Introduction
For many seasonally dry tropical forests (SDTF) in the Neotropics, hurricanes are important disturbance events. It has been documented how hurricanes affect forest structure, environmental conditions, and biotic interactions (Whigham et al., 1991; Zimmerman et al., 1994; Koptur et al., 2002; Van Bloem et al., 2003; Angulo-Sandoval et al., 2004; Bonilla-Moheno, 2012; Jimenez-Rodríguez et al., 2018; Stan and Sánchez-Azofeifa, 2019). Several authors have reported that tree growth rates of surviving trees increase following hurricanes (Whigham et al., 1991; Tanner and Bellingham, 2006; Tanner et al., 2014). This boost in plant growth likely occurs due to the extra and a seasonal precipitation that comes with hurricanes, which is especially important in water-limited ecosystems (Parker et al., 2018). In Jamaica, Tanner and Bellingham (2006) described that stem diameter growth rates increased during the Hurricane Gilbert decade, and that double growth rate occurred in the post-Gilbert decade. In Yucatán, Whigham et al. (1991) also recorded that, for the first year after Gilbert, relative diameter growth for most species was greater than average growth for pre-hurricane years. Hurricanes also change phenology in the forests and it has been reported that their damages increase leaf production, and decrease both herbivory (Koptur et al., 2002; Angulo-Sandoval et al., 2004) and food availability for vertebrates (Wunderle, 1999; Schaffner et al., 2012; Renton et al., 2018). Furthermore, fruiting patterns lag behind, diminish or even are completely suppressed after hurricanes in Puerto Rico (Wunderle, 1999) and in Mexico, in the Yucatán Peninsula (Schaffner et al., 2012) and in the Chamela Cuixmala Biosphere Reserve in the coast of Jalisco (Renton et al., 2018).
Tropical regions where hurricane landfalls used to be rare are experiencing an increment in extreme events. Global warming will likely decrease the frequency of all storms but the frequency of the most intense storms (hurricanes) is projected to increase in the Gulf of Mexico (Biasutti et al., 2012). On September 14th-18th 2010, the SDTF of central Veracruz, Mexico was struck by Hurricane Karl (category 3) causing major disasters throughout the region (Stewart, 2011). However, before Karl, the last recorded event was in 1932, when an unnamed hurricane, category 3 struck this region (Gómez, 2006).
The aims of this study were to examine the impact of Hurricane Karl on tree mortality and the way trees died, and compare diameter growth, and the phenological patterns observed in this SDTF before and after the hurricane event.
Materials and Methods
The study was conducted in a SDTF in central Veracruz, Mexico (19°16’ N, 96°29’ W, elevation: 97-117 m asl, area: 21 ha). Mean total annual precipitation is 932 mm and most rain falls between June and October; mean annual temperature is 25 °C. Dominant tree species are Bursera cinerea Engl., B. fagaroides (Kunth) Engl., B. simaruba (L.) Sarg., Calyptranthes schiedeana O. Berg, Heliocarpus donnellsmithii Rose, Stemmadenia pubescens Benth. and Tabebuia chrysantha (Jacq.) G. Nicholson (Williams-Linera and Lorea, 2009). Monthly precipitation and temperatures were obtained from the nearest meteorological station located at 7 km from the study site.
Along a < 1 m-width trail crossing the forest fragment, at 30 m from the forest edge, one tree near the trail was randomly chosen, and then the nearest neighbor of each consecutive tree was tagged until 165 trees belonging to 16 common species with 3-19 individuals per species were tagged (Table 1).
The status of each phenological phase was observed in each tree from the ground with the aid of binoculars (Swift, SeaHawk 7x50 No. 753). The percentage of the crown that was leafless or producing leaves, flowers or fruits was visually estimated, considering the following categories: 0 (0 %) 1 (1-25 %), 2 (26-50 %), 3 (51-75 %)and 4 (76-100 %). These values were averaged over individuals for each species by month and were used as a phenological index of intensity. Monthly phenological observations were made from October 2007 until September 2010 just a few days before Hurricane Karl struck (Williams-Linera and Álvarez-Aquino, 2016). In early November 2010, the monthly phenological observations were resumed for one more year (until October 2011).
Species | Acr | WD | No. | UR | BE | BR | Dead |
---|---|---|---|---|---|---|---|
Bignoniaceae | |||||||
Tabebuia chrysantha (Jacq.) G. Nicholson | Tab | 0.33 | 19 | 0 | 0 | 1** | 0 |
Bixaceae | |||||||
Cochlospermum vitifolium (Willd.) Spreng. | Coc | 0.48 | 14 | 0 | 2 | 0 | 2 |
Burseraceae | |||||||
Bursera cinerea Engl. | Buc | 0.35 | 11 | 0 | 1 | 0 | 1 |
Bursera simaruba (L.) Sarg. | Bus | 0.78 | 9 | 0 | 1 | 0 | 1 |
Convolvulaceae | |||||||
Ipomoea wolcottiana Rose | Ipo | 0.22 | 11 | 1 | 2 | 1** | 3 |
Fabaceae | |||||||
Gliricidia sepium (Jacq.) Kunth ex Walp. | Gli | 0.58 | 3 | 1 | 0 | 0 | 1 |
Leucaena lanceolata S. Watson | Leu | 0.53 | 7 | 0 | 0 | 0 | 0 |
Senna atomaria (L.) H. S. Irwin & Barneby | Sen | 0.28 | 4 | 2 | 0 | 0 | 2 |
Hernandiaceae | |||||||
Gyrocarpus jatrophifolius Domin | Gyr | 0.16 | 7 | 0 | 0 | 0 | 0 |
Malvaceae | |||||||
Ceiba aesculifolia (Kunth) Britt. & Baker f. | Cei | 0.44 | 10 | 0 | 0 | 0 | 0 |
Guazuma ulmifolia Lam. | Gua | 0.94 | 10 | 0 | 0 | 0 | 0 |
Heliocarpus donnellsmithii Rose | Hel | 0.91 | 16 | 7 | 3* | 1** | 8 |
Luehea candida (DC.) Mart. | Lue | 0.76 | 15 | 1 | 1** | 0 | 1 |
Meliaceae | |||||||
Trichilia trifolia L. | Tri | 0.80 | 8 | 0 | 0 | 0 | 0 |
Moraceae | |||||||
Maclura tinctoria (L.) D. Don ex Steud. | Mac | 1.04 | 10 | 4 | 1** | 1 | 5 |
Myrtaceae | |||||||
Calyptranthes schiedeana O. Berg | Cal | 0.80 | 11 | 1 | 0 | 0 | 1 |
Total | 165 | 17 | 11 | 4 | |||
Dead within a year | 17 | 7 | 1 | 25 |
*Two survived; **One survived; Acr = Acronym; WD = Wood density (g cm-3; Chave et al., 2006); No. = Number of tagged trees before Hurricane Karl, and number of trees that were uprooted (UR), bent (BE), broken (BR), and dead within the year following the hurricane.
Tree diameter was measured using a DBH tape (Forestry Suppliers, Inc.) in June-July 2007 to 2009, and then just after Hurricane Karl in 2010, June 2011 and October 2013. A month later, in October, all the trees that had been tagged for the phenological observations were surveyed, and classified as undamaged, uprooted, and bent or broken trunk.
Data analysis
The response variables (climate, diameter increment and phenology) were tested with analysis of deviance (ANODE) to determine if they differed among years. ANODE is similar to analysis of variance, but the test does not assume normal distribution. Differences in diameter increment before and after the hurricane, and differences in annual precipitation and maximum and minimum temperatures among years were analyzed using generalized linear models (GLM) with normal distribution and log link function. Differences among years for each phenophase intensity for all species together were tested using GLM; phenophase proportions were fitted assuming binomial distribution for the response variable and the logit link function. The model used for diameter increment included species, year (pre- and post-hurricane) and the interaction term. The model for climate and phenology included year only. When significant differences were detected orthogonal planned contrasts were used to determine whether the post-hurricane year was different from each of the three previous years. Data were analyzed using JMP v10.0.0 (SAS Institute, 2012).
Results
Climate
For the hurricane year, rainfall was higher (1 440 mm) than the average (932 mm) for the region (Figure 1). However, precipitation (Χ 2 = 1.84, P = 0.60), maximum temperature (Χ 2 = 1.81, P = 0.61) and minimum temperature (Χ 2 = 0.23, P = 0.97) for the hurricane year (2010) did not differ statistically from the climate variables recorded during pre-hurricane (2007-2009) and the post-hurricane (2011) year (Figure 1).
Tree damage and mortality
Hurricane Karl defoliated all the trees, but damage varied among species. Thirty-two tagged individuals were damaged, however, some of these died while others resprouted after three to twelve months following the hurricane (Table 1). More trees died uprooted (17), than from being bent (7) or broken (1). Five species with bent and broken trunks sprouted, whereas all uprooted trees died (Table 1). The highest mortality occurred in tree species with relatively high wood density (Heliocarpus donnellsmithii, Maclura tinctoria (L.) D. Don ex Steud.) (Table 1), but other species with low wood density also showed intermediate number of damaged trees (e.g, Ipomea wolcottiana Rose, Senna atomaria (L.) H.S. Irwin & Barneby). Other late successional species with high wood density that were partially damaged survived to the hurricane impact, such as Leucaena lanceolata S. Watson and Trichilia trifolia L., or Tabebuia chrysantha that had one individual with a broken trunk that sprouted (Table 1).
Diameter growth
Overall, there were significant differences between pre- (0.68 + 0.09 cm yr-1) and post-hurricane (0.79 + 0.10 cm yr-1) growth rate (Χ 2 = 5.12, P = 0.024) and among species (Χ 2 = 66.36, P < 0.0001, Figure 2). However, the interaction between species and hurricane effect was not significant (Χ 2 = 17.16, P = 0.25). The studied species diameter growth rates before and after the hurricane varied from 0.33 to 1.63 cm yr-1 and 0.30 to 1.89 cm yr-1, respectively (Figure 2). The species with the highest diametric growth was Luehea candida (DC.) Mart., followed by Guazuma ulmifolia Lam. and Senna atomaria.
Phenology
At the community level, leafless status, leafing, and flowering display cyclic annual patterns (Figure 3a, b, c). Phenology within the 3 pre-hurricane years is presented to give an idea of the variation without hurricane disturbance (Figure 3), so it can be observed if the post-hurricane data fall within the year-to-year variation. The forest was leafless during the dry season, and all trees were with leaves in July-September (Figure 3a, b). Leaf fall, leafing and flowering were statistically similar between the pre- and post-hurricane years, but the peak of flowering shifted after the hurricane to April-May (Figure 3c). Fruit production was much lower during the post-hurricane year (Figure 3d).
At the species level, the vegetative phenological patterns remained similar before and after the hurricane. In contrast, intensity of the reproductive phenophases was affected by the hurricane (Figures 3 and 4). For most species (12), flowering intensity decreased in the post-hurricane year, and three species did not flower at all (Calyptranthes schiedeana, Luehea candida and Tabebuia chrysantha), while for two others (Bursera simaruba and Guazuma ulmifolia) (Figure 4a) flowering intensity was higher after the hurricane. Fruiting during the year following the hurricane was null in four species, and lower for most of the species, or the fruiting peak was shifted (Leucaena lanceolata and Senna atomaria); only one species (Gliricidia sepium (Jacq.) Kunth ex Walp.) showed higher fruiting after the hurricane (Figure 4b).
Discussion
Fifteen percent of the trees (25 out of the 165 in %) died within the post-hurricane year; a value at the upper level of the 7-14 % range of values reported after hurricanes at stand level for Neotropical forests (e.g., Brokaw and Walker, 1991; Zimmermann et al., 1994; Bonilla-Moheno, 2012; Stan and Sánchez-Azofeifa, 2019). In the studied forest, most trees died uprooted, followed by bent and broken trunks, but tree species mortality was not related to wood density. In other dry forests where hurricanes are frequent, few trees died uprooted but broken trees experienced the highest mortality (Whigham et al., 1991; Zimmerman et al., 1994; Bonilla-Moheno, 2012). Some authors have reported that early successional species are more susceptible to stem breakage than late successional species (Zimmermann et al., 1994; Ross et al., 2001). In general, tree damage occurs in taller trees as they could be more vulnerable to mechanical damage (Jiménez-Rodríguez et al., 2018), thus larger trees suffered more uprooting and snapping than small ones (Bonilla-Moheno, 2012; García and Siliceo-Cantero, 2019).
The high mortality and degree of uprooting might be related to the fact that Karl was the first hurricane that arrived to this region in more than 60 years (Gómez, 2006). It is possible that this first hurricane in a long time struck the most vulnerable or exposed trees. Bonilla-Moheno (2012) has reported that the impacts of subsequent hurricanes were minor compared to the first one, suggesting that dry forests may have a high degree of recovery depending on the level of subsequent hurricanes.
Tree diameter growth after hurricane Karl was higher than the average growth recorded before the hurricane. Several authors have reported increased diameter growth following hurricanes (Tanner and Bellingham, 2006; Tanner et al., 2014). However, in this study, the high precipitation during the two months preceding the hurricane could also had a positive effect on tree diameter growth. Apparently, change in growth rate depends on the tree species; significant differences among tree species were found. Tanner et al. (2014) also indicated a variation among species, and that the increase was much greater in some light-demanding species (e.g., Alchornea latifolia Sw.) than in others. Whigham et al. (1991) stated that high relative growth rates may reflect the greater availability of nutrients, as large amounts of P, K, Ca, Mg, and Mn were deposited on the forest floor by the storm.
The phenology of the forest was studied for three years before the hurricane (Williams-Linera and Álvarez-Aquino, 2016). Leafing patterns remained similar before and after the hurricane, but at community level, the defoliation caused by Hurricane Karl affected the seasonality timing of leaf fall and the community remained leafless for a month. Flowering was particularly affected because, before the hurricane, significant peaks were apparent at all levels, but after the hurricane there was not seasonality. Shifts in flowering peaks have been documented in Southern Florida where Hurricane Andrew delayed flowering in a tropical shrub for two months, although heavy defoliation might also be related to high flowering (Pascarella, 1998).
Fruit phenology was most affected the year following the hurricane. In Puerto Rico, Wunderle (1999) reported the effect of Hurricane Hugo on fruiting phenology as the lowest fruit production and asynchrony, with trees taking four months for the number of fruiting species to reach a level equivalent to the normal yearly low. Fruit production in the Yucatán Peninsula, was four times higher during the dry season before than after hurricanes (Schaffner et al., 2012). In the Chamela-Cuixmala Biosphere Reserve flowers and fruits were reduced during the first-year after hurricanes Jova and Patricia, forest recovered phenological cycles by the second-year post-Hurricane (Renton et al., 2018).
The impact of hurricanes on phenology may have a strong effect on trophic relationships. Some studies have reported that the forest compensates for damage by increasing leaf production and there are lower levels of herbivory, likely because the disturbance eliminates herbivore insect populations allowing plants to recover (Koptur et al., 2002; Angulo-Sandoval et al., 2004). Also, the shortage of fruits affects the behavior of other herbivores, such as spider monkeys (Schaffner et al., 2012) and other threatened species like parrots (Wunderle, 1999; Renton et al., 2018). Long-term effects of hurricanes can cause shifts in species composition (Tanner et al., 2014); in the studied forest composition can be particularly affected since two species that did not flower or fruited are Calyptranthes schiedeana and Tabebuia chrysantha), the first is endemic to Veracruz; and the second is threatened, according to the Mexican red list (NOM-059-SEMARNAT) (Semarnat, 2010).
Conclusions
Due to hurricane damage, tree mortality was high, but many trees recovered from damage, likely because this forest has not been greatly affected by recent hurricanes and due to the high resilience of this forest type. More research is needed to better understand how the predicted increase in intense hurricanes could influence forest structure and phenology in SDTF in the region. The results suggest that although pre- and post-hurricane vegetative phenological patterns were similar, hurricane Karl clearly had a negative effect on the intensity of the reproductive phenology, which in turn, may shift species composition and have an impact on the trophic relationships and functioning of the forest community.