Detectability of species of Carex varies with abundance, morphology, and site complexity

Abstract

Questions: Are graminoids more poorly detected than other life forms of vascular plants in surveys? How well do observer-, species-, and site-specific variables explain variation in detection of Carex species across forests of different structure? Location: Northeastern Alberta, Canada. Methods: Species inventories were assessed within 50 belt transects, each 100 m in length and 2 m in width. Pseudoturnover was estimated for four life forms and all encountered species. Site-specific factors were then compared with pseudoturnover of all vascular plants and graminoids using generalized linear regression. Carex detection probabilities were compared based on morphological groups. Detection success at a site and delays in detection within a site were assessed using logistic regression with AIC used to rank a-priori hypotheses and standardized variables used to determine effect sizes of parameters related to plant detectability.Results: Pseudoturnover for graminoids was similar to that for other life forms and best related to ground layer cover. Morphological groups related to differences in detection, with short, small-inflorescence Carex most poorly detected. Detection failure was best explained by species abundance and morphology, but delays were more tied to a site's vegetation structure and species abundance than to species morphology.Conclusions: Surveys targeting graminoids, including species of Carex, can achieve high detection rates with high survey effort over small areas, but should consider species-and site-specific biases in detection success. Abundance is likely the most influential factor in determining detection success, and this must be accounted for when searching for low-density species. We recommend that increased effort (time, repeat observations) be applied when searching for morphologically small graminoids. K E Y W O R D S

Observer‐driven pseudoturnover in vegetation monitoring is context‐dependent but does not affect ecological inference

“…Similar to the findings of Dennett and Nielsen (2018), who reported no differences in pseudoturnover across life forms (including graminoids), we found no differences when comparing graminoids vs non‐graminoids. In addition, Dennett and Nielsen (2018) detected only a weak effect of horizontal and total vegetation cover on pseudoturnover, which supports our findings of a non‐significant relationship between pseudoturnover and vegetation cover.…”

“…Similar to the findings of Dennett and Nielsen (2018), who reported no differences in pseudoturnover across life forms (including graminoids), we found no differences when comparing graminoids vs non‐graminoids. In addition, Dennett and Nielsen (2018) detected only a weak effect of horizontal and total vegetation cover on pseudoturnover, which supports our findings of a non‐significant relationship between pseudoturnover and vegetation cover. However, this result is in contrast to that of Vittoz and Guisan (2007), who found significantly higher pseudoturnover in plots with dense vegetation than in plots with open vegetation, indicating that species are easier to detect in open vegetation and that high vegetation coverage might be a source of error and should not be neglected in future studies.…”

“…Similarly, Milberg et al (2008) and Morrison et al (2020), who studied inter‐observer differences during resurveys of permanent plots in Swedish boreal forests and in wetland habitats in Ohio (USA), respectively, reported that most species that were overlooked in double surveys were in the two lowest cover classes. Likewise, Dennett and Nielsen (2018), who studied graminoids vs other life forms in Carex ‐rich vegetation types of Alberta (Canada), found that the abundance of a species was one of the most important factors predicting its detectability. Other studies additionally included investigations of whether species traits might help explain missed occurrences.…”

“…testing whether pseudoturnover was driven by differences in graminoid species between observers (e.g. Dennett & Nielsen, 2018). To explore why a taxon was overlooked by one of the observers, we separated the number of observations summed across all plots per Braun‐Blanquet cover category into shared taxa between the two observers and taxa that were overlooked by either one of the two observers.…”

“…The detectability of species has been proposed to play a major role in this probability of being overlooked. Detectability can be influenced by population size, morphology and phenology (Garrard et al, 2013; Bornand et al, 2014; Dennett & Nielsen, 2018). In addition, there are other factors that can affect pseudoturnover, including seasonal vegetation changes (Kirby et al, 1986; Kapfer et al, 2017), plot size (larger plots have higher pseudoturnover; e.g.…”