Revelation of Coupled Ecosystem Quality and Landscape Patterns for Agroforestry Ecosystem Services Sustainability Improvement in the Karst Desertification Control

Abstract

Since the 1950s, the rapid depletion of natural capital due to human activities has led to a growing global demand for ecological and socioeconomic sustainability, driving the development of agroforestry. Although agroforestry ecosystems can maintain sustainable land resources and maximize land productivity, their quality continues to fluctuate. Moreover, there is no clear review of studies addressing the impact of the evolution of agroforestry landscape patterns on their ecosystems, and to fill this gap, we conducted an inclusive review. A total of 163 publications related to agroforestry ecosystem quality and landscape patterns (ELA) that met a set of inclusion criteria were obtained through the Scopus database using the literature review method of searching, appraisal, and synthesis report. The objectives were to summarize the research progress on ELA, reveal the dynamic coupling mechanism between landscape pattern evolution and ecosystem quality, explore the role of landscape pattern optimization in ecological processes and services in agroforestry, and suggest future research and policy directions. Although the understanding of landscape patterns and ecological processes has been deepened, there are limitations in the study of scales, habitats, and other aspects. It is emphasized that understanding the interaction between agroforestry and other landscape elements in spatiotemporal organization is a prerequisite for promoting sustainable benefits, and that the challenges of spatiotemporal dynamics are integrated to scientifically optimize agroforestry landscape patterns. Finally, it is necessary to gain revelations based on the coupling relationship of ELA, through scientific management of agroforestry landscapes, in order to sustainably consolidate the effectiveness of karst rocky desertification (KRD) control and to enhance human welfare.

1. Introduction

With increased land degradation due to agricultural intensification and expansion [1], agroforestry has been widely adopted as an environmentally friendly and ecologically sustainable land management practice in tropical and subtropical regions worldwide [2]. An important function of agroforestry is to diversify the agricultural landscape by combining perennial woody plants on top of traditional agricultural farming or livestock breeding [3], thus, increasing the overall biodiversity and balancing productivity [4,5]. The ecosystem services research boom initiated by Daily, Costanza, and Millennium Ecosystem Assessment (MA) has led managers to focus on maintaining the quality of agroforestry ecosystems as an important support tool for enhancing service provisioning capacity and has graduallybeen receiving more attention from scholars [6,7]. Studies have indicated that high-quality agroforestry ecosystems provide food, timber, and other products and also suppress soil erosion and control pests and diseases, pollination and biodiversity conservation, and many other ecological services [8,9]. This is due to landscape patterns, as carriers of agroforestry ecological processes, and their dynamics can directly influence the dispersion and provision of ecosystem services [10,11,12] and determine various ecological processes [13]. Within the current published publications, review studies have reported on agroforestry ecosystems, such as environmental benefits [14], ecosystem services [15], structural and functional stability [16], and the influence of generative elements of ecosystem services on supply capacity [17]. These studies have all focused on agroforestry and its scientific understanding and management measures in terms of improving ecosystem service capacity from different perspectives (

Table 1. Comparison of existing review papers with this paper.

Year	Journal	Title	Keywords	Focus Point
2017	Water Resour Manage	Environmental Benefits and Control of Pollution to Surface Water and Groundwater by Agroforestry Systems: A Review [14]	Agroforestry systems, alley cropping, fertilizers, pesticides, control of leachingto groundwater, control of surface runoff pollution	Environmental benefits
2018	Animal Production Science	Agroforestry for ruminants: A review of trees and shrubs as fodder in silvopastoral temperate and tropical production systems [15]	Agroforestry, biodiversity, ecosystem services, landscape ecology, pest regulation, pollination	Summary of ecosystem services
2022	Forests	Structure and Stability of Agroforestry Ecosystems: Insights into the Improvement of Service Supply Capacity of Agroforestry Ecosystems under the Karst Rocky Desertification Control [16]	Agroforestry, ecosystem, structure, stability, progress, insights, rocky desertification control	Structural and functional stability
2022	Science of the Total Environment	A review of agroforestry ecosystem services and its enlightenment on the ecosystem improvement of rocky desertification control [17]	Agroforestry, generating elements, service management, supply capacity, gaps	The influence of service generating factors on supply capacity

). However, few studies have reviewed how the landscape level influences and enhances the capacity of agroforestry ecosystem service provisioning based on a landscape perspective.

For an agroforestry ecosystem, the landscape pattern plays an important role in ensuring the quality of its ecosystem and enhancing the sustainability of its service capacity [13,18]. Agroforestry ecosystems are artificial complex ecosystems composed of multiple species on the same land and with the same environment [6,19,20]. The landscape patterns, mainly forest pastoral, forest food, forest fruit, home garden, and hedgerow intercropping, present a spatially heterogeneous mosaic of landscape complexity, contrast, and dependence of land management diversity [21], which also leads to differences in the composition and types of agroforestry ecosystems. Under human scientific management, the quality change process can be in a dynamic balance and has the ability to be adjusted [22]. The rational optimization of agroforestry landscape patterns is an effective means to improve the quality condition of agroforestry ecosystems [18]. For example, Schmidt et al. (2005) found that the change trends of landscape diversity and species diversity were normally distributed, and an improvement in landscape diversity could promote an increase in uncertain information content [23], which had a good promoting effect on biodiversity restoration and vegetation construction [24,25]. Additionally, the establishment of linear seminatural ecological corridors between landscape patches can promote spatial connectivity, thus, increasing the diversity and abundance of species such as butterflies, beetles and plants [26,27]. However, increased fragmentation due to human disturbance of the landscape can pose a significant threat to the ecological services of water and soil retention, pollination promotion, and pest suppression [28,29]. In essence, an agroforestry landscape pattern can favorably influence ecosystem quality, and the effect depends mainly on the landscape composition and spatial configuration characteristics of the agroforestry ecosystem [30]. When the pattern characteristics of landscape spatial heterogeneity are strongly disturbed by the natural environment and human activities, the agroforestry ecosystem will undergo unstable fluctuations and significant changes beyond the threshold of ecosystem stability [31]. Therefore, it has become a mainstream approach in current research to optimize the number, size, shape, distribution, and combination patterns of patches with different attributes within their landscapes by using a spatial explicit analysis [13,21,29] to improve the efficiency of interactions among species and the stability of synergistic symbiosis in ecosystems [32,33]. However, agroforestry ecosystem quality remains to be at risk of breaching equilibrium thresholds driven by multiple disturbances, making ecosystems more vulnerable [34,35], which in turn, directly affects ecosystem services [36]. For example, Europe is experiencing a series of extreme weather events, including rare high temperatures and historic droughts. Under such events, the question is how can the quality of agroforestry ecosystems be effectively enhanced so that they can sustainably maintain their service provisioning capacity in the long term? Deepening and understanding the relationship between ELA (Ecosystem quality and landscape patterns in agroforestry) from a landscape perspective has become key to addressing risks in agroforestry ecosystems [29].

Especially in ecologically fragile areas represented by KRD, irrational human agricultural development activities that crowd and destroy ecological space still exist, which reduce the benefits of ecological restoration and protection [37], and also intensify the trade-off between ecological environmental protection and socioeconomic development [38]. This is always the focus of much attention globally [39]. The problem of land degradation is a typical case of karst ecologically fragile areas, and is also a key concern for agriculture, ecology, and the environment in the region [40]. In recent years, in karst areas with severe environmental degradation, governments at all levels have been restoring KRD and improving people’s well-being through ongoing governance projects [41]. Among them, the KRD comprehensive control project carried out by the Chinese government at the end of the 20th century has resulted in an overall trend of “continuous net reduction” in environmental degradation, securing a large amount of natural capital and assets for human beings, strengthening the material foundation for human survival and development, and promoting regional ecological recovery and socioeconomic development [41,42,43]. Due to its significant benefits in improving land productivity, preventing soil erosion, and providing livelihood needs, agroforestry is regarded as the preferred multifunctional land-use method in ecologically fragile areas [44]. After years of verification, the agroforestry ecosystem maximizes the ecological benefits of soil and water conservation and land productivity [45,46], and also serves to protect biodiversity and other functions [47,48]. Thus, large areas of agroforestry ecological landscapes have also been formed for the purpose of ecological treatment. However, the unique hydrogeological structure and ecosystems of the KRD region are characterized by complexity and vulnerability as compared with other agroforestry cultivation areas (Figure 1), and the risk of degradation of already fragile ecosystems is exacerbated by irrational human activities. In addition, many studies have shown that the agroforestry ecosystem services after KRD treatment have lagging supply capacity and continuous quality fluctuations, which limit the harmonious development of a human–land relationship [17,18,49]. Under the new situation, how to improve the sustainability of agroforestry benefits and the level of regional ecosystem services in KRD control is imminent.

There has been global research on ecosystem quality [14,50,51], landscape pattern optimization [52,53], benefit coupling mechanisms [13,54], and strategic practices [7,55], which has laid foundational work for exploring the sustainability of agroforestry ecosystem services, but there has not yet been a systematic review (Table 1). Therefore, in this paper, we used a bibliometric statistical approach to systematically review the research progress and landmark results of global ELA [56]. The aims of this study were to deepen the understanding of the synergistic study of landscape patterns and agroforestry ecosystems in support of human well-being, to identify the major limitations and gaps that impede the sustainability of agroforestry ecosystems, and to explore what key scientific questions exist for the future that need to be urgently addressed. Additionally, this research revealed an effective way to enhance the benefits of agroforestry for KRD control, thus, promoting the synergistic development of regional ecology and the social economy and consolidating the ecological security barrier in southern China.

2. Methods

This study is based on a structured literature review of ELA-related research. The goal is to collect a large number of peer-reviewed scientific publications from around the world, from which to identify and catalogue the knowledge field and to provide inspiration for current agroforestry development in the KRD region. To facilitate a scientific quantitative evaluation, we used a systematic literature review framework to trace existing work based on the scientific information provided by the Scopus database [17,57]. The specific framework is as follows:

Protocol Define the scope and purpose of this study;

Search Development of search strategies, identification of search strings, and selection of search databases;

Appraisal To evaluate the quality of retrieved documents and to define the quality standards for inclusion and exclusion of documents;

Synthesis Report Extract data, quantitatively classify and describe the results, summarize the main progress and landmark achievements, determine the key scientific issues to be solved, and give inspiration in combination with practical problems.

The method is systematically logical and has four major advantages: transparency, independence, robustness, and comprehensiveness [58]. Its scientific validity has been confirmed by its use in various research fields [59]. Therefore, we conducted a retrospective study with this method, focusing on four steps of retrieval, i.e., identification, repeat screening, eligibility, and inclusion [60].

2.1. Protocol

To ensure the transparency, transferability, and replicability characteristics of the literature system review effort, a protocol was developed prior to conducting the search. The most central element of the protocol definition was the identification of the review’s purpose, which helped to formulate answerable research questions and to define research boundaries. The main research questions were as follows:

• What are the trends in the number of publications over the years, the types of research topics that can be classified, and the number of each topic?
• What are Global ELA research focus regions and their institutional distribution?
• What are the current major research progress and landmark results?
• What are the key scientific questions that need to be addressed in future research?
• What are the implications of this study in terms of enhancing the sustainability of agroforestry benefits in the KRD region?

All of the above are questions that this study answers by adopting a systematic literature review framework.

2.2. Literature Search Sources

To comprehensively discuss the current status of global ELA research, the literature in this paper was obtained mainly from the Scopus database (

Table 2. Literature search strings.

Database	Retrieval String	Number	Search Date
Scopus	First search string,“ecosystem quality”; second search string, “agroforestry”	375	30 June 2022
	First search string,“landscape pattern”; second search string, “agroforestry”	253	30 June 2022
	First search string:“ecological quality”; second search string, “agroforestry”	217	30 June 2022
Total	ELA	845	30 June 2022

Note that the data here include comments and original articles in all languages.

). Details of the retrieval process are as follows: First, “ecosystem quality”, “ecological quality”, and “landscape pattern” were used as “subject items” in the Scopus database for the initial search. Second, in the above search results, “agroforestry” was entered again for a second search, and a total of 845 documents were retrieved. The retrieval deadline was 30 June 2022, and the time frame of the retrieval was the maximum time frame of the Scopus database.

2.3. Literature Selection Criteria

Using a bibliometric approach [56], the 845 papers obtained by the search engine were reviewed for research content by strictly following the procedure developed in Figure 2. In addition, inclusion and exclusion criteria were applied to the initial search results, and papers that met the inclusion criteria were selected for further review and content assessment [61]. The specific method was based on the following criteria for screening:

• First, all the literature was screened for duplicate items and 689 relevant papers were obtained.
• Second, the literature was examined according to study title, abstract, and keywords, and studies with low relevance to the research topic were eliminated. For example, although the title “Landscape Composition Is More Important than Local Vegetation Structure for Understory Birds in Cocoa Agroforestry Systems” mentions key terms such as agroforestry ecosystems and landscape composition, the abstract states that the focus of the study is on the role of birds, and therefore, was excluded.
• Third, the literature retained from the secondary screening was read in its entirety, and the screening was based on covering at least one ELA-related study for eligibility determination. For example, a full reading of “Conserving Biodiversity Through Certification of Tropical Agroforestry Crops at Local and Landscape Scales” revealed that the study focused on the certification experience of tropical agroforestry crops (coffee and cocoa) and proposed certification at the landscape scale. Therefore, it was eliminated.
• Ultimately, 163 articles that were highly relevant to the research topic after three rounds of screening were included in the process to form the final sample for this study.

2.4. Synthesis Report

In this phase, 163 scientific publications were eventually included and were read intensively, and then relevant variables of interest were extracted and classified in order to draw knowledge and conclusions. With a large increase in population, the supply of agroforestry ecosystem services is increasingly constrained [17]. However, current research on agroforestry ecosystems still has mainly used geographic, botanical, and ecological perspectives to explore the importance of services and classifications, etc. This has tended to overlook the importance of the spatial dimension of the landscape and the supply capacity enhancement [18,49]. Therefore, we believe that summarizing knowledge and conclusions around the above literature and identifying the impacts of landscape elements and external interventions are necessary to mitigate the current situation of limited supply for agroforestry ecosystem services.

According to the ELA-related research, we divided the studies into four main research interests, namely, ecosystem quality, landscape pattern, coupling benefit mechanism, and landscape optimization design. First, in Section 3.1, we conduct a quantitative analysis of the annual distribution and content of the retrieval results; second, in Section 3.2, we identify the global distribution of study areas and their research institutions; thirdly, in Section 4.1 and Section 4.2, the research progress and landmark achievements are summarized, as well as key scientific issues are discussed, and ultimately, insights into the benefit enhancement of agroforestry ecosystems in the KRD region are provided in Section 4.3.

3. Results and Analysis

To obtain the latest progress and signature results of research on ELA and to better reveal the mechanisms that promote sustainable agroforestry ecosystem services, 163 papers were evaluated based on publication year, research content, and regional distribution [62]. Through the four divisions of ecosystem quality, landscape pattern, benefit coupling mechanism, and landscape optimization, the results highlight the gaps and limitations of ELA-related research over the years, and analyze this phenomenon based on the differences between natural economic and social conditions.

3.1. Annual Distribution and Research Contents of the Literature

ELA-related research dates back to 1983 and is now nearly 40 years old. Figure 3a shows that the period 1983–2021 could be roughly divided into three phases. In stage one (1983–2000), the total number of studies did not exceed 11, and this was the budding stage. Stage two (2001–2013), which had a slow but fluctuating upward trend, was the slow growth stage. Stage three (2014–2022) had a rapid growth trend, with an average annual number of studies of 12 and significantly increased interest in research in this field.

To provide precise hotspot research content, we used visualization techniques for word frequency analysis. To generate a word cloud using the titles and keywords of the selected articles, the hotspots and scales of ELA research were summarized. Figure 4 shows the subjective influences from the prior literature retrieval, and terms such as agroforestry, landscape, and ecosystem were excluded. Soil, water, biodiversity, value, biomass, forest, species, etc., have become the most commonly used words in ELA research and are also the main research content. This was followed by high-frequency words such as land use, spatial, conservation, assessment, and management.

To better reveal the sustainable benefits of agroforestry in KRD control, we divided ELA into four main research directions around the main research content (Figure 3b). There were a few studies in the literature that revealed the research topic but the research category was not clear, therefore, these studies were classified as “other research”. First, according to the conditions and pressures of agroforestry ecosystem quality, studies on the quality of the atmosphere, soil, water, climate, and environment were summarized [14], and collectively referred to as “ecosystem quality”; second, landscape-scale studies on spatial connectivity, fragmentation, landscape composition, and ecological processes [53,63,64] were collectively referred to as “landscape patterns”; third, synergistic studies of agroforestry ecosystem quality and landscape patterns [54,65], analyzing the response relationship between the two, were categorized in the literature as “benefit coupling mechanisms”. Finally, research on strategic practices aimed at improving ecosystem quality, such as promoting biological interactions and ecosystem services [7,55], was collectively referred to as “landscape optimization design”.

Regarding the four research hotspots of agroforestry ecosystem quality, landscape pattern, benefit coupling mechanism, and landscape optimization design, we found, through the annual growth trend, that agroforestry ecosystem quality research consistently received increasing attention by scholars and publications (Figure 5). From 2010 to the present, this research has shown a high rate of development, which may be mainly related to the Global 2nd Agroforestry Conference in 2009, which called on scholars to pay attention to the potential role of ecosystems and led to the diversification and high rate of development of agroforestry ecosystem research [49]. Studies related to agroforestry landscape patterns began to receive attention in the early 1990s. Along with the deepening of scholars’ awareness of the importance of landscape patterns, combined studies on benefit coupling mechanisms have begun to emerge, but systematic theory has not yet been formed and is at an exploratory stage. Figure 4b shows that after years of development, research on the coupling mechanism of landscape pattern and benefit has been synergized, and the overall change trend was similar, accounting for 26% and 23% of the total number of published articles, respectively. However, reports on case studies of landscape optimization have consistently fluctuated and did not show a significant growth trend. This phenomenon highlights the gaps in research, suggesting that theory and practice are not developing in tandem and that practice often lags behind theory.

Although the importance of the current landscape pattern to an ecosystem was initially recognized by the academic community, from the landscape perspective, interventions can promote ecosystem functions such as nutrient cycling and energy circulation in agroforestry ecosystems [12,27]. However, There is concern that land use practices on agroforestry landscape optimization have failed to be mainstreamed [65,66]. Through a survey, we found that restraints that existed in traditional agriculture were the main reasons that hindered expansion, which were overly concerned with, and even dangerously dependent on, habits that stem from postwar technological advances [29]. Most farmers are willing to reduce the use of pesticides to restore the ecological environment but are reluctant to change the entire farm pattern. Major barriers are the fear associated with the risk of declining yield and quality, inadequate markets for ecological niche crops, insufficient knowledge of specific practices, and an inherent resistance to change. Another important reason is that, although the impacts of landscape attributes on agroforestry ecosystems have been tested in past studies, better data support should be obtained [11,12,67,68]. These studies have been dominated by seminatural habitats, which could provide food, nesting sites, and breeding sites for plant pollinators and could contribute to the presentation of experimental effects [69]; however, most agroforestry ecosystems were developed based on cropland, with relatively few seminatural habitats, and therefore, current studies have limitations and present obstacles to the practice of landscape optimization in agroforestry.

3.2. Research Region and Institution Distribution

The global perspective reveals that research on ELA is unevenly developed and has prominent regional characteristics. Figure 6 shows the global distribution of ELA-related research regions and research institutions. In terms of the regional distribution of studies, the main concentration of ELA studies has been in United States, China, Germany, Italy, and other north temperate countries, with the number of studies in the literature exceeding 12. However, ELA research in the tropics, where agroforestry (cocoa and coffee) is better developed, does not receive as much attention as research in temperate regions [70,71]. The reasons for this phenomenon are the influence of natural conditions and socioeconomic differences, and also national policy support and the attention of research institutions that play an important role in it [72]. From Figure 6, it can be observed that the highest spatial distribution density of research institutions is located in developed countries, which reflects the higher attention to agroforestry ecosystems and policy development [17,21].

The United States, for example, as the world’s largest agricultural exporter of agricultural products with a fixed international sales market and a highly mechanized agroforestry industry, provides a sufficient research base for ELA research to some extent [5,73]. Therefore, the vast majority of doctoral dissertations on agroforestry complex management originate from U.S. universities, where optimal land resource use and design are the main research components [74]. Additionally, policy support as an important driving factor is an essential link. The United States Department of Agriculture (USDA) released a strategic framework for agroforestry that subdivided its application model into two major categories and five subcategories, namely agroforestry farming systems (including alley cropping, forest farming, and silvopasture) and linear agroforestry practices (including riparian forest buffers and windbreaks). To promote the development of agroforestry, the USDA has made publicity (ensuring that landowners and local communities have access to the latest information and tools for agroforestry operations), research (conducting applied and basic research to promote the development of agroforestry science and technology, leading to increased productivity and effective responses to emergencies), and integration (promoting the integration of agroforestry information, research, tools, and technologies) as strategic goals and has set specific tasks for USDA agencies in the development of agroforestry [75].

China differs significantly from the United States in that the accelerated urban expansion process in China has led to a dramatic reduction in arable land area and a decline in natural habitat quality [76]. Global concerns such as food scarcity, energy crises, and environmental degradation are highlighted under the pressure of a very large population base, prompting people and local governments to deeply recognize the importance of sustainable development [77]. However, agroforestry complex management, as a traditional Chinese characteristic agriculture, is exactly a sustainable agricultural model that meets the current development needs. There has already been a theoretical summary of its agricultural culture in the course of long-term practice, emphasizing the correct handling of the relationship between organisms and the environment, and between organisms and organisms [16]. It contains the biological planar layout of the agroforestry complex management model, which can be divided into various forms such as banded interspecies, cluster mix, landscape layout, water–land interaction, and patch mosaic [74]. In addition, an in-depth exploration of the current distribution of ELA research in China has revealed that the density of ELA research is higher in the northwestern arid zone and the southwestern KRD zone. The reason for taking Guizhou and Shaanxi Provinces as typical areas is that both the Shaanxi Loess Plateau and the Guizhou karst region are ecologically fragile areas with severe soil erosion, where the conflict between humans and land has intensified, and the traditional cropping pattern can no longer meet human needs [41,78]. China’s State Forestry and Grassland Administration has instructed that under the demand of China’s rural agricultural restructuring, the vigorous development of ecological agriculture and sustainable development strategy and the development of agroforestry in KRD areas are of great significance to regional ecological environment construction and economic development. Combining agroforestry with rocky desertification control can be divided into an economic-led rocky desertification control model and an ecological restoration-led rocky desertification control model [47]. As a result, research focusing on ELA has received more attention in China.

However, among the more mature countries with agroforestry development, Australia’s wheat and sheep farming is a typical representative of agroforestry, but there are few studies related to ELA, which may be mainly related to its development model. The southwestern and southeastern regions of Australia (Murray-Darling Basin), as the main agroforestry regions, commonly use mixed wheat and sheep grazing operations [79]. As compared with other countries, Australia has superior development conditions (located in plains and basins, with suitable water and heat conditions, open arable land, fertile soil, and sparsely populated land) and a relatively homogeneous pattern of agroforestry, with sheep grazing and wheat cultivation forming a good ecological balance and a sufficient supply capacity [80]. Although Australia also faces some of the same sustainability issues, such as land degradation and energy use and inputs, the constraints on agroforestry development are much lower than those in other countries [81]. Therefore, ensuring the productivity and sustainability of future rural agricultural landscapes by increasing the number of trees planted in agroforestry has become the main object of research [82], and ELA research aimed at promoting the enhancement of agroforestry ecosystem services has not received much attention in Australia.

4. Retrospect and Prospects

Through a literature review, we summarize the research progress and landmark results of ELA-related research over the years and present the key scientific issues faced in the future research process. Meanwhile, taking the KRD area as a typical case study and synthesizing the summary of ELA research in this paper, the theory and method based on landscape ecology provide enlightenment for the sustainability enhancement of agroforestry ecosystem services in KRD control.

4.1. Research Progress and Landmark Results

Landscape patterns and ecological processes both interact with each other to present certain landscape functions that constitute the bulk of ecosystem services [10]. As artificially governed ecosystems, the evolution of landscape patterns in agroforestry ecosystems is related to their own ecosystem diversity and also profoundly affects the material and energy cycling processes between habitat patches [32]. For example, habitat fragmentation tends to lead to a reduction in connectivity between patches, which compresses the living space and hinders the migration and dispersal of individual organisms, thus, affecting the quality of the whole ecosystem [83]. Moreover, ecosystem function is inherently dependent on biodiversity, yet any organism has a natural tendency to spread and colonize new habitats, and these issues require landscape-scale considerations [29]. Therefore, we systematically reviewed the progress and landmark results of ELA research (

Table 3. Division of research stages.

Study Stages	Key Features	Breakthroughs in Stages
Stage of germination (1983–2000)	Fewer relevant studies in this phase, with a maximum of 3 ELA-related studies in a single year.	1983 “Agroforestry Ecosystems as Human Environment” was published. In the early 1990s, the U.S. Environmental Protection Agency (U.S.EPA) introduced the Environmental Monitoring and Assessment Program (EMAP). In 1995, Fu BJ studied agroecosystems in loess areas with landscape pattern theory [84]. The preliminary foundation of ELA research was laid.
Stage of slow growth (2001–2013)	The number of literature increased during this period, and although there were fluctuations, the overall upward trend was maintained.	3S technology has developed, models have been developed and applied, and quantitative studies combining multiple factors such as vegetation, topography, and climate have matured [85,86]. In 2000, Wu JG published “Landscape Ecology—Concepts and Theory”, and the MA in 2001 was officially launched, which promoted the research process of ELA [1,87].
Stage of high-speed development (2014–present)	The number of studies grew significantly during this period, especially from 2020 to June 2022, when the total number of studies published reaching 46.	Wu (2013) elucidated the dynamic relationship between human well-being and sustainability of ecosystem services from a landscape ecology perspective [83]. Xie et al. (2015) revised the ecosystem assessment system, made progress in quality assessment and dynamic monitoring, and significantly improved the methodology and theory [88].

) to improve a scientific approach to maintain the sustainable benefits of agroforestry.

4.1.1. Ecosystem Quality

Due to the increased demand for ecological and economic benefits, a variety of agroforestry cropping patterns, such as forest fishery, forest agriculture, and forest pastoralism, have emerged [89]. Especially in the tropics, cocoa trees and coffee plantations are typical applications of complex agroforestry systems, which present a farming structure with environmental benefits (conservation of natural forests, biodiversity, and ecosystem health) [14]. However, the inside-out benefits of multiple agroforestry cropping patterns differ significantly, resulting in quality imbalances [90]. Deheuvels et al. (2012) pointed out that agroforestry ecosystems could mitigate differences in benefits by rationally organizing the vertical hierarchy of landscape communities in the aboveground and belowground parts and water by the efficient use of natural resources, including light, temperature, water, and soil [91]. Therefore, early studies related to the quality of agroforestry ecosystems were conducted to guide farmers to rationalize production and to improve the quality level by determining the appropriate agroforestry community structure [92]. For example, Wang et al. (2011) found that agroforestry had ecological benefits such as controlling soil erosion, improving soil structure, increasing soil surface organic matter, and effective water reservoir capacity through fruit-grass cover planting [93]. Pavlidis et al. (2018) noted that tree roots in agroforestry ecosystems were able to reduce soil nitrogen (N) and phosphorus (P) residues from 20% to 100%, efficiently inhibiting pesticide leaching and runoff [14]. However, while reducing agricultural pollution, tree roots have been reported to absorb up to 30% of fertilizers, which inhibited crop growth and constrained ecosystem quality [94].

Sonwa et al. (2017) emphasized that to achieve balanced ecological and socioeconomic synergistic development, it was necessary to pay attention to its horizontal succession and to assess agroforestry ecosystems in an integrated way from horizontal and vertical structures [50]. Marais et al. (2019) found that changes in vegetation composition and planting spacing could directly affect agroforestry ecosystem services based on a sample-scale study [95]. Tscharntke et al. (2014) analyzed tropical agroforestry at the local and landscape scales, and showed that the establishment of agroforestry systems could improve biodiversity and could promote the full recycling of matter and energy [96]. Although it was reasonable to optimize its strip and cluster planting density and planting environment in terms of horizontal structure, which could improve ecosystem services such as pollination, pest control, and soil and water conservation, under the disturbance of heavy rain events in monsoon climate regions, the quality of ecosystems and the ability to provide services were significantly reduced [16,29]. Therefore, combining biological and engineering measures is the best option to maintain the quality of agroforestry ecosystems and their service provisioning capacity. However, carbon sequestration benefits are one of the key indicators of the quality of agroforestry ecosystems [97]. Although agroforestry ecosystems can store CO² in the soil for a long time and reduce atmospheric CO² to some extent, there are significant regional differences [98,99]. Carbon sequestration capacity is lower in arid or semiarid regions than in fertile and humid regions and lower in temperate regions than in the tropics [6]. Therefore, in addition to the difference between vertical and horizontal structures, the key factors affecting the quality of agroforestry ecosystem regional climatic conditions play a key role in their quality.

With the deepening of research, scholars have found that the quality of agroforestry ecosystems has been limited by geographical factors and also affected by differences in their own environmental quality [46,100]. In view of this, Chen et al. (2015) assessed the quality of agroforestry ecosystems based on plant communities, benthic fauna, water environment, and soil environment to capture the influencing factors governing the ecological value of agroforestry [101]. Until constraints are addressed, inappropriate agroforestry development can lead farmers to irreversible errors [29]. Therefore, parametric design and training in the field is necessary to find the best balance between trees and crops or pastures for each combination. However, with the development of remote sensing technology, methods for assessing the quality of agroforestry ecosystems based on remote sensing spatial models are commonly applied [51,102,103]. Therefore, an attempt can be made to explore effective ways to improve the quality of agroforestry ecosystems based on a larger spatial scale using 3S technology, with proper spatial planning and design, and all the above parameters should not be considered to be insurmountable [14].

4.1.2. Landscape Pattern

Landscape patterns are spatially heterogeneous distributions of land use/land cover (LULC) that are important drivers for maintaining regional ecological functions [87]. Agroforestry landscape patterns are a mosaic of mixed crop, monocrop, and non-crop multiple habitat patches [6,29,104]. In early studies of agroforestry landscape patterns, the main focus has been on the spatial and temporal changes in succession, the analysis of driving forces, and how to quantify landscape patterns [105,106]. Naveh (1994) and Wu (2000) systematized the relationship between ecological processes and ecosystems at the landscape level and emphasized that the core concepts of landscape ecology (landscape wholeness and heterogeneity, diversity of landscape scales, mosaic of landscape structures, etc.) were closely linked to ecosystems [87,107,108]. Scholars concerned with agroforestry ecosystems have also begun to experiment with studies based on a landscape perspective.

Some scholars have found that human intervention in agroforestry ecosystems can accelerate changes in landscape patterns, and the resulting changes in landscape and ecological processes can react to species pollination, distribution dynamics, pest control, and community structure, thus, affecting the quality of agroforestry ecosystems [109,110]. Additionally, using the theory and methods of landscape ecology, strategic points of the landscape that are critical for controlling horizontal ecological processes have been captured, and targeting these critical locations for ecological restoration has achieved twice the result with half the effort [111]. For example, Fu (1995) used the basic theory of landscape patterns to analyze agricultural landscapes and pointed out that ecologically sensitive topographic transition zones could enhance diversity and control soil erosion by establishing ecological corridors to enhance landscape continuity [84]. Interestingly, agroforestry, as a new mode of agricultural cultivation developed from traditional agricultural landscapes [3], uses the principles of landscape ecology to increase landscape heterogeneity by changing the three-dimensional mosaic pattern of agriculture, grass, and forest, linking material flows to different landscape attributes [108]. However, it is only a field-scale change and does not focus on a larger scale. With updated technology, the study of agroforestry ecosystems by applying the theory related to landscape ecology gradually tends to develop on a large scale and become diversified. For example, Simoniello et al. (2015), in their study of agroforestry landscapes in southern Italy, observed the evolution trends of their landscape patterns based on a watershed scale and pointed out that the increase in stability was due to a decrease in fragmentation caused by the reduction in small clearings that disrupt the homogeneity of agroforestry [52]. Wang et al. (2019) used GIS spatial simulation to model ecological processes resulting from different landscape patterns of agroforestry on a watershed unit as well as their resulting ecological effects [112]. Additionally, land use and soil erosion were used as discriminatory criteria, the concept of potential soil erosion intensity was introduced, and spatial simulation and land use priority were applied to optimize the landscape pattern of agroforestry and to promote the benefits of soil and water conservation [67,113]. Lin et al. (2019) discussed the ecosystem of the Greater Bay Area through 3S technology, and emphasized that the key to synergistic ecological and socioeconomic sustainable development was in optimizing the landscape pattern [114]. Therefore, using the theory of landscape ecology combined with 3S technology methods to research agroforestry ecosystems is conducive to multiscale assessment, macro-regulation of ecosystem spatial structure, and long-term prediction of future development trends, which coincides with the EMAP proposed by the U.S. EPA.

Forman (1995) proposed an optimal landscape pattern regarding sustainable development [115]. However, existing studies have shown that whether landscape patterns can positively affect agroforestry ecosystems cannot be measured in terms of a certain spatial structure or the state in which they are at a given time. As emphasized by Haines-Young (2000), assessing the quality or sustainability of an ecosystem should be determined through a continuous process of change [116]. In addition, Wu (2013) pointed out that there were multiple steady states in agroforestry ecosystems and that the management of such systems needed to focus on change rather than on equilibrium states, which have been certified long ago in terms of patch dynamics and resilience of socioecological systems [83]. Their arguments all refute the theory of optimal landscape patterns. Therefore, exploring the advantages and disadvantages of agroforestry landscape patterns cannot rely solely on the results presented in current static maps to reach conclusions. Instead, it is necessary to break through the time scale and monitor the continuous effects of landscape ecological processes on the ecosystem over a long period of time.

4.1.3. Benefit Coupling Mechanism

Agroforests, as human-dominated sustainable ecosystems, exhibit complex nonlinear interactions in structural and dynamic scale diversity and self-organizing capacity with changing landscape patterns [3,6,117], thereby, providing their ecosystems with regulatory capacity [19,20]. Changes from simple to complex and from unstable to stable are in dynamic equilibrium [22]. Under external disturbances, landscape and ecological processes change, and the equilibrium point of agroforestry ecosystem quality fluctuates continuously. Once this fluctuation threshold is exceeded, the quality state inevitably develops to another stage [118]. Such changes, if contrary to the evolutionary direction of the ecosystem, will lead to the degradation of the ecosystem service supply capacity [10]. In response to continuous fluctuations in the quality of agroforestry ecosystems, early ecological studies have been at least implicitly concerned with the spatial distribution of organisms [119]. However, the solutions have mainly been based on the “restoration ecology” approach of removing disturbances and accelerating changes in biological components [120,121]. It was not until the 1990s, when spatially explicit models gradually attracted attention, that they began to combine research with traditional landscape ecology [122,123]. However, with the development of agroforestry ecosystems, the risks have exceeded expectations and have even led to increased deterioration [124]. The occurrence of this phenomenon, in addition to the inadequacy of current science, technology and methods, is also a realization that approaches to ecosystem conservation based on the integration of ecosystems at the landscape level have not received much attention [108].

After 2000, scholars gradually began to focus on multispecies interactions at the local spatial level, as well as self-organization and the formation of large-scale spatial patterns in biological distributions. She et al. (2004) and Corry (2019) pointed out, based on a landscape perspective, that agroforestry landscape patterns, while directly influencing landscape processes, also determined various ecological processes whose changes inevitably led to changes in ecosystem quality and services [13,101,106]. For example, beetles are able to obtain subsistence food in patch A but lay eggs in patch B. Although natural enemies are ubiquitous within this landscape, ants in patch B can protect juvenile beetles from natural enemies. If such patch assemblages were disrupted, long-term development would lead to the extinction of beetles within the landscape and the disappearance of predators, which in turn, would affect the disruption of the entire ecosystem cycle and even the breach of the quality threshold [125]. This would affect within-ecosystem energy, nutrient and hydrological cycles, pollutant distribution, and species transport within the ecosystem [11,12]. However, reconnecting them through inter-patch migration may restore stability to the entire spatial ecosystem [126]. Thus, different spatial combinations of patches have different functions, and the size, shape, and connectivity of agroforestry patches affect the species richness, distribution, population viability, and disturbance resistance within the landscape, which in turn improves the dispersal and provision of ecosystem services [106].

After the Second Agroforestry Conference, the potential role of agroforestry ecosystem services has attracted much attention, and whether agroforestry could sustainably play ecosystem services has become a question worth considering in the diversification stage [49]. From the landscape-scale perspective, the evolution of landscape patterns as a key factor for metacommunity structure and persistence in agroforestry landscapes [68], and changes in some indices can result in effective responses in biodiversity and abundance (including butterflies, beetles, birds, and plants) [27]. Pin (2010), Liu et al. (2011), Ge et al. (2012), and Ma (2013) conducted ELA correlation analyses from different spatial and temporal scales, and the results showed that the evolution of landscape patterns was an important factor affecting the quality status of agroforestry ecosystems [78,127,128,129]. In addition, Zhang et al. (2010) and Chen et al. (2015) found that, to some extent, changes in landscape pattern indices could characterize changes in ecosystem quality [101,130]. For instance, areas with a large spatial extent of agroforestry landscape patterns, good connectivity, high-edge density, and low fragmentation could promote material exchange within the ecosystem and its own landscape, with which the corresponding ecosystem quality was higher [101,131]. However, this cannot be used as the only criterion, and it needs to occur in a specific habitat, such as agroforestry landscape under artificial management. Because of this, there are differences in their specific description parameters [29,49], resulting in immature research on the coupling mechanism of ELA. Therefore, it remains to be a key challenge to use landscape ecology theory to provide protection for agroforestry and surrounding habitat relics in different habitats [132] and to promote ecosystem quality and benefits in a sustainable manner [18].

4.1.4. Landscape Optimization Design

Globally, serious deterioration of the natural environment has sounded the alarm and presented significant challenges; an analysis of the causes of these problems has revealed that they were closely related to natural ecosystem changes [63]. Previously, we argued that the relationship between ecosystems and abiotic conditions in landscape planning for the alteration and use of nature has been emphasized, but the importance of landscape pattern optimization for ecosystems has often been overlooked [133]. Previously, we argued that the relationship between ecosystems and abiotic conditions in landscape planning for the alteration and use of nature has been emphasized, but the importance of landscape pattern optimization for ecosystems has often been overlooked [82].

In recent years, the interdependence between agroforestry landscape patterns and biotic and abiotic processes has been understood to some extent, and scholars have paid attention to optimization practices. For example, Chen et al. (2006) used 3S technology to conduct a multielement comprehensive analysis of agroforestry landscapes from four aspects: dynamic change in agroforestry landscape area, patch characteristics of landscape elements, landscape heterogeneity, and spatial relationships of landscape elements [134]. To understand the current situation and dynamic change trend of agroforestry landscape patterns, to carry out landscape planning based on this, and to coordinate the balance between ecological environmental protection and economic development, Huang et al. (2013) reduced water and sand export by optimizing land use structure and agroforestry spatial allocation, thus, promoting the ecological environment toward a virtuous cycle [135]. Burgess et al. (2018) established synergies through the careful selection and planning of components related to agroforestry landscapes to be more resilient and maintain long-term productivity [136]. Jeanneret et al. (2021) also attempted to rethink agroecosystems based on the theory of landscape ecology, by combining functional and hereditary aspects of biodiversity to create ecological corridors for different species [29]. The above cases fully prove that optimizing the landscape pattern of agroforestry is an effective measure to guarantee sustainable ecosystem quality and benefits.

However, solving the problem of agroforestry ecosystem quality in different regions at the landscape level and guaranteeing synergistic and sustainable ecological and socioeconomic development are limited based on theoretical knowledge only. It is also necessary to integrate multiobjective planning and management of agroforestry landscapes, while also taking into consideration the effects of geographically specific factors, such as severe soil erosion, low biodiversity, nutrient loss, and declining land productivity [112]. Thus, we can reveal the mechanism of the optimization of agroforestry landscape patterns on the improvement of ecosystem quality and reduce the obstacles in the process of regional ecological and socioeconomic synergistic sustainable development.

4.2. Key Scientific Questions

4.2.1. Clarify Key Factors Affecting Ecosystem Quality Change, with a Focus on Getting Away from Habitat-Specific Studies

Agroforestry, as an artificially governed ecosystem, has a high rate of input and output of material and energy [6]. Because of this, it is also subject to increased disturbances, and ecosystem stability is difficult to guarantee, leading to a risk of quality breaching the equilibrium threshold. In natural and seminatural habitats, the more complex the structure is, the more it contributes to ecosystem stability [16]. However, Qin et al. (2010) found that species in similar niches in agroforestry ecosystems had resource competition, which became one of the key factors affecting stability [137]. In order to solve the problem of unstable quality of agroforestry ecosystems, it is important to focus on agroforestry ecosystems in different habitats by getting rid of specific habitats, and to explore whether the known key factors are universally applicable and to identify the specific limits of their applicability. Furthermore, a complete monitoring system is needed to accurately identify other factors driving the degradation of agroforestry ecosystem quality and to make timely scientific decisions in response to changes. Clarifying the key factors affecting ecosystem quality changes can better guide the optimization of agroforestry landscape patterns and is a necessary prerequisite for ensuring the gain and quality of agroforestry ecosystem services.

4.2.2. To Solve the Current “Barrel Effect”, It Is Urgent to Clarify the Relationship between Landscape Pattern and Ecological Process in the Special Artificial Ecosystem of Agroforestry

Understanding the landscape patterns and ecological processes of agroforestry is a prerequisite for the scientific planning of agroforestry and a theoretical basis for achieving the efficient and sustainable development of agroforestry ecosystems. Only by building a reasonable structure of complex agroforestry systems (vertical community structure and horizontal spatial structure) (Figure 7) from the intrinsic characteristics of natural resources can the integrity of ecosystem functions and socioeconomic conditions optimize the overall function of agroforestry ecosystems [136]. Some scholars have begun to pay attention to the optimization of agroforestry community structure, but research on landscape patterns as a carrier of ecological processes has stayed at the stage of landscape pattern index analysis, which is only a reflection and evaluation of the existing landscape state [18]. Questions about quantifying the effects of different agroforestry landscape patterns on the realization of their ecosystem functions and how to use the analysis results to guide practice are worthy of further exploration in future research. This is also the key to solving the current series of problems, such as the “barrel effect” of agroforestry ecosystem services, which refers to uneven quality and easy rebounds.

4.2.3. Scientifically Assess the Effect of Landscape Heterogeneity on Agroforestry Ecosystems, Focusing on Multiscale Comprehensive Analysis Combined with Scenario Simulation Techniques

Heterogeneity of agroforestry landscape patches is a major driving factor of species diversity and ecological processes [138]. The promotion of species spillover across habitats and the enhancement of resource complementarity can be achieved by reducing the average area of patches so that crop conformational heterogeneity increases [139]. In practice, however, there may be a risk of increased fragmentation. There are various ways to describe landscape heterogeneity, and considering the spatial distribution based on planting patterns is one effective way [140]. However, exploring the landscape heterogeneity associated with agroforestry patch mosaics at large spatial scales based on this pathway and studying its role on ecosystems is still extremely challenging at the methodological level, especially for the description and mapping of agricultural practices. Moreover, habitat quantity and connectivity are key to maintaining metacommunity structure and persistence in the landscape [68]. How to identify these landscape strategic points among agroforestry landscapes requires multidisciplinary and multimethod intervention based on a landscape perspective integrating biology, ecology, and geography, future prediction with the help of scenario simulation technology, and integrated multiscale comprehensive analysis.

4.2.4. Promoting Agroforestry Landscape Optimization Practices Needs to Pay Attention to the Challenges of Spatial and Temporal Dynamics, and the Key Lies in Conceptual Guidance and Promoting a Change in Thinking

The sustainable development and implementation of agroforestry ecosystems requires a landscape ecology perspective that provides an important role in the spatial positioning and assessment provided by the sustainable development of agroforestry ecosystems [27,112]. This information helps to formulate policies for agroforestry development. However, the practical diffusion of agroforestry landscape pattern optimization strategies is hindered by landowners and farmers [29], mainly from the limitations of traditional perceptions. Therefore, based on our full understanding of agroforestry landscape patterns and ecological processes, we also need the local government to conduct in-depth visits and surveys, weigh and coordinate ecological requirements with farmers’ needs, strengthen farmers’ knowledge of this process, and remove farmers’ psychological barriers. In addition, agroforestry in nonuniform terrain is prone to changes in ecosystem interactions during trade-offs between different market economic positions. This requires that the practice of optimizing agroforestry ecological landscapes faces the challenges of the spatial and temporal dynamics of ecosystems. This is related to whether the overall function optimization and service gain of the agroforestry ecosystem can be achieved sustainably in the future and whether the ecological and social economy can develop in a coordinated and sustainable manner.

4.2.5. Continued Improvement of Service Supply Capacity Urgently Requires Attention to Future Climate Change and the Establishment of a Systematic Assessment System Combined with Resilience Theory

Enhancing the quality of agroforestry ecosystems and promoting ecosystem services requires a landscape perspective, and most ecological problems can be solved in new ways in the field of landscape ecology [29]. However, it is not clear whether the large-scale optimization of landscape patterns would still be effective in improving the stability, resilience, and recovery of agroforestry ecosystems in the event of unforeseen conditions (e.g., climate change and biological disasters) (Figure 8). This calls for the need to combine spatial-scale and temporal-scale research and to focus on resilience studies in future research to strengthen the resilience enhancement of mixed agroforestry ecosystems [141], which can help to refine landscape pattern optimization strategies to resist unknown risks [142]. Additionally, the long-term deployment of agroforestry ecosystem service function enhancement and the sustainability of benefits does not have a systematic assessment system. This requires an ecological assessment based on the landscape level to be combined with the use of 3S technology to consider long-term monitoring studies and to improve the assessment system by taking into consideration climate and other influencing factors.

4.3. Revelation on Enhancing the Sustainability of Agroforestry Benefits in KRD Control Areas

Due to the typical characteristics of the ecological vulnerability of KRD [143], the overall regional ecosystem is less resistant to disturbance, putting the ecosystem quality at risk of breaching the threshold [18]. Uncontrolled human land use and irrational agricultural activities exacerbate habitat disturbance and affect the landscape pattern and ecological processes of agroforestry ecosystems, thereby, undermining the sustainability of agroforestry ecosystem services [13,110]. Therefore, integrating the review work of the above ELA studies in this paper, the main focus is on the difference in natural habitats and the science of landscape planning, highlighting the effectiveness of landscape ecology theories and methods in maintaining sustainable agroforestry benefits of KRD management.

In terms of natural habitat conditions, the various intricate patches of karst landscapes in karst mountains are the result of continuous tectonic uplift, denudation, erosion, and deposition in the mountains over a long period of time [144]. This, coupled with the unique carbonate bedrock, has led to slow soil formation processes and shallow and discontinuous soils in karst mountains, resulting in a severely fragmented landscape (Figure 9b) [143]. This has also become a key limiting factor in enhancing the quality and services of agroforestry ecosystems in KRD areas, especially in terms of stability and supply capacity. However, agroforestry landscape patterns can determine their ecological processes [13], and thus, influence ecosystem quality and service functions [101]. Therefore, in summary, based on the characteristics of the current landscape pattern of agroforestry combined with the special habitat conditions in KRD areas, the strategic points of the landscape pattern to be optimized according to the needs of ecological processes in agroforestry are identified [111], which can be practiced in the following five aspects. First, we selected suitable plants with lithophytic, arid, and calcium-loving characteristics according to soil properties in karst areas [16] and optimized the vertical community structure of agroforestry in response to the competition between soil and water nutrients. Second, soil erosion is a direct factor leading to landscape heterogeneity and increased stone desertification in karst areas [143], and hydrological and soil processes should be further studied to clarify the drivers of agroforestry ecological processes and their variability under different KRD classes. Third, the ecologically sensitive areas within the agroforestry landscape and its surrounding landscapes should be highlighted, as they are the strategic points of the landscape, and the theory of landscape ecology should be used to combine engineering measures with biological measures to optimize the landscape pattern. Fourth, the feature that runoff and shading effects on the surface of rocky bare areas can result in higher surrounding soil moisture and nutrients [145] was fully utilized to analyze the different mosaic combination structures of soil patches and rocky bare patches (Figure 9a) to create different ecological niches for agroforestry [146]. Fifth, we expanded the research on the temporal and spatial scales to comprehensively consider regional limiting factors (such as water in the western region and accumulated temperature in the eastern region) and the ability of agroforestry ecosystems to resist disturbance and self-repair under extreme climate scenarios [147,148], carried out a suitability assessment for the development of agroforestry in the region, and scientifically planned the spatial layout of agroforestry and optimized the landscape pattern. The above ideas will help agroforestry ecosystems to maintain efficient material cycles and energy flows in the long-term evolution of the future [149]. This plays a key role in enhancing the quality of agroforestry ecosystems and refining the sustainability of agroforestry ecosystem services.

With regard to sustainable development, high population densities in karst areas have led to a surge in human demand on ecosystems (e.g., food and timber), and irrational agricultural production activities have exacerbated ecological degradation [77]. Faced with the question of how economic and ecological synergies can be developed, agroforestry has been shown to have synergistic benefits of protecting karst ecosystems and sustaining farmers’ livelihoods [49,150,151]. However, facing the problem of how to maintain the sustainability of the benefits of agroforestry in KRD regions, we emphasized the importance of landscape-level research and analyzed the agroforestry landscape in depth through the above five aspects. By implementing the theories and methods of landscape ecology into practical planning actions, we can transcend the simple cognitive stage and can provide the necessary prerequisites for promoting the sustainable development of agroforestry ecosystem services. At the same time, the transformation of ecological governance thinking in karst areas is also one of the key requirements. The current control of KRD focuses on prevention and control, which is to prevent and control the continuation or aggravation of water and soil loss events, and hope for habitat restoration. This highlights that the current way of thinking is defense, not coexistence. Therefore, we need to change our concepts, expand our thinking, follow the rules, combine the current situation of land use on the basis of prevention and control, and develop more intelligent methods to change our thinking from defensive to accepting. This is about whether ecological and socioeconomic development can be synergistic and sustainable and whether the effectiveness of global KRD control can be consolidated.

4.4. Limitations of the Study

Prior to this study, some of the reviewed studies focused on agroforestry ecosystems, with multifaceted discussions devoted to bridging knowledge gaps and linkages about the ecosystem services provided by agroforestry. For example, Jose (2009) and Moreno et al. (2018) systematically elucidated the multiple benefits of agroforestry ecosystem services through methods such as meta-analysis and revealed that agroforestry ecosystems had functions such as enhancing biodiversity and ecosystem services [6,152,153,154]. A few reviewed studies also focused on the improvement of agroforestry ecosystems. For example, Jiang et al. (2022) elucidated the relationship between the structure and stability of agroforestry ecosystems and proposed effective methods to improve the stability of agroforestry ecosystems [16]. The review by Xiao et al. (2022) focused on how the generative elements and management of agroforestry ecosystems affected agroforestry ecosystem services [17]. In contrast to most current studies, we focused on the quality and landscape patterns of global agroforestry ecosystems based on landscape ecology and explored the drivers of agroforestry ecosystem quality and how optimization of landscape patterns could enhance their ecosystem service provisioning capacity. Therefore, we summarized the landmark results of ELA and discussed, based on key scientific questions, how to optimize agroforestry landscape patterns to enhance ecosystem quality for the sustainable provision of agroforestry ecosystem services. Finally, an effective form of sustainable agroforestry ecosystem services in KRD control areas was clarified based on the results of the study. These revelations will be beneficial for agroforestry-related research based on the landscape perspective to gain more attention and discoveries in the KRD region.

In addition, during the process of searching the literature, although different search engines and screening mechanisms are used, there is always human subjective consciousness that interferes with judgment, which makes the results of the literature search in this paper have a degree of uncertainty. For example, in the “Included” section, the full text was read to determine whether it was highly relevant to ELA and whether it was eligible for inclusion, which makes the characterization of the literature subjective and leads to possible omission of studies in the literature. We plan to try to find a more objective way to determine and optimize this process in our next study.

5. Conclusions

In this paper, a literature search of ELA-related studies was conducted through the Scopus core database, and 163 obtained papers were systematically analyzed and reviewed. The main conclusions were as follows: (1) There was an overall upward trend in the annual number of publications on ELA, which underwent a transition from a nascent to a rapidly growing period, leading to an improved understanding of agroforestry landscape patterns and ecological processes. (2) In terms of research branches, ecosystem quality research accounts for the major part (36%), landscape pattern (26%) and benefit coupling mechanism (23%) research has been developed synergistically, and landscape optimization practice (9%) lags relatively behind other branches, which highlights the failure of synergy between practice and theory. (3) Current progress and landmark results have focused on topics such as cropping patterns, community structure, habitat variation, driver analysis, index response, and method optimization. The focus needs to be on clarifying the interactions of ELA and how they affect and enhance the capacity of agroforestry ecosystem service provisioning. (4) Future research related to agroforestry ecosystems needs to go beyond the field scale, and exploring the sustainability of agroforestry ecosystem benefits based on the landscape scale or a larger spatial scale has become key to addressing the risks. In the research process, it is necessary to combine spatial and temporal scales, cooperate across disciplines and pay attention to regional habitat specificities, to clarify the key drivers of agroforestry landscape patterns and ecological processes in different regional types, and to help improve landscape pattern optimization strategies to guide practice. (5) The practice of optimizing agroforestry landscapes in karst areas needs to face the challenges of spatial and temporal dynamics of ecosystems, to pay attention to possible future contingencies (e.g., the challenges of climate change) and the specificity of rocky desertification habitats, and to integrate biological and engineering measures for scientific planning as an effective way to maintain sustainable agroforestry ecosystem services.

Another important point is that the mindset of karst ecological management needs to change from “prevention” to “coexistence”, respecting the natural base characteristics of karst’s “binary three-dimensional” hydrological structure. The main factor of ecological damage in karst areas is the human–land conflict, i.e., the relationship between human and nature has not been improved. However, with the decoupling of human–land relations due to a massive labor exodus in recent years, we should seize this opportunity to vigorously develop agroforestry and to promote the optimization of landscape structure, and therefore, expand the ecological benefits of agroforestry and to make optimized agroforestry a smarter governance measure, taking advantage of its attributes to see rocky outcrops as friends, and change the governance mindset from defensive to receptive. This requires policy guidance that recognizes the environmental benefits provided by agroforestry ecosystems and encourages widespread implementation of optimization measures. However, individual farmers have limited land and can only be developed and optimized at the farm scale. Therefore, this calls for promotion by policy makers with a higher strategic perspective and learning from developed countries that have advocacy, research, and integration as strategic goals and set specific tasks for the various agencies of the Ministry of Agriculture in the development of agroforestry. Meanwhile, future ELA research work should be based on the relationships between humans and nature and between economic and ecological needs and also be based on the actual situation of geographical habitats in the KRD region. Combined with the main progress and key scientific questions we summarized, this information can be used to inspire policy optimization of decision makers and land management of landowners and to further promote the supply enhancement of agroforestry ecosystem services.