how to compare species diversity

Species diversity is a more complex measure of how many different types of taxa are present in communities. These differences may be a reflection of resource availability or habitat conditions, but may also be simply due to sampling effort (Gotelli and Colwell 2001), as demonstrated in comparisons of logged vs. unlogged forests (Cannon et al. If α‐diversity is dissimilar, use the Raup‐Crick measure (1979) However, we refer to these data with the name i‐Tree Eco hereafter. Try to compare pairwise. Numerous studies have also sought to assess the performance of richness estimators across differing sites and/or sampling schemes (for a review, see Walther and Morand 1998), with the goal of finding methods to defensibly compare across sites. However, sampling scheme is an important consideration; for example, convenience sampling was found to result in higher estimates of species diversity and more rare species, when compared to random sampling in an urban forest (Speak et al. Alpha (within sample) diversity. Estimates of species richness were slightly higher when un‐treed plots were included (Table 2). Finally, we outline an approach for selecting the most appropriate methods for analyses and discuss practical considerations when considering data sampled under different methodologies. Richness is the number of species in each sample. As recommended by Hortal et al. However, traditional multivariate analysis methods, such as MANOVA, make stringent assumptions which are untenable for most ecological datasets (McArdle and Anderson 2001). 2003). Taking a more nuanced view, this interpretation could, however, demonstrate the influence of smaller trees or shrubs; when considering comparisons using basal area as a proxy for tree size, differences in the interactive effect of forest type and province were weaker. While no method is completely robust to these issues, alternative methods, such as rarefaction, can mitigate problems associated with violations in assumptions. When studies involve comparisons of species richness among different sites, there are many methodologies available that standardize richness data using extrapolation or rarefaction techniques (Chao and Chiu 2016). Diversity is often quantified in terms of richness and evenness, as well as community composition. 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. This gives evidence against the hypothesis of ecological homogenization in urban ecosystems, at least in terms of tree diversity (Blood et al. What you want to use very much depends on your interest. 2012, Jenerette et al. To account for the different plot sizes associated with FIA and i‐Tree data, we re‐scaled the axis of accumulation, multiplying by the plot size, before plotting derived species accumulation curves. Still, questions remain about how forest dynamics in rural contexts compare to those of urban environments (Blood et al. The Bray Curtis analyses showed that most measures of vegetation structure and species diversity have recovered >50% compared with the reference site . Second, an area of interest to researchers that we did not review is the vast number of indices used to characterize species evenness and beta diversity. Independent evolution of leaf and root traits within and among temperate grassland plant communities. 1998, Denslow 1995, Imai et al. 0000008263 00000 n Forest composition and tree species diversity have been recognized as primary drivers of ecosystem resilience and function (Jenerette et al. Plot‐based data, where plots are considered as the independent unit of observation, require mixed modeling methods to account for the potential for correlation among trees measured within the same plot (García 2006). 16 0 obj << /Linearized 1 /O 18 /H [ 1102 265 ] /L 19286 /E 10135 /N 2 /T 18848 >> endobj xref 16 35 0000000016 00000 n The Shannon diversity index is a commonly used measure of diversity. 2015, Speak et al. We first developed matrices of the abundances of each species by plot, although these matrices can indicate simple presence/absence (0/1) with equivalent results for most of the analyses described. To make comparisons in this case, the number of species, tree or otherwise, accumulated should be plotted as a function of accumulated individuals instead of samples (Gotelli and Colwell 2001). Thus, non‐parametric methods are preferable. More recently, permutational analysis of variance (PERMANOVA; Anderson 2001) has been used to compare urban forests composition across geographic and urbanization gradients (Blood et al. 2015, McPherson et al. 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. 0000005125 00000 n While specifically studying the effect of grain size (sensu Whittaker et al. Ecological province was defined by USDA Forest Service ecozones (Bailey. But, for most of the sites sampled, urban forests and their peri‐urban counterparts are fairly close, indicating low values of dissimilarity, and therefore indicating a lack of urban homogenization. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. For example, since urban forests and PF often contain different species (Blood et al. 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. To assess this difference, we first compare forested peri‐urban and forested urban, thus excluding plots in urban areas where no trees were recorded. For example, Pearse et al. PERMANOVA, ANOSIM, and the Mantel test in the face of heterogeneous dispersions: What null hypothesis are you testing? While the Roanoke and Abingdon UFs were similar to all UFs and PFs in Virginia, the Falls Church UF was only dissimilar to the Virginia PFs. 2013). No two individuals belonging to the same species are exactly similar. Points in the resulting plots appear close together when the Raup‐Crick dissimilarity metrics indicate their community compositions are similar (Avolio et al. 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. The Charlottesville UF was only dissimilar to the Abingdon PF. Navigating the multiple meanings of β diversity: a roadmap for the practicing ecologist, A framework for quantifying the magnitude and variability of community responses to global change drivers, Biodiverse cities: The nursery industry, homeowners, and neighborhood differences drive urban species composition, Description of the ecoregions of the United States, Differences in the impacts of formal and informal recreational trails on urban forest loss and tree structure, Measuring β‐diversity with species abundance data, How do urban forests compare? Also we do not attempt to address the controversies that exist regarding the appropriate and inappropriate use of diversity indices. 2016). The effective number of species refers to the number of equally abundant species needed to obtain the same mean proportional species abundance as that observed in the dataset of interest (where all species may not be equally abundant). 0000005146 00000 n If α‐diversity is similar, use Jaccard's or Sørensen's index (Koleff et al. 0000002642 00000 n 2016, Kendal et al. While this was a rare (n = 3) occurrence in our case study data, this could be a potential source of error if new exotic species invade peri‐urban locations. Diversity is classically divided into alpha (site level), beta (turnover across multiple sites), and gamma (composite of all sites in a region) components. Richness, however, can be difficult to measure appropriately since more species are recorded as the number and area of samples increases (May 1988). 2018), and a variety of qualitative methods, as well as specifically designed protocols which have included data on herbaceous vegetation and soils (Groffman et al. Investigators define their levels of diversity in different ways. Also, further information is needed on how the increasing use of available plot‐level data in both rural and urban forests can be used to address questions regarding ecological disturbance, functionality, and homogenization (Staudhammer et al. Species diversity is a term used to define the different number of species in an area (Species richness) and its abundance and the distribution of these species in that ecosystem. 2018). 2014, Livesley et al. 2016), this introduces another layer of complexity, as the numbers of species increase as sampling effort increases. 2018). USDA FIA plots located within these areas were identified and extracted. 0000008450 00000 n Another calculation for the rarefaction diversity measurement for different spatial distributions, Measuring beta diversity for presence‐absence data, Influence of plot shape on estimates of tree diversity and community composition in Central Amazonia, The biodiversity of urban and peri‐urban forests and the diverse ecosystem services they provide as socio‐ecological systems, The detection of disease clustering and a generalized regression approach, Fitting multivariate models to community data: a comment on distance‐based redundancy analysis, Urbanization as a major cause of biotic homogenization, The structure function and value of urban forests in California communities, Effects of urbanization on tree species functional diversity in eastern North America, Ground‐based method of assessing urban forest structure and ecosystem services, Toward a mechanistic understanding of prediction of biotic homogenization, Homogenization of plant diversity, composition, and structure in North American urban yards, Experimental design and data analysis for biologists, R: a language and environment for statistical computing, Measurement of faunal similarity in paleontology, Street tree diversity in eastern North America and its potential for tree loss to exotic borers, Trees for urban planting: diversity, uniformity, and common sense, Proceedings, 7th Conference Metropolitan Tree Improvement Alliance (METRIA), Sampling methods for Multiresource Forest Inventory, Plant species diversity in alien black locust forests: a paired comparison with native stands across a north‐Mediterranean range expansion, Nonparametric estimation of species richness, Comparing convenience and probability sampling for urban ecology applications, Predictors, spatial distribution, and occurrence of woody invasive plants in subtropical urban ecosystems, Comparative performance of species richness estimation methods, Spatiotemporal scaling of species richness: patterns, processes, and implications, Scale and species richness: towards a general, hierarchical theory of species richness, FIA database description and users' manual for Phase 2, Exploring land‐use legacy effects on taxonomic and functional diversity of woody plants in a rapidly urbanizing landscape. For example, tree richness can be measured using a management indicator such as the 10‐20‐30 rule (Santamour 1990), which calls for no more than 10% of trees in any single species, 20% in any single genus, and 30% in any single family within a management unit such as a city. Species Richness =Variety of species or the number of different species (or genera, families, etc.). 2016). To our knowledge, there is no i‐Tree Eco check to assure standardization, and there is no systematization nor standardization in species code protocols. The bootstrap estimator (Smith and van Belle 1984) uses repeated resampling (with replacement) from the data, estimating the number of species missed. An ecosystem with a high level of biodiversity is more resistant to the environmental change and such ecosystems are rich in a variety of living organisms. 0000001942 00000 n Assuming that the number of observations and distributional shapes are not drastically different, these quantitative samples may be compared via simple t‐tests, or if data sources are sufficiently different, non‐parametric methods such as the Kolmogorov‐Smirnov or Cramer‐von Mises tests may be more appropriate (Quinn and Keough 2002). 2017). However, we see only very weak evidence in support of the hypothesis (Table 3), at least in terms of tree diversity using both tree counts and tree basal area. Our study, which further incorporated disparate data sources, demonstrates that the Raup‐Crick dissimilarity indices, based on presence/absence data, are robust to sampling differences. 0000001574 00000 n If species exhibit non‐random spatial patterns, such as within‐species clumping and segregation among species, which may occur with planted urban forests, estimates of species pools can be overestimated in small samples, and knowledge about spatial autocorrelation has not been found useful in correcting bias (Collins and Simberloff 2009). Eight urban forest locations sampled in the southeastern United States. 2017, Speak et al. We, however, do note some limitations. species abundance and the more likely it is to only use species richness or higher taxon diversity. 2015), FIA plots are exclusively installed on forested land as defined by the USDA, and explicitly must include trees. 0000001102 00000 n I described this data set in more detail in a recent paper:S.W. This includes non‐parametric estimators, parametric species abundance models, species accumulation curves, and species–area curves. Yet as shown above, the assumptions and statistical methods used with these data can influence results and can have implications for the certainty with which results are communicated regarding urban‐rural ecosystem diversity and homogeneity. Gotelli and Colwell (2001) recommended that raw richness only be compared if species accumulation curves clearly indicate that an asymptote has been reached in both populations of interest, highlighting the need for estimating total population pools. However, we assumed location error to have a minimal impact on analyses. Figure 1 – Sample Index of Diversity. When sampling schemes use identical selection criteria, area‐based variables are unaffected by plot size differences in theory; however, their variances decrease with increases in plot size (Zeide 1980), leading to different levels of uncertainty for each plot size. 2015). 0000006390 00000 n COMPARING SPECIES DIVERSITY AND EVENNESS INDICES By C. HElP AND P. ENGELS Department of Zoology, State University of Ghent, Belgium In a low diversity brackish water habitat the diversity of the copepod community is best measured with the Shannon-Wiener information function and its … 2017). When studies utilize disparate data sources where sampling effort is unequal (Kendal et al. 2017). This type of data is key in estimating and modeling the supply of ecosystem services from urban and peri‐urban forests (PF; Nowak et al. Components of species diversity: species richness and relative abundance. Similarly, data can be used to study the adequacy of human‐dominated landscapes in providing adequate habitat for native tree species and fauna (Livesley et al. PLoS ONE 6(6): e19992. However, trees that split <0.3 m from ground level are assigned the distance and direction corresponding to the approximate location of tree pith, and thus, these locations will not be identical for these multi‐stemmed individuals. Nonetheless, Anderson and Walsh's (2013) simulation study showed that PERMANOVA was much less sensitive to heterogeneity in dispersions than ANOSIM and the Mantel test for balanced designs. The Raup‐Crick is such a measure, allowing for comparisons between communities with varied numbers of species and sampling sizes (Chase et al. Strategic National Urban Forest Inventory for the United States, https://www.itreetools.org/eco/resources/UFORE_Species_List_Apr30_2012.xls, 12.7 cm in 0.0675 ha subplots; 2.54 cm in 4 × 0.00135 ha microplots, Recorded as a single tree, up to 6 largest dbh measured, If species have equal multivariate spread among groups, use Analysis of Similarities (Clarke. However, Winchester and Falls Church, Virginia, indicate different species composition patterns from those of their regional peri‐urban counterparts. With the i‐Tree Eco protocol, multiple stems that originate (or appear to originate) from the same root stock and trees that split below breast height are recorded as single individual trees with multiple diameters. When considering the pool of genera, similar patterns were observed, but with estimates approximately 10% closer to observed estimates (Table 2). For example, the Mantel test (Mantel 1967) has been used to compare tree β‐diversity across a tropical forest (Chust et al. The mean for the SDI value for each site will be calculated from the different samples at each site. The interpretation of Raup‐Crick depends on the potential species pool, and thus, analyses need to consider the implication of the inclusion of species in terms of their impact on hypotheses tested (Chase et al. 2012) or wind (Holzmueller et al. But, differences in the sampling methods underlying these disparate protocols and data sources is a non‐trivial concern in formulating comparative analyses. 2014). Accordingly, hypotheses may be formulated to test for differences among groups of sites (e.g., urban vs. peri‐urban). 2011). 2016). For most methods of species richness comparison, it is assumed that individuals have a random spatial distribution in the environment (Kobayashi 1983), sample sizes are sufficiently large, and populations are sampled in the same manner (Abele and Walters 1979). (2018), who found that native and urban tree population realized climatic niches that had substantial overlap; in most cities we analyzed, we found similar species lists, though urban areas almost always contained more species. Previous urban ecology studies that test the hypothesis of ecological homogenization have used Jaccard's index (McKinney 2006), Sørensen's index (Pearse et al. NMDS1 and NMDS2 refer to the first and second axes, respectively. These results are further corroborated with the corresponding NDMS plot (Fig. 0000006935 00000 n However, species richness increases with sample size. 2012, Avolio et al. Unless one can assume that unobserved species occur in very low numbers—which may be reasonable in some locales—the evaluation of this rule is problematic, as numbers of species as well as individuals will have additional uncertainty. The choice of source data for dissimilarity metrics is an important initial question. 2015, Jenerette et al. Conversely, when using the bootstrap and jackknife estimators, we found that there were significant differences in all locations except Abingdon. If you do not receive an email within 10 minutes, your email address may not be registered, Enter your email address below and we will send you your username, If the address matches an existing account you will receive an email with instructions to retrieve your username. These two conditions would manifest in a significant interaction between forest type and province. For example, Pearse et al. Species diversity is defined as the number of species and abundance of each species that live in a particular location. 0000008067 00000 n Recently, disparate data sources have been used for regional‐scale and even continental‐scale urban‐to‐rural analyses of woody plant community composition, similarity, species richness, and other questions such as ecological homogenization and ecosystem dynamics (Blood et al. 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. 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). If sample plots have different shapes and sizes, sampling bias may be introduced such that particular species are over‐ or under‐sampled (Boulinier et al. 0000008825 00000 n 2008). 1975) utilizing species density should be used to make comparisons, If samples are homogenous among sites, many estimators are unbiased (see Hortal et al. The Bulletin of the Ecological Society of America, I have read and accept the Wiley Online Library Terms and Conditions of Use, The stability‐time hypothesis: reevaluation of the data, A new method for non‐parametric multivariate analysis of variance. Some treat α diversity as one sample whereas others treat α diversity as a 100m x 100m plot.
how to compare species diversity 2021