Phenological cycles in the Pantanal woody communities: Responses to climate and soil moisture seasonality

RESEARCH ARTICLE Phenological cycles in the Pantanal woody communities: Responses to climate and soil moisture seasonality Julia ArieiraID1,2*, Karl-L. SchuchmannID2,3,4, Arnildo Pott5, Michelle D. Lanssanova6, Ana Silvia O. TissianiID2, Osvaldo Borges Pinto JuniorID6, Marinêz Isaac Marques2,4 1 Science Panel for the Amazon, São Jose´ dos Campos, São Paulo, Brazil, 2 National Institute for Science and Technology in Wetlands (INCT-INAU), Centro de Pesquisa do Pantanal–CPP, Federal University of Mato Grosso (UFMT), Computational Bioacoustics Research Unit (CO.BRA), Cuiaba´ , Mato Grosso, Cuiaba´, Brazil, 3 Zoological Research Museum A. Koenig (ZFMK), Ornithology, Bonn, Germany, 4 Institute of Biosciences, Postgraduate Program in Zoology, Federal University of Mato Grosso, Cuiaba´, Mato Grosso, Brazil, 5 Institute of Biosciences, Federal University of Mato Grosso do Sul (UFMS), Campo Grande, Mato Grosso do Sul, Brazil, 6 University of Cuiaba´-UNIC, Cuiaba´ , Mato Grosso, Brazil * juarieira@gmail.com Abstract This study investigated the influences of regional climate and soil moisture conditions on the vegetative and reproductive cycles of seven savanna and forest vegetation communities of the Pantanal Mato-grossense. Circular analysis of the phenological data revealed the occur- rence of interspecific synchronism and seasonal responses in vegetative and reproductive activities, with flowering patterns differing the most between communities. Leaf shedding intensity events in communities were closely linked to climatic seasonality. Over half of the

individuals were semideciduous or deciduous, with leaf drop intense events predominantly triggered by drier, warmer conditions. The annual flood pulse further constrains woody plants, influencing deciduousness and serving as a strategy to mitigate soil water stress. The preceding climatic signal announcing cold fronts was a relevant determinant of flower- ing events for many communities. Climate and soil seasonality had limited influences on fruiting phenology across the various vegetation communities. The asynchronous response of phenological cycles to surface soil moisture seasonality highlights the diverse habitat hydrodynamics and its interactions with the plant communities that may decouple leaf fall, flowering, and fruiting from surface soil water availability. The unique phenological response of the Pantanal’s woody communities to the hydro-climatic cycle sets it apart from other non-flooded savannas of tropical South America. This response involves the intricate inter- play between phenological dynamism and alternating drought and wet-flooded phases. Introduction Tropical ecosystems face unprecedented crises. Accelerated changes in global and regional cli- mate and land cover degradation driven by unsustainable models of development are now PLOS ONE PLOS ONE | https://doi.org/10.1371/journal.pone.0316011 February 7, 2025 1 / 30 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Arieira J, Schuchmann K-L, Pott A, Lanssanova MD, Tissiani ASO, Pinto Junior OB, et al. (2025) Phenological cycles in the Pantanal

woody communities: Responses to climate and soil moisture seasonality. PLoS ONE 20(2): e0316011. https://doi.org/10.1371/journal. pone.0316011 Editor: Angelina Martı´nez-Yrı´zar, UNAM, MEXICO Received: April 4, 2024 Accepted: December 4, 2024 Published: February 7, 2025 Peer Review History: PLOS recognizes the benefits of transparency in the peer review process; therefore, we enable the publication of all of the content of peer review and author responses alongside final, published articles. The editorial history of this article is available here: https://doi.org/10.1371/journal.pone.0316011 Copyright: © 2025 Arieira et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the manuscript and its Supporting Information files.

pushing natural ecosystems in tropical regions to tipping points with loss of resilience [1]. Compared with South American upland rainforests, floodplains and riparian wetlands are among the most important inland wetlands in terms of the ecosystem services provided to peo- ple; vegetation composition and function are directly linked to the services provided, such as nontimber products, erosion control, carbon sequestration, and habitats for species [2–5]. Covering approximately 20% of South America, more than 2 million km2 [6], the extent of nat- ural wetlands continues to decline, and insufficient information on the state and impacts from human activities limits their protection [7,8]. In floodplain wetlands, cyclic biological events in plants, such as leaf fall, flowering, fruiting, and seed germination, are indicators of plant functional strategies and often differ along soil hydrological gradients [9,10]. Flood timing and predictability are influenced by seasonal cli- mate rhythms and are the primary drivers of plant species zonation, assemblages, and season- ality of ecosystem productivity [11–14]. Species adaptations to the flood cycle vary widely: some tree species exhibit dormancy followed by leaf fall during flooding, whereas some ever- green trees tolerate flooding by maintaining their metabolism and leaves [15]. The end of the wet season is known to trigger the production of new leaves and fruits on deciduous plants, whereas for evergreen species, sclerophyllous foliage is an adaptive trait that reduces water loss

via transpiration. The synchronicity between seed dispersal and hydrology is another adapta- tion to the flood cycle, which may increase germination success [12]. Approximately 1,000 species of trees that are tolerant to different flood levels exist in Ama- zon floodplains [16]. In other extensive South American wetlands subjected to predictable monomodal flood pulses, such as the Pantanal Mato-grossense [17], the number of flood-toler- ant trees is much lower: approximately 355 species (47% of the Pantanal trees) [18]. Among the 2,567 known Pantanal flora [19], 756 species are classified as woody plants, representing 39.7% of the higher plants [18]. Unlike in the Amazon, the vegetation of the Pantanal is pre- dominantly savanna, with most trees inhabiting nonfloodable environments and being toler- ant to prolonged dry periods [18,20]. The Pantanal has a savanna climate with marked seasonality defined by a dry season for more than 5 months. In such a hyperseasonal savanna, as described by Eiten in 1982 [21], alternating dry and wet seasons might trigger the pheno- phase of the production of leaves, flowers, and fruits [22]. Savanna formations may undergo a high vapor hydric deficit that is 30% greater than that of forests in the dry season, affecting spe- cies water use, while some forests are more impacted by seasonal hydrological variation [23]. The capacity of habitats to store soil moisture during periods of low rainfall might provide humid microclimates to plants sensitive to drought [24]. The evolutionary response of species

to such seasonal water deficits may involve deep roots and stomatal or chemical control of water loss through evapotranspiration and leaf shedding [23,25]. As in other tropical wetlands [7,8], knowledge about the reproductive and vegetative cycles of woody plants in the Pantanal and their response to seasonality is insufficient [22], limiting our understanding of the func- tions of ecosystems and the impacts of human-induced environmental changes on biodiversity and ecosystem services. We do not yet know, for example, the impact of climate change on the Pantanal landscape composition and configuration, particularly under scenarios of 30% rain- fall reduction and up 7˚C temperature increase by 2100 [26]. Such changes could significantly affect the production of flowers and fruits, which are essential resources for some of the richest herbivore wildlife found in South America [27,28]. The intense wildfires of 2019–2020 in the Pantanal have challenged scientists in understanding the regeneration capability of plant com- munities, which might depend on the speed and direction of reproductive and vegetative phe- nological responses at the intra- and interspecific levels mediated by trade-offs in energy allocation for different parts of the plant [29]. PLOS ONE Phenological cycles in the Pantanal woody communities PLOS ONE | https://doi.org/10.1371/journal.pone.0316011 February 7, 2025 2 / 30 Funding: This work was funded by the Brazilian Science Council (CNPq) awarded to KLS and MIM, Brehm Foundation for International Bird Conservation, Germany, awarded to KLS, National

Institute for Science and Technology in Wetlands (INAU/ UFMT: www.inau.org.br), Brazil, awarded to KLS and MIM, and Instituto Arapyau´ awarded to the Science Panel for the Amazon, Brazil, JA. Competing interests: The authors have declared that no competing interests exist.

In this study, we investigated the existence of seasonal phenological patterns in the distinct woody communities of the northern Pantanal Mato-grossense, which are subjected to similar regional climates but under distinct temporal variabilities in soil moisture availability. To determine the underlying factors influencing plant phenological cycles, we investigated the fol- lowing question: How does the annual seasonality of meteorological conditions and local soil moisture contribute to orchestrating the synchronicity of reproductive and vegetative pheno- logical activities among coexisting species populations? Methods Climate and hydrogeomorphologic conditions of the Pantanal basin The Pantanal basin spans Brazil, Bolivia, and Paraguay in the Southern Hemisphere tropics. It has complex meandering and branching river patterns that flow north to south. The central basin is a lowland (<200 m a.s.l.) formed by Quaternary alluvium and is bordered by the old crystalline Central Brazilian Shield in the east and north, the humid Chaco plains of eastern Bolivia in the west and Paraguay in the south, and the young uplifting Andes in the west [30,31]. Located in the upper Paraguay Basin and extending over 150,502 km2 in the central- western region of Brazil, the climate of the Pantanal is typical of tropical savanna (Aw) accord- ing to the Ko¨ppen (1948) classification [32], with a marked seasonality between the periods of dry winter and rainy summer, leading to a wet-dry climate, with an annual mean temperature

of 26.5˚C and rainfall of 1,100 mm ± 300 mm concentrated from October to April (~80%) [20,33]. Rainfall decreases sharply in winter, with monthly records of 10 mm, resulting in a dry season in which evapotranspiration surpasses precipitation [34]. The Pantanal is considered a floodplain with an annual and predictable pattern of flooding influenced by river overflow and summer rainfall [30,35,36], controlling geomorphology and the great diversity of terrestrial, aquatic, and temporary terrestrial habitats [17,31]. The plain has a low-elevation and low-relief topography, with altitudes varying between 80 and 150 m and a north-south declivity of 2.5 − 5.0 cm.km-1, which allows the seasonal flooding of the Upper Paraguay River to cover an area greater than 130,000 km2 [31]. The level of rivers in the northern Pantanal, such as the Cuiaba´ River, follows the trend of seasonal rainfall, causing annual flooding of adjacent plains, which can last more than 6 months [37]. Even after the dry season, part of the plain remains inundated for an additional 2 months due to the low slope and poorly drained soil [38,39]. The soils of the Pantanal are generally hydromorphic, with textures varying from clayey to sandy [40]. Many soils remain water saturated or inundated for a few days to several months due to seasonal floods, restricted drainage, or a highwater levels (2.5 m on average) [37,41]. Thus, hydromorphic soils are found across the region, and their morphological characteristics are attributed primarily to the reduction and oxidation of iron and manganese, which also

confer their distinctive color [42]. The soils found in the northeastern Pantanal are predomi- nantly clay in texture (between 3% and 65% of clay) and are clayey and in the great domain of the Plinthosol [41]. Alternating annual and multiannual dry and wet periods have resulted in different patterns of discharge and sediment load across the floodplain, creating a mosaic of geomorphologic for- mations linked to subtle changes in relief, such as high levees and paleolevees, reflecting inun- dation regimes and the capacity of the soil to retain water [43]. This mosaic is covered by diverse grassland, savanna and forest vegetation with distinct structural, floristic, and func- tional identities, which constitute the basic units, namely, macrohabitats, for the management and protection of biodiversity [17]. PLOS ONE Phenological cycles in the Pantanal woody communities PLOS ONE | https://doi.org/10.1371/journal.pone.0316011 February 7, 2025 3 / 30

Phenological data collection and analyses Sampling design and collection. The study area is located along the margins and flood- plains of the Cuiaba´ River and its tributary, the Piraim River, within the SESC Baı´a das Pedras. It belongs to the Cuiaba´ fluvial fan system, the second largest megafan of the Pantanal Basin [43]. This area is part of the SESC Pantanal Ecological Station, located in the municipality of Pocone´ (Mato Grosso, Brazil), and it is dedicated to ecotourism and scientific research. Cover- ing 4,200 hectares, it hosts the Advanced Research Base for Pantanal Studies of the Federal University of Mato Grosso. This study was performed during the annual hydroclimatic cycle from July 2017 to July 2018 in the northern Pantanal (16˚30’S and 56˚25’W). This period represents normal to mod- erately wet conditions intercepted with intense rainy periods lasting from nearly 2–4 months [44], considering that the cumulative annual average precipitation in the region was 1,516 mm from 1991–2020 [45]. The vegetation communities of the study area represent mosaics of dif- ferent forest and savanna formations in the northern Pantanal and other subregions [17]. Observations of phenological aspects were made in seven communities, namely, Seasonally Flooded Woodland, also known as Parkland Savanna (PS); Low Tree and Shrub Savanna (Cer- rado Savanna—CS); Low Tree and Shrub Savanna, invaded by Trees (Invaded Savanna—IS); Shrubland (SHB); the Alluvial Seasonal Semideciduous Forest (ASDF); the Lowland Seasonal

Semideciduous Forest (LSDF); and the Semi-evergreen Monodominant Forest of Vochysia divergens (MF) [46]. Parkland Savanna occurs in areas experiencing short-term flooding from rainfall accumu- lation lasting only a few weeks, with depths reaching less than 50 cm, and is distributed along abandoned lobes on fluvial fans [43,47]. These woodland savannas usually encompass a single tree or shrub species over a grassy layer, featuring typical species such as Byrsonima cydoniifo- lia scrub, Curatella americana, Handroanthus heptaphyllus, and T. aurea, which create spots of monodominant formations. The presence of species such as the spiny Machaerium hirtum and the palm Copernicia alba gives these formations physiognomic similarities with the Steppic Park Savanna (Humid Chaco) of the southern Pantanal (Silva and Caputo 2010, Polido and Sartori). The treelet B. cydoniifolia ("canjiqueiral") is associated with sandy soils, whereas C. alba is associated with calcium- or sodium-rich patches [48]. Cerrado Savanna (Low Tree and Shrub Savanna) is composed of herbaceous and shrubby plants, palms, and isolated trees up to 8 m tall, including species such as Cupania vernalis, Ery- throxylum anguifugum, Eugenia florida, Genipa americana, and Tabebuia aurea. This vegeta- tion community typically occurs when flooding lasts less than 3 months and when the floodwater is up to 1 m deep. These dynamic habitats feature species with high tolerance to floods and droughts [47]. The reduction in flooding and soil moisture leads to species turn-

over, with some species, such as the less flood-tolerant Callisthene fasciculata, becoming dominant. Invaded Savanna (locally called campo sujo) by native trees and shrubs exemplifies the Pan- tanal vegetation dynamics regulated by multiannual dry and wet periods that result in a reor- ganization of the spatial distribution of vegetation communities and species [47]. Multiple wet years drive the expansion of pioneer and native riparian tree V. divergens over flooded open- canopy savannas and natural grasslands, reducing herbaceous diversity and evolving into a semi-evergreen monodominant forest [49]. In this woody encroached community, flooding lasts 3–6 months in wet years, and the soil can remain waterlogged in the dry season, with establishment benefitting from low wildfire regimes [50]. Shrublands are usually characterized by encroachment stages of a set of shrubby species over grasslands and savanna areas, such as the entangling Combretum lanceolatum and C. PLOS ONE Phenological cycles in the Pantanal woody communities PLOS ONE | https://doi.org/10.1371/journal.pone.0316011 February 7, 2025 4 / 30

laxum (“pombeiral”). Shrublands dominated by C. lanceolatum are flooded at approximately 2 m and for up to 6 months and have some associated species, such as Triplaris gardneriana, a riparian tree. These shrubs initially grow in isolated clumps but gradually form dense mono- specific stands [49]. Alluvial Seasonal Semi-deciduous Forests are located along permanent watercourses and experience periodic flooding for short to long periods depending on their position along the relief gradient [51]. The canopy height can reach 30 m on better-drained riverbanks near run- ning water [46,48,52]. It is among the richest woody species colonized by highly flood-tolerant species, including Garcinia brasiliensis, Mouriri guianensis, Ocotea suaveolens, Siparuna brasi- liensis, Trichilia catigua, Vitex cymosa, and Zygia inaequalis. Lowland Seasonal Semideciduous Forests may appear contiguous to the Alluvial Seasonal Semideciduous Forests along the relief gradient [52] on flood-free or shortly inundated (1–2 months) ridges or paleolevees [53,54]. Many species, such as Anadenanthera colubrina, Aspi- dosperma cuspa, Astronium fraxinifolium, Attalea phalerata, Cordia glabrata, Dipteryx alata, Pseudobombax marginatum, and Tabebuia roseoalba are frequently shared between these types of forests. In these slightly higher-lying areas, differences in water availability may benefit species with greater tolerance to drier conditions, such as Combretum leprosum [55]. Semi-Evergreen Monodominant Forest of Vochysia divergens, known as “Cambarazal”,

is characterized by the monodominance of V. divergens (>50% dominance) which is usually found on young fluvial deposits [52]. The understory is composed of medium to large trees such as Duroia duckei, Leptobalanus parvifolius, and Nectandra amazonum, riparian species phytogeographically linked with Amazon flora [46]. The canopy can exceed 20 m high, creat- ing dense tree cover with a sparse understory where V. divergens seedlings that are either scat- tered or absent. This forest typically develops under a prolonged flood regime, often lasting over six months [47]. In each of these seven communities, one to three plots measuring 5 m x 80 m were used to monitor phenological aspects at the community level, totaling 18 plots and a sampled area of 7.200 m2 (0.72 hectares). In each plot, all woody individuals with a DBH � 5 cm and palms were identified and tagged, and phenological data were recorded monthly, summing 983 indi- viduals belonging to 59 species across 28 families (17 individuals per species on average). The plant species were collected and deposited in the Herbarium of the Federal University of Mato Grosso (UFMT). Since the research was conducted outside protected areas and did not involve endangered species, no data collection permission from official environmental institutions (i.e., Instituto Chico Mendes de Conservac¸ão da Biodiversidade—ICMBio) was required. We identified plants at the species level by comparison with herbarium specimens and with help from taxonomists. Among the 983 individuals, 156 individuals from 14 species were in the PS,

198 individuals from 21 species were in the CS, 138 individuals from 16 species were moni- tored in the IS, 46 individuals from 5 species were in the SHB, 133 individuals from 34 species were in the ASDF, 180 individuals from 30 species were in the LSDF, and 132 individuals from 15 species were in the MF. The number of species in our study plots is similar to tree richness estimates from other phytosociological inventories conducted in the Pantanal [49,55–57]. We identified vegetative and reproductive phenological events through direct observation with binoculars to evaluate the presence or absence (qualitative) and intensity (quantitative) aspects of phenological dynamics. The vegetative phenophase is represented by leaf fall when canopy gaps can be observed or when leafless branches are present [58]. Flowering encom- passes flower buds and anthesis, and the fruiting phase includes green and ripe fruits. Pheno- logical activity and intensity were measured for each vegetation type to evaluate the synchrony, seasonality, and intensity of phenological events in the species populations. The index of phenological activity was assessed through the frequency of individuals in a certain PLOS ONE Phenological cycles in the Pantanal woody communities PLOS ONE | https://doi.org/10.1371/journal.pone.0316011 February 7, 2025 5 / 30

phenophase, determined by the presence of flower buds and flowers (flowering), green and ripe fruits (fruiting), and leaf fall equal to or greater than 30%, helping to indicate the begin- ning and end of the phenophase [58,59]. The intensity of the phenological events was mea- sured via the semiquantitative method proposed by Fournier (1974) [60]. We scored the percentages of leaf shedding in five classes by dividing the phenological phenomena into five categories according to intensity: 0 = absence of phenological event, 1 = presence of the pheno- logical event at a magnitude between 1% and 25%, 2 = presence of the phenological event at a magnitude between 26% and 50%, 3 = presence of the phenological event at a magnitude between 51% and 75%, and 4 = presence of the phenological event at a magnitude between 76% and 100%. The Fournier percentage of intensity was calculated monthly by summing the individual value categories (1 to 4) in each phenophase divided by the total number of individ- uals found in the community multiplied by four. This analysis allows us to estimate the inten- sity percentage of the phenophase and quantify the abundance of available plant resources and oscillating seasonal productivity [59,61]. The graphical results produced from the seven com- munities allowed comparisons between qualitative and semiquantitative phenological metrics within and between vegetation communities [59,62]. We classified all individuals as deciduous, semideciduous, or evergreen on the basis of intervals of annual leaf loss. Individuals were clas-

sified as deciduous if they lost more than 90% of their leaves at least once a year. Individuals shedding more than 50% of their leaves and less than 90% of their leaves were classified as semideciduous, and evergreen individuals maintained 50–100% of their crowns throughout the year [63,64]. Phenological synchrony and seasonality. We applied circular statistics to evaluate species synchronism and seasonality in phenological activities within the seven studied vegetation types, defining seasonality peaks and annual (in activity once a year) and subannual (in activity more than once a year) seasonal patterns [62,65]. This analysis is based on the Rayleigh test (Z) of the effects of the circular distributions of flowering, fruiting, and leaf fall events on the peak vegetative and reproductive phenologies of plants. Therefore, the distributions of the frequen- cies of flowering, fruiting, and leaf fall in each phenophase were plotted in circular histograms at monthly intervals (from July 2017 to June 2018), with 365 days of the year corresponding to 360˚ of the circumference and months converted into angles of 30˚. The vector length is related to the concentration coefficient (rc) value, which ranges from 0 to 1. The angle where it is plotted indicates the mean angle that corresponds to the phenophase mean date. On the basis of the results, the mean dates (C) and mean angles (μ) of phenophase occurrence were calculated for the month with the highest phenological activity, the concentration of each

event around that date (rc; vector length) and the significance value for the seasonal pattern (μ) of the Rayleigh test (p). The C value can be used to determine the seasonality and frequency distributions of occurrences of each phenophase, and the highest concentration represents the highest seasonality [58,61]. Circular histograms of the frequency of leaf fall, flowering, and fruiting events were generated, and the Watson-Williams test was used to test whether the mean periods of event intensity differed among the seven communities [65,66]. The null hypothesis is that means are equal across groups. These analyses were conducted via the soft- ware Oriana [67]. Abiotic control of phenological events. Assuming that the variables did not have a nor- mal distribution [65], nonparametric Spearman correlations were used to evaluate the rela- tionships between the Fournier intensity data for leaf fall, flowering, and fruiting (response variables) and the regional climate and local soil moisture conditions (explanatory variables). This analysis allows us not to evaluate interactions between variables but rather to contribute to understanding only the most likely phenological triggers, as aimed here [68]. Six regional climate variables were chosen considering their reported influence on phenological dynamics PLOS ONE Phenological cycles in the Pantanal woody communities PLOS ONE | https://doi.org/10.1371/journal.pone.0316011 February 7, 2025 6 / 30

in tropical regions [69,70], namely, monthly average temperature and mean maximum and minimum temperatures (˚C), actual evapotranspiration (mm month-1), total rainfall (mm), and insolation (hours month-1 of incident solar radiation). The climatic data were acquired from the National Agency of Waters on the website ‘Hidroweb’ (http://hidroweb.ana.gov.br) [71], and BDMEP–INMET data, representing the station of Cuiaba´, Mato Grosso (OMM: 83361). This official weather station has operated since 1911 [72] serving as a reference for cli- mate monitoring in the North Pantanal. The correlations were conducted with climatic values from the same month (r0) as the date of the phenological data and with values from the first (r1) and second (r2) months before the observations. This approach allowed us to evaluate delays in phenological response to annual climate variation. The local effect of seasonal variation on soil moisture (%) within each com- munity was also analyzed at a depth of 20 cm, a layer where savanna first lateral root branches grow and where soil nutrients are more available [73]. For this purpose, three soil samples per community were collected with a Dutch corer at 20 cm and taken to the Soil Physics Labora- tory of the Federal University of Mato Grosso (UFMT) for weighing and drying. The samples were oven-dried at 105˚C for 24 h, and the mean of three samples per unit was considered. The gravimetric soil moisture (%) was determined by the difference between the wet weight (WW) and dry weight (DW) via the equation [(WW−DW)/DW]. A soil moisture content of

100% indicates that the site is in a flooded state. The analyses were performed with the R ver- sion 3.6.1 library Hmisc package. Results Annual variability in meteorological and soil moisture data The annual air temperature averaged 27.8˚C, reaching a minimum of 15.0˚C in July and a maximum of 39.5˚C in September (Fig 1). Notably, the dry season, marked by low precipita- tion in July and September, contrasted sharply with the wet season, which accounted for approximately 85% of the total annual rainfall of 1123 mm. These months correspond closely to the annual variations in the water level of the Cuiaba´ River, the primary catalyst for flooding in the studied floodplain. Evapotranspiration reached its zenith during the rainy season (Octo- ber − April), peaking at 188 mm month-1. Its alignment with rainfall pattern was inconsistent, suggesting the possible influence of free water surfaces on evapotranspiration during flooding or the impact of minimum temperature fluctuations. The period of intense insolation spanned from July (257 hours month-1) to September (220 hours month-1), coinciding with the lowest actual monthly evapotranspiration (9–38 mm month-1). Conversely, the rainy season corre- sponded to a 174.3 hour month-1 reduction in insolation in October, which was indicative of increased cloud cover. The soil water moisture content varied between community sites, as illustrated in Fig 2, depicting the intrinsic connection between surface water flux and vegetation cover, soil drain-

age capability, and the cumulative effect of rainfall. The lowest average soil moisture content was found in PS (20%±24) and LSDF (20%±4), followed by CS (29%±25), ASDF (30%±19), MF (40%±39), SHR (40%±42), and IS (51%±38) (S1 Appendix). In some vegetation communi- ties, such as the ASDF, MF, and IS communities, increased precipitation was correlated with increased soil surface moisture and greater evapotranspiration in the previous month. There was no apparent direct influence of climate variables on the annual variation in soil moisture in the LSDF and CS. A higher maximum temperature with a peak in October reduced the soil moisture in the SHB. PLOS ONE Phenological cycles in the Pantanal woody communities PLOS ONE | https://doi.org/10.1371/journal.pone.0316011 February 7, 2025 7 / 30

Plant community phenology Leaf phenology. All seven communities had more than 50% of individuals semideciduous or deciduous, i.e., shedding more than 50% of their leaves at least once a year (Fig 3). The com- munities with a greater number of deciduous individuals were found at both extremes of the soil hydrological gradient. Specifically, IS and MF, located at the wetter end, represented 54% and 23% of deciduous individuals, respectively, whereas PS, which occupied the drier end, rep- resented 51% of deciduous plants. Cerrado savanna accounted for 13% of the deciduous area and 58% of the semideciduous area. The two semideciduous forests contained approximately 50% semideciduous individuals and 10% deciduous individuals, and presented the greatest number of evergreen individuals (LSDF = 39%, ASDF = 35%). Circular analysis of the phenological data revealed interspecific synchronism and seasonal responses to leaf fall in four of the seven communities, namely, the PS (rc = 0.21, p< 0.001), ASDF (rc = 0.19, p< 0.001), SHB (rc = 0.18, p< 0.001), and MF (rc = 0.12, p< 0.001) commu- nities (Fig 4A). Most of these communities are characterized by the dominance of a few spe- cies. On the other hand, sub-annual seasonal patterns of leaf fall were observed in the activity of the populations in other communities and, to a greater extent, in the intensity of the event, reflected in the low values of the concentration coefficient (rc) for these and other communities. The average length of the vector represented the average number of consecutive months

with the highest population activity. The average month with a high concentration of active species in the PS and ASDF was June, which represented the period of reduced rainfall and a decrease in the moisture content of the local soil. During this period, 94% of the PS species, including H. heptaphyllus, C. fasciculata, and M. hirtum, and 75% of the individuals exhibited leaf loss. In the ASDF, Cordia glabrata and Pseudobombax marginatum were among the 54% Fig 1. Hydroclimatic annual variability between July 2017 to July 2018 representing the Cuiaba´ fluvial fan system. A. Precipitation (mm); B. Water level of the Cuiaba´ River (mm); C. Maximum, mean, and minimum temperatures (˚C); D. Actual Evapotranspiration (mm month-1), and relative humidity (%) (sources: National Water Agency (ANA) and the National Institute of Meteorology (INMET)). https://doi.org/10.1371/journal.pone.0316011.g001 PLOS ONE Phenological cycles in the Pantanal woody communities PLOS ONE | https://doi.org/10.1371/journal.pone.0316011 February 7, 2025 8 / 30

Fig 2. Soil moisture content fluctuation (%) in the upper 20 cm of soil across seven plant communities (A–D), alongside precipitation (mm), representing the hydroclimatic cycle from July 2017 to July 2018. https://doi.org/10.1371/journal.pone.0316011.g002 Fig 3. Individuals by leaf fall strategy (%). Individuals were classified as deciduous, semideciduous, or evergreen on the basis of intervals of leaf loss. Individuals were classified as deciduous if they lost more than 90% of their leaves at least once a year. Individuals shedding more than 50% of their leaves and less than 90% of their leaves were classified as semideciduous, and evergreen individuals were those which maintained 50–100% of their crown throughout the year. https://doi.org/10.1371/journal.pone.0316011.g003 PLOS ONE Phenological cycles in the Pantanal woody communities PLOS ONE | https://doi.org/10.1371/journal.pone.0316011 February 7, 2025 9 / 30

Fig 4. Vegetative phenology across seven plant communities. A. Circular histograms of the frequency of leaf fall phenological activity among species. The vector length is related to the concentration coefficient (rc) value, 0 to 1, and the significance value for the seasonal pattern (μ) of the Rayleigh test (p); B. Phenology activity and Fournier intensity of leaf shedding over an annual hydroclimatic cycle (July/2017 to July/2018). https://doi.org/10.1371/journal.pone.0316011.g004 PLOS ONE Phenological cycles in the Pantanal woody communities PLOS ONE | https://doi.org/10.1371/journal.pone.0316011 February 7, 2025 10 / 30

of the species and 36% of the individuals who exhibited leaf-shedding activity. Unlike in the previous communities, in the SHB community, the average month with a high concentration of species exhibiting defoliation activity was October, the transition period between the dry and rainy seasons. During this period, 80% of the species and 96% of the individuals in this community were active, and the event intensity was 49%. The peak community activity coin- cided with the greatest leaf fall of the pioneer and dominant species C. lanceolatum. MF presented increased leaf fall activity during the rainy and flooded seasons on the flood- plain, with the mean month having a greater concentration of species occurring in April. In that month, the frequencies of active individuals (Fig 4B) and species were 68% and 73%, respectively, as observed for the understory species M. guianensis, L. parvifolius, Triplaris americana, E. anguifugum, and Alchornea discolor. The Fournier intensity (Fig 4B) for leaf fall did not follow the same pattern as did the activity, with the dry season (July, 62%) being more intense than the wet period (49%), which was attributed mainly to the activity of the dominant species V. divergens. IS (rc = 0.07, p = 0.06) and CS (rc = 0.07, p = 0.056) presented weak seasonality in terms of leaf fall activity among the species, indicating that fluctuating concentrations of species lost their leaves throughout the year. For the IS, the leaf fall peak occurred during the wet period in

April, with 72% of the species being active, such as E. anguifugum, T. gardneriana, and V. divergens. The Fournier intensity revealed intensity peaks during both the wet-flooded and dry phases. Similarly, CS experienced continuous leaf fall activity year-round, with high concentra- tions of active species occurring in the wet-flooded phase in March and another maximum in the dry phase in August, when 59% of the species were active. The synchronous decrease in leaf fall among individuals was sharper than that between species, with lower percentages of activity occurring during the rainy season (Oct–Feb). The frequency of active individuals increased from 22% in October to 67% in April, when the area flooded. The seasonality of leaf fall in the CS differed from the patterns found in the SHB and LSDF (Table 1). The low soil moisture variation (20%±4%) throughout the year found in the LSDF is reflected in relatively constant leaf fall throughout the year, occurring at levels less than 40% of individuals. The exception occurred in April, when 62% of individuals and 62% of the species were active at a mean Fournier intensity of 47%. The variation in the proportion of individuals who experienced leaf fall throughout the year generally followed its intensity (Fig 4B) but was not within the same magnitude and had some delays. The relationship between intrapopulation activity and intensity seems to be influenced by the species diversity (or species dominance) found in each vegetation type. The most

diverse communities, such as the LSDF and ASDF (30 and 34 species, respectively), had rela- tively lower proportions of individuals in leaf fall during the dry season (33% and 34%, respec- tively, in July). In contrast, the less diverse communities, such as the PS (14 species), MF (15 species), and IS (16 species) communities, had high proportions of individuals with leaf fall activity in the same season (86%, 66%, and 83%, respectively). Reproductive phenology Flowering. The species’ flowering was seasonal for six out of the seven communities (excluding the ASDF community). The highest interspecific synchrony in the flowering period occurred in SHB (rc = 0.98, p = 0.001) and MF (rc = 0.83, p = 0.001) when the concentration of plants was high in August-September and June, respectively (Fig 5A). The flowering activity in the SHB was concentrated in months with low precipitation (<40 mm) and soil moisture (<8%). During this low rainfall period, 20% of the species flowered, although few individuals were active (2.17% in August) (Fig 5B). Among the seven communities, the MF presented the PLOS ONE Phenological cycles in the Pantanal woody communities PLOS ONE | https://doi.org/10.1371/journal.pone.0316011 February 7, 2025 11 / 30

highest Fournier intensity throughout the annual cycle, with the intensity of phenological events reaching 12% in July. The wet-dry transition season (June to July) marked the repro- ductive phase in this community, when 36% and 32% of the species (e.g., L. parvifolius, M. guianensis, and V. divergens) and c. 15% of the individuals were active. The flowering of IS and CS peaked during the dry season, with significant concentrations of species occurring around the mean date in July (rc = 0.68, p = 0.001) and September (rc = 0.28, p = 0.02), respectively. The major flowering event in the IS occurred from June to Sep- tember, during which 20–36% of the species and 6–9% of the individuals exhibited activity at an intensity of 3–5%. For example, T. aurea, V. cymosa, and V. divergens flowered during this period. In CS, despite a weak seasonal pattern of synchrony of flowering events revealed by the low length of the mean vector of the circular analysis (<0.50), the mean date of flower production was significant. There were two peaks of maximum activity in the wet (January, 17% species, 2% individuals) and dry (August, 17% species, 4% individuals) seasons. The Fournier intensity during flowering was low throughout the year, reaching a relatively high value in August (<2%) when the soil moisture was the lowest (8%). Typical Cerrado species, e.g., E. anguifu- gum, E. florida, C. fasciculata, C. americana, and T. aurea, were responsible for the highest indices of activity and intensity of this phenophase.

In the LSDF, the circular analysis revealed strong seasonality and interspecific synchrony (rc = 0.58, p = 0.002), with a mean flowering concentration in the rainy season of December. The flowering activity peaked in the rainy month of January, when 10% of the species, Table 1. Comparison of mean periods of vegetative and reproductive phenological event intensity across the seven communities via the Watson-Williams test. The null hypothesis is that means are equal across groups. Significant differences (p < 0.05) between community phenological patterns are highlighted with asterisks. Leaf fall Flowering Fruiting Community F p F p F p Invaded Savanna x Shrubland 3.46 0.08 2.63 0.12 1.35 0.26 Invaded Savanna x Cerrado Savanna 0.39 0.54 7.19 0.01* 0.48 0.50 Invaded Savanna x Parkland Savanna 0.10 0.75 4.26 0.05* 0.02 0.89 Invaded Savanna x Lowland SD Forest 3.12 0.09 5.49 0.03* 0.70 0.41 Invaded Savanna x Alluvial SD Forest 0.65 0.43 5.83 0.02* 0.10 0.75 Invaded Savanna x Monodominant Forest 0.65 0.43 1.00 0.33 0.36 0.56 Shrubland x Invaded Savanna 3.95 0.06 1.23 0.28 1.04 0.32 Shrubland x Parkland Savanna 2.71 0.11 0.06 0.81 0.17 0.26 Shrubland x Lowland SD Forest 0.60 0.45 0.44 0.52 1.56 0.22 Shrubland x Alluvial SD Forest 3.19 0.09 0.56 0.46 1.44 0.24 Shrubland x Monodominant Forest 2.39 0.14 3.21 0.09 1.06 0.31 Cerrado Savanna x Parkland Savanna 0.06 0.81 5.85 0.02* 0.37 0.55 Cerrado Savanna x Lowland SD Forest 4.35 0.05* 1.05 0.32 1.48 0.24

Cerrado Savanna x Alluvial SD Forest 0.09 0.76 0.90 0.35 0.82 0.38 Cerrado Savanna x Monodominant Forest 0.12 0.73 4.61 0.04* 0.002 0.96 Parkland Savanna x Lowland SD Forest 2.23 0.15 0.79 0.38 1.00 0.33 Parkland Savanna x Alluvial SD Forest 0.20 0.66 1.28 0.27 0.21 0.65 Parkland Savanna x Monodominant Forest 0.22 0.65 3.66 0.07 0.26 0.61 Lowland SD Forest x Alluvial SD Forest 3.17 0.09 0.03 0.88 0.26 0.62 Lowland SD Forest x Monodominant Forest 1.88 0.18 4.12 0.05* 1.17 0.29 Alluvial SD Forest x Monodominant Forest 0.01 0.94 4.22 0.05* 0.64 0.44 https://doi.org/10.1371/journal.pone.0316011.t001 PLOS ONE Phenological cycles in the Pantanal woody communities PLOS ONE | https://doi.org/10.1371/journal.pone.0316011 February 7, 2025 12 / 30

Fig 5. Reproductive phenology across seven plant communities. A. Circular histograms of the frequency of species with respect to flowering and fruiting phenological activity. The vector length is related to the concentration coefficient value (rc), 0–1, and the significance value for the seasonal pattern (μ) of the Rayleigh test (p). B. Phenology activity and Fournier intensity of flowering and fruiting over an annual hydroclimatic cycle (July/2017 –July/2018). https://doi.org/10.1371/journal.pone.0316011.g005 PLOS ONE Phenological cycles in the Pantanal woody communities PLOS ONE | https://doi.org/10.1371/journal.pone.0316011 February 7, 2025 13 / 30

September when the soil moisture was the lowest (10%), indicating the activity of M. guianen- sis and T. americana. Flowering seasonality was the phenological event with the greatest difference between com- munities, as indicated by Watson-Williams F tests (Table 1). The IS showed a similar pattern to that found for MF and SHB, communities characterized by a prolonged flood period and monodominance, with flowering concentrated in the dry season. Fruiting. Except for the PS and ASDF communities, all the other five communities exhib- ited interspecific synchrony during the fruiting cycle. According to the Watson-Williams test, seasonality during the fruiting period did not significantly differ among the communities (Table 1). Fruits were observed in the communities throughout the year, which may be attrib- uted to extended fruiting schedules and staggered ripening times. SHB (rc = 0.59, p<0.001) and MF (rc = 0.41, p<0.001) had the strongest seasonal patterns of interspecific synchrony of fruiting among the communities, with the mean months of fruit- ing occurring in the dry (September) and wet (December) seasons, respectively (Fig 5A). The few recorded individuals flowering in SHR, followed by high-frequency fruiting, may be due to our monthly monitoring missing the preceding flowering events. In the MF, the frequency of species fruiting was as high in wet December as in dry September (36%), with the event being 17% more intense in September (Fig 5B). Species such as M. guianensis and L. parvifolius

presented fruits during the wet and dry seasons, and others had fruits only in the wet phase, such as A. discolor, Myrcia tomentosa, and N. amazonum. The relatively high percentage of individuals fruiting (18%) coincided with the rainy season in December, approximately three to five months after the flowering peak. This pattern was observed in the monodominant spe- cies V. divergens (S1 Table). The mean fruiting date in the IS was dry September (rc = 0.31, p<0.001), when 36% of the species presented fruits, and the Fournier fruiting intensity was 2.8%. From December to March, during the rainy and flooded periods, fruiting was observed in E. anguifugum, G. amer- icana, H. heptaphyllus, and V. divergens. In the CS community, interspecific synchronism among species (rc = 0.21, p = 0.002) was associated with the mean concentration of fruiting in the rainy season (November). On the other hand, two fruiting peaks were recorded in CS (Fig 5B), one during the dry season in Sep- tember (15.3%), when the species C. fasciculata, C. americana, Psidium guineense, and T. aurea PLOS ONE Phenological cycles in the Pantanal woody communities PLOS ONE | https://doi.org/10.1371/journal.pone.0316011 February 7, 2025 14 / 30

were fruiting, and the other during the period of increasing soil moisture in January (9.7%), with fruits observed in E. florida and G. americana. In the flood-free LSDF, the Rayleigh test (rc = 0.36, p<0.001) indicated that the mean date of the fruiting phenophase was in January (Fig 5A), which coincided with the rainy season and the period was the highest soil moisture in this community (49%, Fig 2). The slight difference between the two peaks of Fournier intensity during rainy January (1.7%) and dry September (1.4%) (Fig 5B) did not reflect the increased proportion of active species and individuals. In contrast, the frequencies of active species (20%) (Fig 5A) and individuals (5%) in January were nearly double those reported in September (9.8% and 2.8%, respectively). Anadenanthera colu- brina, Attalea phalerata, Myrcia tomentosa, and Tabebuia roseoalba were among the fruiting species. The weak seasonality in fruiting patterns in the ASDF and PS groups suggests that fruiting events were not significantly concentrated at any particular time of the year (Fig 5A). However, in the ASDF, December was the month with the highest relative activity among the species (11%) and individuals (9%) (Fig 5B). Additionally, with the highest fruiting intensity occurring in December (9%), the importance of the rainy season for fruiting in this alluvial forest is evi- dent. The species that entered the phenophase in December were D. duckei, E. anguifugum, E. florida, M. guianensis, O. suaveolens, and V. cymosa. A unique feature of the fruiting cycle in

PS is the relatively high concentration of species’ fruiting activity during April and May (Fig 5A), marking the abrupt transition from a phase with soil moisture below 20% (March) to an inundated phase (April) (Fig 2). During these months, M. hirtum and Byrsonima cydoniifolia were observed fruiting. Many species populations in this community, including M. hirtum and H. heptaphyllus, begin fruiting during the low precipitation season from July to September and continue throughout the rainy season. Influence of climate and soil moisture on Fournier’s percent index of intensity Leaf fall. The intensity of leaf fall events was correlated with the meteorological variables in all the communities except LSDF (Fig 6). Among all the variables, evapotranspiration had the strongest correlation with leaf fall. The trends of decreasing evapotranspiration, precipita- tion, and temperatures, coupled with increasing insolation toward the austral dry winter, trig- gered leaf fall in most of the observed communities. A decrease in the mean (ranging from 25˚C to 23˚C) and minimum (ranging from 20˚C to 17˚C) temperatures was associated with increasing Fournier intensity of leaf shedding for most communities (Fig 6B). The negative influence of temperature on leaf drop occurred within two months before the date of maxi- mum leaf fall, indicating that cold fronts enter the region during austral winter. The decrease in minimum temperature from April (23˚C) to July (17˚C), during the period when rainfall is

decreasing and flooding is receding, was correlated with an increase in leaf fall in the PS (r0 = -0.85, p = 0.0002), IS (r0 = -0.81, p = 0.001), MF (r0 = - 0.73, p = 0.005), and ASDF (r0 = -0.71, p = 0.01). Even for other communities with weak species synchrony in terms of leaf activity, such as the CS communities, decreasing trends in minimum temperature were related to increased intensity of leaf fall (r0 = -0.80, p = 0.001). The IS (r1 = -0.60, p = 0.03) and PS (r1 = -0.60, p = 0.03) were also influenced by lower minimum temperatures one month before the phenological event. The mean (r0 = 0.62, p = 0.02) and maximum (r0 = 0.75, p = 0.003) tem- peratures were the only climatic variables influencing the intensity of leaf fall in SHB. The months with maximum temperatures from August to October (36 − 38˚C) corresponded with the maximum leaf fall intensity (43 − 50%). PLOS ONE Phenological cycles in the Pantanal woody communities PLOS ONE | https://doi.org/10.1371/journal.pone.0316011 February 7, 2025 15 / 30

Fig 6. Spearman correlation analysis between the Fournier intensity data for reproductive and vegetative phenological events and meteorological and soil moisture variables. The analyses were run considering concomitant (in red), one-month lag (in gray), and two-month lag (in dark blue) prior to the phenological event. The significance of the correlation is indicated by *p<0.01, **p<0.001, and *** p<0.0001. https://doi.org/10.1371/journal.pone.0316011.g006 PLOS ONE Phenological cycles in the Pantanal woody communities PLOS ONE | https://doi.org/10.1371/journal.pone.0316011 February 7, 2025 16 / 30