The invasive Tradescantia zebrina does not inhibit germination of the native Hymenaea courbaril but does modulate its growth

ABSTRACT Background Invasive plant species can modify ecosystem structure and function, potentially leading to the loss of native species, environmental and biotic homogenisation, changes in nutrient cycling, impairment of ecosystem services, and decrease in the rate of regeneration of plant recruits. Aims We assessed the effects of the non-native and invasive vine Tradescantia zebrina on the germination and development of the native tree Hymenaea courbaril in north-eastern Brazil. We hypothesised that the invasive species would negatively impact the germination and development, and, via competition, the growth of the native species. Methods We conducted a germination experiment for six months in seven forest patches. Seeds of H. courbaril were cultivated in four possible combinations of two treatments: under a cover with and without T. zebrina, and on soil that originated from places with and without T. zebrina present. Results Neither T. zebrina cover or soil affected the germination or biomass of H. courbaril during the experiment. H. courbaril seedlings were taller and had longer roots when cultivated in soil that originated from places with T. zebrina. Conclusions T. zebrina presence alters soil properties in a way that triggers morphological responses in H. courbaril, which can be associated to a perceived competition context.


Introduction
A species inhabiting a place outside its natural range of occurrence is considered exotic.If it proliferates, disperses, survives and reproduces at high rates and spreads out across the new range (and thus potentially threatens the invaded ecosystem), then it is called an invasive or invasive alien species (Ziller 2001;Ehrenfeld 2010;Blackburn et al. 2011).Biological invasion, intentionally or accidentally triggered, potentially threatens local biodiversity, and may affect human health and economy (Bellard et al. 2016;Bradshaw et al. 2016;Ricciardi et al. 2017;de Castro et al. 2019).Invasive exotic species can modify ecosystem dynamics in several ways: by affecting the regeneration of native trees (Tecco et al. 2010); increasing the extinction risk for being competitively stronger than native species (de Castro et al. 2019); homogenising the environment (Ehrenfeld 2010); leading to a change in the abundance and richness of native species (Florens et al. 2010); decreasing nutrient inputs and the ratio of nitrogen fixation of autochthonous species (Pysek et al. 2012); increasing mortality of tree recruits (Nishimura et al. 2010); affecting ecosystem services (Matos and Pivello 2009); and resource acquisition, and exhausting seed banks (MacDougall and Turkington 2005).
The main mechanism associated with such ecosystem modifications is the relative competitive advantage of invasive plants over native plants through shading and root/shoot competition (Vila and Weiner 2004;Gioria and Osborne 2014).For example, competition can reduce the shoot and root biomass of native plants when growing in the presence of invasive exotic species (Morrison and Mauck 2007;Kawaletz et al. 2013).In some cases, this effect may also be due to allelochemicals that are released by the invasive plants (Callaway and Ridenour 2004;Kohli et al. 2009).More recently, it has been proposed that interactions with soil microbiota is another important mechanism that enables the success of invasive plants (Inderjit and van der Putten 2010;Dieskau et al. 2020).
Based on the dynamics of global trade and the predicted impacts of climate change, the number of naturalised exotic plants in South America is expected to increase sharply (Pysek et al. 2019), particularly in in the Caatinga in Brazil (Dias et al. 2013;Pinto et al. 2020).The Caatinga, the largest seasonally dry tropical forest in the Neotropics (da Silva et al. 2017), is heterogeneous and includes enclaves of forest which are threatened by several nonnative introduced plant species (Porto et al. 2004;Queiroz et al. 2017;Silva et al. 2017).The conservation of the vegetation of the Caatinga is challenging because its biodiversity is still poorly known and few areas are protected (Santos et al. 2007;Teixeira et al. 2021), areas that also face additional pressures of logging and fire (da Silva et al. 2017).
One of the non-native plant species in the Caatinga is Tradescantia zebrina Hort.ex Bosse (Commelinaceae), a highly invasive succulent creeping herbaceous plant, currently present in several of the main ecoregions of Brazil, including the Cerrado, the Atlantic rain forest, and the Caatinga (Zenni and Ziller 2011;Ziller and Dechoum 2013;Ribeiro et al. 2014;Machado et al. 2019).Tradescantia zebrina develops dense patches, where seedlings of native plant species are virtually absent (Figure S1).This impact raises the question of whether T. zebrina has a relative competitive advantage and is excluding native species due to above-ground competition or to changes in soils that disfavour native species.
In this study we tested (i) if the non-native T. zebrina negatively affects the growth of a native plant species, and (ii) if this impact is related to above-ground competition or to modification of the soil substrate.We used Hymenaea courbaril L. (Fabaceae) as a model species.Hymenaea courbaril is a native tree species that occurs in the forests of the Caatinga at Brejos de Altitude and it may be negatively affected by T. zebrina, which has been shown to interfere with native recruits in tropical forests (Mantoani et al. 2013;de Castro et al. 2019).
Previous studies have reported negative effects of T. zebrina on the germination and postgermination development of native tree species and their recruits, such as Anpidosperma polyneuron and Nectandra megapotanica in southern Brazil (Mantoani et al. 2013;de Castro et al. 2019), but the mechanism have not been identified.To identify the underlying causes of negative effects by T. zebrina, we planted seeds of H. courbaril in a fully factorial design, considering T. zebrina cover (i.e.inside or outside T. zebrina patches) and substrate origin (soil from T. zebrina patches or from forest understorey without T. zebrina).We expected that H. courbaril seeds on T. zebrina substrate or growing inside T. zebrina patches would have lower germination rates and poorer post-germination development (lower biomass, smaller shoot, and root length) than H. courbaril planted with no interference from T. zebrina.

Study species
Tradescantia zebrina probably originates from Mexico and was brought to Brazil as an ornamental plant due to its striking colouration of its leaves (Mantoani et al. 2013).It reproduces readily by stem and root fragments detached from the plant (de Castro et al. 2021) and frequently creates dense mats that may inhibit seedling recruitment (de Castro et al. 2019).Tradescantia zebrina competes with native species (Ribeiro et al. 2014;de Castro et al. 2021), hinders the regeneration of tree species (Mantoani et al. 2013), increases the mortality of seedlings (de Castro et al. 2019), affects the abundance of native species (de Castro et al. 2019), and alters the saprotrophic soil fungal community (Bail et al. 2022).

Study area
The municipality of Areia is in the highest lying region of the Borborema massif in the Agreste Paraibano, at an elevation of 618 m, and exhibits a rugged topography (mIBGE 2019).The study took place in tropical seasonal rain forests enclaves of the Brejos de Altitude area, characterised by deep and relatively fertile soils (Velloso et al. 2002;Queiroz et al. 2017).The climate is humid, with mild temperatures, an average annual temperature of 22°C, average annual rainfall of 1500 mm, and mean annual relative humidity near to 85% (Porto et al. 2004).There are small and medium-sized semiperennial watercourses (da Silva et al. 2017).These forests have been affected by deforestation, predatory hunting, and exploitative forest resource use (Silva et al. 2017).
We identified populations of T. zebrina in seven forest remnants where we subsequently carried out our experiment (areas 1 to 7 in Figure S1).Three of the areas containing T. zebrina were in the Mata do Pau Ferro Protected Area, while four others were located inside the Areia Campus of the Federal University of Paraíba (Figure S1).Each T. zebrina patch was at least 5 m × 5 m and 10 m × 20 m at most (see for example Figure S1).

Experimental design
Hymenaea courbaril seeds were obtained from the Plant Ecology Laboratory, Centre for Agricultural Sciences, Federal University of Paraíba.They were carefully inspected by naked eye for insect infestation and malformation and selected for the experiment.An electric grinder was applied to scarify seeds to speed up the breaking of dormancy (Mayer and Poljakoff-Mayber 1989) and seeds were soaked in water for 48 hours, with the water regularly replaced every 12 hours.This process allows the water to reach the embryo, which facilitates germination around 12 days after this treatment (Souza et al. 2015).Following this treatment, seeds were surface dried outdoors, placed in a closed container, and brought to the experimental sites.
Our experiment had two treatments, with two levels each: (1) soil treatment, soil from places with and without the presence of T. zebrina, and (2) cover treatment, related to current local presence or absence of T. zebrina.We delimited two plots of 1.5 m × 1 m in size in each of the seven experimental areas: one inside T. zebrina patches and one outside T. zebrina patches.We collected soil samples from each plot.Soil was sieved to remove branches, leaves, and stones that could affect seed germination, and used to fill labelled 13 cm × 12 cm plastic bags.
In each of the plots we dug 20 holes, 12 cm wide and 10 cm deep in four columns and five rows (Figure 1).Holes in two columns were filled with the bags containing T. zebrina soil (Soil+Zeb) and the other two columns filled with the bags containing soil without T. zebrina (Soil-Zeb; Figure 1).Lastly, we planted one seed of H. courbaril in each of the 20 soil-filled buried bags in each plot (Figure 1).In total, we sowed 280 seeds distributed in seven areas (Figure S1).Each area had 40 seeds planted, 20 in T. zebrina cover plot (Cover+Zeb) plus 20 in the plot without T. zebrina (Cover-Zeb; Figure 1).Of these 20 seeds in each cover-treatment plot, 10 were grown in soil from areas with T. zebrina (Soil+Zeb) and 10 in soil from areas without T. zebrina (Soil-Zeb; Figure 1 and iv) cover with T. zebrina and soil without T. zebrina (Cover+Zeb/Soil-Zeb).Because of possible site effects related to particularities of the seven forest areas included in the experiment, we considered each forest area as a block in the statistical analysis.
We watered the buried bags every week, both before and after seed germination, using an equivalent amount of water per bag.Initially we added 100 ml of water per bag weekly.When the seedlings reached around 10 cm we added 150 ml of water per bag weekly.We visually inspected them to monitor the germination process and measure plant growth.As seedlings emerged, we measured shoot length with a tape measure, from the ground to the apical bud of the largest branch.Measurements and observations were conducted weekly, over 24 weeks, from May to November 2020.
At the end of week 24, the 190 surviving plants were removed from the plots and placed in labelled plastic bags for processing in the laboratory.Soil was washed off the roots of the seedlings and the remaining water blotted off.The length of the roots and the shoots was measured with a ruler and plant dry weight was measured using a digital scale (Urano® UD6000/1 L).Dry weight was determined after plants were dried for 96 hours at 65°C in an air circulating oven.

Data analyses
To assess the effect of T. zebrina on the germination of H. courbaril seeds we used a generalised linear mixed-effect logistic regression using the function 'glmer' in the R package 'lme4' (Bates et al. 2015).The response variable was the proportion of germinated seeds and thus we used a model of binomial distribution of errors (Crawley 2015).Fixed-effect predictors were the two treatments (T.zebrina soil and cover) and their interaction.Forest patches (statistical blocks), and repeated visits (week 1 to 24) were considered random effects in this model.
To evaluate the effect of T. zebrina soil and canopy cover on H. courbaril (final dry biomass, shoot and root length, and the proportion of root: shoot length), we fitted linear mixed-effect models using the function 'lmer' in the R package 'lme4' (Bates et al. 2015), considering the interaction between T. zebrina soil and canopy cover as fixed effects and forest patch as a random effect.Four dependent variables were separately analysed: plant dry weight, shoot length, root length, and root:shoot length.To achieve normality of the residuals, we transformed dry weight, root length, and the ratio root:shoot length using normal score prior the analyses using the function 'blom' from 'rcompanion' R package (Mangiafico 2022).Shoot length was squared-transformed, i.e. by taking its original value to power 2. We made a graphical comparison of residual quantiles to check approximate normality using the function 'qqnorm' from the R package 'stats' (R Core Team 2022).This visual inspection showed a normal pattern, with no extreme outliers in any of the models (data and R scripts are available at: https://osf.io/a7pmf/).All graphics were made with the ggplot2 package (Wickham 2016) in R v. 4.1.3 (R Core Team 2022).To obtain the coefficient of determination for all statistical models we used the function 'r.squaredGLMM' from 'MuMIn' R package (Barton 2022).We opted to consider the conditional R 2 GLMM , which is interpreted as a variance explained by the entire model, including both fixed and random effects (Nakagawa et al. 2017).

Biomass
Neither cover-treatment nor soil-treatment were significant predictors of plant dry weight in the linear mixed-effect models (Table 1; Figure S2).

Shoot and root length
Final shoot length and root length were both statistically significantly related to soil treatment (Table 1, Figure 2), while the cover treatment and the interaction term were not significant.Plants grown in soil collected under T. zebrina populations (Soil+Zeb) were taller (mean ± SE) by 6% (31.2 ± 0.59 cm) than seedlings grown in Soil-Zeb soil (29.4 ± 0.58 cm).
The statistical model explained 33% of the variation in shoot length.
Similarly, soil-treatment significantly affected root length, while cover-treatment and the interaction term were not significant (Table 1, Figure 2(b)).Soil collected from under T. zebrina populations (Soil+Zeb) resulted in final root length that was 13% more (30.7 ± 0.76 cm) than in Soil-Zeb soil (27.1 ± 0.66 cm).The statistical model explained 18% of the variation in root length.

Discussion
Contrary to our initial expectations, T. zebrina did not affect the germination rate of H. courbaril.Approximately 80% of all sown seeds germinated, Table 1.The results of linear mixed-effects models evaluating the effect of the soil treatment (Soil+zeb/Soil-zeb), cover treatment (Cover+zeb/Cover-zeb) and their interaction term (Soil*cover) on germination percentage, plant dry weight, shoot and root length, and root:shoot length ratio.We used forest patch as random effect predictor in all models.For the logistic model predicting germination percentage we considered the total sample size of 280 seeds sown, while for the other dependent variables we considered the 190 plants which survived until the end of the experiment.Significant results (P < 0.05) are shown in bold.Seedlings planted in soil sourced from areas with T. zebrina (Soil+zeb, represented by blue dots or darker grey in black and white version) exhibited taller shoot lengths (β GLMM : = 142.667,P = 0.017; Table 1).In B, the residuals from the root length model, adjusted for the block effect, are presented.The figure includes the four combinations of treatment levels.Seedlings planted in soil derived from areas with T. zebrina (Soil+zeb) demonstrated longer root lengths (β GLMM : = 0.426, P = 0.021; Table 1).
which falls within the range of germination percentages previously described for this species (69-96%, Cruz et al. 2001;Duarte et al. 2016;Pagliarini et al. 2016).This outcome contradicts previous findings, which demonstrated that the presence of T. zebrina had a negative impact on the regeneration of tree species and the germination of native plant species in the Atlantic Forest (Mantoani et al. 2013;de Castro et al. 2019).Similar adverse effects were observed in studies that investigated the allelopathic effect of T. zebrina extract on the germination and early seedling growth of lettuce (Lactuca sativa) and tomatoes (Lycopersicum esculentum).Lettuce seeds exposed to T. zebrina extract exhibited a reduced germination percentage and slower germination speed (de Moura et al. 2018), and both tomatoes and lettuce displayed reduced root length when cultivated with T. zebrina extract (Martins et al. 2014).It is possible that Hymenaea courbaril has low nutritional requirements during germination and early stages of development (Lorenzi 2002), which could explain the lack of an effect by T. zebrina on the germination time and rate that we observed.Although there were no differences in biomass production, we did observe an increase in both root and shoot length of H. courbaril when grown in soil collected from T. zebrina populations.In contrast, the presence of T. zebrina plants (cover treatment) did not affect the growth of H. courbaril.While the differences in plant height (6% taller) and root length (13% longer) may seem modest, our results indicate that these differences are significant and exceed what would be expected by chance.Consequently, the soil modifications induced by T. zebrina promote growth enhancement in H. courbaril seedlings.Our study suggests that this increase is linked to H. courbaril's ability to detect the presence of competing species through soil cues and adjust its phenotype accordingly.The specific soil cues and the subsequent implications for H. courbaril's fitness resulting from the increased length are yet to be evaluated in future investigations.
Plant competition can induce phenotypic plasticity, leading to increased sapling height (Dudley and Schmitt 1996;Lepik et al. 2005;Nishimura et al. 2010).Plants gain an advantage by extending their root systems or growing taller, thereby improving their access to nutrients and water (Lepik et al. 2005).By rapidly attaining greater height and developing longer roots, H. courbaril may potentially overcome the detrimental effects of shading and nutrient depletion in the soil.H. courbaril has a wide geographic distribution in South America, which gives it the potential for extensive intraspecific variation in morphology, phenology, and genetic diversity (Futuyma 2009).
While we found that H. courbaril was taller and had longer roots when cultivated in T. zebrina soil, there was no detected modification in biomass production (dry weight).This suggests that the soil modifications promoted by T. zebrina do not improve H. courbaril's development.Instead, H. courbaril seedlings simply adjust their growth in response to an intensified competitive environment (Nishimura et al. 2010).
Recent findings from a study conducted in Atlantic Forest fragments have indicated that the presence of T. zebrina alters the soil fungal community (Bail et al. 2022).The abundance and diversity of soil fungi were higher, and the dominance of common fungal species was lower in areas not invaded by T. zebrina compared to areas covered by T. zebrina.The presence of T. zebrina facilitated the growth of non-ligninolytic saprophytic fungi, potentially increasing soil metabolic activity in invaded areas (Bail et al. 2022).Based on our current findings, it appears that soil collected from established populations of T. zebrina had a greater impact on the growth of H. courbaril than the competition for light caused by the presence of T. zebrina cover.Although we did not analyse soil properties in this study, in line with these recent findings, we propose that the major impact of T. zebrina on native populations is due to the altered soil properties caused by its presence.Future investigations examining soil nutrients, allelochemicals, and microbiota in the presence and absence of T. zebrina would enhance our understanding and management of potential plant invasions.

Conclusion
Our experiment revealed that the presence of T. zebrina in moist forest enclaves in the Caatinga did not impact the germination or biomass production of germinated H. courbaril seedlings.This finding contradicts most studies examining invasive species, which have documented negative interactions leading to harm to native species and the ecosystem.In contrast, we observed that the native species exhibited increased shoot height and root length when cultivated in T. zebrina soil.This effect, associated with the soil treatment, suggests that T. zebrina induces significant modifications in the soil, triggering a morphological response in H. courbaril.

Figure 1 .
Figure 1.The figure illustrates the experimental design, showcasing the two treatments, each with two levels.Soil treatment: Hymenaea courbaril (jatobá) seeds were planted in pots using soil sourced from areas with T. zebrina (Soil+zeb) and soil from areas without T. zebrina (Soil-zeb).In the cover treatment the pots containing the seedling were either placed within a T. zebrina patch (Cover+zeb) or located in nearby areas without T. zebrina patches (Cover-zeb).The left panel of the figure represents the cover treatment with T. zebrina and is visually depicted by the inclusion of T. zebrina branches.

Figure 2 .
Figure 2. Residuals shoot and root length models, diamond symbols indicate mean ± standard error.A displays the residuals from the shoot length model after accounting for the block effect.The four combinations of treatment levels are presented.Seedlings planted in soil sourced from areas with T. zebrina (Soil+zeb, represented by blue dots or darker grey in black and white version) exhibited taller shoot lengths (β GLMM : = 142.667,P = 0.017; Table1).In B, the residuals from the root length model, adjusted for the block effect, are presented.The figure includes the four combinations of treatment levels.Seedlings planted in soil derived from areas with T. zebrina (Soil+zeb) demonstrated longer root lengths (β GLMM : = 0.426, P = 0.021; Table1).