Abstract
Figs and fig-pollinating wasps are obligate mutualists that require each other to complete sexual reproduction. However, landscapers can establish populations of fig trees outside their native ranges by propagation through exported seeds, seedlings or cuttings. Once mature, these trees could be colonized by pollinating wasps and/or various non-pollinating wasps that also develop in figs. In recent decades, the Australian endemic Ficus rubiginosa has been planted widely in the Mediterranean region and in parts of the USA. Observation of ripe fruit production suggested that a pollination mutualism has been re-established by pollinating wasps colonizing trees in the plant’s introduced range. We therefore used sampling of pollinators from mainland Spain, Tenerife and California (USA) and molecular studies to characterize the restructured mutualism and compare it with the native range. In the native range, the plant is pollinated by five wasp species that form the Pleistodontes imperialis complex. However, all wasps in the introduced ranges belonged to just one of these species (P. imperialis sp. 1). Moreover, their mtDNA diversity was close to zero and the sequences clearly link them with the native southern population of this species. None of the > 20 non-pollinating wasp species from the native range were found in the introduced ranges. In summary, the restructured mutualism has been dramatically simplified, lacking all non-pollinating wasps and all but one pollinator species from the native range. Moreover, the one pollinator species to establish successfully shows a drastic reduction in genetic diversity relative to its source population.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
Introduction
Symbiosis is a key factor in the evolution of global biodiversity, and many important ecosystem functions, such as pollination and nitrogen fixation depend strongly on symbiotic interactions between species. Although some symbiotic interactions can be diffuse, involving loose sets of species, many others are intimate and highly partner-specific (Leigh Jr 2010). In such cases, the ability of a given species to establish and successfully reproduce in a new range depends heavily on the presence and availability of viable partner species. A classic example is the obligate mutualism between fig trees (Ficus species) and their tiny pollinating wasps in the family Agaonidae (Hymenoptera) (Cook and Rasplus 2003).
There are over 750 Ficus (Moraceae) species and each is pollinated by only a small number (often one, but up to five) of highly host-specific agaonid wasp species (Wiebes 1979; Cook and Rasplus 2003; Darwell et al. 2014). The wasps are the only pollen vectors for the figs, and the wasp offspring can develop only inside fig fruits, where each wasp larva feeds on one fig flower. Ficus is a pantropical genus and figs often act as keystone species by providing resources for a wide variety of frugivorous animals (Shanahan et al. 2001). Figs are also highly valued culturally and used extensively in plantings of parks, roadsides and gardens throughout the warmer parts of the world. This has led to several species being planted extensively outside their natural ranges, including on different continents. Most introduced populations of fig trees also harbour fig wasp species. However, the high host specificity of the wasps, and the fact that most Ficus sections are restricted to particular continents or biogeographic regions, act as major barriers to establishment of new pollination mutualism partnerships. Nevertheless, in some cases, the known pollinators of a given fig species have been deliberately introduced following the establishment of mature fig trees in a new country (e.g. Pemberton 1934; reviewed in McKey 1989).
Ficus rubiginosa Desf. ex. Vent. is endemic to eastern Australia. It is a popular species in the horticultural trade, both within and outside of Australia. In fact, it has been planted widely around the globe, in coastal regions of Spain and in the Canary Islands, as well as in the USA, New Zealand, Italy and Israel (Gardner and Early 1996; Starr et al. 2003). For a long time, it was thought to be pollinated by a single species, Pleistodontes imperialis Saunders (see Lopez-Vaamonde et al. 2002), but Haine et al. (2006) used cytochrome b sequences to reveal four well-differentiated mitochondrial lineages that were suggested to be separate species. Darwell et al. (2014) subsequently identified a complex of five P. imperialis species, supported by data from two mitochondrial markers and the nuclear internal transcribed spacer 2 (ITS2) locus. Interestingly, Darwell et al. (2014) also showed that P. imperialis sp. 1 has disjunct northern and southern populations, separated by > 1000 km. This study was supported by a more recent population genetic study (Sutton et al. 2016), which suggests that the two disjunct P. imperialis sp. 1 populations could represent incipient species.
In this study, we take advantage of the remarkable microcosm of the fig to study reassembly of a symbiotic interaction outside of its natural range. Specifically, we explore the species and genetic diversity of wasps pollinating the Australian endemic F. rubiginosa in its introduced range in mainland Spain, the Canary Islands and San Diego, USA to determine (1) the wasp species present and (2) their likely source region.
Materials and methods
Sample sites and specimen collection
Mature fig fruits were collected from Ficus rubiginosa trees in the introduced range in mainland Spain (Carboneras, Cartagena, Málaga, Mojácar, Moraira and Murcia), Tenerife (Canary Islands), and San Diego (USA) between March 2005 and June 2011. Pleistodontes imperialis wasps either emerged naturally from the fruits or were dissected out from figs that were first preserved in ethanol. All wasps were stored in 95% ethanol at − 20 °C.
DNA extraction
DNA was extracted from wasps using a Chelex extraction procedure detailed in West et al. (1998). A single whole female wasp representing each sample was ground up using a pipette tip and added to a 100 μl Chelex (Bio Rad) solution. Tubes were centrifuged for 2 min at 13,000 rpm and incubated for 3 h at 56 °C, then 15 min at 96 °C. Tubes were then vortexed for 10 s, centrifuged again for 2 min at 13,000 rpm, and the supernatant stored in a new tube for PCR.
Cytochrome b sequencing
A 444 bp segment of mitochondrial cytb was amplified using the CB1/CB2 primers (Jermiin and Crozier 1994). PCRs were performed in a Touchgene Gradient thermocycler (Techne) in 22 μl reactions consisting of 2 μl DNA template, 1X Biomix (Bioline), 1.5 μM MgCl2 and 0.5 μM of each forward and reverse primer. The PCR cycling conditions were: 3 min at 95 °C followed by 30 cycles of 15 s at 94 °C, 20 s at 45 °C and 30 s at 72 °C, and a final extension of 72 °C for 5 min. All PCR products were purified and sequenced by Macrogen (Korea) using the same primers employed in the original PCR.
ITS2 sequencing
The nuclear ITS2 region was sequenced for five Spanish wasps (including wasps from both mainland Spain and Tenerife) and four wasps from San Diego using the ITS2F/ITS2R primers (White et al. 1990). In addition, we sequenced ITS2 for eight other wasps collected previously from the native range of F. rubiginosa (Sutton et al. 2016, 2017). DNA was amplified in 22 μl reactions, comprising 2 μl DNA template, 1X Biomix (Bioline), 0.5 μM MgCl2, and 0.35 μM of each forward and reverse primer. Thermal cycling conditions were as follows: 5 min at 94 °C followed by 35 cycles of 30 s at 94 °C, 40 s at 55 °C and 40 s at 72 °C, and a final extension of 72 °C for 10 min. PCR products were sequenced by Macrogen.
Data analyses
DNA sequences were aligned and edited using BioEdit v. 7.0.9.0 (Hall 1999). Cytochrome b and ITS2 haplotype phylogenies were constructed using MrBayes v. 3.2 (Ronquist et al. 2012) to assign the introduced wasps to one of the five species identified by Darwell et al. (2014). Reference sequences representing each of the five species were included in each phylogeny (Darwell et al. 2014). The most appropriate substitution model for each data set was determined using jModelTest v. 2 (Guindon and Gascuel 2003; Darriba et al. 2012). The cytb haplotype tree was constructed using the symmetrical model (Zharkikh 1994) with gamma distributed rate variation. The ITS2 haplotype tree was constructed using the Kimura 2-parameter model (Kimura 1980). We performed two independent runs of four Metropolis coupled Monte Carlo Markov chains for 10 million generations, sampling every 1000 generations and with a relative burn-in period of 25%. It was assumed the two runs had converged if the standard deviation of the split frequencies of the two runs was below 0.01.
We used DnaSP v. 5 (Librado and Rozas 2009) to estimate nucleotide diversity (π) and gene diversity for introduced and native P. imperialis sp. 1 (south and north populations; Sutton et al. 2016). We also calculated Tajima’s D (Tajima 1989) and Fu’s Fs (Fu 1997) to test for deviations from equilibrium (e.g. demographic expansion, selection).
Results
Cytochrome b phylogeny and diversity
Sequences of 333 bp were obtained for 97 individual wasps from various locations in mainland Spain, Tenerife and San Diego (GenBank accession numbers KJ796700–KJ796788, MG025546–MG025553). Six unique sequences were found for wasps from the introduced ranges. Three haplotypes could be determined fully at all positions, and differed from each other by either one or two base pairs (i.e. 0.3 to 0.6% sequence divergence). The other three haplotypes had double peaks at a small number of sites (7/333; 2.1%). As these haplotypes were almost identical to the others at unambiguous positions, we do not present further analysis of them here. All three fully resolved introduced haplotypes were placed unambiguously within P. imperialis species 1; i.e. only one of the five native pollinator species was found in the introduced range. Moreover, the introduced wasps all clustered with P. imperialis sp. 1 sequences from the native southern Australian population (Fig. 1). One haplotype (Introduced01) was dominant in mainland Spain (57/61 wasps; 93%) and Tenerife (23/31 wasps; 74%). This was also the most common haplotype in P. imperialis sp. 1 (south) in the native range (22/100 wasps; 22%), as well as being the only haplotype in the five San Diego samples. Two haplotypes (Introduced01 and Introduced02) were found in Málaga, Spain, and all three introduced haplotypes were found in Tenerife.
Nucleotide and gene diversity measures were estimated after assignment of individuals to one of the three true haplotypes. Introduced Spanish wasps were far less diverse than sp. 1 in either the south or north populations in the native range (Table 1). Species 1 (south) had the greatest nucleotide (0.008 ± 0.005) and gene (0.925 ± 0.017) diversity, whereas individuals from San Diego had zero diversity. Significant negative values of D and Fs indicate a recent population expansion in the sp. 1 populations in the native range, although no such changes in population size were detected in the introduced wasps.
ITS2 phylogeny
There was considerable base ambiguity in the middle portion of the fragment sequenced, perhaps due to the existence of multiple paralogous ITS sequences, as has been reported for several animal taxa. Consequently, sequences were trimmed from 344 bp to 138 bp, for which alignment and comparison was unambiguous (see Online Resources 1–3 for alignments). All pollinator species were resolved clearly; only one nuclear ITS2 haplotype was identified for the nine wasps from Spain and San Diego. Although the native Australian southern and northern sp. 1 populations group together in the tree that includes other species in the complex, individuals could be assigned to population simply based on a two-bp indel at sequence position 113 (Fig. 2; Online Resource 4). From this, it was apparent that all samples from the introduced range shared the same haplotype as the southern P. imperialis sp. 1 population, providing nuclear DNA support for the above results using mitochondrial cyt b.
Discussion
The reassembly of symbiotic interactions between figs and their pollinating wasps in introduced ranges is critical for the production of ripe figs with seeds and hence, for fig reproduction. This is also important for wider ecological interactions, as figs can provide important year-round resources for diverse frugivores (McKey 1989; Shanahan et al. 2001). Frugivores can aid the spread of plants in introduced ranges and, if environmental conditions support seedling recruitment, introduced figs may become invasive. For example, the seeds of Ficus microcarpa, native to Asia but cultivated widely, can be spread long distances by birds in its introduced range (Williams 1939). Consequently, it has become an aggressive invader in Hawaii, where its pollinator has also been introduced, and is considered to pose a high threat to native flora (Starr et al. 2003). Ficus rubiginosa has also been introduced in several countries and its fruits attract local frugivores in at least parts of its introduced range (Starr et al. 2003). Both pollinator wasps (Gardner and Early 1996) and some non-pollinating wasps (Early 2000) have been reported in F. rubiginosa figs in New Zealand, along with observations of seedlings. Interestingly, Gardner and Early (1996) argued that pollinator wasps probably reached New Zealand by natural long-distance dispersal from Australia, making this case clearly different to wasp arrivals in northern hemisphere locations such as California or Spain.
Our study revealed a severe reduction in pollinator diversity in the F. rubiginosa pollination mutualism following its introduction to Spain, Tenerife and San Diego. We detected only one of the five cryptic Pleistodontes imperialis obligate pollinator species, and all individuals (N = 97) were assigned to the sp. 1 southern population. Given the small sample size obtained from San Diego, however, we cannot rule out the coexistence of other P. imperialis species. Although we do not know the history of wasp introductions in Spain and San Diego, P. imperialis previously became locally established in Hawaii just 1 year following its release (Anonymous 1924). Pollination mutualisms have also been established in some other countries where F. rubiginosa has been planted, e.g. in New Zealand (Gardner and Early 1996), as well as Israel (D. Eisikowitch, pers. comm.), but the identity, diversity and genetic status of the pollinating wasps are currently uncertain. Nuclear ITS2 sequences have been shown to provide a reliable complement to the mitochondrial cytb in fig wasps (e.g. Darwell et al. 2014; Sutton et al. 2016), and in our study both ITS2 and cytb phylogenies support the placement of the introduced wasps with the southern population (New South Wales) of P. imperialis sp. 1 (Sutton et al. 2016), which is the most likely source population for the introduced wasps. Interestingly, if pollinators have reached New Zealand by natural long-distance dispersal (Gardner and Early 1996), we would predict them to be P. imperialis sp. 1 since this is the only species found throughout most of the host plant’s range in New South Wales (e.g. Darwell et al. 2014; Sutton et al. 2016), with the other four species being found farther north, mainly in Queensland.
We detected far lower genetic diversity in wasps in the introduced range than in either of the two sp. 1 populations in the native range. In fact, there was hardly any variation in the introduced populations, although the low genetic diversity revealed in the San Diego wasps may be partly attributable to small sample size. Species 1 south and north populations in the native range showed evidence for a recent population expansion, but the introduced wasps did not. As the introduced wasp populations are probably less than 100 years old, possibly much less, there may have been insufficient time for mutation to replace mtDNA variation that was lost through the colonization process. Ornamental F. rubiginosa trees have been planted in many locations along the Spanish coastline, and represent a fragmented habitat, although some areas already have large numbers of both trees and pollinators. However, the initial populations of pollinators would probably have been bottlenecked at colonization and then subjected to genetic drift and subsequent loss of genetic diversity following introduction (Frankham et al. 2010).
The Spanish and San Diego wasps had almost identical cytb haplotypes, differing by no more than 0.6%. By comparison, we found sp. 1 south and north populations in the native range had up to 3.3 and 5.1% intraspecific cytb divergences, respectively, and a maximum divergence between wasps from these two different populations of 7.5%. It is therefore likely that the introduced pollinators in all regions studied came from the same area in the south of the native range of the host plant. It is also possible that the wasps went through serial colonization events outside of Australia after one original export from the native range. Further studies targeting populations in other countries (e.g. Israel, Italy, New Zealand) and using a wider range of genetic markers (e.g. microsatellites) could be used to test this hypothesis. However, we also cannot exclude the possibility that more than one species (or population) of pollinator was translocated, but that the others have become extinct or have not been sampled. The latter seems unlikely, however, given the broad geographic sampling area of this study and the fact that different members of the P. imperialis complex co-exist in the same sites and even figs in parts of the native range (Sutton et al. 2017).
The fig/pollinator mutualism is also exploited by often diverse communities of non-pollinating fig wasps (NPFW) that develop in fig fruits as gall-inducing seed predators or parasitoids. These parasitic species may (West and Herre 1994) or may not (Bronstein 1991) be in direct competition with fig wasps for fig flowers, but either way, are important components of the system and influence its coevolution (Kerdelhué and Rasplus 1996). Dunn et al. (2008), for example, concluded that parasitism pressure has contributed to pollinator preference for oviposition in central fig ovules. This preference allows outer ovules to develop into fig seeds and, thus, allows for both male and female reproductive functions of the fig to persist.
Introduced species often leave their natural enemies behind (Torchin et al. 2003), and NPFW may also be present in lower densities in introduced ranges (Wang et al. 2015). In our study, we found no NPFW, although in its native range F. rubiginosa hosts over 20 NPFW species (Segar et al. 2013; Darwell and Cook 2017), several of which parasitize the pollinator galls. A few of these species have colonized New Zealand, probably by natural long-distance aerial dispersal (Early 2000), but (as for the pollinators) wasp arrivals in northern hemisphere populations would seemingly require human mediation. This makes an interesting comparison with F. microcarpa, which is probably the fig species planted most widely around the world as an urban tree. Many F. microcarpa populations beyond its native range of Asia and Australasia now have a few of the 43 wasp species associated with this host plant (Wang et al. 2015). In particular, the pollinator and two phytophagous species are often the first wasps recorded in introduced populations. This may reflect the fact that they are common and widespread in the native range and, therefore, the species most likely to be present in transported figs (Wang et al. 2015). In addition, none of these three species (in contrast to e.g. a parasitoid) requires the presence of any other wasp species in order to establish a new population.
The other way for introduced plants to gain insect associates is when local species adopt the new host plant. There are major barriers to this process in fig wasps due to the importance of coevolved insect and plant traits (Cook and Segar 2010), such that it is most likely when the introduced range includes closely related fig species with similar wasps. In a global survey of F. microcarpa, new colonists were very rare, occurring only occasionally in one country (South Africa) that had local figs from the same subgenus (Wang et al. 2015). On this basis, rapid recruitment of new wasp species by F. rubiginosa via host shifts seems very unlikely in the northern hemisphere.
References
Anonymous (1924) Pleistodontes imperialis Saund. Proc Hawaii Entomol Soc 5:200
Bronstein JL (1991) The nonpollinating wasp fauna of Ficus pertusa: exploitation of a mutualism? Oikos 61:175–186
Cook JM, Rasplus J-Y (2003) Mutualists with attitude: coevolving figs wasps and figs. Trends Ecol Evol 18:241–248
Cook JM, Segar ST (2010) Speciation in fig wasps. Ecol Entomol 35:54–66
Darriba D, Taboada GL, Doallo R, Posada D (2012) jModelTest 2: more models, new heuristics and high-performance computing. Nat Methods 9:772
Darwell CT, Cook JM (2017) Cryptic diversity in a fig wasp community—morphologically differentiated species are sympatric but cryptic species are parapatric. Mol Ecol 26:937–950
Darwell CT, al-Beidh S, Cook JM (2014) Molecular species delimitation of a symbiotic fig-pollinating wasp species complex reveals extreme deviation from reciprocal partner specificity. BMC Evol Biol 14:189
Dunn DW, Segar ST, Ridley J, Chan R, Crozier RH, Yu DW, Cook JM, Agrawal AA (2008) A role for parasites in stabilising the fig-pollinator mutualism. PLoS Biol 6:e59
Early JW (2000) Fig wasps (Hymenoptera: Agaonidae and Torymidae) in New Zealand. N Z Entomol 23:27–32
Frankham R, Ballou JD, Briscoe DA (2010) Introduction to conservation genetics. 2nd ed. Cambridge University Press, New York
Fu YX (1997) Statistical tests of neutrality of mutations against population growth, hitchhiking and background selection. Genetics 147:915–925
Gardner RO, Early JW (1996) The naturalisation of banyan figs (Ficus spp., Moraceae) and their pollinating wasps (Hymenoptera: Agaonidae) in New Zealand. N Z J Bot 34:103–110
Guindon S, Gascuel O (2003) A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood. Syst Biol 52:696–704
Haine ER, Martin J, Cook JM (2006) Deep mtDNA divergences indicate cryptic species in a fig-pollinating wasp. BMC Evol Biol 6:83
Hall TA (1999) BioEdit: a user-friendly biological sequence alignment editor and analysis program for Windows 95/98/NT. Nucleic Acids Symp Ser 41:95–98
Jermiin LS, Crozier RH (1994) The cytochrome-b region in the mitochondrial DNA of the ant Tetraponera rufoniger—sequence divergence in the hymenoptera may be associated with nucleotide content. J Mol Evol 38:282–294
Kerdelhué C, Le Clainche I, Rasplus JY (1999) Molecular phylogeny of the Ceratosolen species pollinating Ficus of the subgenus Sycomorus sensu stricto: biogeographical history and origins of the species-specificity breakdown cases. Mol Phylogenet Evol 11:401–414
Kerdelhué C, Rasplus J-Y (1996) Nonpollinating Afro-tropical fig wasps affect the fig-pollinator mutualism in Ficus within the subgenus Sycomorus. Oikos 75:3–14
Kimura M (1980) A simple method for estimating evolutionary rates of base substitutions through comparative studies of nucleotide sequences. J Mol Evol 16:111–120
Leigh EG Jr (2010) The evolution of mutualism. J Evol Biol 12:2507–2528
Librado P, Rozas J (2009) DnaSP Version 5: a software for comprehensive analysis of DNA polymorphism data. Bioinformatics 25:14511452
Lopez-Vaamonde C, Dixon DJ, Cook JM, Rasplus JY (2002) Revision of the Australian species of Pleistodontes (Hymenoptera: Agaonidae) fig-pollinating wasps and their host-plant associations. Zool J Linn Soc 136:637–683
Lopez-Vaamonde C, Rasplus JY, Weiblen GD, Cook JM (2001) Molecular phylogenies of fig wasps: partial cocladogenesis of pollinators and parasites. Mol Phylogenet Evol 21:55–71
McKey D (1989) Population biology of figs: application for conservation. Experientia 45:661–673
Pemberton CE (1934) Fig-wasps established on Kauai. Proc Hawaii Entomol Soc 8:379
Ronquist F, Teslenko M, van der Mark P, Ayres DL, Darling A, Höhna S, Larget B, Liu L, Suchard MA, Huelsenbeck JP (2012) MrBayes 3.2: efficient Bayesian phylogenetic inference and model choice across a large model space. Syst Biol 61:539–542
Segar ST, Pereira RAS, Compton SG, Cook JM, Mooers A (2013) Convergent structure of multitrophic communities over three continents. Ecol Lett 16:1436–1445
Shanahan M, So S, Compton SG, Corlett R (2001) Fig-eating by vertebrate frugivores: a global review. Biol Rev 76:529–572
Starr F, Starr K, Loope L (2003) Ficus microcarpa (Chinese banyan). United States Geological Survey - Biological Resources Division, Haleakala Field Station, Maui, Hawaii, USA
Sutton TL, DeGabriel JL, Riegler M, Cook JM (2017) Local coexistence and genetic isolation of three pollinator species on the same fig tree species. Heredity 118:486–490
Sutton TL, Riegler M, Cook JM (2016) One step ahead: a parasitoid disperses farther and forms a wider geographic population than its fig wasp host. Mol Ecol 25:882–894
Tajima F (1989) Statistical method for testing the neutral mutation hypothesis by DNA polymorphism. Genetics 123:585–595
Torchin ME, Lafferty KD, Dobson AP, McKenzie VJ, Kuris AM (2003) Introduced species and their missing parasites. Nature 421:628–630
Wang R, Aylwin R, Barwell L et al (2015) The fig wasp followers and colonists of a widely introduced fig tree, Ficus microcarpa. Insect Conserv Divers 8:322–336
West SA, Herre EA (1994) The ecology of the New World fig-parasitizing wasps Idarnes and implications for the evolution of the fig-pollinator mutualism. Proc R Soc Lond B Biol Sci 258:67–72
West SA, Cook JM, Werren JH, Godfray HCJ (1998) Wolbachia in two insect host-parasitoid communities. Mol Ecol 7:1457–1465
White TJ, Bruns T, Lee S, Taylor J (1990) Amplification and direct sequencing of fungal ribosomal genes for phylogenies. In: Innis MA, Gelfand DH, Sninsky JJ, White TJ (eds) PCR protocols: a guide to methods and applications. Academic Press, London, pp 315–322
Wiebes J (1979) Co-evolution of figs and their insect pollinators. Annu Rev Ecol Syst 10:1–12
Williams FX (1939) Note on Eupristina verticillata Waterston, first recovery on Oahu. Proc Hawaii Entomol Soc 10:194
Zharkikh A (1994) Estimation of evolutionary distance between nucleotide sequences. J Mol Evol 39:315–329
Acknowledgements
We thank Joshua Kohn for providing wasp samples for analysis, and to two anonymous reviewers for their helpful comments on the manuscript. EJM was supported by a Natural Environment Research Council (UK) Postgraduate Scholarship. TLS was supported by an Australian Postgraduate Award.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors declare that they have no conflict of interest.
Electronic supplementary material
Below is the link to the electronic supplementary material.
Rights and permissions
About this article
Cite this article
Morgan, E.J., Sutton, T.L., Darwell, C.T. et al. Restructuring of a mutualism following introduction of Australian fig trees and pollinating wasps to Europe and the USA. Biol Invasions 20, 3037–3045 (2018). https://doi.org/10.1007/s10530-018-1775-3
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10530-018-1775-3