Oecologia (1992) 91 : 129-133

Oecologia

9 Springer-Verlag 1992

Density and reproductive success in wild populations of DiploCaxis erucoides (Brassicaceae) W.E. Kunin*

Department of Zoology, NJ 15, University of Washington, Seattle, WA 98195, USA Received May 13, 1991 / Accepted in revised form March 13, 1992 Summary. One possible consequence of low population density, particularly in self-incompatible plants, is reproductive failure. I surveyed seed set per flower in two populations of the self-incompatible annual Diplotaxis erucoides (Brassicaceae) in Jerusalem, Israel. Widely spaced plants had lower fruit set and fewer seeds per filled silique than did plants growing close to conspecific neighbors. Such density-dependent reproductive success could help explain the maintanence of spatial patchiness in plant populations, and could also have implications for population dynamics of rare species. Key words: Density dependence - Pollination - Reproductive success - Diplotaxis erucoides

One of the traditional concerns of population biology is the effect of population density on population growth. In classical "density dependence," the survival or reproductive rate of a species is inversely correlated with its density, due to crowding effects, competition for resources, or increased predation (e.g. Smith 1935, Nicholson 1933, 1957; Lack 1954). Such dynamics create a negative feedback loop for population growth, increasing population stability. If populations become too high or low, their conditions deteriorate or improve accordingly, driving them back towards equilibrium. Allee and his coworkers (Allee 1951 ; Allee et al. 1949) first noted a different form of density-dependent effect. Reproductive rates of some organisms behave in a manner opposite to that generally associated with densitydependent processes. In very low density populations, they noticed, reproduction may be hampered by a shortage of potential mates and the difficulties associated with locating and courting them. Such dynamics create a positive feedback loop for population growth, increasing instability. Populations sinking beneath some thresh* Correspondence to: NERC Centre for Population Biology, Imperial College at Silwood Park, Ascot, Berkshire SL5 7PY, UK

old size could begin to spiral towards extinction, with ever lower densities resulting in successive cuts in reproductive success, further reducing density. The study of such density effects, consequently, is important both because of the theoretical interest of the subject, and because they may have applications in population management. Density effects are particularly important in understanding the dynamics of rare populations, and may be one of the causes of their extinction. Improving our understanding of them may be instrumental in refining the art of biological conservation. Self-incompatible plants are ideal subjects for the study of density effects in reproduction - both because they seem especially at risk and because of the ease of studying them. Plants, being sessile, must rely on outside agents (biotic or abiotic) for gamete transfer, and so may be limited in their ability to "search" for rare mates. This is particularly clear in wind-pollinated species (Kunin, unpublished data), where the reduction in pollen transfer at low density is unavoidable. The situation is less straightforward in animal-pollinated plants, as the effectiveness of pollination will depend on the behavioral choices of the pollinators. Models of pollinator foraging (Kunin 1991) suggest that reproductive success in a selfincompatible plant is likely to be strongly influenced by local population density, due to a decline in the number and/or the quality of pollinator visits. The importance of such density effects may loom even larger among annual (or otherwise monocarpic) plants, as poorly pollinated individuals cannot reappropriate reproductive resources into increased growth or storage for future needs. The sessile habit of plants also facilitates research, allowing density and reproductive success to be measured with precision and making it possible to follow even very low density populations with relative ease. Moreover, because of the reliance on external agents for pollen transfer, the mating activities of plants can be readily observed and modelled (in many motile organisms, the search for mates cannot easily be distinguished from search for food or shelter).

130 There have been surprisingly few studies o f reproductive success as a function o f p o p u l a t i o n density in the field. Silander (1978) studied reproductive success in the tropical shrub Cassia biflora, a n d f o u n d p o p u l a t i o n density to be correlated with reproductive success. Similar results have been reported by Platt et al. (1974) a n d H e i t h a u s et al. (1982) for two other shrub species (pollinated b y large bees a n d bats, respectively). Feinsinger a n d his c o w o r k e r s (1986) examined h u m m i n g b i r d visitation rates a n d reproductive success o f a n u m b e r o f shrub species in a C o s t a R i c a n cloud forest. T h a t w o r k , c o m bined with subsequent experimental m a n i p u l a t i o n s o f the system (Feinsinger et al. 1991) suggests that local floral density strongly influences reproductive success in selfincompatible plants (Palicourea lasiorrachis) b u t has little d e m o g r a p h i c influence in a self-compatible species (Besleria triflora). Allison (1990) studied n a t u r a l p o p u l a tions o f A m e r i c a n Y e w (Taxus canadensis) differing in m e a n n e a r e s t n e i g h b o r distance. In this wind-pollinated species as well, m e a n pollination success a n d p e r c e n t s e e d set were strongly correlated with p o p u l a t i o n density. This r e p o r t presents a n o t h e r case study o f a r e p r o d u c tive density effect in nature, taken f r o m a system very different f r o m those published to date. Here I examine seed set per flower in the crucifer Diplotaxis erucoides, a self-incompatible weedy a n n u a l f r o m Israel, as a function o f the distance between neighboring individuals. As an insect-pollinated a n n u a l f r o m a M e d i t e r r a n e a n environment, this species differs in g r o w t h form, pollinators, and h a b i t a t f r o m a n y o f the cases previously studied. This study, like m o s t o f those cited above, involves only a simple field c o m p a r i s o n o f seed p r o d u c t i o n a m o n g individuals. It neither provides the security o f a controlled experiment, n o r does it shed light o n the m e c h a n i s m s responsible for the effects seen (for these, see K u n i n 1991). lts value lies in d o c u m e n t i n g t h a t density effects on plant reproductive success exist in natural populations, a n d that they can be quite powerful.

B. The study populations This study was conducted on two populations on opposite sides of the Givat Ram campus of the Hebrew University of Jerusalem, in Israel. The "Life Sciences" population (LS) was on the north end of a field directly south of the Silberman Life Sciences building, near the western boundary of the Givat Ram campus. Much of this field was densely vegetated, but the Diplotaxis population was concentrated near the sides of a gravel road crossing the field, in areas with little other plant cover. By contrast, the "Botanic Garden" (BG) population grew among high grasses and other annuals in an area slated for expansion of the fledgling Botanic Garden along the eastern boundary of the campus. In each population there were several small clumps of closely spaced plants, surrounded by areas with relatively widely dispersed Diplotaxis individuals.

C. Samplin 9 techniques The LS population was surveyed on 17 March, 1986; the BG population on 10 and 18 March. Twenty plants were chosen haphazardly at each site, including all plants discovered more than 2 m from their nearest conspecific neighbor. I recorded the number of inflorescences on each plant as an index of its size, and measured its distance from its nearest conspecific neighbor ("nearest-neighbor distance" or NND). For each plant, I chose 6 inflorescences by facing away from the plant, reaching behind me, and grasping them at random. In plants with fewer than 6 inflorescences, all inflorescences were included in the sample. I counted the number of expanded fruits, empty fruits, and aborted buds on each of these inftorescences. All expanded fruits were collected from these inflorescences. Later, 5 of these fruits were chosen randomly (using a random number generator) and dissected to count the number of developing seeds they contained. All data were analyzed using the SYSTAT statistical package (SYSTAT Inc, Evanston IL) on an NEC microcomputer.

Results

The p r o p o r t i o n o f n o n - a b o r t e d flowers that set fruit is inversely correlated with the distance to the nearest 1.0

Methods

A. The study species

I A9

o o

~

9

0.8

'~A

9

&A

Diplotaxis erucoides (L.) DC. is a medium-sized (0.2-0.5 m tall) weedy annual crucifer, found throughout much of the Mediterranean area (Zohary 1966). In Israel, it is especially common in disturbed fields in the Jordan river valley where it often forms large dense patches, whereas in the Jerusalem area (where this study was conducted) populations are relatively scattered and often rather sparse. The plant bears numerous conspicuous white flowers in loose spikes. Diplotaxis inflorescences bagged in the field or greenhouse generally set few or no seeds. Experimental hand pollination (Kunin 1991) verifies that these plants require cross-pollination to set seed. In Jerusalem, Diplotaxis flowers were visited by several species of bees and flies, most notably by an unidentified Bombyliid. Fully pollinated flowers develop into papery siliques with up to 60 small (mean: 0.23 mg) seeds. Unpollinated flowers form small, empty siliques which are easily distinguished from aborted buds.

r~ 0.6

0.4 .i

O

~0

0.2

0.0

0.1

'

'

1.0

10.0

9

100.0

NND

Fig. 1. The fraction of non-aborted flowers setting fruits on sampled inflorescences of Diplotaxis erucoides as a function of the distance to the nearest conspecific neighbor (NND). In this and all subsequent figures, circles indicate plants from the LS population, triangles are from the BG site, NND is plotted on a log scale

131

1000

100

il

I

9

iv

9

9

0 o

9

100

0

9t9

10

v

0

"!t 9

o9

10

9 o

1

9

t~

1

O.1

i

I

1.0

10.0

1

0.1

100.0

i

i

1.0

10.0 NND

NND

Fig. 2. Seeds per pollinated fruit as a function of NND. Each point represents the mean of 5 fruits chosen randomly from the inflorescences sampled. Only fnaits with 1 or more seeds were counted. Seed set in this and all subsequent figures is plotted on a log scale

Fig. 4. Plant size, as estimated by inflorescence number, as a function of NND. No significant effect of inter-plant distance is apparent

100000

100.0

9 o

|

100.0

9

o

9

I

10000

9 9149 q~

10.0 0

~D 0.)

1000 o

09 r,t)

o9

1.0 100

.E 0.1

0.1

f

I

1.0

10.0

9

100.0

&o

10 0.1

I

1.0

10.0

100.0

NND

NND

Fig. 3. Seeds per non-aborted flower as a function of NND. These results are the product of those plotted in Figures 1 and 2. Note that the Y-axis has been rescaled

Fig. 5. Estimated total seeds per plant as a function of NND. Seed totals are estimated as (seeds/flower) * (past and present flowers/ inflorescence) * inflorescence number

neighboring conspecific (Fig. 1; P e a r s o n correlation coefficient = - 0.601, p = 4.06 x 10- 5). A similarly strong distance effect appears in the n u m b e r o f seeds per fruit as a function o f N N D , considering only those fruits containing some seed (Fig. 2, P e a r s o n correlation coefficient= - 0 . 5 8 3 , p = 7 . 8 5 x 10-5). T a k i n g b o t h effects together (Fig. 3), seedset per flower d r o p p e d dramatically as the distances between plants rose (Pearson correlation coefficient = - 0.669, p = 2.32 x 10-6). Thus, seedset per flower was cut a p p r o x i m a t e l y in half for each doubling o f the distance between a plant a n d its nearest conspecific neighbor. There was no significant relationship between N N D and plant size, as m e a s u r e d b y inflorescence n u m -

ber (Fig. 4, P e a r s o n correlation coefficient=0.037, p = 0.823). Consequently, the estimated n u m b e r o f seeds per plant (multiplying the average seeds per flower by the average n u m b e r o f flowers per sampled inflorescence and the n u m b e r o f inflorescences per plant) fell sharply as NND grew (Fig. 5, P e a r s o n correlation coefficient = - 0.409, p = 0.009).

Discussion

The results o f this study strongly suggest that r e p r o d u c tive success in Diplotaxis is dependent on local p o p u l a -

132 tion density. There are, however, several weaknesses in this argument. First, the data presented here were gathered during only a few days in the middle of the flowering season. Consequently, they may not accurately represent total reproductive output for the plants involved. Plants with low reproductive success per flower might flower longer and thus compensate for some or all of their losses (Kunin, in prep.). A further weakness of this study stems from my use of naturally occuring differences in local density. Arguably, areas where few Diplotaxis plants grew might have been inherently less suitable for the species (or for its pollinators) than densely populated areas. The differences seen in reproductive output might reflect underlying differences in site quality rather than differences in pollination success. Given the lack of correlation between N N D and plant size, this interpretation seems improbable. Nonetheless, only by experimentally manipulating densities can this problem be formally addressed. Finally, I performed no systematic observations on the number and mix of pollinators visiting plants differing in NND. Without such information, the argument f o r a causal link between local population density and reproductive output is further weakened, since some other interaction might have been responsible for the observed effect (e.g. mycorrhizal interactions, but see Hirrel et al. 1978). All three of these concerns are accounted for in the experiments reported elsewhere (Kunin 1991). Despite its weaknesses, this study has at least one advantage over more rigorous experimental approaches. By examining unmanipulated wild populations, the work reported here provides evidence that density effects in reproduction act within the range of conditions that occur naturally. Given the magnitude of the effects discovered, it seems likely that local population density does have an effect on reproductive success in Diplotaxis. It has become widely accepted among pollination biologists that pollination seldom limits reproductive success under natural conditions (Bawa and Beach 1981). However, the vast majority of pollination studies (indeed of all ecological research) has been performed on populations living at relatively high density, presumably due both to the ease of working on such species and a sense of their ecological importance. The work reported here, in concert with previously published work on temperate (Platt et al. 1974; Allison 1990) and tropical (Silander 1978; Heithaus et al. 1982; Feinsinger et al. 1986) shrubs, suggests that rather different conditions may exist for sparse populations. On one hand, this implies that all else being equal (displays, rewards, breeding biology), sparsely-populated species may be more likely to face pollen limitation than are their commoner counterparts. Conversely, if the reproductive success of such plants is often pollen-limited, they may encounter unusually strong selective pressure favoring either increased displays and rewards or increased self-compatibility. If population density may affect reproductive success in sparse plant population, it may also affect spatial patterns of success within a population, as discovered here. If the reproductive success of individuals in a pop-

ulation varies directly with population density on an individual scale (sensu Lewontin and Levins 1989), then spatial patchiness may be enhanced. Simple computer simulations demonstrate that randomly spaced populations quickly become clumped when density-dependent reproduction is modelled (Kunin, unpublished). Such conditions may also influence the evolution of dispersal distance. Most theoretical discussions of dispersal (e.g. Hamilton and May 1977) have recognized the problems inherent in dispersing too little - intense competition resulting in smaller size and lower reproductive output. The problems of overdispersal, however, if they are discussed at all, are generally couched in terms of unnecessary expense or risk, increased probability of encountering inappropriate habitat (Ellner and Shmida 1981), or the possibility of a genetic "outcrossing depression" effect (Shields 1982). This study and those cited above provide evidence of another significant factor selecting against over-dispersal: density-dependent fertilization. If self-incompatible plants face reproductive difficulties when they grow too far from conspecifics, they should be under selective pressure to keep dispersal distances low. In doing so, they minimize the probability of consigning offspring to the botanical equivalent of spinsterhood. Such low dispersal abilities should further enhance spatial population patchiness in self-incompatible plant species. Acknowledgements. I would like to thank Avi Shmida, Gordon Orians, Doug Schemske, Bob Paine, Dee Boersma, and Mama Sapsowitz for their intellectual contributions to this project and this document. The work reported here was supported by the Lady Davis Fellowship Trust at the Hebrew University of Jerusalem, Israel.

References

Allee WC (1951) The social life of animals, 2 Edition. Beacon Press Boston, 1958 reprint under original title Allee WC, Emerson AE, Park O, Park T, Schmidt KP (1949) Principles of animal ecology. W.B. Saunders Company Philadelphia Allison TD (1990) Pollen production and plant density affect pollination and seed production in Taxus canadensis. Ecology 71 : 516-522 Bawa KS, Beach M (1981)Evolution of sexual systemsin flowering plants. Ann Missouri Bot Gard 68:254-274 Ellner S, Shmida A (1981) Why are adaptations for long-range dispersal rare in desert plants? Oecologia 51 : 133-144 Feinsinger P, Murray KG, Kinsman S, Busby WH (1986) Floral neighborhood and pollination success in four hummingbirdpollinated cloud forest plant species. Ecology 67:449-464 Feinsinger P, Tiebout HM III, Young BE (1991) Do tropical birdpollinated plants exhibit density-dependent interactions? Field experiments. Ecology 72:1953-1963 Hamilton WD, May RM (1977)Dispersal in stable habitats. Nature 269:578-581 Heithaus ER, Stashko E, Anderson PK (1982) Cumulative effects of plant-animal interactions on seed production by Bauhinia ungulata, a neotropical legume. Ecology 63:1294~1302 Hirrel MC, Mehravaran H, Gerdemann JW (1978) Vesiculararbuscular mycorrhizae in the Chenopodiaceae and Cruciferae: do they occur? Can J Bot 56:2813-2817

133 Kunin WE (1991) Few and far between: plant population density and its effects on insect-plant interactions. Ph.D. Thesis. University of Washington Lack D (1954) The natural regulation of animal numbers. Oxford University Press New York Lewontin RC, Levins R (1989) On the characterization of density and resource availability. Am Nat 134:513-524 Nicholson AJ (1933) The balance of animal populations. J Anim Ecol 2:132-178 Nicholson AJ (1957) The self-adjustment of populations to change. Cold Spring Harbor Symp Quant Biol 22:153-173 Platt WJ, Hill GR, Clark S (1974) Seed production in a prairie legume (Astragalus canadensis L.): interactions between pollina-

tion, predispersal seed predation, and plant density. Oecologia 17:55-63 Shields WM (1982) Philopatry, inbreeding, and the evolution of sex. State University of New York Press Albany, NY Silander JW (1978) Density-dependent control of reproductive success in Cassia biflora. Biotropica 10:292-296 Smith HS (1935) The role of biotic factors in the determination of population densities. J Econ Entomol 34:1-12 Zohary M (1966) Flora Palaestina; Part 1 : Equisetaceae to Moringaceae. The Israel Academy of Sciences and Humanities Jerusalem