2015). A community dominated by one or two species is considered to be less diverse than one in which several different species have similar abundance. However, sampling scheme is an important consideration; for exam-ple, convenience sampling was found to result in higher estimates of species diversity and more rare species, when compared … These studies explicitly recognize that species accumulation models indicate that not all species are seen in any sampled site, and hence use species pool functions to better estimate the number of unseen species (Colwell and Coddington 1994). This includes non‐parametric estimators, parametric species abundance models, species accumulation curves, and species–area curves. Moreover, differences tended to be larger when the total number of plots and total number of species detected were lower (e.g., Abingdon, Virginia, USA). 2013). Washington, DC 20036phone 202-833-8773email: esajournals@esa.org. This study assesses methods for analyzing species richness and composition in light of disparate forest data sources, and how their use can affect findings and inference in comparing ecological diversity across different local, regional, and continental areas of interest. 0000006935 00000 n Nonmetric multidimensional scaling plot by location and forest type, utilizing the Raup‐Crick dissimilarity distances. That is, with larger trees having more weight in the analyses, the PF vs. UF tree diversity differences were more consistent by ecological province, as evidenced by the larger P‐value of the interaction. It is basically the variety of species expressed at the genetic level by each individual in a species. 2014). However, species richness increases with sample size. The choice of source data for dissimilarity metrics is an important initial question. I plan on using the Simpson's diversity index (SDI), which combines species richness (number of different species) with the number of each individual to form a number between 0 and 1. 2015). Spe… Having outlined the available methods, we now present a case study as an example of how commonly utilized methods can be applied to disparate data sources addressing urban–rural ecology questions across different scales. Species diversity depends as much on the genetic diversity as on the environmental condition. 2008, Speak et al. We utilize species pool estimators and compare the accuracy and precision of such estimators with and without tree‐less plots. That said, each FIA tree has distance and direction from plot center recorded, and stems from trees that split between 0.3 and 1.37 m are assigned identical distance and direction. Species diversity is a combination of species richness and species abundance. The good climate with good physical geography supports a better species diversity. Species accumulation curves by location for urban (light green) and FIA (dark green) data: (A) excluding plots with zero tree counts (in urban areas only) and (B) including all plots, so that plots with no trees (which occur only in urban areas) are included as zeroes in all species. Thus, a larger measure of uncertainty is obtained when including plots without trees in calculating the bootstrap estimator. Species density or the number of species per m 2 is most commonly used to measure species richness. species diversity using data from disparate sources (Blood et al. What you want to use very much depends on your interest. In PERMANOVA, it is assumed that the observations are exchangeable under the null hypothesis, which implies that the observations are independent and have “similar” distributions. Hunter (2002: 448) defines gamma diversity as "geographic-scale species diversity". Species diversity. However, as the number of samples in the forest sampling protocol increases, the number of genera will always reach an asymptote sooner than that of the number of species, except when samples contain 100% monobasic taxa (Gotelli and Colwell 2001). The question of how many different species exist in a particular environment is central to the understanding of why it is important to promote and preserve species diversity. For instance, such tenants as the 10‐20‐30 rule (Santamour 1990) are subject to misinterpretation when based on data collected using different sampling intensities (Kendal et al. Richness is the number of species in each sample. Still, questions remain about how forest dynamics in rural contexts compare to those of urban environments (Blood et al. An unconditional standard deviation is computed based on the extrapolated number of species in the data (the sample γ‐diversity). 2016), the assumption of multivariate normality cannot be met. The FIA plots consist of groups of four subplots that cover an area of 0.0675 ha (0.167 ac), with a microplot ~0.00135 ha (0.003 ac) in area located within each subplot (total area = 0.0054 ha). Our results suggest that UFs and PFs were very similar within a particular region in the United States. Plot‐based sampling requires different methods from those of individual‐based protocols (sensu Gotelli and Colwell 2001), as plots involve samples of multiple, grouped individuals as replicates, rather than single individuals (Speak et al. 1975). (2018) collected data from 21 to 30 urban household yards and 3 to 6 natural area sites in their study of ecological homogenization across seven metropolitan areas. 2016, Kendal et al. In the United States, this urban forest inventory and monitoring approach is known as the i‐Tree Eco protocol within the i‐Tree software suite, but is now being incorporated as part of the Urban FIA program (https://www.nrs.fs.fed.us/fia/urban/). One advantage of PERMANOVA is that the method is unaffected by correlation among variables (Anderson 2001, Anderson and Walsh 2013), which may occur when species have a tendency to co‐occur. Given the ecological challenges presented in the Anthropocene, robust methods and available datasets are key in understanding the functionality, nativity, and diversity of urban and peri‐urban woody vegetation across all biomes of the world. While the inclusion of plots with no trees would not alter the conclusions associated with comparisons of species richness among communities and/or forest types, the type of estimator used is critical. The matrix of the Raup‐Crick dissimilarity indices showed mostly high values for Winchester's urban forest, indicating that it was very dissimilar to its, and all other regional, PF (Table 4). Given current knowledge of the available methods for situations when there are differing sampling intensities and non‐homogeneous species distributions, and the results of our case study, we recommend specific analytical methods for quantifying diversity using disparate data sources in urban–rural forested contexts and across different scales (Table 5). Previous urban ecology studies that test the hypothesis of ecological homogenization have used Jaccard's index (McKinney 2006), Sørensen's index (Pearse et al. Conversely, when using the bootstrap and jackknife estimators, we found that there were significant differences in all locations except Abingdon. 2016). When considering the pool of genera, similar patterns were observed, but with estimates approximately 10% closer to observed estimates (Table 2). Under the hypothesis of ecological homogenization, we would expect that urban forest locations would be closer to each other than to those of PF in Fig. It takes into account both species richness as well as the dominance/evenness of the species. In contrast, under the FIA protocol, these trees are not single individuals and are recorded as two (or more) trees. 0000008825 00000 n 2018). Richness, however, can be difficult to measure appropriately since more species are recorded as the number and area of samples increases (May 1988). 's (2008) protocol, where each tree or palm with dbh >2.54 cm was measured and its species name recorded within a 0.0404‐ha (0.1 acre) circular plot. While the ecological homogenization hypotheses can be tested under conditions of unequal sampling effort, we suggest robust methods such as PERMANOVA and the Raup‐Crick dissimilarity index. 2017, Avolio et al. Although the conclusion that more sampling is necessary in order to encounter all species present in an area is reached in most urban locations, the implied sample area is more realistic when including these un‐treed plots. H�tS�n�0��+�HMR%ȡ��zI�^�d�vX�T`� ���SnS�+ivvvV�2V�q��`�U�%�ꏔq֓N@�r"9�`. This gives evidence against the hypothesis of ecological homogenization in urban ecosystems, at least in terms of tree diversity (Blood et al. 0000008067 00000 n 2011). Species diversity refers to the diversity at the species level. 0000004529 00000 n Forest land in FIA is defined as having an area of at least 0.4 ha with at least 10% canopy cover of live tree species of any size, either at the time of sampling or in the past, where the land is not subject to non‐forest use which would prevent normal tree regeneration and succession (e.g., regular mowing, or intensive grazing; Woudenberg et al. Such an approach using data from different sampling methods—but within the same general geographic study areas—allows for the evaluation of quantitative methods while isolating variability associated with geography and climate. Our PERMANOVA results utilizing basal area and tree counts were very similar, with both analyses indicating that species distributions were different depending on ecological province (P = 0.001) and forest type (UF vs. PF; P = 0.001; Table 3). (1998) sought a richness estimator that was insensitive to the size and order of the sample, and the unevenness in species distribution. Diversity is a weighted average of the proportions of each species present. 0000006390 00000 n Nonetheless, when considered as a proportion of the total estimated species, these differences are very small (Table 2). Species density = number of species per unit are All rights reserved. We used the function specaccum, which uses as its default method the sample‐based (i.e., plot‐based) exact method to estimate an expected species accumulation curve via sample‐based rarefaction (Chiarucci et al. Since trees in the 2.54–12.7 dbh range are only sampled in ~0.0054 of the total 0.0675‐ha plot, one can avoid methodological complications by limiting studies to trees >12.7 cm dbh. Thus, evaluating the effectiveness of this rule requires certainty when estimating the proportion of individuals in each species, each genus, and each family. Tree data collected included condition, species, dbh, height, and location within plot (for more information on FIA data collection, see Woudenberg et al. $\alpha$ Diversity $\beta$ Diversity $\gamma$ Diversity; Isolation Diversity; Relative Species Abundance; The literature is important on the question of measuring species diversity, ecosystemic function diversity and genetic diversity. Alpha (within sample) diversity. 2016), simulation studies recommend the use of PERMANOVA to account for heterogeneity that may be exacerbated by differences among sampling intensities (Anderson and Walsh 2013). ANOVA‐like test statistics are constructed from matrices of among‐sample resemblances, which may be distances, dissimilarities, or similarities, and P‐values are obtained with randomly generated permutations of observations among groups (Anderson and Walsh 2013). The Bray Curtis analyses showed that most measures of vegetation structure and species diversity have recovered >50% compared with the reference site . When plots are of different sizes, these methods are still valid, but weighting schemes may be appropriate in certain situations (Schreuder et al. (See Smith and Wilson 1996 and Anderson et al. UF, urban forests; PF, peri‐urban forests; WIN, Winchester, Virginia; CHA, Charlottesville, Virginia; ROA, Roanoke, Virginia; ABI, Abingdon, Virginia; FC, Falls Church, Virginia; ATL, Atlanta, Georgia; GNV, Gainesville, Florida; EORL, East Orlando, Florida. Some treat α diversity as one sample whereas others treat α diversity as a 100m x 100m plot. 2016). This has often been referred to as an assumption of equal “multivariate spread” among groups, which is a multivariate analog to the assumption of homoscedasticity in univariate ANOVA. 0000004508 00000 n We demonstrate that this method can indeed be used with species counts or basal area depending on the research hypothesis. Please check your email for instructions on resetting your password. Corresponding Editor: Debra P. C. Peters. We do so by analyzing local‐level urban and peri‐urban forest diversity and composition across a regional gradient using data from two available yet disparate databases. Species richness and composition are often‐reported metrics across ecological disciplines, but have particular importance in urban forests in understanding biodiversity (Livesley et al. A framework for selecting appropriate methods is also discussed. The specific research questions being posed in urban ecology studies should drive the selection of methods and data collected. Using a matrix of comparisons between all pairs of associations, the Raup‐Crick index compares observed numbers of species with the distribution of co‐occurrences generated from 999 Monte Carlo random replicates (Chase et al. Accordingly, rarefaction curves have allowed researchers to compare samples of different sizes through the calculation of expected richness at a standardized size (Heck et al. The function treats the data as binary (presence/absence) regardless of how the matrix is formulated. However, the axis of accumulation is stretched when considering the latter, and thus, comparisons of UF vs. PF are somewhat different. 2015). Following Blood et al. Measuring diversity: the importance of species similarity Tom Leinster 1,2∗ Christina A. Cobbold 1School of Mathematics and Statistics, University of Glasgow, UK 2Boyd Orr Centre for Population and Ecosystem Health, University of Glasgow, UK Abstract Realistic measures of biodiversity should reflect not only the relative abundances of 0000008804 00000 n Investigators define their levels of diversity in different ways. No two individuals belonging to the same species are exactly similar. 2018). However, we assumed location error to have a minimal impact on analyses. For example, unequal sampling intensity of smaller trees in the FIA protocol requires development of a differential measure of uncertainty in richness and composition estimates. Analysis of similarities (ANOSIM; Clarke 1993) has been used to compare urban forest composition in response to hurricane (Burley et al. Evenness is also affected by sampling interval and intensity, but to varying degrees depending on the measure used (White 2007). Finally, using these methods and the urban and peri‐urban data, we analyzed how different assumptions and selected methods can influence results in answering common research questions in urban ecology. So, to compare the magnitudes of two diversities, calculate the effective numbers of species (the exponential of the Shannon entropy, for example) of the two communities so that you can compare them on a linear scale and get an intuitive feel for the difference. Since FIA plot locations are not reported as exact spatial coordinates to comply with privacy issues, extracted locations were between 0.8 and 1.6 km of the actual plot.