Why Are Seed Dispersal Models Rarely Used? Limitations of Scalability and Improvement Measures

Abstract

Seed dispersal studies have primarily relied on prediction model methods, which limit clarity regarding how variables affect movement direction. Suggestions for improvement are limited to specific domains due to the lack of quantitative evaluation of the comprehensive scope of the basic steps, from explaining seed dispersal to its application in related fields. This study aimed to reconsider the classification of seed movement mechanisms, identify research trends from an integrated perspective, and discuss the current implications and improvement measures. The data included 240 studies related to seed dispersal across the observation, generalization, and application phases. By classifying the mechanisms based on the direction of movement, the main variables affecting movement due to gravity or wind can be clearly distinguished. Although seed dispersal models assume gravity as the core principle, only 12.91% of the studies addressed terminal velocity, and a mere 1.25% measured the diaspore area, both of which are essential for accurate prediction. In addition, attempts to utilize seed dispersal for natural regeneration were relatively frequent (14.58%), but they remained at the empirical model stage, relying on data collected in the field, and thus lacked connection with models developed in previous studies. The requirements for enhancing the field usability of seed dispersal based on the current review include (1) improving the data collection system for securing primary data, (2) collecting sufficient field data, and (3) developing a unified model that can be applied to various conditions and species.

1. Introduction

The social demand for seed dispersal has gradually increased for the natural maintenance of vegetation. However, the usability of the results derived from previous research is insufficient because systematic information has not been provided. Seed dispersal is one of the main links between seed production and germination during the process of maintaining and expanding the regeneration niche [1] (Figure 1). As the sustainable maintenance of plant communities comes to the fore through international agreements, such as the Convention on Climate Change and UN Sustainable Development Goals (SDGs), understanding the vegetation succession process is becoming increasingly important [2,3]. While plant communities provide a stable carbon sink, human resources and costs are consumed during the forest management process. Recently, natural regeneration methods using seed dispersal from parent plants have gained attention as a means of reducing consumption costs and maintaining genetic diversity [4,5]. The mechanism by which seeds move has been studied through various approaches; however, the available information is fragmentary, hindering its application in the field. Limitations in the field of seed dispersal stem from the lack of available data, the narrow scope of application for developed models, and the absence of quantitative assessments of detailed research topics. Existing studies, including review papers, have failed to provide comprehensive information because of limitations within specific parameters.

Because the current seed movement classification framework is categorized by the method of model establishment, the description of the main variables and methods according to the movement direction is insufficient. In previous studies, the differences in model algorithms were the main focus in the process of estimating the seed dispersal distance. Although some differences exist in the classification criteria depending on the review paper, they are comprehensively classified into the following frameworks: theoretical, empirical, mechanistic, Eulerian, and Lagrangian models [6,7,8]. This classification offers valuable information to help understand each model in terms of flight distance. However, the seed dispersal distance model was developed as a combination of vertical movement, which determines the fall of seeds from the seed tree, and horizontal movement, which is influenced by external factors. Therefore, to estimate the horizontal movement, the vertical movement must be understood. However, since the main variables that rely on the direction of movement have not been clarified, the physical properties of the seeds and the influence of external forces, such as gravity and wind, have not been strictly distinguished. This not only makes it difficult to easily recognize the importance of each variable but also results in insufficient foundational knowledge for understanding the characteristics of each model.

Seed dispersal has received less attention than other areas of forest regeneration, resulting in a lack of measurement data and an absence of quantitative evaluations for detailed studies in the field. In previous studies, seed dispersal has been used to scientifically elucidate the process of plant succession, and studies on the terminal velocity and dispersal distance according to the physical characteristics of seeds have been conducted. The flight characteristics of seeds are affected by the terminal velocity and drag. Because these can be used as decision variables in situations that exclude environmental variables, they are treated as important in several studies [9,10]. However, terminal velocity measurements have only been performed by a few researchers, and little data are available on the drag-generating wing area [11,12,13]. In addition, because most measurement data have been concentrated on mid-latitude plants, determining whether regional differences exist is challenging. The fundamental limitation of primary data deficiency is particularly evident in field-based research. Because field cases are insufficient, explaining seed dispersal in complex vegetation comprising multiple species is challenging, and each study remains at the level of presenting regression equations using independently measured data. Field-based data should be used to overcome these limitations [14,15,16]. Although basic research is lacking in the area of seed dispersal, the research stages have not been systematically classified and quantitatively evaluated. Detailed studies related to seed dispersal encompass a wide range of subjects, and independent studies have been conducted in various fields. Therefore, differences have been explained based on model characteristics in previous review papers [8,17,18]. This framework distinguishes qualitative approaches effectively but does not provide a comprehensive overview of the field of study. Research areas can be broadly divided into observation, generalization, and application and can be evaluated by quantitatively analyzing the outcome for each item (Figure 2). However, since no such attempts have been made in the field of seed dispersal to date, scientific evidence to support directions for future progress is lacking.

The purpose of this study is to identify the reasons why the results of previous studies on seed dispersal are seldom applied in the field and to discuss improvement measures. The detailed objectives are as follows: (1) to identify the key variables required for seed dispersal prediction based on movement direction, (2) to deduce implications through quantitative evaluation across detailed categories within the field, and (3) to propose improvement measures to enhance the field usability of existing research.

2. Materials and Methods

We reviewed papers published in English primarily using three search keywords: (1) ‘seed dispersal’, (2) ‘terminal velocity’, and (3) ‘wind dispersal’. Our study focused on seed dispersal by wind, and thus, the review excluded studies investigating non-wind seed dispersal processes, such as autochory, hemerochory, nautochory, and zoochory. A total of 240 papers were included after ensuring that each paper contained seed dispersal by wind from data collection to field application (Figure 2, Table S1).

Firstly, we reviewed the methodologies in the published literature to extract the core variables that affect seed dispersal. Dispersal mechanisms were divided into reflecting vertical and reflecting horizontal seed movement. The characteristics and representative results of each were summarized based on physical principles and research methods. Detailed quantifiable variables could be extracted because studies tended to use equations or utilization models rather than statistics. The major variables affecting seed dispersal were identified based on criteria commonly appearing in papers and deemed essential for predicting dispersal distance.

Secondly, seed dispersal and related research trends were categorized into three research phases: observation, generalization, and application [19]. This classification system has been commonly used in the social sciences (e.g., education) [20], and our study is the first to apply it in the realm of seed dispersal research. In terms of seed dispersal, the observation phase was defined as research aimed at collecting baseline data and analyzing seed movement patterns; the generalization phase comprised research incorporating quantitatively analyses and the establishment of prediction models based on environmental conditions; the application phase was considered research investigating the application of seed dispersal models in field studies. The secondary dispersal literature included in this paper were studies explaining the mechanisms by which seeds moved by wind undergo additional movement [21,22,23]. Therefore, it can be considered as a case that utilizes the results of seed dispersal by wind, and it was analyzed by including it in the application phase because it affects the final destination of the seeds. Based on these criteria, papers in each category were evaluated quantitatively to identify research trends.

3. Review of Seed Dispersal Mechanisms

Seed dispersal could be broadly divided into vertical and horizontal seed movements. The main variables influencing seed dispersal include seed morphology, horizontal wind speed, release height, and terminal velocity [7,24] (

Table 1. Methods and associated variables used to estimate seed dispersal distance.

Mechanism	Parameters	Independent Variables	Main Objective (+) and Challenge (−)
Vertical seed movement	Wing loading (disc loading)	· Seed mass · Wing area	(+) Identifying the physical properties of seeds to understand terminal velocity for each species (−) Difficult to apply to multiple species due to lack of diaspore data
	Drag force	· Seed mass · Air velocity · Seed velocity · Air density · Drag coefficient	(+) Measuring the influence of wind on seed movement and changes in terminal velocity (−) Requires large-scale experiments, and modeling is difficult because environmental conditions influence variables
	Leading-edge vortex	· Wing thickness · Wing curvature · Aspect ratio · Ratio of wingspan to chord · Center of rotation	(+) Determining seed rotation mechanisms according to wing shape and curvature (−) Lacks sufficient research cases and requires standardized measurements and general models
	Reynolds number	· Air viscosity · Air density · Seed diameter · Terminal velocity · Seed length, width, and thickness	(+) Identifying the airflow pattern through moving seeds as determined by inertial and viscosity forces (−) Requires micro-measurement equipment and facilities, and values are influenced by environmental conditions
Horizontal seed movement	Dispersal distance	· Wind speed · Seed mass · Wing loading · Release height	(+) Estimating seed dispersal distance according to wind speed and landscape topography (−) Field surveys are costly and time-consuming; data are mostly gathered via interior wind tunnel tests
	Dispersal trajectory	· Horizontal wind · Turbulent wind in vertical updraft · Release height · Terminal velocity	(+) Identifying differences in seed flight trajectories according to seed shape (−) Research on various seed types and evaluations of trajectory changes based on wind speed are lacking
	Long-distance dispersal	· Wind speed · Terminal velocity · Wing loading · Upward vertical air movement	(+) Measuring mechanisms of long-distance seed dispersal (−) Low model accuracy due to complex turbulence variations influencing vertical wind speed

). In the reviewed studies, seed dispersal was investigated using either some of the aforementioned major variables or by applying new variables, depending on the research purpose [25]. Because different variables and equations have been applied to diverse methodologies, comprehensively ranking the importance of the variables and methodological characteristics is challenging. Therefore, in this study, we reviewed only representative mainstream methodologies for deriving the key variables that affect seed movement.

3.1. Vertical Seed Movement

The vertical movement of seeds has mostly been studied using indoor wind tunnel experiments that assess the ratio of seed mass to terminal velocity. This ratio was conceptually subdivided according to the characteristics of the seed wings. Wing loading, which is the weight over the wing area, has been applied to seeds with relatively large wings, resulting in a high loading value [12]. Disc loading, defined as the weight over the rotary disc area, is caused by diaspores that affect rotation and tend to display low loading values [13]. The values of these parameters have been calculated based on the equations developed by Azuma and Yasuda (1989) [26] and Minami and Azuma (2003) [27], separating the wing and disc loading. The authors proposed the following formulas for a simplified disc area calculation:

$$Wing loading={\displaystyle \frac{m\times g}{wing area}},$$

(1)

$$Disc loading={\displaystyle \frac{m\times g}{disc area}},$$

(2)

$$Disc area={\displaystyle \frac{1}{4}}\pi D^2,$$

(3)

where $m$ is the seed mass (mg), $g$ = 9.81 (m s⁻¹) is the gravitational acceleration, and $D$ is the diameter of the seed wing or the pappose disk. The seed area, which included both the seed and its diaspore (which could generate a drag force), was used to represent the wing area. Technological advancements have led to a transition from direct human measurements to the use of scanning techniques to measure seed traits. Micro-computed tomography has recently been introduced to reduce measurement errors and enhance scientific accuracy, and attempts have been made to apply techniques such as scanning electron microscopy to quantify the morphological characteristics of fine seeds [28,29]. Should the time and cost issues related to acquiring diaspore areas be resolved, the establishment of such technical advances would enable an extensive analysis of seed dispersal in future studies.

Some studies have approached seed dispersal from an aerodynamic perspective, in which the vertical movement of seeds is affected by the drag caused by the seed shape and Reynolds number. The drag force resists the movement of an object in a fluid. Generally, the drag coefficient is approximately 0.09 for a streamlined half-body and 0.04 for a streamlined body [30]. In the context of seed dispersal, the drag coefficient is influenced by the degree of bending, shape, and the area of the seed wings. Gan et al. (2022) [31] confirmed that both the flight trajectory and terminal velocity of a seed change according to its location within the diaspore and the curvature of the wing. Therefore, understanding the complex relationship between seed shape and vertical movement is critical because even minor morphological variations can alter the determinants of the drag coefficient. The drag coefficient generated during seed dispersal was calculated as follows [32]:

$$f_{D,s}={\displaystyle \frac{\rho C_{d,s}A}{m}}\left|V_a-V_p\right|\left(V_a-V_p\right),$$

(4)

where $f_{D,s}$ is the drag force acting on the seed, $\rho$ is the air density, $C_{d,s}$ is the drag coefficient acting on the seed surface area $A$, $m$ is the seed mass, $V_a$ is the instantaneous air velocity adjacent to the seed, and $V_p$ is the seed velocity. According to this equation, variations in the seed shape affect the drag coefficient, which increases as the difference between the seed weight and wing shape increases. Jung and Rezgui (2023) [33] estimated the maximum lift coefficient for one-winged Acer nicoense seeds, considering variations caused by the revolving or flapping motions of their wings, and measured the zero-lift drag coefficient to be 0.015 to 0.06. Zero-lift drag is the drag that occurs due to the structure of an object; the higher the coefficient value, the greater the drag force that interferes with the airflow. The zero-lift drag coefficient of Acer nicoense seeds was similar to that 0.01–0.03 calculated for aircrafts [34]. However, unlike the scenario in an aircraft, in which an object moves by propulsion, seed dispersal depends on the natural wind speed, and the curvature and shape of the seed wings differ between species. Therefore, the applicability of this variable to explain the dispersal mechanisms of various plant species is limited.

Research has further confirmed the influence of seed wing shape on dispersal, focusing specifically on the leading-edge vortex (LEV). A LEV study was first introduced by Lentink et al. (2009) [35], and their technique has since been developed via wind tunnel experiments as a method for measuring LEV in various species to establish a dynamic model. However, the study of LEV in samaras is a relatively recent area of research, and the exact mechanism remains under discussion [36,37,38]. LEV research mainly investigates the rotating mechanism in seeds and attempts to increase explanatory power by considering detailed morphological traits, such as the surface roughness of fibrous wings or the greater thickness of the leading edges [13].

Another approach quantified the properties of air passing through a seed using the Reynolds number [39], which is calculated using the following equation [40]:

$$R_e={\displaystyle \frac{{\rho }_a{D}_g{V}_t}{\mu }},$$

(5)

$$D_g={\left(LWT\right)}^{\frac{1}{3}},$$

(6)

where ($R_e$) is the Reynolds number, ${\rho }_a$ is the air density (kg m^–3), $D_g$ is the geometric mean diameter of seeds, $V_t$ is the terminal velocity (m s^–1), $\mu$ is the air viscosity at room temperature, $L$ is seed length (mm), $W$ is seed width (mm), and $T$ is seed thickness (mm). The Reynolds number represents the ratio between inertial and viscous forces, thereby providing an index that describes the airflow pattern through the relative dynamics of these two forces [41]. In general, a small Reynolds number indicates that the laminar flow is dominant because the viscosity is higher than the inertial force. Conversely, higher numbers indicate that the relative contribution of the inertial force increases, the turbulent flow predominates, and the critical Reynolds number is generally assumed as 2100 [42]. In the context of plant seeds, Stevenson et al. (2015) [43] obtained the Reynolds numbers for symmetric double-winged seeds (1313 ± 410), asymmetric double-winged seeds (1095 ± 355), and single-winged seeds (1028 ± 265). The research results confirmed that the airflow passing through the seed wing approximated laminar flow. Therefore, it was concluded that viscous forces dominated the dispersal of winged seeds, and flight times increased owing to the influence of wind.

3.2. Horizontal Movement of Seeds

The mechanism of horizontal seed movement was elucidated through wind tunnel experiments and field investigations. Wind tunnel experiments were typically conducted indoors, where external factors can be easily controlled. They were primarily used to estimate seed dispersal distances or wind disturbances resulting from wind speed interactions with small-scale vegetation and terrain models. In contrast, although environmental variables could not be controlled in field surveys, such experiments provide real-time documentation of phenomena in actual ecosystems. However, conducting field studies requires a relatively large amount of time and a costly budget and has therefore been scant [44]. Wind speed, wind direction, and terrain gradients have been determined to affect the seed dispersal distance in seeds of the same plant species [45,46,47]. However, when environmental variables are controlled for, dispersal distances and trajectory changes related to seed weight, wing loading, and release height can be derived [48]. As these variables change according to the location and time of their measurement, generalized models are required to integrate them with seed dispersal predictions.

The theoretical classification of seed dispersal modeling was largely divided into phenomenological models, mechanistic models, and empirical models. It has evolved into more detailed algorithms with high accuracy, based on the physical characteristics of the seeds used in model development [8,17]. Therefore, various models exist depending on the seed shape, and they tend to show high accuracy for specialized seed types for each model [7,9,49]. However, achieving high accuracy requires detailed information, such as the vertical turbulence, leaf area index, and density and shape of seed appendages, which limits the practicality of these models for various taxa.

Efforts to incorporate external environmental factors into seed dispersal models have evolved to quantify the effects of factors influencing wind flow. The inflection points of vegetation and slope surfaces increased air turbulence by generating turbulence in the atmospheric layer, which was eight times the volume of the vegetation canopy [50,51]. Previous studies identified dispersal mechanisms using large-scale object and point-measurement data. More recently, the relationship between plant systems and wind dynamics using regional-level field data has gained prominence; however, the development of related research remains slow because of insufficient field data being available [25,52]. Large irregularities in wind direction data (with the exception of information on prevailing winds) have limited the researchers’ ability to make generalizations. Accordingly, the effect of wind direction on horizontal seed movement was only considered in a small number of studies, and no generalizable trends were discerned [53].

The flight trajectories were affected by horizontal winds and turbulent fluctuations in vertical updrafts, and higher release heights and slower terminal velocities were associated with a greater probability of trajectory fluctuations. Zhu et al. (2019) [54] showed that the seed and diaspore morphologies of seven Calligonum species can be used to classify seed flight trajectories into horizontal projectile, projectile, straight line, and concave curve types. Calligonum diaspores comprise wings, bristles, balloons, and thorns, with prominent inter- and intraspecific variations. However, winged seeds were limited to round-winged seeds in this genus; therefore, the algorithm developed by Calligonum cannot be applied to seeds with other wing types that display large wing areas, such as one-winged or gliding-winged seeds. Nevertheless, considering that different flight trajectories were discernible within the group of round-winged seeds, it is plausible to infer that trajectory differences were also present in seeds with large wing areas.

Attempts have also been made to quantify long-distance dispersal (LDD), which is a key factor controlling invasion and migration rates during ecosystem regeneration [55]. Elucidating the LDD mechanism is challenging because of the complex fluctuation of turbulence in the vertical wind speed. Accordingly, attempts to estimate the LDD could be categorized based on whether upward vertical air movement is a model variable [56]. Simulations with plumed seeds revealed that a complex model incorporating the drift coefficient and stochastic acceleration produced more variation and longer dispersal distances than a simple stochastic model that utilized wind direction and speed as the main variables [49]. The vertical wind velocity ($w^{\prime }$) in the simple stochastic model could be expressed as:

$$\overline{W}=0,$$

(7)

$$w^{\prime }=W-\overline{W}={\delta }_w\times \epsilon \left(t\right),$$

(8)

$${\delta }_w={\left(\overline{{W^{\prime }}^2}\right)}^{1/2},$$

(9)

where $\overline{W}$ is the mean vertical wind velocity, $W$ is the instantaneous vertical wind velocity, ${\delta }_w$ is the standard deviation of the fluctuations in vertical wind velocity ($w^{\prime }$), and $\epsilon \left(t\right)$ is a Gaussian white-noise random variable with zero mean and unit variance. The Lagrangian stochastic dispersion model was calculated as follows:

$$d{u}_i=a\left(x_i,u_i,t\right)dt+b\left(x_i,u_i,t\right)d\Omega ,$$

(10)

where $u_i$ is the instantaneous wind velocity in direction $x_i$, $a$ is the drift coefficient, $b$ is the stochastic acceleration coefficient, and $d\Omega$ is a Gaussian random variable with zero mean and variance $dt$, with $dt$ dependent on the time-scale of turbulenct air movements. Therefore, the complexity of dispersal predictions increases when the irregularity of wind variables is included in the model.

Katul et al. (2005) [57] applied the Wald analytical LDD (WALD) model, which was based on Lagrangian stochastic approaches, to reduce the complexity of wind flow and organize dynamic relationships [58,59]. The probability $p\left(d\right)$ of a seed-moving distance $d$ was calculated as follows:

$$p\left(d\right)={\left({\displaystyle \frac{\lambda }{2\pi d^3}}\right)}^{{\displaystyle \raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$2$}\right.}}\exp \left[-{\displaystyle \frac{\lambda {\left(d-\mu \right)}^2}{2{\mu }^{2}d}}\right],$$

(11)

$$\lambda =k{h}_c\left(2{\displaystyle \frac{{\sigma }_{\omega }}{\overline{U}}}\right),$$

(12)

$$\mu =h_r{\displaystyle \frac{\overline{U}}{V_t}},$$

(13)

where $k$ is the scaling coefficient, $h_c$ is the canopy height (m), $h_r$ is the height at seed release, $\overline{U}$ is the spatially and temporally averaged wind velocity used to achieve a low turbulence, and $V_t$ is the terminal velocity. When comparing WALD with the mechanistic models for predicting the LDD of forest trees, heathland shrubs, and grassland forbs, the values predicted by the WALD model showed the highest agreement with the actual measurements. The WALD model has contributed to the understanding of LDD mechanisms by integrating the effects of key factors into scalar properties and is still evolving to incorporate multiple species and conditions [60,61]. Many earlier seed dispersal models have been structured to investigate this phenomenon using vector variables under the assumption of a normal distribution. Therefore, a model that can be linked to existing research is required to achieve optimal utility.

4. Research Trends and Limitations in Seed Dispersal

The collected data were categorized by topic, and each category was classified and analyzed according to the research phases of observation, generalization, and application to understand the trends in seed dispersal and related research from a broader perspective [19,20]. The observation phase of seed dispersal consisted of research aimed at collecting primary data and analyzing movement patterns. The generalization phase was classified as research that quantitatively analyses the impact of environmental conditions and establishes models. Finally, the application phase was considered the stage in which seed dispersal was utilized in adjacent fields. Our review classified 240 papers according to research stage, with 111 (46.25%) representing the observation phase, 54 (22.50%) representing the generalization phase, and 75 (31.25%) representing the application phase. Further subdivisions based on research content comprised four categories in the observational phase, five in the generalization phase, and two in the application phase (

Table 2. Classification of research on seed dispersal by wind according to observational, generalization, and application phases.

Research Phase	Topical Subcategory	Description	Number of References
Observation	Field investigations	Measuring and analyzing seed inflow through field surveys	19
	Measurement of physical features	Identifying seed morphology and terminal velocities	50
	Seed type classifications	Classifying seed type and falling behavior	12
	Wind tunnel experiments	Simulating seed dispersal distance according to environmental variables	30
Generalization	Wind effect studies	Simulating seed dispersal patterns by wind speed	28
	Effect of environmental factors	Identifying the effects of environmental conditions on seed dispersal	12
	Modeling and predictions	Developing models for seed dispersal and trajectory	11
	Spatial analysis	Identifying spatial patterns of seed influx	2
	Meta-analysis	Meta-analysis of seed dispersal distances	1
Application	Natural regeneration	Field experiment on the effects of using seed dispersal for natural regeneration	35
	Secondary dispersal	Identifying seed movements occurring after they have settled on the ground	40
Total			240

). In this section, we analyze the research trends according to the research phase and discuss the major implications of each topical subcategory.

4.1. Observation Phase

The main focus of the observation phase research was to measure the physical characteristics related to seed dispersal and analyze movement patterns; however, the available primary data were limited. The variables used directly in seed dispersal were seed mass, terminal velocity, and diaspore surface area [62]. Seed mass has generally been considered an important factor in seed science, so a large amount of data is available. On the other hand, in the previous literature, only 31 studies (12.91%) described terminal velocity. These data were measured in China (112 species), Germany (47 species), South Korea (34 species), and the Netherlands (31 species). Because the accumulated data were concentrated in the mid-latitudes of the Northern Hemisphere and the target species were lacking, there are limitations to understanding the impact of seed dispersal owing to the complex species composition. To compensate for the lack of data, information from plant databases were used; however, most databases were structured in a format that shares data measured by the LEDA Traitbase. This means that there is a high degree of dependence on the information provided by a single platform, and verification of the measurement values is impossible. In the case of diaspore surface area, there were few measurements of diaspores; therefore, researchers tended to measure them themselves or replace them with seed area data [11,12,13]. Therefore, even though there were 111 research results in the observation phase, the level of usable data was low, and primary data on the morphological characteristics of the seeds were lacking.

In seed dispersal by wind, the mechanism responsible for the movement of seeds according to wind speed was the most important variable, eliciting discussions on related measurement methodologies [63,64,65]. Establishing the length of the vertical passage is crucial for measuring terminal velocity and vertical range. Each of the reviewed studies contained unique standards that reflected the seed morphology of the target species; consequently, any comparison of their results raised uncertainties. The distances required to reach the terminal velocity may vary, specifically, for seeds that are large or have developed diaspores, owing to their relatively high drag coefficients [66]. Therefore, standardized criteria must be established to minimize errors, even when applied across diverse species. Considering that terminal velocity is the most important variable for seed dispersal, a scientifically reliable methodology should be supported. Lee et al. (2022) [11] filmed falling seeds, secured that the minimum distance reached the terminal velocity, and measured the velocity based on frame units. In this study, the minimum distance was not set differentially according to the size of the drag, but it could serve as an example of reducing errors in measuring devices and scientifically observing the speed.

Studies aimed at identifying the external factors affecting seed dispersal have mainly simulated changes in wind speed influenced by topography and vegetation. During the observation phase, studies utilizing wind tunnels to investigate the impact of wind speed on seed dispersal were the most prevalent. This was the most active field of inquiry, because the application and intensity of variables could be adjusted depending on the research objectives. In wind tunnel experiments investigating the effects of topography, similar terrains could be reproduced to obtain precise information on changes in speed and air pressure across terrain gradients and canopies [67]. For example, such research has indicated that turbulent winds predominate in the presence of geographical features and canopies. However, the limitation of the wind tunnel experiment was that it could not address the discrepancies between the model and real-world conditions, as the data were based on a scaled-down environment. Therefore, to enable generalization of the findings, the required adjustments using data from the scaled-down model should be quantitatively derived via field measurements.

Outdoor experiments have an advantage over wind tunnel experiments in that they can confirm changes in wind speed at higher altitudes. Nevertheless, they also have limitations, such as high equipment costs and the inability to control environmental variables at a fixed survey site. One study on topographic effects using a flux tower found that hilltops generated turbulence owing to air currents rising from valleys or lowlands, increasing the average distance of seed dispersal by approximately 35% [68,69]. Such a significant increase in dispersal distance underscores the importance of incorporating topographical characteristics as a major contributing environmental factor at field sites.

To determine the influence of vegetation distribution on seed dispersal, young or miniature trees have been used to assess the effects of changes in wind speed within wind tunnels [70]. Wind speed was influenced by the drag of each plant, structural porosity, and leaf and branch configurations, which could increase turbulence intensity by up to 50% [71]. In other words, although the speed of the wind passing through a vegetation zone decreases, its turbulence increases, increasing the complexity of seed dispersal. According to a study measuring the change in wind speed by the tree density and vegetation arrangement in a single forest, the wind speed in the area behind vegetation decreased by up to 20% [72]. Therefore, the seed dispersal distance is affected by the vegetation density and structure. However, because of the lack of related experimental results for environments with complex species and canopy compositions, it is challenging to apply these outcomes in real-world settings.

4.2. Generalization Phase

Research in the generalization phase included quantitative analyses of the impact of environmental conditions on seed dispersal and the establishment of seed dispersal models. In the detailed items of the generalization phase, the effect of the wind subcategory accounted for 28 cases (11.66%), the effect of environmental factors for 12 cases (5.00%), and modeling and prediction for 11 cases, accounting for 4.58%. Studies involving modeling and predictions accounted for 47.43% (37 cases) of this research, while those assessing the effects of wind comprised 35.89% (28 cases). This focus on wind effects can be attributed to wind being one of the main variables influencing seed dispersal through complex mechanisms that require consideration in multiple scenarios [22].

The impact of wind was primarily examined in terms of dispersal distance and spatial spread [73,74]. Because they have been studied independently based on the morphological features of the seeds, the amount of information tends to be concentrated on specific taxa. Accordingly, although research on pappu, one-winged seeds, and rotating falling seeds were active, data on other seed types were relatively lacking. Interspecific differences in seed dispersal were influenced by seed mass, terminal velocity, and seed surface area (including diaspores), because seed flight distance increased as horizontal wind speed increases [75,76]. This principle implied that at the same wind speed and release height, seeds with a lower terminal velocity and higher surface area were capable of a greater dispersal distance. Based on this assumption, previous studies have considered seed morphology a key factor in determining dispersal. Embracing a new perspective, Qin et al. (2022) [77] investigated the blocking of seed dispersal by vegetation canopies, and demonstrated that seeds were more likely to exhibit longer dispersal distances because they were more strongly deflected by canopies at low wind speeds. This suggested that the seed dispersal distance may be extended owing to the generation of updrafts, an increase in vertical wind speed, and a decrease in horizontal wind speed resulting from wind passing through the canopy. Therefore, more accurate seed dispersal predictions can be obtained by including variables that represent changes in wind characteristics caused by features of the surrounding vegetation, such as crown area, vegetation density, and stratification.

In the subcategory of studies investigating the influence of environmental factors on seed dispersal distance, the influences of soil properties and humidity were investigated. Soil properties were viewed as the determining factors of the friction coefficient involved in seed movement during secondary dispersion. However, related research was limited because of the difficulty of tracking individual seeds for measurement [21,78,79]. Whether the relationship between secondary dispersal distances and soil differences could be generalized through additional research requires further evaluation. In terms of humidity, some studies have reported that the flight trajectory and terminal velocity of seeds vary depending on humidity levels, but this phenomenon has only been verified for a limited number of species [79,80]. Therefore, it is necessary to determine whether the results can be extrapolated to a broader range of taxa.

Furthermore, seed dispersal modeling tended to be divided into approaches that consider either the vertical or horizontal movement of seeds. Vertical movement was used to investigate the effect of seed morphology on dispersal via gravity-induced falls and updrafts. Such studies have assessed the wing loading, drag force, and Reynolds number of individual seeds, which are representative indicators of seed dispersal. In the models considering the horizontal movement of seeds, the dispersal distance was estimated by applying the influence of external factors to the variables used in the vertical movement equations. Wind speed and release height were considered important external factors, and a model has been developed to account for various conditions by adjusting wind speed parameters [45]. However, because horizontal mobility models tended to be based on site-specific data, the findings from vertical movement models are not directly applicable. Therefore, to improve the next-generation models, it is necessary to extract essential algorithms by reviewing existing models and integrating their data.

4.3. Application Phase

The application phase consisted of attempts to apply seed dispersal to natural regeneration and identify secondary dispersal caused by various vectors. The main focus of natural regeneration-related research was the monitoring of natural seed influx and sapling development in gap areas within forests [81,82,83]. Generally, the level of seed influx was inversely proportional to the distance from the stock plant and displays interspecific differences in maximum dispersal distance and density patterns [84]. The maximum sapling density is usually located at a small distance from the stock plant and in a location with optimal sunlight, free from the influence of the parent crown [85,86]. Seedling establishment patterns could take many forms, depending on the germination rate and predation. However, because seed influx is the primary factor influencing natural regeneration, it has been the most important parameter in related studies.

While the number of papers on natural regeneration research has been increasing because of high social interest since the 1990s, the seed dispersal mechanism identified in previous studies has not been utilized. Natural regeneration studies included 35 cases (14.58%), and various attempts were made to quantify the effects of seed dispersal. The main variables commonly assumed were the complexity of the species composition, structure of the vegetation patch, and age of the vegetation community. The dependent variable was the number of seeds introduced into the site [82,87,88,89]. They explained this phenomenon by deriving empirical formulae using data collected from a site [90,91]. Therefore, it is dependent on the vegetation structure and environmental factors within the site, which limits its applicability to other sites.

Although social needs have driven research on natural regeneration via seed dispersal, analyses have focused on the number of seeds collected at sites, without considering the seed movement mechanisms involved [89,90]. Consequently, natural regeneration has been explained through the derivation of empirical equations without incorporating the theoretical background of the dispersal process. Therefore, such equations were dependent on site-specific vegetation structures and environmental factors, and their applicability to other sites was low.

Secondary dispersal investigated the additional movement that occurs after seeds settle on the ground, and the mainstream research is on animal vectors. Because the feeding methods for seed dispersing agents are diverse, studies tended to be conducted by taxonomic groups. The most common case was the dung beetle, a member of the endozoochory family that has a complex dispersal mechanism to transport seeds back in the droppings carried by the animal [91,92,93]. This migration was considered ecologically important because it theoretically allows seeds to spread over long distances and form new plant communities. There has also been much interest in myrmecochory, which is the movement of seeds by ants. Because dependence on seed consumption varied depending on the species, discussion is ongoing about its effect on seed dispersal [94]. Secondary dispersal of seeds may not be easily generalized because the vector relationship has not yet been clearly identified, and the extent of spatial influence on dispersal is controlled by feeding characteristics, density, and environmental conditions. Runoff can be a significant agent of secondary dispersal, particularly in hilly topography; however, relevant studies were not identified in the reviewed literature. This gap highlights the need for further attention to secondary seed dispersal via runoff.

5. Discussion

5.1. Implications from Trends in Seed Dispersal Research

Seed dispersal models have been developed in various forms; however, the taxa to which they can be applied are limited because of the lack of primary data. For seed falling, the wing loading, drag force, leading-edge vertex, and Reynolds number methods were used, and seed mass, seed area including diaspores, and terminal velocity were the core variables. Physical measurements of seeds are available from sources, such as the LEDA Traitbase [95], D3 database [15], TRY database [96], UCONN Plant Database [97], and Seed Information Database [98]. Seed mass is a value commonly provided by plant databases, but terminal velocity is dependent on data from the LEDA Traitbase. The terminal velocity is available for only a small number of species for which cross-verification is possible. This is a serious flaw, as it negatively affects the reliability of the statistical analysis results, and it is one of the reasons for the lack of case studies for various species. Therefore, the scientific reliability of this value has not been secured. To increase the accuracy of the vertical movement of seeds, a form that reflects the detailed physical features of the seed appendage should be developed. Additionally, because there is no information on the area of the appendages, such as wings and hair, reliability is not guaranteed when the developed model is applied to various taxa. Information on key variables other than seed mass can be found in only a few databases and individual papers. The horizontal movement of seeds consisted of research on dispersal distance, trajectory, environmental effects, and LDD. The dispersal distance was estimated by adding external factors to the variables used for vertical movement, as external factors, height, and wind speed were considered important, and improvements were made to reflect irregularities in the terrain and wind flow. However, because of the lack of field-based observational data, they could only be applied within the scope of each study [16,99,100,101]. These limitations make it difficult to verify and improve the model and are the main reasons why the developed model is rarely used under various conditions.

The main reasons why seed dispersal models have not been utilized in the field are the narrow scope of application of the results of previous studies and the lack of diversification in approaches to natural regeneration. According to our classification of research phases, the proportion of seed dispersal studies that have focused on model development and field applications was high; however, basic research remains lacking. Based on the data collected in the present study, only 405 species met the key variables required for the seed dispersal model. Thus, currently available data have many limitations when applied to ecosystems with complex distributions of multiple species. Therefore, the effect of linking seed dispersal to adjacent fields stays unclear, and it remains challenging to predict the quantity and species composition of seeds introduced at a specific location. The lack of primary data and limitations of models that only explain specific seed types have resulted in difficulties in introducing previously identified dispersal mechanisms into the field. This could cause serious flaws in recent attempts to introduce natural regeneration into sustainable plant ecosystems. Natural regeneration is a method of maintaining the sustainable health of ecosystems using natural seed dispersal and has the advantage of reducing the human and material resources required for afforestation projects [102,103]. However, in terms of landscape ecology, the effects of seed dispersal have not been quantitatively evaluated because of a lack of research related to habitat fragmentation, overharvesting, biological invasions, and climate change, all of which cause changes in biodiversity [104]. Case studies on the introduction of natural regeneration assumed a simple structure of vegetation patches and did not consider interspecific competition. Therefore, it remains unclear how seed dispersal occurs in an actual environment. This has led to an underestimation of the potential use of seed dispersal to solve ecological problems despite its various advantages for natural regeneration.

5.2. Elements for Improvement in the Field of Seed Dispersal

To increase the field application usability of the seed dispersal mechanism, three requirements should be met. First, the data collection system for securing primary data should be improved. Among the variables used for seed dispersal, available data other than seed mass were limited to specific taxa. In particular, owing to the lack of physical measurements of the seed diaspore, the uncertainty in predicting the terminal velocity was large, and the complexity of the horizontal movement could not be explained [26,105]. The drag force generated from vertical and horizontal movements is affected by the shape and area of the seed appendages and has an important relationship with increasing flight duration compared to their mass [40,43]. However, the measurement value for appendages has been measured directly according to the researcher’s needs, which caused errors due to differences in plant sample management, moisture content, detection equipment, and methods of measurement. Therefore, a method for applying the same standards to directly measure the terminal velocity in the LEDA Traitbase must be prepared for the seed appendage. These data should be implemented in a platform system such as a plant database, which can cover a wide range of taxa beyond the level of measurement by individual researchers. The ideal approach is to construct a system in which the physical properties of the seeds can be cross-verified through the measurement values provided by various platforms. This could be one of the scientific bases for the differences between species and occurrence of ecotypes. To add measurement items to an existing system, there must be a specific list of variables and active recommendations from academia that represent social demands. The list and content of the major variables used in the seed dispersal prediction derived from the results of this study may contribute to this (Table 1).

Second, sufficient field data should be collected. Field-based data are lacking in the sphere of seed dispersal [15,16]. Although many models have been developed to explain seed dispersal, they only have explanatory power for limited plant types and conditions [106,107]. Therefore, the fundamental problem is that the scope of application of seed dispersal models was restricted to field use. Based on the data collected in this study, 35 pilot studies (14.58%) attempted to apply the seed dispersal mechanism to natural regeneration. Although these were cases in which the seed dispersal model was suitable, they tended to develop their own empirical models by employing seed inflow data obtained from field surveys rather than using a developed model [99,100,101]. The continuation of this research trend can be generalized through an inductive method for different vegetation distributions and environmental conditions; however, it takes a lot of time and effort to obtain the required samples. Therefore, it is reasonable to design field investigations based on a theoretical background, and an empirical model established in a specific environment should be carefully compared and verified with existing models. An ideal method for field survey design is to identify the supply and arrival points of seeds through genetic analyses of introduced seeds and seedlings [14,108,109]. Realistically, this method cannot be applied to all field investigations; however, verification through equivalent scientific methods is required to identify general trends regarding typical types of plant communities.

Finally, an integrated model that can be applied to various conditions and species should be developed. To meet the recent social demand for seed dispersal in natural regeneration, a methodology that can be applied to various vegetation structures and environments is required. Natural regeneration by seed dispersal is known to be effective in maintaining biodiversity, but the available models must overcome the limitations of being restricted by region or taxon [102,110]. Forests restored through natural regeneration have a similar form to primeval forests and are highly efficient because their success rate in restoration is higher than that of artificial reforestation [111,112]. Additionally, it plays a role in strengthening the genetic diversity of plants that have settled and survived in a specific area and in passing on genetic information to future generations [113,114,115]. Previously developed models were specialized for specific seed types and have evolved to require detailed variables, such as the drag coefficient, wing thickness, wing curvature, ratio of wingspan, air viscosity, turbulent wind in the vertical updraft, and upward vertical air movement, to increase the explanatory power of the model. This has high value because it guarantees scientific accuracy; however, it is realistically difficult to obtain data because expensive equipment and high-level technology are required. Therefore, to increase usability in fields where various types of seeds are distributed in a complex manner, model integration and variable simplification are required. According to a review of previous studies, it is appropriate for the structure of the integrated model to be consistent with physical characteristics such as seed mass, terminal velocity, and seed area, including diaspores, and external factors such as wind speed and release of height. These are variables that can be obtained relatively easily from plant databases and have the advantage of enabling analysis of various species in the current situation where data are concentrated on some species [116]. Through simplified variables, it is possible to connect with previous research and dramatically expand the scope of application and space range. However, to understand the results intuitively and check for differences, a form that allows spatial analysis based on space and time is ideal [117]. Attempts to utilize seed dispersal models in the field have been effective when combined with spatial models [6,25] and will produce results that can be directly used in establishing policies or management plans.

6. Conclusions

Seed dispersal plays a vital role in plant succession and natural regeneration. However, a widely applicable model remains lacking, as most studies focus on specific sites or species. To support sustainable forest management, seed dispersal effects must be quantified; yet, prediction methods are rarely applied in practice. This study examines why seed dispersal research has not been widely implemented in the field and explores ways to enhance its usability. Specifically, we aim to (1) identify key variables for predicting seed movement, (2) conduct a quantitative assessment within detailed categories, and (3) propose improvements for practical application.

We reviewed 240 papers covering seed physical characteristics, wind tunnel experiments, environmental influences, dispersal models, and natural regeneration, extracting core variables affecting seed dispersal. Mechanisms were classified into vertical and horizontal movement, distinguishing the influences of falling dynamics and external factors. Research was categorized into observational, generalization, and application phases to assess academic trends. Based on these findings, we identified three key areas for improvement: (1) establishing a data collection system to secure primary data, (2) gathering sufficient field data, and (3) developing a unified model applicable across various conditions and species. By addressing gaps in comprehensive reviews and identifying improvement strategies, our study offers valuable insights for field managers, scientists, and policymakers seeking to apply seed dispersal models.