Profiling native and introduced perennial garden plants in Puerto Rican urban residential yards

Abstract

Worldwide the number of non-native species escaping from cultivation into native habitats is steadily increasing with no signs of saturation. Species that eventually become invasive may generate unwanted social and ecological conditions especially in areas of conservation concern. This study built upon prior biodiversity work from 432 residential yards in the San Juan Metropolitan Area of Puerto Rico to evaluate the natural history and functional traits of native and non-native plant species in these green spaces. We reviewed the literature for a total of 361 plant species to extract information on their taxonomy, native distribution range, invasive status (casual, naturalized or invasive), life-form and ecological and biological species attributes. We then evaluated the relationship between their attributes and their probability of escaping cultivation and become invasive. Our results show that non-native species growing in yards are more likely to succeed in becoming invasive if they have vegetative growth, a mixed breeding system, and an unspecialized dispersal mode. We also found that native and non-native species occurring in residential yards share similar adaptive strategy scores. Most plant species that have already become invasive originated from Asia and America a fact that is likely tied to the US nursery trade. We used the combined results of this and prior studies to understand the factors facilitating plant invasion and to generate recommendations for the development of management strategies that may limit the spread of non-native ornamentals with the potential to escape cultivation and become invasive on this island.

Introduction

Modern conservation paradigms consider urban areas as potentially useful spaces for species conservation, the protection of local biodiversity and the provision of ecosystem services in cities (Herzog 2016; Ziter 2016; Locke and McPhearson 2018). Yet, despite showing high levels of biodiversity, many studies evaluating species composition in cities have shown that most urban floras are dominated by non-native species and even more so residential yards and gardens (Vila-Ruiz et al. 2014; Yang et al. 2019). From a conservation perspective, biological invasions are considered one of the most important global biodiversity threats and one of the major drivers of native population declines and extinctions (Pyśek and Richardson 2010; Rojas-Sandoval et al. 2016; Vilà and Hulme 2017). On the other hand, some non-native species, even those that have become invasive, may provide important ecosystem services that may create management conflicts (Lugo 2004; Dickie et al. 2014, Novoa et al. 2018). Therefore, conservation efforts in urban green spaces should attempt to develop tools to evaluate which species have the potential to become invasive and provide management recommendations originated on species-based information.

Horticulture is recognized as the most important introduction pathway of invasive plants worldwide (Dehnen‐Schmutz et al. 2007a; Hulme et al. 2018; van Kleunen et al. 2018) and residential yards and gardens are often characterized by the cultivation of non-native ornamentals (Dehnen‐Schmutz et al. 2007b; Mayer et al. 2017). Indeed, a recent study has suggested that the naturalization success of introduced plant species is positively correlated with their cultivation in domestic yards and that cultivation of non-native species in those spaces may be more important to naturalization success than cultivation in botanical gardens (Guo et al. 2019). Residential yards can occupy considerable portions of the green infrastructure of cities and for that reason have been considered as vital places for species conservation in urban spaces (González-García and Sal 2008; Loram et al. 2011; Meléndez-Ackerman et al. 2014). In the USA, about 50% of all invasive plants and 82% of invasive woody species were introduced for ornamental and landscaping purposes (Reichard 1997; Reichard and White 2001; Mack and Erneberg 2002; Li et al. 2004) and many of the species listed as invasive are still sold in nurseries nationwide (Hall 2000; Burt et al. 2007). New invasive plant species are discovered virtually every year. The repeated sale and widespread introduction (i.e. propagule pressure) of invasive ornamental plants has increased the chance of colonization and naturalization of these plants into natural areas (Enserink 1999). Invasive plants used in horticulture play an important role in changing the look, feel and function of both natural areas and urban green spaces (Li et al. 2004). Therefore, understanding which species occupy these spaces and to the extent by which they could become invasive could not only help us to support plant conservation in general but also facilitate the integration of these spaces into conservation efforts.

Determining the factors associated with naturalization and invasion success is a fundamental goal in invasion ecology (Lockwood et al. 2005; Pyšek and Richardson 2008; Richardson and Pyšek 2012). While predicting what species will invade, when and where may be complex due to the context dependency of species invasions, accumulated research points to the advantage of evaluating plant functional traits and natural histories to understand how non-native species become naturalized and invasive (van Kleunen et al. 2010, 2015; Guo et al. 2018). For example, traits such as rapid growth, early flowering and wide native ranges are often associated with the potential invasiveness of non-native species often cultivated as ornamentals (Haeuser et al. 2019; van Kleunen et al. 2018). Indeed, a previous study assessing the alien flora of Puerto Rico showed that ornamentals escaped from cultivation were the most important pathway of introduction of invasive species on this island, accounting to >50% of all species introduced (Rojas-Sandoval and Acevedo-Rodríguez 2015). In that regard, evaluating functional differences between native and non-native species may be a helpful tool to identify which introduced non-native species should be prioritized for their management and as a way to provide information to local residents on the biological and environmental consequences of the species that they manage.

One approach to evaluate the contribution of plant functional traits explaining naturalization and invasion success is the ordination tool of Pierce et al. (2017) based on Grime’s CSR categories (Grime 1979, 2001; Grime and Pierce 2012). In Grimes’ scheme, three functional categories are defined: (i) competitors -C- species that exploit conditions of low stress and low disturbance, (ii) stress tolerators -S- adapted to conditions of high stress and low disturbance and (iii) and ruderals -R- species adapted to conditions of low stress and high disturbance. In addition, intermediate categories (e.g. CS, SR, CR) are also recognized. Using this classification, Pierce et al. (2017) developed ‘StrateFy’ an ordination tool that calculates continuous quantitative CSR-scores based on the contributions of three leaf traits that represent extremes of plant functional specialization: specific leaf area, leaf area and leaf dry matter content. Recent studies using this approach have effectively demonstrated the global applicability of this tool at explaining the naturalization success of plant species. These studies have revealed that introduced plants, across different life-forms, with relatively high R- and C-scores were positively associated with the probability of becoming naturalized, while high S-scores were instead negatively associated with those events (Guo et al. 2018, 2019).

This study builds upon prior biodiversity work in the San Juan Metropolitan Area of Puerto Rico focused on evaluating vegetation resources on residential yards and gardens of this city with emphasis on perennial plants (García-Montiel et al. 2014; Meléndez-Ackerman et al. 2014; Vila-Ruiz et al. 2014). Previous studies have shown that residential yards of San Juan are biodiversity-rich areas but with a flora dominated by non-native ornamental shrubs and food plants (e.g. 69% non-native species; Meléndez-Ackerman et al. 2014; Vila-Ruiz et al. 2014). In this study, our main goal was to evaluate the natural history and functional traits of native and non-native species occurring in San Juan residential yards. We expected tropical regions and tropical plant families to dominate the taxonomic and geographic origin of the non-native species that have become naturalized and invasive on this island following the premise that a better environmental matching between the non-native species and the invaded environment would increase invasion success (Richardson and Pyšek 2012). We also expected non-native species that are not behaving as invasive to exhibit different functional traits relative to those that have already become invasive (e.g. difference in breeding system, dispersal mode and CSR-scores). Finally, we used our combined results and data on planting intensity (as a proxy of propagule pressure) to generate informed recommendations for the management of non-native plant species with the potential to escape cultivation and become naturalized and potentially invasive on this island.

Methods

Study site

Data were collected in yards from single-family housing units placed at six permanent study plots in San Juan, Puerto Rico. These plots located along the urban cover gradient of the Río Piedras watershed were established by the San Juan ULTRA Network (Urban Long-Term Research Areas; Fig. 1; Vila-Ruiz et al. 2014; Meléndez-Ackerman et al. 2014). The Río Piedras watershed covers an area of 49 km² and is the most urbanized watershed of the Caribbean island of Puerto Rico. It includes the municipalities of San Juan, Guaynabo and Trujillo Alto (Lugo et al. 2011) and lies in the subtropical moist forest zone (Ewel and Whitmore 1973) as described by the life zone system of Holdridge (1967). The Río Piedras watershed has a land cover gradient that ranges from highly urbanized near the coast (lower watershed region) to highly forested around the headwaters (higher watershed region) with mean annual temperatures of 25.7°C and a mean annual rainfall that ranges from 1509 mm on the coast to 1755 mm in the upland (Lugo et al. 2011). More than 50% of the area is covered by volcaniclastic rocks with the remaining geological formation cover including urban fill, mangroves and swamps, alluvium sediments and limestone (Webb and Gómez-Gómez 1998). Within the Río Piedras watershed, residential areas support a biodiversity-rich flora dominated by non-native species (69% corresponded to introduced non-native plant species; Vila-Ruiz et al 2014) that is mostly driven by a combination of social factors at the household scale (Meléndez-Ackerman et al. 2014), historical aspects and plant nursery trade (Torres-Camacho et al. 2017) as well as by a combination of socio-ecological factors external to the watershed system (Meléndez-Ackerman et al. 2016).

Species dataset

Vegetation surveys were conducted at each of the six 2-km diameter circular plots following a stratified sampling scheme with the initial selection of locations based on a general representation of the rural–urban gradient of gray area coverage (Fig. 1). At each plot, streets were randomly selected, and vegetation surveys were conducted in both front- and backyards of 432 households willing to participate in the study. On average, households had a combined yard area of about 400 m², accounting for a total area surveyed of 187,191 m² (Meléndez-Ackerman et al. 2014). Vegetation surveys were conducted by students and volunteers and were focused on surveying woody species (e.g. trees, shrubs, and palms) and large perennial herbs (height between 50 cm and 2 m). Nevertheless, the occurrence of other perennial plant species was also recorded whenever possible. The final dataset comprises a total of 361 spermatophyte species after records were checked for synonyms, unreliable records and hybrids. Species were classified as native or non-natives to the Puerto Rican flora following Axelrod (2011) and Acevedo-Rodríguez and Strong (2012). Species entries were supplemented with information about their taxonomic position (family), native distribution range, invasive status (casual, naturalized or invasive), life-form, and ecological and biological species attributes. The complete list of the parameters evaluated for each species and their definitions is presented in Table 1. If the native distribution range of a species covered more than one continent, it was assigned to each of them. To facilitate comparison with previous studies, we followed the parameters and concepts defined by Rojas-Sandoval and Acevedo-Rodríguez (2015) and Rojas-Sandoval et al. (2017).

Statistical analyses

We used descriptive statistics and contingency analyses to characterize the profile of the native and non-native species growing in the surveyed yards. For each parameter, differences in the observed and expected frequencies of native and non-native species were examined using χ² tests. These analyses were performed only for those parameters with more than five species (n > 5) in each category to fulfill the tests’ assumptions. Generalized linear models (GLM) were used to determine the parameters influencing the likelihood that introduced garden plants would escape from cultivation and become invasive on this island (Rojas-Sandoval and Acevedo-Rodríguez 2015). For these analyses, we used binomial GLM with logit link and logistic regressions with the origin, life history and reproductive parameters described in Table 1 as independent variables. Species were scored as ‘1’ if they were listed as invasive (hereafter ‘non-native/invasive’) or as ‘0’ if they were non-natives but not listed as invasive yet (hereafter ‘non-native/non-invasive’). To fulfill logistic regression assumptions, all parameters used in the analyses were transformed and recorded as dichotomous or continuous variables (Table 2). The best-fit models were selected using the Akaike’s information criterion (Menard 2002; Milbau and Stout 2008). Once models were obtained, we checked for the presence of significant interactions between explanatory variables. The odds ratios were calculated for significant variables as the probability for non-native species to become invasive divided by the probability of failure (Milbau and Stout 2008; Rojas-Sandoval and Acevedo-Rodríguez 2015). All these GLM analyses were carried out using Matlab version R2014a (MathWorks, Inc., Natick, MA, USA). We also calculated the Grime’s CSR-adaptive strategy scores (Grime 2001; Grime and Pierce 2012) for native and non-native species, using the CSR calculator tool ‘StrateFy’ (Pierce et al. 2017) and data on specific leaf area, leaf area and leaf dry matter content obtained from the TRY database (Kattge et al. 2020; www.try-db.org). Within the Grime’s scheme species are classified into three adaptive strategies: competitors (C), stress-tolerators (S) and ruderals (R) (Grime 2001; Grime and Pierce 2012; Pierce et al. 2017). To examine whether the C-, S- and R-scores differed between native and non-native species, we used one-way ANOVA treating each axis (C-, S- and R-) as an independent variable. We then used the R package ‘ggtern’ (Hamilton 2015) to visualize the triangular plot of the C-, S- and R-scores of the species. All these analyses were performed in R (R Development Core Team 2018).

Results

Among the 361 species in our dataset, 102 species (28%) are classified as natives and 259 species (72%) are non-native species that were introduced to Puerto Rico. Regardless the origin, Fabaceae, Rubiaceae and Euphorbiaceae were the most diverse families in the surveys (Fig. 2). Native species were proportionally more common in the Fabaceae, Rubiaceae and Malvaceae while the families with the largest number of non-native species were Fabaceae, Euphorbiacea, Solanaceae and Lamiaceae (Fig. 2). A higher proportion of non-native species that have become invasive is observed in the Zingiberaceae (4 out of 7 non-native species = 57%), Araceae (4 out of 12 non-native species = 33%) and Fabaceae (8 out of 25 non-native species = 32%), whereas no invasive species has been detected in the Solanaceae, Heliconiaceae, Rubiaceae and Malvaceae (Fig. 2). Non-native species growing in the surveyed yards can be sourced to all continents, but a large fraction are species coming from Asia (32%) and continental America (31%) and less common are species coming from Europe (Fig. 3). Among the 259 non-native species growing as ornamentals in the surveyed yards, a total of 50 species have already been listed as invasive for Puerto Rico (Table 3). These 50 invasive species can be sourced to all continents except for Europe (Fig. 3) and included 34 woody species and 16 herbaceous species dominated by trees originating from Asia and the Australia/Pacific region and herbs originating from Asia (Supplementary Fig. S1). Indeed, when all non-native species originated from tropical regions were considered, our data showed that a larger proportion of non-natives that have become invasive came from Asia and the Australia/Pacific region (Fig. 3).

The reproductive strategies as well as the pollination and dispersion types related to native and non-native species are very diverse (Fig. 4). We found that the majority of the non-native species have a mixed breeding system regardless of their invasion status (non-native/non-invasive: 52% out of 206 species, non-native/invasive: 52% out of 50 species, native: 38% out of 102 species, Fig. 4a). In contrast, for natives a larger percentage of species show outcrossing (52%) compared with non-natives regardless of their invasion status (non-invasive: 35%, invasive: 44%, Fig. 4a). We also found that the number of native and non-native species that rely almost exclusively on selfing strategies for reproduction (i.e. apomixis and autogamy) is minimal (1 native and 2 non-native species) with most species reported as relying on animal pollination (Fig 4b). For the non-native species that have become invasive, the dispersion of propagules is predominantly unspecialized, while for native species the dispersion is mostly animal-depended (Fig. 4c). Our data showed that for species that rely on animal dispersion, significantly more native species (∼60%) have large ‘fleshy/drupes’ fruits with large seeds (>5 mm; χ²=4.06; P = 0.03), which need ‘specialized’ animal dispersion, while non-native species tended to have smaller seeds (∼70% have seeds <5 mm). We also detected that a large number of the non-natives (81 non-invasive and 23 invasive) are species with the potential to spread vegetatively by rhizomes, stem fragments, and/or stolon (Fig. 4c). Overall, the frequency distribution of dispersal mechanisms of propagules was significantly different among species groups with a higher than expected frequencies of dispersion by animals among native species and lower than expected frequencies among non-natives that are invasive and higher than expected frequencies of vegetative reproduction in non-natives regardless of their invasion status (χ² = 37.5; P < 0.05).

The GLM best-fit model explaining the probability that non-native plant species growing in yards may escape cultivation and become invaders included three parameters (Table 4) and it was significantly different from the intercept-only model (Likelihood ratio χ² = 9.16; P = 0.01; Table 4). The best-fit model shows that non-native species growing in yards are more likely to become to escape cultivation and become invasive if they have vegetative growth, a mixed breeding system, and an unspecialized dispersal mode. These parameters are arranged in order of importance according to the odds ratios (Table 4).

The C-, S- and R-adaptive strategy scores were estimated for 105 species (38 natives, 48 non-native/non-invasive and 19 non-native/invasive; Fig. 5; Supplementary Table S1). The combined results for these species showed that C-, S- and R-scores were all negatively correlated to each other (R vs S: −0.45; R vs C: −0.24; S vs C: −0.75; all P < 0.01). For natives, non-native/non-invasive, and non-native/invasive, the C-, S- and R-scores were similar and were all clustered around the CS-adaptive strategy with low R-scores (Supplementary Table S2; Fig. 6). Differences in the scores were detected when we accounted the life-form of the species. Trees showed similar C- and S-scores and were clustered around a CS median strategy, with low R-scores (46:43:11%), while shrubs (35:48:17%) and herbaceous species (34:45:21%) showed variation around the CS/CRS strategy (Fig. 5b, c; Supplementary Table S2). When individual C-, S- and R-scores were considered, there were no significant differences among native, non-native/non-invasive and non-native/invasive species (F < 0.56; P > 0.5 in all cases; Fig. 6).

Discussion

Plant taxonomy

We found that, as expected, a higher proportion of the non-native species that turned invasive belonged to tropical plant families such as Zingiberaceae and Araceae (both pantropical monocots) and not surprisingly to the Fabaceae family. Many species in the Zingiberaceae and Araceae are cultivated for their showy flowers and foliage and have become economically important in the horticultural trade (Bown 2000; Prince and Kress 2002). Additionally, species within these two families are typical elements of moist tropical and subtropical forests and have the potential to reproduce vegetatively through rhizomes and tubers, a condition that has been a preferred trait for commercial cultivation (Simão and Scatena 2003; Chomicki 2013). Zingiberaceae and Araceae are also species able to exploit canopy gaps, disturbed sites, and marginal environments often becoming dominant fixtures in the understory of tropical forests (Bown 2000; Prince and Kress 2002). In the case of Fabaceae (legumes), various factors may account for the apparent overabundance of invasive within this family. One possibility is that this result simply reflects a numeric response resulting from the fact that Fabaceae is one of the most diverse plant families across Pantropical forests and the most diverse in the Neotropics (Cardoso et al. 2017; Sosef et al. 2017). However, other two very diverse families also occurring within San Juan urban yards, the Rubiaceae and Malvaceae, did not contain any non-native species behaving as invasive. On the other hand, even when many legumes are capable of nitrogen fixation through symbiotic relationships with nitrogen-fixing bacteria (Corby 1988), a condition that would allow them to invade poor soils, a recent study showed that at a global scale, the presence of this symbiotic interaction can be used to predict invasion success and that being a nitrogen fixer (and thus needing a symbiotic relationship) may limit the capacity of introduced legumes to expand outside their current distribution ranges (Simonsen et al. 2017). Ultimately, the factors leading to a higher proportion of invasive species within Fabaceae may be more complex and involve ecological and socioeconomic factors (see below) that may be unique to each species and should be evaluated in future studies.

Reproductive and dispersal traits

The suite of reproductive traits exhibited by non-native plant species in our surveys is consistent with attributes that would give species a colonization advantage: mixed breeding systems, small seed size and vegetative propagation (e.g. rhizomes, stem fragments, and/or stolon). Mixed breeding systems allow plant species to ensure reproduction through self-compatibility mechanisms, while small seed size provide species with higher dispersal potential (Lake and Leishman 2004; Pyšek and Richardson 2008). Most natives in our yards are reported as outcrossing species and their seeds are animal dispersed and larger than those of non-native species. We also found that a large number of the non-native species are using vegetative growth (e.g. layering, suckering, root resprouting, runners or rhizomes) as methods of natural spread or persistence. Certainly, our combined results suggest that the ability to spread vegetatively is a prevailing trait of non-native species that have become invasive. This result is in agreement with recent studies showing that vegetative propagation is a trait directly linked to invasion success (Cadotte et al. 2006; Klinerová et al. 2018; Nunez-Mir et al. 2019). In addition, regardless of the invasive status, the dispersal mechanism of non-native species in our yards was rather unspecialized and included broad spectrum of abiotic dispersal mechanism (e.g. wind, water, and gravity). All these conditions may provide non-native species in yards with more opportunities to escape cultivation and a higher ability relative to native species to contribute propagules to adjacent green spaces.

CSR-adaptive strategy scores

Plant species growing in the surveyed yards occupied all CSR-adaptive strategies, but more species were represented in the C- (competitor) and S- (stress-tolerant) strategies than in R- (ruderal) strategies. We also found that relative scores for our yard plants were consistent with those reported for tropical and subtropical areas (Pierce et al. 2017). Contrary to our expectation, we found that non-natives (invasive or not) and native species are behaving very similar in terms of their distribution of C-, S- and R-strategies. This finding suggests that both native and non-native species have similar requirements and may occupy similar niches. The distribution of CSR-adaptive strategies in our yards could be explained in part by the survey design that was used in this study and that was focused on woody species rather than herbs. Pierce et al (2017) suggested that different growth habit categories exhibited variation in the CSR-scores. They showed that trees are often clustered around a CS median strategy, with no R-selected trees apparent, quite similar to our results (Pierce et al. 2017; C:S:R = 43:47:10%; this study: C:S:R = 46:43:11%). Additionally, a recent study by Guo et al. (2019) evaluating CSR-adaptive strategies in both botanical and home gardens found that while C-strategy species dominated at botanical gardens, plant species with R-strategy were more common in home gardens due to the higher frequency of herb and shrubs species on these spaces. These authors also showed that cultivation in domestic gardens is one of the most important factors explaining the naturalization success of non-native species (Guo et al. 2019).

Species origin, composition and history of introduction

The origin and composition of the plant species found within San Juan residential yards may reflect the historical, social and economic changes that have been occurring in Puerto Rico during the last century, in which this island changed from a largely agricultural system to a more urbanized economy with emphasis on manufacturing (Thomlinson et al. 1996; Grau et al. 2003). These economic changes also promoted migration from rural to urban areas and the subsequent expansion of residential urban areas that by 1994 had replaced 6% of the island’s prime agricultural lands (López et al. 2001). The course of these changes also reflects ‘waves of interest’ in non-native plant species useful for agriculture that increased in the earliest 1900s but declined after 1940s, when agriculture was abandoned and industry and tourism were promoted as the main economic activities in Puerto Rico. More recent interests in species with horticulture and ornamental value (starting in the end of the 19th century) have been primarily associated with the expansion of urban residential areas (Rojas-Sandoval and Acevedo-Rodríguez 2015; see below). The combination of these ‘waves of interest’ may explain the origin of plant taxa occurring in the surveyed San Juan residential yards.

A subset of the species occurring in the surveyed yards has agricultural and agroforestry value and may have been initially introduced for that purposes. Since its establishment in 1901, the USDA-supported Tropical Agriculture Research Station (TARS) has been developing plant collections for agricultural and horticultural research focusing in tropical species (Goenaga 2002; USDA-ARS 2019). Between 1920s and until 1960s, TARS was involved in the establishment of experimental crops, tree nurseries and the development of large-scale research programs of trial plantings with native and exotic tree species (most of them native to tropical America and tropical Asia; Birdsey and Weaver 1982). In 1934, a major reforestation program in public forests was initiated, and over the next 12 years, about 7800 hectares across the island were planted with 53 tree species, 25 of which were non-native species. In 1939, the USDA-Tropical Forest Experiment Station (now the International Institute of Tropical Forestry IITF) was established and a site adaptability program was started in which more than 350 introduced plant species were tested (Geary and Briscoe 1972; Birdsey and Weaver 1982; Francis and Liogier 1991). Historically TARS and IITF have been worked on experimental projects focused in non-native species valuable for coffee shade, fruit, timber, and other utilitarian purposes (Birdsey and Weaver 1982; Francis and Liogier 1991; Miller and Lugo 2009), some of which were purposely introduced in agricultural areas and forest plantations across the island and later became widely naturalized (Lugo 2004) and now appear in our yard surveys (e.g. from Africa: Spathodea campanulata; from Asia: Citrus sinensis, Mangifera indica and Syzigium jambos; from Asia/Pacific: Erythrina variegata; from Australia/Pacific: Eucalyptus robusta). Another possible explanation is that with the expansion of urban areas into former agricultural lands, it is highly probable that some of the non-native trees and shrubs previously introduced into agricultural lands were kept as ornamental elements and become part of the new urban gardens and yards (Vila-Ruiz et al. 2014).

About two-thirds of the ornamentals cultivated in San Juan residential yards are non-native species (72% current study and 69% according to Vila-Ruiz et al. 2014). A likely introduction pathway for these species may be through the global nursery trade and how it interacts with the USA and its territories. Bradley et al. (2012) described current and emergent nursery trade markets and showed that established sources of nursery plants in the USA are predominantly from Asia and Tropical America (Central and South America with emphasis on Brazil). This pattern of plant trade may therefore explain the origin of the non-native ornamentals growing in our surveyed yards for which we also found that, as expected, is dominated by species originated from tropical America and Asia. An important observation from Bradley et al. (2012) is that climate change and water scarcity may be creating emergent markets for the plant nursery trade where more species are being imported from semiarid regions of Africa and the Middle East. Given the political and economic ties between the USA and Puerto Rico, these market changes may also create different future pathways for the introduction of ornamental garden species and new threats for the island that would need to be explored.

The ‘naturalization-invasion continuum’ model proposes that there are multiple environmental and physical barriers that introduced species most overcome to eventually become naturalized and invasive (Richardson et al. 2000). The combined results of this and prior studies suggest that the social-ecological system that influence vegetation composition in San Juan residential yards may also help overcome the biotic and abiotic barriers that prevent introduced non-native species to become naturalized and invasive. Our findings show that reproductive barriers in the process of ‘plant establishment’ could be reduced through the reproductive and dispersal advantages of non-native plants as well as by the life history affinities of non-natives with native species. A previous study that evaluated the social-ecological mechanisms influencing the occurrence of plant species in San Juan residential yards pointed-out four main drivers directly related to resident gardening practices (Torres-Camacho et al. 2017). First, residents often purchase plants at retail stores and local nurseries. Second, residents exchange plants as gifts. Third, residents maintain plants that were already established when they became owners and last, they allow some plants that colonized yards through natural dispersal to establish and grow. Here, the nursery trade helps overcome both geographic and dispersal filters within the urban matrix. In addition, the median time of house residence in San Juan neighborhoods is 40 years, but for some residents is up to 90 years (Torres-Camacho et al. 2017), so urban yards may be excellent long-term refugia for non-native species. For Puerto Rico, we know that invasive plant species have resident times ranging of about 90–140 years (Rojas-Sandoval and Acevedo-Rodríguez 2015) and thus yards may buy time to non-native species by providing temporal safe heavens. Overall, the social-ecological system of residential urban yards in San Juan, may render gardens not only as a reservoir of potential invasive plants but also as areas that may allow for the movement of yard species through the urban matrix within the residential areas.

Functional traits to profile potential invaders

While performing a through risk assessment evaluation was beyond the scope of this study, here we used our combined results and data available for planting intensity (evaluated as the number of yards in this study carrying each species) as a proxy of propagule pressure to perform a preliminary assessment of the non-native species occurring in San Juan yards that have not been listed as invasive yet (Supplementary Table S3). Our results showed that species classified as non-native/non-invasive with the three main traits identified by the GLM analysis as the most important facilitating invasion success (vegetative growth, mixed breeding system and an unspecialized dispersal mode) occurred in less than 4% of the surveyed yards. Examples of these species are Plectranthus amboinicu, Euphorbia milii, Manihot esculenta, Ravenala madagascariensis and Aphelandra squarrosa (Supplementary Table S4). We also found that for the five most frequent non-natives/non-invasive species, only three have these three main traits that would facilitate invasion success (Hibiscus rosa-sinensis, Annona muricata and Bougainvillea glabra; Supplementary Table S3). Of these three species, Hibiscus rosa-sinensis is the species of major concern as it is a known host of many plant pests (CABI 2019). We also detected that Ixora coccinea, the most common non-native/non-invasive species in San Juan yards (occurring in 40% of the households), did not hold as many traits that favor invasion success (e.g. only shows vegetative growth). This may explain, at least in part, why this species has not yet escaped cultivation locally. On the other hand, it should be noted that many of these ornamentals might be recent introductions that have not had enough residence time to overcome local environmental filters (Pyšek et al. 2009; Davis et al. 2016). In the case of Puerto Rico, data on the date of introduction exist for a minimal fraction of the introduced non-native species; however, we know that a major wave of plant introductions occurred between 1925 and 1970s and new species are introduced every year (Birdsey and Weaver 1982; Goenaga 2002; Rojas-Sandoval and Acevedo-Rodríguez 2015).

General recommendations

At local and global scales, most plant species that have become invasive were originally introduced for horticultural purposes (Rojas-Sandoval and Acevedo-Rodríguez 2015; Hulme et al. 2018; van Kleunen et al. 2018). Thus, horticultural trade is the one of the most important players in the prevention of future invasions. For Puerto Rico we know that non-native species dominate the available stock in retail stores and local nurseries (Torres-Camacho et al. 2017). Therefore, if we want to integrate residential gardens within landscape-level conservation initiatives it is then crucial to involve the private sector related to the nursery trade in order to create a market more focused on native ornamentals. In the meantime, however, it would be advisable for landscaping projects and gardeners to use ornamentals with the lowest invasion risk. Based on our combined results, we recommend planting species with traits such as lower dispersal capability, non-vegetative propagation and obligate outcrossing mechanism as a way to reduce the possibilities that new species may eventually escape and become invasive. In this regard, using our results of the GLM and the CSR-adaptive strategy scores, we generated a list of ornamental non-native species already found within San Juan yards but with relatively low invasiveness risk (Supplementary Table S4). Nonetheless, these recommendations should be complemented with risk assessment protocols and monitoring of species as suggested by literature (Pheloung 2001; Daehler et al. 2004; Dehnen‐Schmutz et al 2011; Faulkner 2014).

On the other hand, the fact that non-native species are exhibiting a close affinity to native species in term of their adaptive strategies presents a challenge for the control and management of non-native and invasive species on this island and calls for the inclusion of parallel strategies to promote and facilitate the use of native ornamentals. Another strategy that could be used is the implementation of environmental policies that regulate the introduction and commercialization of non-native ornamentals. Conservation-oriented cities and jurisdictions should also employ education and outreach campaigns to raise awareness about the potential threats of invasive species and to promote native plantings (Burt et al. 2007; Hulme et al. 2018). Finally, incentives to the private horticultural sector to increase the number of native species in their local inventory, with special emphasis in ornamentals and food species that are the most common elements in San Juan residential yards (Vila-Ruiz et al. 2014) should also be promoted.

Conclusions

San Juan urban residential yards are biodiversity-rich areas, but its flora is dominated by non-native species. The combined results of this and other studies suggest that the ecological and biological traits exhibited by non-native ornamental species that are currently cultivated in these urban yards may facilitate their transition from cultivation to invasion if unchecked. Limiting the introduction-naturalization-invasion process would require incorporating science-based strategies that take advantage of our ecological knowledge of ornamental plants but also strategies and policies directed to regulate introduction pathways and the major factors influencing plant management practices on this city.

Acknowledgements

Data for this study were generated by the San Juan ULTRA collaborative project (sanjuanultra.org) that addressed the vulnerabilities of the Río Piedras Watershed social-ecological system. San Juan ULTRA is supported by NSF-ULTRA (DBI-0948507), NSF-REU (DBI-1062769) and the Center for Applied Tropical Ecology and Conservation (CATEC, NSF-CREST: HRD-0206200) of the University of Puerto Rico (UPR). Data collection could not have been possible without the help of many students and volunteers from the Department of Environmental Sciences and the Graduate School of Planning of the UPR system who helped in every stage of the research and the residents of San Juan and Guaynabo who made field data collection possible.

Supplementary data are available at JUECOL online.

Conflict of interest statement. None declared.

Data Availability

Data used in this study for analyses and additional tables and figures can be found in the Supplementary Materials.

References