AoB PLANTS, 2023, 15, 1–12
https://doi.org/10.1093/aobpla/plad070
Advance access publication 5 November 2023
Studies
Studies
Adaptive constraints at the range edge of a widespread 
and expanding invasive plant
Rebecca A. Fletcher1, Daniel Z. Atwater2, David C. Haak1, , Muthukumar V. Bagavathiannan3, 
Antonio DiTommaso4, Erik Lehnhoff5, Andrew H. Paterson6, Susan Auckland6,  
Prabhu Govindasamy3,7, Cornelia Lemke6, Edward Morris5, Lisa Rainville6 and Jacob N. Barney*1,
1School of Plant and Environmental Sciences, Virginia Tech, 1015 Life Science Circle, Blacksburg, VA 24061, USA
2Department of Animal & Range Sciences, Montana State University, 103 Animal Biosciences Building, Bozeman, MT 59717, USA
3Department of Soil and Crop Sciences, Texas A&M University, 370 Olsen Boulevard, College Station, TX 77843, USA
4School of Integrative Plant Science, Section of Soil and Crop Sciences, Cornell University, Ithaca, NY 14853, USA
5Department of Entomology, Plant Pathology, and Weed Science, New Mexico State University, MSC 3BE, Las Cruces, NM 88003, USA
6Plant Genome Mapping Laboratory, University of Georgia, 111 Riverbend Road, Athens, GA 30602, USA
7Division of Agronomy, ICAR-Indian Agricultural Research Institute, , New Delhi 110012, India
*Corresponding author’s e-mail address: jnbarney@vt.edu
Associate Editor: David Gorchov
Abstract. Identifying the factors that facilitate and limit invasive species’ range expansion has both practical and theoretical importance, espe-
cially at the range edges. Here, we used reciprocal common garden experiments spanning the North/South and East/West range that include 
the North American core, intermediate and range edges of the globally invasive plant, Johnsongrass (Sorghum halepense) to investigate the 
interplay of climate, biotic interactions (i.e. competition) and patterns of adaptation. Our results suggest that the rapid range expansion of 
Johnsongrass into diverse environments across wide geographies occurred largely without local adaptation, but that further range expansion 
may be restricted by a fitness trade-off that limits population growth at the range edge. Interestingly, plant competition strongly dampened 
Johnsongrass growth but did not change the rank order performance of populations within a garden, though this varied among gardens (cli-
mates). Our findings highlight the importance of including the range edge when studying the range dynamics of invasive species, especially as 
we try to understand how invasive species will respond to accelerating global changes.
Keywords: Biotic interactions; fitness trade-off; flowering time; invasive species; range edge; range limits; Sorghum halepense.
Introduction range core versus at the range edge (Geber 2008). Populations 
The factors that prevent species from continual range expan- at the range edge can be geographically distant from the range 
sion have been of great interest to biologists for well over a core and are often rare and occur at low densities (Gaston 
century. In the absence of obvious physical barriers (e.g. a 1990). Thus, local adaptation is hindered by low connectivity 
mountain range) that limit range expansion, a general frame- and insufficient genetic variation (Colautti et al. 2010; Körner 
work has been proposed in which a combination of biotic, et al. 2016). Gene flow from core and intermediate popula-
abiotic and demographic factors interact to set range limits tions (i.e. populations located between the core and edge) can 
(Angert 2009; Sexton et al. 2009; Stanton-Geddes et al. 2012; increase genetic diversity in populations at the edge, but it 
Csergő et al. 2017). Thus, populations persist along an envir- can also have the detrimental effect of introducing maladap-
onmental gradient until population growth is no longer sus- tive genes, hindering adaptation to conditions at the edge of 
tainable (Hargreaves et al. 2014). the range (Kirkpatrick and Barton 1997; Holt et al. 2005; 
The ‘center-periphery hypothesis’ (Pironon et al. 2017) and Bontrager and Angert 2019). Life history trade-offs can fur-
the related ‘abundant center hypothesis’ (Sagarin and Gaines ther constrain adaptation to conditions at the range edge (Holt 
2002) maintain that species are most abundant in the ‘core’ et al. 2005; Sexton et al. 2009; Alexander and Edwards 2010; 
or centre of their range, and get progressively less common/ Chuang and Peterson 2016). However, the supporting evi-
abundant toward the range edge. This has served as the basis dence for these biogeographic hypotheses remains equivocal 
of numerous studies on diverse taxa. Range expansion is (Sagarin and Gaines 2002; Pironon et al. 2017).
driven and limited by a combination of dispersal, population Invasive species introduced to new geographies often 
growth rate and density-dependent biotic factors (Thomas undergo rapid range expansion, even as they face novel gen-
2010), the interactions of which can be quite different at the etic, biotic and abiotic constraints on fitness (Theoharides 
Received: 12 April 2023; Editorial decision: 10 October 2023; Accepted: 3 November 2023
© The Author(s) 2023. Published by Oxford University Press on behalf of the Annals of Botany Company.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), 
which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.
Downloaded from https://academic.oup.com/aobpla/article/15/6/plad070/7352046 by guest on 15 February 2024
2 AoB PLANTS, 2023, Vol. 15, No. 6 
and Dukes 2007; Alexander and Edwards 2010), and can Johnsongrass populations, leading to improved species per-
serve as important systems to test elements of range limits and sistence (Maity et al. 2022).
local adaptation (Sexton et al. 2009). For example, Colautti The core of Johnsongrass’s North American range occurs 
and Barrett (2013) found that, while the evolution of earlier across much of the southeastern USA, with the original site 
flowering time in the invasive plant Lythrum salicaria enabled of introduction in South Carolina likely in the mid-1800s, 
it to expand its range poleward, there was a trade-off of vege- quickly spreading across the southern USA (Sezen et al. 2016). 
tative growth. However, even in invasive study systems, the Currently, Johnsongrass population abundance declines 
relative contributions of abiotic (e.g. climate) and biotic (e.g. sharply around 38°N and 102°W, which we define here as the 
competition) factors on range limits are not well understood ‘intermediate range’ because these populations likely experi-
(Sexton et al. 2009). ence different conditions and population dynamics compared 
At large spatial scales, climate is thought to be the main with the range core and range edge (Chuang and Peterson 
driver of species’ ranges, as demonstrated by decades of cor- 2016). Johnsongrass populations beyond the intermediate 
relative and experimental studies (as reviewed in Sexton et al. range are uncommon, small and isolated (DiTommaso and 
2009). At smaller scales, biotic effects such as competition, Lehnhoff personal observations), which we define as the 
predation and pollination interact in setting geographic limits range edge (Jump and Woodward 2003; Bridle and Vines 
(Bullock et al. 2000; Sexton et al. 2009; Watling and Orrock 2007; Chuang and Peterson 2016). There is anecdotal evi-
2010; Louthan et al. 2015). Biotic effects are often impacted dence that Johnsongrass is unable to overwinter as rhizomes 
by abiotic conditions, appearing to be most important in in the northern edge, and may be limited by arid conditions in 
abiotically less stressful environments (Sanford et al. 2003; the western edge (Warwick et al. 1984, 1986).
Louthan et al. 2017). Thus, biotic interactions may interact Recent studies show that Johnsongrass populations have 
in complicated ways with climate to influence species popu- diverged both phenotypically and genetically. Traits of 
lation growth rates. Johnsongrass populations depend strongly on home climate 
Here, we studied a widespread and range-expanding in- and habitat, suggesting that local adaptation to varying en-
vasive perennial grass to identify factors that govern its vironmental and climatic conditions along with phenotypic 
continent-scale range expansion across two major cli- plasticity have played an important role in the success of this 
mate axes (i.e. temperature North-to-South and precipita- widespread invasive species (Atwater et al. 2015, 2018; Sezen 
tion East-to-West) that included the range edge. We asked et al. 2016; Lakoba and Barney 2020). For example, we have 
the following questions: (i) Do climate and biotic factors shown that Johnsongrass sampled from a small geographic 
interact to limit range expansion?; (ii) Is there evidence of region appears to rely on local adaptation in non-agricultural 
adaptation to conditions along temperature and precipita- habitats and plasticity in agricultural systems (Atwater et al. 
tion gradients? and (iii) Is there evidence that adaptation 2018), but we have not tested this at large spatial scales across 
is constrained at the range edge? Given our previous ex- climate gradients. However, it remains unclear what role local 
perience with local populations (Atwater et al. 2018), we adaptation plays in limiting Johnsongrass range expansion 
expected competition to strongly limit Johnsongrass per- at large spatial scales, which is projected to move poleward 
formance, which we hypothesized would be strongest under with climate change (McDonald et al. 2009; Lakoba et al. 
less favourable growing conditions (i.e. range edges). We 2021a). To elucidate the role that local adaptation plays in 
also expected ‘home’ Johnsongrass populations to gener- response to abiotic and biotic factors, we established five 
ally outperform ‘away’ populations. Alternatively, if local common gardens spanning the western and northern ranges 
adaptation is weak, this supports a previously demonstrated of Johnsongrass using reciprocal transplants grown with and 
‘general-purpose-genotype’ common in Johnsongrass popu- without local competition.
lations (Atwater et al. 2015, 2018; Lakoba and Barney 
2020), allowing strong performance across a range of 
growing conditions. Seed collections
We collected Johnsongrass seeds spanning the western 
and northern range of Johnsongrass near the garden sites. 
Materials and Methods Across the North/South range we collected seeds from core 
(Georgia), intermediate (Virginia) and edge (New York) 
Study system populations. Across the East/West range, we collected seeds 
Johnsongrass [Sorghum halepense], thought to be native from core (Georgia, same as above), intermediate (Texas) and 
to western Asia (Paterson et al. 2020), is one of the most edge (New Mexico) populations. Seeds from GA, VA and 
damaging agricultural weeds globally, commonly invading TX were collected between June and August 2011 and were 
agricultural fields, roadsides, rights-of-way and natural eco- subsequently planted in a greenhouse on the Virginia Tech 
systems (Warwick and Black 1983; Rout et al. 2013). A single campus in Blacksburg, VA where they were maintained as 
plant is capable of producing tens of thousands of seeds in a live germplasm. Cuttings from the plants were transplanted 
single growing season without any clear dispersal mechanism into a seed increase garden in May 2016, and mature seeds 
(likely gravity dispersed), and an extensive rhizome network were harvested in August–October 2016. Seeds from the NY 
(McWhorter 1961). Both seeds and rhizomes contribute to and NM populations were collected from the original source 
the invasiveness of Johnsongrass, as seeds are vital for spread populations in August 2016 and were used directly in the pre-
(Atwater et al. 2017) and rhizomes allow local persistence sent study. In all cases, seeds used in this experiment represent 
and dominance once established (McWhorter 1961) with multiple maternal lines from a single source location. Using 
long-distance dispersal likely a result of contamination (e.g. a mix of seed sources is not ideal, but we were limited logis-
mowers, harvesters). Further, pollen-mediated gene flow can tically to when we received seeds for the edge populations. 
contribute to the dispersal of adaptive alleles to adjacent Results should be interpreted considering this variation.
Downloaded from https://academic.oup.com/aobpla/article/15/6/plad070/7352046 by guest on 15 February 2024
Fletcher et al. – Adaptive constraints at range edge 3
In addition to the five populations, we included a control 
population (‘phytometer’) that is not from any of the garden 
locations and serves as a consistent ‘away’ population in all 
gardens. We are using this population as a phytometer sensu 
Strobl et al. (2018), ‘phytometers are indicator transplants 
that provide information on site conditions based on plant 
survival, growth and reproduction’. Seeds for the phytometer 
were purchased through a commercial vendor (Azlin Seed 
Service, 112 Lilac Dr., Leland, MS 38756), which were lo-
cally sourced in the Mississippi Delta area. We have used this 
phytometer in previous studies (Smith et al. 2021).
Common gardens
We used a series of common gardens to investigate the 
interacting effects of abiotic (e.g. local climate), biotic (e.g. 
competition) and genetic (e.g. invader population source) fac-
tors on local adaptation, population growth and range limits. 
While we are not testing climate gradients per se, our gar-
dens span large temperature (latitudinal) and precipitation 
(longitudinal) gradients, and we have consistently shown that 
Johnsongrass home climate strongly impacts performance 
(Atwater et al. 2015, 2018; Lakoba and Barney 2020). While 
three gardens along each axis are not sufficient to capture 
all variation across these large gradients, significant logistical 
challenges constrain the number of field sites and our study 
has a similar approach to Colautti and Barrett (2013) (and 
others) who established three gardens across a large climate 
gradient representing the existing core-to-edge range of an in-
vasive plant.
Reciprocal transplant experiments are one of the most ef-
fective methods for investigating patterns associated with 
range dynamics (Blanquart et al. 2013), and by studying 
various life history traits, such as growth and flowering time, 
we can assess potential trade-offs that may lead to adaptive 
constraint. We used seedling transplants to standardize plant 
size and to test the survival and performance of the seedling 
stage.
We established five gardens near the locations of the five 
source populations (GA, VA, TX, NY and NM; Table 1) in 
areas where Johnsongrass populations are known to occur, 
but in garden plots that lacked the species. The common 
gardens were established in either March or May 2017, de-
pending on the last frost date of the garden location (Table 1). 
We considered this to be a better test of site climatic effects 
than planting all gardens at the same time and thereby short-
ening the growing season at warmer sites. To improve germin-
ation, seeds from all populations were pre-treated by soaking 
them in 100 % bleach for 4 h followed by a 1-h rinse in tap 
water (Atwater et al. 2018). After pre-treatment, seeds were 
germinated in 128-well flats filled with Miracle-Gro® Potting 
Mix. Seedlings were allowed to grow for approximately 5 
weeks in a greenhouse located in Blacksburg, VA before being 
transplanted into each garden.
Prior to the start of the experiment, each garden was 
tilled so that all transplants were planted into uniform, 
competition-free, bare ground. Each garden consisted of 10 
replicate blocks with each block split into two plots. Each 
8 × 4m plot within a block was randomly assigned to one of 
two treatments: a competition treatment or a bare-ground 
competition-free control treatment. The competition treat-
ment was designed to evaluate one of the important biotic 
limitations to Johnsongrass establishment and performance; 
Table 1. Coordinates, elevation, mean annual temperature (°C; MAT) and total annual precipitation (mm; TAP) of the Johnsongrass source populations and common garden locations, including that for the 
2017 experimental season. The last and first frost dates, experimental planting date and harvesting date are also shown for each of the five common gardens.
Georgia Virginia Texas New York New Mexico
Population Garden 2017 Population Garden 2017 Population Garden 2017 Population Garden 2017 Population Garden 2017
Latitude (DD) 33.883 33.729 37.194 37.194 31.060 30.552 42.764 42.451 32.202 32.314
Longitude (DD) −83.154 −83.302 −80.574 −80.302 −97.342 −96.428 −75.552 −76.461 −106.732 −106.745
Elevation (m) 238 147 518 504 195 66 334 287 1177 1176
MAT (°C) 16.2 16.5 18.1 11.2 11.6 12.1 19 20.1 21.7 6.6 7.5 8.8 16.2 17.5 18.5
TAP (mm) 1230 1220 1420 957 1013 1041 858 994 1326 1049 941 1080 234 213 294
Last Frost Date 24-Mar 4-May 2-Mar 14-May 1-May
Planting Date 29-Mar 29-May 22-Mar 24-May 17-May
First Frost Date 7-Nov 6-Oct 29-Nov 3-Oct 20-Oct
Harvest Date 6-Nov 17-Oct 7-Nov 10-Oct 4-Oct
Downloaded from https://academic.oup.com/aobpla/article/15/6/plad070/7352046 by guest on 15 February 2024
4 AoB PLANTS, 2023, Vol. 15, No. 6 
acknowledging that herbivory and pathogens are other im- The weather of each garden in the experimental year of 
portant biotic interactions, but beyond the scope of this 2017 was slightly warmer and wetter than the long-term 
study. In the competition treatment, the resident flora was averages (Table 1) but trended the same across all gardens. 
allowed to colonize naturally with no other management While executed as a single large experiment, we considered 
applied, and in all cases resulted in a dense weed flora as our North–South and East–West gardens independently for 
would be expected in an agricultural field. We have used this several reasons. First, most climate gradient studies consider 
approach in similar studies where we are not interested in only a single axis, often temperature North–South, and ana-
measuring competition per se (Atwater et al. 2017, 2018); lysing them separately is in line with this—that is, two sep-
rather we are mimicking what Johnsongrass would experi- arate gradients. Second, different mechanisms may be at play 
ence at these locations naturally as a seedling—strong com- in response to temperature and precipitation that would po-
petition with fast-growing resident weeds. The bare-ground tentially muddle interpretation if they were combined and 
treatment was maintained by manually removing weeds on analysed as core, intermediate and edge. Considering them sep-
a weekly basis. In each plot, one seedling from each of the arately allows for independent analyses and interpretations.
five Johnsongrass populations and the phytometer were 
transplanted into a grid, randomly arranged with 2 m spa- Statistical analysis
cing between seedlings. Individuals from all populations Because we were interested in performance and flowering time 
were transplanted in all treatments, in all blocks and in all along two different gradients, we analysed the gardens along 
locations. the latitudinal (Georgia, Virginia, New York) and longitudinal 
Transplants were given 2 weeks to establish in the field, (Georgia, Texas, New Mexico) gradients separately (note that 
during which we supplied supplemental water as needed data from the Georgia garden were included in the analyses 
to mitigate transplant shock. Seedlings that failed to estab- of both gradients as the range core). In addition to recorded 
lish were not included in the analyses (Georgia = 0, New performance traits, we also calculated both aboveground bio-
Mexico = 21, New York = 14, Virginia = 5 and Texas = 0). mass and height of each population relative to the phytometer 
After the establishment period, no additional water was within each garden by taking the logarithm of the ratio between 
added, except in New Mexico, because populations in New the biomass or height value of each plant and the phytometer 
Mexico are almost always found in irrigated agricultural in each respective block and treatment (see Appendix 1 in the 
fields, irrigation and drainage ditches and other locations Supplementary Materials for more details). This allowed us 
that have a consistent water supply during the growing to test for the performance of each Johnsongrass population 
season (Lehnhoff personal observation), we supplied the 10 to a consistent phytometer benchmark, which was always 
experimental blocks in the New Mexico garden with min- an ‘away’ population. We used mixed-effects linear models 
imal supplemental water (irrigation) on an as-needed basis to analyse aboveground biomass (cube-root transformed to 
through the duration of the experiment. In addition to the meet model assumptions), height, flowering time (Julian days 
10 blocks that received irrigation, we included five add- from transplant day to first flower), natural-log transformed 
itional blocks that did not receive any supplemental water biomass relative to phytometer and natural-log transformed 
beyond the initial 2-week establishment period. Because height relative to phytometer. All models were fit using the 
of the limited sample size, the data collected from the five R package ‘lme4’ (Bates et al. 2015). To investigate our first 
extra blocks in New Mexico were not assessed in the main two questions, (i) whether there is evidence that climatic and 
analyses described below. More information and results biotic factors interact to limit Johnsongrass range expansion 
from the no-irrigation treatment in New Mexico can be and (ii) whether there is evidence of adaptation to conditions 
found in Appendix 3 of the Supplementary Materials. As along range gradients, we included the fixed effects (predictor 
an aside, since this experiment concluded we have observed variables) of population, treatment, garden and their inter-
Johnsongrass growing in the dry uplands in neighbouring actions. We also included the random effect of block. The 
Arizona (Barney personal observation) suggesting that effect of competition on Johnsongrass biomass was also cal-
Johnsongrass is capable of persistence in unirrigated por- culated and analysed as above (see Supplementary Appendix 
tions of the study area. 4). Significance was assessed at ∝ = 0.05, and when a signifi-
Beginning 3 weeks after the establishment of each garden, cant effect was present, we implemented pairwise contrasts 
we recorded the date of the first flower on a weekly basis using Tukey correction to compare levels within the predictor 
for each Johnsongrass individual. We also began harvesting variable, as implemented in the package ‘emmeans’ (Lenth 
fully matured panicles to avoid seeds shattering onto the soil. 2019). For significant interactions, we conducted pairwise 
Panicles were labelled and stored and included in the final comparisons within the levels of the interacting predictors. To 
aboveground biomass. Panicle phenotype varies a lot among investigate our third question, whether there is evidence of a 
Johnsongrass populations, and estimating or counting seeds trade-off between life histories that might constrain adapta-
was not feasible. More so, this is a perennial plant with both tion at the range edge, we assessed the relationship between 
sexual and vegetative reproduction, so individual plant fit- height and flowering time by performing a linear regression 
ness is not simply a measure of seed production. Fortunately, with flowering time as a fixed effect and block as a random 
aboveground biomass is a strong predictor of fitness (as re- effect. All analyses were carried out in R Core Team (2018).
viewed in Younginger et al. 2017), and thus we will use bio-
mass and height as proxies for performance. At the end of the 
growing season, as determined by the first frost date of each Results
garden location (Table 1), we recorded survival and culm 
height. Aboveground biomass was harvested, dried at 60 °C For clarity, the Johnsongrass populations will be labelled 
for 3 days, and weighed. The following spring (2018), we re- with state abbreviations (GA, VA, NY, TX, NM) while gar-
corded overwinter survival. dens will be labelled with full state names (Georgia, Virginia, 
Downloaded from https://academic.oup.com/aobpla/article/15/6/plad070/7352046 by guest on 15 February 2024
Fletcher et al. – Adaptive constraints at range edge 5
New York, Texas, New Mexico). The phytometer is labelled Fig. 1A and C) across all three gardens. We also found 
as PHYT. biomass and height of all populations were greatest in the 
Virginia (intermediate) garden compared to the Georgia 
Latitudinal gradient (core) and New York (edge) gardens, where Johnsongrass 
performed about equally (Fig. 1A and C). There was a sig-
Without competition, we found that the PHYT produced nificant interaction between population and garden for 
the most biomass and grew the tallest, while the two edge biomass (Table 2). While the GA, TX and VA populations 
populations (NY and NM) produced the least biomass generally had similar biomass in New York and Georgia, 
and grew the shortest (Table 2; Supplementary Table S1; the VA population had the greatest biomass in its home 
Table 2. Results of mixed-effects models testing the effects of population, garden, treatment and their interactions on cube-root biomass, height and 
flowering time along latitudinal and longitudinal gradients. All models included the random effect of block. The F-statistic was calculated using the Type 
III sum of squares. Values in bold are significant at P < 0.05.
Latitudinal Gradient Longitudinal Gradient
SS df F P SS df F P
Cube-root Biomass
  Population 237.29 5 25.17 <0.001 273.65 5 23.82 <0.001
 G arden 234.40 2 62.17 <0.001 770.80 2 167.71 <0.001
 T reatment 1438.42 1 763.02 <0.001 1066.55 1 464.13 <0.001
 P op. × Gard. 56.16 10 2.98 0.001 71.24 10 3.10 0.001
  Pop. × Treat. 18.39 5 1.95 0.087 8.04 5 0.70 0.624
  Gard. × Treat. 50.69 2 13.44 <0.001 357.03 2 77.68 <0.001
  Pop. × Gard. × Treat. 13.26 10 0.70 0.721 25.44 10 1.11 0.357
ln(Biomass Relative to Phytometer)
Population 11.56 4 3.64 0.007 20.52 4 8.02 <0.001
Garden 1.51 2 0.95 0.401 0.45 2 0.35 0.705
Treatment 8.95 1 11.26 0.001 19.49 1 30.48 <0.001
Pop. × Gard. 8.36 8 1.31 0.239 8.83 8 1.73 0.094
Pop. × Treat. 3.18 4 1.00 0.409 4.12 4 1.61 0.173
Gard. × Treat. 0.98 2 0.61 0.542 7.55 2 5.90 0.003
Pop. × Gard. × Treat. 3.89 8 0.61 0.768 6.77 8 1.32 0.234
Height
  Population 86667 5 27.56 <0.001 59649 5 23.76 <0.001
 G arden 108226 2 86.05 <0.001 308701 2 307.38 <0.001
 T reatment 38796 1 61.70 <0.001 14393 1 28.66 <0.001
  Pop. × Gard. 10470 10 1.66 0.090 10060 10 2.00 0.034
  Pop. × Treat. 4129 5 1.31 0.259 2732 5 1.09 0.367
  Gard. × Treat. 4606 2 3.66 0.027 28210 2 28.09 <0.001
 P op. × Gard. × Treat. 3785 10 0.60 0.812 5910 10 1.18 0.307
 l n(Height Relative to Phytometer)
  Population 4.04 4 14.45 <0.001 2.43 4 10.93 <0.001
  Garden 0.38 2 2.75 0.082 0.33 2 2.97 0.067
  Treatment 0.04 1 0.58 0.448 0.05 1 0.84 0.362
  Pop. × Gard. 0.18 8 0.33 0.954 0.52 8 1.16 0.322
 P op. × Treat. 0.15 4 0.54 0.708 0.10 4 0.44 0.777
  Gard. × Treat. 0.01 2 0.04 0.963 0.47 2 4.24 0.016
  Pop. × Gard. × Treat. 0.28 8 0.50 0.852 0.41 8 0.93 0.496
Flowering Time
 P opulation 3882 5 10.09 <0.001 3860 5 13.48 <0.001
 G arden 111460 2 724.21 <0.001 202108 2 1765.08 <0.001
  Treatment 98 1 1.27 0.260 243 1 4.24 0.040
 P op. × Gard. 792 10 1.03 0.419 1573 10 2.75 0.003
  Pop. × Treat. 234 5 0.61 0.694 250 5 0.87 0.499
 G ard. × Treat. 109 2 0.71 0.493 211 2 1.84 0.161
  Pop. × Gard. × Treat. 317 10 0.41 0.940 285 10 0.50 0.890
Downloaded from https://academic.oup.com/aobpla/article/15/6/plad070/7352046 by guest on 15 February 2024
6 AoB PLANTS, 2023, Vol. 15, No. 6 
Figure 1. Effects of population, treatment and garden on biomass (back-transformed) along a latitudinal (a) and a longitudinal (b) gradient and height 
along a latitudinal (c) and longitudinal (d) gradient of Johnsongrass’s North American range. Separate mixed-effects models were performed for each of 
the two gradients. Points are estimated marginal means and error bars show standard error. See Table 2 for statistical details.
garden with the NY and NM edge populations performing Johnsongrass ln(biomass relative to PHYT) and ln(height 
the worst (Supplementary Table S1; Fig. 1A). Competition relative to PHYT) varied among populations (Table 2; Fig. 
universally decreased both Johnsongrass height and bio- 2A and C). The ln(biomass relative to PHYT) was greatest for 
mass, but the effect of competition on both variables varied the GA population (Table 2) and smallest for the NM popu-
across gardens (Table 2; Fig. 1A and C). Competition lation (Table 1). On the other hand, the VA population had 
greatly reduced biomass but did not change the rank order the greatest ln(height relative to PHYT) and the NY popula-
of Johnsongrass population performance (Fig. 1A). The im- tion had the smallest (Table 2). There was also a significant 
pact of competition on Johnsongrass plant height was more effect of weed competition on ln(biomass relative to PHYT) 
modest relative to that of biomass, and likewise had rela- (Table 2), with ln(biomass relative to PHYT) being lower in 
tively little change into the rank order (Fig. 1C). The results the weed-competition treatment than the bare-ground treat-
of both biomass and height relative to the PHYT followed ment (Fig. 2A). We found no significant interaction between 
similar patterns to those resulting from comparisons of the population and garden. The lnRR differed among gardens 
populations to the PHYT in the Tukey Tests (Supplementary (Supplementary Table S6), and indicated competition from 
Table S6; Fig. S4A and C). Detailed results of biomass and weeds was most intense in Georgia (lnRR ± SE = −3.39 ± 0.21) 
height relative to the PHYT are presented in Appendix 4 in compared to Virginia (lnRR ± SE = −2.07 ± 0.17) and New 
the Supplementary Materials. York (lnRR ± SE = −1.99 ± 0.19) gardens (Supplementary 
Downloaded from https://academic.oup.com/aobpla/article/15/6/plad070/7352046 by guest on 15 February 2024
Fletcher et al. – Adaptive constraints at range edge 7
Figure 2. Effects of population, treatment and garden on ln(biomass relative to phytometer) along a latitudinal (a) and a longitudinal gradient (b), and 
ln(height relative to phytometer) along a latitudinal (c) and longitudinal gradient (d) of Johnsongrass’s North American range. Separate mixed-effects 
models were performed for each of the two gradients. Points are estimated marginal means and error bars show standard error. See Table 2 for 
statistical details.
Fig. S4); however, lnRR did not appear to vary among popu- Plants in competition in Georgia suffered the highest mor-
lations (Supplementary Table S6). tality by the end of the growing season, with only 64 % of 
We found significant variation in flowering time among gar- individuals surviving until the end of the season. Survival at 
dens (Table 2) with plants in Georgia flowering on average 45 the end of the season in the other two gardens was overall 
days earlier than plants in Virginia and New York (Fig. 3A). higher (Virginia: bare-ground = 93 %, weeds = 98 %; New 
We also found a significant effect of population on flowering York: bare-ground = 93 %, weeds = 88 %). Of the plants that 
time (Table 2): along with the PHYT, the two edge populations survived to the end of the first growing season in Georgia, 
(NY and NM) flowered 7 days earlier than that of GA and 74 % of those in bare ground and 62 % in weed competi-
the two intermediate populations (VA and TX) populations tion survived the winter to re-emerge in the spring. There was 
(Fig. 3A; Supplementary Table S1). Results from the linear 100 % rhizome winter kill in New York across all popula-
regression showed a positive relationship between height and tions and treatments. In Virginia, of the plants that survived 
flowering time (Fig. 4A; P < 0.001, Estimate = 0.84, t = 6.0, to the end of the season, 89 % of plants in bare ground and 
df = 80), suggesting a trade-off in that earlier flowering plants 72 % in weed competition survived the winter to re-emerge 
are shorter than later flowering plants; a change in nearly 100 in the spring (see Supplementary Table S3 in Appendix 2 in 
% between early and late flowering individuals. Supplementary Materials for more details).
Downloaded from https://academic.oup.com/aobpla/article/15/6/plad070/7352046 by guest on 15 February 2024
8 AoB PLANTS, 2023, Vol. 15, No. 6 
Figure 3. Effects of population, treatment and garden on Johnsongrass flowering time along a latitudinal (a) and a longitudinal gradient (b) of 
Johnsongrass’s North American range. Separate mixed-effects models were performed for each of the two gradients. Points are estimated marginal 
means and error bars show standard error. See Table 2 for statistical details.
Figure 4. Results of linear regressions of the relationship between Johnsongrass height and flowering time along a latitudinal (a) and a longitudinal 
gradient (b) of Johnsongrass’s North American range. Separate mixed-effects regression models were performed for each of the two gradients. Points 
are estimated raw values and the black line is the model-predicted regression line.
Longitudinal gradient results of both biomass and height relative to the phytometer 
The PHYT again accumulated more biomass than all other followed similar patterns as those found in comparisons of 
populations across all gardens, while the edge NY and NM populations to the phytometer in the Tukey Tests. Detailed 
populations performed worst. Competition decreased both results of biomass and height relative to the phytometer are 
biomass and height, but the magnitude of these effects varied presented in Appendix 4 in the Supplementary Materials.
among gardens (Table 2; Fig. 1B and D). Johnsongrass bio- Both ln(biomass relative to PHYT) and ln(height relative 
mass was most strongly suppressed by weeds in New Mexico to PHYT) varied among populations (Table 2). The two 
(Fig. 1B). In contrast, Johnsongrass height was less affected edge populations had lower ln(biomass relative to PHYT) 
by competition, but did vary among gardens (Fig 1D). The and ln(height relative to PHYT) compared to the other 
Downloaded from https://academic.oup.com/aobpla/article/15/6/plad070/7352046 by guest on 15 February 2024
Fletcher et al. – Adaptive constraints at range edge 9
populations, GA and VA had the largest ln(biomass rela- The core (Georgia) environment elicited early flowering, 
tive to PHYT), and VA had the largest ln(height relative to smaller individuals with no evidence of local adaptation (i.e. 
PHYT) (Fig. 2B and D; Table 2). Biomass and height relative ‘home’ populations did not outperform ‘away’ populations 
to the PHYT also varied between treatments and among gar- in home core garden), while the intermediate and edge gar-
dens (Table 2). We did not find evidence that ln(biomass rela- dens showed later flowering. Despite the ‘aggressive’ nature 
tive to PHYT) was different between the weed-competition of this fast-growing rhizomatous perennial, plant competition 
and bare-ground treatments in Texas, while in the other two is strongly limiting everywhere in the first year (Atwater et 
gardens, weed-competition decreased ln(biomass relative to al. 2018); though we have shown large differences in estab-
PHYT) (Fig. 2B and D; Table 2). On the other hand, ln(height lishment whether starting from seed or rhizome (Atwater et 
relative to PHYT) appeared to be similar in both treatments in al. 2017). We did not evaluate rhizome production, which 
the Georgia and New Mexico gardens, but ln(height relative may have varied across gardens and would be interesting to 
to PHYT) was greater in the bare-ground treatment than the evaluate in future studies. However, our results suggest that 
weed-competition treatment in Texas (Fig. 2B and D; Table the effects of competition are the strongest in intermediate 
2). Competition from weeds was most intense in the Georgia and edge environments. In contrast to northward expansion, 
garden (lnRR ± SE = -3.40 ± 0.20) and least intense in the we could not identify what factor(s) limited its westward ex-
Texas garden (lnRR ± SE = -0.43 ± 0.14) (Supplementary Fig. pansion; transplants in New Mexico exhibited tremendous 
S5). growth potential with or without supplemental water fol-
Flowering time varied among populations and depended lowing initial establishment.
strongly on garden location (Table 2, Fig. 3). Johnsongrass Intermediate populations were able to achieve high per-
flowered earliest in Texas, by 40 days compared to core formance (biomass and height), not only in their home gar-
Georgia and 80 days compared to edge New Mexico, but dens but also in most of the other gardens, displaying an 
there were no differences in flowering time among popu- ability to maintain high performance across a range of en-
lations within the Texas garden (Fig. 3B; Supplementary vironmental conditions. This suggests that the intermediate 
Table S2). Generally, in Georgia and New Mexico, the edge populations may be functioning as ‘general purpose’ (Baker 
populations (NM, NY), along with the PHYT, flowered 1965) or ‘jack-and-master’ genotypes (Richards et al. 2006). 
earliest, whereas VA and GA flowered the latest (Fig. 3B; Previous work with Johnsongrass found that phenotypic plas-
Supplementary Table S2). There was also a significant effect ticity allowed Johnsongrass to perform well in response to 
of treatment on flowering time with plants in the bare-ground different habitats and climatic variations, enabling its rapid 
treatment flowering earlier than in the competition treatment spread in the USA (Atwater et al. 2015).
(Table 2; Fig. 3B). As with the latitudinal gradient, the re-
sults from the linear regression of flowering time and height Adaptation at the range edge
along the longitudinal gradient indicated a positive relation- While we found evidence of ‘Jack-and-Master’ plasticity 
ship (P < 0.001, Estimate = 0.80, t = 5.9, df = 79), suggesting in Johnsongrass, this does not allow any species to ex-
a consistent association between shorter plants and earlier pand its range indefinitely (Chuang and Peterson 2016). We 
flowering (Fig. 4B). found evidence of trait differentiation between edge and 
Survival to the end of the season was high in both New other populations. The edge populations consistently grew 
Mexico (bare-ground = 100 %, weed-competition = 98 %) smaller and shorter than the other three populations across 
and Texas (100 % in both treatments). All plants that sur- all gardens, including their ‘home’ garden. Edge popula-
vived to the end of the growing season in both New Mexico tions also reached reproductive maturity faster than other 
and Texas survived the winter to re-emerge in the spring (see populations across nearly all gardens, indicating that the 
Supplementary Table S3 in Appendix 2 in the Supplementary edge Johnsongrass populations have been selected to flower 
Materials for more details). earlier but perhaps at the expense of growth. Populations at 
the northern edge of their range, where the growing season 
may be shorter due to either cold (northern edge) or mois-
Discussion ture (western edge), often prioritize reproduction over vege-
We used five common gardens across large North/South and tative growth in order to complete their life cycle (Chuang 
East/West environmental gradients spanning the core to edge and Peterson 2016). However, if the selection is acting on 
range of Johnsongrass in the USA to explore the interactive flowering time such that shorter growing seasons favour 
effects of abiotic, biotic and adaptive factors on the range earlier flowering time (Lande and Arnold 1983; Colautti 
limits of a widespread invasive plant. Overall, our results and Barrett 2010), as suggested by our results, we would 
suggest that Johnsongrass’s capacity as a ‘Jack-and-Master’ have expected earlier flowering time in the edge popula-
(sensu Richards et al. 2006) allowed rapid range expansion tions to confer a fitness advantage over the other popula-
following initial introduction (Atwater et al. 2015, 2018; tions in their respective gardens located at the edge. But, 
Lakoba and Barney 2020; Lakoba et al. 2021a), which is we observed the opposite, as the edge populations under-
limited by strong size/flowering time trade-offs at the range performed in their home environments as well as in the 
edges. Northward expansion is limited at the intermediate other environments.
(Mid-Atlantic) range where Johnsongrass is still abundant Life history theory predicts that the covariance between 
and common, dropping precipitously to the edge where the traits, in the absence of sufficient genetic diversity for cor-
climate is too cold for persistence (all rhizomes died in the related traits, will cause the selection to act on one trait to 
New York edge garden), and the short summer season elicits the detriment of others, resulting in a trade-off (Lande and 
strong phenological/fitness trade-offs. At this range edge, even Arnold 1983; Blows and Hoffmann 2005). Populations at the 
the ‘home’ edge population performed poorly. edge often experience bottlenecks that lead to reduced genetic 
Downloaded from https://academic.oup.com/aobpla/article/15/6/plad070/7352046 by guest on 15 February 2024
10 AoB PLANTS, 2023, Vol. 15, No. 6 
diversity and fitness (Sexton et al. 2009). We have observed York. Though there existed considerable population vari-
Johnsongrass populations to be rare, small and isolated be- ation, our results suggest that competition plays a proportion-
yond the intermediate range, likely leading to a lack of genetic ally stronger role in environments where the climate is more 
variation for combinations of traits, resulting in a trade-off favourable for Johnsongrass, which is consistent with the pre-
between flowering time and size. Further, limited potential for diction that biotic interactions become more important in less 
pollen-mediated gene flow among the rather sparsely distrib- stressful abiotic conditions (Louthan et al. 2015). There are 
uted plants could have limited genetic diversity. While we did several possible explanations for this pattern. For example, 
not measure genetic diversity in this study, we did observe less stressful abiotic conditions may be favourable for local 
clear trade-offs between plant size and flowering time. This species as well as introduced species, resulting in increased 
trade-off can result in a decrease in plant performance and competitive pressure on the introduced species or even com-
population growth (Colautti et al. 2010), which acts to re- petitive exclusion (Louthan et al. 2015). Importantly, compe-
strict further range expansion beyond the range edge. tition did not alter the rank order of population performance 
at each common garden. Thus, while it had important eco-
Performance (growth) across the range logical effects, competition did not appear to differentiate 
We expected a decrease in growth at the range edge, as is a the fitness of different populations along the range. The ef-
common assumption in species distribution theory (Sexton et fect of climate is less commonly evaluated in concert with 
al. 2009). However, we found overall Johnsongrass growth biotic interactions (Geber 2008), and we are not aware of any 
varied widely across gardens with no consistent decrease in studies looking at both factors at the range edge of an inva-
performance at the range edge despite the fact that range-edge sive species (Holt 2009). Future studies should incorporate 
accessions themselves were small-statured. Along the longitu- the entire life cycle of Johnsongrass, including seed/rhizome 
dinal gradient, we observed an increase in performance from germination and seedling establishment (Atwater et al. 2017; 
the centre of the range towards the edge. While this should Lakoba et al. 2021b).
be interpreted with caution as the reported New Mexico per-
formance was from irrigated plots, though there was no ap-
preciable difference in Johnsongrass performance between Conclusions
irrigated and non-irrigated plants (Supplementary Table S4-4, The ability to maintain high performance across a wide range 
Supplementary Fig. S1). of environments via ‘Jack-and-master’ phenotypes has en-
In addition to apparent performance trade-offs at the range abled Johnsongrass to invade and rapidly expand its range 
edge (i.e. edge populations under-performed relative to all across a large, ecologically and environmentally diverse swath 
‘away’ populations in their home edge garden), the climates of the USA. However, Johnsongrass has been unable to be-
of the edge gardens posed varying limitations to Johnsongrass come common in drier climates and higher latitudes. These 
demography. We showed that Johnsongrass populations from limits to range expansion may be partially explained by a 
across its range can survive and sexually reproduce in New combination of smaller size, resulting from early flowering in 
Mexico and New York, well beyond the range centre. However, the northern and western edges, and the lack of overwinter 
at the northern edge, no Johnsongrass populations survived survival in the northern edge. Abiotic and biotic limitations 
the winter, which agrees with our previous work showing rhi- do not seem to uniformly impact Johnsongrass range expan-
zome fragments being particularly sensitive to cold temperat- sion and population performance, but they clearly interact 
ures, much more so than seeds (Lakoba et al. 2021b). At the in complex ways across large temperature and precipitation 
western edge, all Johnsongrass populations survived the arid gradients. We show the strong role of plant competition in 
growing season and winter, suggesting we had not yet found limiting Johnsongrass growth across its range; however, bi-
the climatic limitations to range expansion. However, given otic interactions did not change which populations performed 
the rarity of extant Johnsongrass in New Mexico, and the fact best in each garden. Multi-year demographic studies would 
that all known populations occur where water is available help elucidate the life history stages that might be limiting 
yearlong, more information is needed to understand what Johnsongrass population growth at both peripheries (Metcalf 
limits Johnsongrass populations at its Western range edge. and Pavard 2007).
Our previous work has shown that seedling establishment is Latitudinal and longitudinal range edges present unique 
the critical life history transition for Johnsongrass population limitations to future spread. Both edge populations per-
demography (Atwater et al. 2017; Lakoba and Barney 2020), formed poorly relative to other populations across all gar-
and in this study, we transplanted several-week-old seedlings dens, including their home edge gardens, suggesting that 
and watered them for up to 2 weeks. Thus, we hypothesize existing edge populations are not locally adapted and may in 
that seedling germination and/or establishment may play an fact be maladapted, though this requires further study. Our 
important role in arid environments. A targeted study of this study highlights the importance of incorporating range core, 
life stage would be needed to test this hypothesis. intermediate and edges in studies of range expansion in in-
vasive species; such continental-scale studies are often called 
Biotic interactions for but rarely done (e.g. Sexton et al. 2009). Including the 
We found strong evidence that climate interacted with com- range, edge can provide important insight into biological in-
petition to determine performance across the range. Weed vasions as well as species distributions and range limits in 
competition predictably reduced Johnsongrass performance general (Moran and Alexander 2014). A clear understanding 
in all gardens and tended to normalize biomass growth. of the drivers and limitations of invasive species range ex-
Competition was most intense in the Virginia and New pansion is urgently needed as global change is projected to 
Mexico gardens where Johnsongrass grew largest without shift the ranges of invasive species (Bellard et al. 2013) and 
competition and had less impact in Georgia, Texas and New enable some native species to expand beyond their historic 
Downloaded from https://academic.oup.com/aobpla/article/15/6/plad070/7352046 by guest on 15 February 2024
Fletcher et al. – Adaptive constraints at range edge 11
boundaries (Essl et al. 2019). These interactions at the range authors contributed to revisions and the final draft of the 
edge where abiotic, biotic and demographic factors are manuscript.
strongest will determine the fate of range expansion (Sexton 
et al. 2009). Range expansion at the edge will also present 
a suite of complex socio-political challenges as species move Conflicts of Interest
into novel regions (Essl et al. 2019), suggesting more studies None declared.
of range-expanding native and invasive species should be 
prioritized.
Funding
We acknowledge Virginia Tech College of Agriculture and 
Supporting Information Life Sciences and National Institute of Food and Agriculture 
The following additional information is available in the on- grant 2015-68004-23492 to J.N.B. and A.P. for partial sup-
line version of this article – port of this work.
Appendix 1: Calculation of biomass and height relative to 
the phytometer.
Appendix 2: Additional results tables. Data Availability
Table S1. Pair-wise comparisons for significant effects of All raw data are being made publicly available, DOI forth-
the mixed-effects models for the latitudinal gradient.  coming. Supporting information is included.
Table S2: Pair-wise comparisons for significant effects of 
the mixed-effects models for the longitudinal gradient.
Table S3: Proportion of plants that survived in each popu- Literature Cited
lation, treatment, and garden. Alexander JM, Edwards PJ. 2010. Limits to the niche and range mar-
Appendix 3: New Mexico Irrigation Sub-experiment. gins of alien species. Oikos 119:1377–1386.
Table S4: Results of mixed-effects models assessing the Angert AL. 2009. The niche, limits to species’ distributions, and 
effects of population, irrigation, treatment, and their inter- spatiotemporal variation in demography across the elevation 
actions on cube-root biomass, height, and flowering time of ranges of two monkeyflowers. Proceedings of the National Acad-
emy of Sciences of the United States of America 106:19693–19698.
three Johnsongrass populations and one phytometer (control) Atwater DZ, Sezen UU, Goff V, Kong W, Paterson AH, Barney JN. 
population planted in a common garden in New Mexico. 2015. Reconstructing changes in the genotype, phenotype, and cli-
Table S5: Proportion of plants that survived in each popu- matic niche of an introduced species. Ecography 39:894–903.
lation, treatment, and irrigation regime in the New Mexico Atwater DZ, Kim W, Tekiela DR, Barney JN. 2017. Competition and 
garden. propagule density affect sexual and clonal propagation of a weed. 
Figure S1: Effects of population, treatment, and garden on Invasive Plant Science and Management 10:17–25.
Johnsongrass above-ground biomass in the New Mexico gar- Atwater DZ, Fletcher RA, Dickinson CC, Paterson AH, Barney JN. 
den under two irrigation regimes: irrigated and not irrigated. 2018. Evidence for fine-scale habitat specialisation in an invasive 
Figure S2: Effects of population, treatment, and garden on weed. Journal of Plant Ecology 11:rtw124–rtw199.
Johnsongrass height in the New Mexico garden under two Baker HG 1965 Characteristics and modes of origin of weeds. In: 
Baker HG, Stebbins HG, eds. The genetics of colonizing species. 
irrigation regimes: irrigated and not irrigated. New York, NY: Academic Press, 147–172
Figure S3: Effects of population, treatment, and garden Bates D, Maechler M, Bolker B, Walker S. 2015. Fitting linear mixed-
on Johnsongrass flowering time in the New Mexico garden effects models using lme4. Journal of Statistical Software 67: 
under two irrigation regimes: irrigated and not irrigated. 1–48.
Appendix 4: Effects of competition on Johnsongrass bio- Bellard C, Thuiller W, Leroy B, Genovesi P, Bakkenes M, Courchamp 
mass. F. 2013. Will climate change promote future invasions? Global 
Table S6: Tests for the effect of population, garden, and Change Biology 19:3740–3748.
their interaction on competitive effect along a latitudinal and Blanquart F, Kaltz O, Nuismer SL, Gandon S. 2013. A practical guide to 
a longitudinal gradient. measuring local adaptation. Ecology Letters 16:1195–1205.
Figure S4: Results along the latitudinal (temperature) gra- Blows MW, Hoffmann AA. 2005. A reassessment of genetic limits to 
evolutionary change. Ecology 86:1371–1384.
dient of Johnsongrass’ North American range. Bontrager M, Angert AL. 2019. Gene flow improves fitness at a range 
Figure S5: Results along the longitudinal (precipitation) edge under climate change. Evolution Letters 3:55–68.
gradient of Johnsongrass’ North American range. Bridle JR, Vines TH. 2007. Limits to evolution at range margins: when 
and why does adaptation fail? Trends in Ecology & Evolution 
22:140–147.
Acknowledgements Bullock JM, Edwards RJ, Carey PD, Rose RJ. 2000. Geographical sep-
We would like to thank Barney, Lehnhoff, DiTommaso, aration of two Ulex species at three spatial scales: does competition 
Paterson and Bagavathiannan Labs for assistance with field- limit species’ ranges? Ecography 23:257–271.
work and data collection. We thank Anna S. Westbrook for Chuang A, Peterson CR. 2016. Expanding population edges: theories, 
helpful suggestions on an earlier version of this manuscript. traits, and trade-offs. Global Change Biology 22:494–512.
Colautti RI, Barrett SCH. 2010. Natural selection and genetic con-
straints on flowering phenology in an invasive plant. International 
Contributions by the Authors Journal of Plant Sciences 171:960–971.
Colautti RI, Barrett SCH. 2013. Rapid adaptation to climate facilitates 
R.A.F., J.N.B., D.Z.A. and D.C.H. conceived of the idea and range expansion of an invasive plant. Science 342:364–366.
designed the project. R.A.F., M.V.B., A.D., E.L., A.H.P., S.A., Colautti RI, Eckert CG, Barrett SCH. 2010. Evolutionary constraints 
P.G., C.L. and L.R. collected the data. R.A.F. conducted the on adaptive evolution during range expansion in an invasive plant. 
data analysis and wrote the first draft of the manuscript. All Proceedings Biological Sciences 277:1799–1806.
Downloaded from https://academic.oup.com/aobpla/article/15/6/plad070/7352046 by guest on 15 February 2024
12 AoB PLANTS, 2023, Vol. 15, No. 6 
Csergő AM, Salguero-Gómez R, Broennimann O, Coutts SR, Guisan A, Moran EV, Alexander JM. 2014. Evolutionary responses to global 
Angert AL, Welk E, Stott I, Enquist BJ, McGill B, et al. 2017. Less change: lessons from invasive species. Ecology Letters 17:637–649.
favourable climates constrain demographic strategies in plants. Paterson AH, Kong WQ, Johnston RM, Nabukalu P, Wu G, Poehlman 
Ecology Letters 20:969–980. WL, Goff VH, Isaacs K, Lee T-H, Guo H, et al. 2020. The evolution 
Essl F, Dullinger S, Genovesi P, Hulme PE, Jeschke JM, Katsanevakis S, of an invasive plant, Sorghum halepense L (‘Johnsongrass’). Fron-
Kühn I, Lenzner B, Pauchard A, Pyšek P, et al. 2019. A conceptual tiers in Genetics 11:317.
framework for range-expanding species that track human-induced Pironon S, Papuga G, Villellas J, Angert AL, García MB, Thompson 
environmental change. BioScience 69:908–919. JD. 2017. Geographic variation in genetic and demographic per-
Gaston KJ. 1990. Patterns in the geographical ranges of species. Bio- formance: new insights from an old biogeographical paradigm. 
logical Reviews 65:105–129. Biological Reviews of the Cambridge Philosophical Society 
Geber MA. 2008. To the edge: studies of species’ range limits. New 92:1877–1909.
Phytologist 178:228–230. R Core Team 2018 R: a language and environment for statistical com-
Hargreaves AL, Samis KE, Eckert CG. 2014. Are species’ range limits puting. Vienna: R Foundation for Statistical Computing, https://
simply niche limits writ large? A review of transplant experiments www.R-project.org/
beyond the range. The American Naturalist 183:157–173. Richards CL, Bossdorf O, Muth NZ, Gurevitch J, Pigliucci M. 2006. 
Holt RD. 2009. Up against the edge: invasive species as testbeds for Jack of all trades, master of some? On the role of phenotypic plas-
basic questions about evolution in heterogeneous environments. ticity in plant invasions. Ecology Letters 9:981–993.
Molecular Ecology 18:4347–4348. Rout ME, Chrzanowski TH, Smith WK, Gough L. 2013. Ecological 
Holt RD, Keitt TH, Lewis MA, Maurer BA, Taper ML. 2005. Theor- impacts of the invasive grass Sorghum halepense on native tallgrass 
etical models of species’ borders: single species approaches. Oikos prairie. Biological Invasions 15:327–339.
108:18–27. Sagarin RD, Gaines SD. 2002. The ‘abundant centre’ distribution: to 
Jump AS, Woodward FI. 2003. Seed production and population density what extent is it a biogeographical rule? Ecology Letters 5:137–
decline approaching the range-edge of Cirsium species. The New 147.
Phytologist 160:349–358. Sanford E, Roth MS, Johns GC, Wares JP, Somero GN. 2003. Local 
Kirkpatrick M, Barton NH. 1997. Evolution of a species’ range. The selection and latitudinal variation in a marine predator-prey inter-
American Naturalist 150:1–23. action. Science (New York, N.Y.) 300:1135–1137.
Körner C, Basler D, Hoch G, Kollas C, Lenz A, Randin CF, Vitasse Sexton JP, McInyre PJ, Angert AL, Rice KJ. 2009. Evolution and ecol-
Y, Zimmermann NE. 2016. Where, why and how? Explaining the ogy of species range limits. Annual Review of Ecology, Evolution, 
low-temperature range limits of temperate tree species. Journal of and Systematics 40:415–436.
Ecology 104:1076–1088. Sezen UU, Barney JN, Atwater DZ, Pederson GA, Pederson JF, Chand-
Lakoba VT, Barney JN. 2020. Home climate and habitat drive ecotypic ler JM, Cox TS, Cox S, Dotray P, Kopec D, et al. 2016. Multi-phase 
stress response differences in an invasive grass. AoB Plants US spread and habitat switching of a post-columbian invasive, Sor-
12:plaa062. ghum halepense. PLoS One 11:e0164584–e0164514.
Lakoba VT, Atwater DZ, Thomas VE, Strahm BD, Barney JN. 2021a. A Smith AL, Atwater DZ, Kim W, Haak DC, Barney JN. 2021. Inva-
global invader’s niche dynamics with intercontinental introduction, sive plant rhizome production and competitiveness vary based on 
novel habitats, and climate change. Global Ecology and Conserva- neighbor identity. Journal of Plant Ecology 14:638–647.
tion 31:e01848. Stanton-Geddes J, Tiffin P, Shaw RG. 2012. Role of climate and com-
Lakoba VT, Welbaum GE, Seiler JR, Barney JN. 2021b. A perennial petitors in limiting fitness across range edges of an annual plant. 
invader’s seed and rhizome differ in cold tolerance and apparent Ecology 93:1604–1613.
local adaptation. NeoBiota 70:1–21. Strobl K, Schmidt C, Kollmann J. 2018. Selecting plant species and 
Lande R, Arnold SJ. 1983. The measurement of selection on correlated traits for phytometer experiments. The case of peatland restor-
characters. Evolution 37:1210–1226. ation. Ecological Indicators 88:263–273.
Lenth R 2019 Emmeans: estimated marginal means, aka least-squares Theoharides KA, Dukes JS. 2007. Plant invasion across space and time: 
means. R package version 1.3.4. factors affecting nonindigenous species success during four stage of 
Louthan AM, Doak DF, Angert AL. 2015. Where and when do spe- invasion. The New Phytologist 176:256–273.
cies interactions set range limits? Trends in Ecology & Evolution Thomas CD. 2010. Climate, climate change and range boundaries. Di-
30:780–792. versity and Distributions 16:488–495.
Louthan AM, Pringle RM, Goheen JR, Palmer TM, Morris WF, Doak Warwick SI, Black LD. 1983. The biology of Canadian weeds. 61. 
DF. 2017. Aridity weakens population-level effects of multiple spe- Sorghum halepense (L) Pers. Canadian Journal of Plant Science 
cies interactions on Hibiscus meyeri. Proceedings of the National 63:997–1014.
Academy of Sciences 115:543–548. Warwick SI, Thompson BK, Black LD. 1984. Population variation in 
Maity A, Young B, Subramanian N, Bagavathiannan M. 2022. Pollen- Sorghum halepense, Johnson grass, at the northern limits of its 
mediated transfer of herbicide resistance between johnsongrass range. Canadian Journal of Botany 62:1781–1790.
(Sorghum halepense) biotypes. Scientific Reports 12:7663. Warwick SI, Phillips D, Andrews C. 1986. Rhizome depth: the critical 
McDonald A, Riha S, DiTommaso A, DeGaetano A. 2009. Climate factor in winter survival of Sorghum halepense (L) Pers (Johnson 
change and the geography of weed damage: analysis of US maize grass). Weed Research 26:381–388.
systems suggests the potential for significant range transformations. Watling JI, Orrock JL. 2010. Measuring edge contrast using biotic cri-
Agriculture, Ecosystems & Environment 130:131–140. teria helps define edge effects on the density of an invasive plant. 
McWhorter CG. 1961. Morphology and development of Johnsongrass Landscape Ecology 25:69–78.
plants from seeds and rhizomes. Weeds 9:558–562. Younginger BS, Sirová D, Cruzan MB, Ballhorn DJ. 2017. Is biomass 
Metcalf CJE, Pavard S. 2007. Why evolutionary biologists should be a reliable estimate of plant fitness? Applications in Plant Sciences 
demographers. Trends in Ecology and Evolution 22:205–212. 5:1600094.
Downloaded from https://academic.oup.com/aobpla/article/15/6/plad070/7352046 by guest on 15 February 2024