SEARCH

SEARCH BY CITATION

Keywords:

  • genetic diversity;
  • inbreeding;
  • phenology;
  • rarity;
  • reproduction;
  • speciation

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIAL AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. REFERENCES

There are many genera shared between Australian and Papua New Guinean rainforests. Species in the rare rainforest herbaceous genus Romnalda have a relictual and disjunct distribution within the major rainforest blocs of southern Queensland, north Queensland, and New Guinea. There are only four species in this genus: R. strobilacea, R. grallata, and R. sp.‘Cooper Ck’ from Australia, and R. papuana from New Guinea. The Australian species have restricted distributions and high conservation status. Allozymes were used to study the genetic variation and distinctiveness of all four species. Genetic diversity varied significantly amongst the four species. The species in the centre of the genus distribution contained the highest genetic diversity, regardless of rarity. The undescribed R. sp.‘Cooper Ck’ was identified as a clearly distinct species with morphological affinities to R. papuana, but genetic affinities to R. grallata. The study showed that, where the distributions of R. grallata and R. sp.‘Cooper Ck’ overlapped, there was evidence of hybridization. Reproductive participation within populations was typically low with limited flowering synchrony. Populations of all four species were inbred, but higher levels of inbreeding were not correlated with lower genetic diversity. The timing of flowering appeared to be determined by climate. Altitudinal variation in phenological timing in R. sp.‘Cooper Ck’ has led to genetic isolation within the species, but has also limited its genetic introgression with the co-occurring R. grallata. © 2008 The Linnean Society of London, Botanical Journal of the Linnean Society, 2008, 157, 455–474.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIAL AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. REFERENCES

Tropical and subtropical rainforests have high species diversity and contain a large proportion of the rare and threatened plant species in Australia (Tracey, 1981; Environment Protection and Biodiversity Conservation Act, 1999; Hubbell, 2001). Communities are typically characterized by a small number of common species and a much larger number of rare species (Hubbell, 2001; Murray & Lepschi, 2004). A significant amount of rainforest plant diversity, and rare species, are found in the understorey (De Steven et al., 1987; Gentry & Emmons, 1987; Gentry, 1992). Thus, to maintain diversity in rainforests, we need to maintain the diversity of rare species and to focus on understorey species as well as canopy species. In Australian plants, Murray et al. (2002) found that rarity was associated primarily with reduced geographical distribution. Within Australia, Crisp et al. (2001) identified that areas with the highest species richness also had the highest levels of local endemism, and that the rainforest blocs in the northern Queensland and southern Queensland/northern New South Wales regions had both high species richness and high endemism, consistent with relictual refugia. Rainforests in Papua New Guinea (PNG) have been found to be amongst the most diverse in the world (Wright et al., 1997). The forests are very dynamic as a result of natural disturbances, high rainfall, rich soils, and flora with both Gondwanan and Melanesian ancestry contributing to the very high diversity (Wright et al., 1997). There are many genera shared between Australian and Papua New Guinean rainforests (Webb & Tracey, 1981). Ladiges, Udovicic & Nelson (2003) suggested that the distribution of Australian species with affinities to New Guinean rainforest is old and follows relictual rainforest blocs before drying of the climate during the Oligocene.

Species vary in abundance across their geographical range, some being sparse in some locations and abundant in others, with few species being sparse across their geographical range (Murray & Lepschi, 2004). The effective long-term management and conservation of rare and threatened plants requires an understanding of population biology, and the way in which genetic variation is partitioned within and amongst populations (Coates & Hopper, 2000). In a fragmented landscape, populations are frequently more spatially isolated, reducing the potential for demographic or genetic replenishment from outside the population, and thus are more sensitive to chance effects (Hanski & Gilpin, 1997). Therefore, it is expected that species in small, isolated populations may lose genetic diversity through drift and become less fit as a result of increased inbreeding (Ellstrand & Elam, 1993; Byer & Waller, 1999). Increased spatial isolation is also expected to lead to increased genetic differentiation amongst populations (Young & Clarke, 2000). Rare species in small, isolated populations are thus predicted to have lower reproductive success, lower genetic diversity, and higher inbreeding compared with more abundant and more widespread species (Frankham, 1996). However, evidence from studies has found that this is not always the case for plant species, and some species may have existed in a rare state for long periods and developed mechanisms that enable them to persist in the environment (Gitzendanner & Soltis, 2000; Brigham & Schwartz, 2003; Shapcott, 2007). Mating system variation may impact on the long-term potential for persistence for plant species because of impacts on inbreeding and reproductive output, and may be affected by fragmentation and isolation (Broadhurst & Young, 2007; Coates, Sampson & Yates, 2007). However, it has become increasingly clear that the genetics of rare plants are not always easily predicted (Gitzendanner & Soltis, 2000; Shapcott, 2007).

Study species

There are only four known species in the genus Romnalda (Dasypogonaceae), making the whole genus rare. Three species are endemic to Australia: R. strobilacea, R. grallata (R.J.F. Hend.), and an undescribed species R. sp.‘Cooper Ck’. The fourth species, R. papuana (Lauterb.) P.F. Stevens, is endemic to New Guinea (Stevens, 1978; Henderson, 1986, 2002). Romnalda's closest relatives are thought to be the genus Lomandra from which its name was derived as an anagram (Stevens, 1978; Henderson, 1986). This is supported by evidence from Trapezites butterflies (skipper butterflies), which are endemic to Australia and New Guinea and have host-specific associations with both Romnalda and Lomandra species (Atkins, 1975, 2004). Although the evolutionary history of Romnalda is unclear, it appears likely to be of Gondwanan origin (Henderson, 1986; Fay et al., 2000).

The Australian Romnalda species all have restricted distributions and high conservation status. Romnalda strobilacea (vulnerable) (Environment Protection and Biodiversity Conservation Act, 1999; Henderson, 2002) has a limited distribution in the Sunshine Coast region, R. grallata (rare) (Henderson, 2002) is known from only two locations within protected areas at Mt Lewis and the Cape Tribulation National Park, and R. sp.‘Cooper Ck’ (undescribed) is known only from three populations within the Cape Tribulation National Park (Fig. 1). There are currently only five locations documented for R. papuana, which are highly disjunct within New Guinea (Fig. 1). The current conservation status of R. papuana is rare. Using the South-east Asian and Pacific (SEAPAC) Biomization (Pickett et al., 2004), we determined that all Romnalda species currently occur primarily in Warm Temperate Rainforest (WTRF; equivalent to subtropical rainforest), with some populations in lowland Tropical Evergreen Broadleaf Forest (TRFO; equivalent to tropical rainforest). There are major disjunctions both within and between Romnalda species distributions (Fig. 1). Populations are apparently restricted to undisturbed rainforest habitats in mostly upland and highly significant refugial rainforest areas (Webb & Tracey, 1981), indicating that current species distributions may be largely relictual. The distribution of these four Romnalda species appears to be significant in terms of the evolutionary history of rainforest species in Australia and New Guinea.

image

Figure 1. Maps indicating the global distribution of the species in the genus Romnalda and the relative locations of sample sites of each species. Species and site codes are indicated: Romnalda sp.‘Cooper Ck’ (RC), R. papuana (RP), R. grallata (RG), and R. strobilacea (RS). The extent and distribution of remnant vegetation is indicated by shading of the background on the species sample site maps. The geographical reference points and scale are indicated.

Download figure to PowerPoint

Romnalda species are small herbaceous perennials growing up to approximately 50 cm in height, with strap-like leaves, often with distinctive prop-like roots. The robustness of the latter feature varies amongst species (Henderson, 1986; A. Shapcott, pers. observ.). The small cream bisexual flowers produce dehiscent capsules containing up to three to four seeds (Henderson, 1986). The architecture and size of the flower stalk and inflorescence vary amongst species, becoming larger and more robust the more southern the species distribution (Henderson, 1986; A. Shapcott, pers. observ.). Flower stalks are retained on the plant until after the next reproductive season. There is evidence of localized downhill dispersal, potentially caused by gravity, water, ground-scratching animals, or ants (A. Shapcott & B. Bau, pers. observ.), but little evidence of potential vectors for long-distance dispersal. Romnalda strobilacea is known to be a larval host and food plant for Trapezites symmomus, an endemic species of skipper butterfly that lays eggs on and feeds on its leaves (Atkins, 2004).

Aims

Although the palaeobotanical history has been documented in terms of species distributions and taxonomic relationships (for example, Ladiges et al., 2003), there have been few, if any, comparative population genetic studies of an entire genus. This study investigates the effects of ancient fragmentation on all known species of the restricted genus of rainforest herbs Romnalda. The project specifically aims to investigate the genetic distinctiveness of Romnalda species and to clarify the taxonomic status of the undescribed R. sp.‘Cooper Ck’ and its relationship to the other species. The study tests the hypothesis that congeneric species are decreasingly genetically diverse and increasingly inbred with increasing rarity. The hypothesis that species reproductive activity decreases with increasing rarity is also investigated. The potential impacts of past and future climate change on the genus are considered, particularly the potential impacts on phenological synchronicity and the potential significance of hybridization within the genus.

MATERIAL AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIAL AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. REFERENCES

Field methods

Herbarium records and local knowledge were used to determine the known potential distribution of each of the four Romnalda species. Logistics and accessibility then determined the final sample sites. Where possible, species were sampled from either all known populations (for example, R. sp.‘Cooper Ck’) or from across their species range (R. strobilacea, R. grallata). However, in the case of R. papuana, populations from only one of four known locations in PNG were successfully sampled during the project. This was a result of local politics and the inaccessible nature of some sites (two of which were last visited by botanists over 30 years ago).

The location of each species collection site was recorded using a Global Positioning System (GPS) and topographical maps, and the distances between sites were determined using geographical grid co-ordinates (eastings and northings). At R. strobilacea and R. papuana sites, the relative location of every Romnalda plant was mapped relative to belt transects spanning the entire population, and the reproductive status of all plants was recorded. For R. grallata and R. sp.‘Cooper Ck’ sites, only the data for the sample plants were recorded because of time constraints. At each site, plants were sampled to cover the distribution of the population at the site, and were spaced to reduce the likelihood of sampling clones where clumps of plants were found. On average 25–40 plants per site were sampled for genetic analysis. Samples consisted of fresh mature leaf material. Whilst in the field and in transit, samples were kept as cool as possible in insulated bags or cool box. Samples were transferred to a refrigerator at 4 °C as soon as possible. All material was processed whilst fresh. Voucher specimens for each population of the Australian Romnalda species sampled were deposited at the Queensland Herbarium (BRI), and voucher specimens of the sampled populations of R. papuana were lodged at the National Herbarium of PNG at the Forest Research Institute in Lae.

Laboratory methods

Genetic variation was investigated using allozyme analysis, following the standard methods used by Shapcott (2002). Allozymes were selected as they are a reliable genetic method useful for conservation management purposes, and can be obtained in a short time frame for members of a little studied family. The following eight enzymes were clearly and consistently resolved, providing 11 consistently scorable loci: shikimic dehydrogenase (Sdh) (EC 1.1.1.25; Moran & Bell, 1983); isocitric dehydrogenase (Idh) (EC 1.1.1.42; Shaw & Prasad, 1970); phosphose glucose isomerase (Pgi) (EC 5.1.3.1.9; Guries & Ledig, 1978); glucose-6-phosphate dehydrogenase (G6pd) (EC 1.1.1.49); glutamate dehydrogenase (Gdh) (EC 1.4.1.2; Shaw & Prasad, 1970); malate dehydrogenase (Mdh) (EC 1.1.1.37); 6-phosphogluconic acid (6Pg) (EC 1.1.1.44); and aspartate aminotransferase (Aat) (EC 2.6.1.1; Brown et al., 1978). A lithium hydroxide gel and tank buffer system were used (Brewbaker et al., 1968). All runs were cross-referenced by including repeated runs from a subset of individuals from the RS2 population, by repeating some samples to allow cross-referencing of populations and species, and by running samples from multiple sites and/or species on the same gels. When a locus was not present in one species, it was allocated a unique monomorphic allelic designation. This enabled all loci to be included in cross-species comparisons.

Statistical methods

The percentage of each population surveyed that showed any evidence of reproductive activity (R) was determined from the field data. For R. papuana populations, the percentages of the population with evidence of different reproductive stages were calculated separately (%buds, %flowers, %fruit, %old reproductive structures). The ecological densities of populations (D) of each species were calculated when the appropriate data were available, using the number of plants recorded within the area occupied by the population, or, in the case of spatially larger populations, the number of plants within the area surveyed using transect data. When the data were available, the census population size (N) was determined. The x,y co-ordinates (m) for plant relative locations were used to calculate the distances between plants within populations, and these were employed in further analyses. Maps of sample sites for each species were generated using latitude and longitude co-ordinates and available Geographical Information System (GIS) map layers using Arc-GIS (Environmental Systems Research Institute, 2001).

The bands from each enzyme system were assigned to alleles and loci on the basis of theoretical expectations and the observed banding patterns (Gottlieb, 1981; Wendel & Weeden, 1989). The multilocus genotypes of each specimen within populations were compared to look for evidence of clonal spread. The BIOSYS-1 package (Swofford & Selander, 1981) was employed to determine the following for each population: allelic frequencies, mean number of alleles per locus (A), mean percentage of polymorphic loci (Pl), observed mean heterozygosity (Ho), and expected mean heterozygosity based on Hardy–Weinberg expectations (He), using Levene's (1949) correction for small sample sizes. Species level estimates of the mean number of alleles per locus (As) were also determined. The significance of genetic partitioning amongst species within the genus and amongst populations within the species was statistically tested by analysis of molecular variance (AMOVA) using the GenAlEx V6 program (Peakall & Smouse, 2005). Kruskal–Wallis tests were used to determine whether the four species differed significantly in any of the genetic measures of genetic diversity (A, Pl or He), inbreeding (Ho, F) (Wright, 1965), or evidence of reproductive activity (R). Dunnett's T3 post hoc tests and Mann–Whitney U-tests were employed to determine whether there were significant differences between species pairs using the SPSS statistical program.

Several standard genetic distance coefficients were calculated amongst sites using the BIOSYS program. The unweighted pair group method with arithmetic averaging (UPGMA; Sneath & Sokal, 1973) and the distance Wagner procedure (Farris, 1972) were used to construct trees to represent relationships between species and populations. The goodness of fit of both types of dendrogram compared with the input matrix was evaluated using the cophenetic correlation coefficient (Sneath & Sokal, 1973), the F value of Prager & Wilson (1976), the Farris ‘f’ (Farris, 1972), and the standard deviation (Fitch & Margoliash, 1967). The dendrogram and genetic distance matrix with the best fit were then used for further analysis and interpretation. For each species population, allelic frequencies at individual loci were plotted as pie graphs and overlaid on maps to visualize relationships amongst populations in a spatial context. Spearman's rank correlation tests were undertaken to test for relationships between reproductive activity (R) and genetic measures (A, Pl, He, Ho, F) amongst R. strobilacea populations using the SPSS statistical program.

A Mantel test (Mantel, 1967) with 999 permutations was undertaken to test whether there were correlations between genetic distances and geographical distances within the populations of each Romnalda species, following the procedure of Smouse, Long & Sokal (1986) and Smouse & Long (1992). For each Romnalda species, spatial autocorrelation analysis was undertaken within each population to test whether there was geographical clumping in the distribution of like genotypes within the population, or whether they were randomly distributed. This procedure allows the spatial autocorrelation analysis of multilocus genotypes, and these analyses were performed using 999 permutations and even distance classes. The size of the distance classes used for analysis was determined from the distances between sampled plants, such that there were sufficient data for statistically meaningful results and an ecologically meaningful scale for the species given their within-population spatial distributions. Several incrementally increasing scales of distance classes were analysed to investigate the effect of scale on the results, and the distance class giving the best results was used for final analysis and comparisons between populations. Correlograms were produced which plot the autocorrelation coefficient (r) against the distance class. These also indicate the upper and lower 95% confidence limits, and correlation values (r) outside these limits were considered to be significant (P < 0.05). For multilocus correlograms such as these, error bars indicating the 95% confidence intervals around the correlation values (r) were also given. A significant positive autocorrelation (r) indicates significant clumping of genetically like individuals within that distance class, whereas a negative autocorrelation indicates that neighbours within the distance class are more often genetically unlike (Legendre & Fortin, 1989). These tests were undertaken using the GenAlEx V6 program (Smouse & Peakall, 1999). A plot mapping the relative locations of sampled plants of R. grallata and R. sp.‘Cooper Ck’ where they co-occur (RG7/ RC7), and their genotypes at the Aat locus, was produced to investigate species overlap and the potential for gene flow amongst the species.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIAL AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. REFERENCES

Genetic diversity within the genus

The species differed significantly in their within-population genetic diversity across all measures (He: χ2 = 18.76, P < 0.005; A: χ2 = 20.291, P < 0.005; Pl: χ2 = 8.02, P < 0.005; Table 1). Romnalda grallata populations had the highest genetic diversity (A = 2.77, Pl = 100, He = 0.508; Table 1) of all species, but were not significantly more genetically diverse than R. sp.‘Cooper Ck’. Romnalda papuana had significantly lower (P < 0.005) within-population genetic diversity than all the other species (A = 1.6, Pl = 52.5, He = 0.176; Table 1). In this genus, genetic diversity is not related to species abundance, range, or rarity, as R. sp.‘Cooper Ck’ has the narrowest geographical range and lowest total abundance of the four species, but was significantly more genetically diverse than more widespread and abundant Romnalda species. However, R. sp.‘Cooper Ck’ and R. grallata, the two species at the centre of the genus distribution, had significantly higher genetic diversity than those species at the extremes of the genus geographical distribution, R. papuana and R. strobilacea (Table 1; A, Pl, He; P < 0.005).

Table 1.  Summary of mean population measures for the four Romnalda species. Genetic analysis data were averaged across ten allozyme loci for each species, where n is the number of samples per population, A is the mean number of alleles per locus, Pl is the percentage of polymorphic loci, Ho is the observed heterozygosity, and He is the Hardy–Weinberg expected heterozygosity. The mean and standard deviation (SD) values are given for each species. The number of populations sampled (N) and species conservation status codes (E, endangered; R, rare; V, vulnerable) are given. Significant differences amongst species for the variables (Kruskal–Wallis) are indicated:*P < 0.05 (this level not seen), **P < 0.005. Significant differences between species pairs (Dunnett's T3, Mann–Whitney U-test) are indicated using the species code, e.g. superscript ‘C’ means that there is a significant difference between the species and R. sp.‘Cooper Ck’: G, grallata; P, papuana; S, strobilacea
Species (conservation code)NnAPlHoHe
R. sp‘Cooper Ck’ (E?)3     
Mean 32.72.6**PS100**PS0.139**PS0.441**PS
SD 0.90.0820.0000.0390.054
R. papuana (R)4     
Mean 261.6**CGS52.5**CGS0.08**CS0.176**CG
SD 10.10.0714.3300.0110.037
R. grallata (R)6     
Mean 34.22.77**PS100**PS0.141**PS0.508**PS
SD 2.40.0750.0000.0150.006
R. strobilacea (V)12     
Mean 38.81.92**PG68.3**CPG0.074**CG0.231**CG
SD 23.10.1539.8600.0230.044

Taxonomic phylogeography

A significant 56% of genetic variation in the genus Romnalda was found amongst species (PhiRT as measured by AMOVA = 0.558; P = 0.001), with only 9% of the variation in the genus Romnalda found amongst populations within species (PhiRT = 0.199; P = 0.001). Thus, the species are clearly genetically distinct within the genus. The four species are distributed along a north–south axis (Fig. 1); the two species at the geographical extremes of the distribution of the genus were the most genetically distinct, as expected (Fig. 2). The most southern species, R. strobilacea, is clearly the most genetically distinct in the genus, reflecting its greater geographical separation from other members of the genus (Figs 1, 2).

image

Figure 2. Unweighted pair group method with arithmetic averaging (UPGMA) cluster of all four Romnalda species [Romnalda sp.‘Cooper Ck’ (RC), R. papuana (RP), R. grallata (RG), and R. strobilacea (RS)] and their populations using the modified Rogers distance (Wright, 1978). Goodness-of-fit statistics: Farris ‘f’ = 15.189 (Farris, 1972); ‘F’ = 9.350 (Prager & Wilson, 1976); percentage standard deviation = 11.508 (Fitch & Margoliash, 1967); cophenetic correlation = 0.961.

Download figure to PowerPoint

Romnalda strobilacea is genetically separated from all other species by significantly differing band migration distances across several loci and enzyme systems, or the presence or lack of enzyme activity for a locus [for example, 6Pg-2 (uniquely present), Aat (unique D allele), Pgi-1 (unique A allele, alleles D and E absent), Pgi-2 (uniquely absent)], compared with the other three species (Tables 2, 3). The most northern species, R. papuana, from New Guinea is the next most genetically distinctive species (Fig. 2). Although R. papuana is similar to R. sp.‘Cooper Ck’ in terms of gross morphology, it is clearly genetically distinct, indicating that R. sp.‘Cooper Ck’ is probably not conspecific with R. papuana and worthy of species classification (Fig. 2). Romnalda papuana populations were monomorphic for Sdh (allele D) with a significantly different migration distance. This allele was not found in any of the other species. Romnalda papuana was also the only species with Mdh-1 allele D present in the populations. In addition, there were alleles present in the other species that were absent from R. papuana (Tables 2, 3).

Table 2.  Allele frequencies at each locus expressed in Romnalda grallata (RG), R. sp.‘Cooper Ck’ (RC), and R. papuana (RP). The sample size (n) is given for each population
LocusSpecies/population code
RG1RG2RG3RG5RG7RG8RC4RC6RC7RP1RP2RP3RP4
(n)34323233393534323223144225
Sdh
 A0.0000.0000.0630.0150.0000.0000.0000.0000.0000.0000.0000.0000.000
 B0.7350.1720.3280.6670.2180.4000.5590.4690.5310.0000.0000.0000.000
 C0.2650.8280.6090.3180.7820.6000.4410.5310.4690.0000.0000.0000.000
 D0.0000.0000.0000.0000.0000.0000.0000.0000.0001.0001.0001.0001.000
Aat
 A0.3820.4530.2030.3030.0510.1710.8680.7500.3280.0000.0000.0000.000
 B0.5880.5470.7660.6820.3850.5570.1320.2500.5161.0001.0001.0001.000
 C0.0290.0000.0310.0150.5640.2710.0000.0000.1560.0000.0000.0000.000
Gdh
 A0.0290.2340.1090.2120.0000.1710.0000.0630.0000.0000.0000.0000.000
 B0.6470.5470.5940.6060.4620.3860.1180.0780.2810.7390.9290.9050.680
 C0.3240.2190.2970.1820.5380.4430.8820.8590.7190.2610.0710.0950.320
Mdh-1
 A0.2940.3910.7970.2120.2560.1140.0150.1880.0000.0000.0000.0000.000
 B0.3820.4380.2030.3940.3720.5710.0290.2970.1880.1090.0000.3330.580
 C0.3240.1720.0000.3940.3720.3140.9560.5160.8130.7390.8210.6190.360
 D0.0000.0000.0000.0000.0000.0000.0000.0000.0000.1520.1790.0480.060
Mdh-2
 A0.2650.0470.0630.0300.0000.0000.1620.0160.0000.0000.0000.0000.000
 B0.2350.2340.4220.5760.2440.1710.8380.8280.4690.0000.0000.0000.000
 C0.5000.7190.5160.3940.7560.8290.0000.1560.5310.0000.0000.0000.000
G6pd
 A0.4850.4060.2810.5610.5770.8860.3820.6560.6880.0000.0000.0000.000
 B0.5150.5940.7190.4390.4230.1140.6180.3440.3131.0000.9641.0000.880
 C0.0000.0000.0000.0000.0000.0000.0000.0000.0000.0000.0360.0000.120
Pgi-1
 A0.0000.0000.0000.0000.0000.0000.0000.0000.0000.0000.0000.0000.000
 B0.0740.0780.0630.0910.0260.0430.0150.2340.0160.0000.0000.0000.000
 C0.7210.4380.3280.5300.8210.2860.1760.2500.8590.0000.0000.0000.000
 D0.1470.3440.4060.3330.0640.3140.3090.3750.1090.0650.0000.0480.100
 E0.0590.1410.2030.0450.0900.3570.5000.1410.0160.9351.0000.9520.900
Pgi-2
 A0.3970.0940.0470.2270.3330.0290.3240.3750.3590.0000.0000.0000.000
 B0.5150.4690.5780.4390.5000.5710.6030.6090.4840.0000.0000.0000.000
 C0.0880.4380.3750.3330.1670.4000.0740.0160.1561.0001.0001.0001.000
6Pg-1
 A0.3680.1720.3910.5150.4230.3290.1180.5000.2500.8040.6430.4050.700
 B0.2350.5160.5000.4550.3720.5570.6180.4530.3910.0000.0000.0000.000
 C0.3970.3130.1090.0300.2050.1140.2650.0470.3590.0000.0000.0000.000
 D0.0000.0000.0000.0000.0000.0000.0000.0000.0000.1960.3570.5950.300
Idh
 A0.9120.7810.6560.9550.5390.3430.9120.7810.4380.0000.0000.0000.000
 B0.0880.2190.3440.0450.3030.6570.0880.2190.4380.6960.8210.6190.520
 C0.0000.0000.0000.0000.1580.0000.0000.0000.1250.0000.0000.0000.000
 D0.0000.0000.0000.0000.0000.0000.0000.0000.0000.3040.1790.3810.480
Table 3.  Allele frequencies at each locus expressed in Romnalda strobilacea (RS). The sample size (n) is given for each population
LocusPopulation
RS1RS2RS3RS4RS5RS6RS7RS8RS9RS10RS11RS12
(n)104495237297233133431938
Sdh
 A0.9900.9801.0001.0001.0001.0001.0001.0000.9391.0001.0001.000
 B0.0000.0200.0000.0000.0000.0000.0000.0000.0610.0000.0000.000
 C0.0100.0000.0000.0000.0000.0000.0000.0000.0000.0000.0000.000
Aat
 A0.0000.0000.0000.0000.0000.0000.0000.0000.0000.0000.0000.000
 B0.0000.0000.0000.0000.0000.0000.0000.0000.0000.0000.0000.000
 C0.0000.0100.0000.0000.0000.0000.0000.0000.0000.0000.0000.000
 D1.0000.9901.0001.0001.0001.0001.0001.0001.0001.0001.0001.000
Gdh
 A0.0000.0000.1150.0000.0000.0000.0000.0000.0000.0000.0000.000
 B0.8461.0000.8850.9460.9311.0000.4780.8710.9091.0001.0001.000
 C0.1540.0000.0000.0540.0690.0000.5220.1290.0910.0000.0000.000
Mdh-1
 A0.8131.0000.4900.9190.9311.0000.6960.9030.9390.8370.8420.803
 B0.1880.0000.4520.0540.0690.0000.3040.0970.0610.1630.1580.197
 C0.0000.0000.0580.0270.0000.0000.0000.0000.0000.0000.0000.000
Mdh-2
 A0.0000.0000.0000.0000.0000.0000.0000.0000.0000.0000.0000.000
 B0.0820.6530.0000.0000.2410.2860.3480.3870.1060.1860.0260.303
 C0.9180.3471.0001.0000.7590.7140.6520.6130.8940.8140.9740.697
G6pd
 A0.6440.9800.7120.9730.9480.9290.9780.8551.0000.5930.6050.763
 B0.3560.0200.2880.0270.0520.0710.0220.1450.0000.3840.3160.211
 C0.0000.0000.0000.0000.0000.0000.0000.0000.0000.0230.0790.026
Pgi-1
 A0.1630.1120.2120.0950.0690.0000.2170.1450.2120.1510.1320.079
 B0.7790.4800.7400.4460.6900.9290.3910.5970.6060.8370.8680.921
 C0.0580.4080.0480.4590.2410.0710.3910.2580.1820.0120.0000.000
6Pg-1
 A0.0240.9390.0580.1490.4660.1430.2390.3230.3640.8600.5000.684
 B0.9380.0610.8940.7570.2930.4290.5430.3550.6210.1400.4470.289
 C0.0380.0000.0480.0950.2410.4290.2170.3230.0150.0000.0530.026
6Pg-2
 A0.9331.0000.5000.8380.3790.4290.8910.5650.5300.4420.0000.474
 B0.0670.0000.4810.1620.6210.5710.1090.4350.4700.5581.0000.526
 C0.0000.0000.0190.0000.0000.0000.0000.0000.0000.0000.0000.000
Idh
 A0.3510.1940.0000.0000.0000.0000.0000.0000.0000.0810.0000.132
 B0.3990.7960.7210.2301.0001.0001.0001.0001.0000.8371.0000.632
 C0.2500.0100.2790.7700.0000.0000.0000.0000.0000.0810.0000.237

Romnalda grallata and R. sp.‘Cooper Ck’ are morphologically quite distinct to the naked eye; however, they are genetically the most similar species (Fig. 2). They genetically cluster separately, except where populations of the two species are interspersed at the Mt Pieter Botte site (RC7, RG7; Fig. 2). At this site (RG7, RC7), both species populations contain the Idh C allele at low frequency. This allele was not found in other populations of either species (Table 2). However, the populations of R. sp.‘Cooper Ck’ (RC6) and R. grallata (RG5) that were immediately adjacent (approximately 150 m apart), but not interspersed, at Mt Sorrow were genetically distinctive (Fig. 2). At this site, R. grallata (RG5) was replaced by R. sp.‘Cooper Ck’ (RC6) on the slightly higher altitude razor back ridge. The differences between R. sp.‘Cooper Ck’ and R. grallata were mostly a result of differences in allelic frequencies rather than the presence or absence of alleles at the loci scored. However, the Aat C allele was absent from R. sp.‘Cooper Ck’ populations (RC4, RC6), but was present at the RC7 site. This allele was found in most R. grallata populations and was common in the R. grallata RG7 population (Table 2). This may represent evidence of some genetic introgression between these two species via hybridization.

A plot of the distribution of individuals of both species, R. grallata (RG7) and R. sp.‘Cooper Ck’ (RC7), at this site, and showing the Aat genotypes of individual plants, indicates that the two species mostly occupy different adjacent parts of the site but overlap in distribution (Fig. 3). The Aat C allele occurs in the R. sp.‘Cooper Ck’ population in both heterozygous (three plants) and homozygous (three plants) forms, and R. sp.‘Cooper Ck’ plants containing the Aat C allele are located close to R. grallata plants that also contain the allele (Fig. 3). The greatest distance recorded between such plants was approximately 25 m (Fig. 3). Two of the three plants of R. sp.‘Cooper Ck’ that were homozygous for the Aat C allele had associated field notes querying their species identity, and reporting that they had a morphological form that was slightly intermediate with R. grallata, even though they were identified as R. sp.‘Cooper Ck’ in the field. The presence of homozygotes indicates that, if these plants are hybrids, they are second generation, and are self-compatible, crossed with another F1 hybrid, or introgressed back to R. grallata. The field identification to R. sp.‘Cooper Ck’ tends to lend more support to self-fertilization or introgression back to R. sp.‘Cooper Ck’, and thus potentially incorporating the Aat C allele into the R. sp.‘Cooper Ck’ population.

image

Figure 3. Map indicating the relative locations of Romnalda grallata (RG) and R. sp.‘Cooper Ck’ (RC) individuals at the Mt Peiter Botte site RC7/RG7. The x and y axes are not to equal scale. The genotypes of every plant documented for the Aat locus are given for each species (refer to legend). In general, R. grallata plants are denoted by square symbols (e.g. ▪) and R. sp.‘Cooper Ck’ plants are denoted by diamonds (e.g. ♦), except for the seven putative R. sp.‘Cooper Ck’ plants containing the Aat C allele (putative hybrids) which are represented by triangles (e.g. ▴) and encircled. The approximate major zones occupied by the two species are indicated by broken (R. grallata) and dotted (R. sp.‘Cooper Ck’) lines.

Download figure to PowerPoint

Reproduction

Although participation in reproductive activity varied considerably amongst populations within and between species, there was no significant difference (Kruskal–Wallis χ2 = 3.96, P = 0.266) amongst species in population-level participation in reproductive activity (R), with approximately 20% of plants reproductively active in populations of all species (Table 4). In the R. papuana populations, there was a significant positive correlation between population density (D) and the reproductive activity of the populations (R; Spearmans Rho = 0.868; P < 0.05), but these results should be viewed with caution because of the small number of populations (four). Romnalda papuana populations showed relatively high reproductive activity (R; 30%); however, the data indicated that a very low proportion of these populations were synchronous in their reproductive stages, thereby effectively reducing the size of the reproductive population at any one time (Table 5).

Table 4.  Summary of results of tests for inbreeding and spatial genetic structure within populations of the four Romnalda species. Species and site codes are given, as well as the sample size (n). Genetic analysis data are averaged across ten loci for each species: F is Wright's allelic fixation index; R is the percentage of the population with evidence of reproductive activity. Significant differences amongst species for F and R (Kruskal–Wallis) are indicated:*P < 0.05, **P < 0.005. Significant differences between species pairs (Dunnett's T3 or Mann–Whitney U-test) are indicated using the species code, e.g. superscript ‘C’ means that there is a significant difference between the species and R. sp.‘Cooper Ck’: G, grallata; P, papuana; S, strobilacea. Significant results for spatial autocorrelation tests of genetic structure within populations are indicated (ns, not significant; sig, significant), as are the sizes of within-population non-random patches as determined from correlograms. The results of Mantel tests for correlations between the genetic distance and geographical distance of plants within populations are given (Rxy), and the significance is indicated: *P < 0.05, **P < 0.005.
Species/site codeLocationSample size nFRSpatial autocorrelationWithin-population patch sizeMantel
Rxy
R. sp‘Cooper Ck’
 RC4Little Cooper Ck340.7455.88ns00.32**
 RC6Mt Sorrow310.47621.88ns00.26**
 RC7Mt Peiter Botte320.77512.12ns0−0.071
Mean RC  0.665*13.29   
R. papuana
 RP1Bosavi 1230.41430.43ns0−0.026
 RP2Bosavi 2140.31435.56ns00.352*
 RP3Bosavi 3420.51442.29sig15 m0.331**
 RP4Bosavi 4250.67825.45sig15 m0.301**
Mean RP  0.480*G33.43   
R. grallata
 RG1Mt Lewis 1340.75814.71nanana
 RG2Mt Lewis 2320.7236.25ns00.182**
 RG3Mt Lewis 3320.6840.00ns00.012
 RG5Mt Sorrow330.74742.42ns00.02
 RG7Mt Peiter Botte390.73156.41ns00.022
 RG8Mt Lewis 4350.73345.71sig30 m0.246**
Mean RG  0.729*P27.58   
R. strobilacea
 RS1MaryCC1040.82443.40sig60 m0.257**
 RS2Mapleton490.56921.43sig20 m0.048
 RS3Triunia520.65321.74sig30 m0.082
 RS4Nominus370.6726.15ns00.071
 RS5Howells Knob290.6678.33sig30 m0.221**
 RS6Mt Mellum70.4530.00nanana
 RS7Bridge ck230.48435.48ns00.151*
 RS8Witta310.61326.09ns00.032
 RS9Wolli330.85813.04ns00.125*
 RS10Policeman's430.63613.00ns00.082
 RS11High Torr190.42538.89nanana
 RS12 McDonalds380.70122.00ns00.109
Mean RS  0.629*22.46   
Table 5. Romnalda papuana reproductive summary showing the percentage of plants recorded with different reproductive stages present: buds, flowers, fruit, old (or dead) reproductive structures
Site codeN%buds%flowers%fruit%old
RP1923.36.518.58.7
RP2456.717.811.111.1
RP317514.313.126.31.7
RP4550.014.510.93.6

The reproductive timing of the species seemed to vary slightly amongst the species, as recorded by the field notes (which noted the date and reproductive stages present) and reproductive activity data. These preliminary results indicated that the timing of the flowering peak seemed to follow a latitudinal gradient amongst the species. Flowers were recorded in R. strobilacea populations between June and October, with a consistent peak in September. Romnalda grallata appeared to have a slightly later flowering peak in November, and R. papuana seemed to have a mid-reproductive peak when sampled in February, when buds, flowers, fruit and old fruit were all present (Table 5). The flowering time pattern in these species was consistent with a climatic gradient following the geographical gradient in species distributions. Romnalda sp.‘Cooper Ck’ seemed to have a broader, more continuous, and earlier flowering peak, with flowers observed between June and November and, like R. papuana, plants in different reproductive stages were recorded simultaneously. The variation in flowering time in R. sp.‘Cooper Ck’ seemed to follow an altitudinal gradient within the species, with the latest flowering population occurring in RC7 at the highest altitude site (approximately 800 m) and hence coolest climate, and the earliest flowering occurring at the lowest altitude site (RC4, approximately 100 m). The site at which R. sp.‘Cooper Ck’ and R. grallata populations were interspersed (RC7, RG7: Fig. 3) recorded individuals of both species flowering simultaneously, providing the possibility of cross-pollination. Other nearby populations of R. grallata (RG5) and R. sp.‘Cooper Ck’ (RC6) were not flowering simultaneously and were found to be genetically distinct (Figs 1, 2). Differences in the flowering time, both within a species over a short altitudinal gradient (RC) and between species growing over short geographical distances (RC6, RG5; 100 m), can potentially lead to reproductive isolation, and hence genetic isolation, over short distances of both species and populations within species.

Inbreeding and genetic structure

If the recorded reproductive activity is representative of typical levels, one might expect an inverse relationship between reproductive activity levels (R) and the level of inbreeding (F). The R. sp.‘Cooper Ck’ population at Mt Sorrow (RC6) is the largest of the known populations of this species, and had a higher proportion of the population participating in reproductive activity than the other two populations. This population (RC6) was also significantly less inbred (F = 0.476; P < 0.05; Table 4) than the other two populations, possibly indicating that reproductive activity is consistently higher within the population. However, reproductive activity (R) and inbreeding (F) were not significantly correlated in R. strobilacea populations (Spearmans Rho = 0.008, P = 0.982), nor overall within the genus if the data were pooled (P > 0.05).

The observed heterozygosity (Ho) within populations differed significantly amongst the four species (χ2 = 15.9, P < 0.001). Romnalda sp.‘Cooper Ck’ and R. grallata had significantly higher values than R. papuana and R. strobilacea (Table 1). Populations of all species showed high allelic fixation and were inbred (F), but there were significant differences amongst species (Table 4; P < 0.05). Despite the highest genetic diversity and the highest reproductive levels reported in the genus, R. grallata populations also had the highest level of inbreeding (F = 0.73), significantly (P < 0.05) greater than that of R. papuana, which had the least inbred populations (F = 0.48) but the lowest genetic diversity (Tables 1, 4). This indicates that gene flow via pollen appears to be less successful in R. grallata than in R. papuana. This is contrary to what might be expected. The results also indicate that increased inbreeding in this genus is not associated with increasing species rarity (smaller geographical extent, smaller species population; Table 4; Fig. 1).

As the species were able to regenerate vegetatively, it was also of interest whether populations were primarily clonal and if this could have led to the high allelic fixation observed. In most populations of R. grallata and R. sp.‘Cooper Ck’, the multilocus genotypes of individuals were unique, indicating that these populations were not primarily clonal in origin. However, some evidence of potential clonal spread was found in the R. sp.‘Cooper Ck’ population (RC4), in which there were several examples of multiple individuals with the same multilocus genotype, confirming that vegetative spread had occurred on a limited scale. In several populations of R. strobilacea, there was considerable evidence for potential vegetative spread of individual genotypes, although many different genotypes were present, indicating that clonal spread was localized and sexual reproduction was also important. It is unlikely that increased allelic fixation primarily reflects increased clonal spread, as F was not significantly (P > 0.05) correlated with density (D) in R. strobilacea populations.

If the high inbreeding levels (F) observed were a result of either vegetative spread or the formation of family clusters, we should be able to detect spatial genetic structure within populations. There were significant, but weak, correlations between genetic distance and geographical distances amongst individuals in some populations in each of the four species (Table 4). These were most common amongst the R. papuana populations (two of four populations), which was the least inbred of the four species (Table 4). A significant positive spatial autocorrelation of like genotypes was recorded within some populations of three Romnalda species, but not within the R. sp.‘Cooper Ck’ populations (Table 4). The spatial size of significant genetic clumping of like individuals varied, but was usually restricted to short distances (15 m; Table 4), which would be consistent with local clonal spread or family clusters. The most extensive genetic clustering was recorded in RS1, one of the largest R. strobilacea populations (Fig. 4; Table 4). Beyond 15 m, spatial autocorrelation of genetically like individuals was either not significant or, at greater distances, became significantly negative (Fig. 4). There is anecdotal evidence that R. strobilacea is capable of self-pollination leading to fertile offspring when grown in cultivation (P. Bostoc, Queensland Herbarium, Brisbane, pers. comm.); however, it is unknown whether this is achieved by autogamy.

image

Figure 4. Examples of correlograms produced from spatial autocorrelation analysis within Romnalda populations, which plot the autocorrelation coefficient (r) against the distance in metres. These examples show significant positive spatial genetic structure at short distances in one population of each of the species: R. papuana (RP3), R. grallata (RG8), and R. strobilacea (RS1). Upper and lower error bars mark the 95% confidence intervals about the autocorrelation coefficient (r) as determined by bootstrap sampling (999 permutations). Upper (U) and lower (L) confidence limits mark the 95% confidence interval around the null hypothesis of no spatial structure for the combined multilocus data set as determined by permutation (999 permutations).

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIAL AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. REFERENCES

Ancient fragmentation, speciation, and diversity within the Romnalda genus

All extant species in the genus Romnalda appear to be restricted to rainforest habitats. Rainforests (megathermal moist forests) are thought to have evolved in the Cretaceous 100 Mya, and to have connected Australia and New Guinea from at least the Oligocene (approximately 30 Mya) to the mid-Miocene (approximately 13 Mya) when Australia and New Guinea were connected by land bridges (Morley, 2000; Maslin et al., 2005). Bowman & Yeates (2006) identified that the late Miocene to early Pliocene (7–10 Mya) period was critical in establishing the phylogenetic composition of modern Australian flora, being a cooler, drier period sandwiched between warmer, more humid environments. The association of R. strobilacea with Trapezites butterflies suggests a long history of mutualism (Atkins, 2004).

The Pickett et al. (2004) reconstruction of the SEAPAC region during the last glacial maximum, based on pollen records, found that the climate was drier and cooler than at present and that WTRF and TRFO forest types (where Romnalda occurs) were more restricted in distribution than at present. However, this restriction was mostly a reduction in range at higher altitudes, as these areas were more sensitive to climate change, and lowland rainforest persisted more extensively than previously thought (Pickett et al., 2004). However, the Pickett et al. (2004) reconstruction also found that the mid-Holocene climate was slightly warmer and moister than today, with more extensive occurrence of WTRF vegetation than today. This was mainly the result of a higher tree line and lower sea levels between New Guinea and Australia. Thus, the potential Romnalda habitat was probably slightly more continuous than at present.

Ladiges et al. (2003) compared the phylogeny and biogeography of the Australian Eucalyptus and Melaleuca groups in the Myrtaceae with geological history. These are also members of a Gondwanan family whose distribution extends to New Guinea and contains some species with rainforest affinities and similar geographical distributions to Romnalda. Ladiges et al. (2003) concluded that these distributions were related to the geological history of South-east Asia and Australia, and dispersal and colonization. This was aided by rafting on microcontinental fragments by accretion of arc terranes onto New Guinea and land brought in to closer proximity during periods of low sea level from the late Miocene and Pliocene. Heads (2001) used similar patterns to explain the distributions of birds of paradise, and employed the distribution of R. papuana in New Guinea to support this. Van Welzen (1997) also found a strong correlation between species distributions and groups of terranes of 961 endemic New Guinean plants. The highly dynamic geological history plays a significant role in explaining species distributions in New Guinea and may be significant for the highly disjunct Romnalda distribution.

Romnalda species seem to be found in less disturbed vegetation communities that are very high in species diversity, endemic species, and species thought to have relictual distributions. For example, the forests at Cape Tribulation and Mt Lewis have very high levels of species endemism and diversity (Tracey, 1981; Crisp et al., 2001). The Romnalda populations at these locations were also the most genetically diverse within the genus. Although Johns (1986) found that New Guinean forests had high rates of natural disturbance caused by, for example, earthquakes, landslips, and tree falls, R. papuana is found in less disturbed sites (P. Katik, pers. observ.). The forests in a 1 ha plot at 900 m altitude in New Guinea were found by Wright et al. (1997) to have one of the highest species diversity records for such a plot in the world. They attributed this to the frequent natural disturbance, rich soil, and mix of ancestry of the flora with both Gondwanan and Laurasian components (Wright et al., 1997). The areas inhabited by Romnalda appear not to have been affected greatly by past climate change, being continuously occupied by rainforest types, which is thought to have led to higher species diversity (Wright et al., 1997).

Pye & Gadek (2004), in their study of Bunya pine, Araucaria bidwillii, the distribution of which extends across the range of the Australian Romnalda species, found evidence of historical fragmentation and major genetic differences between the northern and southern populations. They found that Mt Lewis populations represented a significant reservoir for genetic diversity in Bunya pine (Pye & Gadek, 2004). The levels of genetic diversity within Romnalda are lowest at the southern and northern extremes of the genus distribution and highest at the centre of the distribution (which includes populations at Mt Lewis; Fig. 1).

The data from this study, together with the gross morphological characteristics (Stevens, 1978; Henderson, 1986), indicate clearly that there are at least four species in the genus Romnalda. Further studies using molecular markers, such as chloroplast DNA, internal transcribed spacer (ITS) sequencing, or microsatellites, are required to confirm the order of species differentiation from the ancestral Romnalda species. However, these initial results and observations point to isolation within the genus as a result of the ancient fragmentation of the three major historical rainforest blocks of New Guinea, north Queensland, and southern Queensland, leading to allopatric speciation and the extant species R. papuana, R. grallata, and R. strobilacea. The gross morphological differences suggest isolation and evolution in significantly different environments. The patterns of geographical distribution, genetic differentiation, and diversity within the genus support this theory (Table 1; Fig. 2).

Although the similarity of gross morphology suggests that R. sp.‘Cooper Ck’ is most closely related to R. papuana, the results of this study indicate that these two species are significantly genetically distinct (Fig. 2; Table 2). Further sampling of the additional known R. papuana populations is needed to determine the genetic relationships between these species more convincingly. The data from this study indicate that, although R. grallata and R. sp.‘Cooper Ck’ are clearly distinctive in gross morphology, they are genetically more closely related than is R. sp.‘Cooper Ck’ to R. papuana (Fig. 2). There is evidence that, at locations in which these species come into immediate geographical contact, there is hybridization and some evidence of introgression (Figs 2, 3). This supports the theory that R. grallata and R. sp.‘Cooper Ck’ originally speciated in geographical isolation and have subsequently come back into contact (Levin, 2000). This is consistent with the possibility that R. papuana migrated south from PNG, possibly during later periods of contact, and subsequently became isolated and differentiated whilst in Australia. Having evolved allopatrically, R. papuana and R. grallata would not necessarily have strong reproductive isolating mechanisms, apart from reproductive timing and habitat preferences. However, occasionally, such as in the Mt Sorrow and Mt Peiter Botte sites, the two species came into contact and some hybridization occurred, followed by introgression, as seen in this study (RG7, RC7). This produced further genetic differentiation of the Australian and New Guinean populations as genes from R. grallata found their way into R. papuana, increasing the differences between the Australian and New Guinean populations and eventually leading to R. sp.‘Cooper Ck’. Romnalda sp.‘Cooper Ck’ thus resulted from a second more recent speciation process, explaining the gross morphological similarity with R. papuana as well as the genetic relationship with R. grallata (Fig. 2). The fact that R. grallata is apparently restricted to altitudes above 600 m, whereas R. sp.‘Cooper Ck’ is found at both higher and lower altitudes, is consistent with the R. papuana climatic profile. Alternatively, R. grallata and R. sp.‘Cooper Ck’ may be sister taxa, possibly evolving by reproductive isolation along an altitudinal gradient, but are incompletely reproductively isolated and have formed a zone of hybridization. Further investigation of the relationships amongst these taxa using a combination of molecular markers would help to clarify the evolutionary history of these species.

Reproduction, hybridization, and diversity

The higher levels of genetic diversity found in R. sp.‘Cooper Ck’ than would be expected on the basis of its small population size and limited distribution (Table 1) may be explained by hybridization and introgression with R. grallata. Hybridization has long been recognized as a major phenomenon promoting genetic diversity in plants (Riesberg, 1997). Ainouche et al. (2003) found that hybridization was important in speciation in the genus Spartina (Poaceae). They found examples of hybridization when geographically separate species came into contact, even in vegetatively spreading species (Ainouche et al., 2003). It has been found that polyploidy is not required for sympatric speciation following hybridization, which can result from backcrossing to one of the parent species combined with sib mating (Riesberg & Carney, 1998). Hardig et al. (2000) found morphological and molecular evidence for hybridization and historic introgression, suggesting that perhaps hybridization occurred in glacial refugia as well as more contemporary hybridization. As these species appear to be associated with rainforest refugia, such scenarios would be consistent with the results of this study. Founder events, in which two species colonize a new location and hybridize, can rapidly form new species, as there is a low likelihood of backcrossing to either parental species populations because of isolation (Riesberg & Carney, 1998). Smissen, Breitwieser & Ward (2007) found an important role for small population size and rarity in the formation of hybrid lineages in New Zealand everlasting daisies. Such scenarios may also explain the Romnalda results.

Riesberg & Carney (1998) found that reproductive isolation is often asymmetrical. For example, some species are more often pollen donors than the maternal parent of hybrids because, if a species that is self-incompatible crosses with a species that is self-compatible, usually only the self-compatible parent can produce viable hybrid offspring. Our study found evidence of allelic spread between R. grallata and R. sp.‘Cooper Ck’ where they co-occurred, but these preliminary results suggest that hybridization is asymmetric with offspring more closely aligned with R. sp.‘Cooper Ck’ than with R. grallata. This is consistent with the findings of Smissen et al. (2007), who also found backcrossing mostly in one direction following hybridization. Our results also indicated that either mating amongst sibs, possibly via selfing, or backcrossing to R. sp.‘Cooper Ck’ had resulted in individuals that were homozygous for R. grallata alleles, thereby introducing new alleles into R. sp.‘Cooper Ck’ (Table 1; Fig. 3). Cogolludo-Agustin, Agundez & Gil (2000) also showed that hybridization led to greater genetic diversity in populations of elm.

Some endangered species have been identified as being under threat as a result of hybridization (Levin, Francisco-Ortega & Jansen, 1996). For example, Schnabel & Krutovskii (2004) have reported that an endangered tree (Gleditsia caspica) in Azerbaijan is threatened because of genetic introgression through hybridization with a related species that is widely cultivated; they found that the populations in one reserve consisted of hybrids. Similarly, Cogolludo-Agustin et al. (2000) found that the native Iberian elm (Ulmus minor) was under threat as a result of hybridization with exotic Siberian elms. Siberian elms and their hybrids are resistant to Dutch elm disease, which has led to their increased abundance in the landscape, but threatens the genetic integrity of native U. minor. They found asymmetric hybridization, in which the hybrids were nearer to the exotic species than to the local native U. minor, and that backcrossing occurred more frequently to the Siberian elm than to the Iberian elm. Given the small number of known individuals of R. sp.‘Cooper Ck’, there is a possibility that continued hybridization could lead to a loss of genetic identity, at least for one population in the future. Further studies utilizing maternally inherited markers, such as chloroplast DNA, would be useful to confirm the direction of hybridization between these species.

Reproductive isolation as a result of the timing of phenology between co-occurring congenetic species has been found to be quite common (Lamont et al., 2003). However, Lamont et al. (2003) found that disturbance altered the phenology of two co-occurring Banksia species, and this led to hybridization. They found that, in undisturbed sites, there was a phenological barrier to gene flow, whereas, in disturbed sites, B. hookeriana flowering was earlier and flowering in B. prionotes was prolonged, breaking the phenological barrier between the co-occurring species and resulting in hybrid swarms (Lamont et al., 2003). The results of the present study indicate that the timing of flowering within Romnalda follows a climatic gradient, with R. sp.‘Cooper Ck’ populations apparently responding to an altitudinal gradient. It appears that differences in flowering time reduce the potential for hybridization between R. grallata and R. sp.‘Cooper Ck’ at mid-altitude sites, but, at the higher altitude site (RC7, RG7), the phenological barrier is removed.

Although phenology is under strong genetic control, new models for the effects of climate change are predicting changes in the onset of flowering in some species (Chuine, Cambon & Comtois, 2000). Osborne et al. (2000) found that olive phenology was likely to change with increased climate warming, specifically leading to both earlier flowering times and greater spatial variation in timing of flowering amongst populations. Menzel et al. (2006) also found evidence for an earlier onset of spring phenological events and greater spatial variability in the timing of plant phenological stages. They predicted that differential changes in the timing of flowering may affect interactions amongst populations. Pickett et al. (2004) found that higher altitude sites were more greatly affected than lower altitude sites by climate change historically. Climate change may thus have a long-term impact on the potential for hybridization in R. sp.‘Cooper Ck’.

Preliminary observations have shown that reproductive synchronicity amongst populations within R. sp.‘Cooper Ck’ could be affected by climate (altitude). Thus, altitude together with distance could lead to genetic isolation of Romnalda populations, hence making small populations more prone to a loss of genetic diversity as a result of drift, and increasing the importance of intermediate altitude populations for gene flow amongst R. sp.‘Cooper Ck’ populations. Gomes et al. (2004) found a weak relationship between genetic variability and altitude in a rare member of the Asteraceae in Brazil. Most variation was found within populations, but 17% of variation was amongst populations from different altitudes. In R. sp.‘Cooper Ck’, there was some evidence of genetic differentiation along an altitudinal gradient. Thus, climate change has the potential to impact on both gene flow within the species and hybridization frequency, and may lead to either increased or decreased phenological synchronicity amongst populations or between species along an altitudinal gradient. Further studies to confirm the timing of flowering of these species along the altitudinal and climate gradients in different years would provide insights to clarify this.

The levels of genetic diversity in all species in the genus Romnalda were quite high compared with other endemic rainforest or herbaceous species (Hamrick & Godt, 1989; Shapcott, 2000; Honnay et al., 2005). Rossetto & Kooyman (2005) suggested that vegetative regeneration in rainforest plants may be a significant factor in slowing the rate of loss of diversity caused by drift as it enables the persistence of genotypes within the population. It has been observed that, under low light conditions, forest herbs may exhibit prolonged clonal growth (Honnay et al., 2005). Richards et al. (2004) found that, even in clonally reproducing species, sexual reproduction is often underestimated, but can be responsible for high diversity within patches assumed to be largely clonal. Sexual reproduction in partially clonal species is important for dispersal between populations (Honnay et al., 2005). The ability of plants to maintain themselves via vegetative means may have been significant in Romnalda to reduce the loss of diversity caused by drift in its small populations.

Populations that are small and isolated are expected to become more inbred, which may lead to a decline in reproductive output and hence population growth (Ellstrand & Elam, 1993). These features leave them more susceptible to demographic chance events. Despite high genetic diversity, all Romnalda species in this study were found to be highly inbred with generally low levels of synchronous reproductive activity (Tables 3, 4). Low levels of synchronous flowering are, however, not uncommon in understorey rainforest species (De Steven et al., 1987; Shapcott, 2000). Culley & Grub (2003) found that increased homozygosity as a result of a decrease in the population size of pollinators led to increased selfing in Viola. The results of the present study, together with anecdotal evidence, suggest that all species are self-compatible to some extent, although it is unknown whether viable seed can be produced by autogamy or whether the species are reliant on pollinators. In Romnalda, the high allelic fixation in the genus appears to be a combined effect of self-compatibility, weak spatial aggregations, and vegetative spread, but not associated with low diversity, as would be expected after founder or bottleneck events (Litrico et al., 2005).

Current conservation

Romnalda sp.‘Cooper Ck’ is currently undescribed. Its distribution is highly restricted and, at present, only three populations are known (Fig. 1); there are probably fewer than 500 plants. Thus, it deserves the highest conservation status of ‘endangered’. Given the inconspicuous nature of the species, it is possible that it has been missed from surveys and more populations are yet to be found. Currently, the RC7 population is potentially under long-term threat of introgression and loss of identity as a result of hybridization with R. grallata. However, given the size of the R. sp.‘Cooper Ck’ populations and their high levels of genetic diversity, these populations appear to be worthy of conservation and are likely to remain viable for some time.

Romnalda papuana has been recorded from only four confirmed sites in PNG and there is one very old record in Irian Jaya (RBG KEW). Two of the PNG sites are remote and inaccessible mountain tops and are likely to have remained reasonably undisturbed since the species collections in the 1970s (P. Katik, pers. observ.). Since this study was completed, an additional population in PNG has been identified, indicating that the species is likely to be more abundant than its current distribution record shows. However, expert opinion (P. Katik, pers. observ.) suggests that the species is not common and should be classified as ‘rare’ until further studies indicate a different conservation status. Given its geographical distribution within PNG, it seems likely that the populations from the disjunct regions are genetically distinct from each other. It also seems likely that the species occurs in clusters of small subpopulations and that these are likely to be associated with relictual areas of low disturbance. Conserving blocks of habitat containing R. papuana populations is likely to be the simplest and most effective means to ensure long-term survival of the species. Romnalda grallata populations contain the highest diversity in the genus and, at present, are known from populations in two disjunct protected areas. In both areas, the species co-occurs with many highly restricted endemic species. Romnalda strobilacea is classified as ‘vulnerable’ and, although more populations are known to exist of this species than of the others, they contain significantly less genetic diversity than the other Australian species (Table 1). The preliminary evidence suggests that the presence of Romnalda species is likely to indicate relictual rainforest which may contain other significant species. However, further studies are needed to confirm this hypothesis. Given that Romnalda species are much easier to observe and identify than most canopy tree species, the usual focus for conservation assessments, this genus may be a potentially useful indicator species of high conservation status rainforest.

ACKNOWLEDGEMENTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIAL AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. REFERENCES

This project was funded by the University of the Sunshine Coast. We would especially like to thank the following people for their assistance: R. Banka and R. Kiapranis (PNG FRI); W. Watt, M. Kuduk, O. Gebia, as well as many individuals who helped us from Bosavi and Kalip villages; M. Powell (USC) for maps; P. Forster (Qld Herbarium EPA); J. Dransfield and W. Baker (RBG KEW) for access to the Kew specimens; B. Conn and J. Croft for taxonomic and logistical advice; QPWS; Caloundra City Council; M. Russell and Barung Landcare for additional site locations and logistical advice; and many private landowners for their permission to access populations. Also, Janelle, Jo, Kathy, Ed, and Heather for field assistance in Australia; plus many others.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIAL AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. REFERENCES
  • Ainouche L, Baumel A, Salmon A, Yannic G. 2003. Hybridization, polyploidy and speciation in Spartina (Poaceae). New Phytologist 161: 165172.
  • Atkins A. 1975. The first record of Rachelia (C. and R. Felder) (Lepidoptera: Hesperiidae: Trapezitinae) in Australia. Journal of the Australian Entomological Society 14: 237241.
  • Atkins A. 2004. New larval foodplant records and notes on the biology of Trapezites symmomus Hubner, T. praxedes (Plotz), T. maheta (Hewitson) and Hesperilla sarina Atkins (Lepidoptera; Hesperiidae: Trapezitinae) from southeast Queensland. Australian Entomologist 31: 137140.
  • Bowman D, Yeates D. 2006. A remarkable moment in Australian biogeography. New Phytologist 170: 20212.
  • Brewbaker JL, Upadhya MD, Makinsen Y, MacDonald T. 1968. Isozyme polymorphism in flowering plants. III Gel electrophoretic methods and applications. Physiologia Plantae 21: 930940.
  • Brigham CA, Schwartz MW. 2003. Population viability in plants: conservation, management and modelling of rare plants. Ecological Studies 165. Wien: Springer.
  • Broadhurst L, Young A. 2007. Seeing the wood and the trees – predicting the future for fragmented plant populations in Australian landscapes. Australian Journal of Botany 55: 250260.
  • Brown AHD, Nevo E, Zohary D, Dagan O. 1978. Genetic variation in natural populations of wild barley (Hordeum spontaneum). Genetics 49: 97108.
  • Byer DL, Waller DM. 1999. Do plant populations purge their genetic load? Effects of population size and mating history in inbreeding depression. Annual Review of Ecology and Systematics 30: 479513.
  • Chuine I, Cambon G, Comtois P. 2000. Scaling phenology from the local to the regional level: advances from species-specific phenological models. Global Change Biology 6: 943952.
  • Coates DJ, Hopper SD. 2000. Genetics and conservation of Australian flora. Australian Journal of Botany 48: 13.
  • Coates DJ, Sampson JF, Yates CJ. 2007. Plant mating systems and assessing population persistence in fragmented landscapes. Australian Journal of Botany 55: 239249.
  • Cogolludo-Agustin MA, Agundez D, Gil L. 2000. Identification of native and hybrid elms in Spain using isozyme gene markers. Heredity 85: 157166.
  • Crisp MD, Laffan S, Linder HP, Monro A. 2001. Endemism in the Australian flora. Journal of Biogeography 28: 183198.
  • Culley TM, Grub TC. 2003. Genetic effects of habitat fragmentation in Viola pubescens (Violaceae) a perennial herb with chasmogamous and cleistogamous flowers. Molecular Ecology 12: 29192930.
  • De Steven D, Windsor DM, Putz FE, De Leon B. 1987. Vegetative and reproductive phenologies of a palm assemblage in Panama. Biotropica 19: 342356.
  • Ellstrand NC, Elam DR. 1993. Population genetic consequences of small population size: implications for plant conservation. Annual Review of Ecology and Systematics 24: 217242.
  • Environment Protection and Biodiversity Conservation Act. 1999. Canberra: Environmental Protection Agency.
  • Environmental Systems Research Institute. 2001. Arc/GIS version 8.1. Redlands, CA: Environmental Systems Research Institute.
  • Farris JS. 1972. Estimating phylogenetic trees from distance matrices. American Naturalist 106: 645658.
  • Fay MF, Rudall PJ, Sullivan S, Stobart KL, De Bruijn AY, Reves G, Qamaruz-Zaman F, Hong W-P, Joseph J, Hahn WJ, Conran JG, Chase MW. 2000. Phylogenetic studies of Asparagales based on four plastid DNA regions. In: WilsonKL, MorrisonDA, eds. Monocots: systematics and evolution. Melbourne: CSIRO, 360371.
  • Fitch WM, Margoliash E. 1967. Construction of phylogenetic trees. Science 155: 279284.
  • Frankham R. 1996. Relationship of genetic variation to population size in wildlife. Conservation Biology 10: 15001508.
  • Gentry AH. 1992. Tropical forest biodiversity: distributional patterns and their conservational significance. OIKOS 63: 1928.
  • Gentry AH, Emmons LH. 1987. Geographical variation in fertility, phenology, and composition of the understorey of tropical forests. Biotropica 19: 216227.
  • Gitzendanner MA, Soltis P. 2000. Patterns of genetic variation in rare and widespread plant congeners. American Journal of Botany 87: 783792.
  • Gomes V, Collevatti RG, Silveira FAO, Fernandes GW. 2004. The distribution of genetic variability in Baccharis concinna (Asteraceae) an endemic dioecious and threatened shrub of rupestrian fields of Brazil. Conservation Genetics 5: 157165.
  • Gottlieb LD. 1981. Electrophoretic evidence and plant populations. Progress in Phytochemistry 7: 146.
  • Guries RP, Ledig FT. 1978. Inheritance of some polymorphic isoenzymes in pitch pine (Pinus rigida Mill.). Heredity 40: 2732.
  • Hamrick JL, Godt MJ. 1989. Allozyme diversity in plant species. In: BrownAHD, CleggMT, KahlerAL, WeirBS, eds. Plant population genetics, breeding and genetic resources. Sunderland, MA: Sinauer Associates Inc., 4363.
  • Hanski I, Gilpin ME. 1997. Metapopulation ecology, genetics and evolution. San Diego, CA: Academic Press.
  • Hardig TM, Brunsfeld SJ, Fritz RS, Morgan M, Orians CM. 2000. Morphological and molecular evidence for hybridization and introgression in a willow (Salix) hybrid zone. Molecular Ecology 9: 924.
  • Heads M. 2001. Birds of paradise, biogeography and ecology in New Guinea: a review. Journal of Biogeography 28: 893925.
  • Henderson RJF. 1986. Romnalda. Flora of Australia 46: 9191.
  • Henderson RJF 2002. Names and distribution of Queensland plants, algae and lichens. Brisbane: Environmental Protection Agency.
  • Honnay O, Jacquemyn H, Bossurt B, Hermy M. 2005. Forest fragmentation effects on patch occupancy and population viability of herbaceous plant species. New Phytologist 166: 723736.
  • Hubbell SP. 2001. Unified neutral theory of biodiversity and biogeography. Princeton, NJ: Princeton University Press.
  • Johns RJ. 1986. The instability of the tropical ecosystem in New Guinea. Blumea 31: 341371.
  • Ladiges PY, Udovicic F, Nelson G. 2003. Australian biogeographical connections and the phylogeny of large genera in plant family Myrtaceae. Journal of Biogeography 30: 989998.
  • Lamont BB, He T, Enright NJ, Krauss SL, Miller BP. 2003. Anthropogenic disturbance promotes hybridization between Banksia species by altering their biology. Journal of Evolutionary Biology 16: 551557.
  • Legendre P, Fortin MJ. 1989. Spatial pattern and ecological analysis. Vegetatio 80: 107138.
  • Levene H. 1949. On a matching problem arising in genetics. Annals of Mathematics and Statistics 20: 9194.
  • Levin DA. 2000. The origin, expansion, and demise of plant species. New York: Oxford University Press.
  • Levin DA, Francisco-Ortega J, Jansen RK. 1996. Hybridization and the extinction of rare plant species. Conservation Biology 10: 1016.
  • Litrico I, Ronfort J, Verlaques R, Thompson JD. 2005. Spatial structure of genetic variation and primary succession in the pioneer tree species Anthirhea borbonica on La Reunion. Molecular Ecology 14: 15751584.
  • Mantel N. 1967. The detection of disease clustering: a generalised regression approach. Cancer Research 27: 209220.
  • Maslin M, Malhi Y, Phillips O, Cowling S. 2005. New views on an old forest; assessing the longevity, resilience and future of the Amazon rainforest. Transactions of the Institute of British Geographers 30: 477499.
  • Menzel A, Sparks TH, Estrella N, Roy DB. 2006. Altered geographic and temporal variability in phenology in response to climate change. Global Ecology and Biogeography 15: 498504.
  • Moran GF, Bell JC. 1983. Eucalyptus. In: TankersleySD, OrtonTJ, eds. Isozymes in plant genetics and breeding, Part B. Amsterdam: Elsevier Science Publishers, 423441.
  • Morley RJ. 2000. Origin and evolution of tropical rain forests. Chichester: Wiley.
  • Murray BR, Lepschi B. 2004. Are locally rare species abundant elsewhere in their geographical range? Austral Ecology 29: 287293.
  • Murray BR, Thrall PH, Gill M, Nicotra AB. 2002. How plant life history and ecological traits relate to species rarity and commonness at varying spatial scales. Austral Ecology 27: 291310.
  • Osborne CP, Chuine I, Viner D, Woodward FI. 2000. Olive phenology as a sensitive indicator of future climatic warming in the Mediterranean. Plant, Cell and Environment 23: 701710.
  • Peakall R, Smouse PE. 2005. GenAlEx V6: genetic analysis in excel. Population genetic software for teaching and research. Canberra: Australian National University.
  • Pickett EJ, Harrison SP, Hope G, Harle K, Dodson JR, Kershaw AP, Prentice IC, Backhouse J, Colhoun EA, D'Costa D, Flenley J, Grindrod J, Haberle S, Hassell C, Kenyon C, Macphail M, Martin H, Martin AH, McKenzie M, Newsome JC, Penny D, Powell J, Raine JI, Southern W, Stevenson J, Sutra J-P, Thomas I, Van Der Kaars S, Ward J. 2004. Pollen-based reconstructions of biome distributions for Australia, Southeast Asia and the Pacific (SEAPAC region) at 0, 6000 and 18000 14C yr BP. Journal of Biogeography 31: 13811444.
  • Prager EM, Wilson AC. 1976. Congruency of phylogenies derived from different proteins. A molecular analysis of the phylogenetic position of cracid birds. Journal of Molecular Evolution 9: 4557.
  • Pye MG, Gadek PA. 2004. Genetic differentiation and conservation in Araucaria bidwillii (Arucariaceae), Australia's Bunya pine. Conservation Genetics 5: 619629.
  • Richards CL, Hamrick JL, Donovan LD, Mauricio R. 2004. Unexpectedly high clonal diversity of two salt marsh perennials across a severe environmental gradient. Ecology Letters 7: 11551162.
  • Riesberg LH. 1997. Hybrid origins of plant species. Annual Review of Ecology and Systematics 28: 359389.
  • Riesberg LH, Carney SE. 1998. Tansley review No. 102 plant hybridization. New Phytologist 140: 599624.
  • Rossetto M, Kooyman R. 2005. The tension between dispersal and persistence regulates the current distribution of rare paleo-endemic rain forest flora: a case study. Journal of Ecology 93: 906917.
  • Schnabel A, Krutovskii KV. 2004. Conservation genetics and evolutionary history of Gleditsia caspica; inferences from allozyme diversity in populations from Azerbaijan. Conservation Genetics 5: 195204.
  • Shapcott A. 2000. Conservation and genetics in the fragmented monsoon rainforest in the northern territory, Australia: a case study of three frugivore dispersed species. Australian Journal of Botany 48: 397407.
  • Shapcott A. 2002. Conservation genetics and ecology of the endangered rainforest shrub Triunia robusta from the Sunshine Coast, Australia. Australian Journal of Botany 50: 93105.
  • Shapcott A. 2007. Does species range and rarity affect population genetics? A case study of four Graptophyllum species from Queensland, Australia. Biotropica 39: 447458.
  • Shaw CR, Prasad R. 1970. Starch gel electrophoresis of enzymes – a compilation of recipes. Biochemical Genetics 4: 297320.
  • Smissen RD, Breitwieser I, Ward JE. 2007. Genetic characterization of hybridization between the New Zealand everlastings Helichrysm lanceolatum and Anaphalioidies bellidioides (Asteraceae; Gnaphalieae). Botanical Journal of the Linnean Society 154: 8998.
  • Smouse PE, Long JC. 1992. Matrix correlation analysis in anthropology and genetics. Yearbook of Physiology and Anthropology 35: 187213.
  • Smouse PE, Long JC, Sokal RR. 1986. Multiple regression and correlation extensions of the Mantel test of matrix correspondence. Systematic Zoology 35: 627632.
  • Smouse PE, Peakall R. 1999. Spatial autocorrelation analysis of individual multiallele and multilocus genetic structure. Heredity 82: 561573.
  • Sneath PHA, Sokal RR. 1973. Numerical taxonomy: the principals and practice of numerical classification. San Francisco, CA: Freeman Press.
  • Stevens P. 1978. An anagram of Lomandra. Journal of the Arnold Arboretum 59: 148.
  • Swofford DL, Selander RK. 1981. BIOSYS-1: a FORTRAN program for the comprehensive analysis of electrophoretic data in population genetics and systematics. Journal of Heredity 79: 281283.
  • Tracey JG. 1981. Australia's rainforest: where are the rare plants and how do we keep them? In: SyngeH, ed. The biological aspects of rare plant conservation. Chichester: John Wiley and Sons Ltd., 165178.
  • Van Welzen P. 1997. Increased speciation in New Guinea: tectonic causes? In: DransfieldJ, CoodeMJE, SimpsonDA, eds. Plant diversity in Melasia III. Kew: Royal Botanic Gardens Kew, 363387.
  • Webb LJ, Tracey JG. 1981. Australian rainforest pattern and change. In: KeastA, ed. Ecological biogeography of Australia. The Hague: Junk, 605694.
  • Wendel JF, Weeden NF. 1989. Visualization and interpretation of plant isozymes. In: SoltisDE, SoltisP, eds. Isozymes in plant biology. London: Chapman and Hall, 545.
  • Wright DD, Jessen JH, Burke P, Garza HG. 1997. Tree and liana enumeration and diversity on a one-hectare plot in Papua New Guinea. Biotropica 29: 250260.
  • Wright S. 1965. The interpretation of population structure by F-statistics with special regard to systems of mating. Evolution 19: 395420.
  • Wright S. 1978. Evolution and the genetics of populations Vol. 4, variability within and among natural populations. Chicago, IL: University of Chicago Press.
  • Young AG, Clarke GM. 2000. Genetics, demography, and viability of fragmented populations. Cambridge: Cambridge University Press.