Unravelling the ecological drivers of Schistostega pennata (Hedw.) F.Weber & D.Mohr on the Iberian Peninsula: distribution and conservation

ABSTRACT Introduction In the Iberian Peninsula, Schistostega pennata is a rare, acidophilic, Holarctic moss with a characteristic glowing protonema when illuminated by sunlight. It is typically found in dark and humid habitats and is categorised as Threatened or Near Threatened on the Iberian Red Lists. The main aim of this study was to improve knowledge of the distribution, ecology and conservation of S. pennata on the Iberian Peninsula. Methods All data that were available to us for Schistostega pennata on the Iberian Peninsula (relating to distribution, ecology, phenology) were compiled from the literature, citizen science platforms, and our own observations. Species distribution models were created in Maxent to provide the first approximation to its potential Iberian distribution. Key results Twenty-six new populations of Schistostega pennata were located, making a total of 73 records for the Iberian Peninsula, mostly from natural and artificial caves, cavities, mines and crevices, and essentially on siliceous substrates. All observations corresponded to areas with abundant rainfall, mild temperatures, and at least some oceanic influence. Species distribution models predicted an expanded potential climatic range in the Iberian temperate region and several montane Mediterranean areas. Conclusions There remains very little information on Schistostega pennata on the Iberian Peninsula. Existing data coincide with the ecology of the species, although other niches are exploited in other parts of the world that do not seem to occur in this area. It is important to promote its conservation, given the fragmentation of its populations and habitat requirements. Monitoring and initiation of exhaustive sampling campaigns involving citizen science could be a good way to approach this.


Introduction
Schistostega pennata (Hedw.)F.Weber & D.Mohr is the only member of the family Schistostegaceae and one of the most intriguing of all mosses.'Luminous moss', 'Goblin's gold' and 'Dragon's gold' are some of its common names, referring to its most interesting feature: its ability to appear to glow in the dark.This property is due to its persistent, emerald-green, shining protonema (Glime and Turzańska 2017;Casas et al. 2020).Specialised cells in the protomena have a single, large, lens-shaped vacuole, whose curved surface is capable of redirecting and focusing light towards chloroplasts.When light comes from a single direction, chloroplasts move to the most intensely lit spot on the inner cell side, generating the characteristic luminosity of the protonema (Glime and Turzańska 2017).This characteristic has promoted many 'fantastic' stories about the taxon (Crum 1973;Berqvist 1991;Glime and Turzańska 2017), and it is one of the few mosses declared as a 'Natural Monument' in Hokkaido, Japan (Kanda 1971;Iwatsuki 1977;Kanda 1988).Beyond this feature, its tiny distichous featherlike green leaves, which go quite unnoticed, are the other diagnostic character that makes S. pennata practically unmistakable.
Schistostega pennata requires slightly humid conditions, and some studies indicate that it usually occurs in areas with a clear oceanic influence (Reinoso Franco 1998;Garcia et al. 2008).Its level of specialisation leads to a certain sensitivity to threats such as soil destabilisation and log removal (Mežaka et al. 2011) and others that affect natural and artificial cavities, such as vandalism, mining and water extraction (Sérgio et al. 2012).Populations of S. pennata are often isolated and highly fragmented (Reinoso Franco et al. 1994;Reinoso Franco 1998;Harpel and Helliwell 2005;Mežaka et al. 2011) or have disappeared from their original natural or artificial habitats due to human pressure (Reinoso Franco 1998;Mežaka et al. 2011).This highlights the need for effective conservation measures to protect the species and its habitats.
Schistostega pennata is categorised as Vulnerable (VU) in the Spanish Red List (Garilleti and Albertos 2012) as well as in the Galicia and Euskadi regional lists (Reinoso Franco et al. 2002; Consejería de Medio Ambiente y Política Territorial (España) 2013), and as Near Threatened (NT) in Portugal (Sérgio et al. 2012).These categories indicate the importance of efforts to conserve the species.
The main goals of the present study were: (i) to update the knowledge of the ecology and distribution of Schistostega pennata on the Iberian Peninsula, and (ii) to elucidate the potential distribution range of S. pennata on the Iberian Peninsula.We further provide several remarks on the conservation of this species.

Study area
The study was carried out on the Iberian Peninsula (southwestern Europe) (Figure 1), a region mainly occupied by Spain and Portugal but also by Andorra, the British territory of Gibraltar, and the southernmost parts of the French Pyrenean Departments.The Iberian Peninsula includes two macroclimates: temperate (20.35% of total area of the Iberian Peninsula and Balearic Islands), defined by a lack of dry summer months, and Mediterranean (79.65%), with at least two dry months in the summer (Rivas-Martínez et al. 2011).A transitional area (known as submediterranean), which has a semiarid summer period, is located between these two bioclimatic regions (Sánchez de Dios et al. 2009, Rivas-Martínez et al. 2017).

Data collection
A compilation of records of Schistostega pennata on biodiversity data platforms (iNaturalist, http://www.inaturalist.org;Global Biodiversity Information Facility, http://www.gbif.org;Biodiversidade, http:// www.biodiversidade.eu;and Observation.org,http:// www.observation.org)was created, and only undoubtful records were retained in a database.Furthermore, an additional bibliographic search was carried out across scientific and grey literature.For all confirmed populations of Schistostega pennata, the following data were obtained where they were available from the information sources: observation month, substrate, elevation, habitat, and climatic conditions (mean annual temperature and annual precipitation from the nearest area weather station).Otherwise, climate data were extracted from https://es.climate-data.org/, and soil substrate from the lithological Iberian maps (Terán and Solé-Sabarís 1978).Distribution maps were generated with ArcMap 10.8 from ArcGis Desktop (ESRI, https://www.esri.com/en-us/arcgis).

Species distribution models
Modelling area.We calibrated the SDMs covering the full global geographical range of Schistostega pennata.The Holarctic biogeographical region was used to define the spatial extent (Barve et al. 2011;Acevedo et al. 2012;Mateo et al. 2015).The boundary of the Holarctic area was delimited using the definitions of Nearctic and Palearctic ecoregions provided by Olson et al. (2001).
Variables were built at a 30 second resolution using functions from the R package raster (Hijmans and Joe 2017).Predictor variables were selected using the function 'corSelect' in the R package fuzzySim (Barbosa 2015).Among each pair of variables correlated at a higher than 0.7 Pearson correlation coefficient (Guisan et al. 2017), the one with the best fit, according to the individual p value, AIC (Akaike's information criterion; Akaike 1973) and BIC (Bayesian information criterion: Schwarz 1978), was selected.A VIF (variance inflation factor: Marquardt 1970; Mansfield and Helms 1982) from R package HH (Heiberger 2018) was then calculated for each selected variable to ensure that multicollinearity was not present in further analyses (VIF score retained < 13.5).A total of seven predictor variables were retained after selection analyses for SDMs construction (Table 1).All analyses were performed in R version 4.2.2 (R Core Team 2022).
Occurrence data.Species distribution models were calibrated using data from different sources.For the distribution outside the Iberian Peninsula, we used a dataset of occurrences obtained at http://www.gbif.org.In addition, the data from http://arctoa.ru/ and Charissou (2014) were also considered as a complement.A list of all these records is included in the Supplemental Material 1.For the Iberian Peninsula, we considered all known and new information compiled in the present study (Appendix 1).For the model calibration, we used only those records with uncertainty in their coordinates less than the resolution of our predictor variables (i.e. 1 km).
Modelling settings.Maximum entropy niche-based modelling (Maxent) was used to predict the potential geographical distribution of Schistostega pennata (Phillips et al. 2004(Phillips et al. , 2006;;Phillips and Dudík 2008).Maxent software 3.4.0(Phillips et al. 2017) was run using the bootstrap strategy (Efron and Tibshirani 1997).Ten bootstrap replicates were chosen in which observations were resampled with replacement.In each repetition, Maxent randomly selected 70% of the occurrence points for training and 30% for validation (552 and 236 points, respectively).All other Maxent default settings for the calibration of the algorithm were maintained.The layer representing the mean probability of the 10 Maxent bootstrap logistic outputs was recreated with Quantum GIS 3.6.3-Noosa(QGIS Geographical Information System, https://www.qgis.org).
Evaluation.The area under receiver operating characteristic curve (hereafter AUC) was used as a thresholdindependent measure of discrimination (Fielding and Bell 1997).The closer the AUC is to 1, the better the model's predictive ability.Models reaching AUC ≥ 0.9 are classified as having an excellent performance (Swets 1988;Araújo et al. 2005).Percentage contribution and permutation importance of each variable to the SDM, and jackknife test results were observed to interpret the importance of the variables for predictive modelling.

Morphological description
Schistostega pennata is a short, dioicous moss with tiny, distichous green leaves that are decurrent and lack a nerve (Figure 2A, B, Figure 3A, B).The fertile stems support sporophytes with erect setae and capsules lacking peristomes (Figure 2D).It has a characteristic emerald-green glowing protonema, caused by reflecting light (Figure 3C, D) due to the large vacuoles, inside specialised cells, that act like lenses (Figure 2G).Protonemal cells develop different forms depending on environmental conditions, acquiring a rounded shape in lowlight conditions and an elongated shape in better illuminated situations (Figure 2H, I) (Edwards 1978).

Distribution of Schistostega pennata
A total of 73 records of Schistostega pennata on the Iberian Peninsula were found (Figure 4; Appendix 1).Of these, 47 (64.4%) were previously published records, 11 (15.1%) were from biodiversity data platforms, and 15 (20.5%) corresponded to new occurrences found during our field trips.Regarding geographical distribution, 34 (46.6%) were from Portugal, 33 (45.2%) from Spain and 6 (8.2%) from the French Pyrenees (see Figure 4).In addition, S. pennata is here reported for the first time from Ourense (Galicia, Spain) and Zamora (Castile and León, Spain).Additional data for several new locations are provided in Supplemental Material 2.

Ecology
A total of 68 (93.1%) records were from inside natural or artificial cavities, caves, mines, cracks or hollows.Only two (2.7%) were from outside such enclosed environments: on a shaded slope (Langreo, Asturias, Spain) and on the root system of a chestnut tree (Muñís, Navia de Suarna, Galicia, Spain).The remaining three records (4.1%) were from unspecified habitats.
All observations were of plants on siliceous substrates.Records of Schistostega pennata were made all year round.The mean (± standard deviation) annual temperature of all observation localities was 12.6 ± 1.5°C; the annual precipitation, 1240 ± 264 mm; and the elevation, 607 ± 370 m a.s.l.Data for all Iberian populations are provided in Appendix 1.

Distribution models
The model adequately discriminated between suitable and unsuitable areas for Schistostega pennata (average training AUC = 0.955; average AUC standard deviation = 0.002).According to variable percentage contributions and permutation importance together with Jackknife test results (Supplemental material 3 and 4), precipitation in the driest quarter (Bio17) was the most important variable for explaining the distribution of S. pennata.Soil pH and annual temperature range (Bio 7) also influenced the species range (Supplemental Material 3-5).
Generally, SDM displayed suitability values within the temperate-oceanic bioclimate area of the Iberian Peninsula, specifically in the northern half of Portugal, in the northern coastal areas of Spain, and in the Pyrenees Departments (Figure 5).Large continuous areas with conditions strongly suited for Schistostega pennata were mostly predicted in the Peneda-Gerês Natural Park and in the Serra da Estrela.Similar conditions occur close to the Serra de Xistral in Galicia and in the Peñas de Aya Natural Park between Euskadi and Navarra.Areas with high suitability values were also found far from currently known populations, such as in the Serra of Barbanza or close to the Serra de Outes (both in Galicia), and in the Sierra de Peña de Francia in Castille and Leon (see Figure 5).The SDM showed that the conditions in the central, southern and eastern Iberian Peninsula (mostly Mediterranean areas) seemed unsuitable for this species to thrive.

Discussion
The findings of the present study represent a significant advance in our knowledge of Schistostega pennata.The study is, to our knowledge, the first to investigate the potential worldwide distribution of the species in detail, and it also represents the most exhaustive floristic study of S. pennata on the Iberian Peninsula.In fact, the number of detected Iberian populations has increased by almost 55%, a considerable value for an apparently patchily distributed moss species.2022) identified a number of important issues as directions for the future of bryological research, one of these being the need to improve our knowledge of biogeographical patterns and processes.Traditionally, Schistostega pennata has been represented as a boreal and temperate moss (Reinoso Franco 1998) with a Holarctic distribution  (Ignatov and Ignatova 2001;Ignatov et al. 2017).However, certain localities, both new and already known, belong to regions with a slight Mediterranean influence (Casas de Puig 1978;Reinoso Franco 1998).This distribution, coupled with its occurrence in Iraq (Aziz 2011), suggests that S. pennata is able survive in a Mediterranean-type climate.In fact, our SDM predicted potential new range within this macroclimatic zone, suggesting that its distribution across the Mediterranean basin could have been underestimated.

Patiño et al. (
A possible explanation for the presence of Schistostega pennata in regions characterised by a Mediterranean climate may be its presence in microhabitats that provide the appropriate microecological conditions for persistence, as suggested for other moss species (Fritz and Heilmann-Clausen 2010;Hespanhol et al. 2011;Kraichak 2014;Goia and Gafta 2019;Táborská et al. 2020;Ren et al. 2021).Thus, variables operating at microscale may have more importance on the distribution of S. pennata than expected, although they are difficult to model.
Our findings are consistent with Schistostega pennata having a preference for acidic substrates, growing mainly on siliceous rocks such as granite, sandstone or gneiss (Lye 1972;Crum and Anderson 1981;Hill et al. 1994;Reinoso Franco 1998;Smith 2004).The importance of soil properties can be observed in northern and central localities where the species occurs on reduced granitic siliceous 'islands' or large terrains (Allorge 1935(Allorge , 1955;;Casas 1993a;Charissou 2014) between calcareous areas (Terán and Solé-Sabarís 1978).In fact, the SDMs captured the acidic soil pH preferences of S. pennata as an important explanatory variable, and although we cannot rule out the possibility of populations occurring in calcareous areas, they should be searched for on siliceous outcrops.
Our SDMs present an approach sometimes used to predict moss distributions (see, e.g., Sérgio et al. 2007;Kruijer et al. 2010;Patiño et al. 2016;Spitale and Mair 2017).It is important to note that the SDM must be considered only as a first step towards further elucidating the potential distribution of Schistostega pennata on the Iberian Peninsula, and new approximations will need to be carried out using other variables (e.g.topography, substrate, geology, humidity, insolation, orientation).Our SDM predictions can be used to optimise resources in the search for new populations of S. pennata, as has been successfully achieved for other species (Edvardsen et al. 2011;Särkinen et al. 2013;Alfaro-Saiz et al. 2015;Fois et al. 2018).

Conservation remarks
Schistostega pennata is considered a rare moss, classified as 'Vulnerable' or 'Endangered' and having some form of statutory protection in several countries (Kalda et al. 1992;Harpel and Helliwell 2005;Hespanhol et al. 2005;Maslovsky 2005;Mežaka et al. 2011;Mišíková et al. 2020).This matches the status of the species in several regions of the Iberian Peninsula (Hespanhol et al. 2005;Garilleti and Albertos 2012;Sérgio et al. 2012).It has been shown that most records occur in natural or artificial caves which are especially sensitive to the transit of people and vehicles (Sérgio et al. 2012).This can be an important threat factor increasing the risk of extinction (Mežaka et al. 2011) and should be taken into account for the Iberian Peninsula.Closing or limiting access to these caves, and thereby minimizing threats such as vandalism, fires, mine exploitation, pollution from vehicles, trampling and water extraction, could be a first step towards preventing further decline of S. pennata (Sérgio et al. 2012).Furthermore, this species, given its apparent ecological requirements, might be affected by the higher temperatures and lower precipitation (He et al. 2016)  Finally, little information is available about demographic trends for Schistostega pennata on the Iberian Peninsula, despite its threat status (Garilleti and Albertos 2012).The situation of the species before the 1900s is unknown, and information from the twentieth century is scarce.Therefore, it is necessary to start monitoring the known populations and studying their demographic features, including dispersal, genetic diversity, and recent changes in effective population size.Complementing that is an urgent need to exhaustively search for new populations, a task for which citizen science platforms could be very useful.In this study, half of the new populations were discovered by contributors to citizen science platforms, which reflects the importance of these new resources.Several fundamental questions are key to the advancement of bryology in the coming decades (Patiño et al. 2022) and identifying the best avenues for studying them is the basis for continuing progress in an increasingly changing world.In fact, although all species are potentially impacted by anthropogenic disturbances, rare and threated species may be the first to disappear locally or to become globally extinct.Mikel Artazkoz is a naturalist and biologist interested in plant and animal ecology.He has worked on plant diversity and conservation in the Cantabrian range, as well as on animalanimal interactions and conservation biology in Mediterranean ecosystems.He is working as a curator of the herbarium of the University of Córdoba.
Finally, several herbaria (University of Oviedo, FCO; Atlantic Botanic Gardens of Gijon, JBAG; University of León, LEB; University Pablo de Olavide, UPOS; University of Sevilla, SEV; and University of Córdoba, COFC) were consulted to find additional occurrences.Only records collected up to 31 December 2022 were included.Additional field sampling was carried out in Galicia, Asturias, Navarra and Euskadi (northern autonomous regions in Spain) between 2019 and 2022 in search of new populations of S. pennata.

Figure 1 .
Figure 1.Distribution of Schistostega pennata around the world.Countries with confirmed presence of the luminous moss (solid grey), countries with no records (white), and the study area, the Iberian Peninsula (red).
recorded in northern Iberian regions due to climatic change (Esteban-Parra et al. 2003; India et al. 2007; Lorenzo and Alvarez 2020).

Figure 5 .
Figure 5. Map of the predictions of Maxent species distribution models (SDMs) of Schistostega pennata (Maxent logistic output recreation).(A) Complete worldwide range including a selection of accurate locations.(B) Iberian Peninsula study area; cold colours indicate areas with predicted low suitable conditions for S. pennata, and warmer colours indicate higher suitable predicted conditions.

Table 1 .
Predictor variables used in the Maxent modelling of the distribution of Schistostega pennata.
a Bio 5 corresponds to the maximum temperature of the warmest month and Bio 6 to the minimum temperature of the coldest month.