FREQUENCIES OF COLOR MORPHS IN FOUR POPULATIONS OF ENOPLOGNA THA OVA TA ( ARANEAE : THERIDIIDAE ) IN EASTERN NORTH AMERICA *

Several types of spiders exhibit conspicuous color polymorphisms. Reiskind (1970) has discovered mimetic polymorphisms among clubionids. Examples in which only one sex is polymorphic occur among the jumping spiders, which have keen vision and often show marked sexual dimorphism (e.g. Galiano 1981a, b). In these species the sex-limited polymorphisms may be related to differential predation pressures upon the sexes and/or sexual selection (Stamps and Gon 1983). Many theridiid spiders are polymorphic (Levi 1957; Stamps and Gon 1983), though in this group the eyes are not well developed and recognition of colors or patterns during courtship probably does not occur. Polymorphisms in theridiid species appear not to be sex-limited, though the frequencies of certain morphs may differ between the sexes (Hippa and Oksala 1977, 1981). By far the most extensively studied color polymorphism among spiders is that found in the theridiid Enoplognatha ovata (Clerck). Populations of E. ovata frequently contain three distinctive types of adults. The most common morph has a pale yellow abdomen, often with several pairs of dorsolateral dark spots, but with no red stripes (form lineata); another form has two dorsolateral red stripes running the length of the abdominal dorsum (form redimita); and the least common morph has a solid red band running the length of the abdominal dorsum (form ovata) (Bristowe 1958). This striking variation in a common species has long fascinated naturalists. Early arachnologists considered the forms to be three species, but by early last century the three varieties were considered to be conspecific


INTRODUCTION
Several types of spiders exhibit conspicuous color polymorphisms.Reiskind (1970) has discovered mimetic polymorphisms among clubionids.Examples in which only one sex is polymorphic occur among the jumping spiders, which have keen vision and often show marked sexual dimorphism (e.g.Galiano 1981a, b).In these species the sex-limited polymorphisms may be related to differential predation pressures upon the sexes and/or sexual selection (Stamps  and Gon 1983).Many theridiid spiders are polymorphic (Levi 1957; Stamps and Gon 1983), though in this group the eyes are not well developed and recognition of colors or patterns during courtship probably does not occur.Polymorphisms in theridiid species appear not to be sex-limited, though the frequencies of certain morphs may differ between the sexes (Hippa and Oksala 1977, 1981).By far the most extensively studied color polymorphism among spiders is that found in the theridiid Enoplognatha ovata (Clerck).
Populations of E. ovata frequently contain three distinctive types of adults.The most common morph has a pale yellow abdomen, often with several pairs of dorsolateral dark spots, but with no red stripes (form lineata); another form has two dorsolateral red stripes running the length of the abdominal dorsum (form redimita); and the least common morph has a solid red band running the length of the abdominal dorsum (form ovata) (Bristowe 1958).This striking variation in a common species has long fascinated naturalists.Early arachnologists considered the forms to be three species, but by early last century the three varieties were considered to be conspecific Psyche [Vol. 92  (Hippa and Oksala 1979).The three morphs will mate readily in the laboratory, and Seligy (1971) discovered that lineata, redimita and ovata could be obtained in the progeny of a single female.Although the morphs are usually distinct, recent research of Hippa and Oksala (1979, 1981, 1982) indicates that the genetics of the color polymorphism is more complicated than previously thought; a brief summary of recent findings appears in the Appendix.
Ecologists have uncovered two intriguing aspects to the distribu- tion of these morphs in Europe.Local populations often differ in the frequencies of the three morphs, yet a given population will maintain remarkably constant frequencies over several generations (Oxford 1976).Furthermore, a latitudinal cline in morph frequen- cies exists, with the red forms tending to be more common in north- ern populations of continental Europe (Hippa and Oksala 1979).
E. ovata also occurs in parts of North America (Levi 1957), but has not been extensively studied on this continent.Seligy (1971) provided information on the natural history of the species in North America and described details of postembryonic development; how- ever, he did not give data on morph frequencies in different locales.
In 1983 we determined morph frequencies in E. ovata populations in Maine, Massachusetts, and two regions of New York.We discov- ered that frequencies varied between local populations within an area, and also found that the overall frequencies for each geographical region were not the same.Most striking was the presence of the ovata morph in Maine and a mountainous area of Massachusetts, and its absence from two populations in widely separated regions of New York State.

M ETItODS
Natural History of E. ovata This brief account is based primarily upon Nielsen (1932), Bris- towe (1958), Seligy (1971) and personal observations.Middle through adult instars are found underneath the leaves of many species of brambles, small saplings and herbs such as golden rod (Solidago).E. ovata occurs along roadside ditches, open fields and in the forest understory.It is often highly aggregated, with dense clumps sometimes being referred to as "colonies".The spider builds a small, inconspicuous tangle web underneath a leaf, the edges of which are often pulled down slightly by silk lines.This retreat is readily noticeable when examining vegetation, and the one con- structed by the female for her egg sac is particularly obvious.Most females deposit a single sac in July or August, which they guard in a rolled leaf.The 2nd instar spiderlings emerge after several weeks and disperse in September through October.These instars overwin- ter in the leaf litter and in early spring build small webs in the curls of dead leaves.At this stage they were the most abundant spider in the litter community of one ecosystem studied (Stevenson and Din- dal 1982; the population that they studied was the same central New York population that we examined).As the juveniles begin to molt they move into living vegetation where they build small webs similar to those of adults, though the younger instars do not consistently curl down the edges of leaves.
Sampling the North American Populations During 13-17 July 1983 we surveyed morph frequencies of penul- timate and adult E. ovata in several local populations in each of four different regions: Mt.Desert Island in Maine, October Mountain State Forest in the Berkshire Mountains of western Massachusetts, Huyck Nature Preserve in eastern New York, and the Lafayette Experimental Station of SUNY Syracuse in central New York (Fig. I).Local populations were defined primarily as dense aggregations of E. ovata in contiguous stands of vegetation.After locating an aggregation, we continued to move systematically through that patch of vegetation until E. ovata was no longer common, or until we had scored approximately 50 spiders.This latter criterion was used primarily in Syracuse, where the vegetation suitable for habita- tion by E. ovata was continuous over large areas of the forest.The local populations were 50m to several km apart.We examined each patch carefully, but did not make an exhaustive survey of each geographical area to determine the number of local populations.Instead, we attempted to examine as many different geographic regions as possible, spending from under a day (Syracuse) to two days (Maine) in each area.We scored a total of 1107 spiders in these four areas, but in presenting frequencies in local populations within each region we have only included those with 20 or more individuals.REStrLTS AND DIS('USSION Pooled local populations from the four geographic regions dif- fered in overall morph frequencies (Fig. 2).Proportions of the yel- low morph, /ineata, differed significantly among the four regions sampled (chi-square 11.41, p < .01,df 3), due to the lower frequency of lineata in Massachusetts compared to the other three areas.The most striking difference involved the solid red morph ovata.This form was absent from both New York populations, but occurred in both Massachusetts and Maine.The differences in the frequency of ovata among the four regions was significant statisti- cally (chi-square 34.49, p < .001,df 3).
We uncovered similar differences in morph frequencies between local populations within one of the regions that we sampled (Fig. 3).Variation in representation of the morphs among populations on Mt.Desert Island in Maine was statistically significant (chi-square 13.2; p < .05,df 5, 2 6 contingency table; comparison of lineata frequencies).The frequency of iineata varied from .70 to .96, and that of ovata from 0 to.15.The difference between population "A" and several of the others was quite striking, in that "A" had no ovata and very few redimita, whereas the others contained large numbers Morph frequencies in the four different regions that were sampled.Indi- viduals from the local populations (i.e.aggregations > 20) and also spiders found in aggregations with fewer than 20 individuals have been pooled to calculate the proportion of each morph.Thus, with the exception of SUNY Syracuse, the sample used to derive proportions for each region is larger than the sum total of local populations that appear in Fig. 3.
of both the striped and the solid-red morphs.Massachusetts differ in the proportion of the ovata phenotype.The three populations from the Huyck Preserve were quite similar; they all lacked the ovata morph and had similar frequencies of redimita.
Populations from the forest in Syracuse, New York also lacked ovata and did not differ significantly in the proportion of redimita (chi-square 6.0, p .4,df 6, 2 7 contingency table ).This constancy likely reflected both the continuity of the habitat and the proximity of the local "populations" to one another.Each group that was defined as a population was no more than 50 m from another, with continuous vegetation also inhabited by E. ovata existing between them.Unlike the local populations in the other three regions, those in the central New York site were not physically separate aggregations.These preliminary observations of morph frequencies in E. ovata populations in eastern North America suggest that patterns similar to those in Europe will be uncovered if more areas are sampled.In the region most extensively researched, Mt.Desert Island, signifi- cant differences in morph frequencies occurred among local populations.European workers have reported similar differences (Oxford   1976, Hippa and Oksala 1979).Morph frequenci6s also differed between widely separated regions.This preliminary evidence is insuf- ficient to judge whether North American E. ovata exhibits a morph- ratio cline similar to that found in Europe by Hippa and Oksala (1979), but the absence of the ovata morph from both regions in New York is striking.A more detailed examination of E. ovata populations along latitudinal and longitudinal gradients in North America might uncover a consistent pattern of variation.
Apparently E. ovata is a relatively recent arrival to North America (Levi 1957).Our initial studies indicate that morph-frequency patterns in North America may be equally as intriguing as those already documented in Europe.More detailed examination of the pattern in a recently colonized area could provide comparative data useful to understanding the evolutionary forces responsible for the patterns of variation in morph frequencies occurring on both continents.

SUMMARY
The spider Enoplognatha ovata exhibits a conspicuous color polymorphism characterized by three morphs: lineata (yellow abdomen with no stripes), redimita (two dorso-lateral red stripes on Psyche [Vol. 92the abdomen) and ovata (medial red band on the dorsum).European researchers have found that local populations often differ in morph frequencies, and that average frequencies differ between geographic regions.We found preliminary evidence of similar vari- ability in North American populations of E. ovata.We recorded the phenotypes of 1107 spiders from several local populations, 50m to several km apart, in each of four areas: eastern Maine, western Massachusetts, eastern New York, and central New York.The overall frequency of lineata was .70 among the spiders from Massachusetts, but was close to .80 in the other three regions.The most striking geographic differences in frequency involved the ovata morph.This form was absent from all New York populations sampled, but occurred in all but two of the local populations from Massachusetts and Maine.Overall frequencies of ovata in the Massachusetts and Maine samples were .03and .06,respectively.In Maine, the most intensively sampled region, morph frequencies dif- fered significantly among local populations.The proportion of lin- eata varied from .70 to .96;ovata frequencies ranged from 0 to. 15.These observations suggest that further examination of E. ovata in North America may uncover morph frequency differences between populations comparable to those in Europe.Since E. ovata is likely a recent immigrant to North America, more detailed information from this continent could yield valuable comparative data on a cu- rious phenomena whose evolutionary basis is still unexplained.APPENDIX Oxford (1976) found variation in coloration and patterning of the three morphs, particularly among laboratory-reared spiders.Hippa and Oksala (1979, 198 l) also found that morphs were not always distinct, and furthermore, that in some groups the red coloration was not expressed until either the third instar or the final molt.Recently, Hippa and Oksala (1982) cited evidence from studies of variability in genitalia structure that some populations of E. ovata may consist of more than one species, contrary to previous interpre- tations of Levi (1957,1967).
This recent information complicates the situation, but does not detract from the usefulness of our findings, for three reasons: (l) Only a very few intermediate forms appeared in our sam- ple of over 1000 spiders.
(2) Only the Maine populations appeared to have had signifi- cant numbers of penultimate females; however, they were a minority.Most spiders appeared to be mature.Develop- ment was more advanced in the southern populations, with many females guarding egg sacs.Thus in Maine we may have slightly underestimated the frequencies of the red morphs (Hippa and Oksala 1979); however, this possible bias makes our conclusion of significant morph frequency differences between regions conservative, since the sample from Maine had the highest frequencies of ovata, the phenotype with the most red coloration.
(3) Definition of new species in the E. ovata group by Hippa and Oksala (1982) is based upon relatively few museum specimens.This type of information does not indicate whether or not expressions of the red morphs differs between E. ovata and the newly recognized Enoplognatha species, nor does it indicate the extent of niche differences.The most extensive sets of data available on morph frequencies in European populations are based on the assumptions of three dis- tinct morphs and one species (eg.Oxford 1976, Hippa and Oksala  1979).Hence our data are directly comparable to that collected by European researchers.Future, more detailed research on both con- tinents will have to take into account the possible existence of more than one species in some E. ovata populations.However, this taxo- nomic problem does not prevent one from making preliminary statements about patterns in morph frequencies.
Fig. I. Sites at which morph frequencies in local populations of Enoplognatha ovata were scored.
Fig.2.Morph frequencies in the four different regions that were sampled.Indi-
The number of popu- lations from October Mt.State Forest in Massachusetts and from the Huyck Preserve in New York is too small to make any generali- zations about local differences in morph frequencies.Additional sampling might indicate that some local populations in western