Geodiversity and biodiversity on a volcanic island: The role of scattered phonolites for plant diversity and reproductive fitness

Oceanic islands are cradles of endemism, contributing substantially to global biodiversity. A similarity in magmatic origin translates into high global comparability of substrates of volcanic islands on the oceanic crust. In some places, the petrography of magmatic rocks is differentiated chemically or physically. Phonolites are examples of rare 20 localities with intermediate chemical traits between felsic and mafic and with diverse textures. They contribute to habitat heterogeneity and offer specific growth conditions in a significantly different matrix of basaltic substrates. The explicit contribution of geodiversity to island biodiversity has been little studied, despite accumulating evidence of its importance on continents. On the island of La Palma, Canary Islands, isolated phonolitic rocks are conspicuous by their light colour and specific shape. Although these outcrops only cover small areas, their unique form and composition increase within-island 25 geodiversity. To investigate how this affects biodiversity on La Palma, we sampled all vascular plant species in 120 plots on four sets of paired sites, comparing phonolitic with basaltic rocks, to test whether phonolitic rocks increase overall plant species richness and promote the reproductive fitness of plants. We recorded species number and abundance, individual plant height, and diameter as proxies for aboveground resource allocation and tested for differences in vegetation cover and species composition between the bedrock types. We found higher species richness and abundance on phonolites than 30 neighbouring basaltic substrates, and individuals of the same species were larger (in height and diameter) on phonolites compared to neighbouring basalt. An endemic woody species with two distinct varieties even appears almost exclusively on the small surfaces of phonolitic rock. Despite extremely limited spatial extent, phonolitic rocks can play an important role for plant biodiversity on islands. https://doi.org/10.5194/bg-2021-107 Preprint. Discussion started: 5 May 2021 c © Author(s) 2021. CC BY 4.0 License.


Introduction 35
Biodiversity is known to depend mainly on abiotic drivers, such as climate and topography (Field et al., 2009). However, the importance and explicit impacts of geodiversity on biodiversity have long been under-researched; recently, they have started to receive more attention (Gray, 2004;Lawler et al., 2015;Bailey et al., 2017;Alahuhta, 2020;Barajas-Barbosa et al., 2020).
Geodiversity is in many respects an abiotic equivalent to biodiversity (Gray, 2011) and represents the variability of geological elements (i.e., composition), surface structure, edaphic and hydrological features (Gray, 2004;Bailey et al., 40 2017). This variability contributes to habitat diversity and thus affects biodiversity patterns via the availability of ecological niches (Liu et al., 2013;Gillespie & Roderick, 2014;Bailey et al., 2017). Geological elements provide unique or distinctive habitats for plants and insects, deliver initial growth conditions for vegetation or fungi formation, and are part of nutrient cycling and soil-atmosphere interactions (Tukiainen et al., 2016).

45
Biodiversity is distributed unevenly throughout the world (Gaston, 2000), and oceanic islands provide much of it due to their high endemic richness (Kier et al., 2009). Substrates that differ in geochemistry and petrography are likely to be relevant for biodiversity on oceanic islands, where most rocks commonly share similar volcanic genesis, resulting in only slight differences in the parent material. Distinct substrates with limited extent, such as individual rock types, may function as a second isolating abiotic filter for populations in addition to the spatial isolation of oceanic islands that are known to be of 50 outstanding importance for speciation at the global scale (Kier et al. 2009). Specific rock habitats, particularly rocks that exhibit petrographic and geochemical substrates such as serpentinites, are known to be rich in habitat-specific endemics (e.g., Harrison et al., 2006;Kazakou et al., 2010). These species evolved specific adaptations to the unique nutrient contents and soil conditions and the presence of heavy metals that cannot be tolerated by other plant species (Harrison & Rajakaruna, 2011). This phenomenon is also known from continental areas, where substrates such as serpentinite and gypsum outcrops 55 host specialised floras and contribute to broad-scale diversity (see, e.g., Chiarucci et al., 1998;Pausas et al. 2003). Highly restricted ranges and small population sizes of insular endemic species make island biotas particularly vulnerable to extinction (Paulay, 1994) and underline the relevance of understanding the importance of geodiversity for insular biodiversity.

60
Phonolites are rocks that occur at volcanic intraplate settings in insular and continental contexts worldwide (Garcia et al., 1986;Ackerman et al., 2015;Hagos et al., 2017). They exist in a variety of forms in the context of volcanic activity. Such outcrops mainly exist on continents, where they are often linked to faults and tectonic activity. Major components of these extrusive igneous rocks (formed from lava with low silica content) are alkaline feldspars together with foid minerals, nepheline, and pyroxene (Ackerman et al., 2015;Abratis et al., 2015) or their conversion products. Distinctive to phonolites 65 is fine-to-medium grain size. On the island of La Palma (Canary Islands, Spain), several phonolitic rock outcrops are embedded into a basaltic matrix (Middlemost, 1972). The dominant rock type found on La Palma is olivine and augite-titanaugite porphyric basalt, resulting from rapidly rising magma from the upper mantle (Middlemost, 1970). In contrast, there were times when a sizeable 70 magmatic chamber below the island enabled differentiation of magma, removal of silica and thus yielded ultramafic, trachytic, and phonolitic rocks (Middlemost, 1970). Phonolite trachytes (showing the exhalation of gases during eruptions) occur on various volcanic islands such as La Palma and St. Helena. On La Palma their distribution is focused on the southern (young) part of the island. The major chemistry of phonolites on La Palma is comparable to that of "average phonolites", as described by Nockolds (1954). 75 Volcanic activity with the production of tephra and lava flows is a noticeably young phenomenon in the southern part of La Palma, with eruptions as recent as 1971. All recorded eruptions in this part took place in historic times. Thus, the remnant phonolitic rocks are the tips of a former land surface that are today embedded in a basaltic matrix of noticeably immature age (Garantje et al., 1998). In consequence, weathering processes on phonolites were active on longer time scales compared to 80 the surrounding rocks. In addition to petrography, differences in weathering and resulting nutrient availability refer to different time scales of exposure. Higher nutrient availability enables higher plant abundances and larger plant size. Porder et al. (2004) found comparable conditions at a catena of different rock ages in the Hawaiian Islands.
Compared to basaltic lava outcrops, phonolites differ in their chemical composition and, additionally, in colour, texture, 85 density, weathering, and formation fracturing (von Fragstein et al., 1988). We observed tafoni-weathering (Formoso et al., 1989) on some phonolitic surfaces, indicating temperature and moisture gradients between the surface and the solid body of rocks (Brandmeier et al., 2011) combined with wind exposure. Circulating leachate reaches the rock's surface and evaporates, exposing its dissolved mineral content and enabling the development of secondary mineral assemblages (Spürgin et al., 2019) that can contribute to plant nutrient supply (Faccini et al., 2015, showed the effectiveness of ground phonolite 90 rock powder as fertilizer). Even if quantitatively small, such processes are of particular importance at nutrient-poor sites. In contrast, the young basalts in the southern part of La Palma are barely weathered (Carracedo et al., 2001), appearing rough and friable with sharp spikes. We expect these petrographic and geochemical differences of parent material to affect vegetation cover and species occurrences.

95
La Palma hosts 149 vascular plant species that are endemic to the archipelago and 49 single island endemics (hereafter SIEs). The hyper-endemic plant species Cheirolophus junonianus, comprising its var. junonianus and var. isoplexiphyllus, (Vitales et al., 2014a;2014b) occurs within a range of only 3500 m², solely on La Palma (Bañares et al., 2004). Within this small range, individuals of this species occur only on a few outcrops, which are almost exclusively phonolitic rocks with a chemical composition different from most of the surrounding substrates. Therefore, the species is very restricted in its range 100 size to just a few small locations (Muer et al., 2016;Atlantis, 2021) and appears to be restricted to phonolites (Fig. 1a).

105
Geologic outcrops, such as phonolites, increase microenvironmental heterogeneity, enhancing species richness at a landscape scale (Hjort et al., 2015). Increased speciation rates on isolated outcrops of scarce rocks are thought to lead to a higher percentage of endemic species than the surrounding matrix (Ricketts, 2001). Geodiversity may thus promote both species richness and endemism. However, relatively little is known about the extent to which phonolites promote species diversity in general and particularly endemism. We aim to address this knowledge gap by investigating plant species richness and 110 abundance on phonolites compared to surrounding basaltic lavas.
Phonolitic outcrops act as small habitat islands within a basaltic matrix (Fig. 1b). The established 'species-area relationship' (SAR) and the 'species-isolation relationship' (SIR) (MacArthur & Wilson, 1967;Rosenzweig, 1995;Giladi et al., 2014) predict a smaller number of species on these small and isolated phonolitic rocks in comparison with basaltic rock outcrops in 115 their surroundings. (From the beginning (MacArthur & Wilson, 1967), such concepts were not only meant for real islands but instead took 'islands' as examples for isolated habitats (or habitat islands) within a terrestrial landscape matrix.) However, the expected higher availability of nutrients would give such habitats more favourable conditions for plant growth.
Besides, it is by no means certain that the phonolitic rocks were permanently separated from each other in southern La Palma's geological evolution. Possibly, a historically much larger phonolitic rock is today largely buried by basaltic 120 eruptions (Garantje et al., 1998). Thus, a few phonolite outcrops may serve as refugia for remnant populations (Eriksson, 1996) of species specialised to phonolitic rocks.
We investigated the occurrences and traits of plant species in a comparative study matching basaltic and phonolitic rock formations on La Palma of comparable size, shape and extent to test the following hypotheses: 125 https://doi.org/10.5194/bg-2021-107 Preprint. Discussion started: 5 May 2021 c Author(s) 2021. CC BY 4.0 License.
i. Species diversity: We expect plant species richness on phonolitic rocks to be higher than on basaltic rocks because phonolites offer more favourable plant growth conditions.
ii. Species reproductive fitness: Plant species populations on phonolites show a larger abundance of individuals that 130 are taller and have greater canopy diameter than neighbouring basalts due to their advantages in resource availability and porosity. Plant height and canopy diameter were used as proxies for aboveground resource allocation and ultimately as a surrogate for fitness.
iii. Island endemism: We expect phonolitic rocks to host more endemic plant species than basaltic rocks because of 135 their high degree of spatial isolation, in combination with the older age of the phonolitic bedrock than the basaltic matrix.

Study site and data sampling
We sampled four phonolitic and four adjacent basaltic rocks in the southern part of La Palma in spring 2018 (Walentowitz et 140 al.,submitted;Fig. 2). Locations were identified in the field based on Middlemost (1972). The sampled phonolitic rocks represented most of the overall extent of this habitat on the island, covering a large gradient of microclimate, aspect, and inclination. We chose comparable neighbouring pairs of phonolite and basalt consisting of one cohesive rock formation each.
Outcrop pairs were chosen to match the size and microclimatic conditions (aspect, slope). For each selected phonolitic and basaltic rock, we recorded plant species composition and abiotic parameters within 15 plots of 2 m x 2 m that were randomly 145 selected within the range of accessibility on the phonolite and basalt. This resulted in a total of 120 plots sampled across the four pairs of phonolitic and basaltic rocks (60 plots on phonolite and 60 plots on basalt in total).  Within each plot, we recorded coordinates, aspect (northerness and easterness), and slope inclination. Then, we estimated rock surface rugosity using thread transects spanning the two plot diagonals: we measured the transect length along the 3Drock surface. Larger values of rugosity indicate higher levels of microtopography (cracks, hollows, uneven slope), while low 155 values indicate smooth, even surfaces.
All vascular plant species within each plot, including ferns, were identified following the taxonomy of "Plants of the World online" (POWO, 2019). The biogeographic status of each species (single-island endemic, multi-island endemic, nonendemic native and introduced) is based on Muer et al. (2016) (see extensive plant list in Appendix A1). The number of 160 individuals per species and plot was counted. Plant height (length from base of the stem to the tip) and canopy diameter (widest part of the plant parallel to the ground) of all single individuals found were measured as traits. Height, diameter, and species abundances were measured for species to ensure that vegetational differences evolved through long-term processes and did not reflect the short-term variability of environmental conditions. We are aware that numbers of flowers and seeds might be more accurate to measure reproductive fitness when monitoring plants over the course of an entire reproductive 165 https://doi.org/10.5194/bg-2021-107 Preprint. Discussion started: 5 May 2021 c Author(s) 2021. CC BY 4.0 License.
cycle, but we chose plant height and width as proxies as these can be measured within a short time frame. Lichen cover, which is abundant on the basalt, was estimated as the percent cover of each plot. Moss cover was negligible in all the plots.

Statistical analysis
Differences in total plant species number and the number of single and multi-island endemics were analysed using Pearson's Chi-squared tests. Percentages of abundance, plant height, diameter, and SIE percentage between plots on phonolites and 170 basalt were analysed using Mann-Whitney U tests. We conducted detrended correspondence analysis (DCA) to investigate the multidimensional aspects of vegetation composition and identify potential fundamental drivers (Appendix A3). We tested for differences in aspect, inclination, rock surface rugosity and lichen cover between phonolite and basalt using Mann-Whitney U tests.

Results 175
We recorded 68 species of vascular plants (pteridophtyes and spermatophtyes) overall. Of these species, nine were Single Island Endemics (SIE) restricted to the island of La Palma, 16 were Multi Island Endemics (MIE) co-occurring also on other islands in the archipelago, 39 were non-endemic natives, and 4 were non-natives. The SIE Cheirolophus junonianus was only found on phonolite, and most individuals of var. junonianus occurred on one isolated outcrop (Roque Teneguía) and individuals of var. isoplexiphyllus on another one (Escarpa del Volcán Teneguía, Fig. 2c). 180 We found higher plant species richness on phonolitic rocks. While 22 species were encountered on both phonolite and basalt, only eleven species were restricted to basalt, but 34 were recorded only on phonolite (Appendix A1). Endemism groupings showed similar patterns (SIEsphonolite: 9, basalt: 5; MIEsphonolite: 15, basalt: 6). Besides the total number of plant species per rock type, we also found higher species richness on phonolite at the plot scale (p = 0.0164, Fig. 3 a), and higher 185 diversity of SIEs (p = 0.00151, Fig. 3 b) and MIEs (p = 0.00727, Fig. 3 d).

195
On phonolitic rocks, we did not find higher total plant abundance (p = 0.169, Fig. 4 a). Moreover, there was no significant difference in abundance when only considering the 23 species found in plots on both substrates (p = 0.179, Fig. 4 b).  For plant species recorded on both rock types, individuals were on average taller and had wider canopies (Fig. 4c-d) on phonolitic than on basaltic rocks (Fig. 4c-d). 205 Plant cover only (excluding lichens) was also significantly higher (p < 0.0001, Appendix A2a) on phonolites than on basalt.
Lichen cover did not show a significant difference (p = 0548, Appendix A2b). The ordination did not show any difference in the species composition, indicating no aspects of beta-diversity at all 210 (Appendix A3). Topographic characteristics of basalt and phonolite plots showed no differences in surface rugosity, aspect (northerness, easterness), or differences in slope inclination (Appendix A4).

Discussion
Phonolites host distinct vegetation compared to equivalent areas of neighbouring basaltic rocks. The vegetation on phonolitic rocks is characterized by higher species richness and higher reproductive fitness of plant individuals and higher total plant 215 cover.
Phonolite rocks on La Palma and other oceanic islands are arranged as habitat islands in a basaltic matrix (Fig. 1b). On La Palma, the total area of phonolite outcrop is tiny compared with the area of the basaltic matrix. Thus, encountering more species on phonolites than on basalt aligns with our hypotheses but defies the area effect on species richness, which predicts 220 species number to be lower in the phonolite habitat. This makes the greater number on phonolite even more remarkable.
However, our findings are congruent with studies which did not find a species-area relationship or relationships with a less steep slope on habitat islands (Matthews et al., 2016;Deák et al., 2018). Reasons might be unrealised niches due to unsaturated evolutionary dynamics in this young and isolated system. Environmental filters (Liu et al., 2020) enhancing growth conditions on phonolite outcrops may also exist, consistent with our findings that the plants are larger on phonolites. 225 Several mutually non-exclusive drivers may explain this phenomenon: 1) phonolite-specific surface texture, 2) colour, 3) chemistry, and 4) age. 1) We observed deeper fractures in phonolitic rocks than in other volcanites on La Palma Island. Besides, phonolitic rocks show a much smoother surface roughness than their surrounding matrix. Basaltic rocks seem to possess a more dynamic 230 relief, mainly attributed to their origin in congealed lava flows, typically found on oceanic islands. When testing rock surface rugosity, there were no significant differences between phonolitic and basaltic rocks. Hence, we argue that surface texture does not play a role for higher plant growth response, richness and abundance observed on phonolites.
2) Most phonolitic rocks in a basaltic matrix are visually brighter than basalt. We expected that phonolites hold a higher 235 albedo than surrounding rocks and therefore expected them to have a reduced surface temperature compared to volcanic outcrops with darker colouring, such as basalt. However, in an experiment with differently coloured bricks Hall et al. (2005) showed that the albedo of white surfaces only leads to significantly lower temperature of the material when the surface temperature falls below air temperature. With monthly temperatures between 17 and 25° C within large areas of oceanic islands (Harter et al., 2015), no major temperature differences between basalt and phonolite surfaces can be expected. We 240 therefore consider that this effect has no major impact on plants' habitat suitability. 3) Grain size and the rather porphyric arrangement of phonolites imply ameliorating plant growing conditions on this rock type. The chemical composition of phonolites has been described as nutrient-rich as it has traditionally been ground up and used as an inorganic fertilizer (von Wilpert & Lukes, 1998;Ramos et al., 2006;Schoen et al., 2016). Phonolites consist of 245 the potassium-rich nepheline, which dissolves much faster than other potassium sources (Manning, 2009). This may explain increased plant growth on phonolites. Weathering processes may act as driving forces for the establishment and germination of plants as they increase plant-available nutrients on these outcrops. Since ground phonolitic rock fertilizer is used to provide crops and trees with potassium (von Wilpert & Lukes, 1998;Schoen et al., 2016), plants growing on phonolites might profit from this rock characteristic. Basaltic rock powder has also been used as fertilizer in rare cases, but according to 250 Manning (2009) this is mainly due to geologically unclear assignments (potassium-rich trachyte is often assigned to basalt).
Furthermore, he points out that basalt is another but smaller source of potassium than phonolite.
4) The age of geological formations influences plant diversity and species composition (Whittaker et al., 2008;Hulshof & Spasojevic, 2020). As noted in previous studies (Carracedo et al., 2001), the Cumbre Vieja rift on La Palma has evolved 255 throughout several eruptions and therefore contains lava formations from different ages as well as slightly different mineralogical compositions. The known phonolite rocks on La Palma are located in the geologically young southern part of the island. Consequently, we can assume that they are all in a stage of high ecological opportunities resulting in unsaturated niches (Whittaker et al. 2008). An interplay between erosion-driven uncovering of lava-covered phonolite rocks and new lava flows may strongly influence the vegetation on those rocks. We observed partly buried phonolites on which the survival 260 of plants or seedlings during volcanic events was improbable (Garantje et al., 1998). Carracedo et al. (2001) showed that the last phonolite formation occurred in 1585, while there have been basaltic outbreaks until 1977. Thus, age differences between phonolitic and basaltic outcrops might influence plant composition and performance. A geological map provided by Carracedo et al. (2001) shows that most phonolitic rocks surrounded by lava flows are much older than 1585.

265
The differential geology itself results in a specialized flora of those habitat islands (Kruckerberg, 1991). Even though the hyper-endemic Cheirolophus junonianus with its two varieties var. junonianus and var. isoplexiphyllus is confined to phonolites, the percentage of single island endemics on phonolites was not significantly higher, refuting our expectations, though the numbers of endemic species were significantly higher. Lessons learned from other outcrops (Kruckerberg, 1991) cannot be applied to phonolitic rocks on La Palma, and the functioning of phonolites as islands of speciation within a sea of 270 basalt does not seem to apply. However, as most individuals of the typical variety of Cheirolophus junonianus occur on one isolated outcrop and individuals of var. isoplexiphyllus on another one (personal observation), a very local allopatric speciation by adaptive radiation or an ongoing genetic drift could be the underlying cause (Vitales et al., 2014a;2014b). One possible explanation for this singularity presumes Cheirolophus junonianus belongs to a relict population of plants that were once widely distributed on phonolitic rock before these were covered mainly in a basaltic matrix. Species composition might 275 https://doi.org/10.5194/bg-2021-107 Preprint. Discussion started: 5 May 2021 c Author(s) 2021. CC BY 4.0 License. be influenced by age differences of underlying rock material as species like Cheirolophus junonianus and its var. junonianus and var. isoplexiphyllus can solely be encountered on phonolitic rocks (Muer et al., 2016). It is well known that habitat diversity on islands leads to higher species richness (Hortal et al., 2009). Phonolites offer an added habitat for plants to grow on and contribute to plant species richness and endemism on La Palma (Irl et al., 2015).

280
When looking at the role of geologic promotion of biodiversity, there are several studies addressing serpentinite rocks and serpentine soils. On those rocks and soils, considerable amounts of toxic heavy metals, low amounts of essential nutrients (N, K, and P), and a low calcium-to-magnesium ratio relate to high rates of endemism and a specialized flora (Chiarucci, 2003;Harrison et al., 2006). These serpentinite-tolerating species are restricted to a harsh environment by dominant competitors in a less harsh matrix. On phonolitic rocks, more favourable growing conditions resulting in higher plant 285 richness and abundances seem to prevail. Compared to serpentinite rock studies, the surrounding matrix is built up by the potentially harsher basaltic rocks resulting in lower plant growth response.
The geological history of basaltic rocks depends on a series of different volcanic eruptions, assuming a more considerable chemical variability, whereas phonolites seem to build upon one or just a few volcanic events (Carracedo et al., 2001). The 290 long chronology of volcanic eruptions on the Canary Islands and their ancestors reveals a high likeliness that most of the eruptions resulted in basaltic, only a few in phonolitic formations. Plants growing on basalt experienced larger environmental gradients since basalt is omnipresent on the islands. Contrasting, plants growing only on phonolites did not experience larger environmental gradients. In accordance, we observed plants on basaltic rocks to be more generalist than plants on phonolitic rocks. 295 We are not aware of other studies conducted in locations where phonolites can be encountered that explore their potential role as exceptional plant habitat islands, even though phonolites can be found all over the world (Garcia et al., 1986;Ackerman et al., 2015;Hagos et al., 2017). Therefore, further investigation is needed to figure out whether the patterns encountered on La Palma may also be found on comparable phonolitic rocks in other areas of the world. Their benefits for 300 biodiversity found in this study need to be recognized and valued. Especially for isolated areas such as islands, phonolites can contribute to small-scale biodiversity hotspots and our findings suggest that they should be conserved.

Conclusion
Phonolites provide unique habitat conditions on oceanic islands and exhibit distinct vegetation compared with surrounding areas. Higher species numbers and abundances as well as higher plant growth responses underline the importance of these 305 rocks for the vegetation on these islands which are globally dispersed. Despite the small area covered by phonolites they play a significant role in enhancing plant biodiversity. Our results contribute to a better understanding of the distribution and plant

Code availability
Only standard tests and plotting commands in R were used for data analysis. The code is available on request from the corresponding author. 335 https://doi.org/10.5194/bg-2021-107 Preprint. Discussion started: 5 May 2021 c Author(s) 2021. CC BY 4.0 License.

Data availability
Any data supporting the findings of this study are available within the supplementary materials of this article and were taken from Walentowitz et al. [submitted].