1 Introduction

Dunes consist of sandy intertidal sediments that, once dry, are transported by wind action toward the continent (Cordazzo et al. 2006). They are characterized by specific environmental conditions, such as strong winds, high light and temperature intensity, and nutrient-poor soil (Britto et al. 1993; García-Mora et al. 2001; Ley Vega de Seoane et al. 2007). The variety of topographic features, and its relation with different environmental gradients, determines the coexistence of different types of plant communities in the space occupied by the dunes (Wilson and Sykes 1999; Frederiksen et al. 2006). Therefore, coastal dunes vegetation is not homogeneous, but arranged in patches, which leads to the formation of complex vegetation mosaics throughout this environment worldwide (Doing 1985; Cordazzo et al. 2006). In addition to their ecological importance, coastal dunes guarantee protection for coastal areas against sea waves, wind, floods, and erosion (Kiehl and Isermann 2007). However, coastal ecosystems, such as dunes, are increasingly threatened globally (Kutiel et al. 2000), mainly due to human actions led by the rapid progress of the urban occupation process (Rosa and Cordazzo 2007; Carboni et al. 2009).

Due to the sensitivity of coastal areas and the accelerated pressure and anthropogenic disturbance, studies on vegetation ecology are important for understanding and conserving these environments (Medeiros et al. 2007). In this respect, phenology stands out as an integrative environmental science and can be used in different approaches related to conservation practices (Morellato et al. 2016). Phenological studies are fundamental for understanding the influence of climate on vegetation and for the analysis of plant-animal interactions, such as pollination and seed dispersal (Williams-Linera and Meave 2002; Morellato 2003; Franco et al. 2014). These interactions are important for understanding the structure and dynamics of the plant community, essential in the elaboration of projects aimed at the conservation of biodiversity (Fournier 1976; Wunderlee 1997; Yamamoto et al. 2007; Dutra et al. 2009). However, studies on the phenological behavior of coastal dune vegetation are still scarce worldwide (Rodríguez-Gallego and Navarro 2015). Brazil follows the global trend, presenting few studies that specifically address the phenology of the coastal dune plant community. In studies carried out with dune vegetation at higher latitudes, temperature was identified as the main factor that influences reproductive phenophases, mainly flowering (Cordazzo e Seeliger 1988; Castellani et al. 1999; Morellato et al. 2013). In areas of coastal dunes in low latitudes and with little accentuated climatic seasonality, the regulation of phenophases may occur differently.

This paper presents the phenodynamic of the woody plant community on coastal dunes in eastern Amazon, as well as its leaf habits and pollination and seed dispersal syndromes, aiming to answer the question: is there any relation between climate factors (photoperiod, precipitation, and temperature) and the phenology of woody species from dunes in the eastern Amazon coast? In addition, in order to characterize the analyzed vegetation and to support discussions, pollination and dispersion syndromes and leaf habits of the species were determined.

2 Materials and methods

Field of study –

This study was carried out on a coastal dune area (02°29′7″ S 44°15′59″ W) located in Maranhão state, northeast Brazil (Fig. 1). This area is close to the urban perimeter and consists of primary and secondary dunes. The primary dunes found are narrow, with an almost completely flat area, probably as a result of disturbances such as the construction of the access road to the beach. On the other hand, the secondary dunes present a steep relief, reaching a height of 35 m (Silva et al. 2016).

Fig. 1
figure 1

Location selected for the phenological research in Maranhão state, northeast Brazil

Full size image

The dune vegetation shows signs of anthropic impacts and is constantly affected by fire, indiscriminate cutting of vegetation, and tourists and beachgoers trampling (Silva et al. 2016). The vegetation extends over a total of about 4 km, with woody vegetation distributed throughout the whole area, whose species display, in general, smaller size and caliber compared to the same species occurring in forest areas (Araújo et al. 2016; Silva et al. 2016).

The area has a tropical savanna climate, type Aw, according to the Köppen climate classification (1948), with higher rainfall trends during the period from January to July, and the dry period between August and December (Alvares et al. 2013). According to historical data (1988 to 2018), temperatures ranged from 22 to 33 °C throughout the years, and the average annual rainfall was 1.977 mm (Fig. 2a) (INMET 2018). Data on total precipitation, average minimum and maximum temperatures (Fig.  2b) (Source: http://www.inmet.gov.br/portal/), and photoperiod (Fig.  2c) (Source: http://www.sci.fi/~benefon/sol.html) were collected monthly, from August 2017 to July 2018, to verify possible associations with phenological data.

Fig. 2
figure 2

a Historical climate data (1988—2018); b Climatic data of the study period (2017–2018): rain (filled area), maximum temperature (continuous line), average temperature (intermediate dashed line), and minimum temperature (dashed line underneath); c Photoperiod of the year of study. Information from an area of coastal dunes located in Maranhão state, northeastern Brazil

Full size image

Phenological data –

To evaluate the phenological data, 19 woody species with a diameter at soil height (DSH) > 3 cm and whole crown visible were selected, including a species of cactus, based on the floristic list by Silva et al. (2016). The names of the species and their respective authors were consulted on the “Flora do Brasil 2020” website; and updates the names of species were carried out. We marked 2 to 15 individuals per species. During 12 consecutive months (August 2017 to July 2018), information on leaf abscission and production, flowering, and fruiting was collected fortnightly for all marked individuals.

The phenophases were estimated using the Fournier intensity index (1974), with values distributed on a semi-quantitative scale, and five categories grouped from 0 to 4, in 25% intervals, for vegetative and reproductive phenophases: leaf fall, new leaves, flowering, and fruiting. The phenophases were defined as (1) leaf abscission, as the proportion of the crown with no leaves, (2) new leaves, emerging or expanding, (3) flowering, the presence of flower buds and anthesis, and (4) fruiting, the presence of immature and ripe fruits. The activity percentage of each phenophase was determined, based on records of the presence or absence of the phenophases in each individual and the percentage of species that presented phenophases over the months observed.

The leaf habit of each species was defined, which were classified as evergreen, brevideciduous, and deciduous, according to Lenza and Kleink (2006), with some adjustments. Species with a crown coverage percentage always above 50% were recognized as evergreen. Species that remain without or with few senescent leaves for less than four weeks were classified as brevideciduous.

Pollination and seed dispersal syndromes –

Pollination and seed dispersal syndromes were determined by consulting specialized literature. For pollination, the following syndromes were considered: anemophily–wind, phalenophily–moths, melittophily–bees, and chiropterophily–bats. For seed dispersal analysis, species were categorized into the following groups: anemochoric–dispersed by the wind, autochoric–explosive or barocoric dispersal, and zoochoric–dispersed by animals.

Statistical analysis –

Directional statistics were applied to determine seasonal trends in the community. In order to do this, the first step was the conversion of the observation dates into angles, where a whole year corresponds to 360° (each day equals 0.986°). Then, the average start and peak dates of all individuals were determined to obtain averages by species and for the community, for each phenophases. The determination of the average data is the average of the degrees, subsequently converted into days, in which each individual manifested the phenophase for the first time (start) and the date when each individual had the highest intensity value for a given phenophase (peak).

The Rayleigh Z test was applied to test whether the circular distribution of the data collected presents seasonality. Significant probability values indicate that the concentration intensity around the average angle, represented by the vector “r,” is considered as a measure of seasonality degree, that ranges from zero (phenological activity evenly distributed throughout the year) to one (phenological activity concentrated around a single date or time of the year) (Morellato et al. 2000; 2010). All circular analyses were performed using the ORIANA 4.1 software (Kovach 2011).

Multiple regression analyses were performed to assess the possible relationships between the percentages of intensity of each phenophase (dependent variable) and the monthly climatic data of precipitation, average maximum and minimum temperatures and photoperiod (independent variables). These analyses were also carried out to verify whether there are relations between a phenological dynamics of the observed month and the climatic factors of the previous month. These analyses will be displayed only significant relationships and their beta values (the effect of the independent variable on the dependent variable) (Zar 2010).

3 Results

The research was conducted on 114 individuals, distributed in 19 species and 16 families (Table 1). Throughout the research, 10 individuals were lost as result of cutting and/or fire, which are some of the anthropic impacts considered as frequent in that area. These individuals were disregarded for statistical analysis, in view of the unfeasibility of observations throughout the study.

Table 1 List of species studied and their respective families, number of individuals sampled (N), leaf habit (EG—evergreen, BVD—brevideciduous), pollination, seed dispersal syndromes and growth form of the plant, recorded in coastal dunes in the state of Maranhão, northeastern Brazil. (-) without leaf habit
Full size table

The community did not present seasonality for the phenophases, start and peak dates, with non-significant values on Rayleigh tests. It was not possible to determine the start dates for leaf fall and new leaves, because the species gained and lost leaves continuously, even though in small quantities (“Appendix 1 and 2”). Although the result was not significant, the angles of phenophases leaf fall and new leaves were concentrated in January. On the other hand, the flowering and fruiting started in April and May, respectively, and the peak for both phenophases was in September (Table 2). Two species did not produce fruit during the research period: Himatanthus obovatus (Müll. Arg.) Woods and Apeiba tibourbou Aubl.

Table 2 Circular statistics for seasonality, peak dates of leaf fall and new leaves, start and peak dates of flowering and fruiting for coastal dune community located in Maranhão, northeastern Brazil
Full size table

The percentage of the number of species showing leaf fall was practically constant throughout the year (about 90%) with a decrease only in January. The species showed a high production of new leaves during the whole rainy season, extending until August, when this phenophase exhibited the highest percentage, reaching almost 100% of activity. After that, the production decreased (Fig. 3a).

Fig. 3
figure 3

Percentage of species a, percentage of activity b, and Fournier intensity c of leaf fall and new leaves phenophases in the coastal dune community of the Amazon coast. Filled bars represent rainy season and unfilled bars represent dry season

Full size image

Regarding the community activity percentage, there was a decrease in the monthly amount of abscission and new leaves production compared to the activity percentage of the species, although the variation on values was similar. Leaf abscission occurred throughout the year but decreased in January, whereas the production of leaves showed a growth pattern in dry/rainy and rainy/dry transition periods (Fig. 3b).

The intensity values determined using the Fournier index showed higher rates of leaf abscission between October and December, with percentages above 30%, followed by a decrease in January (16%). In the succeeding months, these values varied between 22 and 29%. The intensity percentage for new leaves exhibited the same pattern found for the community activity, with an increase during the dry/rainy transition periods (Fig. 3c). Regarding leaf habit, most species were classified as evergreen (83.4%), followed by brevideciduous (16.6%). No deciduous species were found (Table 1).

The species showed reproductive phenophases activity throughout the year. The highest percentages of flowering and fruiting occurred in September, during the dry period, and the lowest in May, during the rainy period. Despite following the same rate of growth and decrease, the values of fruiting dynamics were always lower than those of flowering (Fig. 4a).

Fig. 4
figure 4

Percentage of species a, percentage of activity b, and Fournier intensity c of reproductive phenophases in the coastal dune community of the Amazon coast. Filled bars represent rainy season and unfilled bars represent dry season

Full size image

The highest and lowest flowering activities percentages occurred largely during the dry season, from July to September, and from November to December, respectively. Regarding fruiting, the highest activity was registered during the dry season, from August to October, while the lowest occurred in the rainy season, in January (Fig. 4b). The Fournier intensity percentage for the reproductive phenophases exhibited the highest values during the dry season, both in September. The lowest flowering activity was also observed during the dry season, in November, and the lowest fruiting activity occurred during the rainy season, in April (Fig. 4c).

Regarding pollination syndromes, 15 species (78.9%) were classified as mellitophilous, 2 (10.5%) as phalenophilous, 1 (5.3%) as chiropterophilous, and 1 (5.3%) as anemophilous. As for seed dispersal, 16 species exhibited zoochoric dispersal (84.3%), 2 (10.4%) autochoric, and 1 (5.3%) anemochoric.

According to the multiple regression results, the community has no relationship between the evaluated phenophases (leaf fall, new leaves, flowering, and fruiting) and the climatic data of the observed month, nor with the previous month (rain, temperature, photoperiod) (Table 3).

Table 3 Multiple regression analysis evaluating the effects of climate variables on the phenology of the coastal dune woody community located in Maranhão state, northeastern Brazil. R2 = regression of coefficient; p = probability value
Full size table

4 Discussion

The community did not show seasonality in its vegetative and reproductive phenophases. Besides, there was no relation to the climate elements evaluated (precipitation, temperature, and photoperiod), in the observed month or with the previous month of observation, indicating that other environment factors may be influencing the development of phenophases. The absence of relationships between leaf fall and new leaves, phenophases with climatic variables was also verified by Medeiros et al. (2007) in restinga vegetation in the state of Pernambuco, Brazil. Phenophase leaf fall (Silva et al. 2010) and new leaves (Cordazzo and Seeliger 1988; Bencke and Morellato 2002) have been related to rainfall, in restinga and dunes vegetation. In addition to rainfall, some coastal communities may still have relationships between vegetative phenophases (leaf fall and new leaves) with other climatic factors, such as temperature or photoperiod (Morellato et al. 2000; Marchioretto et al. 2007). This phenodynamic variation found in coastal areas may be related to the wide geographical location of the different communities.

The community evaluated in this study is located in low latitude, near the equator line, with so low variation of temperature and photoperiod. Probably, for this reason, the community did not show any relationship with temperature or photoperiod. Although it also does not show significant results in relation to rainfall, the dry season presented the highest percentages in almost all phenophases evaluated, which indicates that other factors associated with the dry period may be influencing the development of phenophases. In Brazil, coastal vegetation, such as those that develop in coastal dunes, has a wide geographical distribution, being subject to environmental conditions that vary according to latitude and that can influence the dynamic processes of communities (Marques and Oliveira 2004), exhibiting various phenological responses. In coastal areas located at low latitudes, where the annual photoperiod variation is small, the dry season generally tends to determine the phenological pattern (Medeiros et al. 2007; Silva et al. 2010). As the latitude increases, a greater photoperiod variation is noted; and since photoperiod and temperature are closely related, it exerts a greater influence on the phenophases (Marques et al. 2004; Marchioretto et al. 2007).

Regarding leaf habit, most species were classified as evergreen. The studied community is subjected to approximately four months with low or no rainfall, yet most species monitored maintained their leaves during the dry season, with constant leaf production and abscission. Evergreen species are not usually frequent in areas with water deficit, as the energy cost required for maintenance is higher (Holbrook et al. 1995). However, in coastal plains, which are normally environments of low soil fertility, an increase in leaf retention time can mean a nutrient-saving strategy, enabling the transfer of nutrients directly to new leaves (Jordan 1985).

In the flowering analysis, the phenodynamic observed in the community showed that flower production occurred throughout the year, showing an increase in the drier period, despite that, flowering did not show seasonality. These data converge with those obtained in other coastal areas, mainly with those found in lower latitudes (Medeiros et al. 2007; Silva et al. 2010). The occurrence of species in bloom throughout the year and many species with long blooms, as found in this study, is also reported for coastal areas located at higher latitudes (Castellani et al. 1999; Talora and Morellato 2000), where associations with climatic factors were verified, such as temperature and photoperiod (Marchioretto et al. 2007).

The long flowering periods of the species in the dune areas can represent an adaptive advantage to overcome the restriction that the strong action of the winds can induce in the pollinator activity in these areas (Albuquerque et al. 2007). Most of the species in the present study depended on pollinating agents for reproduction, with the majority presenting mellitophilous syndrome, information that converges with another study, previously carried out in the region's dunes (Gottsberger et al. 1988).

Fruiting recorded during all months of the year, and no significant relationship with climatic factors was found. The absence of relationships between this phenophase and the climate variables was also registered for restinga vegetation, in Pernambuco state (Medeiros et al. 2007). It was also verified the occurrence of the relationship between fruiting and precipitation in a restinga vegetation analyzed by Silva et al. (2010). Several ecosystems exhibit such a variable temporal pattern of fruiting (Machado et al. 1997; Justiniano and Fredericksen 2000; Morellato et al. 2013; Lacerda et al. 2017), which may indicate that, at the community level, there are different phenophase-related strategies associated with species seed dispersal, germination, and establishment (Morellato et al. 1989; Schaik et al. 1993; Griz and Machado 2001). These physiological mechanisms can be stronger drivers for fruiting than climatic variables.

As for fruits dispersion, more than 80% of the species in this study presented zoochoric syndrome. In Brazilian coastal areas, the proportion of zoochoric species are very similar to those observed in this study, always above 80% (Moraes and Mondin 2001; Almeida Jr. et al. 2008; Amaral et al. 2015), corroborating the assertion that most tropical species are animal-dispersed (Morellato and Leitão-Filho 1992; Talora and Morellato 2000).

Fruiting, which occurred throughout the year, with a subtle increase in the drier period, did not exceed the flowering percentages, differing from other coastal areas that, in at least some month, the fruiting values ​​exceeded those of flowering (Castellani et al. 1999; Medeiros et al. 2007; Silva et al. 2010). This observation suggests that there may be little fruit production in the community. In addition, the species Himatanthus obovatus (Müll. Arg.) Woodson and Apeiba tibourbou Aubl did not fruiting throughout the study, despite having flowering, unlike studies carried out in different ecosystems in which it is reported that these species have an annual fruiting frequency (Pirani et al. 2009; Barbosa 2014). Such information of low fruit production and failure in the fruiting of some species calls attention to the health of the studied community, a fact that becomes even more worrying because this is a sensitive community due to the strong anthropic action. Anthropic interference is considered one of the causes that can influence the behavior of species, mainly in areas with a greater indication of degradation, there may be a difficulty in obtaining the necessary resources for the maintenance and growth of the plants, due to the scarcity or unavailability of resources in the area (Costa 2002). Therefore, it is important to emphasize that the anthropic interference can harm fruit production that is crucial for the balance of ecosystems, as this guarantees food for dispersers and availability of seed for maintaining the plant community.

Finally, the results of the present study reveal that the coastal dune community did not show a seasonally defined behavior, nor relationship to any of the climatic factors evaluated. Despite this, the dry season had the highest percentages in almost all of the evaluated phenophases, which may indicate that other factors associated with the dry period, such as wind speed, may be responsible for the intensification of these phenophases. The prolonged flowering behavior found in this study seems to be a strategy common to other coastal vegetation. Therefore, other studies that take these factors into account are strongly encouraged. The information collected expands the knowledge about the functioning of this ecosystem and reinforces the need for more attention to coastal lowland areas, through intensification of ecological studies and reduced impact management measures, aiming at coastal vegetation conservation.