Journal of Chemical Ecology, VoI. 15, No. 5, 1989
CHEMICAL INHIBITION OF FIRE-PRONE GRASSES BY FIRE-SENSITIVE SHRUB, Conradina canescens
G.
BRUCE
DONALD
WILLIAMSON, R.
1 NIKOLAUS
H.
FISCHER, 2
3 and ANA
DE
LA
RICHARDSON,
PElqA 2
~Department of Botany 2Department of Chemistry, Louisiana State University Baton Rouge, Louisiana 708003 3Department of Biology, University of South Florida Tampa, Florida 33620 (Received February 16, 1988; accepted July 26, 1988)
A b s t r a c t - - I n an investigation of potential chemical activity of fire-sensitive shrubs in Florida's sand pine scrub community, bioassays of foliar washes of Conradina canescens showed significant inhibitory activity on three native grasses that are known to fuel frequent surface fires; inhibition was concentrated seasonally in spring and summer. Application of runoff from Conradina leaves to one of the grasses caused a 50% reduction in growth over a 20-week period. Isolation of the biologically active fractions from the fresh leaves of C. canescens yielded numerous monoterpenes, a number of which were identified from a GC-MS reference library and/or MS comparison to authentic compounds: 11 from the diethyl ether extract, 11 from steam distillation, and four from the foliar leaf wash. Numerous other monoterpenes present in the extractions were unknown. The terpenoid fraction completely inhibited seed germination of one of the native grasses and of lettuce. Saturated aqueous solutions of nine of the monoterpenes inhibited germination and radicle growth of two native grasses. SEM views of the leaf surfaces of Conradina reveal secretory trichomes that appear to be the source of the monoterpenes as well as the triterpene, ursolic acid. The biological activity of C. canescens as a fire-sensitive component of the scrub community is reviewed in light of the chemical evidence. Key Words--Allelopathy, fire, monoterpenes, trichomes, Conradina canes-
cens, Pinus, Schizachyrium scoparium.
1567 0098-0331/89/0500-1567506.00/0
9 I989 Plenum Publishing Corporation
1568
WILLIAMSON ET AL.
INTRODUCTION
Two different plant communities exist on the upland, well-drained sands of Florida: the sand pine scrub characterized by sand pine [Pinus clausa (Chapm. ex Engelm.) Vasey ex Sarg.] with a dense shrub cover but no herbaceous ground cover, and the sandhill or high pineland dominated by longleaf pine (P. palustris Mill.) with a complete graminoid ground cover but few shrubs (Chapman, 1932; Laessle, 1958; Garren, 1943). Fire is a frequent feature of the sandhill, burning through the deciduous surface fuels every three to eight years (Williamson and Black, 1981). However, healthy scrub, which is evergreen, burns infrequently, about once in 50 years or once per generation of the dominant sand pine, which regenerates via serotinous cones (Harper, 1915; Richardson, 1977). As scrub and sandhill communities both occur on upland sandy soils, they are intermingled and contiguous throughout the Gulf Coastal Plain. The difference ifffire susceptibility is remarkable in that fires sweep through the sandhills until they encounter patches or strands of scrub, where often they are extinguished by changes in the fuel and live vegetation. Webber (1933) called the scrub " a fire-fighting machine," although recognition of the differences between scrub and sandhill trace back to the last century. Nash (1895) noted that "The scrub flora is entirely different from that of the high pine land, hardly a single plant being common to both; in fact these two floras are natural enemies and appear to be constantly fighting each other . . . . . " and Whitney (1898) reiterated that "it is an impressive sight to stand at the border line between the scrub and the high pine land and notice the difference in the character of the vegetation." In the absence of surface fires in the sandhills, the woody species from the scrub will colonize (Laessle, 1958; Veno, 1976). They grow well in the sandhills, often faster than in the scrub, unless a surface fire sweeps through the graminoid ground cover and kills them (Veno, 1976; Hebb, 1982). Large shrubs and especially groups of shrubs may escape mortality, if graminoid growth around them has been suppressed. Therefore, we have been investigating the hypothesis that shrubs from the scrub produce chemicals that inhibit the growth of the grasses and perhaps pines, which provide the fuel for surface fires that otherwise would kill the shrubs. In particular, we are focusing on the early colonizers of the scrub community. One such species is Conradina canescens (Torr. & Gray) (Lamiaceae), a small, evergreen mint, that is endemic to the sand pine scrub community. Its distribution is highly disjunct with two major populations--one on the Gulf coast from Horn Island, Mississippi, to just south of Tallahassee, Florida, and the other concentrated on Florida's Central Ridge in Highlands and Polk counties. In both locations the species is a relatively common shrub, reaching 2 m in height where scrub vegetation is colonizing and in full sunlight, but it diminishes in abundance under the canopy of sand pine and the subcanopy of oaks,
CHEMICAL INHIBITION OF FIRE-PRONE GRASSES
1569
characteristic of mature scrub. The small (2 x 10 mm), recurred leaves of Conradina release a strong mintlike or terpenoid odor when crushed, whereas undamaged leaves emit no odor.
METHODS AND MATERIALS
Foliage from several Conradina shrubs was collected monthly from a site near Sun Ray, Florida. Care was taken to cut the stems so as not to tear the leaves in order to prevent the release of internally bound compounds. In the laboratory, 50 g of foliage was placed in 500 ml of distilled water for 24 hr under refrigeration at 8~ The resulting leaf wash was filtered, and 5 ml was added to 30 target seeds placed on a sheet of Whatman No. 1 filter paper in a 10-cm-diameter Petri dish (Richardson, 1985; Williamson et al., 1988). Seeds of three native sandhill grasses, Schizachyrium scoparium (Michx.) Nash, Andropogon gyrans Ashe, and Leptochloa dubia (HBK.) Nees, were used as target species, each in four replicate dishes kept in the dark at 26~ Control dishes given distilled water were not adjusted for pH or osmolarity because the Conradina leachates were within ranges (5-7 pH, 2-9 m0sm) that did not influence seed germination and growth. Results of the treatment means (T) were compared to the control means (C) each month in a one-way analysis of variance (ANOVA) with four samples per monthly treatment. The magnitude of inhibition or stimulation is reported as RI, a response index that gives the percentage inhibition of the treatment relative to the control (a negative RI) or the percentage inhibition of the control relative to the treatment (a positive RI, when stimulation occurs). RI values allowed parametric comparisons (Williamson and Richardson, 1988). A second bioassay was performed by misting water (6 mm/hr) over potted Conradina shrubs in a growth chamber and collecting the foliar runoff in glass funnels. Then the runoff was used to water flats (5 x 25 x 60 cm) of sandhill soil sown with 300 seeds of Andropogon gyrans. Control flats were given equal amounts of water. Flats were fertilized to field capacity every 18 days with Peters 2 0 : 2 0 : 2 0 (N/P/K). Throughout the experiment the seedlings were thinned only when a plant's leaves reached beyond the stems of its neighbors, so thinning occurred when plants reached a common size, but not necessarily at the same time. After 140 days, each remaining plant was removed from the flat and oven-dried at 105 ~ for three days, and the weights of leaves and roots were recorded. Plant material for chemical analysis was collected from Dauphin Island, Alabama, on October 8, 1983, and again from Perdido Key, Alabama, on June 15, 1984. The sites are within 50 km of each other and probably are part of the same population.
1570
WILLIAMSON ET AL.
Crude extracts of air-dried stems and of air-dried leaves were obtained by sequential use of petroleum ether (PE), diethyl ether (EE), and ethyl alcohol (EtOH) as follows: plant material was soaked in the solvent for 48 hr at 8~ then filtered by suction and dried before soaking in the next solvent (de la Pefia, 1985). Thereafter, the plant material was washed with 60~ ethyl alcohol and finally EtOH-Soxhlet extracted. Solvents were rotary evaporated below 50~ Saturated aqueous solutions were prepared from each of the five residues of the crude extracts of both stems and leaves and then tested for biological activity with Petri dish bioassays on Schizachyrium scoparium, the same native grass employed in the monthly bioassays. Replication varied from two to four dishes of 30 seeds. In addition, parallel bioassays were performed with commercial lettuce (Lactuca sativa). Inhibitory activity was concentrated predominantly in the EE fraction from the leaves, so further chemical analysis was concentrated there. Silica gel column chromatography of the EE fraction yielded 80 fractions, which were combined according to results of thin-layer chromoatography (TLC) into 11 fractions for subsequent bioassays. Biological activity was again concentrated in one sample, EE 6-7, which caused nearly complete inhibition of seed germination. Fractionation of EE 6-7 on a silica gel column yielded 59 fractions, which were combined according to TLC results to give 19 fractions for bioassays; from these, one sample, EE 6.16-6.24 was as inhibitory as EE 6-7. The EE 6.166.24 was analyzed using gas chromotography-mass spectral analysis (GC-MS) to identify the components of the mixture. The GC-MS analyses were performed on a Hewlett-Packard model 5985 instrument. A 30-m bonded silica capillary column was used for the GC-MS analysis, under the following conditions: injection temperature 250~ column temperature program 60 ~ for 1 min., followed by 5~ increases to 210 ~ temperature retained at 210 ~ for 5 min. Compounds were identified by using the computerized MS spectral search system from Hewlett-Packard: Flavor-Fragrance, 59817D National Bureau of Standards Subset Library. Identities of some compounds were further corroborated by comparison of retention times and mass spectra to known standards. Because many of the compounds in the EE active fraction appeared to be volatile organics, a steam distillation of Conradina foliage was undertaken. Fresh leaves in distilled water were boiled with the vapors trapped in an attached condenser. The distillate was extracted with diethyl ether, dried with anhydrous magnesium sulfate, filtered, and concentrated in a rotatory evaporator. The residue, labeled SD (steam distillation), was analyzed (GC-MS) for identification of compounds. In order to determine compounds that might leach from foliage in precipitation, fresh leaves of Conradina were extracted with distilled water as follows: 360 g were twice soaked in 3 liters of distilled water for 3 days at 8 ~ and the aqueous extract reextracted with ethyl acetate; the organic layer was dried with
CHEMICAL INHIBITION OF FIRE-PRONE GRASSES
1571
anhydrous magnesium sulfate, filtered, and the solvent rotary evaporated. The 0.6-g residue, labeled Soak, was analyzed (GC-MS) to identify components of the mixture. Nine of the monoterpenes, identified from the active fractions of Conradina, were tested for biological activity on two native grasses, Leptochloa and Schizachyrium, in triplicate Petri dish bioassays, as described earlier, 30 seeds per dish. Most of these monoterpenes are low-molecular-weight, nonpolar entities, whose solubilities in water have never been determined, so saturated aqueous solutions were prepared for the bioassays here. Scanning electron micrographs of the surface of the leaves were examined for possible trichomes, as sources of production of active compounds. Both fresh leaves and fresh leaves dipped in an organic solvent, dichloromethane (DCM), were examined. RESULTS AND DISCUSSION
The monthly bioassays of Conradina leaf washes showed similar effects on the three sandhill grasses, with Leptochloa experiencing the most inhibition, Andropogon the least, and Schizachyrium intermediate (Table 1). Simple ANOVAs of each month' s data produced significant inhibition of germination in 52 % of the monthly tests--in 88 % for Leptochloa, 44 % for Schizachyrium, and 25 % for Andropogon. Inhibition of radicle length was about half as frequent--in 28 % of the monthly tests overall, 50 % for Leptochloa, 22 % for Schizachyrium, and 12% for Andropogon. The frequency of the significant results suggested that inhibition occurred regularly, but not every month. Species differences in the magnitude of the monthly responses followed
TABLE ][. GERMINATION AND RADICLE LENGTH RESPONSES OF THREE SANDHILL GRASSES TO LEAF WASHES OF Conradina canescens RELATIVE TO WATER CONTROLS
Leptochloa Months tested Significant inhibition Of germination Of radicle length Mean monthly response (RI) Of germination Of radicle length
Schizachyrium
Andropogon
Total
8
9
8
25
7 4
4 2
2 1
13 7
- 0 . 3 4 _+ 0.17 a - 0 . 1 3 _+ 0.14 b
- 0 . 1 7 + 0.37 - 0 . 1 0 _+ 0.22
- 0 . 0 9 + 0.17 +0.03 _+ 0.39
- 0 . 2 2 + 0.28 a - 0 . 0 8 _+ 0.25 b
aTreatment was significantly different from the control at P -< 0.01. bTreatment was significantly different from the control at P
CHEMICAL INHIBITION OF FIRE-PRONE GRASSES BY FIRE-SENSITIVE SHRUB, Conradina canescens
G.
BRUCE
DONALD
WILLIAMSON, R.
1 NIKOLAUS
H.
FISCHER, 2
3 and ANA
DE
LA
RICHARDSON,
PElqA 2
~Department of Botany 2Department of Chemistry, Louisiana State University Baton Rouge, Louisiana 708003 3Department of Biology, University of South Florida Tampa, Florida 33620 (Received February 16, 1988; accepted July 26, 1988)
A b s t r a c t - - I n an investigation of potential chemical activity of fire-sensitive shrubs in Florida's sand pine scrub community, bioassays of foliar washes of Conradina canescens showed significant inhibitory activity on three native grasses that are known to fuel frequent surface fires; inhibition was concentrated seasonally in spring and summer. Application of runoff from Conradina leaves to one of the grasses caused a 50% reduction in growth over a 20-week period. Isolation of the biologically active fractions from the fresh leaves of C. canescens yielded numerous monoterpenes, a number of which were identified from a GC-MS reference library and/or MS comparison to authentic compounds: 11 from the diethyl ether extract, 11 from steam distillation, and four from the foliar leaf wash. Numerous other monoterpenes present in the extractions were unknown. The terpenoid fraction completely inhibited seed germination of one of the native grasses and of lettuce. Saturated aqueous solutions of nine of the monoterpenes inhibited germination and radicle growth of two native grasses. SEM views of the leaf surfaces of Conradina reveal secretory trichomes that appear to be the source of the monoterpenes as well as the triterpene, ursolic acid. The biological activity of C. canescens as a fire-sensitive component of the scrub community is reviewed in light of the chemical evidence. Key Words--Allelopathy, fire, monoterpenes, trichomes, Conradina canes-
cens, Pinus, Schizachyrium scoparium.
1567 0098-0331/89/0500-1567506.00/0
9 I989 Plenum Publishing Corporation
1568
WILLIAMSON ET AL.
INTRODUCTION
Two different plant communities exist on the upland, well-drained sands of Florida: the sand pine scrub characterized by sand pine [Pinus clausa (Chapm. ex Engelm.) Vasey ex Sarg.] with a dense shrub cover but no herbaceous ground cover, and the sandhill or high pineland dominated by longleaf pine (P. palustris Mill.) with a complete graminoid ground cover but few shrubs (Chapman, 1932; Laessle, 1958; Garren, 1943). Fire is a frequent feature of the sandhill, burning through the deciduous surface fuels every three to eight years (Williamson and Black, 1981). However, healthy scrub, which is evergreen, burns infrequently, about once in 50 years or once per generation of the dominant sand pine, which regenerates via serotinous cones (Harper, 1915; Richardson, 1977). As scrub and sandhill communities both occur on upland sandy soils, they are intermingled and contiguous throughout the Gulf Coastal Plain. The difference ifffire susceptibility is remarkable in that fires sweep through the sandhills until they encounter patches or strands of scrub, where often they are extinguished by changes in the fuel and live vegetation. Webber (1933) called the scrub " a fire-fighting machine," although recognition of the differences between scrub and sandhill trace back to the last century. Nash (1895) noted that "The scrub flora is entirely different from that of the high pine land, hardly a single plant being common to both; in fact these two floras are natural enemies and appear to be constantly fighting each other . . . . . " and Whitney (1898) reiterated that "it is an impressive sight to stand at the border line between the scrub and the high pine land and notice the difference in the character of the vegetation." In the absence of surface fires in the sandhills, the woody species from the scrub will colonize (Laessle, 1958; Veno, 1976). They grow well in the sandhills, often faster than in the scrub, unless a surface fire sweeps through the graminoid ground cover and kills them (Veno, 1976; Hebb, 1982). Large shrubs and especially groups of shrubs may escape mortality, if graminoid growth around them has been suppressed. Therefore, we have been investigating the hypothesis that shrubs from the scrub produce chemicals that inhibit the growth of the grasses and perhaps pines, which provide the fuel for surface fires that otherwise would kill the shrubs. In particular, we are focusing on the early colonizers of the scrub community. One such species is Conradina canescens (Torr. & Gray) (Lamiaceae), a small, evergreen mint, that is endemic to the sand pine scrub community. Its distribution is highly disjunct with two major populations--one on the Gulf coast from Horn Island, Mississippi, to just south of Tallahassee, Florida, and the other concentrated on Florida's Central Ridge in Highlands and Polk counties. In both locations the species is a relatively common shrub, reaching 2 m in height where scrub vegetation is colonizing and in full sunlight, but it diminishes in abundance under the canopy of sand pine and the subcanopy of oaks,
CHEMICAL INHIBITION OF FIRE-PRONE GRASSES
1569
characteristic of mature scrub. The small (2 x 10 mm), recurred leaves of Conradina release a strong mintlike or terpenoid odor when crushed, whereas undamaged leaves emit no odor.
METHODS AND MATERIALS
Foliage from several Conradina shrubs was collected monthly from a site near Sun Ray, Florida. Care was taken to cut the stems so as not to tear the leaves in order to prevent the release of internally bound compounds. In the laboratory, 50 g of foliage was placed in 500 ml of distilled water for 24 hr under refrigeration at 8~ The resulting leaf wash was filtered, and 5 ml was added to 30 target seeds placed on a sheet of Whatman No. 1 filter paper in a 10-cm-diameter Petri dish (Richardson, 1985; Williamson et al., 1988). Seeds of three native sandhill grasses, Schizachyrium scoparium (Michx.) Nash, Andropogon gyrans Ashe, and Leptochloa dubia (HBK.) Nees, were used as target species, each in four replicate dishes kept in the dark at 26~ Control dishes given distilled water were not adjusted for pH or osmolarity because the Conradina leachates were within ranges (5-7 pH, 2-9 m0sm) that did not influence seed germination and growth. Results of the treatment means (T) were compared to the control means (C) each month in a one-way analysis of variance (ANOVA) with four samples per monthly treatment. The magnitude of inhibition or stimulation is reported as RI, a response index that gives the percentage inhibition of the treatment relative to the control (a negative RI) or the percentage inhibition of the control relative to the treatment (a positive RI, when stimulation occurs). RI values allowed parametric comparisons (Williamson and Richardson, 1988). A second bioassay was performed by misting water (6 mm/hr) over potted Conradina shrubs in a growth chamber and collecting the foliar runoff in glass funnels. Then the runoff was used to water flats (5 x 25 x 60 cm) of sandhill soil sown with 300 seeds of Andropogon gyrans. Control flats were given equal amounts of water. Flats were fertilized to field capacity every 18 days with Peters 2 0 : 2 0 : 2 0 (N/P/K). Throughout the experiment the seedlings were thinned only when a plant's leaves reached beyond the stems of its neighbors, so thinning occurred when plants reached a common size, but not necessarily at the same time. After 140 days, each remaining plant was removed from the flat and oven-dried at 105 ~ for three days, and the weights of leaves and roots were recorded. Plant material for chemical analysis was collected from Dauphin Island, Alabama, on October 8, 1983, and again from Perdido Key, Alabama, on June 15, 1984. The sites are within 50 km of each other and probably are part of the same population.
1570
WILLIAMSON ET AL.
Crude extracts of air-dried stems and of air-dried leaves were obtained by sequential use of petroleum ether (PE), diethyl ether (EE), and ethyl alcohol (EtOH) as follows: plant material was soaked in the solvent for 48 hr at 8~ then filtered by suction and dried before soaking in the next solvent (de la Pefia, 1985). Thereafter, the plant material was washed with 60~ ethyl alcohol and finally EtOH-Soxhlet extracted. Solvents were rotary evaporated below 50~ Saturated aqueous solutions were prepared from each of the five residues of the crude extracts of both stems and leaves and then tested for biological activity with Petri dish bioassays on Schizachyrium scoparium, the same native grass employed in the monthly bioassays. Replication varied from two to four dishes of 30 seeds. In addition, parallel bioassays were performed with commercial lettuce (Lactuca sativa). Inhibitory activity was concentrated predominantly in the EE fraction from the leaves, so further chemical analysis was concentrated there. Silica gel column chromatography of the EE fraction yielded 80 fractions, which were combined according to results of thin-layer chromoatography (TLC) into 11 fractions for subsequent bioassays. Biological activity was again concentrated in one sample, EE 6-7, which caused nearly complete inhibition of seed germination. Fractionation of EE 6-7 on a silica gel column yielded 59 fractions, which were combined according to TLC results to give 19 fractions for bioassays; from these, one sample, EE 6.16-6.24 was as inhibitory as EE 6-7. The EE 6.166.24 was analyzed using gas chromotography-mass spectral analysis (GC-MS) to identify the components of the mixture. The GC-MS analyses were performed on a Hewlett-Packard model 5985 instrument. A 30-m bonded silica capillary column was used for the GC-MS analysis, under the following conditions: injection temperature 250~ column temperature program 60 ~ for 1 min., followed by 5~ increases to 210 ~ temperature retained at 210 ~ for 5 min. Compounds were identified by using the computerized MS spectral search system from Hewlett-Packard: Flavor-Fragrance, 59817D National Bureau of Standards Subset Library. Identities of some compounds were further corroborated by comparison of retention times and mass spectra to known standards. Because many of the compounds in the EE active fraction appeared to be volatile organics, a steam distillation of Conradina foliage was undertaken. Fresh leaves in distilled water were boiled with the vapors trapped in an attached condenser. The distillate was extracted with diethyl ether, dried with anhydrous magnesium sulfate, filtered, and concentrated in a rotatory evaporator. The residue, labeled SD (steam distillation), was analyzed (GC-MS) for identification of compounds. In order to determine compounds that might leach from foliage in precipitation, fresh leaves of Conradina were extracted with distilled water as follows: 360 g were twice soaked in 3 liters of distilled water for 3 days at 8 ~ and the aqueous extract reextracted with ethyl acetate; the organic layer was dried with
CHEMICAL INHIBITION OF FIRE-PRONE GRASSES
1571
anhydrous magnesium sulfate, filtered, and the solvent rotary evaporated. The 0.6-g residue, labeled Soak, was analyzed (GC-MS) to identify components of the mixture. Nine of the monoterpenes, identified from the active fractions of Conradina, were tested for biological activity on two native grasses, Leptochloa and Schizachyrium, in triplicate Petri dish bioassays, as described earlier, 30 seeds per dish. Most of these monoterpenes are low-molecular-weight, nonpolar entities, whose solubilities in water have never been determined, so saturated aqueous solutions were prepared for the bioassays here. Scanning electron micrographs of the surface of the leaves were examined for possible trichomes, as sources of production of active compounds. Both fresh leaves and fresh leaves dipped in an organic solvent, dichloromethane (DCM), were examined. RESULTS AND DISCUSSION
The monthly bioassays of Conradina leaf washes showed similar effects on the three sandhill grasses, with Leptochloa experiencing the most inhibition, Andropogon the least, and Schizachyrium intermediate (Table 1). Simple ANOVAs of each month' s data produced significant inhibition of germination in 52 % of the monthly tests--in 88 % for Leptochloa, 44 % for Schizachyrium, and 25 % for Andropogon. Inhibition of radicle length was about half as frequent--in 28 % of the monthly tests overall, 50 % for Leptochloa, 22 % for Schizachyrium, and 12% for Andropogon. The frequency of the significant results suggested that inhibition occurred regularly, but not every month. Species differences in the magnitude of the monthly responses followed
TABLE ][. GERMINATION AND RADICLE LENGTH RESPONSES OF THREE SANDHILL GRASSES TO LEAF WASHES OF Conradina canescens RELATIVE TO WATER CONTROLS
Leptochloa Months tested Significant inhibition Of germination Of radicle length Mean monthly response (RI) Of germination Of radicle length
Schizachyrium
Andropogon
Total
8
9
8
25
7 4
4 2
2 1
13 7
- 0 . 3 4 _+ 0.17 a - 0 . 1 3 _+ 0.14 b
- 0 . 1 7 + 0.37 - 0 . 1 0 _+ 0.22
- 0 . 0 9 + 0.17 +0.03 _+ 0.39
- 0 . 2 2 + 0.28 a - 0 . 0 8 _+ 0.25 b
aTreatment was significantly different from the control at P -< 0.01. bTreatment was significantly different from the control at P
