Diversity maintenance mechanism changes with ... - Wiley Online Library

3 downloads 486 Views 3MB Size Report
Oct 17, 2016 - results suggest that species diversity maintenance mechanisms of the ... Key words: diversity maintenance mechanism; neutral model; niche ...
Diversity maintenance mechanism changes with vegetation type and the community size in a tropical nature reserve Shuzi Zhang,1,2 Runguo Zang,1,2,† Yunfeng Huang,1 Yi Ding,1,2 Jihong Huang,1,2 Xinghui Lu,1,2 Wande Liu,1 Wenxing Long,1 Junyan Zhang,1 and Yong Jiang1 1Key

Laboratory of Forest Ecology and Environment, The State Forestry Administration; Institute of Forest Ecology, Environment and Protection, Chinese Academy of Forestry, Beijing 100091 China 2Co-Innovation Center for Sustainable Forestry in Southern China, Nanjing Forestry University, Nanjing, Jiangsu 210037 China Citation: Zhang, S., R. Zang, Y. Huang, Y. Ding, J. Huang, X. Lu, W. Liu, W. Long, J. Zhang, and Y. Jiang. 2016. Diversity maintenance mechanism changes with vegetation type and the community size in a tropical nature reserve. Ecosphere 7(10):e01526. 10.1002/ecs2.1526

Abstract. Species-­abundance distribution (SAD) is an essential tool to explain the mechanism of diversity

maintenance in ecological communities. Most of the studies on diversity maintenance in a specific forest dynamics plot just consider stems with a certain minimum size class to include into the tree community to be examined. However, the species in the juvenile stage are easily disturbed by a variety of factors; here, we define the minimum stem size to tag trees in a community as the community size (DBHmintag), which implies that the communities with different minimum diameter at breast height (DBH) sizes to tag trees are tree assemblages containing tree populations of different minimum DBH. We used data from 17 1-­ha forest dynamics plots across six old-­growth forest types in a tropical nature reserve to explore diversity maintenance mechanism by SAD curves (at three levels of DBHmintag) fitting to neutral model and niche preemption model. We found that the SADs of the two zonal vegetation types (tropical montane rain forest and tropical lowland rain forest) were best fitted by neutral model at each level of DBHmintag; meanwhile, the best fitted models for the four azonal vegetation types (tropical coniferous forest, tropical deciduous monsoon rain forest, tropical montane evergreen forest, and tropical montane dwarf forest) varied with DBHmintag levels, for communities with DBHmintag ≥ 1 cm and DBHmintag ≥ 5 cm, the fitting effect of neutral model was better than niche preemption model’s for the forest dynamics plots in the four azonal vegetation types, and for communities with DBHmintag ≥ 10 cm, the four azonal vegetation types were all best fitted by the niche model. Our results suggest that species diversity maintenance mechanisms of the two zonal vegetation types derived from the neutral model increased the predictive accuracy at each level of DBHmintag, and meanwhile, the four azonal vegetation types derived from the neutral model increased the predictive accuracy at smaller community size; however, with the increase in DBHmintag, these communities derived from the niche theory model increased the predictive accuracy. Habitat heterogeneity might be the major constraints in determining the neutral or niche process for diversity maintenance of a specific forest community. Key words: diversity maintenance mechanism; neutral model; niche model; species-abundance distribution; tropical rain forest. Received 3 August 2016; revised 4 August 2016; accepted 19 August 2016. Corresponding Editor: Debra P. C. Peters. Copyright: © 2016 Zhang et al. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. † E-mail: [email protected]

Introduction

different communities (Whittaker 1965, 1972, Chao et  al. 2015). Species-­abundance relationship is one of the most fundamental aspects of community structure (Watkins and Wilson 1994), and abundance is a well measure of the effects a

A major research aim of ecology is to understand the mechanisms that generate and shape the differences among species abundances in  v www.esajournals.org

1

October 2016 v Volume 7(10) v Article e01526



Zhang et al.

species has on its inhabiting ecosystem (Ehrlén and Morris 2015). A species-­abundance distribution (SAD) can well display the abundance of all species recorded within a biotic assemblage (Ulrich et al. 2010, Matthews et al. 2014). The SAD has been defined as a vector of the abundances of all species presented in a community (McGill et al. 2007). It is one of the most basic information to further elucidate the composition, structure, and dynamics of an ecosystem (Tokeshi 1990, 1993, Morlon et al. 2009). As SAD can reflect some underlying processes that assemble ecological communities, it has been used to test alternative hypotheses on the determinants of community structure and species coexistence (Morlon et  al. 2009). SAD-­based models, which combine mathematical theories and ecological features, have been developed to characterize the pattern of distribution curve and potential mechanisms. Niche theory and neutral theory are two mainstreams of theoretical models to explain patterns of biodiversity (Chisholm and Pacala 2010). Neutral and niche theories propose contrasting explanations for the maintenance of species diversity (Brown et al. 2013). Classical niche theory having developed over 100  years is one of the main theories to explain species coexistence of natural communities (Silvertown 2004, Chase 2005). The theory assumes that different species are associated with different combinations of abiotic and biotic factors and that the diversity of species is mainly caused by the spatiotemporal heterogeneity in environment (Tilman 2004, Chisholm and Pacala 2010). Gause (1934) proposed the competitive exclusion principle and suggested that niche differentiation is a necessary condition for maintaining species coexistence. Otherwise species with the same niche could not coexist because of competition for the common resources (Vandermeer 1972). Competition can reduce diversity if superior competitors increase in abundance and exclude other species (Wright 2002). Variation among individuals and species in communities can respond to or affect the environment (Lowe and McPeek 2014). Niche differentiation is thus vital in regulating species coexistence at different scales (Chesson 2000, Chase and Leibold 2003, Leibold and McPeek 2006, Adler et al. 2007). The niche theory models closely link species abundance to the spatial and temporal distribution  v www.esajournals.org

of niche. Although niche theory models have well explained some SAD patterns of communities, traditional niche theory meets a big challenge when the niche theory models are applied to explain coexistence phenomenon in habitats with less limited resources and nonobvious niche differentiation; thus, neutral theory model was proposed as an alternative to niche theory model (Hubbell 2001, 2006). Since Hubbell (2001) proposed the neutral theory of biodiversity, it has become one of the most concerned while debated ecological theory (Bell 2001). The neutral theory of biodiversity is a convenient theoretical framework for linking biodiversity patterns to the fundamental processes of population demography (Volkov et al. 2005), which offer a simple phenomenological model of explanation for community features such as SAD and species–area curves (Etienne 2005, He and Hu 2005), without considering any functional differences among species (Brown et al. 2013). Under the neutral theory framework, all individuals within a particular trophic level have the same chances of per capita reproduction and death regardless of their species identity (Hubbell 2001, Rosindell et al. 2011), niche differences are irrelevant, and the relative abundance of each species in a community is determined by a “random-­walk” process, in which dead individuals should be replaced by either migrants from surrounding metacommunities or offsprings of extant individuals (Silvertown 2004, Munoz et al. 2007). The neutral model is very contentious, with some ecologists showing evidence supporting it (Etienne 2005, 2007, Zhou and Zhang 2008), and others offering evidence that falsify it (Ricklefs 2003, Adler 2004, Stanley Harpole and Tilman 2006). Many ecologists indicate that neutral process is only one of the diversity maintenance mechanisms and may not be applicable for all cases (McGill 2003, Tilman 2004). Species-­abundance distribution curves are now widely used to describe diversity status across terrestrial, aquatic, and marine ecosystems (Elith and Leathwick 2009). In different types of ecosystems, habitat variability has strong impacts on the distribution, abundance, and diversity of different species (Ehrlén and Morris 2015). Magurran and Henderson (2012) suggested that ecological communities are composed of species that exploit available habitat in different ways. 2

October 2016 v Volume 7(10) v Article e01526



Zhang et al.

SAD curves can capture varied natures of SADs within an ecological community. Using different theoretical models to fit SAD patterns in different vegetation types is an important approach to understand biodiversity maintenance mechanisms under varied biotic and abiotic conditions. Tropical forests are one of the most species-­rich ecosystems on the earth (Sterck et al. 2014). The coexistence of vast numbers of species in tropical forest has fascinated many ecologists over the world (Wilson et al. 2012). However, the high diversity of tropical forest has been challenging for diversity maintenance theories (May et  al. 2015). Some ecologists argued that neutral processes drive patterns of relative species abundance in high-­diversity communities such as the tropical rain forests, even when niche structures are assumed to exist; however, neutral theory cannot be used to infer an absence of niche structure (Chisholm and Pacala 2010). Up to present, most studies on diversity maintenance in tropical forests have examined the robustness of different coexistence processes (i.e., neutral or niche) for the same vegetation type (e.g., a big forest dynamics plot). But few of them compared the expected diversity maintenance across different vegetation types distributed in a tropical landscape. Furthermore, almost all of the studies just consider stems with a certain minimum size class to include into the tree community to be examined. However, the species in the juvenile stage are easily disturbed by a variety of factors—juveniles had higher mortality in suboptimal habitats (Paoli et  al. 2006), so it could help us to understand the diversity maintenance mechanism via defining the community size using the DBH. Here, we define the minimum stem size to tag trees in a community as the community size (DBHmintag), which implies that the communities with different minimum DBH sizes to tag trees are tree assemblages containing tree populations of different minimum DBH. On the basis of the investigation on tree stems in 17 1-­ha forest dynamics plots across six old-­growth tropical forest types in a biodiversity hotspot on Hainan island of China, we examined the species-­abundance patterns and their best fit models for each tree community in each forest dynamics plot at three different minimum diameter at breast height (DBH) classes (DBHmintag ≥ 1 cm, DBHmintag ≥ 5 cm, and  v www.esajournals.org

DBHmintag  ≥  10  cm) to disclose the common or differing underlying mechanisms of biodiversity maintenance among different vegetation types. We specifically explore the following questions: (1) “Is the best fit species-­abundance model vegetation type specific or DBHmintag specific?”— “Does the best fit model vary with vegetation type or also with the different DBHmintag within the same vegetation type?”, (2) “As per the best fit models, is the diversity of the same community maintained by only one single mechanism or by a combination of several mechanisms?”, and (3) “What are the main possible determinants of SAD patterns across the six vegetation types?”

Materials and Methods Study area

The study area is located in the Bawangling National Nature Reserve (BNNR, 18°57′– 19°11′ N, 109°03′–109°17′  E) on Hainan Island, China (Fig.  1), which is at the northern edge of the Asian tropical forest zone. The area is about 500 km2, with an elevation range of 100–1654 m a.s.l. (Ding et  al. 2012). The region is characterized by a tropical monsoon climate, the average annual temperature is 23.6°C, and the average annual precipitation is 1750 mm, with a distinct wet season from May to October and a dry season from November to April (Zhang et al. 2012). In the BNNR, the areas below 800 m in elevation have three types of tropical vegetation, tropical lowland rain forest (TLRF), tropical coniferous forest (TCF), and tropical deciduous monsoon rain forest (TDMRF). The low elevation tropical forests have similar amounts of precipitation; however, the local terrain and soil heterogeneity for each of these forest types are highly variable (Jiang et  al. 2015). TLRF is the zonal vegetation type in the low elevational areas of Hainan island, and it is distributed in better soil conditions and occupies the largest area in the low elevations (800  m) and low (≤800  m) elevations, respectively, had the faster rate of species accumulation for each DBHmintag classes, whereas the other four vegetation types had slower rate of species accumulations. The rank-­abundance curves (Fig. 2) showed a similar order of diversity change with the species-­abundance accumulation curves. Fig. 3 shows that the number of stems at both DBHmintag  ≥  5  cm and DBHmintag  ≥  10  cm only accounted for a small proportion of all the individuals in the communities with DBHmintag ≥ 1 cm.

Results Changes in species diversity and environmental factors across the six old-­growth forest vegetation types

The species-­abundance accumulation curve of the 17 1-­ha forest dynamics plots with three  v www.esajournals.org

6

October 2016 v Volume 7(10) v Article e01526



Zhang et al.

However, the number of species decreased relatively slowly with the increase in the DBHmintag. Table 1 shows that the CO was highest in the TMDF, followed by the TCF. The TMRF had the highest SWC and soil TK. The soil TN was highest in the TDMRF. Soil pH, SOM, and soil TP showed a similar pattern of change, with mean values in the TMRF and the TLRF at moderate level. The coefficients of variation in environmental factors in the TMRF and the TLRF were relatively lower than in other vegetation types except CO.

The model fitting for species-­abundance curves of the 17 1-­ha forest dynamics plots at different community sizes across the six old-­growth forest vegetation types

The statistical results of goodness of fit for the SAD curves in the 17 forest dynamics plots at three levels of DBHmintag across the six old-­ growth vegetation types are listed in Tables 2, 3, and the fitted SAD curves for models are summarized in Figs.  4–6. In the TMRF, neutral model was well fitted for the SADs of all the forest dynamics plots at levels of DBHmintag (Figs. 4–6). Through chi-­square tests, neutral model was not significantly different for the forest dynamics plots in TMRF except TMRF1 when all stems with DBHmintag  ≥  1  cm were included and the neutral model was accepted by chi-­square tests and also accepted by niche preemption model when DBHmintag ≥ 5 cm. Considering only large trees (DBHmintag ≥ 10 cm), the neutral model and niche preemption model were accepted by chi-­ square tests. The neutral model was the best fitting model for the SAD curves at three levels

Fig.  3. Comparison of the number of individuals (a) and the number of species (b) with three levels of DBHmintag in the 17 1-­ha forest dynamics plots across the six old-­grown forest types. TMRF, tropical montane rain forest; TLRF, tropical lowland rain forest; TCF, tropical coniferous forest; TDMRF, tropical deciduous monsoon rain forest; TMEF, tropical montane evergreen forest; TMDF, tropical montane dwarf forest, and number followed the vegetation types is the plot number.

Table 1. Average values and coefficient of variation (CV) for environmental factors of the six old-­growth forest vegetation types. Vegetation TMRF TLRF TCF TDMRF TMEF TMDF

CO (%)

SWC (%)

pH

SOM (g/kg)

TN (g/kg)

TP (g/kg)

TK (g/kg)

11.76 (27.85) 6.71 (27.7) 16.81 (14.85) 21.93 (46.06) 10.59 (36.56) 42.59 (28.03)

29.67 (12.74) 18.31 (19.81) 14.07 (20.73) 14.64 (30.15) 20.22 (29.09) 20.08 (32.85)

4.44 (3.39) 4.74 (3.29) 4.71 (3.75) 5.8 (6.39) 4.19 (3.89) 3.95 (9.01)

43.76 (12.96) 31.87 (17.67) 22.88 (20.74) 56.91 (10.88) 54.79 (18.39) 53.69 (52.22)

1.95 (15.95) 1.07 (17.11) 1.19 (22.34) 2.83 (29.7) 1.72 (28.93) 1.66 (31.77)

0.33 (26.83) 0.2 (33.76) 0.24 (20.49) 0.62 (50.64) 0.88 (83.13) 0.35 (49.73)

7.97 (10.64) 8.73 (8.81) 9.21 (11.36) 11.25 (14.08) 9.69 (17.01) 12.68 (21.42)

Notes: Canopy openness (CO, %), soil water content (SWC, %), soil organic matter (SOM, g/kg), soil total nitrogen (TN, g/ kg), soil total phosphorus (TP, g/kg), soil total potassium (TK, g/kg). TMRF, tropical montane rain forest; TLRF, tropical lowland rain forest; TCF, tropical coniferous forest; TDMRF, tropical deciduous monsoon rain forest; TMEF, tropical montane evergreen forest; TMDF, tropical montane dwarf forest, and the numbers in brackets represent coefficient of variation.

 v www.esajournals.org

7

October 2016 v Volume 7(10) v Article e01526



Zhang et al.

Table 2. The result of fitting to models of 17 1-­ha forest dynamics plots at three levels of DBHmintag across the six old-­grown forest types. DBHmintag ≥ 1 cm

DBHmintag ≥ 5 cm

DBHmintag ≥ 10 cm

Forest dynamics plots

S

J

θ

m

α

S

J

θ

m

α

S

J

θ

m

α

TMRF1 TMRF2 TMRF3 TMRF4 TLRF1 TLRF2 TLRF3 TLRF4 TCF1 TCF2 TCF3 TDMRF1 TDMRF2 TMEF1 TMEF2 TMDF1 TMDF2

232 192 254 237 194 175 206 198 163 149 167 141 171 127 142 103 109

6939 7287 6582 5378 5249 4270 5897 4629 6664 10996 7845 15871 14023 8966 7507 10183 9723

46.3 37.3 53.9 55.8 40.2 38.6 41.4 46.3 30.9 25.6 30.4 23.4 30 21.6 25.4 17.8 19.5

0.9 0.8 0.8 0.5 0.9 0.7 0.9 0.5 0.8 0.7 0.9 0.4 0.4 0.8 0.9 0.4 0.3

0.04 0.04 0.03 0.03 0.04 0.04 0.04 0.04 0.06 0.09 0.07 0.11 0.07 0.08 0.06 0.08 0.06

149 144 180 163 125 121 149 136 74 85 87 63 104 74 93 69 86

1548 1804 1667 1385 1446 1322 1556 1370 1582 2555 2397 2088 2424 2962 2774 2940 3447

43.6 36.8 54.4 48 45.1 32.4 42.8 37.5 16.1 16.9 17.7 12.2 22.1 13.8 18.5 14.9 16.4

0.7 0.9 0.8 0.9 0.2 0.9 0.8 0.9 0.9 0.9 0.9 0.9 0.9 0.9 0.9 0.3 0.8

0.04 0.05 0.03 0.03 0.04 0.05 0.04 0.05 0.11 0.14 0.1 0.16 0.08 0.12 0.08 0.1 0.07

109 99 131 116 89 81 105 100 47 51 64 40 52 62 71 53 65

681 782 832 656 696 631 749 659 742 791 1082 707 736 1399 1366 1121 1204

38.9 37.3 49.3 41.8 58.4 24.7 33.2 32.9 11.1 12.2 14.9 9.2 12.8 13.3 15.9 12 74

0.8 0.4 0.6 0.9 0.1 0.9 0.9 0.9 0.9 0.9 0.9 0.9 0.9 0.9 0.9 0.8 0.1

0.04 0.04 0.03 0.04 0.04 0.06 0.05 0.05 0.18 0.16 0.13 0.15 0.13 0.14 0.09 0.14 0.07

Notes: TMRF, tropical montane rain forest; TLRF, tropical lowland rain forest; TCF, tropical coniferous forest; TDMRF, tropical deciduous monsoon rain forest; TMEF, tropical montane evergreen forest; TMDF, tropical montane dwarf forest; and number followed the vegetation types is the plot number. S is the number of observed species; J, the number of individuals in the community; θ, the biodiversity parameter; m, immigration rate; and α, a constant rate.

and there were no clear best models for DBHmintag ≥ 5 cm and DBHmintag ≥ 10 cm classes. While, for TMDF2 at three levels of DBHmintag, the niche preemption model was the best fitting model according to chi-­square tests and AIC.

DBHmintag according to AIC in TMRF. The TLRF showed the similar result as TMRF. In the TCF, according to the chi-­square tests, there was no clear best model for all stems with DBHmintag  ≥  1  cm; however, the neutral model has minimum AIC value. For DBHmintag ≥ 5 cm classes, TCF3’s SAD was well fitted only by the neutral model. However, as for the TCF2 and TCF3, the niche preemption model was the best fit model for DBHmintag ≥ 10 cm according to chi-­ square test and AIC. The TDMRF showed the similar result as TCF with DBHmintag ≥ 1 cm, and only the niche preemption model could well fit the TDMRF2’s SAD with DBHmintag  ≥  5  cm. Considering only large trees (DBHmintag ≥ 10 cm), the best fit model was the niche preemption model through chi-­square tests and AIC in the TDMRF. In the TMEF, the fitting effect of neutral model was better than niche preemption model’s according to chi-­square test and AIC for both DBHmintag ≥ 1 cm and DBHmintag ≥ 5 cm classes; however, for large trees (DBHmintag ≥ 10 cm), the niche preemption was the best fitting model. In the TMDF, the result of the chi-­square tests showed that neutral model was not significantly different for the TMDF1 with DBHmintag ≥ 1 cm;  v www.esajournals.org

Discussion Mechanism of biodiversity maintenance implicated by the best fitting models

A central question in community ecology is the mechanism of species coexistence at small spatial scales (Wright 2002). Species-­abundance curve is one of the quantitative approaches to describe community composition. Meanwhile, the fitting of the species-­abundance patterns to the two competing hypotheses (such as neutral-­based model, niche-­based model) can reflect some underlying ecological processes in community assembly and diversity maintenance. Our study of utilizing different models to fit SAD patterns of the 17 1-­ha forest dynamics plots with three different DBHmintag classes across different vegetation types demonstrated that different forest types in our study region had different best fitting models for SAD curves. The SAD curves of 8

October 2016 v Volume 7(10) v Article e01526



Zhang et al.

Table 3. Goodness-­of-­fit test of models of 17 1-­ha forest dynamics plots at three levels of DBHmintag across the six old-­grown forest types. Forest dynamics plots TMRF1 TMRF1 TMRF2 TMRF2 TMRF3 TMRF3 TMRF4 TMRF4 TLRF1 TLRF1 TLRF2 TLRF2 TLRF3 TLRF3 TLRF4 TLRF4 TCF1 TCF1 TCF2 TCF2 TCF3 TCF3 TDMRF1 TDMRF1 TDMRF2 TDMRF2 TMEF1 TMEF1 TMEF2 TMEF2 TMDF1 TMDF1 TMDF2 TMDF2

DBHmintag ≥ 1 cm

DBHmintag ≥ 5 cm

DBHmintag ≥ 10 cm

Test method

Niche preemption

Neutral model

Niche preemption

Neutral model

Niche preemption

Neutral model

AIC χ2 AIC χ2 AIC χ2 AIC χ2 AIC χ2 AIC χ2 AIC χ2 AIC χ2 AIC χ2 AIC χ2 AIC χ2 AIC χ2 AIC χ2 AIC χ2 AIC χ2 AIC χ2 AIC χ2

1923.35 1421.00* 1495.11 787.95* 1775.81 645.84* 1601.55 587.20* 1381.17 556.87* 1256.69 523.53* 1577.77 776.94* 1786.29 1623.30* 1246.05 513.01* 1536.46 2185.80* 1344.92 668.01* 1606.68 3766.40* 1690.60 2040.30* 1063.59 577.61* 1145.81 482.05* 891.24 523.81* 627.18 69.32

1762.31 544.96* 1242.33 174.64 1434.35 118.89 1303.22 124.67 1121.02 112.16 1090.03 142.98 1386.41 221.91 1739.96 969.72* 1084.34 224.26* 1508.46 1695.70* 1242.71 444.88* 1595.57 3253.40* 1633.72 1350.80* 919.94 224.00* 979.04 138.73 700.32 101.50 938.69 533.66*

935.29 272.32* 974.04 335.46* 589.73 73.21 803.77 158.87 534.51 65.12 712.64 181.23* 726.42 129.558* 826.21 302.73* 537.72 316.32* 667.48 387.78* 567.58 172.96* 468.63 291.18* 438.51 81.08 494.82 174.76* 693.74 264.53* 572.44 454.84* 456.91 52.52

873.18 134.59 910.77 163.26 239.77 11.05 656.67 42.66 314.20 10.05 642.78 73.85 540.64 27.56 757.14 131.94 513.40 181.78* 668.60 404.83* 503.87 95.38 458.25 221.31* 502.37 128.18* 455.51 95.297* 603.47 84.61 507.90 177.60* 532.92 107.68*

281.65 24.24 256.43 19.99 252.41 24.87 287.63 39.80 138.97 11.97 158.51 14.11 240.27 23.83 346.17 61.51 347.49 286.35* 230.53 55.89 272.89 60.99 109.93 13.18 147.92 18.85 267.48 40.10 240.66 17.79 398.75 299.81* 154.50 7.00

60.48 4.96 2.99 4.28 58.77 4.86 119.14 8.00 64.68 5.30 124.37 7.78 147.03 7.83 219.52 14.92 342.29 202.19* 247.72 67.76* 283.67 70.11 152.22 18.87 159.28 25.63 291.48 66.33 271.28 20.17 376.34 155.00* 203.03 10.36

Notes: TMRF, tropical montane rain forest; TLRF, tropical lowland rain forest; TCF, tropical coniferous forest; TDMRF, tropical deciduous monsoon rain forest; TMEF, tropical montane evergreen forest; TMDF, tropical montane dwarf forest; and number followed the vegetation types is the plot number. AIC, Akaike information criterion. * P