Rediscovering two Isoetes species in the Brazilian Amazon and Cerrado after 167 years

Abstract Isoetes amazonica and I. gardneriana were the first two species of the genus to be collected from Brazil. Isoetes amazonica was gathered by Richard Spruce in the Amazon basin near Santarém in the state of Pará in 1850. Isoetes gardneriana was collected by George Gardner in the current Dianópolis in Tocantins State in 1843. Despite being known for a long time by botanists, these species have not been recollected since then, which raised questions about their taxonomic recognition, current distribution ranges and conservation status. Fieldwork efforts led to the rediscovery of I. amazonica and I. gardneriana after 167 years. These collections enrich our understanding of their habitats and morphologies. We provide here re-descriptions for these species. Based on IUCN criteria, Isoetes amazonica and I. gardneriana should be assigned as data deficient (DD) and endangered (EN), respectively. The rediscovery of these species raises hopes that other areas in Amazon and Cerrado biomes harbour I. amazonica and I. gardneriana, respectively. This study will serve as a basis towards the conservation of these species.


Introduction
Brazil presents the greatest diversity of plants in the world (Forzza et al. 2012), which partially reflects its large quantity of habitats. Particularly, habitats of its two largest biomes -Amazon and Cerrado -are undergoing a rapid reduction due to deforesta-tion and large scale agriculture, including soybean and cattle farming and construction of hydroelectric dams (Laurance et al. 2000;Carvalho et al. 2009). At the same time, these areas remain largely unexplored botanically (Sousa-Baena et al. 2014), which raises conservation concerns about the numerous "lost plant species" that have been known only from type specimens.
The lycophyte genus Isoetes L. is globally distributed with an estimated 250 species (Troia et al. 2016), 22 of them being endemic to Brazil (Prado et al. 2015). The genus is frequently overlooked by botanists due to its resemblance to grasses or sedges (Taylor and Hickey 1992) and due to its aquatic habitat occurring semi-to fully submerged up to 6-7 m deep in water (Middelboe and Markager 1997). As a result, many species are known only from type specimens (e.g. Pereira et al. 2016Pereira et al. , 2017Hickey et al. 2009).
Isoetes amazonica A. Braun and I. gardneriana Kunze ex A. Braun were the first two Isoetes species to be collected and described from Brazil. Isoetes amazonica was first collected by Richard Spruce in September 1850 from inundated shores of the Tapajós river near Santarém municipality in the state of Pará (Kuhn 1884). Isoetes gardneriana was first found by George Gardner (1849: 236) in 1843 in a marsh by the side of the river Preto, Mission of Duro, in the state of Goiás (currently on the border between Tocantins and Bahia in the municipalities of Dianópolis and Formosa do Rio Preto, respectively). Isoetes amazonica was published in 1880, I. gardneriana in 1862 (see Troia et al. 2016) and further information about them was compiled in the "Flora Brasiliensis" of Martius by Kuhn (1884). Despite having been collected and known for a long time, I. amazonica and I. gardneriana have not been recollected for 167 years. Our lack of knowledge about these species raises questions about their taxonomic recognition, current distribution ranges and their conservation status.
Motivated by these issues, we embarked on an attempt to rediscover these species in both the type localities and other similar environments in Amazon basin and Brazilian Cerrado.

Material and methods
For Isoetes amazonica, fieldwork was carried out along both banks of the Tapajós river, near the district Alter do Chão, municipality of Santarém, in the state of Pará, Brazil, in September 2016 and July 2017. For Isoetes gardneriana, fieldwork efforts were carried out along the margins of the Preto river in Formosa do Rio Preto (Bahia) in January 2018. Additional efforts to find this species took place in other Brazilian Cerrado areas: Ondas river, Barreiras (Bahia) -ca. 200 km away from the type location -in January 2018; Parque Nacional Serra da Mesa (Maranhão) -500 km away from the type locationin November 2017; Parque Nacional da Serra do Cipó -900 km away from the type location -in June 2018; Fazenda Modelo, Campo Experimental da Embrapa, Terenos (Mato Grosso do Sul) -1200 km away from the type location -in November 2017.
Besides field trips, specimens from the following herbaria were consulted to check for previous records of these species (acronyms following Thiers 2018): CGMS, MG, RB and UPCB (Brazil); B, E, M, HBG, P and K (Europe). These materials were compared to type specimens of I. amazonica (Spruce 1081, K [K000574506]) and I. gardneriana (Gardner 3563, B and E [E00429095]).
We checked the total monthly precipitation and average monthly maximum and minimum temperatures of the environments of these species' localities to understand the influence of both flooding and drought in their habitats and life forms. For I. amazonica, the climatic data were collected from the meteorological station located in Belterra in the state of Pará and made available by "Instituto Nacional de Meteorologia" (INMET 2019). The climatic data for I. gardneriana were obtained from Campo Grande in the state of Mato Grosso do Sul (MS) and made available by the "Centro de Monitoramento do Tempo e Clima, MS" (CEMTEC/MS 2019).
Habitat, life form, colour, size and ornamentation of the mega-and microspores, the proportion of the sporangium wall covered by the velum and the sporangial wall colouration were used in the identification of the species. The megaspores and microspores were analysed using scanning electron microscopy (SEM). Images of the spores were made by transferring the spores to aluminium stubs coated with a carbon adhesive. The stubs were then coated with gold-palladium-alloy in a sputter-coater for 180 s and then digitally imaged using a Zeiss SIGMA VP.
Since megaspore ornamentations are essential for the correct species identification, the absence of detailed images of spores during the determination process may have potentially led to the name I. gardneriana being misused for several collections of I. panamensis Maxon & C.V.Morton sensu lato. We consulted these materials to check whether the identification was correct or not in these cases. Amongst these materials were collections from: Paraguay in 1878 (Balansa 3294, P [P00170381, P00573953, P04459456]); municipality of Barreiras in Bahia, Brazil, in 1971 (Irwin 31615, P [P01591973]); an area next to type location of I. gardneriana in the municipality of Formosa do Rio Preto in Bahia, Brazil, in 2015 (Labiak 5783, UPCB with duplicates in NY [NY2697584]). In this step, megaspores of these materials were removed, images were taken using SEM and then compared with the type of I. gardneriana. We used both qualitative and quantitative characters to identify the species. The terminology used for the description of the spores follows that of Punt et al. (2007), with some modification using Pereira et al. (2016). Boxplots of the megaspore macro-ornamentation projects were generated using an R script (v. 3.0.2; R Core Team 2013).
On the other hand, none of the analysed herbarium collections appeared (Balansa 3294, Irwin 31615 and Labiak 5783) to be I. gardneriana. The megaspores of these collections are both qualitatively and quantitatively distinct from the type of I. gardneriana. The Balansa and Labiak collections have baculate-tuberculate megaspores and Irwin's collection revealed baculate-clavate megaspores (Fig. 3), which confirm that these materials represent variants of I. panamensis s.l. Additionally, the macro-ornamentation projections of the megaspores of I. gardneriana are, in general, narrower and shorter than those found in I. panamensis s.l. (Fig. 4).

Discussion
Although fieldwork investigation is fundamental to improve our understanding about how human impacts on biological systems can be recognised, mitigated or averted, fieldwork has considerably decreased in the past decades with negative implications for global biodiversity conservation (Ríos-Saldaña et al. 2018). The rediscovering of these species was only possible due to intense fieldwork; otherwise, they would have remained little known to science.
Both proper habitat and taxonomic identification of species are the first steps towards conserving biodiversity. Amongst the aquatic macrophytes, Isoetes is one of the most threatened groups (Murphy et al. 2019). However, difficulties related to finding species in the field, identifying them morphologically and, consequently, establishing their geographical distribution, hamper efforts to assess their current conservation status. Isoetes amazonica and I. gardneriana were known only from their type materials collected 167 years ago, which raised questions about their current occurrences and morphological distinction. Our rediscoveries provide a basis for a better understanding of the distribution and taxonomy of these species, which will help develop a plan to conserve these plants.
Even though I. amazonica was collected only during the dry season, we can make inferences about its habitat conditions and life forms during the year, using climatic data (see INMET 2019). Isoetes amazonica was collected as a terrestrial at the beginning of the dry season in July. However, its life form may oscillate between terrestrial and completely aquatic due to the alternating flooding and drought conditions in the Amazon basin during the year (see Marengo and Espinoza 2016). Additionally, during the driest and hottest period in August-November, its habitat may entirely dry out and this species may lose its leaves due to the combination of low precipitation, decreasing of the water table above the surface and high temperature. On the other hand, during the peak of the rainy season in March, its habitat is flooded and I. amazonica may become a completely submerged aquatic. Similarly, I. gardneriana occurs in an area which undergoes dry and rainy seasons (see CEMTEC/MS 2019 for climatic data). However, I. gardneriana grows in the deepest part of a small pond just by the side of the "vereda" grassland, which stays waterlogged year-round and feeds this pond in the dry season (see Moreira et al. 2011). This factor leads its habitat to be marshy and flooded throughout the year and I. gardneriana may rarely be found as terrestrial.
Despite the importance of habitat data for species characterisation, they provide a limited amount of information for species distinction if two or more similar species occupy the same habitat and/or show morphological convergence due to habitat adaptation (e.g. Taylor and Hickey 1992;Jiménez-Mejías et al. 2017). Both Isoetes gardneriana and I. panamensis s.l. are found in areas of "veredas" in Cerrado, which partially contribute to taxonomic difficulties involving these two species. However, they can be distinguished by qualitative and quantitative characters of megaspores.
Additionally, an Isoetes population from Itaparica lake in Xique-Xique (Bahia State) in north-eastern Brazil was tentatively identified as I. amazonica (Harley 19109, K). However, despite its resemblance to I. amazonica by size of megaspores and number and size of leaves, the presence of brown sporangium (vs. hyaline) and its occurrence in Caatinga (vs. Amazon) leads us to believe that this population is either a variant of I. luetzelburgii U. Weber or an undescribed species.
The geographical distribution of the species is crucial in assessing their conservation status (IUCN 2016). In Isoetes, the proportion of species with narrow-range distributions is remarkably high (Prado et al. 2015). The same extreme restricted distribution patterns are also found in several other aquatic macrophytes, such as Podostemaceae, Araceae, especially Cryptocoryne spp., Cyperaceae and Eriocaulaceae (Murphy et al. 2019). However, in several cases, it appears unclear whether this pattern occurs due to endemism (driven by biological factors) or collection deficiency. Although more fieldwork efforts are needed to address this question in I. amazonica, this study revealed that I. gardneriana shows a much wider distribution than previously known.
Isoetes amazonica is currently known from a single locality next to a cattle farm and, thus, it is prone to the effects of human activities within a short time. However, given its potential occurrence in other areas in the Amazon basin and the lack of current knowledge about its distribution range, I. amazonica should be assessed as data deficient (DD), according to IUCN criteria (IUCN 2016). On the other hand, I. gardneriana -which is endemic to Cerrado -is clearly undergoing a population size re-duction due to the loss of suitable habitats. The agri-business expansion, infrastructure development, weak legal protection and limited conservation incentives have led to the loss of 46% of Cerrado native vegetation and, by 2050, Cerrado may lose up to 34% of its remaining area (Strassburg et al. 2017). This habitat reduction will have a direct impact on I. gardneriana and the population size of this species may likely substantially decrease in the next years. Thus, I. gardneriana should be assigned as endangered (EN), according to IUCN criteria (IUCN 2016).
In conclusion, the rediscovering of these species raises hopes that other areas in Amazon and Cerrado biomes still harbour Isoetes amazonica and I. gardneriana, respectively. We hope that these rediscoveries spark research towards a deeper understanding of the life history of Isoetes and provide information for any future efforts to protect Isoetes amazonica and I. gardneriana from extinction.