Study Area.

This study was conducted in P. ponderosa forests of the Arapaho and Roosevelt National Forests (ARNF) in the Colorado Front Range after the 2020 Cameron Peak Fire, which is considered one of the largest wildfires in the history of Colorado (NPS 2021) (Fig. 1). The ARNF covers about 1.5 million acres and is primarily coniferous. Pinus ponderosa dominates at elevations between 1,800 and 2,500 m, accompanied by understory species such as graminoids, forbs, and shrubs (Fryer 2018). Other conifers like Pinus contorta Douglas ex Loudon (lodgepole pine), Pseudotsuga menziesii (Mirb.) Franco (Douglas-fir), Pinus flexilis James (limber pine), and Juniperus scopulorum Sarg. (Rocky Mountain juniper) appear at mid-high elevations (2000–2500 m). Plots receive an average annual precipitation of 38.3 cm, with temperatures varying from 2 °C in January to 32 °C in June (data from 1991 to 2020; Western Regional Climate Center 2007).

Fig. 1.

Map of the 40 plots sampled from understory vegetation distributed within burned (n = 20) and unburned (n = 20) Pinus ponderosa stands in the Colorado Front Range.

Field and Understory Vegetation Measurements.

Sampling area was delineated and informed based upon the combination of historical fire data in the Colorado Front Range and on-site inspection. Prior to onfield visit, we reviewed historical data on the Cameron Peak wildfire to provide guidance on sampling locations and understand the spatial extent and severity of the fire (ArcGIS 2023). We then systematically drove through P. ponderosa stands within the potential sampling area to inspect potential burned and unburned plots. Based upon onfield inspection, the 20 burned plots were sampled by assessing P. ponderosa stands that had experienced complete canopy fire consumption during the wildfire event. Similarly, the 20 unburned plots used as controls were sampled plots that had not experienced fire damage in at least the last two decades (Fig. 2). Although sampling techniques relied on visual assessment during field visits, the locations of plots were recorded using GPS devices (Garmin eTrex 32x, Garmin Ltd., Olathe, KS) to ensure accurate locations of plots. Plot sampling was conducted in a systematic manner to minimize bias and ensure representation across the study area. Each of the 40 total burned and unburned plots was 10 m × 10 m, with plots at least 50 m apart. In every sampled plot, we took the following environmental data from the plot center: elevation, slope, aspect, longitude, and latitude (Appendix 1).

Fig. 2.

Photograph of the study area showing burned (A) and unburned (B) Pinus ponderosa stands in the Colorado Front Range.

Past studies assert that using plant cover data effectively characterizes plant communities, describes plant-environment interactions, and monitors plant ecosystems through time (Mueller-Dombois and Ellenberg 1974; Anderson 1986). Visual estimates of relative percent cover were done using the ocular method of estimating percent cover for all plant species in each plot, both clonal and nonclonal (Helm and Mead 2004). We estimated percent cover of each plant species separately within each plot by pooling plant species abundances before calculating relative percent cover. The choice of percent cover over an ordinal cover scale is supported by a recent study by Dengler and Dembicz (2023). In their study, the authors found that direct percent cover estimation produces lower mean absolute and relative errors compared to the use of ordinal scales, particularly for cover values above 1% (Dengler and Dembicz 2023). If a plant could not be identified, we followed the Assessment Inventory and Monitoring methods (Toevs et al. 2011) and initially named that plant as an unknown species. The plant was collected, with detailed descriptions and photos, brought to the University of Northern Colorado herbarium to be keyed and identified, and once identified, reclassified to species in the data set. Taxonomic identification followed Ackerfield (2015). CLO-PLA 3 (Klimeš and Klimešová 1999; Klimešová and Klimeš 2008), and TRY plant trait databases (Kattge et al. 2020) were used to label species as clonal or nonclonal based on their dominant CGO such as epigeogenous rhizome, hypogeogenous rhizome, bulb, root splitter, and horizontal aboveground stem. In addition, the USDA PLANTS Database (USDA 2024) was used to provide other ecological traits of species, including life form (forb, shrub or graminoid) and the photosynthetic pathway that each species utilizes (C3, C4, or CAM).

Clonal Traits Measurements.

Ten plots each from burned and unburned areas were selected to study clonal traits' responses in A. uva-ursi and A. parvifolia. These plots were chosen based on their distribution. We aimed to compare lateral spread and ramet length between burned and unburned plots, as these traits indicate clonal reproduction and recovery postdisturbance (Klimeš and Klimešová 2008; Pausas et al. 2018). Measurements followed Klimešová et al. (2019) protocols, with lateral spread measured between offspring and rooting units, and ramet length from base to tip. Five measurements per plant were used to calculate an average, and five individuals were averaged for each plot.

Data Analyses.

We were primarily interested in assessing the relative percent cover (i.e., cover of all species in each plot summed to 100%) and diversity of clonal plants in burned and unburned plots of P. ponderosa stands. Diversity variables (i.e., richness and evenness) were calculated using relative cover of full species matrix for 40 total plots with 27 clonal taxa/species in PC-ORD. Since percent cover and diversity data did not meet assumptions of normality and homoscedasticity of variances, unpaired nonparametric Mann-Whitney U-tests were conducted to assess whether each of our response variables—relative percent cover, richness, and evenness—differed between burned versus unburned plots for clonal plants. For clonal plant trait measurements, a MANOVA (Multivariate Analysis of Variance) was used to detect the effects of burned versus unburned plots on lateral spread and ramet length for each clonal plant A. uva-ursi and A. parvifolia in a combined model. Although lateral spread and ramet length may respond differently, analyzing them together allowed for a comprehensive assessment of trait responses. All analyses were conducted using R Statistical Software (version 4.3.1; R Core Team 2021, Vienna, Austria).

PC-ORD 7.10 (McCune and Mefford 1999) was used to conduct multivariate analyses to study species composition differences in burned and unburned plots and the roles of clonal and nonclonal plants. We used four multivariate analyses: NMS (nonmetric multidimensional scaling), MRPP (multiresponse permutation procedures), ISA (indicator species analysis), and FCA (fourth-corner analysis) to explore trait-environment relationships.

Nonmetric multidimensional scaling ordinates points (burned and unburned plots in our case) in a reduced-dimensional space, preserving the pairwise dissimilarity among them. This helps to visualize patterns and relationships in ecological communities (McCune and Grace 2002). Nonmetric multidimensional scaling was used because the data did not follow linear, parametric assumptions for normality and homogeneity of variances, which is common with ecological community data (Clarke 1993; McCune and Grace 2002). Autopilot with the “Slow and Thorough” setting using “Sørensen distance” measure was used to run 50 iterations to create a configuration for the ordination including 104 clonal and nonclonal taxa/species in the primary species matrix and 40 total burned and unburned plots (Appendix 2). The second matrix had five environmental variables collected at each plot from the center (elevation, slope, aspect, latitude, and longitude). Multi-response permutation procedures and ISA were performed to complement and help interpret the NMS ordination.

Multi-response permutation procedures is a nonparametric statistical test that determines whether there are significant differences between groups based on multivariate response variables (Pillar 2013). In this case, we used MRPP to assess the dissimilarity in species composition between burned and unburned plots using the “Sørensen distance” measure. Based on the weighting technique, the analysis can generate the T statistic, which measures the difference between groups, with more negative values indicating stronger separation; the A statistic, which reflects within-group homogeneity, ranging from –1 (no agreement) to 1 (perfect agreement); and P, which is the probability of smaller or equal delta (McCune and Mefford 1999).

Indicator species analysis was used to determine which plants were significantly associated with burned and unburned plots. The analysis helps to determine which species (clonal or nonclonal) are indicators of these specific environmental conditions (Dufrêne and Legendre 1997; McCune and Grace 2002).

The fourth-corner analysis explores species' trait-environment relationships. It is called the “fourth-corner” because it combines three corners of species ecological traits, environment variables, and species cover data, and then leaves the fourth corner to examine the relationship among species, environmental variables, and plant ecological traits (Dray et al. 2014). Specifically in this study, FCA was used to determine whether environmental variables had any relationship with species ecological traits and with differences in species composition observed in the NMS ordination.

Relative Percent Cover and Diversity Area of Clonal Plants.

We identified 104 understory species (27 clonal and 77 nonclonal) in all 40 burned and unburned plots. Invasive species were rare in all plots. Clonal plants (Table 1) and nonclonal plants were mainly forbs, with a few shrubs and graminoids. Both plot types had an overstory dominated by P. ponderosa but included P. flexilis and P. menziesii to a lesser extent. Achillea millefolium L. (common yarrow), Campanula rotundifolia L. (bluebell bellflower), and A. parvifolia showed the greatest cover among forbs. Arctostaphylos uva-ursi, Artemisia ludoviciana Nutt. (white sagebrush), Artemisia frigida Willd. (prairie sagewort), Cercocarpus montanus Raf. (alder-leaf mountain mahogany), and Purshia tridentata (Pursh.) D.C. (antelope bitterbrush) were the shrubs with the most cover. In terms of graminoids, Carex rossii Boott (Ross' sedge), Bromus tectorum L. (cheatgrass), and Poa bulbosa L. (bulbous bluegrass) were the species with the greatest cover in our plots.

Table 1.

Clonal plants with dominant clonal growth organs (CGOs) in sampled plots of the Pinus ponderosa stands in the Colorado Front Range (from CLO-PLA 3 & TRY databases).

Clonal plant percent cover differed significantly between burned and unburned plots (U = 332; p < 0.001), being greater in burned plots (Fig. 3). Burned plots averaged 12 species, whereas unburned had 6, with standard deviations of 1.883 and 1.071, respectively. Clonal plant richness was significantly higher in burned plots (U = 397; p < 0.001; Fig. 4), but evenness did not differ between plot types (U = 203.5; p = 0.935; Fig. 5).

Fig. 3.

Relative percent cover of clonal plants from burned versus unburned Pinus ponderosa stands in the Colorado Front Range.

Fig. 4.

Richness of clonal plants from burned versus unburned Pinus ponderosa stands in the Colorado Front Range.

Fig. 5.

Evenness of clonal plants from burned versus unburned Pinus ponderosa stands in the Colorado Front Range.

Community Structure.

Species community composition in burned and unburned plots using NMS ordination yielded a three-dimensional solution explaining 64.3% of the variance (p = 0.004). The final stress level was 18.11 with a solution instability of < 0.0001 after 102 iterations. Nonmetric multidimensional scaling axis 1 explained 25.2% of the variance, capturing a significant portion of species composition variation. The positive end of axis 1 correlated with higher C. rossii, Pascopyrum Á. Löve (wheatgrass), Helianthus pumilus Nutt. (little sunflower), P. bulbosa, and Mertensia ciliata E. James ex. Torr G. Don (tall fringed bluebells). Conversely, the negative end of axis 1 correlated with Cirsium arvense (L.) Scop. (Canada thistle), Geranium sp. (geranium), and A. millefolium. The only environmental variable that was positively correlated with NMS axis 1 was slope, whereas elevation and latitude were negatively correlated. Plots formed distinct burned and unburned groups in ordination space (Fig. 6).

Fig. 6.

NMS ordination of the understory species composition for all 40 burned and unburned plots of Pinus ponderosa stands in the Colorado Front Range. A three-dimensional solution was recommended, but only two dimensions are shown here. Burned plots are marked by red triangles, whereas unburned plots are marked by light green triangles with five environmental variables (elevation, slope, aspect, latitude, and longitude).

Nonmetric multidimensional scaling axis 2 contributed an additional 23.4% of the variance. Species strongly correlated with the positive end of axis 2 included Taraxacum officinale F. H. Wigg. (common dandelion), Potentilla fissa Nutt. (big-flower cinquefoil), and C. rossii. Viburnum edule (Michx.) Raf. (squashberry), C. rotundifolia, and A. millefolium showed weaker positive correlations with axis 2. Species negatively correlated with this axis included Erigeron eximius Greene (sprucefir fleabane), Sedum lanceolatum Torr. (spearleaf stonecrop), P. bulbosa, and to a lesser extent, Draba graminea Greene (Rocky Mountain draba). All environmental variables had weak correlations with axis 2. However, the unburned plots tended to cluster at the negative end of axis 2, whereas burned plots clustered toward the positive end of this axis.

The third NMS axis accounted for 13.4% of the explained variance. The positive end of axis 3 had strong correlations with Castilleja puberula Rydb. (shortflower Indian paintbrush) and V. edule. The negative end was strongly correlated with A. ludoviciana, P. tridentata, Heterotheca villosa (Pursh) Shinners (hairy false goldenaster), H. pumilus, Bromus inermis Leyss. (smooth brome), and C. montanus. Slope was strongly negatively correlated, whereas elevation, aspect, longitude, and longitude all had negative correlations.

Multi-response permutation procedures (T = –9.970, A = 0.021, p ≤ 0.001) analysis revealed statistically significant differences in vegetation composition between burned versus unburned plots, albeit a low A value questions ecological significance. Indicator species analysis identified 16 indicator species for both burned and unburned plots out of the 104 taxa/species in the species matrix (p ≤ 0.05) with 63% clonal indicating burned plots and one clonal species indicating unburned plots (Table 2).

Table 2.

Indicator species analysis (ISA) results for species across study plots from Pinus ponderosa stands in the Colorado Front Range.

Fourth-corner analysis revealed no significant relationships between elevation, slope, latitude, longitude, and life form. Environmental variables had weak correlations with clonality (all p > 0.05). However, CGOs were associated with elevation and slope. Specifically, elevation related to horizontal aboveground stems, epigeogenous rhizomes, and root splitters (p = 0.025, p = 0.019, and p = 0.032, respectively), whereas slope related to epigeogenous rhizomes (p = 0.023). Environmental variables did not relate with photosynthetic type (all p > 0.05). Overall, elevation, slope, and fire history significantly influenced vegetation structure, consistent with NMS ordination results (Table 3).

Table 3.

Results of fourth-corner analysis (FCA) examining the relationship of environmental variables with species ecological traits from Pinus ponderosa stands in the Front Range of Colorado.

Clonal Traits Response in Burned and Unburned Plots.

MANOVA results suggested that the combined responses of lateral spread (LS) and ramet length (RL) of A. urva-ursi were significantly higher in burned plots compared to unburned plots (Pillai Trace F = 0.5624, p < 0.001; Fig. 7). Although these traits in A. urva-ursi varied in plasticity after fire in burned plots, unburned plots showed less variability. In contrast, A. parvifolia showed no significant differences in LS or RL between burned and unburned plots (Pillai Trace F = 0.1042, p = 0.393; Fig. 8), with unburned plots also displaying less trait variability.

Fig. 7.

Response of lateral spread and ramet size of Arctostaphylos uva-ursi for burned versus unburned Pinus ponderosa stands in the Colorado Front Range.

Fig. 8.

Response of lateral spread and ramet size of Antennaria parvifolia for burned versus unburned Pinus ponderosa stands in the Colorado Front Range.

Relative Percent Cover and Diversity of Clonal Plants.

Our study focused on the response of understory clonal plants in P. ponderosa stands in the Colorado Front Range following fire. We expected burned plots to show greater clonal plant cover and diversity than unburned plots. We also predicted that the different responses of clonal and nonclonal plants would result in significant changes in overall species composition between burned and unburned plots. In addition, we hypothesized that clonal traits of A. uva-ursi and A. parvifolia would differ in their clonal traits' responses between burned and unburned plots. Overall, the findings partially supported our hypotheses. Although burned plots did show increased clonal plant cover and diversity as predicted, the changes in species composition may not be ecologically significant. Additionally, the clonal traits responses of the two studied plants did not differ much between burned and unburned plots.

Several studies indicate that plant clonality plays a key role in ecosystem recovery and community restoration after disturbances (e.g., Fahrig et al. 1994; Oborny et al. 2000; Van Mantgem et al. 2020). Here, we found greater clonal plant cover in burned plots, attributing this to their resilience, colonization ability, and vegetative reproduction from strong bud banks and CGOs like rhizomes and stolons (Klimešová and Martínková 2022). In addition, postfire environments often exhibit increased heterogeneity, affecting soil, microclimate, and vegetation at broader scales (Schoennagel et al. 2008). Although our study focused on community-level effects, clonal plants thrive in these varied postfire landscapes, benefiting from quick regrowth and resource exploitation (Svensson et al. 2013). This increased cover is also linked to traits like physiological integration and resource allocation, crucial for clonal plant dominance in disturbed ecosystems (Liu et al. 2016).

Our data suggest that clonal plant species richness increases postfire. Clonal plants had more species in burned plots compared to unburned ones, consistent with other studies (e.g., McMahon et al. 2017; Pausas et al. 2018). However, clonal plant evenness did not differ between burned and unburned plots. Similar evenness suggests fire does not change the relative abundance of clonal species. This evenness could influence community stability and resilience postfire, affecting recovery ability (Sasaki and Lauenroth 2011; Battisti et al. 2016).

Community Structure.

Nonmetric multidimensional scaling ordination axis 1 primarily represents elevation gradient in vegetation community structure, likely influenced by topography, especially in unburned ponderosa pine plots. On the positive end of axis 1, species such as C. rossii, Pascopyrum sp., H. pumilus, P. bulbosa, and M. ciliata were associated at higher elevations, indicating cooler conditions. These species likely thrive in cooler conditions typical of higher elevations in the unburned ponderosa communities. The negative end of axis 1 was correlated with C. arvense, Geranium sp., and A. millefolium linked to lower elevations, possibly due to the dense P. ponderosa canopy cover. Although elevation emerges as the main driver of species along axis 1, particularly in unburned plots, there might be an interaction between elevation and fire history in shaping vegetation community structure.

Nonmetric multidimensional scaling axis 2 represents a postfire disturbance and species-specific response gradient. Burned plots mainly cluster at the positive end, including species such as T. officinale, C. rossii, and P. fissa, which can thrive postfire due to clonality or wind-dispersed seeds (Toth 1991). These species benefit from postfire microhabitats and establish quickly. On the negative end of axis 2, species like E. eximius, S. lanceolatum, P. bulbosa, and D. graminea apparently preferred less disturbed habitats. Further, unburned plots mostly clustered at the negative end of this axis and we believe fire regimes in the Colorado Front Range likely influenced these patterns. It should be noted that “unburned” really implies burned less recently, as all plots have burned at some time in this P. ponderosa ecosystem.

Nonmetric multidimensional scaling axis 3 corresponds to variation in species cover related to postfire vegetation structure. The positive end had only C. puberula and V. edule, which are not indicative of postfire environments in the literature. However, at the negative end, we saw species with greater cover commonly found in postfire environments. These species included A. ludoviciana, P. tridentata, H. villosa, B. inermis, and C. montanus. Species cover along this axis also correlated negatively with slope, indicating slope's role in shaping species distribution. Steeper slopes might experience different fire behavior, affecting postfire community structure (Bassett et al. 2015).

The data suggests that there are slight compositional differences between burned and unburned plots, but we interpret these cautiously. Indicator species analysis identified 16 significant indicator species, but only three had values above 60. This low indicator value is likely due to many species being widespread across plots. Clonal plants like A. millefolium, C. rotundifolia, C. arvense, Oxytropis campestris (L.) DC. (field locoweed), P. tridentata, T. officinale, and Poa pratensis L. (Kentucky bluegrass) were significant indicators in burned plots, aligning with our expectation that clonal plants respond more directly to fire disturbance. This supports the idea that clonal plants enhance postfire ecosystem resilience (Falk et al. 2019).

We found no relationships between clonality, life form, photosynthetic type, and environmental variables like elevation, slope, aspect, longitude, and latitude. This aligns with our prediction that these traits are relatively stable across species regardless of environmental variations. However, responses within these traits can vary; for example, clonal plants may exhibit diverse strategies postdisturbance.

Significantly, elevation and slope were related to certain CGOs, like horizontal aboveground stems and rhizomes. This suggests clonal plants allocate resources to these CGOs more at higher elevations or steeper slopes, possibly enhancing their ability to spread and access resources postdisturbance (Štõastná et al. 2012). Elevation also strongly correlated with NMS axis 1 and was related to CGOs in the FCA. This indicates that elevation and fire history interact to shape vegetation in these Colorado Front Range ecosystems, suggesting a nuanced response to environmental conditions (Klimešová et al. 2012).

Clonal Traits Response in Burned and Unburned Plots.

The responses of clonal plant species A. uva-ursi and A. parvifolia to fire varied, highlighting the importance of understanding each species' unique postdisturbance behavior (Franklin et al. 2020; Bell et al. 2023). Specifically, A. uva-ursi exhibited increased lateral spread and ramet length in burned plots, potentially aiding its recovery and growth postfire. Although few studies focus on these species' responses to fire, Hutchings and de Kroon (1994) found that clonal traits like lateral spread can vary based on environmental conditions and genetics. This trait plasticity in response to disturbance is vital for resource competition and foraging (Dong and de Kroon 1994; Hutchings and de Kroon 1994; Dietz and Steinlein 2001). Such variability has practical implications for ecosystem management, especially in fire-prone areas. It underscores the need for tailored management plans that account for each species' specific needs postfire.

In summary, this study highlights that elevation, slope, and fire history collectively influence the understory vegetation of P. ponderosa stands in the Colorado Front Range. Three years after the fire, burned plots showed increased relative cover and diversity of clonal plants, displaying their resilience and adaptability to disturbances. Our findings are important for foresters, managers, and modelers to understand clonal plant performance and response to changing fire regimes. Given the importance of clonal plants in postdisturbance recovery, future research should delve into the mechanisms behind their success. Long-term monitoring of postfire ecosystems will further elucidate the sustainability of clonal plants and their traits.

©Copyright 2025 by The Torrey Botanical Society