Introduction

Successful germination and seedling establishment are important regeneration traits that lead to effective processes of biological invasion (^{Pyšek & Richardson 2007}; ^{Gioria & Pyšek 2017}). In disturbed ecosystems, potentially invasive species are often more adapted to disturbance regimes when compared to the native ones (^{Facon et al. 2006}). Fire is a principal disturbance in many ecosystems, and its effects depend on its origin (natural, by a lightning strike, or accidental, by human actions), season and intensity (^{Whelan 1995}), as well as on the particular adaptations of the community species (^{Huston 2004}; ^{Lockwood et al. 2007}).

At a community level, fire could affect plants by altering species richness, composition and structure (^{Whelan 1995}; ^{Fidelis et al. 2012}). At the individual level, some species (such as grasses) could have a rapid regrowth, increased productivity, flowering and seed production after fire (^{Whelan 1995}). Fire also could affect the survival, viability and germination of seeds (^{Moreira et al. 2010}; ^{Clarke & French 2005}).

Seed exposure to fire-related germination cues may enhance recruitment of both invasive and native species (^{Adkins & Peters 2001}; ^{Figueroa et al. 2009}; ^{Arán et al. 2013}). Heat shock and plant-derived smoke solutions, for instance, can stimulate germination of many grass species in fire-prone ecosystems (^{Clarke & French 2005}). On the other hand, little effect of heat shock and exposure to smoke solution in germination has been reported for herbaceous species from Brazilian grasslands and savannas (^{Overbeck et al. 2006}; ^{Stradic et al. 2015}; ^{Fichino et al. 2016}; ^{Ramos et al. 2016}; ^{Paredes et al. 2018}).

However, seeds of many Cerrado (^{Ribeiro et al. 2013}; ^{Fichino et al. 2016}) and Chaco species (^{Jaureguiberry & Díaz 2015}) may tolerate exposure to temperatures up to 100 °C. Therefore, seed survival to fire has been suggested as a critical trait for post-fire regeneration (^{Pausas et al. 2004}; ^{Paula & Pausas 2008}; ^{Ramos et al. 2016}). Species that have fire-tolerant seeds and are able to promptly germinate after fire may have a competitive advantage in post-fire environments (^{Ramula et al. 2015}). Hence, in Brazil, seed tolerance to high temperatures, rather than fire-stimulated germination, appears to promote species’ survival in fire-prone environments (^{Fichino et al. 2016}; ^{Fidelis et al. 2016}) and may foster biological invasion after fire events (^{Paredes et al. 2018}).

Fire has been present in the Cerrado for millions of years, and has received some academic and scientific attention (^{Simon et al. 2009}). Nevertheless, few studies have investigated the direct effects of fire on germination of Cerrado grass species (but see ^{Ramos et al. 2016}; ^{Paredes et al. 2018}). In addition to the native species, C₄ African grasses (such as Urochloa spp., previously Brachiaria) were introduced in Brazil for cattle forage (^{Pivello et al. 1999}a; ^b; ^{Durigan et al. 2007}; ^{Ziller & Dechoum 2013}). Such African grasses have spread from the pastures and became a threaten to Cerrado native herbs by reducing their biomass and diversity (^{Pivello et al. 1999}a; ^b; ^{Damasceno et al. 2018}).

Invasive grasses can also change fire behaviour by enhancing flame height and temperature, thus increasing the mortality of native species (^{Mistry & Berardi 2005}; ^{Gorgone-Barbosa et al. 2015}). Improved understanding of the direct effects of fire on seed germination and on seedling recruitment may provide useful insight to control these species in protected areas (^{Paredes et al. 2018}). Therefore, understanding fire effects at early stages of plant development is fundamental to elucidating biological invasion processes (^{Pyšek & Richardson 2007}). Studies of post-fire regeneration strategies of invasive and native species are also crucial for planning of conservation and management programs (^{Paredes et al. 2018}). Such programs may include prescribed fire management in order to identify mechanisms of post-fire competitive advantage of the invasives, and thus avoid their proliferation in burned areas.

Because Cerrado is subjected to frequent fires, we initially expected seeds of native and invasive grasses to respond to fire-related cues (high temperature and smoke water solution, a proxy for smoke) by tolerating or benefitting from fire. Given that several African grasses outcompete Cerrado grasses and become invasive (^{Pivello et al. 1999}a; ^b) we hypothesized higher seed survival/germination of such African grasses under fire events compared to native Cerrado grasses. Therefore, in this study we aimed to evaluate the effects of heat shock and smoke (fire-related cues) on seed viability, germination percentage, and mean germination time (MGT) of three invasive (Melinis minutiflora, Urochloa decumbens and Urochloa brizantha), and three native (Axonopus pressus, Aristida setifolia and Gymnopogon foliosus) grass species frequently found in open Cerrado. Specifically, we assessed (1) the effect of heat shock (60, 100 and 200 °C) and (2) the effect of smoke water solution on seed germination.

Materials and methods

Focal species

We tested germination of six grasses that commonly co-occur in the Cerrado: three native species: Axonopus pressus (Nees ex Steud.) Parodi, Aristida setifolia Kunth and Gymnopogon foliosus (Willd.) Nees, and three of the most aggressive invasives: Melinis minutiflora P. Beauv, Urochloa decumbens (Stapf) R.D. Webster and Urochloa brizantha (Stapf) Webster (^{Pivello et al. 1999}a; ^b). Urochloa species are perennial grasses that copiously resprout after fire, but their seeds might also germinate under temperatures up to 50 °C (^{Gorgone-Barbosa et al. 2016}). Melinis minutiflora, although able to resprout, apparently regenerates primarily through seeds (^{Lorenzi 2008}). Among the native species, A. pressus and A. setifolia can resprout after fire, while G. foliosus is an annual grass. Germination percentages of A. setifolia and G. foliosus are usually higher than 50 % (^{Carmona et al. 1998}; ¹⁹⁹⁹; ^{Kolb et al. 2016}), while A. pressus shows a high proportion of empty seeds (E Gorgone-Barbosa unpubl. res.), and thus low germination percentage (^{Kolb et al. 2016}).

Seed collection

Since the three invasive species are widely commercialized, we used commercial seeds obtained from certified producers. The selected native species were collected in patches of campo sujo, which is an open Cerrado physiognomy dominated by grasses and forbs with scattered shrubs and dwarf trees (^{Coutinho 1978}). Axonopus pressus and G. foliosus were collected in Itirapina Ecological Station (22°15’S 47°49’W; Southeastern Brazil), and A. setifolia was collected in a private property (close to the Jalapão State Park) in the Tocantins state (10°22’S 46°40’W; Northern Brazil).

Seeds of each species were collected from at least 20 individuals and stored at room temperature for less than three months until the beginning of the experiments. Heat shock experiments were conducted first, and smoke solution effects tested with a difference of around two months later. Prior to the experiments, the seeds were visually screened to avoid malformed and/or empty seeds, which were discarded. Detachable parts of the propagules (usually the glumes and palea) were also carefully removed, avoiding any mechanical damage to the seed.

Germination experiments

To test seed responses to fire cues, we subjected them to heat shocks and smoke solutions, under laboratory conditions. We subjected seeds to dry heat shocks of 60 °C, 100 °C and 200 °C for one minute in a pre-heated electric oven. These treatments were chosen based on field-recorded fire temperatures (^{Coutinho 1978}; ^{Miranda et al. 1993}; ^{Daibes et al. 2018}). Belowground temperatures reach up to 60 °C in the soil seed bank and increase with aboveground fuel load (^{Daibes et al. 2017}). When exposed at the soil surface, seeds face temperatures >100 °C, mostly being killed by temperatures >200 °C (^{Daibes et al. 2018}). Moreover, Cerrado fires are fast and rapidly consume the herbaceous biomass, therefore the heat pulse rarely reaches more than a few minutes (^{Miranda et al. 1993}; ²⁰⁰²).

For the heat shock treatments, we used five replicates of 20 seeds per treatment for each of the six species. We placed each replicate separately into the oven to avoid pseudo-replication (^{Morrison & Morris 2000}). Seeds of the different study species were exposed to the treatments all together at the same time, but separately for each replicate. Control seeds were not exposed to heat shock (untreated seeds).

We used smoke water as a proxy for smoke to examine smoke effects. We prepared aqueous smoke solutions following the methodology adapted by ^{Moreira et al. (2010}) and ^{Fichino et al. (2016}). We heated 5 g of dry aboveground biomass - composed mostly of native grasses from an open Cerrado savanna - in an oven for 30 min at 200 °C. Immediately after removing the biomass from the oven, we added 50 ml of distilled water into the material and, after 10 minutes, we filtered the solution in filter paper (Whatman n°1). We prepared one solution for each replicate to avoid pseudo-replication. Subsequently, seeds were soaked in the aqueous smoke solution for 24 hours. Control seeds were soaked in distilled water for the same period.

After both experiments, seeds were placed in Petri dishes (60 mm diameter) with moistened filter paper (distilled water) and placed in germination chambers at a constant temperature of 27 ºC and light cycle of 12 h (according to ^{Fichino et al. 2016}). Seeds were observed every two days for 30 days. We counted and removed germinated seeds that showed primary root protrusion (^{Bewley et al. 2013}). By the end of germination experiments, we tested non-germinated seeds of the control and heat shock experiments for viability, using a 1 % tetrazolium salt solution (pH 7) (^{Hilhorst 2011}).

Data analysis

We calculated the percentage of germination and the mean germination time (MGT), according to ^{Ranal & Santana (2006}). Besides germination, seed survival could also be affected in the heat shock treatments, and therefore we calculated the total seed viability as the germination percentage + the seeds stained as viable by the tetrazolium tests. Separately for each species, we performed one-way analysis of variance (ANOVA) with permutation tests (10,000 iterations, Euclidean distance between sampling units) to verify the effect of treatments (heat shocks and smoke solution) on seed viability, germination percentage, and MGT. Because germination times are crucial to determine the early steps of invasion process (^{Gioria & Pyšek 2017}; ^{Gioria et al. 2018}), we also verified differences in the MGT among species, independently of the treatment. In all analyses, we used permutation tests since there are no assumptions regarding normal data distribution. All statistical analyses were performed using the software MULTIV (^{Pillar 2006}).

Results

The invasive M. minutiflora showed germination percentages around 60 %, regardless of the heat shock treatment (Tab. 1), and average MGT of 5.1 days (Tab. 2). Additionally, M. minutiflora seed viability was not affected by any of the heat shocks (Tab. 3). Seed germination of U. decumbens varied from 25 to 42 %, but showed no significant difference among temperatures (Tab. 1), while MGT was < 4.3 days in all heat shock treatments (Tab. 2). Seed viability of U. decumbens was significantly reduced at 100 and 200 °C (Tab. 3). Seeds of U. brizantha showed germination percentages around 60 %, but it was reduced to <20 % when seeds were subjected to 200 °C (Tab. 1). Average MGT was of 4.7 days and it was not affected by heat shock treatments (Tab. 2). Viability of U. brizantha also showed values ca. 15 % higher than seed germination in the control and showed an abrupt reduction in the hottest treatment (Tab. 3).

Heat shock did not affect germination of native grasses. Aristida setifolia seeds showed high germination percentages (from 63 to 75 %) and viability higher than 70 %, both not affected by any of the temperatures (Tabs. 1 and 3 respectively). MGT of A. setifolia varied from six to nine days, with no difference among treatments (Tab. 2). On the other hand, low germination percentages and viability (≤ 20 %, Tabs. 1, 3) were observed in A. pressus in all treatments, showing a slow MGT (14.3 days on average, Tab. 2). Gymonopogon foliosus had germination percentages always higher than 60 % and seed viability of nearly 100 % in all treatments (Tabs. 1 and 3, respectively). In contrast to the other two native species, G. foliosus seeds germinated very fast - on average 2.5 days - with no difference among treatments (Tab. 2).

Neither germination percentage nor MGT of invasive and native species were affected by the smoke solution treatment. Despite the variation, values were not statistically different (Tab. 4). Native species germination percentages were ca. 60 % for G. foliosus and A. setifolia (from 54 to 64 %) and lower for A. pressus (from 26 to 32 %) when subjected to smoke solution, indicating no treatment effect (Tab. 4).

When comparing MGT among the six species independently of treatment (combining heat shock and smoke), we observed that seeds of invasive species germinate approximately twice as fast as the native ones (P <0.0001), except for the native G. foliosus, which had the lowest MGT value (2.5 days, P <0.001 - Tab. 5).

Discussion

Fire cues did not trigger germination of the study species, regardless of whether they were invasive or native grasses. On the contrary, heat shock decreased germination and/or viability of the invasive Urochloa spp. Such results contrast with other invasive species elsewhere (^{Arán et al. 2013}; ^{Cóbar-Carranza et al. 2015}), and therefore fire-related cues apparently do not directly facilitate the invasion of these African grasses in the Cerrado savannas. Seeds of the invasive Melinis minutiflora and the native species were not affected by high temperatures, corroborating similar results described for native and invasive grasses in the Cerrado, under temperatures from 50 to 150°C (^{Paredes et al. 2018}).

Because seeds of the native grasses retained their viability similar to controls under the hottest treatment (200 °C), their tolerance to high temperatures can be considered as an important trait for persisting under Cerrado fires (^{Ramos et al. 2016}). Hence, fire survival may provide propagule supply for the species regeneration in frequently burned environments (^{Overbeck et al. 2006}; ^{Fichino et al. 2016}). However, fire temperatures just below the soil surface increase only a few degrees during Cerrado typical fires (^{Miranda et al. 1993}; ^{Daibes et al. 2017}). Therefore, most propagules of either invasive or native species would survive during fires when incorporated to the soil seed banks.

Persistence of buried propagules in the soil seed banks would thus allow the recolonization of the recently burned environment. Hence, recruitment from seed occurs with the onset of the well-marked rainy season in Cerrado (^{Andrade & Miranda 2014}; ^{Ramos et al. 2017}). Once seeds survived fire heat shock, indirect effects of disturbance could thus benefit their germination in the post-fire environment. Such effects include enhanced temperature fluctuations, which may break physiological dormancy of invasive grasses (^{Musso et al. 2015}; ^{Gorgone-Barbosa et al. 2016}). Other factors related to biomass removal could be related to reduced competition and/or favourable microsites, but competitive interactions between invasive and native species are still poorly understood and difficult to measure in the field (^{Zouhar et al. 2008}).

Our results showed no effect of smoke water on germination, similar to the lack of smoke-mediated germination which has been reported for Cerrado species in the literature (^{Stradic et al. 2015}; ^{Fichino et al. 2016}). This absence of a positive response strongly contrasts to germination of Australian grasses, which show smoke-stimulated germination (^{Clarke & French 2005}). However, smoke has recently been suggested to enhance root growth of Cerrado seedlings (^{Ramos et al. 2019}). Moreover, the effect of smoke solution depends on concentration and environmental interactions, as reported for Mediterranean ecosystems (^{Crosti et al. 2006}; ^{Moreira & Pausas 2018}). Because germination of grasses can be light-dependent (^{Carmona et al. 1998}; ^{Baskin & Baskin, 1998}), further studies should also consider the role of smoke on stimulating germination under dark conditions (see ^{Gardner et al. 2001}). Such an approach could help to explain the post-fire germination of propagules buried in soil seed banks.

Irrespective of fire-related cues, a shorter germination time has been demonstrated to enhance species’ invasiveness (^{Pyšek & Richardson 2007}; ^{Gioria & Pyšek 2017}). In accordance with that, our results showed that MGT was usually lower for the invasive species compared to the native ones. In disturbed environments, seeds shall at first place survive fires and then find conditions for germination to occur, thus the timing of germination is actually one aspect of a more complex process. Therefore, invasive species could show a competitive advantage for quickly recolonizing the post-fire environment (^{Jauni et al. 2015}; Gioria et al. 2018). The exception was G. foliosus, a native species that showed the lowest MGT of all studied species. Being an annual species, its population persistence after fire relies only on regeneration from seeds (^{Pausas & Keeley 2014}) and, therefore, germination time becomes critical for this species’ survival.

Hence, germination of both M. minutiflora and G. foliosus may be favoured in the post-fire environment, given their seed tolerance to high temperatures and rapid germination. However, several other regeneration traits, such as high seed production, have been recognized to increase invasiveness (^{Pyšek & Richardson 2007}). Melinis minutiflora produces more than 70,000 seeds/m² (^{Martins et al. 2009}) with high viability and persistence in the soil seed bank - for more than three years (^{Carmona & Martins 2010}). These traits may promote a rapid post-fire establishment, increase competitiveness over the natives and facilitate the invasion (^{Lake & Leishman 2004}; ^{Tierney & Cushman 2006}).

Resprouting ability is also a post-fire regeneration strategy (^{Pausas et al. 2004}). Therefore, many species from fire-prone environments may persist by forming a bud bank (^{Fidelis et al. 2014}), rather than post-fire recruitment from the seeds (^{Paula & Pausas 2008}). This could be the case of Urochloa spp., considering that - despite the reduced germination and viability percentages under the highest temperatures - U. brizantha has shown a remarkable capacity for post-fire resprouting (^{Gorgone-Barbosa 2016}). Hence, once seeds have colonized (and established in) a certain area, resprouting seems to be a major strategy to warrant Urochloa persistence after fire. This ability is so effective that may provide advantage to Urochloa species even over M. minutiflora in the invaded habitats where they co-occur (^{Pivello et al. 1999}a).

In summary, seeds of native grasses are tolerant to Cerrado fires, as well as seeds of the invasive M. minutiflora. On the other hand, U. brizantha and U. decumbens seeds are sensitive to hotter temperatures (100 and 200 °C) and rely mostly on resprouting to persist in the burned savannas, although a proportion of seeds would probably tolerate fire temperatures if incorporated into the soil seed bank. In general, invasive species showed the ability to germinate faster than most natives, which helps to explain their invasiveness in the Cerrado. Smoke solution showed no effect either on invasive or native species, but further research is needed to explore this topic. The information obtained in this research is fundamental to elucidate post-fire invasion process in the Cerrado and may guide management strategies for controlling invasive species.