Addendum: IMPACTS OF A NON-INDIGENOUS ECOSYSTEM ENGINEER, THE AMERICAN BEAVER (CASTOR CANADENSIS), IN A BIODIVERSITY HOTSPOT

Jonathan Richmond^1* Camm C. Swift² Thomas Wake³ Cheryl S. Brehme¹ Kristine L. Preston¹ Barbara Kus¹ Edward L. Ervin⁴ Scott Tremor⁵ Tritia Matsuda¹ Robert N. Fisher¹

• ¹Western Ecological Research Center, United States Geological Survey (USGS), United States
• ²Natural History Museum of Los Angeles County, United States
• ³Cotsen Institute of Archaeology, College of Social Sciences, University of California, Los Angeles, United States
• ⁴Merkel & Associates, Inc, United States
• ⁵San Diego Natural History Museum, United States

Omission of a Data Availability Statement
The published article did not contain a Data Availability statement. We have added the Data Availability statement below.

Data Availability Statement
Data for the arroyo toad habitat occupancy analysis and R code for the Poisson generalized linear mixed effects model (GLM) and categorical regression analyses can be downloaded from the journal website.

The authors apologize for the omission and state that this does not change the scientific conclusions of the article in any way. The original article has been updated.

Missing Supplementary Material
In the published article, the following material was not included in the Supplemental File. The missing material appears below:

To further evaluate the relationship between the non-native species index (NNI) and beaver dam presence, we performed two sets of analyses using Bayesian multilevel models in the R package brms (Bürkner 2017). First, we fit a Poisson GLM, which included a fixed intercept, fixed effect of beaver presence, and a random effect for year to account for repeated measures (Duarte et al. 2018). We evaluated model fit by ensuring that all model parameters had Gelman-Rubin values <1.1, by visually inspecting the posterior samples for convergence, and by generating posterior predictive plots to measure within-sample predictive performance. We specified default priors and ran the model across 4 chains for 2000 iterations, with a warmup of 1000 iterations, and a thinning rate of 1, totaling 4000 post-warmup samples. This model showed that the probability of observing a higher NNI score increased in the presence of a beaver dam, with the estimated NNI being nearly three times greater (mean = 2.9; 95%CI = 2.1–3.9) in areas where beaver dams were present (mean = 1.9, 95%CI = 1.3–2.6) compared to areas where they were absent (mean = 0.7, 95%CI= 0.5–0.8).
We also performed a categorical regression treating NNI as a discrete state variable (Congdon 2005), with NNI modeled as a function of an intercept and beaver dam presence. We included each year as a random effect, and specified default priors for the intercepts at each NNI level and the effect of beaver dams on NNI. We evaluated model fit and performed the analysis using the same run specifications as described above. Default priors used for these analyses are viewable in the model object after it is run in R. This analysis confirmed that a higher NNI score was more probable when a beaver dam was present (Fig 2B, main text).

Bürkner, P.-C. (2017). brms: An R package for Bayesian multilevel models using Stan. J. Stat. Softw. 80, 1–28.
Congdon, P. 2005. Bayesian models for categorical data. New York, NY: John Wiley & Sons, Ltd.
Duarte, A., Adams, M.J., and Peterson, J.T. (2018). Fitting N-mixture models to count data with unmodeled heterogeneity: Bias, diagnostics, and alternative approaches. Ecol. Model. 374, 51–59.

The authors apologize for this omission and state that this does not change the scientific conclusions of the article in any way. The original article has been updated.

Corrigendum: IMPACTS OF A NON-INDIGENOUS ECOSYSTEM ENGINEER, THE AMERICAN BEAVER (CASTOR CANADENSIS), IN A BIODIVERSITY HOTSPOT

* Correspondence: jrichmond@usgs.gov

Keywords: same as original article

Corrigendum on: Richmond, J. Q., C. C. Swift, T. A. Wake, C. S. Brehme, K. L. Preston, B. E. Kus, E. L. Ervin, S. Tremor, T. Matsuda, and R. N. Fisher. 2021. Impacts of a non-indigenous ecosystem engineer, the American Beaver (Castor canadensis), in a biodiversity hotspot. Frontiers in Conservation Science 2 - 2021|https://doi.org/10.3389/fcosc.2021.752400

Incorrect Supplementary Material
In the published article, there was an error in the Supplementary File. In preparing an addendum to provide the original data for the arroyo toad habitat occupancy analyses, we discovered (1) that a beaver dam recorded in 2004 was inadvertently not included in the dataset, and (2) that removal of mosquitofish (Gambusia affinis) from the non-native index (NNI) in the 2013–2018 dataset (due to its ubiquity across the survey area) was not applied to the 2003–2012 dataset. This led to an NNI ranging from 0 to 3 in the 2013–2018 dataset and 0 to 4 in the 2003-2012 dataset.
To correct this mismatch in the combined dataset and test for any effect on our conclusions in the published article, we removed mosquitofish from the NNI across all years and reran the occupancy models in PRESENCE, then re-analyzed the correlation between beaver dam presence and NNI using a Spearman rank test for averaged annual covariate values across sites.

The authors apologize for this error and state that this does not change the scientific conclusions of the article in any way. The original article has been updated.

In the published article, there was a statement in the Supplementary File (Arroyo toad field Surveys and data analysis, paragraph 3) that required clarification about the use of correlated variables. The corrected statement appears below:

To reduce model complexity, we initially removed correlated variables by category (water, aquatic vegetation, etc.; Table S1) based on Pearson and Spearman rank tests (Bonferroni adjusted p < 0.05 and r/ > 0.25; Bonett and Wright 2000). If certain correlated variables were of particular interest (e.g., NNI, specific non-native taxa, etc.), we tested them in separate models.

In the published article, there was labeling error in Supplementary Table S1 involving the covariate “Hydroperiod (current year & previous year)”.” The clarification appears below:

Hydroperiod Ephemeral, perennial n/a Both

Error in Figure
Re-analysis of the arroyo toad occupancy data as described above required updates to Figure 2 as published. The corrected Figure 2 and its caption appear below.



FIGURE 2 | (A) Effect of the non-native aquatic species index [(NNI) i.e., number of non-native species observed at a site, either American bullfrog, red swamp crayfish, or large predatory fish (e.g., Micropterus sp.)] on the probability of occupancy, colonization, and local extinction of arroyo toad in coastal San Diego county from 2003–2018. (B) Results of the categorical regression showing the probability of the NNI score relative to beaver dam presence/absence (error bars represent the upper 95% credible intervals of the posterior probability distributions; see Supplemental File for further details).

Text Correction
Re-analysis of the arroyo toad occupancy data as described above required minor edits to the published article. A correction has been made to NATIVE FAUNA AND FLORA UNDER
THREAT, Arroyo Toad, paragraph 1. The text previously stated:

“…(Fig 2). The presence of a beaver dam was also a significant predictor of extirpation, with the probability of local extinction being 2.5 (95% CI 0.9–3.4) times higher when a dam was present. The probability of extinction grew to nearly 0.70 when a beaver dam and all three non-native species were present, nearly seven times higher than in reaches without any invasive species (Fig 2).”

The corrected sentence appears below:
“…(Fig 2A). Data averaged across years further showed that a non-native species index (NNI: the number of non-native species recorded at a site, either American bullfrog, red swamp crayfish, or large predatory fish [e.g., Micropterus sp.]) and beaver dam presence were significantly correlated (Spearman’s  = 0.62, P < 0.001). The probability of observing higher numbers of non-native species also increased in the presence of beaver dams (Fig 2B), with the estimated NNI being nearly three times greater (mean = 2.9; 95% credible interval = 2.1–3.9) at sites where beaver dams were present versus sites where they were absent (see the Supplemental File for further details). Thus, not only are dams detrimental to arroyo toad breeding and habitat occupancy dynamics, their presence is also associated with increased richness of invasive predators.”

In the published article, the reference for [Anderson et al. 2009] was incorrectly written as [Wright et al. 2006].

A correction has been made to Effects of novel ecological disturbance on native fauna and flora, paragraph no.6. This sentence previously stated:
“…beavers are present within the system (Wright et al. 2009)”
The corrected sentence appears below:
“…beavers are present within the system (Anderson et al. 2009)”

The authors apologize for these errors and state that this does not change the scientific conclusions of the article in any way. The original article has been updated.