Pollen limitation and Allee effect related to ... - Wiley Online Library

Plant Biology ISSN 1435-8603

RESEARCH PAPER

Pollen limitation and Allee effect related to population size and sex ratio in the endangered Ottelia acuminata (Hydrocharitaceae): implications for conservation and reintroduction J. Xia1,2, J. Lu1,3, Z. X. Wang1,3, B. B. Hao1,3, H. B. Wang1,3 & G. H. Liu1 1 Key Laboratory of Aquatic Botany and Watershed Ecology, Wuhan Botanical Garden, the Chinese Academy of Sciences, Wuhan, China 2 Engineering Research Centre for Protection and Utilization of Bioresource in Ethnic Area of Southern China, College of Life Sciences, South-Central University for Nationalities, Wuhan, China 3 Graduate School of the Chinese Academy of Sciences, Beijing, China

Keywords Ottelia acuminata; pollen limitation; population size; reintroduction; sex ratio. Correspondence J. Xia & G. H. Liu, Key Laboratory of Aquatic Botany and Watershed Ecology, Wuhan Botanical Gardens, Wuhan 430074, China. E-mail: [email protected], [email protected] Editor A. Dafni Received: 4 August 2011; Accepted: 11 June 2012 doi:10.1111/j.1438-8677.2012.00653.x

ABSTRACT Small populations may suffer more severe pollen limitation and result in Allee effects. Sex ratio may also affect pollination and reproduction success in dioecious species, which is always overlooked when performing conservation and reintroduction tasks. In this study, we investigated whether and how population size and sex ratio affected pollen limitation and reproduction in the endangered Ottelia acuminata, a dioecious submerged species. We established experimental plots with increasing population size and male sex ratio. We observed insect visitation, estimated pollen limitation by hand-pollinations and counted fruit set and seed production per fruit. Fruit set and seed production decreased significantly in small populations due to pollinator scarcity and thus suffered more severe pollen limitation. Although frequently visited, female-biased larger populations also suffered severe pollen limitation due to few effective visits and insufficient pollen availability. Rising male ratio enhanced pollination service and hence reproduction. Unexpectedly, pollinator preferences did not cause reduced reproduction in male-biased populations because of high pollen availability. However, reproductive outputs showed more variability in severe male-biased populations. Our results revealed two component Allee effects in fruit set and seed production, mediated by pollen limitation in O. acuminata. Moreover, reproduction decreased significantly in larger female-biased populations, increasing the risk of an Allee effect.

INTRODUCTION More and more plant species, especially those vulnerable to current global change, have been drawn into extinction vortex due to population explosions (Brook et al. 2008). Therefore, conserving plant diversity is becoming a critical concern in ecology. Botanical gardens around the world play an important role in in-situ and ex-situ conservation of plant diversity (Huang et al. 2002; Chen et al. 2009). Moreover, as the final goal of ex-situ conservation, reintroduction of plants into their original habitat, from which they had disappeared, is an increasingly important tool for scientific wildlife management (Deredec & Courchamp 2007). However, there are a distressingly large number of failures in conservation and reintroduction programmes because of reduced fitness or growth rate, especially in small founder populations (Courchamp et al. 1999, 2008; Stephens & Sutherland 1999; Stephens et al. 1999; Berec et al. 2007). Therefore, it is crucial to assess factors influencing reproductive success and determine the minimum viable population size in specific contexts (Deredec & Courchamp 2007). 376

In flowering plants, successful pollination is a prerequisite for sexual reproduction. However, plants under natural pollination conditions often suffer from pollen limitation, especially in small populations (Ashman et al. 2004; Knight et al. 2005; Le Cadre et al. 2008; Wagenius & Lyon 2010). For example, plants in small populations attract fewer pollinators, resulting in reduced stigma pollen load and thus decreased seed set. Such a reduction in fecundity is called pollen limitation, creating an Allee effect, i.e. the size or density dependence of fitness components or growth rate (Waites & Agren 2004; Aizen & Harder 2007; Elam et al. 2007; Wagenius & Lyon 2010). On the other hand, the vulnerability of plant species to Allee effects largely depends on their mating system (Ghazoul 2005). Self-incompatible species are more likely to suffer pollen limitation than self-compatible species because the former cannot compensate for reduced pollinator services through selfing (Aizen et al. 2002; Knight et al. 2005; Aguilar et al. 2006). Dioecious plants cannot self-pollinate and their sexual reproduction depends absolutely on pollen transfer from male to female, making them vulnerable to reduced reproduction ¨ ster & Eriksson 2007). Consethrough pollen limitation (O

Plant Biology 15 (2013) 376–383 ª 2012 German Botanical Society and The Royal Botanical Society of the Netherlands

Xia, Lu, Wang, Hao, Wang & Liu

quently, sex ratio, as well as population size, may affect male abundance and further reproductive success. The Allee effect occurs more frequently in female-biased populations, and such populations cannot survive if the density or number of males decreases below some threshold (Carlsson-Graner et al. 1998; Boukal & Berec 2002; Engen et al. 2003; Rankin & Kokko 2007; Shelton 2008). Furthermore, pollinators usually prefer male flowers because of sexual dimorphism, which often evolved to ensure that males are more attractive to pollinators than females (Corff et al. 1998; Hemborg & Bond 2005; Huang et al. 2006). Reduced reproduction can be expected if pollinators either ignore populations with few males or avoid visiting female flowers within populations that have a surfeit of males. Therefore, it is necessary to consider the effects of sex ratio when conserving or reintroducing dioecious populations. However, few studies have investigated the role of the sex ratio in the Allee effect in conservation or reintroduction programmes (Deredec & Courchamp 2007; Hardwick et al. 2011). In this study, we investigated whether and how population size and sex ratio affected pollen limitation and reproduction in the endangered Ottelia acuminata, a dioecious submerged aquatic species. This species was once dominant but has now disappeared from many lakes (e.g. Dianchi Lake) in southwest China, mainly because of water pollution (Yu et al. 2000; Godo et al. 2003). O. acuminata is sensitive to deterioration of water quality, is considered an indicator species, and thus is a priority species in lake restoration for government decision-making (http://www.shidi.org/sf_C0677C6AF 68D4FD8A29C6F04293BEAAA_151_ynsd.html). First, we asked whether there is an Allee effect mediated by pollen limitation. Second, we examined how the ratio of males in populations affected pollination and reproduction. Finally, we assessed whether a biased sex ratio could increase the risk of the Allee effect in O. acuminata, given the same population size. MATERIAL AND METHODS Study species and site

Ottelia acuminata (Hydrocharitaceae), a perennial submerged species endemic to China, is dioecious, insect-pollinated and endangered in the wild. O. acuminata shows sexual dimorphism in flower display. Each male inflorescence can produce 40–50 male flowers, while female spathes only have two to nine female flowers. Both male and female flowers are white with a yellow base, attracting visits from generalist pollinators. The single-day flowers open in the morning sun and close in the afternoon. We carried out the experiments and observations at Huichang village (2444¢14.64¢¢ N, 10235¢21.37¢¢ E) close to Dianchi Lake, where a workstation has been established to restore submerged plant populations in the lake. Plants used in the experiments were collected from Heilong Lake, Dali city, Yunnan Province. We transplanted the plants into 20-cm diameter pots, placed in cement pools with a water depth of 40 cm. Experimental design

To estimate the effects of population size and plant sex ratio on pollination success and final reproductive output, we

Pollen limitation and Allee effect related to population size and sex ratio

established experimental plots of increasing sizes (two, four, ten and 20 individuals in plots termed S2, S4, S10 and S20, respectively) and increasing male plant sex ratio (10%, 30%, 50%, 70% and 90%; plots termed R10, R30, R50, R70 and R90, respectively) from May to October 2010. S2 and S4 plots were only arranged with a 50% male ratio and replicated five times, while the other plots contained all five sex ratios and three replicates each. In total, we had 40 arrays and replicates in different observation periods. We arranged males and females alternately in 50% male ratio plots, but randomly in other plots. Plants within populations were placed 0.3 m from the nearest neighbour. Neighbouring populations were separated by at least 100 m when more than two assays were involved. In experimental populations, we labelled each flowering individual, transferred it to a plastic drum (30-cm high) filled with water, and then left the drums under natural pollination conditions. Unopened flowering individuals were replaced with open-flowered individuals every day before sunrise. Pollination observations lasted one week for each replicate. Males usually had larger floral displays than females, resulting in a more male-biased floral sex ratio compared with the overall plant sex ratio (paired t230, 0.05 = 15.645, P = 0.000; Fig. 1). Floral sex ratio did not differ significantly across population sizes, except when the male plant sex ratio was 30% (t45, 0.05 = 9.306, P = 0.000; Fig. 1). Floral sex ratio was female-biased in S20–R30 populations, but often malebiased in S10–R30 populations (Fig. 1). The number of male flowers did not differ significantly between S10 and S20 populations unless populations were male-biased at the plant level (Fig. 1). Pollination observation

To determine the effects of population size and sex ratio on pollination service, pollination observations were made between 08:00 and 13:00 h on days when weather conditions were optimal. Each day was divided into 15 20-min observation periods. A single observer sat approximately 1 m north of the focal population to avoid casting a shadow over it. Pollinators were tracked visually for as long as possible until they left the focal population, and their foraging sequence within and between plants was recorded during each observation period. A pollinator foraging in a flower was considered as a visit. For dioecious species, successful pollination requires pollen dispersal between males and females. The foraging sequence from male to female is thus essential for pollen transfer. Therefore, we defined an effective visit as when a female flower was visited after pollinator visits to male flowers. Finally, flower display size of each plant was counted after the observations. We assessed pollination success using the following three parameters. (i) At population level, pollinator approaches were assessed as the number of pollinators visiting a focal population per 20-min interval, indicating pollinator flower attractiveness. (ii) At individual level, we calculated visits and effective visits per female flower per day (number of visits or effective visits to flowers on the focal female plant ⁄ floral display size) for each female plant, which implied the level of pollination achieved. (iii) Pollen availability (PA) for seed set was also calculated as: PA = M ⁄ F,


377



Fig. 1. Population characteristics in Ottelia acuminata with increasing number (size 2, 4, 10 and 20) and male plant sex ratio (10%, 30%, 50%, 70% and 90%). (A) Number of open male and female flowers on each day. (B) Relationships between male flower and plant sex ratio. Males usually had larger displays than females, resulting in a more male-biased flower sex ratio compared with the plant sex ratio (above diagonal). Data points represent values calculated on each day.

where M represents total number of visits to donor male flowers during foraging trips when the focal female individual was effectively visited, and F represents floral display size of the focal female plant. We considered M as the number of open male flowers if the value of M was more than the total number of male flowers. Pollen limitation and female reproductive success

To determine whether pollination limited reproductive success and to assess the effects of population size and sex ratio, we conducted experimental hand-pollination during the 1week pollination observations. On a given day, one or more female flowers were subjected to supplemental pollination while the other flowers were left for natural pollination by insects. For hand-pollination, we harvested male flowers from the greenhouse, rubbed the donor anthers against the bifid styles of target flowers until there were sufficient pollen grains on the surface then marked the hand-pollinated flowers with red string. At least one flower was hand-pollinated on each female plant by the end of the 1-week observation for each population. One month after treatments, mature fruits on each study inflorescence were collected and the number of seeds per fruit were counted. Finally, two components of female reproductive success were measured: (i) fruit set (ratio of fruits to flowers per plant in each population); and (ii) seed production (number of seeds per fruit). Seed production per fruit under natural and supplemental pollination was used to calculate the intensity of pollen limitation (PL%) expressed as: PL = (1)C ⁄ X)*100, where C and X represent seed production per fruit under natural and supplemental pollination (modified from Gonza´lez-Varo & Traveset 2010). The value of PL varied from 0 (no pollen limitation) to 100 (full pollen limitation). 378

Data analyses

To determine the effects of population size and plant sex ratio on pollinator attractiveness, approaches of total, major and minor pollinators were analysed using manova, with population size and plant sex ratio as fixed effects. Because pollinator approaches of different insect types may be intercorrelated, manova was used initially to control for Type I errors. A significant manova was followed by univariate anova, using the Bonferroni method for multiple comparisons. Two-way anova was used to determine the effects of population size and sex ratio on total visits per female flower, effective visits per female flower and pollen availability. To test for the potential mechanism of size and ratio effects on pollination success and the possibility of a pollinator foraging preference for male flowers, step-wise multiple linear regression analyses were then performed. Population size, plant and floral sex ratio, and number of male and female flowers were set as independent variables, and the measures of pollination success (pollinator approaches, total and effective visits per female flower, pollen availability) were dependent variables. As the regression analyses on pollinator approaches were based on day means, data were weighted by the number of 20-min observation periods on each day. All predictor factors were excluded from the step-wise regression models on total visits per female flower, which was thus removed from the analysis. Because pollen limitation and seed production might be inter-correlated, manova and following univariate anova were also used with population size and plant sex ratio as fixed effects. A two-way anova was then used to determine the effects of population size, sex ratio and fruit set. Size · Ratio interactions were not significant in any case, so were dropped them from the models. All statistical analyses were performed with spss 16.0 (SPSS, Chicago, IL, USA).




Table 1. Effects of population size, plant sex ratio and their interaction (size · ratio) on pollinator attractiveness (number of pollinators per 20-min interval) for different insect types (analysed with MANOVA).

MANOVA

(df = 3732)

all pollinators (df = 1867)

major pollinators (df = 1867)

minor pollinators (df = 1867)

factor

Wilks’ k

F

MS

F

MS

F

MS

F

population size plant sex ratio size · ratio

0.81 0.92 0.90

70.13** 20.31** 24.22**

912.55 72.27 31.94

144.02** 11.41** 5.04**

25.89 31.59 30.41

22.65** 27.63** 26.60**

632.36 54.26 80.82

137.27** 11.78** 17.54**

Asterisks indicate

MANOVA

or univariate

ANOVA

statistically significant: **£0.001.

RESULTS Pollinator approaches

In total, there were 16 insect species that visited flowers of O. acuminata. According to body size and foraging behaviour, these visitors were grouped into two pollinator types: major pollinators (body size > 1 cm, suitable for the size of the pistil and androecium, frequent between-plant foraging, mainly Eristalis tenax) and minor pollinators (tiny flies, common but usually staying on a single flower for a long time and rare between-plant foraging). Using a manova, we found population size and plant sex ratio had significant effects on pollinator approaches for all insect types (Table 1). Also, we detected a significant interaction between population size and sex ratio (Table 1). Regardless of pollinator type, pollinator attractiveness dramatically declined in small S2 and S4 populations (Fig. 2); there were only 0.1 and 0.6 pollinator visiting S2 and S4 population

per 20 min, respectively, significantly less than for larger population sizes (Fig. 2). S20 populations attracted significantly more pollinators than S10 populations at all sex ratios except R90, for which the difference in approach rate between S20 and S10 populations was not significant (F = 2.487, P = 0.115; P < 0.01 for each others; Fig. 2). Pollinator attractiveness also significantly decreased in severely female-biased R10 populations. Increased male sex ratio resulted in a significant increase in pollinator approaches (F4, 1234 = 7.284, P = 0.000; Fig. 2). Both major and minor pollinators rarely visited small populations (Fig. 2). No major pollinators visited S2 populations and only visited S4 populations 0.1 times per 20 min. Minor pollinators visited S2 and S4 populations 0.1 and 0.5 times per 20 min, but visited 1.5–4.4 times in larger populations (Fig. 2). Multiple linear regression revealed that major pollinators had a preference for foraging on male flowers, while minor pollinators did not. The number of male flowers in a population was the only factor responsible for major pollinator approaches (Table 2). Major pollinators, which most likely make inter-plant movements, tended to avoid populations with large numbers of females. As a result, major pollinator approaches declined dramatically in female-biased populations (Fig. 2). Less than 0.2 major pollinators foraged in R10 populations, while 0.5–1.6 major pollinators visited larger populations (Fig. 2). In contrast, minor pollinator approaches were largely determined by population size (Table 2). Among S10 populations, minor pollinator approach only significantly increased in 90% male populations (F4, 1867 = 5.929, P < 0.001; Fig. 2). Among S20 populations, however, minor pollinator approaches initially increased with male sex ratio then declined significantly in male-biased populations (F4, 1867 = 20.4887, P < 0.001; Fig. 2). Pollination rate and pollen availability

Fig. 2. Mean number of pollinator visits (approaches) per 20-min interval in Ottelia acuminata with increasing population size (S2, 4, 10 and 20) and male plant sex ratio (10%, 30%, 50%, 70% and 90%). Pollinators were grouped into two types: major pollinators with larger body size and frequent between-plant movement, and minor pollinators with small size and rarely foraging between plants.

Two-way anovas revealed significant effects of population size and plant sex ratio on total visits per female flower per day, but there were no significant interaction effects between population size and sex ratio (Table 3). Total visits per flower per day in small populations were significant lower than in larger populations (F3, 849 = 6.811, P = 0.000; Fig. 3A). Female flowers were visited, on average, 0.4 times in S2 plots and once in S4 plots. By contrast, female flowers were visited on average 1.7 and 1.5 times in S10 and S20 populations, respectively. In S10 and S20 populations, the number of visits per flower was not monotonic to population sex ratio, with the highest visit frequency


379



Table 2. Multiple linear regression analyses of population size, plant and floral sex ratio, number of male and female flowers on major and minor pollinator approaches, effective visits per female flower per day and pollen availability (step-wise methods were used). regression models response variable

r

df

F

P

predictor variables

b

t

P

major pollinators minor pollinators effective visits pollen availability

0.351 0.352 0.065 0.224

197 197 852 142

106.044 106.327 59.098 40.639

0.000 0.000 0.000 0.000

male flowers population size floral sex ratio male flowers

0.593 0.593 0.255 0.473

10.298 10.311 7.688 6.375

0.000 0.000 0.000 0.000

Table 3. Effects of population size, sex ratio and their interaction (size · ratio) on total visits and effective visits (visiting females after males) per female flower per day, and pollen availability for Ottelia acuminata populations (analysed with two-way ANOVAs). total visits

effective visits

pollen availability

source

df

MS

F

P

df

MS

F

P

df

MS

F

P

population size plant sex ratio size · ratio Error

3 4 4 840

44.72 27.53 1.07 3.54

12.64 7.78 0.30

0.000 0.000 0.876

3 4 4 840

3.30 7.35 6.49 0.64

5.18 11.54 10.18

0.002 0.000 0.000

3 4 4 130

15.11 706.39 57.11 87.72

0.172 8.05 0.65

0.915 0.000 0.627

Fig. 3. Visits per flower per day (A), effective visits per flower per day (B), pollen availability (C), pollen limitation (D), fruit set (E) and seed production per fruit (F) of Ottelia acuminata with increasing population size (S2, S4, S10 and S20) and male sex ratio (R10, R30, R50, R70 and R90). Bars represent SE.

380




Table 4. Effects of population size and plant sex ratio on pollen limitation and seed production per fruit (analysed with MANOVA), also their effects on fruit set in Ottelia acuminata (analysed with a two-way ANOVA). Size · Ratio interactions were not significant in any case, so were dropped from the models. MANOVA

(df = 1102)

pollen limitation (df = 552)

seed production (df = 552)

fruit set (df = 322)

factor

Wilks’ k

F

MS

F

MS

F

MS

F

population size plant sex ratio

0.95 0.82

5.06** 14.90**

5592.68 15687.25

6.17** 17.30**

9807.66 36539.64

7.28** 27.12**

7882 20,635

5.31** 13.91**

Asterisks indicate

MANOVA

or univariate

ANOVA

statistically significant: **£0.001.

occurring in the R50 population (F4, 753 = 7.757, P = 0.000; Fig. 3A). Effective visits per flower were also significantly affected by population size and male sex ratio, and there were significant Size · Ratio interactions (Table 3, Fig. 3B). Regression analysis demonstrated that differences in effective visits were largely explained by floral sex ratio (Table 2). Effective visits per flower significantly decreased in female-biased floral ratios (t486.4, 0.05 = 9.764, P < 0.001). For example, almost no female plants achieved effective pollination per day when the floral ratio was female-biased: 372 of 380 plant observations had no effective visits, and effective visits were less than once for the other 18 observations. When the floral ratio was even or male-biased, however, female plants achieved one or more visits per flower per day in 28% of the 473 observations (ranging from one to nine). In contrast to the total and effective visit number, pollen availability was significantly affected by sex ratio rather than by population size (Table 3, Fig. 3C). Pollen availability was significantly higher in male-biased populations than in other populations, while there were no significant differences among population sizes (Fig. 3C). The following regression analysis revealed that the detected sex ratio effect was induced by a difference in number of male flowers (Table 2). Pollen availability was significantly positively correlated with male flowers number (r = 0.473, P < 0.001). Pollen limitation and female reproductive success

The manova and univariate anova revealed that both population size and plant sex ratio significantly affected pollen limitation and reproductive output (Table 4, Fig. 3D–F). Small populations and female-biased populations suffered significantly more severe pollen limitation than other populations (Fig. 3D). As a result, fruit set and seed production per fruit decreased significantly in small and in femalebiased populations (Fig. 3E and F). Fruit set and seed production in the S2 populations was half that in larger even sex ratio populations and was more variable (Fig. 3E and F). We detected no significant differences in fruit set between S10 and S20 populations (F1, 285 = 3.536, P = 0.061; Fig. 3E). However, seed production per fruit in S10 populations was higher than in S20 populations when the male ratio was