Accumulation of Local Biodiversity in Spacetime
Total Page:16
File Type:pdf, Size:1020Kb
Integration of local and regional species-area relationships from space-time species accumulation Jason D. Fridley1, Robert K. Peet2, Eddy van der Maarel3 & Jo H. Willems4 1Department of Biology, CB 3280, Coker Hall, University of North Carolina, Chapel Hill, NC 27599-3280, USA, [email protected] 2Department of Biology, CB 3280, Coker Hall, University of North Carolina, Chapel Hill, NC 27599-3280, USA, [email protected] 3Community and Conservation Ecology Group, University of Groningen, P.B. 14, NL - 9750 AA Haren, The Netherlands, [email protected] 4Department of Plant Ecology, Utrecht University, Sorbonnelaan 16, NL - 3508 TB, Utrecht, The Netherlands, [email protected] Revised submission to American Naturalist as an Article Keywords: species-area curve; species-time curve; Preston; species richness; scale-dependence; grassland Running head: Plant species richness in space and time No items for expanded online content 2 1 ABSTRACT 2 A long-standing observation in community ecology is that the scaling of species richness, as 3 exemplified by species-area curves, differs on local and regional scales. This decoupling of 4 scales may be due largely to sampling processes (the increasing constraint imposed by sampling 5 fewer individuals at fine scales) as distinct from ecological processes such as environmental 6 heterogeneity that operate across scales. Removal of the constraint of sampling from fine-scale 7 estimates of richness should yield species-area curves that behave like those of the regions in 8 which they are imbedded, but an effective method for doing so has not been available. We 9 suggest an approach that incorporates how small areas accumulate species over time as a way to 10 remove the signature of sampling processes from fine-scale species-area curves. We report for 11 three species-rich grasslands from two continents how local plant species richness is distributed 12 through time at multiple, nested spatial scales, and ask whether sampling-corrected curves reflect 13 the spatial scaling of richness of each larger floristic province. Our analysis suggests that fine- 14 scale values of richness are highly constrained by sampling processes, but once these constraints 15 are removed the spatial scaling of species richness can be seen to be consistent from the scale of 16 individuals to that of an entire province. 3 1 INTRODUCTION 2 All biological communities are spatially and temporally variable. Ecologists have long 3 been fascinated with species-area curves and have documented how species richness 4 accumulates with area, from the scale of individuals to the globe (Rosenzweig 1995, Hubbell 5 2001). Communities are also temporally dynamic, in that they gain and lose species over time 6 (MacArthur and Wilson 1967, van der Maarel and Sykes 1993). Although rarely recognized by 7 ecologists, the spatial and temporal aspects of biodiversity are not independent (Preston 1960, 8 van der Maarel 1993, Rosenzweig 1998, Adler and Lauenroth 2003, Adler 2004) in that the 9 species composition of smaller areas must fluctuate more rapidly than that of larger contiguous 10 areas. This interdependence of the spatial and temporal scaling of species richness has important 11 implications for the investigation of the causes and consequences of biodiversity. The study of 12 species-area curves is a case in point; much controversy has been associated with the spatial 13 dependence of species richness at fine scales (<10,000 m2) versus larger, regional scales (Connor 14 and McCoy 1979, Rosenzweig 1995). Fine-scale spatial patterns exhibit a faster rate of species 15 accumulation in space than larger scales (Williams 1964, Rosenzweig 1995) and are often 16 described by different models (Arrhenius 1921, Gleason 1922, Kylin 1926, He and Legendre 17 1996, Fridley et al. 2005). Although these inconsistencies have been explained away on both 18 biological and statistical grounds (Connor and McCoy 1979, Rosenzweig 1995, He and Legendre 19 1996, Hubbell 2001), few have explicitly recognized the greater temporal dependence of fine- 20 scale data or considered whether species-area relationships should fundamentally depend on the 21 temporal stability of populations at different spatial scales. 4 1 The species richness of small samples—those captured in small areas or small sampling 2 durations—is constrained by the total number of individuals in the sample, regardless of the 3 importance of ecological processes such as competition or local niche partitioning (Fisher et al. 4 1943, Gotelli and Colwell 2001, White 2004). Williams (1943) and Preston (1960) were among 5 the first to recognize that both species-area and species-time curves should be dominated by such 6 “sampling” processes at small areas or durations, and Preston (1960) explicitly addressed 7 whether species accumulation in space and time are related—his “ergodic conjecture” 8 (Rosenzweig 1998). Recent renewed interest in Preston’s hypothesis (Rosenzweig 1998, Adler 9 and Lauenroth 2003, Adler 2004, White 2004) has supported Preston’s case that species-area and 10 species-time curves are qualitatively similar and underlain by the same processes. However, 11 Preston’s hypothesis also suggests that classic debates over the behaviour of species-area curves, 12 and particularly those for plants (Arrhenius 1921, Gleason 1922, Williams 1964, Rosenzweig 13 1995, He and Legendre 1996, Hubbell 2001) may be resolved by incorporating the increasing 14 dependence of smaller-area samples on the temporal duration of community surveys. Adler and 15 Lauenroth (2003) and Adler et al. (2005) presented evidence that fine-scale accumulation rates of 16 plant richness may decrease over time, although its significance to full-scale species-area curves 17 was not addressed in their studies. Following Williams (1943) and Preston (1960), we further 18 suggest that the accumulation of species over time in small-area samples can be used to estimate 19 and subsequently correct for the influence of sampling processes in fine-scale species areas 20 curves, and thus potentially reconcile the spatial scaling of species richness at local and regional 21 scales (Smith et al. 2005). 22 In light of increasing interest in the relationship between accumulation patterns of species 23 richness in space and time (Rosenzweig 1998, Adler and Lauenroth 2003, Adler 2004, White 5 1 2004), we ask whether “correcting” fine-scale spatial richness data by accumulating richness 2 over time in small spatial samples reconciles fine-scale species-area curves and those of the 3 regions in which they are imbedded. Addition of a temporal component to species-area curves 4 should be especially important in communities that exhibit sufficient year-to-year fluctuation in 5 species composition to allow temporal accumulations over the scale of a few years to generate a 6 much larger sample size for small quadrats. We use plant community survey data from three 7 floristically separate perennial grassland communities that, by means of regular disturbance from 8 fire, grazing, or drought, display yearly turnover in fine-scale species composition. We 9 hypothesize that fine-scale species-area curves are strongly constrained by the sampling of too 10 few individuals, and that such constraints can be alleviated through the analysis of temporal 11 species accumulation for quadrats of different sizes. 12 REMOVING SAMPLING FROM FINE-SCALE RICHNESS VALUES 13 As with species-area relationships, the rate at which species richness accumulates over 14 time for a given area is a function of both sampling and ecological processes (Rosenzweig 1998, 15 Adler and Lauenroth 2003, Adler et al. 2005, White 2004). Although the accumulation of 16 species over relatively short temporal durations is thought to be largely the result of sampling 17 processes (Preston 1960, White 2004, Fridley et al. 2005, White et al. 2006), there remains the 18 possibility that some ecological process—such as a shift in local environmental properties— 19 contributes to the rate at which species accumulate from year to year. The problem becomes one 20 of isolating the portion of richness increase that comes only from obtaining a larger sample of 21 individuals (sampling) from that that would occur even if the number of individuals in one 6 1 sample were very large (ecological). This can be done by holding the influence of one process 2 constant, and calculating the rate at which richness increases by systematically varying the other. 3 If a large proportion of the individuals in a sample turnover from year to year and the 4 density of individuals (N) in the sample is relatively constant over time, then the accumulated 5 richness from adding one temporal survey of a fixed quadrat to another (∆S) should bear a 6 constant signature of sampling (i.e., N to 2N), regardless of the temporal extent between surveys. 7 However, the ecological influence on ∆S should be a function of temporal extent, assuming that 8 environments on average become less similar with the passage of time (analogous to the spatial 9 concept of distance decay of similarity; Nekola and White 1999). Thus, comparison of how 10 richness accumulates for a variety of two-time surveys of a single quadrat (e.g., comparing the 11 number of new species gained after 5 years versus that after 10 years) allows an estimate of the 12 unique contribution of ecological processes to the richness increase. Subsequent removal of this 13 component from ∆S suggests the rate that richness increases from one sample to two samples in 14 the absence of ecological processes, and this rate can be used in a sampling-only model to 15 estimate the richness of sample given infinite individuals (i.e., a sampling-independent estimate 16 of richness for a given quadrat size). 17 The approach of partitioning sampling and ecological components illustrated in Figure 1, 18 where a single quadrat is surveyed at three times.