Study area

The garden studied was called "Huerta UNAL" (UNAL Garden) within the framework of the Agrobiodiverse Biosystems project funded by "Convocatoria Nacional de Alianzas Interdisciplinares" (National Call for Interdisciplinary Alliances). The garden was located in a plastic-covered greenhouse of the Faculty of Agricultural Sciences of the Universidad Nacional de Colombia - Bogota campus (4°38'8.9" N, 74°05'21.7" W) at 2640 m a.s.l., with an area of 280 m². The average maximum temperatures external and internal to the greenhouse fluctuated between 18°C and 20°C and 22°C and 24°C, respectively. At dawn, the minimum external temperature was between 8°C and 10°C, and the average total annual rainfall in the area is 797 mm, distributed in two dry seasons and two rainy seasons. The characteristics of the soil are shown in Tables 2 and 3.

For the physical-chemical analysis of the soil, samples were collected from the first 30 cm of soil depth; the first 2 cm of soil was removed, and the sample was extracted, taking about 10 sub-samples throughout the length and width of the area, which were then mixed in the bucket to form one soil sample. Following the ^{IGAC (2021)} methodology, a zigzag pattern was formed throughout the area, as it is the most appropriate and simple method to recollect the 10 soil sub-samples. The samples were sent to the Agrosoil laboratory (Bogotá) for analysis. The soil texture was determined according to the ^{Bouyoucos (1936)} method, the soil organic matter content was quantified according to the ^{Walkley and Black (1934)}, the bulk density was determined with the paraffin-coated clod method according to ^{Blake and Hartge (1986)}, and the real density was taken from ^{FAO (2022)}. Furthermore, the description of soil structure was based on characteristics defined by ^{Jaramillo (2002)}, in which the soil structure is classified by shape, kind and evolution grade.

Soil preparation

The soil was initially very compacted and degraded and had high contents of clay (52%) and a low content of organic matter (3.1%) and nitrogen (0.21%) (Tab. 2). The interpretation of compaction was made by taking the value of bulk density of the studied soil of 1.3 g cm^-3 and comparing it with the reference values presented by ^{Jaramillo (2002)}, who describes that those predominant soils of fine texture greater than 1.3 g cm^-3 indicate soil compaction. Potassium (0.88 cmol kg^-1) and calcium (6.38 cmol kg^-1) were in excess, unbalancing the ionic relations in the soil (Tab. 3). To prepare the soil, intensive amendments with organic matter were applied 15 d before transplanting, using products from commercial brands Lombritenjo^® granulated worm humus (Vermiculture of Tenjo, Tenjo, Colombia, 2 kg m^-1), Solidblend^® organic fertilizer with biochar (Biodiversal, Bogotá, Colombia, 2 kg m^-1) or material from other greenhouses which was transported on wheelbarrows to be evenly incorporated into the soil of the study area using a hoe.

Before transplanting, the double digging method was used and weeds were removed manually based on the Las Cañadas Agroecological Center (^{Centro Agroecológico las Cañadas, 2009}) methodology, using a biointensive approach (^{Jeavons, 2001}) to ensure that the crop management was in line with agroecological principles. After building the sections, 1 kg m^-2 of commercial biochar (^{Pentón Fernández et al., 2021}) was added to the soil 15 d before transplanting, with beneficial microorganisms such as phosphorus-solubilizing fungi Penicillium janthinellum (Fosfobiol®, Biocultivos, Ibagué, Colombia - 5 ml L^-1 - soil drench application) and plant-pathogenic fungi controllers such as Trichoderma spp. (Fitotripen®, Natural Control, La Ceja, Antioquia, Colombia - 2 g L^-1 - soil drench application) applied 15 d after transplanting.

Garden design and establishment

The 280 m² area was divided into two half-greenhouses measuring 14 m x 20 m. The first half was covered with a plastic roof, and the second was left uncovered. Inside each part of the greenhouse, six beds of 6 m x 3 m were spaced 0.4 m apart (Fig. 1); this was done to observe the emergence of the distinct kinds of weeds in the garden where several species were planted at a conventional density for each species (Tab. 1) since production was for self-consumption. The plant species established were vegetables, aromatic herbs, fruit trees, tubers, legumes, and cereals, with the largest number of species in the first two categories. Inside the main cultivated species were tomato, corn, potato, vegetables such as lettuce, carrot, radish, and herbs such as basil and oregano. These species were planted with the biointensive cultivation method, with the distances between plants and rows reduced to achieve a higher density of plants per 1 m² and boost yield (^{Pérez & Hernández, 2022}).

FIGURE 1 Greenhouse sections layout. The 280 m² area was divided into two half-greenhouses measuring 14 m x 20 m. The first part of the greenhouse was covered with a plastic roof (gray background), and the second was an open space (brown background). Some of the species cultivated are shown. 

The seedlings were germinated and cared for before transplanting to the seedling production area of the Universidad Nacional de Colombia. Their planting and distribution were previously designed considering some relationships between plant species; for example, plants of the family Lamiaceae, such as Mentha spicata, that exhibit allelopathy (^{Islam et al., 2022}) that would facilitate management and reduction of phytosanitary problems, such as pests, through its insecticidal, antimicrobial, and antifungal capacity inside the garden.

For the transplanting, maintenance, and harvesting processes, collaborative workdays (mingas, in Spanish) were promoted through different means to invite university students to participate. As a result, the activity would strengthen learning about the establishment and care of an urban garden among people unfamiliar with the basic processes of agricultural production.

Biological products were used for pest and disease management. For pest management, some of these products were applied at different times of the day, including Bacillus thuringiensis (Subticrop^®, Bio-Crop, Palmira, Colombia - 2 ml L^-1 - soil drench application), Beauveria bassiana, Lecanicillium lecanii, Metarhizium anisopliae and Bacillus thuringiensis (Sáfermix WP^®, Sáfer, Medellín, Colombia - 2 g L^-1 - soil drench application), potassium soap (Bonfyton^®, Biorracionales de Colombia, Bogotá, Colombia - 2 ml L^-1 - foliar application), and for disease management Trichoderma harzianum (Fitotripen^®, Natural Control, La Ceja, Antioquia - 2 g L^-1 - soil drench application), and copper sulfate (Antrasín^®, Sáfer, Medellín, Colombia - 2 g L^-1 -foliar application) were used.

Finally, crop fertilization was based on the application of products such as Actiphyl Kfruto^® and Nutriponic^® (Ingeplant, Colombia) and compost obtained from biodigester bales located in a greenhouse near the garden, prepared openly and in piles, as schematized by FAO for its preparation in Latin America (^{Román et al., 2013}).

Soil seed bank behavior

The germination conditions mirrored those of the garden location, with a temperature of 22°C±4°C, relative humidity ranging from 40 to 90%, and a standard photoperiod of12 h. The analysis was conducted under conditions similar to those of the garden to gain insight into the species that germinate more rapidly and predominate in the garden's soil.

The weed characterization was made to identify invasive and competitive species in the garden soil. This information was used to plan the planting of species in the next planting cycle, selecting those that could outcompete the rapidly emerging weeds. Additionally, it allowed implementation of a preventive weed management strategy.

For the analysis of weed behavior, soil samples were collected in the uncovered area of the garden, after three months without any intervention and prior to the next planting cycle, on March 14, 2022. The samples were gathered from the first 30 cm at four points in the garden (^{Cardenal Rubio et al., 2016}) and placed in 20 pots of 10 cm in diameter. Irrigation was carried out daily to allow the evaluation of the seed bank of weeds in the soil with a maximum temperature of 24°C and a minimum of 8°C. Maximum and minimum humidity of soil were 73% and 32%, respectively. The weed species were then observed, identified, and quantified. A first destructive sampling (S1) was carried out using 10 pots when achieving 100% coverage to identify species that stopped growing due to competition with others until 30 d after seeding (DAS). Subsequently, all weeds were removed from the pots. A second destructive sampling (S2) was carried out 60 DAS in the same pots to observe the species of the bank that were more competitive and established based on the ^{Cardenal Rubio et al. (2016)} methodology. Finally, Equation 1 for Relative representation (Rr) was utilized to calculate the importance of each weed species in the ecosystem, where the Species value represents the assessment of a species, and Total density of all species is the total number of species found in the sample (^{Moreno-Preciado et al., 2021}):

Finally, the weed identification process was carried out through morphological characterization of seedlings both in a controlled environment and in the field. We used the reference materials "Seedlings of common weed species in the central zone of Colombia" (^{Fuentes et al., 2006}) and the "Illustrated guide to weedy plants of the Marengo agricultural center" (^{Gámez et al., 2018}).

Limitations associated with soil poor structure and texture

The clay content was predominantly higher in the studied soil. These clays can be defined as low-activity clays based on the observations in the field that evidenced deficiencies as a result of the lack of exchangeable elements available for the plant (^{Jaramillo, 2002}). Additionally, the absence of soil structure prevents the breakdown of inorganic materials, which in turn inhibits the production of elements that plants can metabolize (Tab. 2).

Also, the soil exhibited small pores related to fine textures and poor structure without sand soil. ^{Jaramillo (2002)} attributed these soil features to compaction. The soil had clayey clods of moderate resistance and larger than 80 mm throughout the intervened area. This is the product of anthropogenic actions associated with the material in the area, which is poor in weatherable materials and sometimes contains rubble of different sizes and plastic debris linked to the low soil evolution degree. ^{Del Valle Neder et al. (2010)} describe the anthropic effect as a cause of loss of productivity in soils, stating that human intervention acts as a modifying factor of the environment, forming soils of anthropogenic origin. This garden is a clear example of such an effect.

These characteristics make the establishment of cultivated plants even more difficult, affecting the optimum physical properties of the soil for the root growth, such as pore space, air capacity, moisture retention, actual and bulk density, and resistance to compaction (^{Antúnez et al., 2015}).

Therefore, soils with anthropic influence that are typical of urban spaces must be renovated and conditioned to be used in urban gardens. It is important to define the ideal soil practices that can help to transform low-fertility soils into areas with the desired fertility and the ability to sustain plant growth. An ideal scenario is described by ^{Torres Sanabria and Cuartas Ricaurte (2013)}, who mention that the pre-Columbian cultures of the Amazon transformed low-fertility soils into highly productive anthropic soils called Tierras prietas. One of the numerous examples of soil conditioning is the use of aromatic plants that improve soil conditions such as pH and organic carbon availability (^{Dikr, 2022}).

Alternatives to soil limitations

While the addition of organic matter to the studied soil improved its structural characteristics, such as porosity, texture and friability, various authors propose alternatives that expand the options for soil enrichment and which may be considered in the future. For instance, ^{Pozza et al. (2020)} state that the use of cover crops and the incorporation of organic matter not only maintain soil structure and prevent soil exposure, but also promote diversity within the system. They also suggest that the addition of organic matter to soil should go hand in hand with the cultivation of resilient crop varieties and the incorporation of nitrogen-fixing legumes to reduce the use of fertilizers.

The product Lombritenjo^® was applied using vermicom-posting, in which earthworms convert organic materials into humus. It enhances soil fertility, provides more nutrients to plants, and adds more porosity, density, aeration, and water retention to the soil (^{Lim et al., 2015}). The biochar Solidblend^® improves the structure of compacted soil. ^{Allohverdi et al. (2021)} mentioned that compost mixed with biochar produces higher biomass degradation that provides nutrients to plants rapidly. Thus, using biochar's porosity improves soil aeration and increases porosity.

Fosfobiol^®, which contains Penicillium janthinellum isolates as part of the Penicillium species, was also used. ^{Hao et al. (2020)} identified its capacity to solubilize inorganic P, making it available for the plant. Fitotripen^®, which contains Trichoderma spp., was used to increase the soil health. ^{Zin and Badaluddin (2020)} stated that it works as a plant growth promoter and increases the decomposition of organic matter to enhance the availability of nutrients in the soil.

Finally, the utilization of arbuscular mycorrhizae frequently establishes beneficial associations that enhance nutrient uptake by plants through a symbiotic relationship, ultimately ameliorating the fitness and growth of the plants integrated within the system (^{Alyokhin et al., 2020}).

Limitations associated with the presence and abundance of invasive species

Thirteen weed species were found in the soil of the UNAL Garden (Tab. 4). These species corresponded to 10 families that were recorded throughout the experiment, but only 8 remained until the end. Of these weed species, 84.6% were broad-leaved and 15.4% were narrow-leaved (Tab. 4). The most represented species calculated with the Equation 1 of relative representation (Rr) in the first sampling was Veronica spp. (43.6%) (Fig. 3E), followed by Lolium temulentum (21.6%) and Oxalis corniculata (11.7%) (Figs. 2 and 3G).

In the second sampling, the most relative representative species was also Veronica spp. (33.3%), followed by Lolium temulentum (21.6%) and Cenchrus clandestinus (17.7%). All the species representations tended to decrease, except Spergula arvensis (Fig. 3D) and C. clandestinus (Fig. 3H). In addition, in the first sampling, C. clandestinus represented only 2.7% of the number of individuals found in the garden but was the most difficult to manage. Although the other species were more represented in the samplings, they were easily controlled manually (Fig. 3A).

The characteristics of Veronica spp. as a highly relative representation are explained by ^{Bond et al. (2007)} who observed that this species grew on calcareous and acid soils. These features were found in the soil during the experiment, which could explain the predominance in the relative representation. The authors claimed that Veronica spp. makes up a persistent seed bank which could explain its major predominance in the soil bank of weeds.

FIGURE 2 Weed species found in the UNAL Garden (Bogota). Samples "S1" and "S2" were collected up to 30 and 60 d after seeding (DAS), respectively. The x-axis shows the different weed species and the y-axis shows the number of individuals per number of analyzed samples, with n = 10. 

The results of the bank of weeds in the soil test contrasted with what was observed in the study area. During the establishment of the garden, the weed C. clandestinus (Kikuyu grass) became a major problem since it was rooted up to 30 cm deep and showed an abundance of rhizomes (Fig. 3H) (Cordero ^{Rodríguez et al., 2021}). These characteristics make С. clandestinus a species with excellent reproductive potential which, added to its ecology (a perennial species with a wide range of adaptability to different environments), results in a very invasive grass species. It reproduces by rhizomes and stolons, allowing it to cover the entire area where it is found (^{Contexto Ganadero, 2022}). This behavior negatively affects the biosystem and diversity which is of utmost importance for the maintenance and conservation of agroecosystems (^{Camacho-Ballesteros, 2018}; ^{Rodriguez, 2021}).

Although competition and existing allelopathy were considered for the cultivable species, an in-depth diagnosis of the bank of weeds in the soil was not performed, which would have prevented competition and drastic invasion by C. clandestinus (Fig. 3A). ^{Camacho-Ballesteros (2018)} pointed out that this species of grass competes with cultivated plants for water, light, and space. The use of water resources is the most critical problem, not only because of competition but also because this species prevents water infiltration, seriously affecting water availability in reservoirs and the soil.

FIGURE 3 Photographic record of the presence of weeds in the UNAL Garden. A) Cenchrus clandestinus growing around cabbage, chard and lettuce sections in the UNAL Garden, B) Galinsoga quadriradiata, C) Trifolium spp., D) Spergula arvensis, E) Veronica spp., F) pots of the soil bank of weeds, G) Oxalis corniculata, and H) Cenchrus clandestinus rhizome found during the trial. 

The relative representation by Veronica spp. during the experiment does not relate with the establishment of the different species of crops over the field. Since the biology of C. clandestinus depends on the competition of water and nutrients, it might have a higher impact on the production system. The opposite occurred with Veronica spp., which had a low persistence in the field, although it was the most representative in the experiment in pots. This shows the importance of knowing the biology and weather response of these different weeds in the production system.

Although the problems of managing and maintaining a biointensive garden can affect the continuity of urban agriculture projects, there are relevant factors that are rarely considered. For instance, compacted soil can be improved by adding various materials, such as organic matter from diverse sources, to enhance the soil structural integrity and, in some instances, supply essential mineral elements for plant growth.

Additionally, soil conditioning techniques can be implemented and the bank of weeds managed by selecting beneficial weed species, creating the possibility to use the qualities of distinct plants in a way that allows soil improvement and the establishment of different crops. Finally, plant growth-promoting rhizobacteria can be used along with the addition of vegetal material and mycorrhizae to create a symbiosis with plants.