Skip to main content
Dryad logo

Ecological data from: Combining botanical collections and ecological data to better describe plant community diversity


Alba, Christina; Levy, Richard; Hufft, Rebecca (2020), Ecological data from: Combining botanical collections and ecological data to better describe plant community diversity, Dryad, Dataset,


In this age of rapid biodiversity loss, we must continue to refine our approaches to describing variation in life on Earth. Combining knowledge and research tools from multiple disciplines is one way to better describe complex natural systems. Understanding plant community diversity requires documenting both pattern and process. We must first know which species exist, and where (i.e., taxonomic and biogeographic patterns), before we can determine why they exist there (i.e., ecological and evolutionary processes). Floristic botanists often use collections-based approaches to elucidate biodiversity patterns, while plant ecologists use hypothesis-driven statistical approaches to describe underlying processes. Because of these different disciplinary histories and research goals, floristic botanists and plant ecologists often remain siloed in their work. Here, using a case study from an urban greenway in Colorado, USA, we illustrate that the collections-based, opportunistic sampling of floristic botanists is highly complementary to the transect- or plot-based sampling of plant ecologists. We found that floristic sampling captured a community species pool four times larger than that captured using ecological transects, with rarefaction and non-parametric species estimation indicating that it would be prohibitive to capture the “true” community species pool if constrained to sampling within transects. We further illustrate that the discrepancy in species pool size between approaches led to a different interpretation of the greenway’s ecological condition in some cases (e.g., transects missed uncommon cultivated species escaping from nearby gardens) but not others (e.g., plant species distributions among functional groups were similar between species pools). Finally, we show that while using transects to estimate plant relative abundances necessarily trades off with a fuller assessment of the species pool, it is an indispensable indicator of ecosystem health, as evidenced by three non-native grasses contributing to 50% of plant cover along the highly modified urban greenway. We suggest that actively fostering collaborations between floristic botanists and ecologists can create new insights into the maintenance of species diversity at the community scale.


We sampled plant communities along the High Line Canal greenway, a 66-mile recreational trail that passes through 11 municipalities in the Denver-Metro area of Colorado. The trail runs alongside a 71-mile earthen canal (owned by Denver Water), which was excavated in the late 1800s to support agriculture and human settlement in what was historically native plains and foothills shrubland vegetation. The greenway thus represents a human-created waterway that is highly managed, yet supports a species pool that contains native flora.

The greenway’s length (with the Canal’s inception located at 39.48362, -105.11293) is demarcated by mile markers that we used to generate a random subset of 45 locations at which the botanical and ecological field crews could synchronize their data collection. As is typical for collections-based floristic surveys, the botanical crew sampled exhaustively from early spring (7 May) through late summer (28 September) to capture early-, middle-, and late-season species. Starting at each of the 45 mile markers, the crew walked in the Canal’s downstream direction, searching the greenway for newly encountered species to collect and accession to the Kathryn Kalmbach Herbarium (KHD) at Denver Botanic Gardens . Most mile marker locations were sampled once during the inventory, but a few were revisited if they occurred in a vegetation type that would not be re-encountered later in the season at the other mile markers (e.g., mile markers zero and one at the inception of the Canal were the only locations in the foothills shrubland Ecoregion).

We used a staggered sampling design in which 5 mile markers spanning the southwestern to northeastern extent of the Canal were sampled every other week from May to September. The floristic survey was carried out over 57 days, comprising 850 search-hours and an estimated distance covered of 42 miles (calculated from our daily starting and stopping waypoints logged with a GPS unit). The botanical crew consisted of two botanists trained in the local flora and one to two additional non-botanists who assisted with specimen collection. All members of the crew searched for species within an ~50 to 75-foot-wide viewshed moving from the bed of the Canal, up the Canal bank, across the greenway trail, and over to the property line that marked the end of Denver Water’s ownership. High-veracity (with identifying structures) herbarium specimens were accessioned for every species encountered during the floristic survey (numbering 1570 specimens, including duplicates; collections data available). Identifications were made using Ackerfield's Flora of Colorado (2015), Wingate's Illustrated Keys to the Grasses of Colorado (1994), and Wingate's Sedges of Colorado (2017).

The ecological sampling was carried out over 10 days, from May 22, 2018 to June 6, 2018, to capture a snapshot of plant communities around peak biomass. This method of deploying a concerted sampling effort over a short time period is common in ecological sampling, because it is often of interest to detect treatment differences that could be obscured by confounding time lags between sample dates (as opposed to the floristic botany goal of exhaustively delineating a species pool over time). At each of the 45 miles markers, we laid a 12 m × 2 m transect, the length of which captured habitat variation across the greenway corridor. We used the line-point intercept method to make field observations of plant species presence every 0.25 m along the 12 m transect (as well as bare ground, plant litter, and rocks, which we do not report herein). In the associated data set, the “first hit” was used to generate the reported percent cover estimates (number of hits per species per total number of hits), while the “second hit” was used to add species to our presence list. We also searched each of the two, one-meter-wide belt transects for additional species that were not encountered along the line-point transect. Voucher specimens were collected for the species encountered during the observational ecological sampling (collected outside the transects so as not to influence long-term sampling). However, given the short time period of the ecological sampling, not all specimens had flowers or fruits, and therefore were not of sufficient quality to be curated. All specimens were kept during the field season and subsequent analyses to facilitate identifications, but only higher quality specimens were accessioned to the herbarium. Please note that one example specimen exists for potentially hundreds of field observations (i.e., each time a species was encountered along the transects).

We chose the line-point method as the most appropriate for our system, with its narrow and steep canal bank that could not accommodate other plot designs. Additionally, our aim with the ecological transects was to estimate not only species presence, but also composition. For questions about composition, the line-point method is highly repeatable across individuals and rapidly deployed, thereby maximizing sampling replicates across many locations in a single season. Such transects will capture fewer species than other methods (e.g., Modified-Whitaker plots); however, any bounded sampling approach will cover considerably less area than can be achieved with opportunistic sampling based in the floristic tradition of using the habitat itself as the sampling unit. Importantly, we note that it was not our goal to equalize the temporal or spatial scales of the two sampling approaches (which in our experience is not often done in practice), but rather to sample in a manner that is broadly consistent with collections-based versus ecological disciplines. The particulars of our comparison, such as the sampling window and the use of transects rather than any number of plot types, contextualize the results.

Usage Notes

Transect Observations Data

This data set contains each measurement taken via the line-point intercept method in each transect surveyed as well as each additional species observed within the 2 meter wide belt transects. InOut values indicate whether the species was observed within the transect via the line-point intercept method (I) or as an additional species within the 2 meter wide belt transect (O). Transect.Point.Hit values indicate the distance from the transect's origin at which the observaiton was recorded. Hit.Number values indicate whether the species was detected as top canopy (1) or lower canopy (2) per the line-point intercept method. NULL values recorded in Species.Hit indicate the detection of  ground surface (non-plant value, such as bare soil). Transect.Latitude indicates that latitude at which the site of the transect was established. Transect.Longitude indicates the longitude at which the site of the transect was established.