Seasonality and forest edge as drivers of Tradescantia zebrina Hort . ex Bosse invasion in the Atlantic Forest

Invasive alien species are one of the major causes of species loss (Newbold et al., 2015), threatening biodiversity conservation (Wittenberg and Cock, 2001) and the provision of ecosystem services (Walsh et al., 2016). In natural ecosystems, the impacts of biological invasions are complex and can permanently alter the structure and function of communities (Panetta and Gooden, 2017), as well as cause local extinctions and homogenize the invaded environment (Sakai et al., 2001). Environmental degradation has caused biodiversity loss, and has shifted the dynamics of ecosystems, thus promoting the establishment and spread of invasive alien species (Valéry et al., 2008; Zenni, 2010). Nevertheless, alien plants may affect ecosystem processes by changing nutrient cycling, trophic interactions, community structure, and species functional attributes (Dogra et al., 2010), but also by affecting physical properties of ecosystems such as erosion rates, sedimentation, and the water cycle (Wittenberg and Cock, 2001; Ziller, 2001; GISP, 2005; Currie et al., 2014). Biodiversity loss and landscape changes caused by invasive alien species may be particularly severe in protected areas (FICMNEW, 1998). No Brazilian protected area is known to be free of alien species, although such areas possibly exist (Leão et al., 2011). Ziller and Dechoum (2013) reported Abstract As a result of biodiversity and ecosystem service losses associated with biological invasions, there has been growing interest in basic and applied research on invasive species aiming to improve management strategies. Tradescantia zebrina is a herbaceous species increasingly reported as invasive in the understory of disturbed forest ecosystems. In this study, we assess the effect of spatial and seasonal variation on biological attributes of this species in the Atlantic Forest. To this end, we measured attributes of T. zebrina associated with plant growth and stress in the four seasons at the forest edge and in the forest interior of invaded sites in the Iguaçu National Park, Southern Brazil. The invasive plant had higher growth at the forest edge than in the forest interior and lower leaf asymmetry and herbivory in the winter than in the summer. Our findings suggest that the forest edge environment favours the growth of T. zebrina. This invasive species is highly competitive in the understory of semi-deciduous seasonal forests all over the year. Our study contributes to the management of T. zebrina by showing that the summer is the best season for controlling this species.


Introduction
Invasive alien species are one of the major causes of species loss (Newbold et al., 2015), threatening biodiversity conservation (Wittenberg and Cock, 2001) and the provision of ecosystem services (Walsh et al., 2016). In natural ecosystems, the impacts of biological invasions are complex and can permanently alter the structure and function of communities (Panetta and Gooden, 2017), as well as cause local extinctions and homogenize the invaded environment (Sakai et al., 2001).
Environmental degradation has caused biodiversity loss, and has shifted the dynamics of ecosystems, thus promoting the establishment and spread of invasive alien species (Valéry et al., 2008;Zenni, 2010). Nevertheless, alien plants may affect ecosystem processes by changing nutrient cycling, trophic interactions, community structure, and species functional attributes (Dogra et al., 2010), but also by affecting physical properties of ecosystems such as erosion rates, sedimentation, and the water cycle (Wittenberg and Cock, 2001;Ziller, 2001;GISP, 2005;Currie et al., 2014).
Biodiversity loss and landscape changes caused by invasive alien species may be particularly severe in protected areas (FICMNEW, 1998). No Brazilian protected area is known to be free of alien species, although such areas possibly exist (Leão et al., 2011). Ziller and Dechoum (2013) reported Abstract As a result of biodiversity and ecosystem service losses associated with biological invasions, there has been growing interest in basic and applied research on invasive species aiming to improve management strategies. Tradescantia zebrina is a herbaceous species increasingly reported as invasive in the understory of disturbed forest ecosystems. In this study, we assess the effect of spatial and seasonal variation on biological attributes of this species in the Atlantic Forest. To this end, we measured attributes of T. zebrina associated with plant growth and stress in the four seasons at the forest edge and in the forest interior of invaded sites in the Iguaçu National Park, Southern Brazil. The invasive plant had higher growth at the forest edge than in the forest interior and lower leaf asymmetry and herbivory in the winter than in the summer. Our findings suggest that the forest edge environment favours the growth of T. zebrina. This invasive species is highly competitive in the understory of semi-deciduous seasonal forests all over the year. Our study contributes to the management of T. zebrina by showing that the summer is the best season for controlling this species.

Seasonality and forest edge as drivers of Tradescantia zebrina
Hort. ex Bosse invasion in the Atlantic Forest 1,170 invasive alien species occurrences in 227 protected areas in Brazil, associated with 167 species of alien plants and 42 species of alien vertebrates. Despite the fact that the Iguaçu National Park is one of the most important subtropical protected areas in Brazil, it also has a relatively high number (29) of invasive species (Sampaio and Schmidt, 2014).
The wandering jew (Tradescantia zebrina Hort. ex Bosse) is one of the most common invasive alien species in protected areas in Brazil (e.g., Dechoum, 2010;Dickfeldt et al., 2013;Horowitz et al., 2014). This invasive plant has been reported in the Cerrado (Carpanezzi, 2011;Zenni and Ziller, 2011) and in the Atlantic Forest (Rodolfo et al., 2008;Zenni and Ziller, 2011). Tradescantia zebrina is a monotiledoneous plant from the Commelinaceae family likely native to Mexico and Northern Central America (Mantoani et al., 2013), but is currently widely distributed in the world as an introduced species (Instituto Hórus, 2020). It is a 15 cm to 25 cm tall perennial herbaceous species, with succulent stems and glabrous leaves with adaxial silver longitudinal stripes and abaxial purple epiderm (Lorenzi and Souza, 2008). Tradescantia zebrina has purple-pink trimerous flowers and produces seeds that are naturally dispersed by physical agents, such as water and wind (Lorenzi and Souza, 2008). The species shows effective vegetative reproduction by stolons throughout the year. Similarly to the congeneric species Tradescantia fluminensis Vell., which is an aggressive invasive species in New Zealand (Standish et al., 2001) and Australia (Dugdale et al., 2015), T. zebrina is more common in forests that are degraded, disturbed, or undergoing natural regeneration (Mantoani et al., 2013;Chiba de Castro et al., 2019).
Considering that invasive alien species often pose a severe threat to protected areas, obtaining basic biological information on these species is crucial (Brugnoli et al., 2009). However, approaches to control of biological invasions have rarely considered the attributes of each target species (Zalba, 2010;Ziller, 2010) Furthermore, delayed actions and poor management decisions often enable invasive species to become widespread, thus making control economically unviable (Simberloff, 2014). Tradescantia zebrina is already considered an invasive species in many continents, including the United States (Burns, 2004), South Africa (Foxcroft et al., 2008), China (Weber et al., 2008), Australia (Biosecurity Queensland, 2016), and Brazil (Zenni and Ziller, 2011). Nevertheless, the mechanisms enabling this species to become dominant in invaded areas are still poorly known, and the little information available is widely scattered (e.g., Burns and Winn, 2006;Carpanezzi, 2011;Mantoani et al., 2013;Chiba de Castro et al., 2019). Detailed information on the biology of invasive species and on how their attributes interact with the invaded environment are critical for developing and validating management strategies (Yokomizo et al., 2009;Simberloff, 2008;Panetta and Gooden, 2017).
In this study, we examine the effect of forest edge and seasonality on biological attributes of T. zebrina in a semideciduous seasonal forest in Southern Brazil. We believe that this invasive species grows better at the border and in the summer, given that it is an important invader at forest edges (Chiba de Castro et al., 2019), and shows high invasiveness under high resource availability (Burns and Winn, 2006). Moreover, we believe that T. zebrina is subjected to more stress in the forest interior and during the winter. Our findings may contribute to the management of T. zebrina by showing i) where and ii) when the species should be controlled.

Material and Methods
We carried out the study in sites dominated by T. zebrina in the Iguaçu National Park. The Park has an area of 185,262.5 ha, and is located in the western region of the Parana state, Southern Brazil (25°05' to 25°41' S and 53°40' to 54°38' W) (IBAMA, 2018). The climate in the study region is subtropical Cfa (Köppen's classification) with hot summers (Alvares et al., 2013). The mean annual precipitation ranges from 1,600 to 1,800 mm, with 500 to 600 mm distributed in the summer, 400 to 500 mm in the autumn, 300 to 400 mm in the winter, and 500 to 600 mm in the spring. The mean annual temperature ranges from 22 to 23 °C, including 25 to 27 °C in the summer, 21 to 23 °C in the autumn, 14 to 18 °C in the winter and 22 to 24 °C in the spring (IAPAR, 2019). The Iguaçu National Park has the largest remaining area of Atlantic Forest in Southern Brazil, which comprises semi-deciduous seasonal forests, mixed ombrophilous forests and early successional alluvial forests (IBAMA, 2018).
The sites selected for this study are in a region close to the main entrance of the Iguaçu Falls, which is the major tourist attraction in the Iguaçu National Park. All sites were within a radius of 1.5 km from the following position: 25°37'45"S and 54°27'52"W. The natural vegetation in this region, which is also the dominant vegetation type in the park, is a submontane semi-deciduous seasonal forest. These forests experience a leaf-fall of 20 to 25% of the trees in the winter, and occur in plain to slightly undulated areas with maximum elevation of 400 m, among other characteristics (IBAMA, 2018).
We sampled five sites with a high level of invasion by T. zebrina (over 80% of dominance assessed by visual estimation), which were selected based on the following criteria: i) separated by at least 300 m from each other; ii) belonged to distinct invasions (i.e. discontinuous invaded sites representing different populations); iii) invaded both the forest edge and interior (at least 20 m towards the forest interior). In each site we considered i) forest edge; and ii) forest interior as distinct spatial treatments, at least 10 m apart from each other. Three 0.5 m × 0.5 m quadrats (i.e. 0.75 m 2 area) were then randomly allocated in each spatial treatment at each site in the following periods: i) spring (October 2015); ii) summer (January 2016); iii) autumn (May 2016); and iv) winter (September 2016).
In each quadrat we sampled biomass, and morphological and herbivory data of T. zebrina. We were unable to obtain the dry biomass of all T. zebrina individuals inside the quadrants due to limitations of laboratory space. Instead, dry biomass data were estimated from fresh biomass measurements performed in the field, in which all individuals of T. zebrina were carefully removed from 0.25 × 0.25 m quadrats and subsequently weighed. The dry biomass was then estimated using linear regressions for each sampling site. These analyses were based on approximately seven fresh biomass samples for each sampling, comprising a variety of weights representative of the biomass variation in our sampling. Fresh biomass samples were oven dried at 60 °C to constant weight and the resulting dry biomass values were related to the initial fresh biomass values using simple linear regression. We also estimated the number of leaves in a randomly placed biomass subsample within each experimental unit, where the leaves were counted in the laboratory on the following day.
We also obtained leaf area, leaf asymmetry, and herbivory data in each of the three subsamples by quadrat. These measurements were based on ten undamaged stolons from each subsample (30 subsamples across all sites), whose sixth leaf from the apex to the base removed. This approach was derived from field observations aiming to identify the position along stolons of T. zebrina that would better reflect the last three months of plant growth. The 30 leaves from each subsample were placed on a white background under transparent glass and photographed with a scale bar. We then measured leaf area and leaf asymmetry (largest size (A)/smallest size (B)) in each image using ImageJ™. Finally, we quantified the herbivory in each subsample by visually searching for herbivory signs among the 1st and 6th leaf of the same ten stolons. The amount of herbivory in each subsample was then calculated as the average percent of leaves showing signs of herbivory.
We assessed the effect of the spatial and seasonal treatments on each attribute of T. zebrina in a randomized block design using Generalized Linear Mixed Models (GLMMs) (Bolker et al., 2009). Response variables were the number of leaves, dry biomass, herbivory rate, average leaf area, and leaf asymmetry in each 0.25 x 0.25 quadrat. We assumed a Poisson error distribution in models for the number of leaves, and a Gaussian distribution for other response variables. The fixed effects in each model were spatial treatment, seasonal treatment, and their interaction, and random effects were i) sampling site, ii) spatial treatment nested in site, and iii) seasonal treatment nested in spatial treatment. We performed all analyses in the statistical environment R (R Core Team, 2020). GLMMs for the number of leaves were conducted in the package "lme4" (Bates et al., 2015), whereas models for the remaining variables were performed in the package "nlme" (Pinheiro et al., 2017); in both cases we used Restricted Maximum Likelihood as the estimation method. When there was a significant effect of seasonal treatment, we used post-hoc pairwise tests in the package "multcomp" (Hothorn et al., 2008) to compare all combinations of treatment levels .

Results
Our analyses showed that both spatial and seasonal treatments had a significant effect on multiple attributes of T. zebrina, but indicated no significant effect of the interaction between these treatments (Table 1). Regarding the spatial treatment, T. zebrina had a greater number of leaves and higher dry biomass at the forest edge compared to the forest interior ( Figure 1). However, there were no significant differences in herbivory, average leaf area, or asymmetry between forest edge and interior (Table 1). For the temporal treatment, pairwise post-hoc comparisons between the four seasons indicated that in the winter T. zebrina had lower dry biomass ( Figure 2A) and leaf asymmetry ( Figure 2D) than in the summer (summer > winter = spring >= autumn = summer), lower herbivory than in all other seasons (summer = spring = autumn >winter, Table 1. Effects of spatial (forest edge and interior) and seasonal treatments (Winter, Spring, Summer and Autun) and their interaction on the average values of the number of leaves, dry biomass, herbivory, leaf area and leaf asymmetry of Tradescantia zebrina in invaded areas of the Iguaçu National Park. Figure as 2B), and lower leaf area than in the summer and in the autumn (summer > autumn> winter = spring = autumn, Figure 2C). The number of leaves of T. zebrina did not differ significantly among the four seasons (Table 1).

Discussion
Our results demonstrate that both forest edge effects and climatic seasonality affect biological attributes of T. zebrina. We found that this invader has a greater number of leaves and produces more biomass at the forest edge than in the forest interior, suggesting that growing conditions at the edge of semi-deciduous forests favour this species. Edges of forest fragments are often vulnerable to invasions by alien plants, especially along roads, where native vegetation cover and species richness are generally low (Catford et al., 2011). In our study the high dominance of T. zebrina at the forest edge may be associated with increased photosynthetic and relative growth rates under high resource availability (Burns, 2004), particularly as a result of high light availability compared to the forest interior (Yamashita et al., 2000). This preference for forest edges also suggests that the control of this species in the Atlantic Forest is more likely to succeed in the forest interior, as in this environment the regenerative potential of native tree species tends to be high, and both the biomass production and ability to recovery from mechanical removal shown by T. zebrina are lower in the interior than at the forest edge (Chiba de Castro et al., 2019). Nevertheless, controlling invasions both at the forest edge and in the forest interior can increase the likelihood of management success, given that forest edges invaded by T. zebrina may act as a large source of propagules to invasions in the forest interior, as has been reported for other invasive plants (Cadenasso and Pickett, 2001;Lambrinos, 2006;Thomas and Moloney, 2015).
Seasonality effects, especially the influence of seasonal treatments on the number of leaves of T. zebrina, were the most surprising results of our study, as they did not corroborate our initial hypothesis. Although we sampled a semi-deciduous forest that experiences relatively low temperature and precipitation in the winter, in our study the number of leaves of T. zebrina did not show seasonal variation. Other studies also investigated the role of seasonality in the competitive ability of invasive herbaceous species. For example, Bromus tectorum L. is a highly invasive grass in the western United States, where its seeds germinate in the autumn; this seasonal pattern is advantageous for the species, as it matches with the beginning of the senescence of native grasses in this region (Knapp, 1996). Likewise, early winter growth compared to the native community is the main competitive advantage of the invasive grass Arundo donax L. (Decruyenaere and Holt, 2005). Considering that T. zebrina also showed low biomass and leaf area in the winter, we believe that the lack of seasonal variation in the number of leaves shown by this species is associated with rapid leaf replacement; therefore, small leaves sampled in the winter were possibly newly produced and hence we found no annual variation in the number of leaves. A similar phenomenon was described for Hedychium coronarium J. Koenig, a rhizomatous herbaceous species whose ramets are replaced during the winter, and thereby suppress the germination of native species in the spring (Chiba de Castro et al., 2016, 2020. Therefore, we believe that replacing leaves in the winter may confer a competitive advantage to T. zebrina over native species in the understory of the Atlantic Forest. Contrasting with our initial hypothesis, T. zebrina exhibited maximum leaf asymmetry in the summer, suggesting that conditions are more stressful to this species in this season than in the winter. Leaf asymmetry can be influenced by multiple abiotic factors, such as level of pollution, altitude, soil moisture, and temperature (Beasley et al., 2013;Maldonado-López et al., 2019). For instance, a variety of studies have found close association between low soil moisture and high leaf asymmetry (Orians and Solbrig, 1977;Nagamitsu et al., 2004;Souza et al., 2005). Likewise, the alien tree Ficus carica L. exhibited high leaf asymmetry early in the spring in the southeastern United States, where the winter is significantly colder than in its native range, thus also suggesting a relationship between temperate and leaf stress (Cowart and Graham, 1999). While we expected that T. zebrina would be more subjected to stress in the winter, given that in the subtropical Atlantic Forest this season is considerably colder and drier than in the summer, and T. zebrina is native to the tropics, the species exhibited maximum leaf asymmetry in the summer. We believe that this pattern is associated with high leaf herbivory in the summer. Multiple studies have found positive association between leaf asymmetry and herbivory rate (Cornelissen and Stiling, 2005;Cuevas-Reyes et al., 2011;Beasley et al., 2013;Maldonado-López et al., 2019). Mari and Galassi (2010) demonstrated that herbivory pressure increased in the summer and decreased in the cold season. Importantly, despite the fact that T. zebrina produced more biomass at the border than in the forest interior, the amount of herbivory did not differ between these environments. This finding suggests that herbivory pressure was consistent with plant density and hence high plant biomass at the forest edge did not attract more herbivores. Furthermore, it is well known that higher temperatures may stimulate herbivory activity (Heiler et al., 2008;O'Connor, 2009;Lemoine et al., 2014), which is also in agreement with our results. In addition to warm temperatures, rainy seasons also favour the development of herbivorous insects (Araujo et al., 2013). For example, Cerrado insects are seasonally distributed over the year, with most of them occurring in the rainy season (81.1%) (Oliveira and Frizzas, 2008). In December 2015, which coincides with the summer treatment in our experiment, the precipitation in our study region ranged from 350 to 400 mm. This precipitation accumulation in a single month represents 20 to 25% of the average annual precipitation. Therefore, in our study region both high temperature and high precipitation in the summer seem to provide optimal conditions for herbivorous insects.
Our results suggest that in the Atlantic Forest T. zebrina should be controlled in the summer. This recommendation is mostly based both on i) the lack of seasonal variation in the number of leaves; and ii) on the higher herbivory pressure in the summer. Results on the leaf area and number of leaves of T. zebrina in our study are consistent with those found by Chiba de Castro et al. (2019), which reported rapid leaf regrowth for this species. This effective recovery from defoliation is also corroborated by the failure of previous attempts to use mechanical removal for controlling T. zebrina in the Iguaçu National Park (ICMBio, personal communication). Effective leaf regrowth is also expected following management actions in the summer, so that even after being removed during this season, T. zebrina may show a rapid recovery supported by newly grown leaves. However, it is well established that herbivory pressure tends to be much higher in young than in mature leaves (Crawley, 1989). Considering that even herbivory on mature leaves may lead to stress, as suggested by our leaf asymmetry findings, this negative effect is likely even higher on young leaves. Therefore, herbivory possibly decreases the performance of T. zebrina in the summer and hence would increase the efficiency of managing the species during this season.
Overall, our findings show that conditions at the forest edge are more suitable to the growth of T. zebrina than those in the forest interior. Maintaining a high number of leaves throughout the year enables this species to be an excellent competitor in the understory of subtropical semi-deciduous seasonal forests High biomass and leaf area in the summer suggest that most of the leaves of T. zebrina reach maturity in this season, when herbivory is maximal and appears to increase leaf asymmetry. Our study contributes to a more effective management of T. zebrina by suggesting that the summer is the best season to control the species in the Atlantic Forest. BRUGNOLI, E., MASCIADRI, S. and MUNIZ, P., 2009