Fifty shades of grey: black beech and mountain beech

0 downloads 0 Views 3MB Size Report
Oct 15, 2018 - parts of New Zealand's South Island (and sometimes in the North Island). .... of 60°C. Markers notssr18 (VIC) and notssr22 (6-FAM) were ..... lower elevation in the South Island provinces of Marlborough (Mirza Creek, Isolation. Creek) .... for technical assistance; Duckchul Park and Julia Allwood for capillary ...

New Zealand Journal of Botany

ISSN: 0028-825X (Print) 1175-8643 (Online) Journal homepage:

Fifty shades of grey: black beech and mountain beech are genetically distinct but locally admixed Rob D. Smissen, Sarah J. Richardson & Peter B. Heenan To cite this article: Rob D. Smissen, Sarah J. Richardson & Peter B. Heenan (2018): Fifty shades of grey: black beech and mountain beech are genetically distinct but locally admixed, New Zealand Journal of Botany, DOI: 10.1080/0028825X.2018.1530683 To link to this article:

View supplementary material

Published online: 15 Oct 2018.

Submit your article to this journal

View Crossmark data

Full Terms & Conditions of access and use can be found at



Fifty shades of grey: black beech and mountain beech are genetically distinct but locally admixed Rob D. Smissen


, Sarah J. Richardsonb and Peter B. Heenan


Allan Herbarium, Manaaki Whenua – Landcare Research, Lincoln, New Zealand; bManaaki Whenua – Landcare Research, Lincoln, New Zealand




Nothofagaceae (southern beech) dominate two-thirds of the remaining indigenous forest of New Zealand. Four of the five species indigenous to New Zealand belong to the genus Fuscospora, and the remaining species to Lophozonia. Two species, Fuscospora cliffortioides (mountain beech) and Fuscospora solandri (black beech), are characterised by their small, entire leaves, characters not found in other species of the genus. Their taxonomic status and rank have been a persistent problem. While individuals of each species are often well differentiated by leaf shape and growth habit, intermediate plants and populations are common. They have been described as constituting a cline with morphological variation correlated with both elevation and latitude, and have been treated by many botanists as varieties of a single species. We assessed relatedness among populations of the two taxa by NeighborNet analysis and model-based inference of population structure using simple sequence repeat data. Two gene pools corresponding to the two taxa can be recognised, and these co-exist at a regional scale. The strongest correlations between environmental variables and membership of modelled populations corresponding broadly with mountain and black beech were with mean annual temperature, elevation, and longitude. Although past (and presumably ongoing) gene flow between them has occurred in areas of contact, the hybrid zone appears to be narrow, at least in some areas. It is appropriate to recognise these recently diverged taxa at species rank, despite local gene flow.

Received 20 March 2018 Accepted 28 September 2018 First published online 01 October 2018 ASSOCIATE EDITOR

Leon Perrie KEYWORDS

Fuscospora; F. cliffortioides; F. solandri; hybridisation; Nothofagaceae; Nothofagus solandri; N. cliffortioides; N. solandri var. cliffortioides; microsatellites; New Zealand

Introduction Classification of organisms into species is complicated when the organisms fail to fall into discrete clusters, regardless of whether they are grouped by morphological, genetic, or historical (phylogenetic) criteria. This can lead to conflict among taxonomists as to the number and circumscription of taxa. One challenging situation occurs when a group of organisms includes two or more distinguishable sets of individuals, but also individuals of intermediate character such that variation is essentially continuous. Such patterns of variation can arise through very different processes, and distinguishing them can CONTACT Rob D. Smissen [email protected] Supplemental data for this article can be accessed © 2018 The Royal Society of New Zealand



provide crucial evidence to inform taxonomic judgement. At least in some cases, these processes lead to differing patterns in the distribution of neutral genetic variation, allowing DNA markers to be used to distinguish among them. In some instances (case 1), morphological (phenotypic) variation is caused by environmental conditions acting directly on the development of organisms (phenotypic plasticity). Individual organisms developing at extreme ends of an environmental gradient may look very different, but may be linked by intermediate morphologies expressed by individuals developing under intermediate conditions. In such cases, variation in morphology observed in nature will not be heritable, and variation in genetic characters will not generally correlate with that in morphological characters. However, if an environmental gradient is strongly correlated with a geographic axis, and there is genetic structure in the population due to isolation by distance (Wright 1943), then genetic and morphological variation will correlate as a result of each being linked to geography (see Meirmans 2012). In other instances (case 2), variation in environmental conditions can lead to corresponding morphological variation among populations of organisms indirectly through natural selection. Populations experiencing similar selective pressure can evolve towards the same morphological end point by different genetic pathways. In such cases, different phenotypes will be heritable in the face of changes to environmental conditions, but may not be robust to interbreeding between morphologically similar but genetically distinct (i.e. convergent) populations because of genetic recombination. Because many ecologically and taxonomically relevant traits are affected by multiple genetic loci, it is possible for marked differences in morphology to evolve among populations even in the face of substantial gene flow due to the additivity of small differences in allele frequency at many loci (Latta 2003). In these cases, genetic differences will not correlate with morphological differences (e.g. Galbany-Casals et al. 2011) except in the case of loci that are linked to the traits in question. In another set of examples (case 3), morphological variation is caused by hybridisation between distinct (or previously distinct) gene pools. In the absence of effective reproductive isolating mechanisms, hybridisation can lead to the merging of populations and a loss of their genetic and morphological differences. However, sufficient reproductive isolation to maintain distinct gene pools in the face of hybridisation can be achieved through a reduction in hybrid fertility or viability (either in the first or later generations of hybrids). Moreover, in a patchy landscape with fragmented distributions, selection against hybrid genotypes outside of spatially restricted intermediate habitats can be sufficient to maintain distinct species even when there appear to be no intrinsic barriers to gene flow between them (Briggs 1962). In these cases, morphological differences will be heritable and should correlate not only with differences at genetic loci underpinning the differences, but also with differences at neutrally evolving loci. In this paper we examine a group of forest trees that show continuous and sometimes subtle variation in morphological characters associated with elevation and latitude over a wide geographic area to consider which of these cases best explains the variation observed. Although the morphological extremes of these trees are readily distinguished, intermediates are widespread and frequent. We consider that taxonomic recognition of the two morphological forms is only appropriate if they constitute distinct gene pools with some level of reproductive isolation between them (i.e. the biological species concept, Mayr 1963).



Here we follow the taxonomy of Heenan and Smissen (2013), which recognised four genera in Nothofagaceae, despite the argument of Hill et al. (2015) that recognition of a single genus, Nothofagus Blume, should be preferred because some fossil Nothofagaceae cannot be assigned to the more narrowly defined genera. Black beech (Fuscospora solandri (Hook.f.) Heenan & Smissen) and mountain beech (Fuscospora cliffortioides (Hook.f.) Heenan & Smissen) have been considered either as two species (e.g. Cockayne 1926; Molloy and St George 1994; Molloy et al. 1999; in all cases as Nothofagus solandri (Hook.f.) Oerst. and Nothofagus cliffortioides (Hook.f.) Oerst.), or as extreme forms of a single, continuously variable species appropriate for recognition as varieties (N. solandri var. solandri and N. solandri var. cliffortioides (Hook.f) Poole; Poole 1958; Moore 1961, p. 1022). In part this reflects shifts in species concepts, with Cockayne adhering to an essentially typological concept and later authors, including Poole, explicitly accepting genetic polymorphism as a property of species and stressing that the ‘interbreeding population goes to make up the species’. However, differences in species concept are only part of the picture. Cockayne asserted that the majority of specimens could be readily and unambiguously ascribed to one or the other species using leaf characters, particularly the leaf base. Conversely, after analysing leaf characters, Poole characterised the variation in the group as continuous and clinal, and the number of individuals and areas of forest where no confident assignment to one or the other taxon could be made as a greater problem. Morphological characters used to distinguish the taxa are shown in Table 1 and some are illustrated in Figure 1. Further details can be found in Heenan and Smissen (2013) and images in Ford et al. (2016). Although both taxa hybridise with the other Fuscospora (R. S. Hill & J. Read) Heenan & Smissen species in New Zealand (Fuscospora fusca (Hook.f.) Heenan & Smissen and Fuscospora truncata (Colenso) Heenan & Smissen), there may be no gene flow from those species (Smissen et al. 2014, 2015). Mountain beech favours higher-elevation environments and forms the treeline in many parts of New Zealand’s South Island (and sometimes in the North Island). However, in the southern part of its range it descends to near sea level. By contrast, black beech is predominantly a lowland species, although in the north of its range it occurs up to at least 800 m above sea level. The geographic distribution of each taxon and the regions of particular ambiguity are well documented in Wardle (1984). Trees of intermediate morphology often occur when the two forms are in contact, particularly at intermediate elevations. Some intermediate stands are composed of trees with more or less uniformly intermediate morphology. However, others have been characterised as having trees with leaf characters

Table 1. Morphological characters used to distinguish black and mountain beech. Character

Black beech

Mature leaf lamina shape Mature leaf base Mature leaf tip Tree form

Oblong to elliptic, more or less symmetric Equally tapered, obtuse to cuneate Obtuse to subacute Large tree with spreading crown

Juvenile habit Cupule and flower

Branches interlacing Densely pilose

Mountain beech Ovate to triangular ovate, often strongly asymmetric Oblique Subacute to acute Prostrate shrub to medium-sized tree with spreading crown Erect Sparsely pilose



Figure 1. F. cliffortioides (A, C, E) and F. solandri (B, D, F). A and B, mature trees. C and D, leafy shoots showing typical leaf shapes. E, F, female inflorescences (one male inflorescence at bottom of F). Photos KA Ford (A–B, D–F), PB Heenan (C).

typical of one or the other species alongside individuals of intermediate morphology (Molloy et al. 1999). Cultivation of both mountain and black beech in common gardens confirms that the morphological difference between them is not entirely the result of phenotypic plasticity (case 1 above), although this does contribute to at least some of the difference in stature between them (Wilcox and Ledgard 1983). Given that the morphological differences are at



least in part heritable (genetic), our focus here is to test whether patterns of variation in neutral genetic markers better fit a case of local selection acting on a common wider gene pool (case 2 above) or hybridisation between distinct gene pools corresponding to the morphological extremes (case 3 above). To do this, we have genotyped 352 trees from 39 locations over most of the altitudinal and latitudinal range of the two species using 10 previously characterised simple sequence repeat (microsatellite, SSR) loci (Jones et al. 2004; Smissen et al. 2012, 2015). These data have been analysed using NeighborNet (Bryant and Moulton 2003) analysis of population genetic distances as well as Bayesian admixture analysis of genotypes (Pritchard et al. 2000). A previous study of SSR variation in Fuscospora species in New Zealand (Smissen et al. 2014) addressed these issues from a smaller number of samples, but here we include trees from additional sites in areas of contact between the taxa and data for two additional SSR loci. We also present analyses relating genetic variation in these species to environmental and geographic variables. These analyses explore how black and mountain beech are distributed throughout New Zealand along well-defined environmental gradients (e.g. mean annual temperature) and according to latitude, longitude, and elevation. These analyses resolve how the two species are separated geographically and in relation to environmental gradients.

Materials and methods Collection sites were chosen to represent the geographic range of each species, to include nearby sites at high and low elevation, and to include stands with intermediate morphology, following in part the distribution maps of Wardle (1984). Between four and 16 trees were sampled from each site. Plant samples were collected in the field into plastic bags and frozen at −20°C within 3 days for later DNA extraction. DNA extraction used either the Maxwell® DNA extraction system (Promega, Madison, WI, USA) or the CTAB method (Doyle and Dickson 1987) from leaf, twig, or leaf-bud (preferred if present). DNA samples were further purified by phenol–chloroform–isoamyl alcohol extraction, and the DNA recovered using Zymo spin columns (Zymo Research Corp., Irvine, CA, USA). Each sample was genotyped for the nine loci (notssr2, notssr6, notssr8, notssr13, notssr18, notssr22, notssr26, notssr36, and BC9) described in Smissen et al. 2012, 2015, and for the locus Ncutas13 (Jones et al. 2004) using primers labelled with fluorophores 6-FAM, VIC, NED or PET and sized on an ABI 3500 genetic analyser using ROX labelled GS500 size standard. Markers notssr26 (NED) and notssr36 (VIC) were multiplexed and amplified with an annealing temperature of 50°C. Markers notssr2 (VIC), notssr8 (6FAM), and notssr13 (PET) were multiplexed and amplified with an annealing temperature of 60°C. Markers notssr18 (VIC) and notssr22 (6-FAM) were multiplexed and amplified with an annealing temperature of 65°C. The other markers were each amplified separately at the annealing temperatures previously published. Markers notssr2, notssr8, notssr13, notssr26, and notssr36 were pooled for genotyping together, as were markers notssr6 (NED), BC9 (PET), Ncutas13 (6-FAM), notssr18 (VIC), and notssr22 (6-FAM). Fragment size analysis files were viewed using GeneMarker 1.51 software (SoftGenetics LLC) and allele sizes recorded manually. Samples with missing data for more than four loci



(failed reactions or ambiguous genotypes) were removed from analyses, leaving 352 individuals representing 39 populations (Table 2). Summary statistics and Nei’s (1972) population genetic distances were calculated using PowerMarker 3.0 (Liu and Muse 2005). Barplots of ancestry estimates and a location map were generated using R v. 3.3.3 (R Core Team 2017). A NeighborNet (Bryant and Moulton 2003) graph was generated from Nei’s (1972) genetic distances in Splitstree 4.13.1 (Huson and Bryant 2006). Bayesian admixture analysis used the software Structure 2.3.3 (Pritchard et al. 2000). The model settings used were default, except that runs were conducted both using and not using sampling location information. Our hypotheses dictate the appropriate number of clusters (K) to test is two. However, we none the less ran Structure with K varying from 1 to 10. Twenty replicate runs of 100,000 sample generations preceded by a burn-in of 100,000 generations for each value of K were combined using Clumpak (Kopelman et al. 2015) and we examined changes in posterior probabilities and the adhoc statistic ΔK using STRUCTURE HARVESTER (Earl and von Holdt 2012). The ancestry estimates for populations shown in Figures 3 and 4 and used in statistical analyses with environmental variables were taken from 20 replicate K = 2 runs of Structure with a burnin of 100,000 and sample of 1,000,000 using location information and combined with Clumpak. Correlation between genetic and geographic distances (isolation by distance) was tested in R v. 3.3.3 (R Core Team 2017) with a Mantel test using the mantel.rtest function in the ade4 package (Chessel et al. 2004), with distances between sampling sites calculated using the earth.dist function in the fossil package (Vavrek 2011). Correlation between genetic and environmental distances (isolation by niche partitioning) was also tested with a Mantel test, with distances among populations in environmental space between sampling sites calculated using standardised and centred values of elevation, mean annual temperature, mean annual rainfall, October vapour pressure deficit (VPD) and the ratio of potential evapotranspiration (PET) to mean annual rainfall. Elevation was derived from topographic maps, and the remaining variables were estimated from the Land Environments of New Zealand (Leathwick et al. 2003). Correlations between membership of populations modelled by Structure and each environmental variable were tested for using Pearson’s correlation coefficient.

Results Summary statistics for the genotyped markers are given in Table S1. Missing data (low number of observations and availability scores for some loci) are generally the result of failed amplifications rather than the presence of null alleles (see Smissen et al. 2012). Although null alleles were suggested to be frequent for notSSR8 by Smissen et al. (2014) in another species (F. fusca) the low availability for notssr8 in this study is the result of ambiguous profiles that were not scored. Observed heterozygosity was consistently lower than expected heterozygosity (gene diversity), but this is to be expected in a data set with genetic structure. A NeighborNet graph for genetic distances among populations (Figure 2) reveals considerable structure in the data but also considerable conflict. In Figure 2, morphologically typical black beech populations appear towards the left of the diagram and morphologically typical mountain beech samples towards the right, with numerous populations in



Table 2. Sampled locations and voucher details. Pop#




Elevation (m)




Arthur’s Pass










3 4 5

Banks Peninsula Cobb Valley Gladstone Heron

Morse, Richardson, Buxton s. n., 20 Mar. 2010 (CHR 635344) No voucher

–41.10 –41.12 –43.33

172.69 175.78 171.12

820 195 700

10 4 10








Isolation Creek





8 9 10

Kaweka Korere Leatham Valley

–39.28 –41.54 –41.80

176.43 172.80 173.19

1100 290 650

11 10 9


Lees Valley Road Lewis Pass













–38.63 –39.42 –42.19 –39.36

176.87 175.45 172.22 175.47

500 665 340 893

11 9 8 7


MacFarlane Mound Mangapae Mangateitei Maruia River Maungahuia Stream Mirza Creek





19 20

Mt Grey Mt Richardson

–43.14 –43.17

172.52 172.26

360 850

4 11


Mt Thomas












Piano Flat





24 25

Rahu Saddle Ruapehu

–42.31 –39.36

172.11 175.47

670 916

10 6







27 28

Sharplin Falls Sunrise Hut track Sunrise Hut track Sunrise Hut track Tākaka Hill Te Marua Tom Creek

–43.64 –39.79

171.41 176.20

460 638

8 13

Heenan s. n., Oct. 2014 (CHR 589015) Smissen 622, 8 Dec. 2011 (CHR 623690) Morse, Richardson, Buxton s. n., 10 Mar. 2010 (CHR 6353465 & CHR 635346) Morse, Richardson, Buxton s. n., 9 Mar. 2010 (CHR 635349) Smissen 1250 & KA Ford, 9 Dec. 2014 (CHR 589017) Buxton s.n., Nov. 2012 (CHR 589025) Heenan s.n., Oct. 2014 (CHR 589018) Smissen 1307, Ford, Breitwieser, 27 Jan. 2015 (CHR589009); Smissen 1309, Ford, Breitwieser, 27 Jan. 2015 (CHR589010); Smissen 1314, Ford, Breitwieser, 27 Jan. 2015 (CHR589011); Smissen 1316, Ford, Breitwieser, 27 Jan. 2015 (CHR589012) Smissen 598, 9 Jun. 2011 (CHR 623677), Smissen 599, 9 Jun. 2011 (CHR 623680) Morse, Richardson, Buxton s.n., 16 Apr. 2010 (CHR 635301) Morse, Richardson, Buxton s.n., 3 Mar. 2010 (CHR 635334) no voucher Smissen 652, 9 Dec. 2011 (CHR 623687) Heenan s. n., Oct. 2014 (CHR 589019) Smissen & PM Novis, 13 Dec. 2013 (CHR 589026) Morse, Richardson, Buxton s. n., 23 Apr. 2010 (CHR 635294–635298) No voucher Richardson s. n., 25 Jan. 2015 (CHR 589013 & CHR 589014) Smissen 607, 9 Jun. 2011 (CHR 623684); Smissen 608, 9 Jun. 2011 (CHR 623682); Smissen 609, 9 Jun. 2011 (CHR 623678); Smissen 610, 9 Jun. 2011 (CHR 623681) Smissen 1216, Nov. 2014 (CHR 589008); Smissen 1218, Nov. 2014 (CHR 589007) Morse, Richardson, Buxton s. n., 7 Mar. 2010 (CHR 635342) Heenan s. n., Oct. 2014 (CHR 589020) Smissen 1048, Novis, Rainforth, 13 Dec. 2013 (CHR 589024) Smissen 1031, Novis, Rainforth, 13 Dec. 2013 (CHR 589023) No voucher No voucher





No voucher





No voucher

–41.02 –41.09 –45.95

172.90 175.15 169.47

620 200 25

8 4 9

12 13 14 15 16 17

29 30 31 32 33

Heenan s.n., Oct. 2014 (CHR 589016) No voucher (Continued )



Table 2. Continued. Pop#

34 35 36 37 38 39


Tongariro River Top Leatham Hut Tōtara Flats Waihohonu Stream Wangapeka River Woolshed Hill



Elevation (m)



–39.23 –41.97

175.78 173.08

900 1020

10 11

Morse, Richardson, Buxton s. n., 8 Aug. 2010 (CHR 635324); Morse, Richardson, Buxton s. n., 8 Aug. 2010 (CHR 635325) Smissen 642, 8 Dec. 2011 (CHR 623688) No voucher

–40.93 –39.21

175.40 175.74

200 920

11 6

Richardson s.n., 28 Oct. 2014 (CHR 589022) No voucher





Heenan s.n., Oct. 2014 (CHR 589021)





Morse, Richardson, Buxton s. n., 1 Mar. 2010 (CHR 635328); Morse, Richardson, Buxton s. n., 1 Mar. 2010 (CHR 635329)

between, including some that are morphologically intermediate. Selected populations spanning the morphological and altitudinal range of mountain and black beech over short geographic distances are highlighted using coloured type. The two populations we sampled from Mt Ruapehu and the Mangateitei population (blue) all occur within approximately 12 km of each other. The lowest-elevation sample of these three (Mangateitei, 665 m) appears to the left of the NeighborNet graph, the lower of two Ruapehu (916 m) samplings further toward the right side of the graph, and the higher Ruapehu sample (near treeline, 1320 m) further again to the right of the graph. Similar patterns can be observed for three samples taken at different elevations along the steeply climbing track to Sunrise Hut (Ruahine Range) from within about 3 km of each other (green), and

Figure 2. NeighborNet graph derived from Nei’s population distance among populations. Locations highlighted with coloured type signify geographically close sites of varying elevation.



Figure 3. Bar plot of Structure population assignment for each sampling site, based on membership of population 1. Each population is shaded according to their estimated membership of Structure population 1 (corresponding with mountain beech). Populations with k1 ≥ 0.6 are arbitrarily designated as mountain, populations with k1 < 0.4 are designated as black beech. Populations with values of k1 > 0.4 and