Arboreal Seed Removal and Insect Damage in Three California Oaks

We investigated arboreal removal and insect damage to acorns in an undisturbed oak woodland in central coastal California. Arboreal seed removal was determined for four to eight individual Quercus lobata trees over a period of 14 years by comparing visual estimates of the acorn crop with the number of acorns caught in seed traps. Insect damage was assessed by sampling acorns from trees of all three species of oaks common in the study site (Quercus lobata, Q. douglasii, and Q. agrifolia). Patterns were generally similar for both sets of data: more acorns, but a smaller proportion of the crop, were removed or damaged as the productivity of an individual tree increased. However, we found no evidence that trees outproducing local conspecifics attracted a disproportionate number of arboreal seed or insect predators. Acorn removal was not significantly correlated with population sizes of either California scrub-jays (Aphelocoma coerulescens) or acorn woodpeckers (Melanerpes formicivorus), two common arboreal seed removers that are also potentially important dispersal agents. These patterns are partially in accord with predator satiation, but not the attraction of seed dispersers, being an important factor potentially influencing the reproductive strategies of oaks in central coastal California. Introduction Patterns of seed production, including reproductive synchrony on a geographically large scale or masting (Kelly and Sork [In press]), Silvertown 1980), may be explained by two general hypotheses. First is resource matching, which proposes that masting is associated with years in which resources are more available, and second are economies of scale, which propose that synchrony arises from the potential energetic advantages to individuals within a population of investing more into reproduction synchronously every few years rather than less each year, given an overall constraint to the total level of reproductive effort (Norton and Kelly 1988). Economies of scale include energetic advantages related to wind pollination (Norton and Kelly 1988, Smith and others 1990), predator satiation (Ims 1990a, 1990b; Janzen 1971), and the attraction of seed dispersers (Givnish 1980, Sork and others 1983). As traditionally envisioned, the importance of predator satiation is 1 An abbreviated version of this paper was presented at the Fifth Symposium on Oak Woodlands: Oaks in California’s Changing Landscape, October 22-25, 2001, San Diego, California. 2 Research Zoologist, Hastings Reservation and Museum of Vertebrate Zoology, University of California, Berkeley, 38601 E. Carmel Valley Road, Carmel Valley, CA 93924 (e-mail: [email protected]) 3 Assistant Professor, Department of Biological Sciences, University of Nebraska, 348 Manter Hall, Lincoln, NE 68588. 4 Principal, 145 Eldridge Ave., Mill Valley, California 94941. USDA Forest Service Gen. Tech. Rep. PSW-GTR-184. 2002. 193 Seed Predation—Koenig, Knops, and Carmen dependent on the ability of highly productive trees to overwhelm seed predators, thereby resulting in a negative correlation between the size of the seed crop and the proportion of seeds depredated (Norton and Kelly 1988). Selection for reproductive synchrony should be particularly intense depending on neighborhood effects, that is, the crop of a particular tree relative to the area as a whole (Sargent 1990). Trees outproducing the local area should suffer increased seed predation and there will be selection to synchronize reproductive effort within the population. Alternatively, if the attraction of seed dispersers is important, trees should produce fruit in ways which differentially attract potential seed dispersers and lead to a relatively high proportion of seeds removed being cached rather than eaten. Thus, the optimal pattern of seed production for an individual tree depends on the potentially complex and generally unknown relationship between the proportion of seeds cached by animals, the efficiency with which cached seeds are recovered, the reproductive effort of an individual tree, and overall seed availability in a particular year. If neighborhood effects are strong, some trees should invest heavily in reproduction each year in order to outcompete other conspecifics and ensure the attraction of potential dispersal agents. On the other hand, an individual tree producing a large crop in a year when all other trees fail will presumably suffer heavy predation as animals attracted to the tree eat rather than cache most of the acorns they remove. The expected result of these conflicting selection pressures is not likely to be straightforward (Ims 1990a). Presumably these competing considerations are likely to result in moderate to low reproductive synchrony (Koenig and others 1994a). With respect to the size of the acorn crop per se, we can expect a positive correlation between a tree’s seed crop and not only the number but also the proportion of acorns removed if productive trees successfully attract proportionately more dispersers than unproductive trees. A third possibility is that primary seed predators are generalists switching from alternative food resources, in which case Ims (1990a, 1990b) has shown that the expected pattern of reproduction is one of asynchrony rather than synchrony. Given the vast array of both vertebrate and invertebrate acorn predators, it is thus not possible to confidently predict the optimal pattern of reproduction in response to predation in this system (Koenig and others 1994a). Consequently, we restrict ourselves to the more traditional form of predator satiation discussed above. Invertebrate users of oaks (genus Quercus, family Fagaceae) are particularly extensive: as many as 5,000 insect species are associated with oaks in California, of which approximately 800 use some portion for food (Pavlik and others 1991). In addition, over 170 species of birds and mammals are dependent on oak habitats in California and nearly 100 species are known to feed on acorns in the United States, making Quercus one of the most important genera of woody plants to wildlife in North America (Barrett 1980, Christisen and Korschgen 1955, Martin and others 1951, Van Dersal 1940, Verner 1980). Given this impressive assemblage of animals, it is likely that predation is important to reproductive patterns of oaks. In return, oaks disperse their seeds largely by taking advantage of animals that fail to remove some proportion of acorns that they have moved and cached (Smith and Folmer 1972, Sork and others 1983). Particularly significant in this respect are jays (family Corvidae); birds of this family are common seed removers (Darley-Hill and Johnson 1981) and, since they store acorns in the ground, are excellent dispersal agents (Carmen [In press], Grinnell 1936, Vander Wall 1990). USDA Forest Service Gen. Tech. Rep. PSW-GTR-184. 2002. 194 Seed Predation—Koenig, Knops, and Carmen This paper is one of a series devoted to understanding acorn production patterns of California oaks. Here we discuss the factors influencing arboreal acorn removal, primarily by birds, and predation on acorns, again prior to acorn fall, by insects; we do not discuss seed predation on sound acorns after they have fallen from the tree. We focus on three questions. First, what is the intensity of arboreal acorn removal and and of insect predation? Second, what factors influence the extent of these phenomena? Third, does the pattern of acorn removal and insect damage suggest an important role for either predator satiation or the attraction of seed dispersers in the evolution of oak reproductive patterns? Methods The study was conducted at Hastings Reservation, a 900-ha reserve located in the Santa Lucia Mountains of central coastal California, approximately 42 km inland from Monterey. Elevation at the study site ranges from 460 to 950 m. This area experiences a Mediterranean climate in which virtually no rain falls during the summer and early fall (June-September). Annual rainfall ranges from 26.1 to 111.2 cm, with a 50-yr mean of 55 cm. In all areas of the study site oak (Quercus) is the dominant genus of tree. Five species are common, but only three are widespread throughout the site: Quercus lobata (valley oak), Q. douglasii (blue oak), and Q. agrifolia (coast live oak). These are joined locally, mostly at higher elevations, by Q. chrysolepis (canyon live oak) and Q. kelloggii (California black oak). Arboreal Seed Removal Eight large, mature Quercus lobata trees were used. Four, sampled from 19801989 and 1992-1996, were located on level ground in the floodplain about 0.5 km from the reserve headquarters, while the other four, sampled from 1992-1996 only, were located in an old field on a hill 0.75 km from headquarters. For each tree, we estimated the extent of arboreal seed removal by comparing visual estimates of the acorn crop made prior to acorn fall with the number of acorns collected from seed traps. Visual estimates involved having two experienced observers scan different areas of each tree’s canopy and count as many apparently viable acorns as possible in 15 seconds. Counts were made each autumn in September or early October at the height of the acorn crop prior to acorn fall. The total number of acorns counted by both observers was added to yield acorns counted per 30 seconds and then log-transformed to reduce the correlation between the mean and variance (Sokal and Rolf 1969). For further details on this survey method, see Koenig and others (1994b). Log-transformed values for the four (or eight) trees were averaged to estimate the mean crop of the focal trees for each year. Acorn fall under each of the focal trees was sampled using seed traps consisting of plastic garbage bags held in place by hogwire frames. Each trap was approximately 0.25 m in area and permanently located around the tree about halfway between the trunk to the edge of the tree’s canopy. Each trap was assumed to sample an approximately equal volume of canopy. Four traps per tree were used. Traps were checked at weekly intervals throughout the period of acorn fall and the total number of acorns caught summed for all traps for a given tree throughout the season. USDA Forest Service Gen. Tech. Rep. PSW-GTR-184. 2002. 195 Seed Predation—Koenig, Knops, and Carmen Arboreal seed removal was estimated as follows. First, we plotted the relationship between the two measures of the acorn crop for the individual trees (figure 1). The expected number of acorns trapped was estimated from a line drawn from the origin to the point yielding a line of maximum slope (point A); the formula for this line is y = 1.259 x, where y is the number of acorns trapped and x is the number of acorns counted (both log-transformed). This conservatively assumes that point A represents no seed removal and, given this assumption, the deviation of the other points from the line provides an estimate of the extent of arboreal seed removal. Values for the expected and observed numbers of acorns trapped were then backtransformed (to acorns m), from which we estimated both the total number of acorns (per trap) that were apparently removed by arboreal seed predators (the “total” number of acorns removed) and the proportion of acorns produced that were removed by arboreal seed predators (the “relative proportion” of acorns removed). Values were then averaged within years to derive mean annual values. Trees and years in which no acorns were produced were excluded, since the number and proportion of acorns removed in such cases were indeterminate. This left samples for 61 trees over a period of 14 years (none of the trees sampled produced any acorns in 1983). In addition to the focal trees, an additional 37 Q. lobata trees within 1 km of the focal trees were visually sampled at Hastings each autumn in order to assess the overall acorn crop. The average number of (log-transformed) acorns counted in these trees was used as a measure of the overall acorn crop. Results were not substantively changed if all sampled trees (including 25 Q. douglasii and 27 Q. agrifolia) within 1 km were used instead. Neighborhood effects, that is, the effect of local seed density on removal from individual trees (Sargent 1990) were investigated by comparing the estimated proportion of acorns removed from trees that outproduced an average tree in the full Q. lobata sample with those that produced fewer acorns than the full sample, based on the number of acorns counted per 30 seconds.