Study species

Acrocomia species are endemic to the subhumid tropical regions of the Americas and demonstrate a high adaptability to diverse soil and climate conditions. Acrocomia occurs naturally in regions with annual mean temperatures ranging from 15°C to 35°C and annual rainfall between 1000 and 1900 mm [45], while regions with higher precipitation are less favorable. In areas with annual precipitation below 1000 mm, Acrocomia is primarily found in moist wetlands and along water bodies with well-draining soils [45–48]. Its natural occurrence is typically associated with tropical seasonally dry areas, such as regions classified according to Köppen-Geiger as having a tropical climate with a dry winter (Aw) and tropical monsoon (Af) [47,49]. Acrocomia can endure periods of water deficit and high temperatures [45,46,50,51].

The most widespread species is A. aculeata, which ranges from southeastern Brazil to southern Mexico, with major population centers in the Mata Atlântica and Cerrado regions. A. totai and A. intumescens, however, are distributed in distinct regions of South America. A. totai occurs primarily in the southwest of Brazil, Paraguay, Bolivia and Northern Argentina, especially within the Pantanal and Chaco biomes. In contrast, A. intumescens is restricted to the northeast of Brazil, where it is associated with the Caatinga biome and related ecotones.

Acrocomia is a monoecious and protogynous palm. Its large panicle inflorescences are protected by a peduncular and woody bract, which splits abaxially (Fig 1). The rachillae bear both staminate and pistillate flowers, with the pistillate flowers becoming receptive before the staminate flowers reach anthesis. This favors cross-pollination facilitated mainly by insect pollinators, including flower weevils (Andranthobius spp.; Derelomini) and small sap beetles (Mystrops sp.; Mystropini) [30,52]. Therefore, at least two inflorescences are necessary for a successful pollination event, one acting as pollen donor and the other as pollen recipient.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Inflorescence of Acrocomia aculeata.

While emerging, the inflorescences are enclosed in a bract, which is woody and covered in brown trichomes on the outside with the occasional aculeus. The bract bursts at an abaxial, longitudinal slit and the inflorescence is released (= flowering event). At that point, pistillate flowers are already receptive, whereas staminate anthesis starts around 24 hours after bract opening. Small picture: Emerging, closed bracts in the palm crown. A. aculeata is here the representative of the studied species; bract-opening mechanics and overall inflorescence structure do not differ substantially from A. intumescens and A. totai [35].

https://doi.org/10.1371/journal.pone.0352981.g001

Study site

The study was carried out at the Active Germplasm Bank of Acrocomia (BAG-Macaúba), an ex-situ collection of over 300 maternal families (Accessions) located at the experimental station of the Federal University of Viçosa in Araponga, Minas Gerais, in the South-East of Brazil (latitude 20°40′1′′S; longitude 42°31′15′′W; altitude of ~1000 m). The BAG-Macaúba is registered as an official repository by the Brazilian Board of Genetic Heritage (# 084/2013-SECEX/CEGEN). The climate is characterized as a humid highland climate with rainy, warm summers and dry winters and is classified under Köppen as Cwb [53]. The soil in the area is considered a red-yellow Latosol (FAO/WRB: Ferralsol) with a sandy-clay texture [54].

Six accessions (31 palms in total) were selected according to the criteria outlined in Meyer et al. (2024) [35] which included (1) confirmed flowering in 2018, (2) the number of palm trees available per accession, (3) the height of the palm trees for labor safety reasons, and (4) the geographic origin of each accession (Fig 2). The accessions were named based on the Acrocomia species they were allocated to, as a 3-letter prefix (INT = intumescens; ACL = aculeata; TOT = totai) followed by their accession number from the BAG-Macaúba database (Motoike S., unpublished data, 2018).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. The regions of origin of the six Acrocomia accessions used in this study.

The map shows the major South American biomes in which Acrocomia occurs [40,47,55,56]. The Köppen-Geiger climate classification, annual mean temperature and annual precipitation for each accession’s region of origin are provided. In addition, information is provided on the planting time in the BAG-Macaúba and the number of plants that were evaluated. The climate information and Köppen-Geiger climate classification per region are derived, respectively, from climate-data.org and koppen.earth (both accessed 24 Sept. 2025). Map created by the author in QGIS 3.40.0-Bratislava (QGIS Development Team, 2025). Ecoregion boundaries were obtained from Olson et al. (2001) [57] (WWF ecoregions dataset); river centerlines and physical and cultural boundaries were sourced from Natural Earth [58]. The distribution points are derived from GBIF.org (accessed 24. Sept. 2025), where only living specimens outside conservation sites (e.g., botanical gardens) were considered. All cartographic styling were performed by the author.

https://doi.org/10.1371/journal.pone.0352981.g002

Inclusivity in global research

The study was conducted under the framework of a Memorandum of Agreement between the University of Hohenheim (UHOH) and Federal University of Viçosa (UFV), signed in 2018. In addition, a Material Transfer Term was signed between UFV and UHOH on March 4^th, 2020 for any exchange of plant, soil or other research material between the two institutions. Additional information regarding the ethical, cultural, and scientific considerations specific to inclusivity in global research related to this study is included in the Supporting Information (S7 Checklist).

Monitoring of flowering events

In the context of this study, the term “flowering event” will hereafter be used to specifically refer to the event of bract opening for each individual inflorescence (Fig 1). This marks the start of flowering for the given inflorescence and serves as the key time point for tracking its reproductive development throughout the flowering season. The term “flowering onset” is defined as the first flowering event of each palm within each accession and season. The flowering events of the accessions were monitored from September to February over the years 2019, 2020, and 2021. These periods are referred to as FS2019, FS2020, and FS2021, collectively termed “flowering seasons” throughout the paper. In total, 382 inflorescences were recorded across all flowering seasons. Flowering was monitored two to three times per week under favorable weather conditions guaranteeing the accessibility to the plantation, and with at least one assessment conducted weekly under unfavorable weather conditions. Because not all flowering events could be dated precisely to the day, all observations were aggregated into half-month periods for analysis. Half-month refers to two periods of each month: the 1st to the 15th (1H) and the 16th to the end of the month (2H). In the following, each half-month will be abbreviated as 1H or 2H, followed by a three-letter month abbreviation, and if needed, the year (e.g., 2HDec2019 for the second half of December 2019)

Circular statistics for flowering phenology

To assess the flowering pattern within and between seasons, we used circular statistics. Originally developed for data such as wind direction or animal migration patterns [59], the scope of circular statistics has since expanded to include phenological studies. In contrast to linear methods, which introduce discontinuities and artificial boundaries like the start and end of a year, circular statistics maintain the continuity of time, reducing bias and misinterpretation when dealing with circular data [1,59,60]. This is beneficial in tropical and subtropical ecosystems in particular of the Southern hemisphere, where phenophases often extend across the entire calendar year [1,61]. For circular statistics, time periods were converted into angles, with each day of the year (DOY) corresponding to approximately 1° and a half-month interval representing 15° on the circle.

For DOY (1–365) to angles (in degrees):

(1)

For half-month (1–24) to angles in degrees:

(2)

Using the “circular” package in R (Posit Software, Version 2024.09.0 + 375), we calculated the circular mean, circular median, angular deviation, and mean resultant length at the level of the flowering season. The mean resultant length (rho) is a measure for the concentration or dispersion of circular data, ranging from 0 (uniformly distributed) to 1 (perfectly clustered at a single angle).

The onset of flowering for each palm tree within an accession was defined as the date of its first flowering event. These individual onset dates were then used to calculate the circular mean and mean resultant length of flowering onset. For graphical presentation, the circular statistical measures were back transformed into half-month or DOY, where necessary.

To test for differences in flowering phenology, flowering dates were analyzed as angular data to capture their seasonal distribution among accessions within seasons and between seasons (FS2019, FS2020, FS2021). Two analyses were performed: one using all flowering events to calculate mean flowering dates, and another using only the first flowering event per palm (flowering onset). Data were grouped either by accession within each season (within season assessment) or by season within each accession (between season assessment). Differences in mean flowering direction were assessed using multivariate analysis of variance (MANOVA) on sine and cosine transformed angular flowering data (package npmv, R, Posit Software, Version 2024.09.0 + 375) [62]. Differences in angular dispersion among accessions were evaluated with the Wallraff test (package circular, R, Posit Software, Version 2024.09.0 + 375), which compares concentration of angles around the mean using the Kruskal-Wallis test (R Documentation, 2024).

Climate and photoperiod data

The daily weather data were obtained from the AgEra5 dataset [63] through the Climate Engine platform [64] for the period of 01/01/2016 to 31/12/2022 (Fig 3). The photoperiod data were downloaded from the NASA Goddard Institute for Space Studies [65] (Fig 4).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. The climate variables in Araponga, Minas Gerais, Brazil from 2017 to 2022.

They are shown in half-month aggregations from top to bottom: Standardized Precipitation-Evapotranspiration Index (SPEI), solar radiation, degree days, precipitation, mean temperature and minimum and maximum temperature. El Niño Southern Oscillation events (as defined by the NOAA Climate Prediction Center [66]) are shown as horizontal bars. El Niño is typically associated with drier conditions, while La Niña is associated with increased rainfall in south-eastern Brazil.

https://doi.org/10.1371/journal.pone.0352981.g003

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Photoperiod metrics for Araponga, MG, Brazil, calculated using NASA Goddard Institute for Space Studies data [65].

Top: difference between sunrise and sunset times over a half-month period. A negative sunrise/sunset delta means that the time of the event was earlier at the end of the period than at the beginning, and vice versa. A difference of 0 would indicate no change between the start and end of the period. The higher the absolute value, the more pronounced the difference and the greater the change. Bottom: Mean normalized day length and mean cosine of the solar zenith angle at solar noon.

https://doi.org/10.1371/journal.pone.0352981.g004

From the full set of AgERA5 climate variables, we selected those that were known to influence palm phenology and were also considered relevant from ecological and agricultural perspectives. To avoid multicollinearity, we examined the Spearman’s rank correlation between all the candidate variables. The final selection included precipitation, solar radiation, degree days, and the Standardized Precipitation-Evapotranspiration Index (SPEI). Degree days were calculated from daily minimum and maximum temperatures in accordance with the methodology proposed by Suresh et al. (2021) [67]:

(3)

As the base temperature of Acrocomia is unknown, the base temperature of 15°C of the African oil palm (Suresh 2020) was used. As both palms share a similar temperature range, we considered it the closest alternative.

In line with Borchert et al.’s (2005, 2015) [41,68] recommendations for tropical and subtropical regions, we retained normalized day length, sunrise and sunset times, and the daily insolation-weighted cosine of the zenith angle as photoperiod variables. The zenith angle describes the position of the sun relative to the vertical. Incorporating its cosine, weighted by daily insolation, provides a more physiologically meaningful approximation of the radiation incident on the canopy surface, influencing light availability, photosynthesis and growth patterns [69]. The day length was normalized according to the following formula

(4)

where day length is the number of daylight hours on a given day, day length_{short, year} and day length_{long, year} are, respectively, the daylight hours of the shortest and longest days of the year.

All the selected climate and photoperiod factors were then aggregated to the half-month as summarized in Table 1, to align with the time scale of the flowering phenology dataset. The SPEI was calculated using the SPEI R package (Posit Software, version 2024.09.0 + 375) to identify periods of drought and wetness based on the aggregated precipitation and Hargreaves–Samani potential evapotranspiration.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. An overview of the aggregation methods of all environmental factors used.

https://doi.org/10.1371/journal.pone.0352981.t001

We calculated the two-week change in sunrise and sunset times (delta sunrise and delta sunset) as the difference between their times at the start and end of each half-month period [41].

Seasonal decomposition of climate factors

Exploration of the climate data showed a hierarchical clustering of the climate data into winter (April to September) and summer periods (October to March) (Fig 5). Therefore, we performed a seasonal decomposition of time series using LOESS in R (Posit Software, Version 2024.09.0 + 375) on the 168 half-month periods (1HJan2016–2HDec2022). To minimize border effects in the seasonal decomposition model, one year of data preceding and following the period of interest of 2017–2021 was included, ensuring a more stable estimation of the seasonal decomposition components within the focal period (S1 Fig). Seasonal decomposition separates a time series into three components: the trend, representing long-term changes, the seasonal component, capturing regular repeating patterns, and the residuals, containing short-term irregular deviations not explained by trend or seasonality. These residual deviations are ecologically relevant because anomalies in temperature, precipitation, or drought (SPEI) can disrupt the timing (delay or advance) and magnitude of phenological events depending on species and phenophase, acting as environmental triggers [71,72]. For our analysis, we retained only the trend and residual components, excluding the seasonal signal, to focus on both long-term patterns and potential impacts of short-term anomalies on phenology.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. Half-month aggregated climatic factors, standardized using z-scores, plotted against normalized day length.

The mean values of the standardized factors are shown alongside the half-month data points. The equinoxes are marked at a normalized day length of 0.5. The dataset, which covers the period from January 2016 to December 2022, was derived from AgERA5 [63,64]. Normalized day length was calculated using NASA Goddard Institute for Space Studies data [65].

https://doi.org/10.1371/journal.pone.0352981.g005

Moving window correlation analysis

We conducted a moving window correlation analysis using Spearman’s rank correlation to assess the relationship between the peak flowering period and aggregated environmental factors, as well as trends and residual components extracted from seasonal decomposition. As for different palm species, the impact of environmental factors was found to be up to two years prior to the phenophase [23,28,29], we extended the analysis up to two and a half years before the flowering peak and performed it separately for each season and factor. The flowering event window was fixed at five half-months, spanning from 2HOct to 2HDec for each season, where the average half-month of the seasons’ peak was the middle (2HNov). The environmental factor window, also five half-months long, was shifted back by an increment of one half-month for a total of 60 half-months. Before analysis, inflorescence counts were log10-transformed to standardize flowering intensity across seasons, highlighting relative temporal patterns. Raw climate data (precipitation, degree days, solar radiation) were z-score standardized, while trend and residual components from seasonal decomposition were used unstandardized. To assess the significance of Spearman’s rank correlation coefficients, confidence intervals were calculated using the method described by Bonett and Wright (2000) [73]. All analyses, including the generation of heatmaps, were performed in R (Posit Software, Version 2024.09.0 + 375).

Limited short- and long-term climate effects

While the pronounced flowering seasonality links to photoperiod, climatic factors likely shape fine-scale variation in flowering timing via longer-term developmental processes, while short-term climate effects take a minimal role.

Precipitation events in August prior to flowering showed a positive correlation with flowering peaks. This suggests a potential role of precipitation in modulating the timing of bract opening within a photoperiodically-defined flowering window, as the first bract-enclosed inflorescences were observed as early as mid-August. However, this signal was not consistently detected across the broader period preceding flowering onset and up to the flowering peak, where associations with climate factors were generally weak and variable between years. These findings show only partial agreement with the results reported by Lorenzi (2006) [33], who found no correlation between temperature, precipitation, and flowering timing in the three months preceding flowering in Acrocomia populations in Paraná. Despite the evident similarity between the climatic (i.e., periods of dry winter and rainy summer) and the flowering seasons (peaking in November) in Paraná and in Minas Gerais, it is plausible that the discrepancy between the studies can be attributed to the methodological differences in the length of the chosen observation windows (three months in Lorenzi (2006) [33], and half a month in this study). Whilst certain studies have found no link between flowering and preceding temperature or precipitation in a range of palm species [5,33,82], others have identified such effects in other palm species including Mauritiella armata [17], Mauritia flexuosa [18,19], Attalea phalerata [15,16], A. funifera [83], Bactris glaucescens [15], Butia purpurascens [14], Cocos nucifera [84], and Sygarus romanzoffiana [85]. Overall, this evidence suggests that responses to short-term climatic variation are species-specific and even differ between members of the same genus [5,82]. In Araponga, rainfall during the dry winter occurs mainly as convective thunderstorms, and the observed signal in August likely reflects these short moisture pulses, which provide temporary relief from drought stress. In seasonally dry ecosystems, brief rainfall events have been shown to relieve water deficit and stimulate photosynthesis, growth, and reproduction [86,87], potentially influencing pre-floral development and fine-tuning the timing of flowering events [86,88]. The inconsistency in the link between flowering timing and short-term precipitation suggests that water availability likely plays only a minor role in modulating flowering seasonality of the Acrocomia accessions in Araponga.

While precipitation does not play a major role in the timing of the flowerings season window, correlations with precipitation and SPEI, and their deseasonalized anomalies, one to two years before flowering indicate that climatic variability is potentially linked to processes occurring during reproductive development. However, these correlations are not accompanied by consistent patterns in the magnitude or direction of the climatic variables across flowering seasons, indicating that the reason for these associations is not straightforward and remains unclear. Despite the lack of consistency, strong positive correlations with rainfall and drought anomalies 24 half-months prior to the flowering peak (S4 Fig) suggest that short moisture deviations may affect early developmental stages of leaves or inflorescences in the apical meristem. In pleonanthic species of the Arecaceae, each leaf has the capacity to develop an axillary inflorescence bud [28]. Thus, the number of inflorescence buds is associated with the number of the subtending leaves, which is further influenced by resource availability [12,28,29]. Consequently, the leaf emergence rhythm dictates the timing of inflorescence formation. In species such as Elaeis guineensis, inflorescence initiation occurs approximately 24–30 months prior to anthesis, followed by a prolonged phase of differentiation and expansion before flowering [12,29,89,90]. Similarly, extended developmental timelines exceeding two years have been described for C. nucifera [91], which, like E. guineensis, undergoes a sequence of several growth phases of reproductive development at different growth rates [29,91]. It remains unclear whether the strong positive correlation with water availability 22 months prior to the peak primarily corresponds to the timing of inflorescence initiation, leaf primordia formation that will subtend future inflorescences, or both, as the developmental timeline of Acrocomia is currently unknown. Despite this uncertainty, the timing and consistency of these long-term climatic associations with the flowering peak support a multi-year developmental process from floral initiation to bract opening in Acrocomia.

The marked seasonality in precipitation, temperature, and solar radiation (Fig 3) in Araponga, regulates the vegetative and reproductive development of the Acrocomia accessions in the BAG-Macaúba. Independent of the accession, leaves that expanded from around the summer solstice until shortly after the fall equinox produced inflorescences (S5 and S6 Figs), coinciding with periods of abundant water, radiation, and temperature. The process of leaf emergence was suppressed during the late winter months and resumed with increasing rainfall and radiation around the spring equinox (see S3 Fig). Beside the leaf emergence rhythm, the flowering periodicity in pleonanthic Arecaeae depends on two additional processes: the initiation and survival of inflorescence buds and units, and their expansion leading to bract opening and pollen release [28]. Resource availability, metabolic activity, sugar levels in the stem, and hormonal signal have been identified as contributing factors to inflorescence development in Arecaceae [12,90,92–95]. Explicitly, findings in E. guineensis suggest that the limited availability of resources, particularly water and radiation, may regulate the development and abortion of inflorescences (9–11 months before harvest) by affecting the carbohydrate status of plants and resulting in source-sink imbalances [11,12,24,89]. The absence of inflorescences in leaves expanding in winter and spring suggests abortion under conditions of resource stress. Inflorescences did not always flower in strict sequence with leaf age (S6 Fig), further indicating delays in pre-floral development possibly linked to resource availability. Thus, resource limitations may cause inflorescence abortion or slow expansion at various crucial developmental stages in Acrocomia. Primordia dissection studies [29] would be required to verify these assumptions.

Accession-specific variation in flowering event distribution

Despite the considerable geographic distances between their origins, the accessions flowered within the same seasonal window, suggesting that support for a single climatic factor acting as a universal trigger is not evident. There were no interannual differences in mean flowering time; however, accessions differed in their flowering period range, likely due to staggered flowering onset within accessions, the number of inflorescences produced per plant (more inflorescences resulting in a longer flowering period), and the flowering event rate (time between flowering events). Interannual variability was only evident in two accessions: ACL270, which flowered only in 2020; and TOT301, which showed a delayed onset and less clustered flowering events in 2019. At the start of this study, the plants were 6 and 8 years old, respectively, and still showed irregular reproductive behavior. We observed subtle but consistent differences in flowering onset, mean, and range among the various accessions (Fig. 6), as well as at the level of individual palms (see S2 Fig). Overall, this indicates that flowering seasonality is strongly synchronized and most likely driven by photoperiod, while accession-specific differences are mainly expressed in flowering intensity and event distribution within the season rather than in flowering timing within the year. While this distribution could in principle be influenced by climatic conditions through effects on short developmental phases such as inflorescence expansion or bract opening, any such effects would likely operate at a finer temporal scale (i.e., days) than the two-week resolution of this dataset. In addition, accessions may differ in their sensitivity to seasonal changes in light cues, including day length, sunrise or sunset timing, and solar zenith angle, which could contribute to the observed variation in flowering dynamics within the seasonal window. Such differences may reflect genetic adaptation to regional seasonal regimes [96,97], although the extent to which they represent genotype × environment interactions cannot be resolved within the single-site design of this study.

A number of limitations should be considered when extrapolating the discussed results beyond the study system. Firstly, the short study period of two to three flowering seasons captured only part of the interannual variability in flowering intensity, which was insufficient for some accessions. Secondly, while the inclusion of multiple genotypes from different origins within a common environment is a key strength, allowing partial separation of genetic and environmental effects, the absence of reciprocal trials across multiple sites limits the ability to assess the consistency of genotype-specific responses under contrasting conditions. Furthermore, as all accessions originated between 7° S and 25° S due to the availability of material in the BAG-Macaúba, including material from a broader latitudinal range, spanning both hemispheres, would strengthen the evaluation of how phenological patterns vary across environments and accessions. These constraints highlight the need for longer-term, multi-site experiments incorporating accessions across the species’ full distributional range. Such approaches would enable more robust testing of phenological patterns across environments and help resolve their drivers in Acrocomia beyond a correlational statistic, particularly when coupled with efforts to elucidate the underlying developmental mechanisms.

Implications for cultivation and breeding

Flowering periodicity is a central component of Acrocomia’s reproductive biology and has direct implications for its commercial cultivation as a palm crop for vegetable oil. Understanding the species’ flowering phenology is critical for plantation design, as self-pollination often leads to fruit abortion due to inbreeding depression [30,101,102]. Sequential flowering within individual palms reduces the likelihood of self-pollination, while synchrony among palms in a population and continuous flowering throughout the season enhance cross-pollination [36]. This combination of traits promotes gene flow among genetically diverse wild plants and reduces the risk of reproductive failure [36,47].

Meyer et al. (2024) [35] reported substantial phenotypic variation among Acrocomia genotypes, with inflorescence traits differing across clones and species, highlighting opportunities to exploit this diversity to improve plantation resilience and yield stability. Effective plantation management must therefore consider Acrocomia’s reproductive biology to maximize reproductive success. Based on our results and prior studies, and the limitations outlined above, we provide several considerations for the design and establishment of commercial plantations.

1. Planting only closely related or genetically uniform material (e.g., clones) may lead to poor fruit set and yield instability. However, planting genetically diverse material—particularly important when using clones—can mitigate the effects of dichogamy and ensure effective cross-pollination, since selfing rarely results in fruit set [30,38,102]. The deliberate use of phenological diversity, as noted by Scariot et al. (1991) [36], enhances mating opportunities and supports reproductive success. We suggest planting genetically diverse material and, in the case of clones, using several clones originating from different regions. As hybridization between species is observed [39,103,104], they can act as pollen donor to each other. So, a mixture of clones of A. totai, A. intumescens, and A. aculeata, which all show diversity in inflorescence and fruit traits [35] and phenological patterns, may be particularly valuable.
2. As photoperiod appears to be the key environmental cue of flowering in Acrocomia, with climate factors playing a secondary role, integrating genotypes from different geographic origins or Acrocomia species into plantations is feasible. Flowering periods will seasonally align despite local adaptations, allowing use of diverse genetic material without disrupting flowering synchrony. From a breeding perspective, these findings imply a high degree of transferability of flowering seasonality across latitudes, while still allowing for selection on traits influencing flowering duration, flowering intensity and synchrony, which may be relevant for yield distribution and harvest management.
3. Integrating genotypes with earlier or later flowering windows, as well as narrower and wider flowering windows within the same plantation, has the potential to extend the flowering and harvest periods. On the one hand, this permits adaptation to specific production systems, ranging from monocropping to silvopastoral and silvoarable systems, to complex multistrata agroforestry, while ensuring sufficient overlap for cross-pollination. For instance, introducing various Acrocomia genotypes into silvopastoral systems could lead to staggered harvest windows, where livestock can be kept on site while being kept apart from active harvesting areas. Conversely, staggering harvests, which are the most labor-intensive periods, can enhance predictability and flexibility in labor and post-harvest management.

Alternating production years, a common feature in Acrocomia, should also be considered. Year-to-year fluctuations in inflorescence production (Fig 6) are influenced by plant carbon (C) reserves, which depend on photosynthetic activity [105,106]. Long-term climatic patterns affect resource accumulation and allocation, ultimately determining the number of inflorescences produced [13,90,107,108]. However, the causes and magnitude of alternating production years in Acrocomia remain unclear due to the lack of long-term data on yield formation and C reserves.