Never underestimate biodiversity: how undersampling affects Bray–Curtis similarity estimates and a possible countermeasure

Figures & data

Table I. Original abundance (species richness), percentage abundance difference for the tested dataset, number of shared species, and Bray–Curtis similarity values.

Paper dataset	Assemblage 1 abundance (No. of species)	Assemblage 2 abundance (No. of species)	Δ% between A2 and A1	No. of shared species	Bray–Curtis similarity
Andresen (2003)	1851 (49)	4113 (45)	55.0%	40	0.53
Bonham et al. (2002)	2050 (85)	2279 (72)	10.0%	64	0.73
Boulton et al. (2005)	2456 (17)	2693 (16)	8.8%	13	0.80
Collier et al. (1997)	772 (35)	2448 (25)	68.5%	19	0.14
Gristina et al. (2006)	5000 (38)	50,207 (28)	90.0%	23	0.12
Hardersen et al. (2014)	1428 (61)	2947 (73)	51.5%	45	0.23
Heck et al. (1995)	4265 (29)	9873 (31)	56.8%	24	0.38
Horváth et al. (2013)	2233 (192)	3270 (203)	31.7%	138	0.54
Jokimäki and Kaisanlahti-Jokimäki (2003)	504 (14)	705 (13)	28.5%	10	0.56
Maes et al. (2016)	612,934 (63)	761,660 (70)	19.5%	55	0.85
Mueter and Norcross (2000)	2655 (23)	4790 (22)	44.6%	21	0.35
Muturi et al. (2006)	6106 (22)	7699 (20)	20.7%	17	0.73
Niemelä et al. (2002)	2210 (36)	3125 (45)	29.3%	28	0.60
Shah et al. (2003)	6322 (39)	15,790 (37)	60.0%	31	0.51
Stark et al. (2003)	3863 (39)	4539 (23)	14.9%	13	0.05
Thorn et al. (2014)	9528 (150)	24,268 (145)	60.7%	116	0.50

Figure 1. Similarity boxplot of Muturi et al. (2006) dataset for random sub-samples of 10, 30, 100, 300, 900, 3000, 6000 individuals. The red dashed line indicates the similarity value calculated when considering the complete dataset. The plots for the other datasets are presented in the Online Resource 2.

Figure 2. Barplot of mean variance ± SE of similarity for random sub-samples of 10, 30, 100, 300, 900, 3000, 6000 individuals obtained adding zero, one, three or five dummy species.

Table II. P-values of Levene’s test obtained by comparing variances estimated adding zero, one, three or five dummy species at different numbers of individuals drawn from the original dataset.

	Resampled individuals
Paper dataset	10	30	100	300	900	3000	6000
Andresen (2003)	<0.0001	0.054	0.465	0.394	0.667	–	–
Bonham et al. (2002)	<0.0001	<0.0001	0.280	0.935	0.538	–	–
Boulton et al. (2005)	<0.0001	0.001	0.020	0.427	0.560	–	–
Collier et al. (1997)	<0.0001	0.522	0.759	0.221	–	–	–
Gristina et al. (2006)	0.004	0.210	0.416	0.704	0.850	0.595	–
Hardersen et al. (2014)	<0.0001	0.001	0.012	0.127	0.173	–	–
Heck et al. (1995)	<0.0001	0.091	0.301	0.304	0.646	0.976	–
Horváth et al. (2013)	<0.0001	0.010	0.021	0.777	0.487	–	–
Jokimäki and Kaisanlahti-Jokimäki (2003)	<0.0001	0.017	0.798	0.289	–	–	–
Maes et al. (2016)	<0.0001	<0.0001	0.467	0.543	0.212	0.956	0.914
Mueter and Norcross (2000)	<0.0001	0.140	0.328	0.991	0.033	–	–
Muturi et al. (2006)	<0.0001	<0.0001	0.110	0.662	0.275	0.210	0.837
Niemelä et al. (2002)	<0.0001	0.001	0.413	0.327	0.813	–	–
Shah et al. (2003)	<0.0001	<0.0001	0.771	0.377	0.220	0.379	0.667
Stark et al. (2003)	<0.0001	0.002	0.398	0.775	0.538	0.911	–
Thorn et al. (2014)	<0.0001	0.003	0.151	0.060	0.644	0.722	0.477

Table III. Mean values (±sd) of Good-Turing sample completeness at the increasing number of individuals resampled from the species of the original dataset.

	Resampled individuals
Assemblage column in each dataset (cf. Table I)	10	30	100	300	900	3000	6000
Assemblage 1	0.78 ± 0.17	0.88 ± 0.10	0.95 ± 0.05	0.98 ± 0.03	0.99 ± 0.01	1.00 ± 0.00	1.00 ± 0.00
Assemblage 2	0.77 ± 0.18	0.89 ± 0.10	0.95 ± 0.06	0.98 ± 0.03	0.99 ± 0.01	1.00 ± 0.00	1.00 ± 0.00

(-) Values could not be calculated.

Figure 3. Accuracy plots, which indicate the proportion of similarity values, calculated for increasing numbers of individuals drawn from the 16 datasets. Black point-ranges represent mean values (±sd) that fall within the range of ±10% around the similarity value calculated with the complete dataset; red dashed lines represent the mean rate of underestimation and blue dashed lines mean rate of overestimation.

Figure 4. Mean errors on 16 datasets with an increasing number of (i) resampled individuals (left-right) and (ii) dummy species (top-down). All data are means of 1000 random draws and unbiased estimates have a mean error of zero (indicated by a red dotted line). Correlation coefficients r are reported in labels.

Figure 5. Boxplot of mean errors with an increasing number of dummy species and regression lines. All data are means of 1000 random draws. Correlation coefficients and P values are reported in labels.

Andresen E. 2003. Effect of forest fragmentation on dung beetle communities and functional consequences for plant regeneration. Ecography 26:87–97. DOI: 10.1034/j.1600-0587.2003.03362.x.

 Web of Science ®Google Scholar

Bonham KJ, Mesibov R, Bashford R. 2002. Diversity and abundance of some ground-dwelling invertebrates in plantation vs. native forests in Tasmania, Australia. Forest Ecology and Management 158:237–247. DOI: 10.1016/S0378-1127(00)00717-9.

 Web of Science ®Google Scholar

Boulton AM, Davies KF, Ward PS. 2005. Species richness, abundance, and composition of ground-dwelling ants in Northern California grasslands: Role of plants, soil, and grazing. Environmental Entomology 34:96–104. DOI: 10.1603/0046-225X-34.1.96.

 Web of Science ®Google Scholar

Collier KJ, Smith BJ, Baillie BR. 1997. Summer light‐trap catches of adult Trichoptera in hill‐country catchments of contrasting land use, Waikato, New Zealand. New Zealand Journal of Marine and Freshwater Research 31:623–634. DOI: 10.1080/00288330.1997.9516794.

 Web of Science ®Google Scholar

Gristina M, Bahri T, Fiorentino F, Garofalo G. 2006. Comparison of demersal fish assemblages in three areas of the Strait of Sicily under different trawling pressure. Fisheries Research 81:60–71. DOI: 10.1016/j.fishres.2006.05.010.

 Web of Science ®Google Scholar

Hardersen S, Curletti G, Leseigneur L, Platia G, Liberti G, Leo P, Cornacchia P, Gatti E. 2014. Spatio-temporal analysis of beetles from the canopy and ground layer in an Italian lowland forest. Bulletin of Insectology 67:87–97.

 Web of Science ®Google Scholar

Heck KL, Able KW, Roman CT, Fahay MP. 1995. Composition, abundance, biomass, and production of macrofauna in a New England estuary: Comparisons among eelgrass meadows and other nursery habitats. Estuaries 18:379. DOI: 10.2307/1352320.

 Google Scholar

Horváth B, Tóth V, Kovács G. 2013. The effect of herb layer on nocturnal Macrolepidoptera (Lepidoptera: Macroheterocera) communities. Acta Silvatica et Lignaria Hungarica 9:43–56. DOI: 10.2478/aslh-2013-0004.

 Google Scholar

Jokimäki J, Kaisanlahti-Jokimäki ML. 2003. Spatial similarity of urban bird communities: A multiscale approach. Journal of Biogeography 30:1183–1193. DOI: 10.1046/j.1365-2699.2003.00896.x.

 Web of Science ®Google Scholar

Maes D, Vanreusel W, Herremans M, Vantieghem P, Brosens D, Gielen K, Beck O, Van Dyck H, Desmet P, Natuurpunt V. 2016. A database on the distribution of butterflies (Lepidoptera) in northern Belgium (Flanders and the Brussels Capital Region). ZooKeys 585:143–156. DOI: 10.3897/zookeys.585.8019.

 Google Scholar

Mueter FJ, Norcross BL. 2000. Species composition and abundance of juvenile groundfishes around Steller Sea Lion Eumetopias jubatus rookeries in the Gulf of Alaska. Alaska Fishery Research Bulletin 7:33–43.

 Google Scholar

Muturi EJ, Shililu J, Jacob B, Gu W, Githure J, Novak R. 2006. Mosquito species diversity and abundance in relation to land use in a riceland agroecosystem in Mwea, Kenya. Journal of Vector Ecology 31:129–137. DOI: 10.3376/1081-1710(2006)31[129:MSDAAI]2.0.CO;2.

 PubMed Web of Science ®Google Scholar

Niemelä J, Kotze DJ, Venn S, Penev L, Stoyanov I, Spence J, Hartley D, De Oca EM. 2002. Carabid beetle assemblages (Coleoptera, Carabidae) across urban-rural gradients: An international comparison. Landscape Ecology 17:387–401. DOI: 10.1023/A:1021270121630.

 Web of Science ®Google Scholar

Shah PA, Brooks DR, Ashby JE, Perry JN, Woiwod IP. 2003. Diversity and abundance of the coleopteran fauna from organic and conventional management systems in southern England. Forest Agriculture Enterprises 5:51–60. DOI: 10.1046/j.1461-9563.2003.00162.x.

 Web of Science ®Google Scholar

Stark JS, Riddle MJ, Simpson RD. 2003. Human impacts in soft-sediment assemblages at Casey Station, East Antarctica: Spatial variation, taxonomic resolution and data transformation. Austral Ecology 28:287–304. DOI: 10.1046/j.1442-9993.2003.01289.x.

 Web of Science ®Google Scholar

Thorn S, Bässler C, Gottschalk T, Hothorn T, Bussler H, Raffa K, Müller J. 2014. New insights into the consequences of post-windthrow salvage logging revealed by functional structure of saproxylic beetles assemblages. PLoS ONE 9:7. DOI: 10.1371/journal.pone.0101757.

 Web of Science ®Google Scholar

Data availability statement

Source dataset are available in the GitHub repository “bioglp/bc2000.git” (https://github.com/bioglp/bc2000).