Search

The loss of the intromittent organ (IO) in the majority of birds remains unexplained. Here, I propose that IO loss results from sperm competition in the context of the unique avian sperm storage system. The first stage of fertilization is the movement of sperm through the vagina from the site of ejaculation to the sperm storage tubules (SSTs) at the uterovaginal junction. In a second stage, sperm are released from the SSTs and move through the uterus to the site of fertilization in the infundibulum. Last-male fertilization precedence can occur if sperm that arrive later at the uterovaginal junction occupy uterus-side SSTs, which then have a head start in the race to the infundibulum with each ovulation. Under such "last-in, first-out" conditions, there is strong sperm competition to arrive later at the SSTs. Consequently, the optimal male strategy would be placement of the ejaculate at the cloacal opening to allow any other sperm in the vagina to reach the SSTs first. Cloacal placement is effectively achieved by loss of the IO. The evolution of altricial development in Neoaves, the largest clade that lacks IOs, created conditions that favor IO loss. Specifically, the smaller clutch sizes and hatching asynchrony of altricial birds increase the intensity of sperm competition for fertilization of early eggs in the laying sequence and thus the selective advantage of later arrival at the SSTs. The rarity of IO loss among all animals suggests that the complex mechanism of avian fertilization creates unique conditions for sperm competition.

Antonín Dvořák composed the String Quartet in F major (op. 96, the ‘American’) in June 1893 while visiting the town of Spillville, Iowa, USA. Based on accounts from Dvořák's secretary Josef Kovařík, portions of the quartet's Scherzo movement were inspired by birdsong Dvořák heard in Spillville. Kovařík's account included both Dvořák's visual description of the bird that inspired him and a transcribed fragment of the birdsong Dvořák reportedly used. In 1954, Clapham identified the bird as a scarlet tanager (Piranga olivacea), based mainly on the visual description. This widely accepted identification was questionable because of the poor match of the tanager's song to musical material in the Scherzo. In 2016, Floyd proposed the red-eyed vireo (Vireo olivaceus) as Dvořák's inspiration, based on song rather than appearance. The vireo's song is clearly identifiable in the Scherzo as well in the birdsong fragment recorded by Kovařík. The mismatch between the visual description and the transcribed song fragment could have occurred when Dvořák heard and recorded the persistent singing of a cryptic red-eyed vireo, but misidentified the song's source when he caught sight of a brightly coloured tanager. Correct identification of the birdsong changes the interpretation of the movement. Previously, birdsong has been considered the source of a short secondary motif. But red-eyed vireo songs are highly variable, and a slight variation of Kovařík's song fragment could be the basis of the main theme of the Scherzo. Therefore, Dvořák may have used American birdsong in the movement considerably more than previously appreciated.

As a consequence of extensive habitat loss and fragmentation, grassland bird populations have declined substantially across much of North America. To assess management strategies for restoring and maintaining suitable habitat for grassland birds, we conducted a four-year survey of grassland-breeding passerines in the restored tallgrass prairies of the Carleton College Cowling Arboretum in Northfield, Minnesota. Between 2010 and 2013, we sampled bird populations weekly during the breeding season in ten adjacent management units that differed in size and spatial configuration. We used principal components analysis (PCA) to compare avian community composition among the units and to examine the association between habitat properties and makeup of the community. The size of a management unit had no effect on the bird community, but there was a strong impact of forest edge on community composition. Management units with less exposure to forest edge had greater abundance of grassland-obligate but not gr...

Amphibian populations have been declining worldwide, with multiple potential causes. At La Selva field station in north-eastern Costa Rica, previous work has shown that populations of many amphibians have decreased significantly since the 1970s, especially in primary forest. Starting in 1998, we investigated one of the most common frog species at La Selva, the poison-dart frog Oophaga pumilio (= Dendrobates pumilio). In a survey of 50 plots of 100 m2 in 1998, adult frogs were 4.6 times more abundant in secondary forest than in primary forest. Tadpoles were found only in secondary-forest plots. Almost all (89%) of the tadpoles were found in leaf axils of Dieffenbachia spp., which were much more abundant in secondary-forest than in primary-forest plots. The greater abundance of Dieffenbachia spp. in secondary forest was confirmed in a broad survey of ~11 km of trails within La Selva in 2002. When the same trails were resampled in 2012, Dieffenbachia spp. had been extirpated from 72% of the 50-m segments where plants were present in 2002; abundance was greatly reduced in the few trail segments where any Dieffenbachia spp. remained in 2012. The loss of Dieffenbachia spp., especially in secondary forest, removed the species most often used by O. pumilio for tadpole rearing. Based on counts of calling frogs in 2010, there was no difference in O. pumilio abundance in primary versus secondary forest, in striking contrast to multiple earlier surveys that found much greater frog abundance in secondary forest. We propose that the reason for the rapid decline in Dieffenbachia spp. is herbivory by the collared peccary (Pecari tajacu), which has increased in abundance at La Selva in recent years. A likely consequence is continued reduction in O. pumilio populations.

Understanding patterns of soil C and N accumulation in restored prairies is necessary to determine the sequestration potential of these systems. Estimates of C and N accumulation following prairie restoration are generally based on chronosequence studies. However, few studies have compared a chronosequence approach to direct measures of accumulation over time. Restored prairies in the Cowling Arboretum of Carleton College, Northfield, MN, were established annually from 1995 to 2007. In 2000 and 2010–2011, prairies were sampled for soil %C and %N and soil bulk density, allowing for a comparison of the effects of prairie age using a chronosequence approach to direct measures of C and N accumulation rates. Soil C and N content increased in all restorations, although there was no effect of field age across the chronosequence on C or N pools. However, rates of soil C and N accumulation decreased with restoration age, suggesting a rapid increase in the years immediately following agricultural conversion. This rapid accumulation was likely due to the turnover of annual and biennial species that dominated plant cover in the early in succession. This suggestion was supported by an observed increase in δ¹³C values with increasing prairie age. Thus, contributions from early successional species may result in rapid recovery of soil pools following restoration from agriculture. Direct measures of soil C and N accumulation may be necessary when initial variability among sites limit the ability of a chronosequence approach to detect significant changes in soil C and N content over short time periods.

The once-extensive tallgrass prairie community of North America has been reduced to small rem- nants, many of which are surrounded by intensive corn (Zea mays) agriculture. We investigated adult corn- rootworm beetles (Chrysomelidae: Diabrotica spp.), important pests of corn, on sunflowers (Asteraceae: He- lianthus spp.) in prairie remnants in southeast Minnesota. Large numbers of beetles invaded the prairie from surrounding corn fields in late summer. D. barberi and D. virgifera were captured on sticky traps in all loca- tions in the prairie, but abundance was much greater near the edge adjacent to corn. We observed D, barberi (but not D. virgifera) feeding extensively on sunflower pollen and occasionally on other flower parts, such as petak. Sunflowers located nearer corn fields sustained morefloral damage than those farther from om. To de- termine the effect of beetle &mage on seed set, we enclosed sunflower heads in bags with either zero, two, or four D. barberi adults. Seed set was reduced in heads enclosed with D. barberi Thus, this agriculturalpest may interfere with the successful reproduction of sunflowers and possibly otherprairie composites thatflower in late summer. Given the small size of mostprairie remnants and the abundance of thisflowerrfeeding beetle in land- scapes dominated by corn agriculture, D. barberi may affect the sustainability ofprairie plantpopulations.

Recent null models that place species ranges randomly within a bounded domain have produced controversial results. Many such geometric constraint models predict a peak in species richness in the centre of domains in the absence of underlying environmental gradients or interspecific interactions. We used two-dimensional simulation models to explore different ways that species ranges could interact with the domain boundary. In the rejection model, a randomly generated range that overlaps a domain boundary is removed from the simulation. In the reshaping model, a range that overlaps the domain boundary is reshaped so that the entire range is placed within the domain. The truncation model allows potential ranges to extend across the boundary, but only that portion of the range within the domain is included in the realized range. Both rejection and reshaping models produced a drop in species richness near domain boundaries, though the effect was less pronounced in the reshaping model. Our truncation model did not produce any spatial pattern in species richness. Thus the random placement of species ranges within a bounded domain does not necessarily lead to a mid-domain effect. Range truncation is consistent with bioclimate envelope models, which can successfully predict a species range in response to the availability of appropriate climate conditions. We argue that such flexible range sizes are more realistic than the assumption that range size is an unvarying characteristic of a species. Other range characteristics, including size and shape, can change near domain boundaries in the null models, including the truncation model. A broader consideration of range characteristics near domain boundaries could be productive.

Changes in the plant community and ecosystem properties that follow the conversion of agriculture to restored tallgrass prairies are poorly understood. Beginning in 1995, we established a species-rich, restored prairie chronosequence where ∼3 ha of agricultural land have been converted to tallgrass prairie each year. Our goals were to examine differences in ecosystem properties between these restored prairies and adjacent agricultural fields and to determine changes in, and potential interactions between, the plant community and ecosystem properties that occur over time in the restored prairies. During the summers of 2000–2002, we examined species cover, soil C and N, potential net C and N mineralization, litter mass, soil texture, and bulk density across the 6- to 8-year-old prairie chronosequence and adjacent agricultural fields in southern Minnesota. We also established experimentally fertilized, watered, and control plots in the prairie chronosequence to examine the degree of nitrogen limitation on aboveground and belowground net primary production (ANPP and BNPP). Large shifts in functional diversity occurred within three growing seasons. First-year prairies were dominated by annuals and biennials. By the second growing season, perennial native composites had become dominant, followed by a significant shift to warm-season C4 grasses in prairies ≥3 yr old. Ecosystem properties that changed with the rise of C4 grasses included increased BNPP, litter mass, and C mineralization rates and decreased N mineralization rates. ANPP increased significantly with N fertilization but did not vary between young and old prairies with dramatically different plant community composition. Total soil C and N were not significantly different between prairie and agricultural soils in the depths examined (0–10, 10–20, 20–35, 35–50, 50–65 cm). We compared the results from our species-rich prairie restoration to published data on ecosystem function in other restored grasslands, such as Conservation Reserve Program (CRP) and old-field successional sites. Results suggest that rapid changes in functional diversity can have large impacts on ecosystem-level properties, causing community- and system-level dynamics in species-rich prairie restorations to converge with those from low-diversity managed grasslands.

In some leaf-cutting ant species, minim workers ride on the fragments of leaves as they are carried back to the nest from the cutting site. There is convincing evidence that these ‘‘hitchhikers’’ can protect the leaf carriers from attack by phorid (Diptera: Phoridae) parasitoids, but we consider the possibility of other functions for the hitchhiking behavior. It has been hypothesized that the hitchhikers (1) feed on leaf sap from the edges of the cut leaves; (2) ride back to the nest to save energy; (3) get caught on the fragments as they are cut, and hitchhike because they cannot (or will not) get off; and (4) begin the process of preparing the leaf to enter the fungal gardens in the nest, perhaps by removing microbial contaminants. We observed hitchhikers of Atta cephalotes in 14 nests at the La Selva Biological Station in Costa Rica. There was no difference in the proportion of leaf carriers with hitchhikers between day and night. Because the nests we observed were largely nocturnal, more than 90 percent of the hitchhiking occurred at night. The phorid parasitoids are usually considered to be diurnal, so the preponderance of nocturnal hitchhiking suggests other functions in addition to parasitoid defense. Hitchhikers spent more time in the defensive head-up posture during the day, but spent more time in the head-down posture at night. The head-down posture may indicate cleaning or other leaf preparation. The hitchhikers were never observed feeding on sap. Hitchhikers frequently got onto and off of the fragments, and so they were not ‘‘marooned.’’ Few hitchhikers rode all the way back to the nest and were often moving on the leaf fragment; these observations make the energy conservation hypothesis less likely, although we cannot reject it. We conclude that parasitoid defense is an important function of hitchhiking but also that there are probably other functions when parasitoids are absent. Based on available data, the most likely possibility is preparation of the leaf fragment before it enters the nest.

Inflorescences of New Zealand Chionochloa species are attacked by at least three insects, two flies and a moth. There has been disagreement about the identity of various life stages of these insects. We followed the seasonal pattern of occurrence of the two fly species that fed in Chionochloa pallens inflorescences in a population on Mt Hutt, Canterbury. Eggs and larvae of Diplotoxa similis (Diptera: Chloropidae) appeared in the inflorescences as soon as they emerged. The larvae are principally flower feeders, and most D. similis individuals had pupated by the end of the flowering period. Diplotoxa similis adults emerged from the puparia at the end of the season, and probably overwintered as adults. The second fly was an undescribed cecidomyiid (Diptera: Cecidomyiidae). Eggs of the cecidomyiid are laid into the C. pallens florets at the time of flowering, and hatch into mobile, translucent, early‐instar larvae. Late‐instar larvae were less mobile and opaque orange, and probably dropped from the i...

Proctolaelaps kirmsei (Mesostigmata: Ascidae) is a specialist flower mite that feeds on pollen and nectar in flowers of Hamelia patens (Rubiaceae). We offered mites a choice between pairs of manipulated H. patens flowers in petri dishes in the laboratory, then counted the number of mites inside the flowers after 1 hour. Relative to flowers with stamens and nectaries removed, there were more mites in flowers with only stamens, with only nectaries, and with both stamens and nectaries. When offered a choice between flowers with stamens but no nectaries versus flowers with nectaries but no stamens, the mites did not show a preference. Mites preferred H. patens flowers with H. patens stamens over those with the stamens of a coexisting non-host plant (Heliconia pogonantha; Heliconiaceae) inserted, so the identity of the pollen is important in flower choice. These results and previous work on the importance of nectar show that the presence of both pollen and nectar can be used by P. kirmsei to discrimin...

Global surface temperatures are expected to increase by several degrees in the next century, with potentially large but poorly understood impacts on ecological interactions. Here we propose potential effects of increased temperatures on ecologically dominant New Zealand grasses (Chionochloa spp.) that mass flower and mast seed. Twenty-two years’ data from five masting Chionochloa species in New Zealand showed that the cue for heavy flowering was unusually high temperature in the summer of the year before flowering. Attack by predispersal insect seed predators was much reduced in mast years, apparently because predator populations were satiated. Increased temperatures would greatly decrease interannual variation in Chionochloa flowering, allowing seed predator populations to increase and potentially to devastate the seed crop annually. Similar responses are likely in masting species worldwide. This previously unrecognized effect of global warming could have widespread impacts on temperate ecosystems.

It has been proposed that some wind-pollinated plants have the necessary conditions for an optimal sex allocation that is male biased, though there are few data that address this prediction. We determined that two prairie grass species ( Andropogon gerardii and Sorghastrum nutans) had reproductive characteristics that theoretically would result in a male-biased allocation: both species were self-incompatible and neither species had increased seed set after supplemental hand pollination. The relative allocation to pollen and seed production was measured in terms of biomass, energy, nitrogen, phosphorus, potassium, magnesium, and calcium. Sex allocation in A. gerardii was significantly male biased (from 60 to 89% male) when measured in currencies of biomass, energy, potassium, and calcium; there was no significant bias in the sex allocation (from 49 to 57% male) when measured in currencies of nitrogen, phosphorus, and magnesium. Sex allocation in S. nutans was significantly male biased (from 69 to 81% male) for all currencies except phosphorus (61% male). This is the first evidence for male-biased sex allocation in any plant or animal hermaphrodite. Though the necessary conditions may be uncommon, male-biased allocation may be found in other species with similar reproductive biology.

Nests of leaf-cutting ants (Hymenoptera: Formicidae: Attini) are abundant disturbances in Neotropical rain forests, and could affect the plant community both while the nests are active and after they are abandoned. We measured the diversity and abundance of understorey plants (>1 m in height) in the area around active and abandoned nests of leaf-cutting ants (Atta cephalotes) at the La Selva Biological Station in Costa Rica. Sampel quadrats on active nests had reduced diversity (number of morphospecies) and abundance of both small (height >10 cm) and large (10 cm–1 m) understorey plants, when compared to the nearby forest floor (3 and 13 m from the nest edge). Abandoned nests had greater diversity and marginally greater abundance of small understorey plants relative to nearby forest; there was no difference in diversity or abundance of large understorey plants. Leaf-cutting ant nests create gaps in the plant understorey when active, but serve as centres of recruitment for small plants after they are abandoned. Thus, like canopy gaps, ant nests could play an important role in recruitment of new individuals and maintenance of plant species diversity in tropical forests.

In Chionochloa pallens and C. macra (Poaceae), pollen tube growth on excised styles planted on agar showed that self-pollen was as viable as cross-pollen. While selfing is therefore possible, other evidence (including pollen-ovule ratios) suggests it is not the rule in Chionochloa.

Levels of seed predation were recorded between 1986 and 1992 in Chionochloa pattens, C. oreophila, C. flavescens, C. macra, and C. crassiuscula populations at various altitudes at three sites in Canterbury. There were consistently high levels of attack by a chloropid fly, Diplotoxa similis: between 22 and 94% of florets showed evidence of fly damage, except at higher altitudes (decreasing to 2% in C. pattens at 1620 m) and in C. oreophila (6 – 10%). Since D. similis emerges rapidly from the pupa and since larvae possibly move between florets, the number of insects found underestimated the number of florets predated by up to 90%. Seed set varied from 2% to 59% and was negatively correlated with seed predation. Two other Chionochloa seed predators have been reported previously: a moth, Megacraspedus calamogonus, and an unidentified cecidomyiid (gall midge) larva. In this study, larvae similar to Megacraspedus were very rare, and we suggest that the unidentified cecidomyiid is actually a misidentifi...

SUMMARY Intraspecific variation in pollen production and pollen size were studied within a population of Chionochloa pallens, an alpine snow-tussock grass of New Zealand. I sampled plants from two sites at the altitudinal extremes (1620 and 1070 m) of the distribution of C. pallens on Mt. Hutt, South Island. There was a highly skewed distribution of pollen production among plants; the largest 10% of plants produced over a third of the pollen. In contrast to some other species, pollen production in C. pallens can be accurately estimated by simple measures such as tussock area and number of flowering shoots. Other components of yield (number of spikelets per shoot, number of florets per spikelet, and average anther length) varied significantly among individuals, but these characteristics had a small impact on total pollen yield. There was continuous variation in pollen size among individuals; between the plants with the largest and the smallest pollen grains, there was a 21 % difference in pollen diameter and a 76% difference in pollen volume. Interplant differences in pollen size have been found in a number of plants, although the biological significance of this variation is unknown. There were few differences in pollen characteristics between sites. Plants at the low-altitude site had slightly greater pollen yield because total plant size was greater.

It is well known that some species produce offspring carrying only female chromosomes by processes such as apomixis and parthenogenesis (generically termed "gynogenesis"). There are also several cases of natural reproduction by androgenesis in which diploid offspring carry nuclear chromosomes from only the male parent. We used population genetics models to investigate the conditions for invasion of rare androgenesis alleles and the consequences of their spread. Our models predict that androgenesis alleles often spread to fixation. If fixation causes the loss of females or female function in the population, population extinction occurs. Therefore, androgenesis alleles represent a new class of selfish genetic elements. Extinction is more likely in dioecious species than in hermaphrodites. Within dioecious species, extinction is more likely when androgenesis occurs via paternal apomixis (vs. fusion or doubling of haploid nuclei) and when females are the heterogametic sex (vs. male heterogamety). The apparent rarity of androgenesis compared to gynogenesis could be because androgenesis is harder to detect and more often leads to population extinction. Also, there could be greater evolutionary constraints on the origin of mutations for androgenesis. We suggest characteristics of groups in which further cases of androgenesis are more likely to be found.

Many of the venomous New World coral snakes (Micrurus and Micruroides) have a distinctive pattern of red, black, and yellow rings (Campbell and Lamar 1989; Savage and Slowinski 1992), which typically appear in the sequence red-yellow-black-yellow (the "tricolor monad" of Savage and Slowinski 1992) repeated multiple times on each snake. Relatively harmless coexisting snakes in several different genera have a similar appearance, and most recent investigations have concluded that these are cases of Batesian mimicry (Greene and McDiarmid 1981; Pough 1988a; Campbell and Lamar 1989; Savage and Slowinski 1992). Though it is clear that a precise mimic of a dangerous or unpalatable model often gains protection from predation (see Waldbauer 1988 for a review), the gradual evolution of mimicry requires that partial mimics gain some fitness benefit from even a poor resemblance to model species (Fisher 1958; Sheppard 1959). There is evidence that partial mimics gain limited protection from predation in some insect systems (e.g., Morrell and Turner 1970; Pilecki and O'Donald 1971; Shideler 1973). A large number of Neotropical snakes have some elements of the coral snake pattern (Pough 1988b), but there is very little information on the protective effects of partial mimics, largely due to the extreme difficulty of observing predation events in the field. Brodie (1993, following Madsen 1987) pioneered the use of plasticine replicas of coral snakes to gather data on rates of predation by free-ranging birds in the natural habitat of coral snakes. The soft plasticine retains the imprint of any attempted predation, which can be used to identify the predator as bird, mammal, etc. (Brodie 1993). Using this method, Brodie (1993) showed for the first time that the coral snake pattern reduces the rate of avian predation for replicas of both true coral snakes and coral snake mimics. Here we extend Brodie's (1993) method to determine bird attack rates on partial coral snake mimics that have color and pattern combinations not found in any living snake. The coloration of coral snakes includes a number of elements that can vary independently: ring color, ring width, and order of the arrangement of the rings. There is no historical infor

Darwin pointed out that plants with vertical inflorescences are likely to be outcrossed if the inflorescence is acropetalous (flowers from the bottom up), the flowers are protandrous (pollen is dispersed before stigmas are receptive), and pollinators move upward on the inflorescence. This syndrome is common in species pollinated by bees and flies, and very few exceptions are known. We investigated flowering phenology and pollinator behavior in Besseya bullii (Scrophulariaceae) and found that it did not fit Darwin's syndrome. The vertical inflorescence was acropetalous but the flowers were distinctly protogynous, so flowers with newly receptive stigmas appeared on the inflorescence above those with dehiscing anthers. A number of small insects visited B. bullii; bees in the family Halictidae (Augochlorella striata and Dialictus spp.) were most common. When insects moved between gender phases within inflorescences, they moved up more often than down (61% versus 39% of observations, respectively) but this difference was only marginally significant. Most visits were to male-phase flowers only, and this preference was more pronounced for pollen-foraging insects than for nectar-foraging insects. B. bullii was self-compatible, so its flowering characteristics potentially could result in considerable self-pollination. However, an average of 38% of the lowermost flowers opened before any pollen was available on the same inflorescence; these “solo females” had a high probability of outcrossing (though fruit set was relatively low in the bottom portion of the inflorescence). Upper flowers may also be outcrossed because downward insect movement was not uncommon. Therefore protogyny in B. bullii may not necessarily lead to more selfing than would protandry.

In our paper on experiments conducted at multiple sites, we suggested that there are two options for statistical analysis (McKone and Lively 1993). One option is the standard ANOVA where sites are crossed with treatment, and the F-ratio for testing treatment effects is calculated using the treatment-by-site interaction. This analysis is obviously preferable when the number of sites is large, because it has the advantage of being able to test for treatment effects averaged over all sites. However, when the number of sites is small, there is very little power for testing for a treatment effect in this way. We therefore recommended that treatment be analyzed as nested within site when there are few sites. Nesting allows for the assessment of treatment effects within individual sites, but has the drawback that the effect of treatment cannot be generalized to the population of sites. Greenwood (1994) thinks that our recommendation is mistaken, precisely because one cannot generalize using the nested approach. But he does not address what we see as the real problem. We agree unconditionally that it would always be preferable to generalize. Our main point is that it will often be counterproductive to attempt generalization from a handful of sites. The question here is not which is better in a perfect world with a large random sample of sites. The answer to that is clear. The question is: what should we do when only a few sites are manageable, due to unavoidable constraints? We see three options: (1) recognize the futility of generalizing from these few sites, and simply not conduct the experiment; (2) analyze the data using a factorial ANOVA, and hope that the interaction term is not significant and therefore possibly combinable with the error term (Sokal and Rohlf 1981), or (3) use the nested approach to analyze site effects and treatment effects within sites. Nothing is learned from the first approach; the second approach lacks statistical power, and has a high risk of type II error. The third (nested) approach is not generalizable, but something is learned. In our example of the nested method, herbivores had significant effects at all three sites. This is an important finding for these sites, whether or not the result is later shown to be general. We stick by our recommendation.

Ecological phenomena are often spatially variable, and hence it is desirable to simultaneously conduct experiments at multiple sites. However, the number of sites is often constrained to be few, resulting in many more replicates within sites than there are sites. Such experiments are commonly analyzed by mixed model ANOVAs (where site is the random effect), which are very powerful at detecting differences among sites and siteby-treatment interactions. But, since few sites are sampled, there is little statistical power to make inferences concerning the treatment effects themselves. This lack of power is unfortunate whenever the primary reason for conducting the experiment is to test for differences among treatments, as is true for most ecological field experiments. In this note we suggest that, when there are few sites, the treatment effect should be examined only within sites using a nested ANOVA.

Five species of Bromus were tested for self-fertility when bagged. Bromus inermis was selfincompatible. Bromus tectorum, B. kalmEi, B. ciliatus, and B. Iatiglumis were capable of selffertilization. Further evidence from flowering behavior suggested that B. tectorum almost always self-pollinated and that the remaining self-fertile species sometimes outcrossed. Florets in different positions on a spikelet varied in the proportion that set seed. Open-pollinated B. inermis plants had lower seed set than the self-fertile species, in agreement with the pattern in a variety of hermaphroditic plants. THE RATE of self-pollination can have a profound effiect on the reproductive and population biology of hermaphroditic plants. In comparison with closely related outcrossing taxa plants that have high rates of self-fertilization can be expected to have lower rates of gene flow via pollen, smaller neighborhood size more homozygosity, and greater genetic differentiation and ecotypic specialization among populations (Stebbins, 1957; Allard, Jain and Workman, 1968; Schoen, 1982; Layton and Ganders,1984). Ornduff(1969) has listed many other correlates of breeding system. In addition, Sutherland and Delph (1984) have recently shown a relationship between the breeding system and the rate of seed set in many difEerent plant species. To test the generality of these patterns and as part of a study of the reproductive biology of several species of bromegrass (Bromus L.) I investigated the breeding system and pattern of seed set in five congeneric species that show wide variation in breeding system. Like most grasses, species of the genus Bromus are wind pollinated. The flower is much reduced and, with the two subtending bracts is referred to as a floret; florets are grouped together on the plant into spikelets. Each floret

Variation in pollen production was measured within five hermaphrodite species of bromegrass (Bromus). Anther length is an excellent predictor of pollen production in this genus (R2 = 0.97). Anther length varied considerably within each of the species, both among and within individual plants. Within plants, most of the variation occurred among florets within spikelets; florets in upper spikelet positions were smaller and produced less pollen. In B. inermis, pollen production was decreased by defoliation and increased in shoots that grew on thatching ant (Formica obscuripes) mounds. Whole-shoot pollen yield was determined by spikelet number, number of florets per spikelet, and pollen production per floret. All of these yield components must be considered in attempts to estimate pollen production accurately. IN CONTRAST to the voluminous literature on intraspecific variation in fruit and seed yield, relatively little is known about variation in pollen production (Bertin, 1988). Yet pollen yield can have as much impact on fitness as seed yield, and recent empirical and theoretical studies have emphasized the importance of paternal fitness in the evolution of attractive structures in angiosperms (e.g., Bell, 1985; Charnov and Bull, 1986; Stanton, Snow, and Handel, 1986). As part of a study of the reproductive biology of bromegrass (Poaceae: Bromus L.), I measured intraspecific variation in pollen production in five Bromus species. Lloyd (1980) suggested that a plant's maternal allocation is affected by a series of adjustments through the flowering season; total seed production could be modified by changes in flower number, proportion of pistillate flowers, number of ovules per flower, fruit and seed set, and seed weight. Similarly, serial adjustment of a plant's pollen Received for publication 28 December 1987; revision accepted 16 June 1988. E. J. Cushing, L. F. Delph, D. G. Lloyd, P. A. Morrow, and C. J. Webb made helpful comments on various versions of the manuscript. Financial assistance was provided by several bodies at the University of Minnesota, including the Dayton-Wilkie Fund, the Field Biology Program, the Graduate School (which awarded a Doctoral Dissertation Fellowship and a Doctoral Dissertation Special Grant), and the Department of Ecology and Behavioral Biology. This work was also partially supported by NSF Grant DEB8114302-A02 to D. Tilman and J. Tester for Long-Term Ecological Research at Cedar Creek Natural History Area. Partial support while preparing the manuscript was provided by NSF Grant INT-8603569 and by a grant from the Miss E. L. Hellaby Indigenous Grasslands Research Trust of Dunedin, New Zealand. 2 Current address: Department of Biology, Carleton College, Northfield, MN 55057. production is possible by means of changes in flower number, proportion of staminate flowers, number of stamens per flower, and number of pollen grains per stamen. Bromus flowers are hermaphrodite and always contain three stamens; thus, the remaining options for adjustment of pollen yield are flower number and pollen grains per stamen. I measured these parameters in field populations of five species, and calculated their relative importance in determining pollen yield. MATERIALS AND METHODS-Bromus tectorum L. plants were collected in 1982 from a disturbed site near the Botany Greenhouse of the University of Minnesota, Minneapolis. Bromus inermis Leyss., B. kalmii Gray, and B. ciliatus L. were collected in 1983 from populations in the Cedar Creek Natural History Area, about 50 km north of Minneapolis. The B. latiglumis Shear (Hitchc.) population was sampled in 1982 in Fort Snelling State Park, St. Paul. Voucher specimens of the five species have been placed in the University of Minnesota Herbarium, St. Paul. The five Bromus species differ in a number of reproductive characteristics, including average anther length, pollen size, and sex allocation (McKone, 1985, 1987). The species span a broad range of outcrossing rates (McKone, 1985), from obligate outcrossing in B. inermis, through partial selfing in B. kalmii, B. ciliatus, and B. latiglumis, to nearly complete selfing in B. tectorum. Bromus tectorum is an annual; the other four species are perennial. Pollen production was estimated for individual flowering shoots (culms). I will refer to shoots as plants; except for B. inermis, the species studied are not rhizomatous, and only

The roots of seven species of goldenrod (Asteraceae: Solidago ) were assayed for associative nitrogen fixation. We used the acetylene reduction method to measure N 2 fixation rates in sampling jars that contained roots of sampled plants and soil. The rate of C 2 H 2 reduction in jars with S. rigida (12.9 ± 1.2 nmol h −1 , mean ± SE) and S. canadensis (6.0 ± 2.0 nmol h −1 ) was significantly higher than in soil-only control jars. There was measurable C 2 H 2 reduction in jars of the other five species, but the rates were not significantly different from controls. The rates of C 2 H 2 reduction per root dry weight and per total plant dry weight also were higher for S. rigida and S. canadensis than for the other species.

We measured variation in gender among individuals within populations of ragweed (Ambrosia artemisiifolia) in an abandoned old-field and in the greenhouse. There was great variability in sex expression, from all-female to approximately 78% male. Plants differed significantly in gender in different locations within the field. Plants in an area abandoned from agriculture one year previously were more male than plants in a nearby area abandoned four years previously. In the greenhouse, soil moisture treatments and levels of attack by spittlebugs (Philaenus spumarius) did not affect gender. Height was positively correlated with relative maleness in both populations. Plants with greater shoot weight were relatively more male in the greenhouse, but not in the field. The gender variation we observed either has a genetic basis or is controlled by environmental variables other than those we investigated.

The size of pollen grains from male and hermaphrodite flowers was compared in two species, Lignocarpa diversifolia (Cheeseman) J . Wyndham Dawson and Gingidia harveyana (F. Muell.) J . Wyndham Dawson (Apiaceae). In L. diversifolia, pollen grains from hermaphrodite flowers were on average 20% greater in volume than the grains from male flowers; in G. harveyana, pollen grains from hermaphrodite flowers were on average 13% greater in volume than grains from male flowers. There was also considerable variation in pollen size among flowers of each type, both within and among plants of both species. Conditions during pollen development could be important; the early arrest of development of the gynoecium in male flowers could have a general effect on the flower and result in smaller pollen. Alternatively, the size difference could be adaptive if pollen from hermaphrodite flowers fertilises flowers with longer styles than does the pollen from male flowers. The size difference also could result from selection for an increased number of pollen grains in male flowers if there is a trade off between pollen size and number.

Predictions of sex-allocation theory were tested by comparisons among hermaphroditic bromegrass (Bromus) species that differed in outcrossing rate. Relative maternal and paternal investment were calculated using both the ratio of pollen to seed production, and absolute allocations in units of energy, nitrogen, phosphorus, potassium, magnesium, and calcium. Outcrossing rate had a large effect on sex allocation; species having greater outcrossing rates had relatively more paternal reproductive effort. Bromus inermis was obligately outcrossing, and nearly half of its reproductive effort was devoted to pollen production. Three partially outcrossed species, B. kalmii, B. ciliatus, and B. latiglumis, invested between 5% and 11% of reproductive effort in pollen production. Paternal investment was less than 2% in the selling species B. tectorum. Estimates of sex allocation were relatively unaffected by the resource currency used in calculation. The differences among species in sex allocation were mostly due to differences in anther size and seed set.>.