Ecohydrological impacts of oil palm expansion: a systematic review

Global demand for vegetable oil and high oil palm yield have driven the rapid expansion of oil palm plantations in tropical countries. The research literature widely recognizes the effects of forest cover conversion into oil palm on biodiversity, deforestation, and carbon dynamics. However, research on the ecohydrological impacts of oil palm plantations is sparse, despite studies indicating that oil palm development may reshape land-water interactions and the availability and movement of water at different spatial and temporal scales. We address this gap by conducting a systematic literature review on oil palm development and its relation to ecohydrological processes. We found 139 relevant papers up to the year 2021, addressing different ecohydrological processes related to oil palm. We reviewed their spatiotemporal scales, geographic distribution, oil palm species and age, and the effects of land conversion from forest, cropland, and pastures. We also incorporated societal aspects regarding community perceptions of water. Our review highlights the effects of oil palm plantations on three main components of the water cycle: (i) land-atmosphere, (ii) fluvial systems, and (iii) soils and groundwater. Most studies include analyses of the Indo-Malayan and Australasian biogeographical regions (113), followed by the Neotropics (49) and the Afrotropics (15). Compared to rainforests, oil palm monocultures are warmer and drier. They have higher evapotranspiration (ET) rates, lower runoff regulation and infiltration capacity, and lower soil organic carbon (SOC). Although less often implemented, alternative oil palm management practices, including oil palm agroforestry, can help to mitigate some of these effects. Forest to oil palm conversion is the most studied land transition, while conversions from croplands, pastures, and grasslands are less studied. Overall, we identify gaps in understanding the long-term effects of management on ecohydrological processes under different land conversions, especially in the Neotropics and the Afrotropics, precluding research-informed policy to manage impacts of this expanding crop.


Introduction
The extensive use of palm oil in biodiesel and consumable goods has resulted in widespread global demand for this vegetable oil, leading to accelerated land-use change across tropical countries toward oil palm plantation establishment and expansion (Pirker et al 2016, Bessou et al 2017. The total harvested oil palm area increased by over 600% between 1961 and 2020, from approximately 3 Mha to over 28 Mha (FAOSTAT 2022). Moreover, between 1972 and 2015, about half of oil palm development expanded into previously forested lands, while the other half replaced other land uses (Meijaard et al 2018) displacing other crops onto non-cultivated lands (Fitzherbert et al 2008, Koh andWilcove 2008).
These findings indicate that, in tropical countries, oil palm is directly related to deforestation and indirectly causes further land-use change. However, patterns of oil palm expansion can also follow other land cover change pathways. In the Neotropics, for example, oil palm often replaces grasslands and agriculture (Furumo and Aide 2017). The expansion of oil palm cultivation is expected to continue based on the current palm oil market demands (von Geibler 2013, Bessou et al 2017 and as national governments promote these plantations as a driver of economic growth by establishing policies such as tax exemptions and subsidy programs (Castiblanco et al 2013, Dislich et al 2017, Furumo and Aide 2017. The surface alterations related to the cultivation lifecycle of the oil palm crop result in modifications of water and energy fluxes. The establishment of oil palm plantations in previously forested land typically involves terrain clearing, which is sometimes done by slashing and burning, draining waterlogged soils and peatlands, terracing in areas with high slopes, and building drainage channels and access roads (Dislich et al 2017). This process results in removal of soil organic matter, carbon losses, soil degradation and compaction, and biodiversity loss, among other negative impacts (Dislich et al 2017, Meijaard et al 2018. Planted areas generally constitute large-scale monocrops with homogeneous canopy structures, as plants are of uniform age, and understory vegetation is commonly cleared (Meijaard et al 2018).
Understanding the impacts of oil palm establishment and development due to these changes requires an evaluation of the water and energy cycles mediated by vegetation change. In this context, an ecohydrological perspective is an ideal framework to achieve a comprehensive view of these impacts. Ecohydrology examines the hydrological processes that underlie interactions between soil, vegetation, and climate (Rodriguez-Iturbe 2000). Ecohydrology also investigates the effects of hydrological processes on ecosystem composition, structure, and function and the effects of biotic processes on the water cycle (Nuttle 2002). The feedbacks produced by interactions among soil, vegetation, and water are highly sensitive to land use and climate, with implications for land and watershed management (Asbjornsen et al 2011). Given the nature of agricultural settings, social dynamics coupled with ecological and hydrological changes are critical for understanding ecohydrological feedbacks, especially as water availability is of utmost concern for human societies (D'Odorico et al 2010). Oil palm cultivation has been strongly linked to socio-ecological impacts on local communities (Reiss-Woolever et al 2021). For some communities, the negative impacts on water quality and availability are often not offset by economic gains, which only a few stakeholders commonly retain (Larsen et al 2014, Castiblanco et al 2015. Although the number of publications regarding the effects of oil palm cultivation and palm oil production has been growing over the last few decades, a systematic review framed in an ecohydrological context has, to the best of our knowledge, not yet been produced. Available reviews partially address some ecohydrological impacts of palm expansion, but their perspectives are not centered on ecohydrology. Ecohydrological topics addressed by current reviews include plant water requirements and irrigation (Carr 2011) and the impacts of oil palm on surface hydrologic fluxes (Comte et al 2012), and ecosystem functions (Dislich et al 2017). Some of the gaps in ecohydrological processes identified in these reviews highlight: (1) the need to quantify changes in the water balance partition between throughfall, stemflow, and rainfall interception across different oil palm ages, ET at immature oil palm stages, and long-term runoff changes (Comte et al 2012); (2) limited understanding of biophysical changes (e.g. albedo, surface energy fluxes, and microclimate) associated with oil palm driven land use change (Dislich et al 2017); and (3) the lack of information regarding both the maximum water table depths oil palms can tolerate and the different water management strategies in oil palm productivity (Carr 2011). In addition, previous reviews focus on publications that limit their geographical scope mostly to Indonesia and Malaysia, the largest palm oil producers (Fitzherbert et al 2008, Wilcove and Koh 2010, FAOSTAT 2022, and center on the land transition from forest to oil palm plantations. While this focus was reasonable given these countries' large amounts of palm oil production, production is also increasing in the Neotropics and Afrotropics. As oil palm planting areas continue expanding to countries outside Southeast Asia, oil palm adaptation to local environmental conditions becomes critical. The expansion includes planting in environmental settings under different ecological and social backgrounds, with various cultural practices, and experimenting with other oil palm species. The genus Elaeis has three species. Elaeis guineensis Jacq. or African oil palm (E. guineensis) has been the oil palm species commercially used for palm oil production, while Elaeis Oleifera Cortés or American oil palm (E. oleifera) and E. odora have not been used on their own for commercial purposes (Corley andTinker 2003, Barcelos et al 2015). As genetic sampling of E. oleifera was thoroughly established around 40 years ago, oil palm breeders became interested in its agronomic potential (Barcelos et al 2015). Hybrids between E. guineensis and E. oleifera, called OxG interspecific hybrids, were created in the 1980s due to the sexual compatibility between these two species, and E. oleifera's high resistance to the bud rot disease, a highly lethal fungus infection (Duff 1962, Torres et al 2010. Different Elaeis species exhibit physiological, structural, and morphological variations that impact growth rates, canopy cover, fruit production, and tolerance to disease and climate extremes (e.g. drought and flooding) (Corley and Tinker 2003). Taken together, the diversity of environmental settings, oil palm species, and hybrids may constrain the applicability of conclusions raised in Indonesia and Malaysia to different regions given the potential variations in ecohydrological feedbacks.
Here, we review the state of the knowledge and highlight emerging understanding associated with the effects of oil palm cultivation on ecohydrological processes across the tropics. We perform a systematic literature search and review with qualitative and quantitative elements (Grant and Booth 2009). Our review involves three main aspects not previously explored in a global systematic review: (1) it summarizes work that examines different land transitions to oil palm at different spatial scales across the tropics to identify trends of change; (2) it characterizes the state of knowledge of ecohydrological processes occurring within the crop and how these processes change through time and space; and (3) it examines the extent to which different Elaeis species and hybrids have been studied. Our literature review also assesses whether and how social perceptions in oil palm cultivation have been linked to ecohydrological processes. We evaluate and synthesize available information at several spatial and temporal scales to determine the extent of oil palm research on ecohydrological processes in the three main components of the hydrologic cycle: land-atmosphere; fluvial systems; and soils and groundwater. We also summarize the effects of oil palm on the water budget and how these effects link to other ecosystem processes and to scenarios of land transition (we considered tropical forests, croplands, and pastureland or grasslands). Finally, this review describes areas of potential future research on ecohydrological fluxes that still need to be addressed.

Material and methods
We developed a systematic search and review to cover the spread and unequal distribution of oil palm research in ecohydrological processes across the tropics. We used a systematic literature search and review approach based on the search, appraisal, synthesis, and analysis (SALSA) framework (Grant andBooth 2009, Samnani et al 2017). This framework established five steps to perform literature reviews and to decrease the risk of bias: SALSA. Each of these steps is described in detail as follows.

Literature search and selection
We conducted a literature search across nine databases and one search engine: Redalyc, Scielo, Dialnet, IEEE, Springer Link, Scopus, Science Direct, Web of Science, PubMed, and Google Scholar. Considering the expansion of oil palm cultivation in Latin America (Furumo and Aide 2017), we decided to include literature in Spanish and Portuguese to ensure that relevant research from Latin American countries was not filtered out due to a language barrier. Redalyc, Scielo, and Dialnet databases were used mainly to find literature written in these languages. Literature on oil palm that explicitly includes the term 'ecohydrology' is not very common. Therefore, the search query across the different platforms was done using combinations of search terms associated with ecohydrology, especially those related to water and energy balance in the soil-plant-atmosphere continuum (see section 1 at supplementary material, table S1). Our search was performed between November 2018 and March 2021. During the preparation of the manuscript, we included relevant and more recent literature that was not systematically collected. We did not limit the search to any year or period.
The resulting publications were filtered in three primary stages after removing duplicate records: (1) selection by title and abstract, (2) filtering by group to do complete reading, and (3) classification, detailed data extraction, and data analysis by category. In the first stage, we screened the search results by title and abstract (figure 1, part 1). In the second stage, we grouped the documents retained from the first filter into three groups according to the information included in the title and abstract and after reading the material (figure 1, part 2). Groups A and B categorize papers according to the biogeophysical ecohydrological processes they study. Group C includes papers that link physical processes with societal aspects of water. While groups A and B focus on strictly physical processes, we wanted to identify perceptions of ecohydrological processes in social-related literature. This identification aimed to establish if and how linkages between ecohydrology and social perceptions of water use and availability have been addressed. In the third stage, we read papers from each group in detail and classified each paper based on its specific topic into one or more of four categories of analysis (oil palm distribution and expansion, landatmosphere, fluvial processes, and soil and groundwater). For each group we extracted data using a Google forms survey (http://forms.google.com), with specific questions derived from the main questions of each group (see supplementary material, section 2).
The initial groups were defined as follows: Group A: hydrological papers, including work examining current or expected oil palm expansion scenarios and their hydrologic effects. The articles could address one or several processes in the hydrologic cycle or water balance at one or more scales. Guiding questions: how do studies consider coupled hydrologic-vegetation function? What are the effects of oil palm on surface energy and water balances? Which types of land use and land cover transitions have been studied? Figure 1. Systematic review paper filtering stages, data extraction, and data analysis. 1. General descriptive information obtained at the beginning of reading each paper (first filtering stage). 2. Grouping and complete reading (second filtering stage). 3. Classifying in categories of analysis and final data extraction. In part 2, group A: hydrologic processes (blue shapes); group B: biogeophysical processes (green shapes); group C, sustainability and sociopolitical aspects of oil palm development (yellow shapes). After surveying papers by group, papers were classified and analyzed into one or more categories blue-green shapes): oil palm distribution, land-atmosphere, fluvial, and soil and groundwater processes. The final numbers of relevant papers for each subcategory are presented in the figure. Some papers overlap categories (see supplementary table S3).
Group B: research on oil palm expansion patterns, including papers quantifying and mapping oil palm and its biophysical characteristics and structural attributes, such as leaf area index, other canopyrelated metrics, and water use efficiency. Guiding questions: what is the geographical distribution and expansion of oil palm? Which types of land use and land cover transitions have been studied? How do the biophysical and structural attributes of oil palm impact environmental conditions?
Group C: research on political, economic, sustainability, and social aspects related to linkages between physical processes in ecohydrology and society associated with communities' perceptions of water availability. Guiding questions: what are the major impacts on water resources? What changes in water availability are perceived by communities? Are the perceptions from the community being addressed in the scientific literature?

Data synthesis and analysis
We collected general (e.g. year, authors, study area, and temporal and spatial scale of the analysis) and topic-specific information from each paper (figure 1, part 3). Additionally, we identified the study area of each paper to verify whether the studies are concentrated in specific regions or areas of interest. Given the variety of definitions of spatial scale, we classified papers using the scheme developed by Becker and Nemec (1987), who define spatial scale as plant (at a specific part of the plant), plot (1 m 2 -1 km 2 ), regional (>1 km 2 -100 km 2 ), mesoscale (>100 km 2 -10 000 km 2 ), and macroscale (>10 000 km 2 ). Papers with study areas covering the entire tropics were considered macroscale. We based the selection of specific information variables for groups A and B on the works of Bonan (2008) andD'Odorico et al (2010). These authors described biogeophysical ecosystem processes, in particular, how they are integrated to treat the biosphere as a coupled system (Bonan 2008), and land ecohydrological processes and their role in environmental change (D'Odorico et al 2010). For papers in group C, we selected specific information regarding the methods used for gathering communities' perceptions. We aimed to identify whether the paper specifically addresses a water-related issue and information on the perceived impacts on water resources and the potential connections to oil palm cultivation (see section 2 supplementary material).
Following the components of the water cycle and their interactions with vegetation, we distributed groups A, B, and C papers into final categories of analysis (figure 1, part 3) defined as (i) oil palm distribution and expansion, which brings insights to understanding patterns of land transition and distribution of oil palm species; (ii) landatmosphere, which describes microclimatic variables, energy balance partition, and ET; (c) fluvial processes, indicating the hydrologic connectivity of the landscape through sediment transport and surface runoff, including streamflow; and (iv) soils and groundwater processes, which includes the soil properties that play a critical role in regulating the water balance (Rodriguez-Iturbe 2000), and water table fluctuation in groundwater systems. Papers from group C were included in these categories accordingly.
To identify the differences in ecohydrological processes under different scenarios of land transition into oil palm plantations, we designed a coding system (see supplementary material, table S2) that differentiates the direction of change (i.e. increase, decrease, no change, inconclusive or unknown), type of land transition, oil palm species or hybrids, oil palm age, and the scale of the analysis. We analyzed oil palm conversion from three land cover types: forest, grassland or pastureland, and cropland (differentiating between short-term and long-term crops). Oil palm species and hybrids included E. guineensis, E. oleifera, OxG interspecific hybrids, and non-specified species or hybrids. Regarding oil palm age, we classified processes occurring at the seedling stage of oil palm (less than two years old), the juvenile stage (2-6 years), and the mature stage (>6 years), where oil palm dominates the canopy. When papers compared processes between oil palm crops and other land cover types, we identified three types of analysis: pair-comparison (i.e. comparison among sites in different land covers), land use-land cover transition through time, and construction of land use-land cover change scenarios. We also extracted the type of approach the study followed, separating fieldwork and experimental studies from modeling approaches, either physical or statistically based.
We included the mean values of each variable analyzed for each relevant paper, the type of land transition analysis, and the direction of change (see supplementary data, table S4). We also included a summary of papers that did not include land transition and explored differences in fluxes within the plantation. Given the differences in approaches, units, and ways to present error metrics, values from different studies could not be averaged or aggregated. For papers that did not include results in tables or explicitly in the text but only in figures, we relied on the figures to extract maximum, average, and minimum values for the variable(s) analyzed.

Exclusion criteria and appraisal
Given the various databases used in this review, we discarded non-peer-reviewed papers. During the different filtering stages, we excluded papers that did not cover topics related to the objectives and guiding questions of the literature review. We filtered out papers referring to the industrial aspects of palm oil production, detailed agronomic studies, or food industry studies (including topics such as waste and yield management, pest control, and nutritional analysis of oil palm), as well as articles that did not include any explicit mention or analysis of processes related to oil palm. Additionally, papers evaluating impacts on biodiversity, sustainable palm oil certification, biomass and carbon budget, and nutrient uptake and cycling were excluded. While we acknowledge the relevance of carbon and nutrient cycling in ecohydrological processes as they relate to plant growth and development, our review is centered on water and energy fluxes and the variables influencing these processes following the scope of the definition found in Nuttle (2002) and detailed further in Moore et al (2015). To help with the analysis and to select the relevant papers, we defined a scale of relevance between 1 and 5. We define the reasons to consider the paper relevant based on one or more criteria: data reported, method implemented, unique in the region/country/area, unique in the type of ecosystem/biome evaluated, and relevant descriptions based on the guiding questions. Papers ranking between 3 and 5 were selected as relevant. (see supplementary material section 2).

Results
A total of 139 relevant articles remained in the analysis after the first (712 Papers), second (331 papers), and third filtering stages (figure 1). For the complete references of the 139 papers, see supplementary material, table S6. Of the 139 papers, 109 have not been included in any previous reviews that we used as references for evaluating knowledge gaps in ecohydrological research (Carr 2011, Comte et al 2012, Dislich et al 2017. Six references overlapped with Carr (2011); two with Comte et al (2012), and 25 with Dislich et al (2017). The oldest relevant study identified in this analysis was from 1992 (Dufrene et al 1992), after that year, research only started being published consistently after 2005, with a sharp increase in the number of publications after 2010. The number of publications decreased in 2020, potentially attributable to the global COVID-19 pandemic. Papers were collected systematically up to March 2021, which explains the low number of papers in 2021 (figure 2(A)). However, we included more recent relevant material in the preparation of the last version of this manuscript (16 papers, see supplementary material, table S6), though they did not undergo the same formal analysis as the papers that were systematically selected. Most of the selected literature was written in English, but we found three relevant papers written in Spanish and one in Portuguese.
Unsurprisingly, the vast majority of the 139 publications, 113 papers (81%), have studied areas in the Indo-Malayan and Australasian regions, mainly in Indonesia (58 papers, 42%) and Malaysia (35 papers, 25%) (figure 3(A)). In the Indo-Malayan region, the Jambi province (Sumatra) and Kalimantan in Indonesia hold the majority of the studies (20 and 18, respectively). The Australasian region, including Papua New Guinea (which is the sixth global oil palm producer), accounts for seven relevant papers, five of them specific to the region and two more that are part of a macroscale analysis. A substantial percentage of the 139 studies (49 papers, 35%) have study areas in the Neotropics, a region that has had an increase in publications since 2008. In this region, the state of Pará, Brazil (seven studies), and the Meta department, Colombia (three studies) comprise the largest number of studies (see the complete list in table S7 in the supplementary material). The Afrotropics region contribute the smallest number of studies, relative to its contribution to palm oil production (15 papers, 11%). Despite Nigeria's status as the world's fourthlargest palm oil producer, we only found two studies including this country: the review by Carr (2011) and a review on the environmental impacts of oil palm (Meijaard et al 2020). In terms of global studies, we found three covering the entire tropics and addressing land suitability for oil palm expansion (Paterson et al 2016, Pirker et al 2016, Vijay et al 2016, and one study that presents a global map of closed-canopy smallholder and industrial oil palm plantations (Descals et al 2021).
We found high variability in the temporal scale of the studies. It ranges from a few days to four years in experimental and field-based methods and multiple years in modeling approaches (data not shown). Most of the work with experimental and field-based methods had short-term data collection (days to months that sum less than a year). Regarding the spatial scale, the highest number of studies were performed at the plot scale, followed by regional and mesoscale. We found studies addressing land-atmosphere and fluvial processes at all spatial scales. In contrast, soil and groundwater studies were performed at plant to regional scales (figure 4 and figure 2(C)). Most studies identifying oil palm distribution use remote sensing techniques, especially at the regional scale (1 km 2 -100 km 2 ), while studies in fluvial and soils and groundwater processes use more field-experimental analysis (figure 4).
Modeling frameworks have also been employed to address ecohydrological processes. Some of the model frameworks used include the Soil and Water Assessment Tool, SWAT (Borah and Bera 2003), Tethys & Chloris (Fatichi et al 2012), SIMGRO (Querner 1997), SEBAL (Bastiaanssen et al 1998), and CLM-Palm (Fan et al 2015(Fan et al , 2019. SWAT has been used to compare the hydrologic response of different land covers (Heidari et al 2020), to identify variations in hydrologic fluxes in oil palm expansion scenarios (Babel et al 2011, Tarigan et al 2018, and to evaluate crop management mitigation options to reduce surface runoff (Tarigan et al 2016(Tarigan et al , 2020. The ecohydrological effects of oil palm age and its comparison to forest has been modeled using Tethys & Chloris in Manoli et al (2018) or with multiple plot fieldwork approaches . Subsidence and groundwater table depletion has been analyzed with SIM-GRO (Wösten et al 2006). Oil palm growth, canopy structure, yield, and their relations with water and energy cycles have been simulated with CLM-Palm (Fan et al 2015(Fan et al , 2019, a model built within the Community Land Model, CLM, framework. This model has been used at plot (Fan et al 2015), and regional scales (Fan et al 2019). Analyses of land conversion to oil palm plantation and its effects on ecohydrological processes appeared in 39 publications. Thirty out of the thirty nine publications performed paired comparisons among different land covers, six evaluated changes through time, and three conducted analyses   of land conversion scenarios. For forest conversion, the forest type is rarely specified in the literature (e.g. rainforest, mangrove forest, peatland forest, etc).
Within the literature on sociopolitical and sustainability aspects of oil palm associated with water (16 papers), we found a few examples that conduct a comprehensive analysis of community perceptions on water-related issues linked to oil palm cultivation (a total of four papers). Available studies use structured or semi-structured interviews with different stakeholders (including indigenous groups, smallholders, government officials, and private sector organizations) to identify positive and negative effects of oil palm cultivation and expansion on environmental and socio-economic conditions of local communities (Larsen et al 2014, Tittor 2017, Damiani et al 2020, Hervas 2020. Additionally, we found two studies, classified in the fluvial processes category, that use interviews with villagers and instrument data to empirically validate the communities' perceptions of water issues (Merten et al 2016(Merten et al , 2020. Several studies centered on identifying how different management practices impact hydrological fluxes. Specifically, these studies analyzed the reduction of runoff through silt pits and frond piles (Tarigan et al 2016(Tarigan et al , 2020, biopores (Devianti et al 2020), or understory vegetation (Satriawan et al 2016, Wawan et al 2019, and the benefits of riparian zones (Luke et al 2017, Chellaiah and Yule 2018a, Horton et al 2018.
About half (49%) of the examined studies did not report oil palm species and hybrids. We found occasional descriptions of oil palm varieties (i.e. variations of the same oil palm species). However, this information is rare in the studies. By looking closely at the categories of analysis in which the species were not reported, we found that 63.4% of the studies were classified in the oil palm distribution category, 24.3% in land-atmosphere processes, 7.32% in fluvial and 24.39% in soil and groundwater processes. In the 50% of the studies which did identify the species, we found the species reported as either E. guineensis (49% of studies) or the OxG interspecific hybrid (1% of studies, i.e. one study in Indonesia and one in Colombia) (figure 2(D)). In the Afrotropical region, the lack of peer-reviewed literature in English, Spanish or Portuguese impedes the establishment of overall oil palm-related deforestation trends at the national and regional scales, making it difficult to clearly assess current land-use trends. Nevertheless, future expansion of oil palm crops is expected to result in similar deforestation trends as in the Indo-Malayan region due to socioeconomic and political conditions that could facilitate uncontrolled crop expansion (Amigun et al 2011, Burton et al 2017, Folefack et al 2019. Some studies on land-use change corroborated these expected deforestation trends in specific study areas (Asubonteng et al 2018, Ordway et al 2019.

Global expansion of oil palm plantations
Available research on global oil palm expansion also points to differences in land tenure and cultivation methods in different biogeographic areas and their impacts on land-use change patterns and environmental services, particularly in smallholders versus large industrial plantations. For the Neotropics and Indo-Malayan regions, large-scale industrial plantations have the largest association with deforestation, while smallholders' plantations typically occupy already degraded land (Erniwati et al 2017, Hernandez-Rojas et al 2018, Glinskis and Gutierrez-Velez 2019. Nevertheless, in the Indo-Malayan region, smallholders cultivate between 30% and 60% of oil palm area (Meijaard et al 2020) and could become an important driver of deforestation. Schoneveld et al (2019) report that conversion patterns by smallholders in Indonesia could be changing towards higher deforestation because of the lack of incentives to subscribe to zero-deforestation agreements and lack of enforcement of the country's land management policies on smallholders. For the Afrotropical region, small-scale agriculture dominates palm oil production (Feintrenie et al 2016). In this region, oil palm expansion and related deforestation is mainly driven by smallholders (Ordway et al 2019). In Nigeria, the fourth-largest palm oil producer (FAOSTAT 2022), 94% of oil palm area is cultivated by smallholders (Meijaard et al 2020).

Land-atmosphere processes
Initial work on land-atmosphere processes in land transition investigated the use of water by oil palm relative to other crops, under different moisture conditions (Radersma and de Ridder 1996) and to forest cover (Luskin and Potts 2011). Given the relevance of deforestation in Indonesia and Malaysia, later works focus mainly on forest transition to oil palm (Hardwick et al 2015, Sabajo et al 2017, Manoli et al 2018 or in comparing oil palm with undisturbed forest and different stages of logged forest (Nainar et al 2018). More recently, in the Neotropics (Mexico), we found one study comparing pastures or grasslands to oil palm conversion (Heidari et al 2020). Additional studies have been performed comparing microclimatic conditions of oil palm with forest (Hardwick et al 2015. Changes in surface and air temperature, vapor pressure deficit (VPD), albedo, latent heat, and sensible heat fluxes relative to other land covers, and oil palm age have been mainly studied in the Indo-Malayan region ( Manoli et al 2018). In this region one study was reported on OxG interspecific hybrid (Fowler et al 2011). In addition, we found literature for this region assessing transpiration, ET, water stress, and effects of canopy structure in the hydrologic partition throughout oil palm plant or crop (Henson et al 2005, Legros et al 2009 or comparing this crop with other land covers (Fan et al 2015, Jucker et al 2018, Tarigan et al 2020. Recent literature also evaluates the microclimatic effect of including diverse vegetation in an oil palm agroforest system (Donfack et al 2021). In the Neotropics, available studies focus on assessing transpiration , Brum et al 2021 within the plantation or on comparing oil palm ET to other land covers (Heidari et al 2020). In the Afrotropics, we did not find recent studies on land-atmosphere processes, although we found work on ET in oil palm plantations in the context of irrigation and water stress (Dufrene et al 1992, Radersma andde Ridder N 1996).
The available literature on forest land transitions indicates that oil palm creates drier and warmer microclimatic conditions compared to rainforests, while other land cover transitions have been less studied (figure 5). In the early stages of crop establishment after replacing rainforest, open canopies caused the land surface temperature in oil palm crops to increase by 6.0 ± 1.9 • C in Indonesia (Sabajo et al 2017). Luskin and Potts (2011) reported mean diurnal air temperatures, at 0.1 m above the ground, between 2.03 and 4.08 • C higher in 8 and 22 years-old oil palm plantations, respectively, than in rainforests in Malaysia. Another study measuring temperature at 2 m height showed that diurnal air temperatures range between 22.2 ± 0.1-30.6 ± 0.3 • C in oil palm plantations with a closed canopy while temperatures in forests and agroforest (specifically jungle rubber) range between 22.4 ± 0.1-28.3 ± 0.2, and 22.2 ± 0.1-29.8 ± 0.3 • C respectively, indicating better climate buffering capacities for these land covers . Similarly, with measurements taken at 1.5 m height, old-growth-logged forests and nonlogged forests reported maximum hourly air temperature between around 3.5 and 4.5 • C lower than oil palm plantations (Hardwick et al 2015). Changes in air temperature also relate to relative humidity, which is lower in closed-canopy oil palm plantations (e.g. mean 91.3 ± 0.8% in Meijide et al (2018), and about 75% at the minimum value in Hardwick et al (2015)) than in forests (e.g. 95.6 ± 1.0% in Meijide et al (2018), and about 92% at the minimum value in Hardwick et al (2015)).
Regarding albedo, studies found only small differences within plantations, despite the observed changes in temperature. Albedo did not show significant changes between 1-and 12 years old oil palms (0.16 ± 0.02 and 0.14 ± 0.01, respectively) . Similarly, albedo showed a weak influence on the land surface temperature and very small differences between forest and oil palm plantations (Sabajo et al 2017). The short temporal scale of these studies does not allow to evaluate albedo through the entire oil palm life cycle or of seasonal variations.
As oil palm grows, canopy closure is responsible for the differences in surface roughness and temperature distribution across the vertical profile, which relates to changes in microclimatic conditions (Ramdani et (2017) reported air temperature differences of about 1 • C between 1-and 12 years old plantations in Indonesia. In that same study, the diurnal cycle of air temperature showed higher temperatures during longer periods for the more open 1 year old plantation. By modeling the ecohydrology of oil palm at plot scale using the Tethys & Chloris model and the data collected by Meijide et al (2017), Manoli et al (2018) found that surface temperature in juvenile oil palms ranges between 23 • C-26 • C, while temperature decreases to 22 • C-23 • C when oil palm is mature and the canopy is closed. In addition to canopy closure, stand structural complexity (i.e. the heterogeneity of vegetation structure) and riparian zone vegetation are suggested to have a key role in reducing temperature and increasing relative humidity when present in oil palm plantations (Donfack et al 2021, Williamson et al 2021 as taller and denser canopies are associated with lower mean surface temperature and lower VPD (Jucker et al 2018).
At the plant scale, previous reviews emphasized VPD's impact on stomatal openness, which regulates water and gas exchange (Carr 2011). More recent studies addressed the variability of these conditions across oil palm ages and the diurnal cycle. Within E. guineensis plantations, VPD is high in juvenile oil palm and exhibits more variability in mature plantations. A one-year-old oil palm plantation in Indonesia experienced a maximum VPD of around 16 hPa while the maximum VPD in a 12 years old oil palm plantation was around 14 hPa . Röll et al (2015) found VPD in oil palm plantations between 2-and 9 years old to be around 15.3 ± 0.95 hPa, and a high VPD variability in plantations with palms older than 10 years (15.4 ± 3.35 hPa). In a 12 days experiment during the wet season for a 5 years old E. guineensis plantation in Colombia, the average VPD was 26 ± 8.3 hPa (Bayona-Rodríguez and Romero 2016), similar to results found in a 12 years old oil palm plantation in Indonesia (maximum VPD around 30 hPa) during the wet season (Niu et  In a comparison among different oil palm ages, transpiration showed a rapid increase during the first seven years after planting, followed by a steady-state associated with canopy closure . Regarding ET, Röll et al (2015) used sap flow and eddy covariance measurements in two Indonesian plots and found that ET was 2.8 mm day −1 in a 2 years old oil palm and 4.7 mm day −1 in a 12 years-old oil palm (transpiration represented 8% and 53% of ET, respectively). An ecohydrological modeling approach reported ET between 1000 and 1600 mm yr −1 in juvenile and 1200-2000 mm yr −1 in mature plantations (Manoli et al 2018). Other studies reported annual ET of 918 ± 46 mm yr −1 in a 1 year old plantation, and a value of 1216 ± 34 mm yr −1 for a 12 years old oil palm plantation in Indonesia ; and daily ET in Indonesia at the regional scale 3.88 mm day −1 using the CLM model (Fan et al 2019).  soil moisture) which influence oil palm physiological response through its water use, and gas exchange, impacting transpiration and ET. Soil water deficit disrupts the metabolism of oil palm, altering photosynthesis and causing stomatal closure, which in turn reduces transpiration in E. guineensis (Rivera-Mendes et al 2016, Najihah et al 2019, Brum et al 2021). Although under drier conditions irrigation guarantees optimal palm oil productivity, a study in Brazil showed that plants adapt to conserve water under drought conditions (Brum et al 2021). Radersma and de Ridder (1996) tested different water supply conditions in a closed canopy oil palm plantation. They found that under suboptimal or drier conditions of water supply during the dry and wet seasons, ET is higher in oil palm plantations than in cacao, rice, and maize, except for cacao in the dry season. Hardanto et al (2017) tested different topographic (i.e. upland and valley topographies) and flooded conditions in rubber and oil palm plantations. They found high transpiration in oil palm located at the non-flooded valley compared to rubber plantations, potentially explained by the horizontal oil palm root distribution.

Differences in ET between oil palm plantations and other crops and forests varied among the studies examined. While most studies report higher ET in
We found a few examples addressing the effect of oil palm plantations on precipitation partitioning through quantification of throughfall (Dufrene et al 1992, Banabas et al 2008, Agusta et al 2019 and interception (Tarigan et al 2018, Fan et al 2019. Measured in the field, gradients of throughfall in an 8 years old oil palm plantation were found to be 35.3%-39.4% of the total precipitation (Agusta et al 2019), while Fan et al (2019) found interception loss between 17%-27% by using the CLM-Palm land surface model ( figure 6). In watersheds where forest has been converted to agriculture that include oil palm plantations, areas of oil palm experienced less precipitation during the dry season after conversion while an increase in precipitation occurs in the wet season, suggesting a change in intensity of extreme events (Adnan and Atkinson 2011) (figure 5).

Fluvial processes
Sediment yield increases in the early stages of oil palm establishment when compared to forest cover and agroforest (Carlson et al 2014, Nainar et al 2017. An increase has been also observed in soil erosion in oil palm compared to forest (Gharibreza et al 2013, Nainar et al 2017. Erosion and sediment concentration in streams increase due to oil palm establishment, after forest conversion. Reasons for these increases include the construction of roads (e.g. roads, harvesting paths) to facilitate crop management (Carlson et al 2014), and the absence of vegetated riparian zones (Carlson et al 2014). The loss of understory vegetation and low-stand structural complexity causes sediment yield to remain higher in oil palm crops than in forests, even in adult plantations (figure 6). For example, compared to rainforest, sediment yield increased with differences between 19 000 ± 3400 mg h −1 ha −1 in juvenile plantations and 8000 ± 2000 mg h −1 ha −1 in mature plantations (Carlson et al 2014). In another study, annual suspended sediment concentration discharge in an oil palm catchment was estimated to be 4-12 times greater than primary and multiple-logged forest catchments (Nainar et al 2017). Similarly, in a 5 years old oil palm plantation, erosion was higher (46 kg ha −1 ) than in forests (13 kg ha −1 ) (Jaya et al 2018).
Compared to other types of plantations, erosion and sediment transport have been found to be higher in oil palm crops. In a scenario-based analysis with oil palm and other biofuel crops, sediment loss was higher in oil palm compared to cassava and soybeans (Babel et al 2011). Similarly, compared to mixedspecies logged agroforest, stream sediment yield was greater in oil palm plantations more juvenile than three years old, and in plantations older than 10 years (Carlson et al 2014). Guillaume et al (2015) evaluated soil erosion at the plot scale in four land-use types, including forests, agroforest (i.e. jungle with rubber trees), rubber plantations, and oil palm plantations. They found that the oil palm and rubber plantations exhibited higher erosion (with a maximum of 35 ± 8, and 33 ± 10 cm, respectively) than the agroforest (14 ± 14 cm). The fact that the agroforest does not require land preparation makes this land use less susceptible to alterations in the soil conditions. We did not find studies that contrast sediment concentration in streams draining pasturelands with oil palm plantations ( figure 5).
Runoff, baseflow, and streamflow generation responses in oil palm plantations vary as a function of previous land cover, soil, and topographic conditions, showing a general increase compared to forest cover (Babel et al 2011, Algeet-Abarquero et al 2015, Tarigan et al 2016, 2018, Jaya et al 2018, Kurniawan et al 2018. However, trends are not conclusive in transitioning from cropland, grasslands, or pastures (figure 5). Nainar et al (2018) analyzed three types of land cover in the post-logging phase: two types of logged forest and a mature 22 years old oil palm, plus an undisturbed primary forest and virgin jungle reserve slightly disturbed in its surroundings. They found that total daily discharge, runoff, and baseflow were lower in oil palm plantations than in other land cover types. In contrast, in a study in Costa Rica, oil palm showed the highest runoff response (runoff coefficient 32.6%), twice the mean of grassland (runoff coefficient 15.3%) (Algeet-Abarquero et al 2015). This was attributed to soil compaction resulting from oil palm establishment and former grazing management. In another study, simulated runoff increased in oil palm plantations (31% of precipitation) when compared to forests (16% of precipitation) and rubber (22% of precipitation) in loam and clay acrisols (Kurniawan et Devianti et al 2020), as well as preserving riparian zones to control stream properties (Luke et al 2017, Chellaiah and Yule 2018a, 2018b, Merten et al 2020. These studies are motivated by the trends in runoff regeneration, and soil erosion, and soil, when oil palm replaces forests. The runoff coefficient in oil palm plantations decreased from 63% to 50% when a combination of frond piles and silt pits was used as a mitigation strategy (Tarigan et al 2016). Different understory vegetation can be planted to prevent changes in infiltration and runoff, and losses of fertile soils that are rich in organic matter (Satriawan et al 2016). Preserving or planting riparian zones contributes to decreasing stream temperature (Luke et al 2017, Chellaiah andYule 2018a), bank erosion while increasing oil palm yield (Horton et al 2018), and to increasing microbial activity (Chelliah and Yule 2018b) near oil palm plantations. Modeling approaches have also been implemented to find oil palm limits of expansion in watersheds associated with regulation capacity. Tarigan et al (2018) compared watersheds to identify how the baseflow index and the runoff coefficient vary under different dominant land covers and land cover changes in Indonesia. They found a percentage of area threshold (minimum 30% forest and maximum 40% oil palm), after which catchments can lose streamflow regulation capacity.

Soils and groundwater
Soil properties have been studied in peatland (Wösten et al 2006, Couwenberg and (Tarigan et al 2016, Damiani et al 2020. Elevated values of soil bulk density, an indicator of low infiltration rates and high compaction, is found oil palm plantations' soils compared to the soil in forests (Kenye et al 2019, Khasanah and van Noordwijk 2019, Tarigan et al 2020 and other croplands (Bruun et al 2013, Guillaume et al 2016. However, in grassland conversion, significant changes were found in soil pH and exchangeable magnesium, but no statistically significant variations were detected in bulk density, soil carbon, or nitrogen content (Nelson et al 2014). Within oil palm plantations, soil bulk density is highly heterogeneous and particularly high in the traffic zone (i.e. the zone in which oil palm bunches are collected) due to harvesting (Sato et al 2017). Bulk density also varies with soil depth (Matysek et al 2018), and plantation age (Guillaume et al 2016).
Infiltration rates in oil palm plantations can be improved by adjusting management practices. Infiltration rates have been measured for different land use and land covers and for different mitigation techniques to prevent erosion and overland flow in oil palm (Satriawan et al 2016, Tarigan et al 2016. In Indonesia, field measurements using doublering infiltrometers found low infiltration rates in oil palm plantations with no management treatment (around 10 cm h −1 ) and rubber plantations (around 20 cm h −1 ), while high infiltration rates were obtained in secondary forest (90 cm h −1 ) (Tarigan et al 2016). By implementing front piles as a mitigation strategy, the study reported an increase in infiltration to around 40 cm h −1 . Satriawan et al (2016) measured infiltration volume and estimated infiltration capacity (i.e. the maximum soil infiltration rate) in plots where palms were between 5 months and 2 years old, and under different understory vegetation used as a soil conservation technique to prevent erosion. They found that soil conservation techniques did not significantly affect infiltration capacity, while the infiltration volume increased significantly.
SOC is generally lower in oil palm plantations than in rainforest (Sommer et (2014), working in Brazil, measured SOC in mineral oxisol soils (using the USDA classification) under rainforests, mixed forests, and oil palm. They found that mixed forest has the lowest SOC among the land covers. However, within two different oil palm ages (23 and 34 years old), and across all the soil depths evaluated, they found lower SOC in the younger plantation and higher SOC in the older plantation. Couwenberg and Hooijer (2013) analyzed carbon losses and subsidence in acacia and oil palm plantations. Annual carbon losses are similar in acacia plantations and in 5-and 19 years old oil palm plantations when comparing the rate of carbon loss to the subsidence rate. SOC losses are also detected in repetitive plantation cycles after replanting (Matysek et al 2018). One study conducted in active deforestation regions of Indonesia, Cameroon, and Peru indicates that the more SOC present in the original forests, the more SOC lost after conversion (van Straaten et al 2015) (figure 6). However, analyzing SOC and bulk density together, Bruun et al (2013) found inconclusive trends when comparing oil palm with swidden systems, (i.e. rotational agriculture that regenerates the land after several cultivation cycles). The combined analysis of SOC and bulk density indicated that while SOC decreases in oil palm plantations when compared to swidden systems, an increase in bulk density does not allow conclusions about SOC losses (Bruun et al 2013). In the land conversion of grassland to oil palm in Papua New Guinea, significant changes were found in soil pH and exchangeable Magnesium, but no statistically significant variations were detected in bulk density, soil carbon, or nitrogen content (Nelson et al 2014). More recently, Quezada et al (2019) analyzed soil carbon stocks in a range of 60 years after oil palm conversion from pastures. While SOC in the first 50 cm of depth reported losses of 39 ± 8% Mg C ha −1 after 56 years of oil palm cultivation, it stabilized after this period in the second crop rotation.
Less studied in the literature are processes associated with groundwater dynamics in oil palm plantations. Typically, studies looked at the first 60 cm of soil depth. However, analysis of water table variation and subsidence has been done in peatland soils (Wösten et al 2006, Couwenberg andHoojier 2013). Due to drainage from previous oil palm establishments, depth to water table significantly decreases in oil palm plantations, causing subsidence. For example, Wösten et al (2006) found subsidence increasing between 2 and 3 m reducing water table and catchment discharge. Another study found that constant water drainage in oil palm plantations led to water table subsidence (between 3.7 and 3.9 cm y −1 ) (Couwenberg and Hooijer 2013). Water table reduction and subsidence have an impact on soil water storage. In Malaysia, a study reported higher soil water content in peat swamp forests compared to mature oil palm (Tonks et al 2017) (figure 5).

Discussion
Our review summarizes relevant findings in oil palm research associated with ecohydrological processes. These findings are in the context of changes in energy and water cycles due to oil palm plantations in the three main components of the hydrologic cycle: (1) land-atmosphere, which covers processes within the troposphere; (2) fluvial systems covering streams and land processes associated with erosion; and (3) soils and groundwater. We included oil palm distribution, as identifying patterns of expansion and distribution is an integral part of understanding ecohydrological processes. In addition, given the nature of agricultural systems, we explored literature on the social aspects of oil palm associated with perceptions of water availability and sustainability. All these aspects were analyzed in the context of three land transition groups (oil palm replacing or compared to forests, planted forest and short-term cropland, and pastures and grassland) and oil palm plantation ages. We also reported on the differences between oil palm species and their hybrids, when reported in the literature, and the mitigation strategies implemented to reduce negative ecohydrological impacts. Through this review, we have identified the following five major challenges that require additional research to advance the field.

To identify land use change trajectories in understudied regions
Due to the environmental impacts of massive deforestation in Indonesia and Malaysia from oil palm establishment, oil palm transition from forests has been extensively studied. Available literature on the transition from short-term and permanent crop plantations is scarce, particularly in short-term croplands. This lack of information impedes identifying land trajectories and land fragmentation. Those trajectories may differentially impact ecohydrological processes at the meso-and macro-scales. We recommend increasing attention in characterizing these trajectories since these are essential inputs to examine long-term effects of oil palm development (e.g. understanding the drivers of oil palm water use and its contribution to ET). A global dataset of closedcanopy oil palm plantations (Descals et al 2021) represents a valuable tool for testing ecohydrological questions at the meso-and macro-scales.
While we recognize the progress made in identifying the impacts of oil palm establishment, development, and expansion in ecohydrological processes, we recommend increasing the attention to regions outside Southeast Asia. This is particularly critical in the Afrotropics, where socioeconomic and political conditions could promote uncontrolled crop expansion (Amigun et al 2011, Burton et al 2017, Folefack et al 2019. Although plantations in the Afrotropics represent 19% (around 5.3 million ha in 2020) of the global oil palm harvested area (FAOSTAT 2022), only 10% of the relevant publications have study areas in this region. This uneven distribution strongly contrasts with the Indo-Malay region, which has 70% of the global oil palm plantations (FAOSTAT 2022) and accounts for 81% of the relevant literature. Accurate information on land-use trajectories for the Afrotropics region should be a relevant research objective in the near future, particularly considering the potential adverse effects of oil palm-driven deforestation on biodiversity and carbon stocks (Wich et al 2014, Burton et al 2017.

To advance in the understanding of ecohydrological consequences of different land transitions and oil palm ages
We found progress in addressing the gaps reported in previous reviews regarding this topic. More attention has been dedicated to evaluating biophysical changes (e.g. albedo, surface energy fluxes, and microclimate) associated with land use change, particularly in the transition from forests to oil palm or in the comparison between them. The results from these studies indicate that oil palm plantations are warmer and drier than forests, while no clear changes in albedo have been observed. Warmer environmental conditions are noticeable from the early stages of oil palm establishment all the way to mature oil palm stages. Drier conditions are observable in oil palm plantations where relative humidity is lower than in forests at the hour of maximum temperature. However, it is not possible to identify patterns in non-forest land transitions as studies have yet to address these changes (particularly outside the Indo-Malayan region, where plantations mostly replace land cover different from forests). Progress in these lines of research will facilitate the creation of comparable or contrasting scenarios relative to forest conversion and the inclusion of those insights in the identification of limits of oil palm expansion.
Notably, the effects of oil palm plantation turnover required after a certain age and considering different environmental conditions are understudied. Research on plantation turnover may help to differentiate the effect of land transition from the effects of oil palm development after establishment. Importantly, the effects of land transition (e.g. deforestation) do not disappear right after oil palm is planted, but rather water and energy fluxes continue to be affected by the legacies of the previous land cover, the extent of disturbance, and the presence of the new crop. Nevertheless, long-term records of measurements and future research could help in separating the relative importance of these effects.
The second major gap previous reviews have indicated is the need to identify changes in the water balance partitioning, such as throughfall and rainfall interception across different oil palm ages, and ET at immature oil palm stages. We observed an increase in the literature addressing these topics, particularly ET at immature oil palm stages. We recommend directing attention to identifying plant transpiration under different environmental conditions and oil palm species and hybrids. In addition, increasing the spatial resolution of ET analysis will contribute to a more accurate assessment of water balance in the tropics. In this regard, remote sensing and earth system modeling offer opportunities for improvement (e.g. Fan et al 2019, Ellsäßer et al 2020. Given its high water use, oil palm establishment in grasslands may lead to decreases in soil moisture and groundwater recharge, which may potentially impact water availability (Cunningham et al 2015). Current studies on the transformation of peatlands into oil palm evaluate carbon losses and stocks, as well as subsidence. In addition, expansion of oil palm into non forested land, former pastures and grasslands has the potential to be beneficial in terms of carbon sequestration (Quezada et al 2019, Garcia-Ulloa et al 2012. However, research is needed to identify the hydrological effects of this expansion. Similarly, new research needs to include the effect of draining peatlands on streams (e.g. changing water quality), changes in soil hydraulic properties and groundwater recharge, as well as variations in groundwater storage. Overall, additional research is needed to understand hydrological changes due to the transformation of peatlands and grasslands into oil palm plantations.
In fluvial processes and runoff generation, besides addressing different environmental settings, an improved understanding of how oil palm water use places pressure on community water supply is needed, given the high water requirements of mature oil palms. This understanding includes evaluating irrigation systems and their dynamics during the wet and dry seasons, particularly in agricultural regions where oil palm is not the only crop.

Increasing long-term monitoring of ecohydrological processes
We found a gap in addressing long-term oil palm ecohydrological processes relative to evaluating microclimatic conditions under different climatic settings, land uses and soil types; plant transpiration in different oil palm species and hybrids; runoff and sediment transport generation, especially contrasting different configurations of riparian zones; alteration of soil properties; and groundwater depletion, particularly in peatlands. Most studies were only performed for very short time periods, with a few exceptions in the recent literature. Given that ecohydrological responses vary with oil palm age, long-term analysis of ecohydrological processes is necessary, including oil palm rotation cycles. This analysis would allow quantifying trends of change and the effect of oil palm age and replanting cycles. This quantification is important to dissociate land transition impacts from the impacts caused by long-term oil palm establishment.
We recognize the complexities and costs of long-term ground monitoring systems. Partnerships with oil palm growers at industrial and smallholder plantations may contribute to addressing these challenges, which further will help mitigate adverse long-term environmental effects. In this sense, promoting multipurpose long-term monitoring programs with open-data access aims to characterize the plantations and address local needs at the same time.
These monitoring programs may increase the spatiotemporal distribution of ground observations.

To recognize the importance of specifying the effects of different oil palm species and hybrids
We found a lack of specification of oil palm species in the literature ( figure 2(D)). While the limited information about the species studies is understandable given the historical commercial dominance E. guineensis, commercial attention to hybrids is growing rapidly as varieties of oil from OxG interspecific hybrids are being recognized for their nutritional properties (Mozzon et al 2020). There are currently very few studies with an ecohydrological focus on these hybrids, and the ecohydrological implications of their use are largely unknown.
Most of the literature on OxG interspecific hybrid suggests it has been cultivated mainly in the Neotropics (e.g. nearly 80 000 ha have been cultivated in Colombia (https://web.fedepalma.org/), although we did not find reports that map the distribution of hybrids. Only one study in Indonesia was reported on a plantation growing the OxG interspecific hybrid (Fowler et al 2011). Given the number of studies where species or hybrids are not reported, it is possible that OxG interspecific hybrids are planted in regions other than the Neotropics.
Some of the agronomic characteristics of OxG interspecific hybrids that make it commercially attractive may have different effects on ecohydrological fluxes compared to those identified in E. guineensis. For instance, OxG interspecific hybrid has a slower growth than E. guineensis, which increases its life cycle and makes it easier to harvest, and it has higher environmental adaptability (Barcelos et al 2015). To determine the constraints of OxG interspecific hybrids expansion and the strategies in water management, it is necessary to address the effects on land suitability and ecohydrological processes in these plantations in comparison with E.guineensis. Thus far, work has been concentrated on understanding physiological (Rivera-Mendes et al 2013, Bayona-Rodriguez et al 2016) and phenological (Rivera-Mendes et al 2012) aspects of OxG interspecific hybrids varieties at seedling and nursery oil palm stages. However, there have been very few studies at the plantation, regional, meso-and macro-scales.
To overcome these knowledge gaps, we suggest: (1) increasing the reporting of the specific species studied in research publications; (2) additional studies evaluating the impacts of hybrids in ecohydrological processes, the differences in ecohydrological responses to agricultural practices, and comparing ecohydrological responses between E. guineensis and OxG interspecific hybrids. Additional understanding of the specific ecohydrological impacts of different species and hybrids would help develop better management practices and suitable sustainable conditions for oil palm expansion.

Increase the evaluation of linkages among social dynamics and ecohydrological processes
We found a limited number of articles that specifically address the perceptions of local communities on the links between oil palm cultivation and alterations in ecohydrological processes. This research gap indicates that local knowledge and community perspectives in oil palm-producing regions still need to be more present in questions motivating the scientific literature. These perspectives can help advance knowledge of hydrological and ecological processes in oil palm research while also linking social and environmental dynamics.
Including communities' experiences, knowledge, and perceptions in research design is of high relevance, particularly considering that local communities clearly identify alterations in ecohydrological processes due to oil palm. Examples of this identification include surface and groundwater quality degradation and water scarcity (Larsen et al 2014, Merten et al 2016, Tittor 2017, Damiani et al 2020, Hervas 2020), water contamination due to erosion and runoff from oil palm plantations and mills and toxins from pesticides (Damiani et al 2020), reduced surface water in rivers during the dry season and declining water tables in land adjacent to oil palm plantations (Larsen et al 2014), flooding related to deforestation with increased peak flow, flash floods in the rainy season, and reductions in fishing capacity (Tarigan et al 2016, Merten et al 2020. This progress indicates that local knowledge could help identify ecohydrological impacts that are not being tracked by current research. Additionally, actively engaging local communities in research could lead to a better design of sustainable practices and a better commitment of the community to those practices. Despite the challenges associated with interdisciplinary research, involving local actors in evaluating ecohydrological processes in agricultural settings provides a more holistic way that may positively impact their local communities and the environment (Reiss-Woolever et al 2021).

Limitations of this review
Though the inclusion of literature in Spanish and Portuguese highlights local academic production in the Neotropics in different environmental conditions, we could not address scientific literature written in all the languages of tropical regions. We recognize that including all the scientific literature produced in the field would require a multi-language search wider than the one we performed, and this limitation may have hidden progress made in other regions (e.g. Afrotropics). However, the number of studies in Spanish and Portuguese was comparatively lower than the number of studies in English. This may be an indication of a still low scientific production published even in local journals.
Because we focused on water and energy cycles, we did not include nutrient cycles to describe ecohydrological mechanisms associated with plant growth. The extensive literature on this topic that overlaps environmental and agronomic studies would require a separate systematic review that also considers a multilanguage approach. As an observation, during the first stage of this review, we identified a number of agronomic studies written in Spanish and Portuguese, perhaps reflecting the local interest in these topics.
This systematic search and review allowed us to identify the progress in the field for various spatiotemporal scales, different approaches, and a range of oil palm ages. We extracted the range of the average values of each of the ecohydrological fluxes (supplementary data table S4). This extraction, along with the description of the current knowledge, allowed us to present a synthesis of ecohydrological processes within the crop (figure 6) and relative to other land cover types (figure 5). Still, the heterogeneity of the studies across the tropics did not allow us to quantify the overall trends of ecohydrological processes. Literature on ecohydrological processes has been increasing in the last decade. As literature continues increasing, soon it will be possible to perform meta-analysis of the research produced and to identify, with adequate degrees of accuracy, the magnitude of ecohydrological changes.

Concluding remarks
Our study synthesizes progress toward understanding ecohydrological processes in oil palm plantations at different temporal and spatial scales across the tropics. We focused our search on different scenarios of land transition, the three main components of the hydrologic cycle, and different oil palm species and hybrids (i.e. the commercial African oil palm E. guineensis Jacq. and the OxG interspecific hybrid (E. oleifera cortés-E. guineensis Jacq.). We found important efforts that have identified hydroclimatic processes in oil palm related to plant transpiration, shifts in air temperature, and changes in water vapor pressure within different stages of oil palm growth and different land covers. Still, most of the progress has been at the plot and regional scale and at shortterm temporal scales. Similarly, most of the work to understand runoff and sediment yield has been done at regional scales or smaller. Although we recognize the logistics and cost-related complexities of longterm monitoring in oil palm plantations, this lack of continuous data constitutes a limitation to understanding ecohydrological processes. While we found examples in which community perceptions have been included while addressing scientific questions and attending to community concerns, this approach is still very limited considering the human-dominated nature of oil palm. In addition, as oil palm expansion is likely to increase in countries outside the Indo-Malayan region, it is becoming necessary to increase the evaluation of ecohydrological processes in these countries. More studies are needed on oil palm interspecific hybrids, which are likely to continue expanding due to their commercial value and resistance to extreme environmental conditions (i.e. plant disease, drought, and flooding). Critical thresholds of water use requirements in water-limited scenarios need further exploration, especially in landscapes where oil palm is not the only crop, crops require irrigation to gain optimal yields, and land transitions different from forests have occurred. Further research in these types of landscapes will help to identify the limits of oil palm expansion from an ecohydrological perspective.

Data availability statement
All data that support the findings of this study are included within the article (and any supplementary files).

Acknowledgments
We acknowledge the support of the Minciencias-Fulbright fellowship, Colombia, awarded to Angélica M Gómez and Adriana Parra, and the Faculty for the Future fellowship awarded to Angélica M Gómez. Special thanks to Catherine Lopera and to undergraduate student Sofia Vela at Universidad de Antioquia who contributed to populating the supplementary tables S4 and S5. Thanks to Giovanny Ruiz for his contributions to the graphic design of figure 6.