Study Area

The research was conducted during 2017 in the municipalities of Ventaquemada (5° 22' 20" North - 73° 31' 22" West), Turmequé (5° 19' 26" North - 73° 29' 26" West) and Tibasosa (5° 44' 46" North - 73° 00' 04" West), located in the central altiplano of the department of Boyacá, Eastern Cordillera of the Colombian Andes (Fig. 1). The climate is temperate according to the Köppen classification, with an average annual temperature of 14.6 °C and an average annual rainfall of 939 mm (climate-data.org). The farms surveyed in this study are distributed between 2380 and 3050 m altitude, which corresponds to Montane humid forest (^{Holdridge 1987}) with a predominance of the plant families Asteraceae, Melastomataceae, Rosaceae, and Poaceae.

Figure 1 Location of the municipalities and farms studied (in green). 

Agriculture and livestock production are the main economic activities of this region. Some of the most commonly cultivated species are corn (Zea mays L.), cubios (Tropaeolum tuberosum Ruiz & Pav.), squash (Cucúrbita pepo L.), potatoes (Solanum tuberosum L.), papayuela (Vasconcellea pubescens A.DC.), broad beans (Vicia faba L.), arracacha (Arracacia xanthorrhiza Bancr.), peach (Prunus persica (L.) Batsch), and beans (Phaseolus vulgaris L.) (^{Sánchez 2018}). These agricultural systems have gone through a historical transformation. According to ^{Sánchez (2018)}, during the 1960s traditional agricultural systems used native seeds and organic inputs. But in the mid-1970s the implementation of a commercial production model increased the use of agrochemicals and machinery, promoting the elimination of native forests to expand the agricultural frontier. At this time, reforestation activities were also promoted using exotic tree species such as pine (Pinus patula Schiede ex Schltdl. & Cham.), eucalyptus (Eucalyptus globulus Labill.) and acacia (Acacia melanoxylon R.Br.). These conditions lasted for a few decades; however, farmers began to see a significant decrease in their natural resources vital for food production. This situation marked the current trend of conversion to family smallholder systems, adopting more environmentally friendly practices. Nowadays, farmers prefer to use organic inputs, as well as soil and water conservation practices, planting trees as living fences or conserving natural vegetation relics (^{Sánchez 2018}).

Data collection

The selection of the smallholder farms was carried out during three workshops with the community, one in each municipality. Members of various community organizations participated in these workshops. In each municipality, eight farms were chosen, for a total of 24 in the study. For their selection, the following was considered: that the family had only one property; that the farm owner was a member of the family group; and that agricultural activities contributed to the family incomes.

To characterize socio-economic traits of the families and physical aspects of the farms, semi-structured interviews were conducted with the farm owners. Variables that were taken into account for the analysis were the following: owner's age and schooling, household members working at the farm, farm's age (owner tenure time), area, distance to natural vegetation and distance to the closest urban center. Aspects of farm management such as the implementation of polycultures or reforestation activities were also documented.

Through the technique of ethnobotanical walks (^{Martín 2001}), tree species were located and questionnaires were applied to record each species, its growth habit (shrub or tree), use and management category (wild or cultivated), and the area of the farm where it was located. Three categories were defined to classify areas within each farm: (1) the production areas (includes the garden, the cultivation and grazing areas); (2) living fences; and (3) areas with wild vegetation (includes forest relics, thickets, and creek edges). Uses were classified into seven categories following ^{Pérez and Matiz-Guerra (2017)}: (1) agroecological (includes use as living fences); (2) food; (3) environmental (includes uses for protection of water sources); (4) handicrafts; (5) construction; (6) medicinal; and (7) ornamental.

The species were preliminarily identified in the field and botanical samples were collected for complete identification in the Herbarium of the Pontificia Universidad Javeriana (HPUJ). The nomenclature of the species and the information on their origin (native or exotic) was based on the Catálogo de Plantas y Líquenes de Colombia (^{Bernal et al. c2019}).

Kruskal-Wallis tests were used to compare species richness, and farm size and altitude between the three municipalities. Correlations between species richness and owner's age and schooling (from the most basic to the most advanced, primary, secondary and technical), household members working at the farm, farm age and area, distance to the natural vegetation relicts, and distance to the closest urban center were assessed using Spearman's rank correlation coefficient.

Structure of households and farms studied

The size of the farms ranged from 0.02 to 8 ha (X = 1.96; SD = 1.86) (Table 1). Both the size and altitude at which the farms were located did not vary among the three municipalities (P > 0.05, Kruskal-Wallis test). The distance of the farms to the closest urban center ranged between 0.92 and 6.96 km (X = 3.45; SD = 1.61). Polycultures were planted in 23 farms, and in twelve farms there were areas with natural vegetation or the perimeter of the farm bordered areas with natural vegetation, such as forest relics or riverine vegetation. The distance of the farms to the natural vegetation varied between 0.01 to 1.21 km (X = 0.36; SD = 0.39).

Farm owners' ages ranged from 28 to 72 years (X = 54.63; SD = 12.24) (Table 1). The families were composed of three to nine members, but only between one and five participated in the farm work. The level of schooling was variable: half of the owners had only primary education, while three had not studied and nine had secondary, technical, or professional education (Table 1). The families lived between 5 and 33 years on the farms. The majority (79 %) of the owners dedicated all their work time to the farm.

Tree and shrub richness and composition

In total, 142 species of trees and shrubs were identified (Table 2). The plant family with the greatest richness was Asteraceae (17 species), followed by Solanaceae (eleven species), Fabaceae and Rosaceae (nine species each), Myrtaceae (eight species), Melastomataceae (seven species), Verbenaceae (five species), and Ericaceae, Malvaceae and Rutaceae (four species each). The most species-rich genera were Baccharis and Solanum with four species each, and Citrus, Fuchsia, Miconia, Prunus and Syzygium with three species each.

Table 2 List of trees and shrubs in smallholder farms of the central altiplano of Boyacá, Colombia. *Endemic species. Use categories: Agroecologica (Agr), Food, Environmental (Env), Handicrafts (Han), Construction (Con), Medicinal (Med), Ornamental (Orn). 

The number of tree and shrub species per farm ranged between four and 40 (X = 25.17; SD = 10.13) (Table 1). Species richness was homogeneous among the farms of the three municipalities (P > 0.05, Kruskal-Wallis test). The most frequent species in the farms were Sambucus nigra L. and Vasconcellea pubescens, which were recorded from 18 farms, Eucalyptus globulus and Fuchsia boliviana Carrière in 17 farms, Acca sellowiana (O.Berg) Burret in 16 farms, Prunus persica in fifteen farms and Solanum betaceum Cav. in thirteen farms (Table 2). In general, these are species used for their edible fruits or as living fences. In contrast, about 48 % of the species (68) were registered only once or twice in the smallholder farms.

Most of the species documented on the farms were shrubs (68), 60 were trees, and fourteen were recognized as trees or shrubs. Most of the species were native (88), of which eight were endemic to Colombia (Table 2); 54 species were introduced from other regions of the world. Likewise, most species were cultivated (75), 52 were wild and fifteen species were wild or sometimes cultivated. The most frequent wild species were Oreopanax incises (Schult.) Decne. & Planch., Miconia squamulosa Triana, Myrsine guianensis (Aubl.) Kuntze, and Monnina aestuans (L.f.) DC., while the most frequently cultivated species were Vasconcellea pubescens, Eucalyptus globulus, Acca sellowiana, and Prunus persica. Some of the species that are both wild and cultivated on farms were Sambucus nigra, Alnus acuminata Kunth, Myrcianthes leucoxyla (Ortega) McVaugh and Fuchsia boliviana (Table 2).

The distribution of these species in the farms was as follows: 117 species were found in productive areas - that is, in home gardens, cultivated areas or scattered among pastures -, 70 of which are native. We also found 54 species in living fences, 39 of which are native, and 43 species in the unmodified original vegetation stands that are present on some of the farms, 41 of which are native.

We documented the use of 96 species which were classified into seven categories. Ninety-eight percent of cultivated species and 23 % of wild species had at least one use; no uses were reported for two cultivated species. The most frequent use category was Agroecological since 39 species were used as living fences, including Sambucus nigra, Eucalyptus globulus, Alnus acuminata, and Fuchsia magellanica Lam. We found 37 species with ornamental uses; the most frequent were Sambucus nigra, Fuchsia boliviana, Fuchsia magellanica, and Pittosporum undulatum Vent. We recorded 29 species used as foods, especially fruits; the most frequent were Vasconcellea pubescens, Acca sellowiana, Fuchsia boliviana, Prunus persica, and Solanum betaceum. Fifteen species were registered as medicinal, of which the most frequently used were Sam-bucus nigra, Smallanthus sonchifolius (Poepp.) H.Rob., Cupressus lusitanica Mill., Rosmarinus officinalis L., Eucalyptus globulus and Aloysia citriodora Palau. Only two species (Miconia squamulosa and Myrsine guianen-sis) were used for construction, other two species (Escallonia paniculata (Ruiz & Pav.) Schult. and Abatia parviflora Ruiz & Pav.) were used for environmental purposes, and only Erythrina rubrinervia Kunth was used to make handicrafts (Table 2).

Socio-economic and agroecological factors and species richness

A positive correlation was found between species richness and farm size (r_s = 0.664, P << 0.001) and a negative correlation between species richness and distance to areas of natural vegetation (r _s = -0.515, P = 0.01). As for the other socioeconomic and structural variables, no correlation was found in relation to species richness.

Diversity

The results show an important diversity of tree species associated with smallholder farms. In these spaces, wild native species converge with native or exotic species planted by the owners. Wild species correspond to elements of the original vegetation that were not destroyed or to new elements that have arrived through natural dispersal. It has been documented that smallholder farms can act as sinks for wild species (^{Rooduijn et al. 2018}). Plants dispersed by wind or birds and with a high tolerance for solar radiation can be successful in these types of transformed environments (^{Thijs et al. 2015}). These traits likely explain the occurrence of species of the Asteraceae, Ericaceae or Me-lastomataceae families among the wild plants of the farms.

Another interesting aspect of tree and shrub diversity on the farms is the predominance of native species over exotic species, consistent with what is reported in the literature for smallholder farming systems in the Neo-tropics (^{Albuquerque et al. 2005}, ^{Garen et al. 2011}, ^{Abebe et al. 2013}, ^{Pérez and Matiz-Guerra 2017}, ^{Rooduijn et al. 2018}). However, although exotic species were less diverse than native ones, they are the most frequent on farms. In fact, of the seven most frequent species, six were exotic and only one was native. This lower diversity but high frequency of exotic species may be related to the fact that in human-modified systems exotic species tend to offer many more services (^{Garen et al. 2011}). In fact, 98 % of the exotic species were useful in the studied farms. In contrast, native species were less frequently utilized and were used in a smaller proportion (23 %). The most frequent native species were found in no more than half of the farms and corresponded to cultivated species, all with some use-value for the owners.

Regardless of the origin and the diversity of tree and shrub species found in smallholder farms, our results suggest that these productive spaces may have a key role in biodiversity conservation. In this sense, several authors have reported that environments altered by humans, such as home gardens, are spaces for the conservation of wild species and are germplasm reservoirs of a wide variety of cultivated plants (^{Bhagwat et al. 2008}, ^{Pulido et al. 2008}, ^{Webb and Kabir 2009}, ^{Rooduijn et al. 2018}). Some of the most characteristic genera of the Andean flora, such as Baccharis, Solanum, Fuchsia, Miconia, Piper, Chromolaena, Cestrum, Oreopanax, Tibouchina, Myrsine or Monnina (^{García et al. 2006}, ^{Fernández-Alonso and Hernández-Schmidt 2007}, ^{González-M and López-Camacho 2012}) are represented by one or several species within the farms studied (Table 1). Likewise, it is significant that eight of the wild species are endemic to Colombia (^{Bernal et al. c2019}). Additionally, some of the genera of cultivated plants frequently reported in Neotropical agricultural systems such as Abutilon, Annona, Citrus, Inga, Manguifera, Prunus, Psidium, Solanum or Syzygium (^{Pulido et al. 2008}, ^{Pérez and Matiz-Guerra 2017}, ^{Pinto Rayol et al. 2017}, ^{Villa and García 2018}), are also represented in the farms studied with one or more species (Table 1). These results reinforce the idea that smallholder farms can play a significant role in biodiversity conservation strategies.

What are the characteristics of the smallholder farming systems that favor this tree and shrub diversity? One of the practices that favor the existence of wild diversity in the farms is the maintenance of the original native vegetation or allowing the establishment of wild plants that arrive by natural dispersal. The motivations of owners who allow the establishment of these wild species may be related to their perception of the environment. The tendency for organic practices on these farms implies that the owners value the services that wild species can offer, such as shade in pastures, a supply of food for birds and insects or the protection of hatcheries or water sources. The establishment of wild species is influenced by the surrounding flora, but their maintenance on farms depends on the actions or practices of the owners (^{Rooduijn et al. 2018}).

Another aspect that favors tree diversity on farms is the presence of various spaces within the farms, such as home gardens, living fences or forest relics. This complex structure is characteristic of smallholder farming systems and has been described in various family production systems around the world (^{Kehlenbeck et al. 2007}, ^{Pulido et al. 2008}, ^{Abebe et al. 2013}, ^{Mohri et al. 2013}). In the studied farms, most of the tree species diversity was found in productive areas, which included home gardens and cultivated or grazing areas. The fact that these are precisely the areas of greatest human intervention indicates that the presence of trees and shrubs depends on the decisions of the owners. This confirms the statement by ^{Thijs et al. (2015)} who recognize that the occurrence of tree species in agricultural systems is highly influenced by human interventions. In Panama, for example, farmers perform various tree management practices, assigning value to native species (^{Garen et al. 2011}). Living fences also turn out to be another space that protects a variety of trees and shrubs, since these are important spaces for the establishment of wild plants along with the species planted by the owners. The role of living fences as areas for the establishment and conservation of tree species has already been documented for the Colombian Andes (^{Pulido-Santacruz and Renjifo 2011}). Spaces with remaining native vegetation, such as forest relics, also contribute significantly to maintaining tree diversity on these farms.

Uses

Our results showed that farmers maintain species that provide them with some service. The uses reported for tree species on these farms, in general, coincided with those reported in other studies in smallholder farming systems (^{Pulido et al. 2008}, ^{Garen et al. 2011}, ^{Abebe et al. 2013}, ^{Thijs et al. 2015}, ^{Pinto Rayol et al. 2017}). The predominance of species planted to form living fences or due to their ornamental value may be related to the proximity of farms to urban areas. This trend has been documented in studies of home gardens (^{Pulido et al. 2008}, ^{Caballero-Serrano 2016}), which highlight that the distance to urban centers or markets influences the composition of the flora, increasing the richness of ornamental species that are usually exotic. Despite this trend, it is important to highlight that the trees of these farms not only represented an aesthetic or recreational value, but they also had a productive subsistence purpose, which is shown by the 29 species that had use as food, the majority producing fruits. In general, the management of various arboreal species for food purposes is a characteristic of small farming systems (^{Pulido et al. 2008}). Virtually all edible species are cultivated by the owners, and the tree-species documented here are used as food throughout the Colombian high Andes (^{Pérez and Matiz-Guerra 2017}). This could indicate an owner's interest in maintaining a food source for the local market and for family consumption, which is a distinctive feature of smallholder farming (^{Thijs et al. 2015}, ^{Ortiz et al. 2018}).

The medicinal use of tree species was less frequent. It may be related to the fact that only the farm's owner was interviewed, which could imply a bias since people familiar with medicinal plants could not have participated in the study. Another aspect that is relevant is the sporadic use of tree species as timber for construction or the lack of reports of their use as fuels, which corresponds to use categories frequently reported in the literature for woody plants (^{Garen et al. 2011}). This may be related to the environmental and socio-economic conditions of this region since households have other combustible sources such as gas or coal; and, in general, their dependence on wood and firewood has decreased.

It is worth noting that at least one-third of the species recorded are conserved in these farms despite not having an obvious use. Within this group are wild species that are dispersed in productive areas or are part of living fences or forest relics. Although the owners do not assign them a particular use, they may be recognized as having some environmental or aesthetic value by the owners. It is precisely in smallholder farming systems where such cultural values, which go beyond commercial production, can influence the diversity of species (^{Ortiz et al. 2018}).

Factors that influence species richness

We found that only size of the farms and proximity to areas with natural vegetation significantly influenced species richness. These results concur with other papers on smallholder farming systems (^{Caballero-Serrano et al. 2016}, ^{Pinto Rayol et al. 2017}, ^{Rooduijn et al. 2018}). To the extent that the farms are larger, they can hold more tree species (^{Pinto Rayol et al. 2017}). Likewise, to the extent that the remaining vegetation surrounds the farms, there is a greater chance of dispersal of wild species (^{Rooduijn et al. 2018}). It seems that forest fragmentation can increase the establishment of wild species in this type of agricultural system due to a clustering effect (^{Rooduijn et al. 2018}). In this sense, the remaining forests act as a source of propagules that end up being established in neighboring agroecosystems (^{Pinto Rayol et al. 2017}). In this panorama, it is relevant to explore the role of smallholder farms as areas of connectivity for the biodiversity of the Colombian Andean region and to recognize the importance of spaces with human intervention for biodiversity conservation strategies (^{Thijs et al. 2015}).