RHODORA, Vol. 103, No. 913, 96-1 2001 pp. 16, AND DEMOGRAPHY, REPRODUCTIVE CYCLE, LIFE BIOLOGY OF HERB ROBERT {GERANIUM ROBERTIANUM) ROBKRT BERTIN I. MA Holy Cross College, Worcester, 01610 e-mail: [email protected] abstract. For nine years studied the cycle, demography, and life re- I productive biology of a woodland herb. Geranium robertianum, in central These Massachusetts. plants are facultative biennials that typically flower in the year after germination. Of 408 seeds experimentally sown natural hab- in 26% gave seedlings within Of 1352 marked, itat, rise to six years. naturally 53% occurring seedlings, survived to the end of their growing season, first 13% of plants alive at the end of their first growing season survived to year 54% two, and of these flowered. Plants germinating early spring were more in likely to survive to the end of year one and to flower than those germinating Low summer summer later in the year. rainfall appeared to reduce survival and low winter temperatures combined with snow cover seemed little to re- duce winter survival. Floral structure and development, pollen-ovule ratios, and pollinator activity all suggest that plants of this species are highly self- pollinated. Herb Robert is one of only a few biennials that occupy undisturbed forest habitat. Key Words: demography, Geranium biennial, facultative biennial. rober- tianum, herb Robert, cycle, history, self-pollination life life Biennials are rare, comprising only 1.4% of the 14,500 species in the provisional checklist of the Flora of North America (Hart 1977). Plants with this life cycle pattern often are said to be associated with disturbed habitats (Kelly 1985; Silvertown 1984), although few a biennial herbs of relatively undisturbed habitats, known such (Morgan as forests, are 1971). Numerous hypotheses have been proposed regarding the adap- tive significance of bienniality (Bender and Baskin 1994; Hart Meidjen 1977; et al. 1992; Schaffer and Gadgil 1975; Silvertown 1984), but our current knowledge of biennial species insuffi- is cient to decide which, if any, of these hypotheses correct. Given is this state of knowledge, examination of the natural history of additional biennial species desirable. is Here present an ecological investigation of a putative bien- I Geranium robertianum em- nial, L. (herb Robert, Geraniaceae), phasizing and two the species' cycle areas potentially relevant life 96 — Herb Robert {Geranium robertianum) 97 Bertin 2001] demography and to interpreting the life cycle: reproductive bi- My ology. specific objectives were to determine what cycle 1) life pattern this species possesses; 2) understand the timing and caus- es of mortality; 3) investigate the relationship between germina- tion time, vegetative growth, and flowering; and 4) describe the flowering phenology and breeding system. Geranium robertianum a forest herb that has been referred is to as a biennial, but also as an annual, a winter annual, and a monocarpic or polycarpic perennial (Baker 1956; Falinska and Piroznikow 1983 and references therein; Fernald 1950; Gleason Yeo Yeo and Cronquist 1991; Widler-Kiefer and 1987; 1973). Even within North America, the species has been described as an annual (Fernald 1950), an annual or winter annual (Great Plains Flora Association 1986), and an annual or biennial (Gleason and Cronquist 1991). Herb Robert occurs northeastern North America, Eurasia, in Yeo and northern Africa (Baker 1957; Widler-Kiefer and 1987). The North American range extends from Newfoundland Mary- to The land, west to Illinois (Fernald 1950). species has been intro- New duced western North America, Zealand, and Singapore to (Baker 1957). Herb Robert typically occurs shady or partly shaded habitats, in and often associated with rocky ground and nutrient-rich or is basic (Baker 1956; Falinska and Piroznikow 1983; Voss soils 1985). In North America, the habitat of herb Robert described is "damp woods" (Gleason and Cronquist "rocky as rich 1991), woods, ravines and gravelly shores" (Fernald 1950), and "rich woods deciduous along roads and rocky clearings, trails . . . . . . New openings, gravelly shores and rubble" (Voss 1985). In Eng- Seymour land the species listed as "occasional" by (1969). is It absent from several Massachusetts coastal counties and rare is is Rhode Somers in Island (George 1996; Sorrie and 1999). In cen- and Massachusetts, the species occurs sporadically restrict- tral is wooded ed largely to rocky slopes with soils that are seemingly richer and/or less acid than the norm. AND METHODS MATERIALS Field work. conducted work two Princeton, field at sites in I Massachusetts (42°28'N, 71°54'W). Herb Robert plants both at among much grew boulders rocky woodland. Because of sites in 98 Rhodora [Vol. 103 the ground surface was occupied by boulders, plots were placed randomly arbitrarily rather than to avoid including large areas of unvegetated rock. m One study was an elevation of 350 on a south-facing site at Meadow slope in Wachusett Wildlife Sanctuary of the Massachu- Audubon Dominant setts Society. trees at this site were shagbark hickory \Carya ovata (Miller) K. Koch], white ash (Fraxinus americana and sugar maple (Acer saccharum Marshall), with L.), common hornbeam [Ostrya virginiana (Miller) K. Koch] the in Two were on understory. plots arbitrarily selected this slope in A m the spring of 1990. Plot occupied 0.72 2 with vegetation little m B other than herb Robert present. Plot was 12 from plot A, m comprised 1.82 2 and contained a fern [Dryopteris marginalis , (L.) A. Gray], Virginia creeper [Parthenocissus quinquefolia (L.) Planchon] and the goldenrod Solidago caesia L. km The second was from Wachusett 2.7 the site first site, in Mountain 430 The dom- State Reservation, an elevation of m. at woody inant plants on ESE-facing slope were white this ash, sugar maple, shagbark hickory, hornbeam, red elderberry (Sam- bucus racemosa and dogwood (Cornus alternate-leaved L.), al- m was Plot C, established 1990, 0.82 2 and ternifolia L.f.). in contained Jack-in-the-pulpit [Arisaema triphyllum (L.) Schott.], enchanter's nightshade (Circaea lutetiana and fringed bind- L.), A weed (Polygonum cilinode Michx.). second plot (D) was es- number tablished at this site in 1995 to increase the of plants being monitored. included vegetation besides herb Robert. It little May were Plots visited starting in late April or early in each year between 1990 (1995 for plot D) and 1998. The plots were weekly June and two- three-week visited until late at to intervals September thereafter until late or early October. also visited the I plots four times in 1999 to determine the fate of the 1998 cohort. Newly germinated plants were recognized by cotyledons and their Each was small monitored throughout following size. its life marking with a numbered plastic stake secured to the plant with During number a plastic twist each recorded the of tie. visit I leaves on each plant as an indicator of plant For reproductive size. plants, recorded the numbers of flowers and fruits and used the I maximum sum recorded of flowers and fruits as a conservative estimate of total flower production for each plant. Substrate depth was measured for each plant the time of death by pressing at its cm marked down a plastic knitting needle in increments into 1 — Herb Robert (Geranium robertianum) 99 Bertin 2001| humus the the base of the plant until reached a rock or a at it The depth of 12 cm. length of the concealed portion of the needle was taken as the substrate depth. To examine the demography of dispersed seeds, sowed seeds I cm Meadow twelve 60 diameter Wachusett 1993. into plots at in A Plots were chosen on the same rocky hillside as plots and B, was marked numbered and the center of each with a plastic stake. m Each demography was from seed 2.0 the nearest plot least at herb Robert plant to minimize the possible presence of any nat- was urally occurring herb Robert seeds, but in habitat that ap- peared to be appropriate for the species. Seeds were collected Meadow sown from growing Wachusett and naturally plants at Each on on received 10 seeds July 7 July in the plots. plot 14, on 4 on and 4 on August of 9 July July 25, for a total 16, 19, 8, 34 The numbers seeds. variation these reflects the natural var- in summer. Seeds were on seed supply during the scattered iation in would the surface of the leaf as they be during natural litter, cm The were dispersal, within 30 of the center of the plot. plots week monitored for seedlings 1-3 intervals during the remain- at 1994-1998 der of the 1993 growing season and in each of the The between growing seasons, and four times during 1999. area cm cm 30 and 150 from the plot center also was monitored during these visits to detect any additional seedlings that might suggest bank the presence of a naturally occurring seed in or near these plots. made development and observations of pollinator field floral I activity using 38 marked flowers on a total of twelve plants near A B Meadow. plots and in early June, 1993 at Wachusett col- I lected one flower from each of five plants at each site to deter- mine pollen/ovule ratios (see below). was from two Information on weather conditions obtained lo- Temperature and records were obtained cations. precipitation from weather Worcester This the station the airport. station at is m km 300 above sea level and 22 S of the study sites. obtained I daily records of winter snow cover from this station until October, when 1995 weather ceased recording information. the station this From October, 995 through April, 999 snow cover records were 1 1 recorded residence Paxton, Massachusetts, 15 the author's in at SSW km of the study an elevation of 318 m. sites, at Massachu- Laboratory studies. Seeds from the Princeton, 100 Rhodora [Vol. 103 setts populations were unavailable when laboratory investigations were conducted; instead used eight plants established from seeds I cm collected Hinesburg, Vermont. were grown in Plants 8 in square pots containing Metromix 350, a soil-less growth medium. were Plants overwintered greenhouse temperature in a at a just above freezing, and were then moved growth chamber to a for The further study. temperature the chamber was 25°C during in each 14 day and 15°C during each 10 hr. night. hr. The of flowers was examined ability to self-pollinate in a growth chamber in the absence of insects. used four experi- I mental treatments: hand hand 1) self-pollination, 2) cross-polli- nation, 3) no pollination, and 4) emasculation and no pollination. Treatments were assigned randomly within two or three blocks of four flowers on each of eight plants. The stigmas of hand- pollinated flowers received large loads of pollen from toothpick a on all five stigmas. Emasculation involved removing anthers with Any fine forceps shortly after the flowers opened. pollen inad- vertently deposited on stigmas was removed by wiping with moist cotton under a dissecting microscope. Seeds were counted at fruit maturation. One flower from each of four Vermont was plants collected and, in addition to the five flowers from each Massachusetts site, used determine The number pollen/ovule to of ovules ratios. in each flower was counted under One a dissecting microscope. an- ther from each flower was emptied into a small Petri dish con- 70% taining alcohol and the pollen grains, which are large in this mm mm X species, were counted over a 2 2 grid using a dis- secting microscope. Data Sample analysis* from were sizes individual field plots too small most lumped for analyses, so data for four I all plots. lumping This violates the assumption of independence of obser- and shortcoming vations, this should be considered in interpreting However, were results. general trends usually similar in the four C plots. For example, correlations between plots A, B, and in numbers of seedlings over nine years were significantly pos- all > < P No D between itive (r 0.67, 0.05). correlations plot and were the other plots significant, though these were not very tests powerful because had only four years of data for this plot. I used three kinds of analyses used for field data. G-tests to I I evaluate independence of variables for frequency data num- (e.g., — Herb Robert (Geranium robertianum) 101 200 Bertin 1 ] 400 - Seedlings 300 - D Second- and Third </) Year Plants ° 200 - n E 100 - 1990 1991 1992 1993 1994 1995 1996 1997 1998 Years Figure Numbers of seedlings and older (2nd- and 3rd-year) plants in 1. A-C plots in nine years. Second and third year plants represent survivors from earlier years. used ber of survivors and non-survivors in different years). I between product-moment correlations to evaluate associations amount and continuous variables, such as proportion surviving of compare used precipitation in different years. Finally, a t-test to I numbers produced by and late-emerging of flowers early- the seedlings. RESULTS A was marked Plant demography. of 1352 seedlings total was monitored between 1990 and 1998, and each seedling for its A-C, which were monitored Considering for entire just plots life. from numbers of emerging seedlings varied the entire study, the Numbers 37 1998 343 1990 (Figure of older (2nd- in to in 1). among from and 3rd-year) plants also varied dramatically years, C one plant for plots A, B, and combined in 1992 to 34 plants May, Emergence was concentrated espe- 1991 (Figure in in 1). low through October and June, but continued a level cially, at (Figure 2). 53% Of 1352 marked survived the end of the seedlings, until 102 Rhodora [Vol. 103 8OO-1 d) 600 - CO T3 <D O) E 400- 5 0) 200 2 - -Q E 3 May May < 16 16-31 June 1-15 June 16-30 -Aug. Jul. Sept. Oct. - Emergence Date Figure 9 Numbers of newly emerged seedlings at different times of the A-D year. Data are combined for plots and for years 990-1 998. 1 7% growing the first season and survived their winter and first produced one Of at least leaf the following year (Figure the 3). 51% 98 plants surviving their winter, flowered during first the 30% second and year subsequently died, died during second their 14% year without flowering, and died during the subsequent win- 5% The remaining did not flower during second ter. their year, Of but survived to their third year. these five individuals, three flowered and died during their third year and two died without flowering. In 53 of the 1352 marked individuals flowered all, at some time, 50 (94%) in their second year and three in their third year. Yearly differences in survivorship. Survival of seedlings through their first growing season and winter varied consid- first from erably year to year (Figure Survival of seedlings 4). to the = end of their year varied from 5.5% 1995 first (n 109) in to = 65.7% (n 201) in 1996. Overwinter survival ranged from 0.0% = = G 21.3% (n 59) in 1991 to (n 150) 1992. in tests for inde- pendence of year and survival revealed significant departures from independence both for seedling survival during the first = < growing season (G P 148.48, 8 df, 0.001) and the winter first — Herb Robert (Geranium robertianum) 103 Bertin 2001] 3.5 50 3 CO | 40 CL 2.5 </) o > F30 2 2: o CO LL 1.5 20 O o 1 0) 10 | 0.5 year spring year 2 spring year 3 initial fall 1 seedlings Time marked Figure Survival and flowering of 1352 plants as seedlings in 3. was A-D and 1990-1998. Survivorship for years 2 3 during years plots the No survived year measured the beginning of the year. plants to 4. at 0.7 0.6 c > 0.5 0.4 (/> C o 0.3 o a 0.2 - o a. 0.1 - 1990 1991 1992 1993 1994 1995 1996 1997 1998 Emergence Year of Figure 4. Yearly variation in survival of seedlings emerging in nine years. A-D proportion of combined. Overwinter survival the Data for plots are is of plants alive at the end of their first year that survived to spring the fol- lowing year. 4 Rhodora 1() [Vol. 103 = P < (G 44.46, 4 The second df, 0.001). analysis has fewer degrees of freedom than the because excludes first four years it of which data for the expected values were below 5.0. The yearly differences seem in survival in part to reflect pre- weather vailing conditions, especially precipitation during the summer, and temperature and snow depth during the winter. Sur- vival of seedlings to the end of their year was positively first > P correlated (though not significantly so, 0.05) with the amount May = = of precipitation in (r 0.46), June 0.57), and August (r = = (r 0.36), but not in July -0.15). Survival was, however, (r significantly correlated with the amount of May- precipitation in = and May-August P < July (r 0.68, 0.74, respectively, 0.05). The lowest rainfall for any of these months in the nine years of the study was June when cm in of 1995, only 3.8 of rain fell, 42% 94% of June Over the average. of seedlings died by end the of compared this year, to a nine-year average mortality of 52%. Overwinter survival of plants alive at the end of the grow- first was ing season positively correlated with winter snowfall total (r > > although P 0.52), not significantly so (0.20 The 0.10). winter with was greatest survival 1992-93, which had also a greater snowfall than any winter in the preceding century (3.1 m). exposure If to harsh winter conditions reduces survival, it would seem that a combination of low winter temperatures and minimal snow cover would more be highly correlated with winter snow mortality than cover alone. This suggested by the nearly is = > > significant negative correlation -0.64, 0.10 P (r 0.05) between and number survival the of winter days with both tem- peratures below -12°C cm (10°F) and than less 2.5 of in.) (1 snow The 0% cover. winter with the lowest survival (1991-92, survival) also had number the largest of such cold days with bare ground compared 1-7 (15, to other in years). The substantial variation in numbers of seedlings emerging in different years caused the density of seedlings vary to dramati- among = cally years and plots (range 1-191/m 2 This density ). was however, not, related to survival or flowering. Correlation between coefficients seedling density and survival end to the of the year, overwinter and first survival, proportion of overwinter = = survivors flowered were that 0.23 -0.05 (n and 31), (n 29) = 0.45 (n respectively. 15), Performance of seedlings. Emergence was date strongly as- — Bertin Herb Robert {Geranium robertianum) 105 2001] 7 -i 6 - CD f • 3 4 CD E 3 3 E x 2 - 1 < May 16 May 16-31 June 1-15 June 16-30 -Aug. Sept.-Oct. Jul. Emergence Date of maximum Figure 5. Average leaf numbers produced during their first year by plants emerging different times of the year. at maximum number) by sociated with the size (as represented leaf attained by year plants, with their likelihood of surviving to first the end of the growing season, and with the numbers of first emerging and flowers that they produced. Plants in late April the May maximum half of bore an average of 5.9 leaves during first maximum number their season, while the average for the first number was more other time intervals no than half this (Figure May 56% Plants emerging on or before 15 had a chance of 5). surviving to the last census date of the year, while the probability May was was emerging This for those after 15 0.36. latter figure inflated by the high survival o( seedlings that emerged in Sep- many them tember and October of (0.93), reflecting the fact that needed to survive for only 2-3 weeks to live to the end of the growing season. Plants surviving the end of the season had a slightly to first emerged on greater chance of surviving the winter they or if May (18% emerged before 15 than they later in the year vs. if G = P < However, emerging 1%; seedlings 4.46, 0.05). df, if 1 1 September and October are excluded, the survival of late- in emerging seedlings increases 14%, which not significantly to is 18% from (G different 1.20, df). 1