characterized using the planars intersections technique. Results: A total of eight tree species, two shrubs, ve lianas and two herbaceous species were found in tropical freshwater forested wetlands. The vertical structure concentrates the highest proportion of trees between 2 and 12 m high, including the rst two dened height classes. The horizontal structure denotes a higher percentage of trees with a normal diameter between 2.5 to 7.5 cm (61.4%) of the total. The sites none disturbance presented the highest arboreal density (2,686 ind. ha − 1 ), however the highest basal area was found in the sites with medium disturbance (39.41 m 2 ha − 1 ). The richness and diversity of species shows that the Fisher, Margalef, Shannon and Simpson α indices were higher in the sites undisturbed, while the Berger-Parker index shows greater dominance in the sites high disturbance. Pachira aquatica Aubl. was the species with the highest importance value index, and contributes the most to the fuel beds. The average accumulation of dead fuel beds was higher in sites with high disturbance (222.18 ± 33.62 t ha − 1 ), with the largest accumulations of woody fuels occurring in the 1 h, 10 h and 1000 h classes, the latter in a state of fermentation. Conclusions: It’s important to consider the probability of occurrence of surface res is high. In these tropical freshwater forested wetlands, independent of disturbance, underground res have the same probability and the intensity will depend on the depth of the organic matter layer. This study contributes dene re-prone areas in these ecosystems. The results are of great importance to design re prevention strategies.

Conclusions: It's important to consider the probability of occurrence of surface res is high. In these tropical freshwater forested wetlands, independent of disturbance, underground res have the same probability and the intensity will depend on the depth of the organic matter layer. This study contributes de ne re-prone areas in these ecosystems. The results are of great importance to design re prevention strategies.

Background
Tropical freshwater forested wetlands (TFFW) are regularly found close to mangrove forests within tropical regions occupying swampy terrain, gently sloping and near the river bank, where freshwater in uence is greatest ( ood pulses) (Ewel, 2010;Infante et al. 2011a; Torres et al. 2018). These plant communities group a large number of tree, shrub and climbing species (Moreno and Infante, 2016). Its complexity is very high, since in addition to providing numerous environmental services, they signi cantly favor the natural processes of nutrient supply (Moreno and Infante, 2009). In hydrological and terrestrial cycles, TFFWs function as a natural lter for pollutants and as an important refuge for wildlife (Infante et al. 2011a). They also constitute important carbon (C) sinks and even store greater amounts of C in the soil than mangroves (Moreno and Infante, 2016), but unlike these, TFFW are a type of tropical arboreal vegetation that has been little studied.
TFFW are located in freshwater-in uenced areas where one or two species commonly dominate (Infante et al. 2011a;Silva et al. 2012). A species of great importance in TFFW is Pachira aquatica Aubl. (water zapote), which by its physiological and morphological characteristics could belong to the mangrove group, since they have the particularity of being oating trees that form a false forest oor with organic matter produced and retained (Infante Mata et al. 2011). TFFW dominated by P. aquatica are pure communities that only share their space with certain species tolerant to oods (Barrios-Calderón, 2019). In Mexico and Mesoamerica, the only ood forests dominated by P. aquatica are those distributed in Huimanguillo, Tabasco (Ascencio, 1994), Ciénaga del Fuerte, Laguna Chica, El Apompal in Veracruz (Infante et al. 2011b) and La Encrucijada in Chiapas (Infante et al. 2011;Rincón, 2014; Barrios-Calderón, 2015). In this last, the TFFW of P. aquatica come to be associated with mangrove species (Moreno and Infante, 2016), however human in uence has caused the gradual disappearance of these ecosystems.
One of the disturbance agents that promote the loss of plant communities in TTFW are forest res, either by natural or anthropogenic causes. These play an important role in the dynamics of vegetation and the change of land use (Rodríguez y Fulé, 2003;FAO, 2007;Jardel et al. 2009). Therefore, these coastal wetlands are considered fragile ecosystems, which are threatened by the frequency and magnitude of re (Page et al. 2009;Garcia et al. 2014). Although TFFW remain ooded for most of the year, they are highly productive and store woody forest fuels (i.e. branches, twigs, logs, fallen trees) and litterfall accumulated. This latter is at different levels of decomposition on the soil surface (Sales et al. 2020). It's precisely in the dry seasons that these ecosystems can lead to combustion processes when the water level fades, the soil dries and all accumulated organic material is available to burn (Goldammer, 1999;Kaal et al. 2011). This explains why this ecosystem has presented res on a recurring basis, largely managed by the quality and quantity of woody fuel and litterfall.
Forest fuels are the source of energy, in combination with topography, climate and a source of ignition that control the magnitude and re spread (Scott et al. 2014). Because res are an important environmental driver for ecosystem processes and biodiversity losses, studies to characterize and estimate forest fuels are of great relevance to re management and prediction (Rodríguez et al. 2002). The characterization of fuel beds integrates the analysis of the quantity and quality of forest fuel, taking into account the spatial distribution both vertical and horizontal (Dentoni and Muñoz, 2001).
A rst step towards this is the description of the fuel beds resulting from the strati cation of live fuels (Flowers, 2001;Sandberg et al. 2001) and the quanti cation of dead fuels (Villers, 2006). Additional information to be considered is the frequency of res, due to anthropogenic activities such as illegal wood extraction, the use of re for poaching and the proximity of population center (Turner et al. 2001), among others that have caused the fragmentation of these forest ecosystems.
Particularly in areas such as the La Encrucijada Biosphere Reserve (LEBRE) Chiapas, these disturbance agents are a threat to ecosystems such as TFFW. CONANP (2018) and CONAFOR (2018)  . However, few studies have focused on exploring tropical freshwater forested wetlands with different levels of disturbance. Therefore, the objective of this research was to characterize the structure and composition of living and dead fuels, from a study approach to fuel beds of TFFW dominated by Pachira aquatica in El Castaño (LEBRE), with different levels of disturbance (high, medium, none). The following hypotheses were raised: (i) Species diversity will be greater in undisturbed sites with respect to sites of medium and high disturbance. In addition, (ii) P. aquatica will have higher importance value and major dominance in tropical freshwater forested wetlands with high disturbance, this as a result of decreased in species richness in these conditions. (iii) Despite having less strati cation and mixing of species, the accumulation of dead fuels is greater in sites with high disturbance.

Study area
The natural protected area La Encrucijada (LEBRE) is also a RAMSAR site (1996), located in the southeast Paci c coast of Mexico. The surface area of the LEBRE is 144, 868 ha, 24% (36,216 ha) are classi ed as core areas, while 75.1% (108,651 ha) are classi ed as buffer areas (Diario O cial de la Federación, 1995). TFFW of P. aquatica are found surrounding mangroves, freshwater marshes and other coastal ecosystems (Fig. 1), these ecosystems are connected through a complex hydrology network. In this wetland system, P. aquatica is far from the salt water inlets of the estuaries and the main source of water is freshwater from the San Nicolas River. Climate is tropical warm-humid with summer rains Am(w), mean annual temperature is 28 °C, and annual precipitation ranges from 1300 mm to 3000 mm (INE, 1999). The dry period goes from February to May, during this period water level is below the soil surface (Rincón, 2014).
Field study Based on hydroperiod information by Rincón (2014), the eld campaigns for this study were planned from February to Abril in 2017 and 2018. Following a strati ed randomized experimental designal design, 17 circular sampling units (SU) of 600 m 2 (radius = 13.8 m) were established in TFFW with three different disturbance level (Fig. 1). Disturbance levels was de ned based on the following criteria: 1) canopy openness, 2) distance from roads, 3) human activities, 4) illegal extraction wood, 5) hurricane/wind impact, 6) res, and 7) forest cover ( Table 1). Based on this information and eld validation sites were categorized as None disturbance (n = 7), Medium disturbance (n = 5), and High disturbance (n = 5) (Fig. 2).

Statistical analysis
The software Paleontological Statistics (Past 3.26) (Hammer, 2016) was used to analyze diversity taking into consideration the disturbance level, for all fuel components (trees, shrubs, herbs, and lianas) studied here. Species richness was estimated with the Margalef index (DMg). At the structural level, Fisher's alpha indexes (α) and Shannon-Wiener (H') were used to assess the proportional abundance and the Berger-Parker (d) and Simpson (D) indexes to calcule species dominance for each disturbance level. Difference among disturbance levels was evaluated with an ANOVA and Tukey-Kramer at 95% con dence for comparison of means, using Statistical Analysis System (2018) (SAS, version 14). To de ne possible differences between dead fuel load (woody and litterfall) was also applied an ANOVA and mean comparison test by Tukey-Kramer, with a signi cance level of 0.05 using the SAS statistical package.

Structure and composition of fuel beds
The tree stratum registered eight species in the TFFW studied at the LEBRE, were P. aquatica was dominant. Vertical structure of fuel beds In sites with high disturbance there was a larger number of trees with heights between 2 and 7 m (75%). While, in sites with none disturbance, 65% of trees belonged to the same as above category, and in medium disturbance only 49%. Thus, this was the dominant height class in the three levels of disturbance evaluated ( Table 3).The percentage for the 7 to 12 m height class varies in sites where the high disturbance has only 20% of the trees in this category, in sites with medium disturbance 25%, and in sites with none disturbance the percentage increases to 32%. Sites with medium disturbance have the highest percentage for classes 12-17 m (14.48%), 17.1-22 m (6.43%), and > 22 m (6.43%), with respect to none disturbance and high disturbance sites (Table 3). So much so that the sum of the percentages of these three classes corresponds to 24%, while in the other two disturbance conditions (high and null), the height classes greater than 12 m only represent 3.29 and 3.16%, respectively.  Table 4). The species composition decreases with disturbance, in this way, sites with none disturbance have more species (n = 12) than sites with high disturbance (n = 5). 33% of the species are present across all sites independent from the disturbance level. In order of importance, these species were P. aquatica, Zygia conzattii, Laguncularia racemosa, and Rhizophora mangle while Hampea macrocarpa is present only in sites with high and none disturbance. P. aquatica was registered at all three disturbance levels and it becomes more dominant as disturbance increases, which is re ected in its forest value index. Holdridge index shows that as the number of species decreases, complexity is less (Table 4). Legend: Absolute (D A ) and relative (D R ) densities, absolute (F A ) and relative (F R ) frequencies, basal area (BA), Importance value index (IVI), forest value index (FVI) and Holdrige complexity index (HCI) Species richness, was highest for none disturbance sites (19), followed by medium disturbance (14), and nally high disturbance (9) sites. Alpha Fisher index, as well as Margalef, and Shannon and Simpson were all higher in none disturbed sites. While, the Berger-Parker index was higher for P. aquatica in high disturbed sites (Table 5).

Dead fuels
The woody fuels stratum of fuel beds in TFFW different among classes 1 h, 10 h, and 1000 h (rotten). The highest fuel load was presented in sites with high disturbance, while in the remaining categories there are no signi cant differences. For its part, the litterfall fuels (SL and FL) did not show differences in the three disturbance conditions studied (  (Fig. 4).  (Rincón, 2014). Given that in this region, P. aquatica is a very adaptive species, so much so that it grows in association with some mangrove species ( . This pressure on the ecosystem could lead to a successional stage, where the tree stratum further reduces its cover percentage, while shrubs and herbs may be more abundant (Kellogg et al. 2003). From this perspective, the characterization of the structure and composition of the forest fuels bed is of great interest, so they should be studied over time. Precisely, an important eld of application of studies related to the structural composition of forests and jungles is the re ecology, mainly in the TFFW. These ecosystems have been impacted by a regime of irresponsible use of re (Chen, 2006) and re dynamics have also contributed to changes in its vegetation (Randerson et  1% of the canopy strata goes from 6 to 17 m, while smaller trees represent only 6%. However, this is not the case in high disturbance areas were the most abundant height (76%) ranges from 2 to 7 m, thereby verifying that the disturbance has an effect on the vertical structure, which is evident with the supression of the trees.
Medium disturbance in the TFFW of the Paci c, results in P. aquatica growing higher than in sites none disturbance. This suggests that low re frequency could be bene cial for the ecosystem (Keane y Karau, 2010; Bowman et al. 2011), acting like a renewal mechanism. In high disturbance areas, P. aquatica is more dominant however does not have great high. Therefore, the level of disturbance will in uence vertical and horizontal structure, and in sites with more density of P. aquatica the forest will not be as tall. The most common diameter class of live fuels on the horizontal plane goes from 2.5 to 7.5 cm, independent from disturbance. Similar observations have been reported in other sites of the LEBRE e.j. El Jicaro (Rincón, 2014). Similarly, in TFFW from the Gulf of Mexico in Veracruz, the most frequent class is from 3 to 13 cm (Infante et al. 2011a). This is specially observed in El Castaño in high disturbance sites. The density of trees in this study has a similar range to that reported in sites at the Gulf of Mexico (Infante, 2011). However, in other LEBRE sites with a higher state of conservation such as Brisas del Hueyate, the density can go up to 3,310 ind. ha − 1 and the IVI of P. aquatica (285.28%) is higher (Barrios-Calderón, 2015).
It is important to note that in the present study, the density of trees that make up the live fuels stratum is higher in less disturbed sites, while woody and litter fuels show an opposite pattern, having more accumulation in sites with greater disturbance. However, the average value of dead fuels (woody and litterfall) reveals that tropical freshwater forested wetlands from the Paci c coast can have more fuel than other forested ecosystems in the tropics ( rm (14.18 ± 9.33 t ha − 1 ), these same authors report loads similar to those obtained at no disturbance TFFW sites in the present investigation. Thus, the TFFW with medium and high disturbance represent a greater potential to present res, under favorable temperature and weather conditions. Mainly high disturbance sites that would represent hot-spots for res with worst implication for the ecosystem (Flores, 2017).
Litterfall fuels, especially the super cial litterfall (SL) from all studied sites is higher than the ones reported by Rodríguez et al. (2011) in Quintana Roo, Campeche and Yucatan, with loads up to 17.2 t ha − 1 . Litterfall productivity in TFFW from the Gulf of Mexico (Veracruz) goes from 9.3 ± 0.5 t ha-1 (Apompal) to 14.9 ± 1.0 t ha − 1 (Chica) (Infante et al. 2011b), which are similar to the observed in this study. In both sites Gulf of Mexico and the southeast Paci c (LEBRE, El Castaño), litterfall fuels is the result of the productivity of trees and lianas. As Souza et al (2019) point out, high litterfall accumulation is determined by climatic factors that affect the vegetative phenology of tree species. In this way, the amount of litterfall fuels in the soil of the study area doubles the amount accumulated in TFFW of Veracruz in the Gulf of México, therefore the accumulation rate is higher in the in the southeast Paci c. This in terms of re potential also represents a greater danger for the El Castaño area, due to the high accumulation of these litterfall fuels (Westcott et al. 2014;Varner et al. 2015). However, the litterfall fuels from the fermented layer are lower in comparison with sited in the Yucatan Peninsula (Rodríguez et al. 2011), with maximum loads of 53.89 t ha − 1 . This layer is especially important because the combustion is low, but the energy is high, therefore underground res are more severe than the surface res (Neri et al. 2009). Observations of this fuel layer in this study shows no signi cant differences in the three conditions evaluated, therefore independent of disturbance level TFFW are vulnerable to underground res in the LEBRE.
Some authors such as Rodriguez (2014) point out that the amount of woody fuels and litterfall decreases with frequent res. However, despite the fact that in the las 10 years, two res of low intensity have been recorded (CONANP, 2018), not much of a decrease of down dead fuels was observed in sites with high and medium disturbance. This is due to disturbance caused by wood extraction and opening of new roads, according to information provided by forest rangers.
Regarding the average load of dead fuels (woody and litterfall), Barrios-Calderón et al. (2018) came to obtain an average load of 225.06 t ha − 1 at other study sites in the TFFW of El Castaño. These loads are very similar to obtained in the sites of disturbance high at the present investigation, which represent the highest average accumulation of dead fuels. In another study carried out at the TFFW in Calakmul, Campeche, Contreras et al.
(2006) report total loads ranging from 43.15 to 154.5 t ha − 1 , which are within the range obtained for none disturbance and medium disturbance sites. However, the total amount of fuels dead is higher than other TFFW dominated by P. aquatica dominated in the Yucatan peninsula, Mexico (16.46 t ha − 1 ) obtained by Reyes and Coli (2009), which are the lowest loads that could be considered in relation to those obtained in this work.
Recently, Flores et al. (2018) determined that the load or biomass of these dead fuels contains between 47.5% of C (for litter) to 67% of C (in fuels of 1000 h rm). In this way, the total amount of C released when the TFFW are burned contributes a greater amount of greenhouse gases, compared to other ecosystems that have lower amounts of dead fuels (necromass), since the ood forests are important C sinks. Thus, is important role of these coastal ecosystem to the C cycle at local and regional scale.
Fuel beds at high disturbance sites have higher ignition potential and C emissions through ne fuels (1 h y 10 h) and litterfall, while the medium fuels (100 h) propagate the re and coarse fuels (1000 h) are related to its intensity. Furthermore, dead fuel available in any level of forest disturbance, will be related to propagation potential and movement of the re on stairway fuels (Chávez et al. 2016). The latter is the result of the vertical continuity of the fuels arranged in these ecosystems, which goes from herbs, shrubs, lianas and trees connecting one stratum with the next. In general, sites with higher disturbance will have higher re potential, these sites will require the implementation of strategies to prevent and mitigate re. Even more so if you have a wide dominance of P. aquatica, a type of softwood that complies with the established by Haruk et al (2020) when developed in humid areas with high temperatures and annual precipitation greater than 1837 mm, and that favor the decomposition rates of this type of woody material. In this way the complex of woody fuels forms the necromass, that is, the dead wood that is available and susceptible to ignite. Finally, the probabilities of underground re are similar in the three TFFW conditions, because there are no differences in the depth of the litterfall layer and amount of organic material.
The information generated in this study related to the characterization of fuel beds, is a starting point for further studies which will allow predict the ignition, propagation, and impact of res in these ecosystems. In general, the potential for res in the TFFW conditions evaluated is more evident in areas with high disturbance that require the implementation of preventive and e cient. This to counteract the possible presence of res that could spread to adjacent ecosystems (mangroves, freshwater marshes and palmares). All information referring to the study of fuel beds and their composition, constitutes a starting point at the regional level to predict or diagnose the start, spread and impact of res in these ecosystems, in the face of anthropogenic disturbances that have fragmented the connectivity and ecosystem functionality of these coastal wetlands.

Conclusions
According to the results obtained in the present study, a greater diversity of species was found in undisturbed sites, while in medium and high disturbance sites richness decreases. Although the Pachira aquatica species has an importance value higher that makes it a dominant species under conditions of disturbance, its vertical and horizontal structure does not show optimal development. Large amounts of woody fuels give a higher probability of occurrence of high to severe category res. In this way, the composition of fuel beds in high and middle disturbance forests have greater accumulation of forest fuels. The proximity to population center, the opening of roads or access routes, the illegal extraction of trees, shrubs and the historical record of res have become causes of the high accumulation of fuels forest. Therefore, it will be important to consider that the probability of occurrence of surface res increase. However, the three conditions of TFFW evaluated have the same possibility of presenting underground res, where the intensity will depend on the depth of the organic material (fermented litterfall) that is distributed equally in the three evaluated disturbance conditions. TFFW's of coastal zones are important C sink, that during a re event could release considerable amounts of greenhouse gases to the atmosphere. The information from this study helps to de ne and prioritize areas that need different management strategies in these ecosystems for which there are not many investigation that refer to the topic addressed. However, it will be important in future studies to increase the sampling points to achieve greater precision in the results. The datasets used and/or analysed during the current study are available from the corresponding autor on reasonable request. Ethics approval and consent to participate Not applicable.

Consent for publication
Not applicable.

Competing interests
The authors declare that they have no competing interests.
Funding Page 17/26 The author(s) received no nancial support for the research, authorship, and/or publication of this article.   Tropical freshwater forested wetlands of P. aquatica: a) None disturbance, b) Medium disturbance, and c) High disturbance Page   Relative histogram by diameter class for each disturbance level.