Management Implications
Woody encroachment is an important issue in grasslands globally. Although ecologists are usually more concerned with invasions by alien species into a foreign habitat, native woody species can aggressively invade grasslands. The encroaching of native woody species, like Juniperus virginiana var. virginiana (eastern redcedar; hereafter ERC), threatens native flora, reduces species diversity, changes the quality and quantity of light reaching the understory, and affects ecosystem functions. To understand how native species expand their ranges, managers should know whether population growth is driven by the reproductive activity of previously established trees or by the continual arrival of propagules from distant source populations. If seeds are transported successfully over long distances, rapid dispersal of invasions from small satellite populations can occur and increase the rate of range expansion for the species. On the other hand, if seeds travel only short distances, managers can focus control efforts on the edge of the range to prevent the establishment of woody encroachers in intact grassland ecosystems. As clearing satellite populations before individuals become sexually mature is the most cost-effective form of management, understanding range expansion patterns provides a framework for better prioritization of management efforts.
In the case of ERC, our data indicate that encroachment results primarily from local seed production and intermediate-distance seed dispersal by animals, with occasional arrival of propagules via long-distance dispersal facilitated by humans. Therefore, adequate management requires consistent monitoring of grasslands to detect and eradicate newly established foci before initiating seed production with a particular focus on grasslands near other populations of ERC. Because the youngest recorded cone-producing female tree in our study was 6 yr old, managers could survey grasslands and eradicate ERC every 5 yr to most efficiently prevent the establishment of new ERC stands. Furthermore, managers should restrict the use of known invasive woody plants like ERC for windbreaks and landscaping. In addition, we recommend that managers limit seed source usage in nurseries, avoiding using seeds of distant origin. Limiting seed sources minimizes the risk of intraspecific hybridization among different ecotypes, which may broaden the range of environmental conditions suitable for the species and thus promote the local expansion of their ranges.
Introduction
There is ample evidence that native and nonnative species invasions have similar impacts on biodiversity (Sagoff Reference Sagoff1999; Schnelle Reference Schnelle2019; Simberloff et al. Reference Simberloff, Souza, Nunez, Barrios-Garcia and Bunn2012; Yazlik and Ambarli Reference Yazlik and Ambarli2022). While most invasive species are not native, there are native species that have become too abundant and have been implicated as drivers of recent extinctions (Blackburn et al. Reference Blackburn, Bellard and Ricciardi2019). Some native invaders became problematic after intracontinental movements outside their native ranges into new suitable habitats. However, other species expand their ranges from the edges and become increasingly dominant because of their prolific fruit production, seed dispersal by birds and mammals, anthropogenic disturbances, overgrazing of pastures, and tolerance of environmental extremes (Aththanayaka et al. Reference Aththanayaka, Siyasinghe, Prakash, Bloch and Surasinghe2023; Schnelle Reference Schnelle2019). Range expansion by woody native plants has been documented globally (Gettys and Schnelle Reference Gettys and Schnelle2018; Negi et al. Reference Negi, Maletha, Pathak and Maikhuri2021; Schnelle Reference Schnelle2019; Ward et al. Reference Ward, Pillay, Mbongwa, Kirkman, Hansen and Van Achterbergh2022).
Woody encroachment by native species is an important issue in grasslands globally resulting from overgrazing by cattle, anthropogenic interventions, and reduced fire frequency (Briggs et al. Reference Briggs, Hoch and Johnson2002; Eldridge et al. Reference Eldridge, Bowker, Maestre, Roger, Reynolds and Whitford2011; Ratajczak et al. Reference Ratajczak, Nippert and Collins2012; Ward Reference Ward2020). Encroachment of woody plants into grasslands reduces the species diversity and changes the quality and quantity of light reaching the understory (Ratajczak et al. Reference Ratajczak, Nippert and Collins2012). In addition, woody encroachment modifies nutrient cycling, ecosystem productivity, soil chemistry, natural disturbance regimes, and local hydrology (Donovan et al. Reference Donovan, Burnett, Bielski, Birgé, Bevans, Twidwell and Allen2018; Eldridge et al. Reference Eldridge, Bowker, Maestre, Roger, Reynolds and Whitford2011; Ratajczak et al. Reference Ratajczak, Nippert and Collins2012; Ward Reference Ward2020). Climate change may also exacerbate the problem, as the increase in global CO2 levels may benefit C3 woody species more than C4 grasses, allowing for further grassland transition to shrubland (Eldridge et al. Reference Eldridge, Bowker, Maestre, Roger, Reynolds and Whitford2011; Schnelle Reference Schnelle2019; Tunnell et al. Reference Tunnell, Stubbendieck, Huddle and Brollier2004).
Eastern redcedar (Juniperus virginiana L. var. virginiana; hereafter ERC) is a widespread native tree in North America (Ward Reference Ward2021). ERC is currently expanding its range out of its traditional habitat and into open areas like grasslands (Briggs et al. Reference Briggs, Hoch and Johnson2002; Vasiliauskas and Aarssen Reference Vasiliauskas and Aarssen1992; Ward Reference Ward2021). It is commonly used as a windbreak surrounding agricultural fields and can aggressively colonize neighboring grasslands, rapidly transforming them into closed-canopy ERC forests (Briggs et al. Reference Briggs, Hoch and Johnson2002; Donovan et al. Reference Donovan, Burnett, Bielski, Birgé, Bevans, Twidwell and Allen2018). Although ERC’s historical niche is limestone soils and cliffsides, it can survive in many environments and thrives in xeric environments where competition with other plants is reduced (Lawton and Cothran Reference Lawton and Cothran2000; Sangüesa-Barreda et al. Reference Sangüesa-Barreda, García-Cervigón, García-Hidalgo, Rozas, Martín-Esquive, Martín-Carbaja, Martínez and Olano2021; Ward Reference Ward2020). In addition, ERC is an animal-dispersed species, and seeds can be dispersed away from the mother tree, allowing further movement into new areas (Holthuijzen and Sharik Reference Holthuijzen and Sharik1985; Horncastle et al. Reference Horncastle, Hellgren, Mayer, Engle and Leslie2004). Like other successful invaders, ERC trees become sexually mature early in their lives; female ERC trees produce many seed-containing fleshy cones when they are 10 yr old, and males develop pollen-producing cones when they are 6 yr old, leading to rapid reproduction and dispersal (Aronson et al. Reference Aronson, Handel and Clemants2007; Van Haverbeke and Read Reference Van Haverbeke and Read1976; Wickert et al. Reference Wickert, O’Neal, Davis and Kasson2017; K Shvach, personal communication). Moreover, increasing severe droughts brought on by climate change could give ERC an even more decisive competitive advantage over native grassland plant species (Kaskie et al. Reference Kaskie, Wimberly and Bauman2019).
One area affected by ERC invasion is Lakeside Daisy State Nature Preserve (LDSNP). Located in Ohio on the Marblehead Peninsula on the shore of Lake Erie, LDSNP is a unique preserve, including extensive wetlands and a small prairie traditionally dominated by short grasses and wildflowers (Figure 1). This preserve was likely historically maintained by its xeric environment and shallow soil, limiting competition with the native flora by other plant species. Observation of historical satellite imagery reveals clumps of established trees in the 1990s, just shortly after the preserve’s founding in 1988 (Ohio Department of Natural Resources n.d.). Possible reasons for the establishment of ERC trees at LDSNP could include urbanization of the peninsula and subsequent landscaping of privately owned land using ERC or anthropogenic disturbance caused by limestone quarrying at the site. Additionally, ERC trees populate the sides of highways and hills surrounding LDSNP and could be dispersing into the preserve from these areas. ERC’s affinity for xeric environments where competition with other plants is low and shade is minimal has allowed this population to rapidly colonize the prairie at LDSNP following its introduction (Lawton and Cothran Reference Lawton and Cothran2000; Sangüesa-Barreda et al. Reference Sangüesa-Barreda, García-Cervigón, García-Hidalgo, Rozas, Martín-Esquive, Martín-Carbaja, Martínez and Olano2021; Ward Reference Ward2020).
Figure 1. Lakeside Daisy State Nature Preserve is located on the eastern end of Marblehead Peninsula in northern Ohio. The sampling area (i), where the population of encroaching Juniperus virginiana var. virginiana is located, is approximately 22 acres out of the entire (ii) 136-acre preserve. Created using ArcGIS (ESRI 2011).
Grasslands like LDSNP are considered among the most endangered ecosystems globally (Kaskie et al. Reference Kaskie, Wimberly and Bauman2019; Leis et al. Reference Leis, Blocksome, Twidwell, Fuhlendorf, Briggs and Sanders2017). Changes in plant community composition linked to woody encroachment can modify the belowground biomass of these ecosystems, reducing native grasses’ soil-stabilizing effects (Watson et al. Reference Watson, Alexander and Moczygemba2019). In semiarid environments like grasslands, soil stabilization is critical to prevent erosion and retain the little available water (Eldridge et al. Reference Eldridge, Bowker, Maestre, Roger, Reynolds and Whitford2011; Kaur et al. Reference Kaur, Joshi and Will2020; Knapp et al. Reference Knapp, Chen, Griffin-Nolan, Baur, Carroll, Gray, Hoffman, Li, Post, Slette, Collins, Luo and Smith2020). Additionally, the fragmentation of grasslands resulting from anthropogenic influence and woody species encroachment can result in biodiversity losses (Leis Reference Leis, Blocksome, Twidwell, Fuhlendorf, Briggs and Sanders2017). Up to 330 million ha of grasslands were experiencing shrub encroachment in 2011, negatively affecting many of the economic and ecological resources that grasslands possess (Eldridge et al. Reference Eldridge, Bowker, Maestre, Roger, Reynolds and Whitford2011).
Two theoretical models describe how invasions typically occur: outward diffusion from the edge of the range and long-distance dispersal followed by local expansion (Auld and Coote Reference Auld and Coote1980; Campbell and Dooley Reference Campbell and Dooley1992; Gorchov et al. Reference Gorchov, Castellano and Noe2014). Diffusion of plant species is characterized by seed and pollen traveling short distances from the edge of the range and may result in local adaptation due to isolated reproduction in different areas (Auld and Coote Reference Auld and Coote1980; Campbell and Dooley Reference Campbell and Dooley1992; Keller et al. Reference Keller, Chhatre and Fitzpatrick2017). On the other hand, long-distance dispersal may have a homogenizing effect on genetic diversity, because it allows for sharing of genetic material through pollen and seeds between geographically separate areas of the range (Barriball et al. Reference Barriball, McNutt, Gorchov and Rocha2015; Campbell and Dooley Reference Campbell and Dooley1992). Additionally, long-distance dispersal may lead to an increased rate of spread, as new adult trees in an area can facilitate the deposition of more seeds by providing perches for birds, the main dispersers of some woody encroachers (Higgins et al. Reference Higgins, Richardson and Cowling2000; Holthuijzen and Sharik Reference Holthuijzen and Sharik1985; Horncastle et al. Reference Horncastle, Hellgren, Mayer, Engle and Leslie2004). If seeds are transported successfully over long distances, range expansion can occur even more rapidly for the species (Moody and Mack Reference Moody and Mack1988). It is therefore essential to identify patterns of a range expansion in order to model and predict them more accurately. Better prediction of spread rate and pattern can give managers the ability to target management efforts on new colonies before they expand further (Gorchov et al. Reference Gorchov, Castellano and Noe2014; Higgins et al. Reference Higgins, Richardson and Cowling2000; Moody and Mack Reference Moody and Mack1988).
Using neutral genetic markers provides one way to infer the historical seed dispersal rates and patterns (Hamrick and Trapnell Reference Hamrick and Trapnell2011). Dispersal patterns of seeds and pollen affect the genetic diversity and structure of plant populations, which is a driver of plant evolution and determines the population’s response to selection pressures (Chybicki and Oleksa Reference Chybicki and Oleksa2018; Sork and Smouse Reference Sork and Smouse2006). Therefore, studying the changes in the genetic structure over time can provide a more detailed understanding of how a newly established population of ERC persists over time (Céspedes et al. Reference Céspedes, Gutierrez, Holbrook and Rocha2003; Roser et al. Reference Roser, Ferreyra, Saidman and Vilardi2017). We suggest that higher genetic diversity allows the population to respond to environmental variation and may affect the success of ERC trees in different environmental conditions (Fuchs et al. Reference Fuchs, Robles and Hamrick2013; Gonzales et al. Reference Gonzales, Hamrick, Smouse, Trapnell and Peakall2009). We used genetic similarity among ERC trees over a temporal scale in the LDSNP to address the question: Is ERC’s pattern of range expansion driven by diffusion from the edge of the range or by long-distance dispersal events? Answering this question will help us estimate patterns of dispersal and colonization of ERC and inform a model of ERC’s range expansion in grasslands and other open areas. Such models can help determine where to prioritize control measures and proactively manage expanding populations (Donovan et al. Reference Donovan, Burnett, Bielski, Birgé, Bevans, Twidwell and Allen2018; Kaskie et al. Reference Kaskie, Wimberly and Bauman2019).
Materials and Methods
Study Site
The LDSNP is at the eastern end of the Marblehead Peninsula on Lake Erie in Ohio, USA (41.53°N, 82.73°W) (Figure 1). The 55-hectare preserve was established to protect the only naturally occurring population of the federally threatened lakeside daisy (Tetraneuris herbacea Greene) in the United States and includes an old limestone quarry of the Marblehead geological series (Ohio Department of Natural Resources n.d.). Our study focuses on the 9-ha parcel of prairie habitat with a high presence of ERC (Figure 1, i), because much of the rest of the 155-ha preserve (Figure 1, ii) is made up of wetlands where ERC does not have a dominant presence. ERC presence in this parcel of prairie habitat threatens the native flora of LDSNP, including T. herbacea. Mean annual precipitation ranges from 686 to 914 mm, and mean annual air temperature ranges from 7 to 11 C.
Sampling Protocol
Leaf samples from 189 ERC trees from LDSNP were collected and stored in resealable plastic bags in a −20 C freezer until DNA extraction. We sampled 170 ERC trees using a grid-like pattern from the north to the south end of the 9-ha sampling area. Eleven sampling transects run from east to west every 50 m, and we collected leaf tissue from the closest tree every 10 m along each transect. The transect length varied, as we stopped the transect when we reached the wetland where ERC trees were absent. Overall, the sampled area had a perimeter of 1,389 m, with transect two being the longest (24 trees sampled). Tree height, coordinates, presence or absence of female cones, and diameter at breast height (DBH) for tall trees or base diameter for short trees were noted. The youngest tree found bearing female cones, indicating sexual maturity, was determined to be 6 yr old.
Additionally, 9 large trees found in a clump in one portion of the 9-ha sampled area and 10 large trees along the roadside on the edge of the sampling area were targeted as candidates for founding members of the population at LDSNP. These trees were sampled outside the grid sampling scheme, but they were sampled in the same way and included in our data analysis together with the rest of the trees sampled. LDSNP is located in the less-developed eastern end of the Marblehead Peninsula and is surrounded by a town with private residences (Figure 1). Large trees can be seen in the town of Marblehead, on other islands close to LDSNP, and along the highways in northeast Ohio, but these were located on privately owned property and therefore could not be sampled.
Wood core samples were collected using an increment borer from 45 trees for which a DBH measurement was taken; meanwhile, stem cross sections were collected from 50 trees that had a base diameter taken. Cores and cross sections were frozen in a −20 C freezer until processed. The annual rings in each tree core or stem cross section were counted under a dissecting microscope to determine the age of each tree. A core or cross section could not be obtained for all trees, as some trees were missing the tags that were placed during the initial sampling and could not be reliably located using GPS coordinates. Linear relationships between stem diameter and age of the tree, based on growth ring count, were used to estimate the age of those trees for which we did not have a core (41 tall trees) or stem cross section (53 small trees). ERC growth rates are highly variable across sexes and microsite conditions, especially once trees are sexually mature (Quinn and Meiners Reference Quinn and Meiners2004), so this estimation may have resulted in miscalculation of the age of trees and may have contributed to the disproportionate sample sizes of each age group. The trees were grouped into five age groups of 10-yr increments rather than on a continuous age scale to help account for any error resulting from age estimation.
DNA Extraction and Microsatellite Analysis
Total genomic DNA was extracted from the leaf tissue of each ERC tree using a modification of the CTAB protocol described by Doyle and Doyle (Reference Doyle and Doyle1987). Subsequently, all DNA samples were diluted to approximately 10 ng µl−1 for polymerase chain reaction (PCR) amplification. PCR was performed using eight microsatellite markers, seven of these were developed in our lab for genetic analysis of ERC and one was developed for genetic analysis of common juniper (Juniperus communis L.) (Michalczyk et al. Reference Michalczyk, Sebastiani, Buonamici, Cremer, Mengel, Ziegenhagen and Vendramin2006) (Table 1). PCR was performed in a final volume of 10 µl, containing approximately 15 ng of genomic DNA, 10 mM Tris buffer with KCl, pH 8.8, 1.88 mM MgCl2, 0.2 mM dNTPs, 0.2 µM of each primer, and 1 unit of Taq polymerase using a PTC-200 thermal cycler (MJ Research, Watertown, MA, USA). The PCR was conducted using a touchdown annealing approach to improve the specificity of primer binding to the target DNA. Annealing temperature decreased by 1 C every three cycles from 60 to 56 C during the first 15 cycles, then annealing temperature remained at 55 C for the remaining 30 cycles. The thermocycling profile consisted of initial denaturation at 94 C for 2 min, followed by 15 cycles of 94 C for 30 s, 60 to 56 C for 30 s, 72 C for 30 s, followed by 30 cycles of 94 C for 30 s, 55 C for 30 s, 72 C for 30 s, a final extension of 72 C for 3 min, and a holding temperature of 5 C. All forward primers were labeled using the fluorescent dyes FAM or TET. The PCR product size was determined using an ABI 3730 DNA sequencer (Applied Biosystems, Waltham, MA, USA) at MC Lab (San Francisco, CA, USA). The program Peak Scanner was used to determine the multilocus genotype of all trees.
Table 1. Locus name, oligonucleotide primer sequences, repeat motifs, and PCR product size range for each of the seven microsatellite loci for genetic analysis of Juniperus virginiana var. virginiana and one microsatellite locus developed for Juniperus communis.
Temporal Variation in Genetic Diversity
The program GenAlEx (v. 6.4; Peakall and Smouse Reference Peakall and Smouse2006) was used to determine the levels of genetic variation found within and among ERC trees from different age categories. This program was used to calculate common indicators of genetic diversity such as the number of alleles per locus (N a), effective number of alleles (N e), number of unique alleles in each age group, and observed and unbiased expected heterozygosity (H o and H e, respectively). N e indicates the number of equally frequent alleles that it would take to achieve the same H e in each age group, and alleles unique to an age group are present only in that age group. Deviations from Hardy-Weinberg equilibrium (HWE) and linkage disequilibrium were determined using Genepop (v. 4.7; Raymond and Rousset Reference Raymond and Rousset1995; Rousset Reference Rousset2008).
We also used GenAlEx to estimate Wright’s F coefficients and to conduct an analysis of molecular variance (AMOVA) to assess the distribution of genetic diversity among ERC established each decade to determine levels of differentiation between age groups (Peakall and Smouse Reference Peakall and Smouse2006). Because genetic diversity estimates are positively correlated with sample size, we obtained 30 random subsamples of 10 individuals for age groups 10 to 19, 20 to 29, and 30 to 39 to compare to the 0 to 9 and 40 and greater (40+) age groups in order to account for the disparity in sample size among age groups in our study (Ward and Jasieniuk Reference Ward and Jasieniuk2009). After random subsampling, we calculated N a, N e, the number of unique alleles, H o, H e, and pairwise genetic differentiation index (F st) among all age groups. In addition, we conducted one-way ANOVAs using the 30 random samples taken from age groups 10 to 19, 20 to 29, and 30 to 39 to determine whether there were significant differences in the means of N a, N e, and H o among groups. We use pairwise post hoc Tukey tests among age groups for these three diversity indicators. We also used GenAlEx to conduct principal coordinate analysis (PCoA) to visualize the genetic relationships among trees of different age groups.
Genetic Structure
The genetic structure of ERC was analyzed with the use of Bayesian model–based clustering with the program STRUCTURE (v. 2.3.4; Hubisz et al. Reference Hubisz, Falush, Stephens and Pritchard2009; Pritchard et al. Reference Pritchard, Stephens and Donnelly2000). This analysis was conducted to determine whether there are changes in the genetic structure among age groups. This program predicts the most likely number of subpopulation clusters for the populations sampled and recalculates F-statistics. All STRUCTURE runs used a burn-in length of 100,000 followed by 1,000,000 Markov chain Monte Carlo repetitions. A graphic representation of the results generated by STRUCTURE was presented using the program DISTRUCT (v. 1.1; Rosenberg Reference Rosenberg2003). Options were selected to allow admixture, assume independence among loci, and ignore population affiliations when defining clusters. To determine the likeliest number of subpopulation clusters (K), we followed the methodology proposed by Evanno et al. (Reference Evanno, Regnaut and Goudet2005). All probable K values were run 20 times to obtain ΔK, which is an ad hoc measure based upon the second-order rate of change of the likelihood function with respect to each K value (Evanno et al. Reference Evanno, Regnaut and Goudet2005). According to this procedure, the modal value of the ΔK can be used as an indicator of the number of ancestral population clusters in the area. The program STRUCTURE HARVESTER (v. 6.0; Earl and von Holdt Reference Earl and von Holdt2012) was used for calculating parameters of Evanno et al. (Reference Evanno, Regnaut and Goudet2005).
Results and Discussion
Levels of Genetic Diversity
The results of this study showed that the genetic diversity of a population of ERC in LDSNP is increasing over time. We found a total of 63 alleles among the 189 ERC trees examined using eight microsatellite loci. After random subsampling, average N a values ranged from 2.125 ± 0.227 in the 40+-yr-old group to 3.250 ± 0.453 in the 0- to 9-yr-old group (Figure 2; Table 2). N e values showed a similar trend, with averages ranging from 1.671 ± 0.153 in the 40+-yr-old group to 2.079 ± 0.221 in the 0- to 9-yr-old group (Figure 2; Table 2). There is a significant overall effect of age group on N a, N e, and H o (MANOVA, Wilks’s lambda = 0.1188, F(8, 288) = 68.46, P < 0.0001). ANOVA and Tukey’s honest significant difference (HSD) tests revealed significant differences in mean values of N a and N e between different age groups of ERC (N a: F(4, 145) = 197.515; P < 0.0001) (N e: F(4, 145) = 46.538; P < 0.0001) (Figure 2). Our findings indicate that, considering all trees with no subsampling, the average number of unique alleles per tree increased from 0 in the oldest group to 0.44 in the youngest group (Table 3). When random subsampling was conducted in age groups with more than 10 individuals, we found that the number of unique alleles in each age group increased from 0.033 unique alleles in the 40+-yr-old group to 3.667 unique alleles in the 10- to 19-yr old group (Table 2). Although there were no unique alleles in the 40+-yr-old age group when considering all trees in the data set, our random subsampling of the data set revealed that there is there a low likelihood of finding a unique allele among the oldest group of trees (unique allele = 0.033 ± 0.033) (Tables 2 and 3). N e values increased only slightly among all ages, but the difference between the mean N e values of the youngest group and the older groups was significant (Figure 2). These findings suggest that, although there is a greater number of unique alleles found in younger populations, the new alleles are in low frequency within all populations, suggesting the arrival of new variants over time (Table 2).
Figure 2. Average number of alleles (N a) and average number of effective alleles (N e) per locus for all age groups of Juniperus virginiana var. virginiana estimated from 30 random subsamples of age groups 10–19, 20–29, and 30–39. Values for N a and N e for age groups 0–9 and 40+ were calculated using all individuals (N = 9 and N = 5, respectively). Sample size disparity among age groups may impact N a and N e values. Error bars correspond to standard errors resulting from ANOVA. Letters indicate significantly different means determined using Tukey’s honest significant difference (HSD) pairwise comparisons.
Table 2. Sample size after subsampling (N), number of alleles (N a), number of effective alleles (N e), observed heterozygosity (H o), expected heterozygosity (H e), number of unique alleles, and fixation index (F) across all loci for each Juniperus virginiana var. virginiana age group.
Table 3. Number of unique alleles present in each locus (JV1–JV7, JC1) for each Juniperus virginiana var. virginiana age group and total number of unique alleles per age group, considering all individuals in each age group.
Our analysis also showed that two of the eight loci are not in HWE in the sample as a whole, namely, JV4 and JV10 (Hardy-Weinberg exact test P-values = 0.0014 ± 0.0014 and 0.0095 ± 0.0050, respectively). Departure from HWE may occur due to various causes, including purifying selection, inbreeding, population substructure, copy-number variation, or genotyping error. For example, deviations from HWE are likely to occur in expanding populations and structured populations such as ERC in the LDSNP (Chen et al. Reference Chen, Cole and Grond-Ginsbach2017; Meisner and Albrechtsen Reference Meisner and Albrechtsen2019). Moreover, genotype frequency deviation in finite random-mating populations may also result from the difference between the gene frequencies of male and female gametes, which is determined by two independent causes: the gene frequency difference between male and female parents and the sampling error due to the finite number of offspring (Wang Reference Wang1996). Given that this population was recently established and that seeds from nearby trees continue to arrive at LDSNP, it is reasonable to expect that the population will not be in equilibrium.
Similarly, there are indications of linkage disequilibrium among loci JV4 and JV7 (P-values = 0.0256 ± 0.0064) and loci JV11 and JV8 (P-values = 0.0460 ± 0.0127). Disequilibrium in expanding populations can also result from the arrival of individuals from different provenances or regions with unusual genomic patterns, which can distort the inference of the genetic structure of the population (Liu et al. Reference Liu, Wu and Wang2022). Furthermore, we found that pairs of loci showing linkage disequilibrium were inconsistent among age groups.
The results of this study help to elucidate the historical patterns of encroaching of ERC in the LDSNP and provide insight into how dispersal and genetic admixture can contribute to its range expansion. Our findings show that despite the size of the study area, there is considerable genetic diversity in the ERC population. Furthermore, our findings suggest that genetic diversity has continued to build over the last four to five decades. Because genetic diversity is essential for population survival and adaptation to changing environments, our findings suggest that increasing genetic diversity during ERC invasion of open areas could be a critical driver for its range expansion.
Ward et al. (Reference Ward, Gaskin and Wilson2008) reviewed published studies that examined the changes in allelic diversity of alien species expanding in their new range. The invasion processes of nonnative species imply an initial founder effect into a novel range; therefore, the new populations typically have lower diversity relative to populations in their native ranges (Uller and Leimu Reference Uller and Leimu2011; Ward et al. Reference Ward, Gaskin and Wilson2008; Zhao et al. Reference Zhao, Solís-Montero, Lou and Vallejo-Marín2013; Zimmermann et al. Reference Zimmermann, Ritz, Hirsch, Renison, Wesche and Hensen2010). Moreover, environmental filtering also causes diversity loss in alien species after arrival at their new locations, further lowering genetic diversity (Dar et al. Reference Dar, Bhat, Khuroo, Verna and Islam2020; Dormontt et al. Reference Dormontt, Gardner, Breed, Rodger, Prentis and Lowe2014; Vyšniauskienė et al. Reference Vyšniauskienė, Rančelienė, Žvingila and Patamsytė2011). In contrast, native species that become invasive do not always experience loss of genetic diversity due to environmental filtering (Negi et al. Reference Negi, Maletha, Pathak and Maikhuri2021). Native invasive species, like ERC, have more sources of genetic diversity close to their invasive range than most nonnative invasive species and are therefore more likely to experience intraspecific hybridization with other ecotypes, which allows the species to cope with stresses resulting from invading new habitat (Castillo et al. Reference Castillo, Schaffner, van Wilgen, Manuel Montaño, Bustamante, Cosacov, Mathese and Le Roux2021; Glisson and Larkin Reference Glisson and Larkin2021; Negi et al. Reference Negi, Maletha, Pathak and Maikhuri2021).
Changes in the genetic diversity of native species increasing their ranges by invading new environments have received little attention. Our results show that changes in genetic diversity can occur within a short time, with the average number of alleles per locus increasing by 65% in three decades (Table 2; Figure 2). Excoffier et al. (Reference Excoffier, Foll and Petit2009) examined the genetic consequences of range expansion. They proposed that range expansions could promote the surfing of rare variants into newly occupied territories. Our findings revealed the accumulation of unique low-frequency alleles in the LDSNP as time progressed (Tables 2–4). As expected for native species experiencing range expansion, our findings indicate that the newly established population was not isolated from other nearby populations, which explains the arrival of additional genetic diversity in the population. We have observed nearby populations of ERC on the Marblehead Peninsula, with some less than 1 km away, and on surrounding islands, including Kelley’s Island, approximately 8 km away.
Table 4. List of unique allele size (bp) found in each Juniperus virginiana var. virginiana age group, considering all individuals in each age group.
Genetic Structure
The AMOVA revealed that only 1% molecular variance was among age groups. Our analysis also showed that genetic differentiation among age groups increased slightly as the age difference between age groups increased (Table 5). We found that H o was highest in the youngest age group (H o = 0.581 ± 0.119) and lowest in the oldest age group (H o = 0.458 ± 0.107) (Table 2). While this trend is interesting, it is possible that the smaller sample size of the oldest group contributed to the lower H o value (Ward and Jasieniuk Reference Ward and Jasieniuk2009). On the other hand, individuals in the older age groups may have died off over time; therefore, we may see an underrepresentation of genetic diversity for that age group. Although the H o values do not show a clear trend over time, our analysis using ANOVA and Tukey’s HSD test revealed significantly different mean H o values for the 0 to 9, 10 to 19, 20 to 29, and 40+ age groups, but the 30 to 39 age group was not significantly different from either the 20 to 29 or 40+ age groups (F(4, 145) = 60.117; P < 0.0001). The significant increase of H o seen in the youngest two age groups further indicates the increase in genetic diversity in younger trees. Overall, we saw an excess in heterozygosity and a low level of differentiation among age groups (F st = 0.037 ± 0.009) (Table 2). An excess in heterozygosity in a population indicates little biparental inbreeding is occurring in a population. Accompanied by the fact that ERC is a dioecious species, it is likely that an influx of new lineages is invading LDSNP.
Table 5. Pairwise F st values demonstrating similarity between pairs of age groups of Juniperus virginiana var. virginiana.
We used the Bayesian model-cluster program STRUCTURE to describe the genetic structure among age groups. Our analysis showed that the genetic structure could be best described by four genetically distinct clusters (K = 4) (Figure 3A). However, our analysis showed that all four clusters are present in all age groups, and all clusters showed a similar contribution to the genetic diversity of all age groups (Figure 3B). Moreover, this finding suggests that all individuals originated from previously admixed populations of the four clusters. Despite the similar contribution of all lineages in all age groups, a PCoA shows two distinct groups of trees when the first two axes are plotted (Figure 4). It is worth noting that individuals from all age groups are represented in both groups in the PCoA in similar proportions (Figure 4). Principal coordinates 1 and 2 explain 16.8% and 11.3% of the variation, respectively.
Figure 3. STRUCTURE estimates of cryptic population structure of the Juniperus virginiana var. virginiana in the Lakeside Daisy State Nature Preserve. (A) Calculation of the second-order rate of change (ΔK), determined by the modal peak. The modal peak for natural populations is at K = 4. (B) STRUCTURE plot of ancestral subpopulations from the natural populations, with different colors representing the four population clusters and each line on the x axis representing a single individual arranged in age groups, with the percent of its genome identified by the y axis (Cluster 1: yellow; Cluster 2: blue; Cluster 3: green; Cluster 4: red).
Figure 4. The principal coordinate analysis (PCoA) of (A) individual Juniperus virginiana var. virginiana and (B) age groups at Lakeside Daisy State Nature Preserve based on eight polymorphic microsatellite markers.
Genetic diversity in invasive plant populations accumulates through multiple introductions, gene flow, mutation, and hybridization among plants of different origins (Carr et al. Reference Carr, Hooper and Dukes2019; Espeland Reference Espeland2013; Gaskin and Schaal Reference Gaskin and Schaal2002; Jeschke and Starzer Reference Jeschke and Starzer2018). However, the results of the STRUCTURE analysis show that all individuals at LDSNP contain the same admixture of four genetic clusters, suggesting that all individuals are derived from the same admixed source population. In addition, the high heterozygosity and the lack of inbreeding maintains the four genetically distinct clusters, as ERC is a dioecious wind-pollinated species.
Seed dispersal outside a local area has been documented for numerous invasive species (Horncastle et al. Reference Horncastle, Hellgren, Mayer, Engle and Leslie2004; Martinod and Gorchov Reference Martinod and Gorchov2017; Nathan et al. Reference Nathan, Schurr, Spiegel, Steinitz, Trakhtenbrot and Tsoar2008). This type of seed dispersal is especially common among species with fruiting phenologies that mature their fruits in the late fall (Barriball et al. Reference Barriball, McNutt, Gorchov and Rocha2015; Bartowitz and Orrock Reference Bartowitz and Orrock2016; McNeish and McEwan Reference McNeish and McEwan2016). ERC trees initiate fruit maturation in September and bear mature fruits through the winter months in LDSNP (K Shvach, personal communication). These fruits are available to be dispersed by a diverse array of birds and mammals over the winter when resources are scarce (Horncastle et al. Reference Horncastle, Hellgren, Mayer, Engle and Leslie2004). Avian dispersers of ERC can disperse seeds short and intermediate distances from the source and are primarily resident or nomadic birds, traveling with no regular pattern to follow resource availability (Holthuijzen and Sharik Reference Holthuijzen and Sharik1985; Horncastle et al. Reference Horncastle, Hellgren, Mayer, Engle and Leslie2004; K Shvach, personal communication). Additionally, small mammals are likely to carry seeds short distances (<100 m), while medium-sized mammals may spread seeds longer distances (1 km or more) (Horncastle et al. Reference Horncastle, Hellgren, Mayer, Engle and Leslie2004). ERC can be distributed by humans as well through use of the tree for landscaping and windbreaks and are often distributed very long distances from the seed source, resulting in long-distance dispersal (Donovan et al. Reference Donovan, Burnett, Bielski, Birgé, Bevans, Twidwell and Allen2018). Therefore, the behavior of ERC’s seed dispersers could result in the colonization of LDSNP by ERC seeds from within LDSNP and from surrounding populations from varying distances. This seed dispersal pattern likely explains ERC’s rapid encroachment of LDSNP and the increase in allelic diversity and heterozygosity in the younger age groups.
Once introduced to a new suitable habitat, ERC can quickly dominate through its strong competitive abilities, similar to many nonnative invasive plants (Briggs et al. Reference Briggs, Hoch and Johnson2002; Donovan et al. Reference Donovan, Burnett, Bielski, Birgé, Bevans, Twidwell and Allen2018). ERC thrives in environments where competition with other plants for sunlight is low and can tolerate many moisture levels, especially xeric environments (Hamati Reference Hamati2022; Ward Reference Ward2020). Additionally, birds are more likely to deposit seeds near perching sites, making previously colonized open areas more likely to be sites of future seed deposition than undisturbed open areas (Higgins et al. Reference Higgins, Richardson and Cowling2000; Holthuijzen and Sharik Reference Holthuijzen and Sharik1985; Horncastle et al. Reference Horncastle, Hellgren, Mayer, Engle and Leslie2004). When long-distance dispersal occurs, ERC becomes established in a new area, and the acceleration of invasion quickly increases (Moody and Mack Reference Moody and Mack1988). Therefore, it is imperative to prevent long-distance dispersal and establishment of ERC into undisturbed grasslands to maintain the integrity of the ecosystem and prevent further invasion.
Conclusions
Overall, we found that the invasion pattern of the native ERC has similarities and differences from that of nonnative invasive plants. First, contrary to most nonnative invasive plants, there was an increase in allelic diversity in the ERC population at the LDSNP in just a few decades. This finding follows the predictions of Excoffier et al. (Reference Excoffier, Foll and Petit2009); the new alleles are rare (low frequency). Moreover, the results of our STRUCTURE analysis and trends in genetic diversity also suggest sustained gene flow from small clusters of trees neighboring LDSNP originating from previously admixed source populations now found along highways, farmland, and in yards in and around Marblehead. Because ERC is a dioecious tree, outcrossing promotes admixture among clusters, increasing the range of environmental conditions where it can thrive. Hybridization among different ecotypes is often a condition that facilitates invasion of both nonnative and native invaders (Castillo et al. Reference Castillo, Schaffner, van Wilgen, Manuel Montaño, Bustamante, Cosacov, Mathese and Le Roux2021; Ellstrand Reference Ellstrand2009; Vilà et al. Reference Vilà, Weber and Antonio2000).
Finally, while we could not identify a clear advancing front, the level of admixture observed in all ages in our analysis suggests that diffusion from newly established foci trees through seeds dispersed at short and intermediate distances drives ERC’s range expansion. Additionally, anthropogenic distribution resulting from using ERC for landscaping can result in the long-distance dispersal of trees from different origins, which in turn facilitates hybridization and further invasion of grasslands and open areas. The new allelic variants seen in the younger individuals in our data set are present at low frequencies and are likely to result from introgression with external sources in this admixed population through pollen and seed dispersal from neighboring populations. Specifically, in the case of LDSNP, ERC has likely arrived through a combination of animal- and human-induced means from neighboring populations and has subsequently disseminated through the preserve’s prairie from those introductions. At LDSNP, ERC is quickly dominating the dry, open areas of the preserve but is being outcompeted in wetter areas.
It is imperative that managers control the spread of ERC to new areas in order to prevent the establishment of these fast-growing populations. To forestall the establishment of new stands of ERC most effectively, managers should eliminate satellite populations before individual trees are able to reach sexual maturity. Additionally, humans have the responsibility to prevent long-distance dispersal of invasive plants through anthropogenic means like landscaping. By preventing the long-distance dispersal of ERC, managers can limit the amount of intraspecific hybridization occurring among different ecotypes with the goal of limiting genetic diversity and potential for evolution of beneficial traits in invasive plants.
Acknowledgments
The authors are thankful for and acknowledge the contribution of Katie Shvach for her company and help in our fieldwork. We also thank William Hass for his comments on the previous version of this article. We thank the anonymous reviewers of our article for their time spent reviewing drafts and for their many insightful comments. Finally, we also thank the National Science Foundation (grant no. DEB 2214146 to OJR), the Graduate Student Senate at Kent State University, and the Art and Margaret Herrick Aquatic Ecology Research Facility Student Research Grant (to HMH) for funding this work. The authors declare no conflict of interest and they have no relevant financial or non-financial interests to declare.
Open access
