Sarcocornia pillansii[i] (Moss) AJ Scott

31 Pages·2017·1.7 MB·English

The influence of Pleistocene dynamics on the South African
salt marsh species Sarcocornia pillansii (Moss) A. J. Scott
(Amaranthaceae): Inferences from phylogeography and
species distribution modelling
Dimitri A Veldkornet Corresp., 1 , Anusha Rajkaran 1 , Janine B Adams 2
1 Department of Biodiversity and Conservation Biology, University of the Western Cape, Cape Town, South Africa
2 Botany, Nelson Mandela University, Port Elizabeth, South Africa
Corresponding Author: Dimitri A Veldkornet
Email address: [email protected]
Glacial–interglacial climate oscillations during the Pleistocene played a significant role in
shifting species distributions. During this period (26 500 - 19 000 years ago) the sea level
was 120 m lower than it is currently with large areas of the Southern African continental
shelf being exposed. This formed a barrier to cold-water dispersal of various aquatic
organisms between the west and east coast. This study explores the influence of past
climatic conditions on the salt marsh species Sarcocornia pillansii (Moss) A. J. Scott using
species distribution modelling and multi-locus phylogeography. The area under curve
(AUC) values were considered ‘good’ (> 0.80), indicating that the models had high
specificity and sensitivity. The AUC was greater for the Maxent model (AUC = 0.881)
compared to the Bioclim model (AUC = 0.837) under current conditions. Climate
simulation of the Last Glacial Maximum (LGM) indicated greatest habitat suitability in
estuaries along the west (Orange River Estuary to Langebaan) and east (Algoa Bay to
Keiskamma) coast of South Africa. This pattern is reflected in the phylogeographic analysis
where a greater number of haplotypes were found in estuaries west and east of the
greater continental shelf. The nuclear DNA dataset that included 97 sequences eight
ribotypes whereas the chloroplast DNA for 94 sequences that were resolved into four
haplotypes. The results suggest that species survived in these estuaries (as refugia) during
Pleistocene climate cycles. Post-LGM increases in sea level along the south coast allowed
confluence between isolated river systems, offering opportunities for dispersal among
populations.
PeerJ Preprints | https://doi.org/10.7287/peerj.preprints.26927v1 | CC BY 4.0 Open Access | rec: 12 May 2018, publ: 12 May 2018
1 The influence of Pleistocene dynamics on the South African salt marsh species
2 Sarcocornia pillansii (Moss) A. J. Scott (Amaranthaceae): Inferences from
3 phylogeography and species distribution modelling.
4 Veldkornet DA*1, Rajkaran A1, Adams JB2
5 *[email protected]
6 1Department of Biodiversity and Conservation Biology, University of the Western Cape, Bellville, 7535
7 2Department of Botany, Nelson Mandela Metropolitan University, PO Box 77000, Port Elizabeth, 6031
8
9
10
11 Corresponding Author:
12 Dimitri Veldkornet 1
13 email address: [email protected]
PeerJ Preprints | https://doi.org/10.7287/peerj.preprints.26927v1 | CC BY 4.0 Open Access | rec: 12 May 2018, publ: 12 May 2018
14 Abstract
15
16 Glacial–interglacial climate oscillations during the Pleistocene played a significant role in shifting species
17 distributions. During this period (26 500 - 19 000 years ago) the sea level was 120 m lower than it is
18 currently with large areas of the Southern African continental shelf being exposed. This formed a barrier
19 to cold-water dispersal of various aquatic organisms between the west and east coast. This study explores
20 the influence of past climatic conditions on the salt marsh species Sarcocornia pillansii (Moss) A. J. Scott
21 using species distribution modelling and multi-locus phylogeography. The area under curve (AUC) values
22 were considered ‘good’ (> 0.80), indicating that the models had high specificity and sensitivity. The the
23 Maxent model (AUC = 0.881) performed better compared to the Bioclim model (AUC = 0.837) under
24 current conditions. Climate simulation of the Last Glacial Maximum (LGM) indicated greatest habitat
25 suitability in estuaries along the west (Orange River Estuary to Langebaan) and east (Algoa Bay to
26 Keiskamma) coast of South Africa. This pattern is reflected in the phylogeographic analysis where a greater
27 number of haplotypes were found in estuaries west and east of the greater continental shelf. The nuclear
28 DNA dataset that included 97 sequences eight ribotypes whereas the chloroplast DNA for 94 sequences
29 that were resolved into four haplotypes. The results suggest that species survived in these estuaries (as
30 refugia) during Pleistocene climate cycles. Post-LGM increases in sea level along the south coast resulted
31 in the meeting between rivers, offering opportunities for dispersal among populations.
32 Key words: cold-water dispersal; habitat suitability; sea level; ancestral haplotypes, population
33 bottlenecks
34 Introduction
35 Phylogeography was first introduced by Avise et al. (1987) to describe the phylogenetic structure of
36 genealogical lineages within a species across the geographical landscape. Phylogeography provides insight
37 into the current patterns of evolutionary history of subdivisions within species and species complexes and
38 provide new insights into the relationship between earth history and biotic diversification (Hedenas,
39 2011). It can also help in conservation of evolutionary significant units (Chung et al., 2014) where species
40 with low haplotype diversity and population extant can be considered as more vulnerable. Historical
41 considerations about the distribution of genetic diversity have also contributed to our understanding of
42 invasion biology and can be used to address issues surrounding cryptic biodiversity (Beheregaray, 2008).
43 The possible scenarios that are responsible for the current phylogeographical patterns in species are often
44 difficult to disentangle (Avise, 2000; Potts et al., 2013). Along the Southern African coastline two major
45 forces has been suggested to have influenced the phylogeography of most taxa: marine biogeography and
PeerJ Preprints | https://doi.org/10.7287/peerj.preprints.26927v1 | CC BY 4.0 Open Access | rec: 12 May 2018, publ: 12 May 2018
46 climate oscillations during the Pleistocene (Teske et al., 2011). Teske et al. (2011) reviewed the current
47 knowledge on South African coastal phylogeography and concluded that most coastal taxa are divided
48 into regionally confined genetic lineages whose distributions are linked to southern Africa’s three marine
49 biogeographic provinces. On the southwest coast, phylogeographic breaks that coincide with the
50 biogeographic disjunction between cool temperate and warm-temperate biota have been reported near
51 Cape Town (e.g. Teske et al., 2006 the northernmost breaks in this region have been identified on the
52 central southeast coast and the southernmost breaks were reported near Port Elizabeth (Mmonwa et al.,
53 2015). The third area coincides approximately with the transition zone between subtropical and tropical
54 biotas on the east coast (e.g. Teske et al., 2007) (See Fig. 1).
55 The influence of biogeographic boundaries have been well documented in other taxa along the South
56 African coastline, where phylogeographic patterns in coastal marine organisms reflect historic properties,
57 driven by dispersal abilities (Reynolds et al., 2014; Toms et al., 2014). Along the South African coastline
58 there is little evidence suggesting major geological vicariances that could have driven species divergence.
59 However, climate oscillations during the Pleistocene had a major effect on the coastal morphology. During
60 the Last Glacial Maximum (26 500 – 19 000 years ago) the sea level was 120 m lower than it currently is.
61 This allowed for large areas of continental shelf to be exposed, particularly south of Cape Agulhas (Swartz
62 et al., 2009; Carr et al., 2010; Fisher et al., 2010; Teske et al., 2011). The exposure of the Agulhas Bank
63 resulted in the southern tip of Africa being about 200 km south from current and colder water
64 temperatures in the region during that time may have presented a cold-water dispersal barrier (Teske et
65 al., 2011).
66 Recently Potts et al., (2016) explored the phylogeographic pattern of the upper intertidal salt marsh plant
67 species, Juncus kraussii Hochst., along the west, south and east coast of South Africa. A phylogeographic
68 break was found along the south coast within the warm temperate coastal biogeographic zone. Their
69 results suggest that this break may have been driven by the rapid shifts in shoreline along the south coast
70 in response to Pleistocene climate cycles. It was also recommended that the phylogeography of South
71 African salt marsh species be explored to aid conservation planning in terms of priority protection and
72 management. In this study, we explore the influence of past climatic conditions on the supratidal salt
73 marsh species Sarcocornia pillansii using species distribution modelling and multi-locus phylogeography.
PeerJ Preprints | https://doi.org/10.7287/peerj.preprints.26927v1 | CC BY 4.0 Open Access | rec: 12 May 2018, publ: 12 May 2018
74 Materials and Methods
75 Taxonomic clarification and leaf material collection
76 Sarcocornia pillansii is an erect to decumbent shrub that can grow up to 70 cm high (Plate S1,
77 supplementary information). The shape of the segments is cylindrical to slightly obconical. Segments
78 appear strongly succulent 8 to 20 mm long and 2 to 4 mm (dried and fresh) wide. There are three flowers
79 per cyme with the seeds being approximately 0.9-1.1 mm long, 0.7-0.9 mm broad, testa tuberculate,
80 papillae 30 μm long (Steffen et al., 2010). According to Kadereit et al. (2006) members of the subfamily
81 Salicornioideae (Amaranthaceae), to which S. pillansii belong, often occur sympatrically or are sometimes
82 organised along ecological gradients such as increasing salinity or flooding. Steffen et al. (2010) found that
83 Sarcocornia pillansii occurs on sandy to clayey soils in saline habitats such as inland salt pans, saline
84 bottoms of intermittent rivers in seasonally arid regions, and it is regularly found dominating supratidal
85 terraces of salt marshes in most of the estuaries spanning the West and East Coast of South Africa. The
86 habitats preferred by S. pillansii are always outside direct influence of tides.
87 Sarcocornia pillansii is a widespread species along the South African coasts where it can be found from
88 the Orange River mouth (straddling border between Northern Cape and Namibia) to Umlalazi in KwaZulu-
89 Natal. Mucina et al. (2006) classify the vegetation types bearing this species as Arid Estuarine Salt Marshes,
90 Cape Estuarine Salt Marshes, Cape Inland Salt Pans, and Subtropical Estuarine Salt Marshes (Steffen et al.,
91 2010). According to the Adams et al. (2016) this species is distributed in 31 out of 277 estuaries. Leaf
92 material for Sarcocornia pillansii was collected in 31 estuaries (1-6 per estuary) stretching from the Orange
93 River Estuary to the Nahoon Estuary (Fig. 1; Table S1, supplementary information).
94 Species Distribution Modelling
95 The assumption of this study is that modelling of the climate envelope, or niche, assumes that the current
96 species or community distribution is in equilibrium with its environment (Phillips et al., 2006). Within the
97 estuarine environment, the important variables affecting species geographical distribution is related to
98 temperature and rainfall (Zhu et al., 2013; James et al., 2016). Climate (present and past) information was
99 obtained from the Worldclim database (version 1.4; Hijmans et al., 2005; http://www.worldclim.org).
100 These climatic layers (19) are based on weather conditions recorded over 50 years from 1950 to 2000.
101 From the 19 layers, six layers were selected that are deemed important in defining species distributions
102 and represent summaries of annual trends for temperature and precipitation, aspects of seasonality and
103 extreme or potentially limiting environmental factors (supplementary information, Table S2). The climate
104 variables were selected as it will reflect the environmental dimensions in which to characterize the climate
105 niche of Sarcocornia pillansii, and is described in Mucina et al. (2006) and Slenzka et al. (2013). These are
PeerJ Preprints | https://doi.org/10.7287/peerj.preprints.26927v1 | CC BY 4.0 Open Access | rec: 12 May 2018, publ: 12 May 2018
106 summarized in Table S2 (supplementary information). There are other climatic variables that affect the
107 distribution of salt marsh species (e.g. elevation relative to mean sea-level and sediment salinity;
108 Veldkornet et al., 2016). However, data on these variables are not available over a large geographical
109 scale. Data layers (vector layer) were downscaled based on the Model the for Interdisciplinary Research
110 on Climate MIROC (Hasumi and Emori, 2004).
111 In this study two distribution modelling algorithms were used: BIOCLIM (profile model) and Maxent
112 (machine learning) (Franklin, 2010). These models were used to calculate current climate suitability, and
113 then projected onto past climate data to develop climate suitability maps the LGM. In order to reduce
114 sampling bias and to facilitate the comparison of outputs from different modelling algorithms the
115 ‘probability of occurrence’ maps were reclassified to binary data (‘present’ or ‘absent’) using the
116 maximum sensitivity plus specificity threshold criterion (Liu et al., 2005). A mask of the estuarine
117 functional zone (EFZ, Veldkornet et al., 2015) was used to sample 2000 background points from the
118 climate layers; Maxent require data of absences (known as background or “pseudo-absences”). Model
119 performance was evaluated for each combination using a standard statistical measure of predictive
120 ability, the area under the receiver operating characteristic curve (AUC). The AUC statistic ranges from 0.5
121 (model prediction is no better than random) to 1.0 (perfect model prediction of presence versus absence).
122
123 DNA extraction, amplification
124 Total genomic DNA was isolated by using the Qiagen DNeasy Plant Mini Kit(www.qiagen.com) and
125 followed the kit’s protocol. PCR was performed using an EmeraldAmp PCR Master Mix (Takara Bio Inc.,
126 Shiga, Japan) with 0.5 mM of each primer. The PCRs were performed in volumes of 25 µl containing 1 µl
127 of template. The PCR protocol consisted of an initial 2 min denaturing step at 94 °C; 40 cycles, each
128 comprising 94 °C for 1.45 min, 55 °C for 30 s, 72 °C for 2 min; and a final 6 min extension step at 72 °C. All
129 PCRs were performed on a GeneAmp 2700 PCR System (Applied Biosystems, Foster City, CA, USA) and
130 were sequenced using both forward and reverse primers. Non-coding chloroplast intergenic spacers
131 sequences were combined and analyzed. The chloroplast gene region trnQ-5’-rps16 (trnQ(UUG) 5'-
132 GCGTGGCCAAGYGGTAAGGC-3'; rpS16x1: 5'-TTGCTTTYTACCACATCGTTT -3') and the 18S-26S cistron, using
133 primers ITS5m (5'-GGAAGGAGAAGTCGTAACAAGG-3') and ITS4 (5'-TCCTCCGCTTATTGATATGC-3') were
134 used in this study (Sang et al., 1995). Sequencing was performed at Inqaba Biotechnical Industries,
135 Pretoria, South Africa). All sequences to be uploaded to GenBank.
PeerJ Preprints | https://doi.org/10.7287/peerj.preprints.26927v1 | CC BY 4.0 Open Access | rec: 12 May 2018, publ: 12 May 2018
136 Sequence assembly, alignment and networks
137 Sequences generated were assembled with CODONCODE ALIGNER v5.4 (Codon Code Corp, Barnstable,
138 MA, USA, http://www.codoncode.com) and automatically aligned using the software’s in-built alignment
139 algorithm and then manually checked; potentially misleading sites such as homopolymer repeats were
140 removed for subsequent analyses. Polymorphic sites were verified by eye and then coded using IUPCA
141 ambiguous sites. Haplotypes were identified as any base substitution or indels (insertion or deletion).
142 Haplotypes were split into different lineages for species that had highly divergent sequences. The
143 genealogical relationships among cpDNA haplotypes were estimated using the statistical parsimony
144 method (SP) in TCS version 1.4b1 (Clement et al., 2000).
145
146 Results
147 Climate models performed well at predicting the distribution of the species (Fig. 2). Area under curve
148 (AUC) values were considered ‘good’ (> 0.80), indicating that the models had high specificity (true positive
149 rate) and sensitivity (false positive rate). The AUC was greater for the Maxent model (AUC = 0.881)
150 compared to the Bioclim model (AUC = 0.837). Generally, the Maxent model performed better at
151 predicting the current distribution of Sarcocornia pillansii than the Bioclim model. The LGM model
152 indicates cooler, drier conditions with greater suitability along the west and east coast of the country. The
153 lower water levels in combination with colder water temperatures in the region during the LGM,
154 presented a cold-water dispersal barrier similar to that on the west coast.
155 The nuclear DNA datasets for Sarcocornia pillansii included 97 sequences (Table 1, Table S3,
156 supplementary information) that resolved into eight ribotypes in the statistical parsimony (SP) network
157 (Table 3, Fig. 3). Ribotype A was the most abundant and widespread along the cool and warm temperate
158 region. Ribotype C and D were restricted to the west and southeast coast, respectively. The other three
159 ribotypes (B, F and G) had highly divergent sequences and were found within individual estuaries along
160 the coast.
161 Analysis of chloroplast DNA for Sarcocornia pillansii included 94 sequences (Table 2, Table S4,
162 supplementary information). The dataset was resolved into four haplotypes in the SP network (Table 4,
163 Fig. 4). Two haplotypes (A and B) were widely distributed. Haplotype A was distributed along the coast
164 line in nine estuaries; whereas haplotype B was distributed in ten estuaries). Haplotype C was restricted
PeerJ Preprints | https://doi.org/10.7287/peerj.preprints.26927v1 | CC BY 4.0 Open Access | rec: 12 May 2018, publ: 12 May 2018
165 to estuaries along the southeast coast. The Uilkraals Estuary contained one unique haplotype (Haplotype
166 D).
167 Discussion
168 Globally, the Pleistocene glacial–interglacial cycles often had severe impacts on species' range dynamics
169 and evolutionary history (Teske et al., 2011). The ensemble SDM predictions under LGM conditions
170 suggest that the species experienced a dramatic range reduction where conditions along the west and
171 southeast coast had greater suitability for the species. It is therefore likely that the estuaries along the
172 south coast were unfavorable for species persistence and estuaries along the west coast estuaries acted
173 as refugia. This agrees with Steffen et al. (2015) noted that the genus Sarcocornia shows large-scale
174 colonization and probably reached the west coast of South Africa during the Pleistocene.
175 Results from the SDM of Sarcocornia pillansii is mirrored in the phylogeographic patterns observed in the
176 genetic data. The distribution of S. pillansii is restricted to the cool and warm temperate marine regions
177 of South Africa but showed no cool and warm temperate biogeographic separation of haplotypes. It can
178 then be concluded that the current phylogeography of S. pillansii is related to past climatic conditions.
179 Post-LGM formation of links between the Overberg (south coast) and the west coast estuaries (as well as
180 interregional movements of birds) may have resulted in dispersal of the common haplotypes (Potts et al.,
181 2013; Slenzka et al., 2013; Teske et al. 2011). Historical links between estuaries or river systems can also
182 explain the abundance of common haplotypes along the south coast. Swartz et al. (2009) explained that
183 along the south coast, lower sea levels than present allowed the confluence between currently isolated
184 river systems, offering opportunities for dispersal among populations.
185 The nuclear dataset also had great divergent sequences, that could also be related to population isolation.
186 Potts et al. (2013) suggests that extremely high sequence divergence between different populations is
187 caused by i) genetic drift during long-term interruption of seed flow or during periods of very small
188 population sizes, ii) lineage sorting of divergent ancestral haplotypes during isolation, or iii) chloroplast
189 capture, i.e., fixation of the chloroplast genome captured from another species into another species. In
190 Sarcocornia pillansii, the highly divergent sequences of high haplotype diversity that suggests a population
191 bottleneck followed by rapid population growth and accumulation of mutations (Mmonwa et al., 2015).
192 Results from this study is similar to that of Potts et al., (2016) for Juncus kraussii Hochst. They suggest that
193 a break shoreline along the south coast may have been driven by the rapid shifts in response to
194 Pleistocene climate cycles. The south coast has the widest section of Agulhas Bank and the intertidal
PeerJ Preprints | https://doi.org/10.7287/peerj.preprints.26927v1 | CC BY 4.0 Open Access | rec: 12 May 2018, publ: 12 May 2018
195 ecosystems would have experienced the greatest degree of lateral displacement. Thus, the south coast
196 may have been a long-term killing zone for J. kraussii. In another study Slenzka et al. (2013) found that
197 there are four geographical groupings of Salicornia (an intertidal annual species) lineages: the south coast,
198 west coast, inland salt pans and Overberg including a distinct molecular lineage in the Knysna Estuary. The
199 authors also relate this to regular fragmentation by marine transgression/regression cycles during the
200 Pleistocene. This would explain the distinct molecular difference in the Uilkraals haplotypes.
201 Conclusion
202 The combination of species distribution modelling and genetic analysis clearly highlighted the influence
203 of climatic oscillations during the Pleistocene on the phylogeography of Sarcocornia pillansii. This
204 supratidal species shows unique haplotypes (with divergent sequences) along the west and east coast.
205 These estuaries were refugia for the species and warrants some conservation prioritization. Estuaries such
206 Olifants, Verlorenvlei, Uilkraals, Hartenbos, Knysna, Kariega, Bushmans and Keiskamma need to be
207 prioritized for conservation as it contains ancient rare haplotypes, to main their ecological resilience in
208 the face of climate change. We also suggest that other estuarine species (lower intertidal and across the
209 biogeographic provinces) be investigated as they have different life history strategies dispersal
210 mechanisms.
211 Acknowledgements
212 The authors thanks Dr Alastair Potts (Centre for Coastal Palaeoscience, Nelson Mandela University South
213 Africa) for assistance in statistical analysis.
214
215 References
216
217 Adams JB, Veldkornet D, Tabot P. 2016. Distribution of macrophyte species and habitats in South African
218 estuaries. South African Journal of Botany 107: 5-11. doi.10.1016/j.sajb.2016.08.001.
219 Avise JC. 2000. Phylogeography: the history and formation of species. Harvard university press.
220 Avise JC, Arnold J, Ball RM, Bermingham E, Lamb T, Neigel JE, Reeb CA, Saunders NC. 1987. Intraspecific
221 phylogeography: the mitochondrial DNA bridge between population genetics and systematics. Annual
222 Review of Ecology and Systematics 18: 489-522.
223 Beheregaray LB. 2008. Twenty years of phylogeography: the state of the field and the challenges for the
224 Southern Hemisphere. Molecular Ecology 17: 3754-3774. doi:10.1111/j.1365-294X.2008.03857.x.
PeerJ Preprints | https://doi.org/10.7287/peerj.preprints.26927v1 | CC BY 4.0 Open Access | rec: 12 May 2018, publ: 12 May 2018
225 Carr AS, Bateman MD, Roberts DL, Murray-Wallace CV, Jacobs Z, Holmes PJ. 2010. The last interglacial
226 sea-level high stand on the southern Cape coastline of South Africa. Quaternary Research 73: 351-363.
227 doi:10.1016/j.yqres.2009.08.006.
228 Chung PP, Chu I, Ballard JWO. 2014. Assessment of temporal genetic variability of two epibenthic
229 amphipod species in an eastern Australian estuarine environment and their suitability as biological
230 monitors. Australian Journal of Zoology 62: 206-215.
231 Clement M, Posada DCKA, Crandall KA. 2000. TCS: a computer program to estimate gene genealogies.
232 Molecular Ecology 9:1657-1659.
233 Fisher EC, Bar-Matthews M, Jerardino A, Marean CW. 2010. Middle and Late Pleistocene paleoscape
234 modelling along the southern coast of South Africa. Quaternary Science Reviews 29: 1382-1398.
235 doi:10.1016/j.quascirev.2010.01.015.
236 Franklin J. 2010. Mapping species distributions: spatial inference and prediction. Cambridge University
237 Press.
238 Hasumi H, Emori S. 2004. Kl coupled model (MIROC) description. Technical Report 1. University of Tokyo,
239 Center for Climate System Research, Tokyo, Japan.
240 Hedenas L. 2011. Global phylogeography in Sanionia uncinata (Amblystegiaceae: Bryophyta). Botanical
241 Journal of the Linnean Society 168: 19-42.
242 Hijmans RJ, Cameron SE, Parra JL, Jones PG, Jarvis A. 2005. Very high resolution interpolated climate
243 surfaces for global land areas. International Journal of Climatology 25: 1965-1978.
244 http://dx.doi.org/10.1002/joc.1276.
245 James NC, Whitfield AK, Harrison TD. 2016. Grey mullet (Mugilidae) as possible indicators of global
246 warming in South African estuaries and coastal waters. Marine Environmental Research 122: 188-195.
247 http://dx.doi.org/10.1016/j.marenvres.2016.11.002.
248 Kadereit G, Mucina L, Freitag H. 2006. Phylogeny of Salicornioideae (Chenopodiaceae): diversification,
249 biogeography, and evolutionary trends in leaf and flower morphology. Taxon 55: 617-642.
250 Li X, Ren L, Liu Y, Craft C, Mander Ü, Yang S. 2014. The impact of the change in vegetation structure on the
251 ecological functions of salt marshes: the example of the Yangtze estuary. Regional Environmental Change
252 14: 623-632. DOI:10.1007/s10113-013-0520-9.
PeerJ Preprints | https://doi.org/10.7287/peerj.preprints.26927v1 | CC BY 4.0 Open Access | rec: 12 May 2018, publ: 12 May 2018