Wilson Bull., 108(4), 1996, pp. 783-796 THE USE OF COASTAL AGRICULTURAL FIELDS IN VIRGINIA AS FORAGING HABITAT BY SHOREBIRDS Stephen C. Rottenborn Abstract. I studied temporal abundance patterns and use of cover types by shorebirds foraging in coastal croplands on the eastern shore of Virginia from March 1991 through February 1992. A total of 21,254 shorebirds of 21 species was observed foraging in agri- cultural croplands. Shorebird abundance reached a peak during spring migration and was lower during fall and winter. Shorebird species richness was highest in fall and lower in spring and winter. Some species appeared to use fields primarily as alternate foraging habitat when preferred intertidal habitats were covered by high tides, whereas other species foraged in fields regardless of tidal height. Most species showed highly significant positive associ- ations with plowed fields (bare earth) and negative associations with herbaceous vegetation >10 cm tall. Associations with vegetated cover of <10 cm varied seasonally, tending to be positive in spring and negative (or with no association) in fall and winter for most species. Because most shorebirds foraged on the cover type providing the least cover from predators (plowed fields), the observed cover associations may reflect differences in predatordetection or foraging efficiency among the cover types. Received 20 Nov. 1995, accepted 13 April 1996. Most research on habitat use and foraging strategies of shorebirds has focused on the use of intertidal habitats, where they may choose different foraging sites and strategies depending on tidal conditions (cf Burger et al. 1977, Connors et al. 1981). When preferred intertidal habitats are inundated by high tides, shorebirds may forage temporarily in alternate areas, including agricultural fields (Atkinson 1976, Gerstenberg 1979, Page et al. 1979). Previous studies of the use of agricultural croplands by shorebirds have dealt primarily with species that prefer upland habitats and rarely move to intertidal areas (Fuller and Youngman 1979, Barnard and Thompson 1985, Milsom et al. 1985) or were conducted in inland regions where tidal regime had no influence on shorebird distribution and behavior (Ohmart et al. 1985). In California, a number of shorebird spe- cies move from intertidal areas to pastures during high tides (Gerstenberg 1979, Page et al. 1979), and some of the habitat features associated with the birds’ use of coastal pastures have been described (Colwell and Dodd 1995). Despite the potential importance of cropland as alternate foraging habitat for shorebirds (Ohmart et al. 1985), the use of coastal cropland by foraging shorebirds has not been described. The objectives of this study were (1) to determine the species composition and temporal abun- dance patterns of shorebirds foraging in coastal cropland on the Eastern Center for Conservation Biology, Dept, of Biological Sciences, Stanford Univ., Stanford, California 94305-5020 . 783 784 THE WILSON BULLETIN • Vol. 108, No. 4, December 1996 Shore of Virginia, and (2) to determine patterns of cover use by these shorebirds. STUDY AREA AND METHODS I conducted this study on the southern tip of the Eastern Shore of Virginia, within an area of approximately of 170 km^. The study area is bounded to the east by extensive tidal salt marsh, mud flats, and barrier islands and to the west by narrow sandy beaches (with several tidal creeks) along Chesapeake Bay. Large numbers of shorebirds forage in these tidal areas, especially at low tide when beaches and mudflats are exposed. Previous obser- vations indicated that individuals of some shorebird species moved from these intertidal habitats to agricultural croplands to forage during high tide. In order to determine the relationship between tidal height and the abundance of certain shorebird species in agricultural fields, I conducted four surveys of a small number offields in the study area (6-8 fields/survey). The days on which these censuses were conducted, 29 May, 12 and 31 August, and 2 November 1991, were chosen so that the timing of high and low tides varied among surveys. Such variation in the timing of high and low tides allowed me to distinguish possible relationships between shorebird abundance and time of day from relationships between shorebird abundance and tidal height. During these surveys, I visited each field at one- to two-h intervals between dawn and dusk, recording the number of individuals of each species in each field during each round of censuses. All fields were visited the same number of times on a given date, but the number of census rounds varied from seven to nine among different survey dates. I divided the species observed in fields into two groups based on my previous experience with shorebirds in intertidal habitats around the study area and life history information (Stout 1967, Hayman et al. 1986): field specialists, thought to forage primarily in fields and rarely visiting intertidal habitats, and field exploiters, species that regularly forage in intertidal habitats during low tide but oc- casionally forage in fields during high tide. Eor each field, the abundance of shorebirds in each of these groups was determined for the censuses conducted closest to the times cor- responding to high and low tide at Cape Charles, Virginia (U.S. Department of Commerce 1990). Paired comparisons r-tests were used to test the null hypothesis that the abundance of shorebirds in these fields did not differ between high tide and low tide (Sokal and Rohlf 1981). In order to determine seasonal patterns of shorebird abundance and the use of various cover types by shorebirds foraging in coastal croplands, I conducted thirty censuses of croplands (comprising 4598.8 ha) in the study area from March 1991 through Eebruary 1992. For the purpose of determining the relative availability of cover types, I defined a field as a continuous plot with homogeneous cover of bare earth or herbaceous vegetation of the same height class (<10 cm or >10 cm) not divided by a paved road or by woody vegetation. Because high and low tides were separated by approximately 6.0-6.5 h, I began each census 4 h before high tide so that all fields would be surveyed during the highest 2/3 of the tidal cycle. Each census lasted approximately 8 h. To randomize the tide height during my observation of any particular field, I began each census at a location chosen randomly from a pool of 20 points along the census route. Censuses were not conducted if visibility was hampered by fog or precipitation. During each census, I drove along the census route, stopping at each field to scan for shorebirds. I counted shorebirds from one or more points at the edge of each field using 8.5X binoculars. For larger fields and for fields with taller vegetation, I scanned from several points along the edge ofeach field, using a 22X spotting scope as necessary. Scanning along rows of vegetation and listening for the birds’ calls (given frequently by foraging flocks) Rottenborn • AGRICULTURAL FIELD USE BY SHOREBIRDS 785 aided in the detection ot shorebirds. Eor large or widely dispersed flocks, I made replicate counts ot shorebirds and recorded the mean of the original and replicate counts. Most of the shore^rds observed on croplands were actively foraging, with only a small proportion roosting. Observations made during the four dawn-to-dusk surveys and at other times sug- gested that very few shorebirds, if any, used these fields for roosting without spending some time foraging. Therefore, all shorebirds recorded during censuses were assumed to be for- aging in the fields. I was able to corroborate the accuracy of counts for cover types of <10 cm vegetation and plowed fields by locating three fields with each of these cover types that contained large flocks of foraging shorebirds, counting the birds from the edge of each field, and then walking through the fields to flush and count every bird. In all six cases, the error in these replicate counts was less than two percent. I was unable to corroborate directly my counts for >10 cm cover, but on one occasion a disturbance in a field of tall vegetation flushed the shorebirds within the field, confirming my previous count. During each census, I recorded the type and height of cover in each field. I recognized three cover types in these croplands: plowed (bare earth with no vegetation), herbaceous vegetation <10 cm tall, and herbaceous vegetation >10 cm tall. No distinction was made between row crops (which included cotton, soybeans, potatoes, string beans, and several other crops) and mat crops (mostly cereals). Eields with weedy herbaceous vegetation al- lowed to grow in the interval between the harvest of one crop and seeding of another were also included in this study, although I excluded fields that lay fallow for the entire study period. All crop types reached heights exceeding 10 cm, so vegetation height (<10 cm or >10 cm) was not dependent on crop type. I did not include any pastureland in this study, as very little was present within the study area. The area of each field was measured with a planimeter from field-checked 7.5-min to- pographic maps. For each census, I summed the areas of all fields with a particular cover type to determine the total availability of each cover type in the study area. I then deter- mined, for each census, the number of individuals of each shorebird species expected to occur in fields with each cover type if shorebird distribution over the study area were independent ofcover type. These numbers were the products ofthe total numberofobserved individuals ofeach species and the proportion offields having each ofthe three cover types. I summed the numbers of shorebirds observed and expected on each cover type over all censuses in each season in order to determine the total number ofIndividuals ofeach species that were observed and expected on each cover type by season. In pooling these numbers by season, I assumed that patterns of cover use did not differ significantly among different censuses within the same season. For the purposes of this study, I delimited seasons a posteriori by examining abundance patterns of species known to occur in coastal Virginia as summer or winter residents or as northbound (spring) or southbound (fall) migrants (Kain 1987). I defined summer as the period from early June until mid-July, during which the only species observed in the study area was a local breeder. Fall lasted from the arrival of the first southbound migrants in late July until the departure of the last southbound transients (i.e., species that do not overwinter in Virginia) from the study area in mid-November. Winter extended from mid-November until late March. Spring began with the first arrival of northbound migrants in early April and ended in late May. Because I observed only two individuals of one locally breeding species during summer, analyses presented here are restricted to fall, winter, and spring. Sampling intensity differed among seasons due to differences in season length; fourcensuses were conducted during summer, 10 during fall, 1 1 during winter, and five during spring. For the species represented by at least 30 individuals during a given season, I u.sed a test to test the null hypothesis that shorebird distribution was independent of cover type. I 786 THE WILSON BULLETIN • Vol. 108, No. 4, December 1996 performed these tests separately for each season. In a few cases, the expected number of individuals on both plowed and <10 cm vegetation was less than five; in such cases, the numbers of shorebirds observed and expected were pooled for these cover types, allowing for a valid test (Sokal and Rohlf 1981). Each total value was the sum ofthree “component X"” values, one for each of the three cover types. If the total x^ exceeded the critical value (rejecting the null hypothesis), then each “component x^” was examined individually to determine which of the shorebird-cover associations were significant. If a “component x^” for a single cover type was so high that it exceeded the critical value required for significance for the entire test, then the association between the species and the cover type in question was deemed significant. This method provided a conservative approach to as- sessing individual species-cover associations. The nature of each significant association was identified by determining whether the number of individuals observed on a cover type was greater than or less than the number expected on that cover type (indicating a positive or negative association, respectively). Standard y^ tests assume independence of individuals, an assumption that may be violated in species, such as shorebirds, that frequently move in flocks. In this study, I was rarely able to determine the number of independent groups of shorebirds that selected a field for foraging. However, to treat the entire shorebird assemblage in each field as a single unit composed of interdependent individuals may unnecessarily reduce the number of observa- tions available for analysis if, indeed, the shorebirds individually display their own cover preferences. In order to determine whether cover use may have been affected by social interactions, I examined the results of the x^ tests a posteriori. If flocking, rather than se- lection of a desired cover type, had been the primary influence on the birds’ choice offields, 1 would have expected a preponderance of insignificant, weak, or inconsistent shorebird- cover associations. However, the consistent, strong cover associations observed in this study indicate that social interactions did not mask actual patterns of cover preference or avoid- ance, supporting the validity of the x^ tests as they were conducted. I conducted an analysis of variance (in SYSTAT, Wilkinson 1990) to test the null hypotheses that mean species richness and shorebird abundance per census did not differ among the seasons. RESULTS — Relationship between shorebird abundance and tidal height During . the four dawn-to-dusk surveys, those species thought to prefer foraging in intertidal habitats (field exploiters) were most abundant on croplands during the higher portion of the tidal cycle and were scarce or absent during the lower portion (Fig. 1). During these surveys, such species included Semipalmated Plover, Black-bellied Plover, Whimbrel, Willet, Ruddy Turnstone, Semipalmated Sandpiper, Least Sandpiper, and Dunlin (scientific names in Table 1). Abundance of these shorebirds in fields at high tide and low was significantly higher during censuses conducted near high tide on 29 May {t = 2.670, df = 5, P < 0.05), 12 August {t = 3.940, df = 6, P < 0.01), 31 August {t = 4.255, df = 7, P < 0.01), and 2 November (r = 2.449, df = 6, P < 0.05). According to Fig. 1, relatively few field exploiters were present on censuses conducted more than two hours preceding or following high tide, although peak abundance did not always coincide with the census conducted nearest high tide (e.g., 29 May). Peak shorebird abundance occurred at different times during the Rottenborn • AGRICULTURAL FIELD USE BY SHOREBIRDS 787 2 NOVEMBER Fig. 1. Plots of shorebird abundance vs time of day. Solid circles represent individuals of “field exploiter” species, while open circles represent “field specialist” species (see “Methods” for descriptions of these two groups). Labels on the lower axis indicate hours after midnight. Solid and open arrows along the lower axis signify the times of high and low tides, respectively. different surveys, indicating that these dynamics were related to tidal height and not time of day. In contrast, the abundance of species thought to forage almost exclu- sively in fields (field specialists), including Killdeer, American Golden- 788 THE WILSON BULLETIN • Vol. 108, No. 4, December 1996 r^OOOOOOOOO O O O O O O O — o IT) O <N o 00 (—N r- CM CM . m m O — — o — On Gv ON 00 (n Tmj* (N — rN O On sD in in ON in <n (n in ON C—N 00 in (N in OOOOCNOOO TtCN(N — (NOOOOO — O — o o — o o NO in CN c—n no in Season nd CmN 00 in in CN and ^OpppOOO NOininONNO^d’^dcndcndmocNo^ o— ddddoooo Species C t^cNONTtcN — oo H o by o m o 1 2 1 4 4 4 1 1 1 1 Fields "o 3f3 0C0N C—N OrO^ni G cn ON N0O0 N7tO in <r-n- m (n — CiNn "O ON On CN CN Z6'>5 NO CN 1 Agricultural Om-^OOOOONONONDONNO •rfTi*-^fncNm»— — ^ — O m Table in Foraging Shorebirds 5 o Cj 5 of -^c is ^ 5 ^ O -Cs) S: 53 O S. 2 S '>2 ro; 0R0 •2 5 s Abundance •S :Ucgs .C1^ -".gS5. -S s^; >1 Sos:'i -.R22 ^^ •"R2o R& •KSS 'CCcS!: s "0oS0 Coiu(>uU i'<2SC3s3;. 50•gu«OuM,o ^Ug§2g T0"DQacc««ED0S3. sQ=Oo^uo: :"u-CJDw25OJ. ^EC>goCOC2:1ll s 2COcocL "U7^Cu(a33U. .*7fa0c903e).- C7cuOCCC0Q3n3LL hCOO£.APD =>S^H -.^'cCLs5(§cgis5UU)l :CuQCgj. tCo>oCCS/S135L « I’7wr0(QECS30UO.'. .co^<0RRCcEnD0- ^ Qc=3 ^CEl 22D(u aEu. Is^oS: 0UOo<30 ^—>>4-« '<0O(Eu3U F ^ 7DOC033. Jo(yUi 6^LSP.3 .x^LrEE:-j CJ„S1^^3) L5SO>X33' _T=3 2-S3 UiEo ca c/5 1/3 0, C/5 Oi QQ Rottenborn • AGRICULTURAL FIELD USE BY SHOREBIRDS 789 Fig. 2. Number of individual shorebirds foraging in agricultural fields during each cen- sus (Mar. 1991-Feb. 1992). Plover, Upland Sandpiper, and Buff-breasted Sandpiper, fluctuated little during each of the four dawn-to-dusk surveys (Fig. 1). There were no significant differences in the abundance of these species in fields between high and low tides on 12 August (t - 1.016, df = 6, P > 0.05), 31 August (t = 0.429, df = 7, P > 0.05), and 2 November (r = 1.000, df = 6, P > 0.05). None of these species was observed during the 29 May survey. — Species composition and seasonal abundance patterns. During the study of seasonal abundance patterns and cover use, I recorded a total of 21,254 individuals of 21 species of shorebirds foraging on croplands (Ta- ble 1). Shorebirds were observed foraging in agricultural fields on 24 of 30 censuses; no shorebirds were observed on two censuses in June and one each in April, July, January, and February. Only two individuals of one species (Willet) were observed during the four censuses in summer. Total shorebird species richness reached a peak of 18 species in fall, compared to 10 in spring and six in winter. This pattern also held for the mean number of species/census, being highest in fall {x = 8.4 ± 6.9 [SD]), intermediate in spring {x = 6.2 ± 2.7), and lowest in winter {x = 2.0 ± 3.0). In contrast, mean abundance/census was highest during spring ^ (x = 2300.2 ± 2320.8 individuals/census) and lower in fall (x 555.1 ± 536.1) and winter (x = 381.8 ± 663.7). ANOVAs confirmed that mean species richness/census (P = 24.5, P < 0.001) and mean shorebird abun- dance/census (F = 4.83, P < 0.02) did differ among seasons. In addition. 790 THE WILSON BULLETIN • Vol. 108, No. 4, December 1996 CENSUS DATE Eig. 3. Number of shorebird species foraging in agricultural fields during each census (Mar. 1991-Eeb. 1992). there was substantial variation in shorebird species richness and abun- dance within seasons as well. The four species which were recorded on the most censuses (Black- bellied Plover, Killdeer, Semipalmated Plover, and Dunlin) were also the most abundant species, accounting for 94.3% of all the individuals ob- served. Most species were represented by relatively few individuals. Only species were observed on more than four censuses, and only six spe- 1 1 cies were represented by more than 90 individuals. Whereas most species foraged in agricultural fields in only one or two seasons. Black-bellied Plover, Killdeer, Dunlin, and Short-billed Dow- itcher were present during spring, fall, and winter. There were no species recorded year-round, as the only species present during summer (Willet) was not recorded in fall or winter. Of the remaining eight species repre- sented by at least 30 individuals, four occurred strictly as fall transients (Pectoral Sandpiper, American Golden-Plover, Buff-breasted Sandpiper, and Upland Sandpiper), three were present only as spring and fall tran- sients (Semipalmated Plover, Semipalmated Sandpiper, and Ruddy Turn- stone), and one was present only during fall and winter (Western Sand- piper). Because few of the fields were irrigated and rainfall was generally lower than average during the study period (U.S. Dept, of Commerce 1991 standing water was rarely present in the fields. As a result, species ), that require open water, such as Greater Yellowlegs, Lesser Yellowlegs, Rottenborn • AGRICULTURAL FIELD USE BY SHOREBIRDS 791 and Solitary Sandpiper, were poorly represented on surveys. All other shorebirds usually foraged on relatively dry substrates, quite different from the saturated intertidal habitats with which the “field exploiter” species are mo—re commonly associated. Cover use. A total of 19 species foraged on plowed fields, 17 species on fields with <10 cm cover, and seven species on fields with >10 cm cover. Although the number of species recorded on plowed fields was only slightly higher than the number on fields with <10 cm cover, far more individuals were observed foraging on plowed fields than on the A other cover types. total of 17,714 shorebirds, or 83.3% of all birds recorded during the study, foraged on plowed fields. In contrast, 3432 individuals (16.2% of total) foraged on fields with <10 cm vegetation and 108 birds (0.5% of total) on fields with >10 cm vegetation. Of the species sufficiently numerous for analysis, all consistently showed highly significant positive assocations with plowed fields and negative assocations with fields with >10 cm cover except for Willet in spring and Upland Sandpiper in fall (Table 2). No species seemed actively to avoid plowed fields, and none was shown to prefer fields with >10 cm cover. The data for fields with <10 cm cover suggest that use of this cover type varied not only among species but also among seasons within a species. The proportion of species analyzed for cover use that showed positive associations with fields of <10 cm cover decreased from 0.71 in spring to 0.09 in fall and to 0.00 in winter. Conversely, the proportion of species negatively associated with fields of <10 cm cover increased from 0.14 in spring to 0.45 in fall and 1.00 in winter. For no species was the association with fields of <10 cm cover consistent among all seasons. DISCUSSION Agricultural fields on Virginia’s Eastern Shore seem to be important A foraging areas for migrating and overwintering shorebirds. few species, such as Killdeer, American Golden-Plover, Upland Sandpiper, and Buff- breasted Sandpiper, are known to prefer such habitats to intertidal areas (Stout 1967, Hayman et al. 1986). These field specialists were never seen moving from fields to intertidal areas, and their numbers did not seem to vary with tidal height. Most of the species observed in this study, how- ever, are known to prefer intertidal habitats when they are available. These field exploiters use fields as alternate foraging sites when mudflats and beaches are inundated by high water (Goss-Custard 1969, Page and Whi- tacre 1975, Gerstenberg 1979, Page et al. 1979). Data from the four dawn-to-dusk surveys indicated that the abundance of field exploiters on cropland was high only within approximately two hours of high tide. Therefore, the surveys conducted to determine sea- 792 THE WILSON BULLETIN • Vol. 108, No. 4, December 1996 Table 2 Cover Associations of Shorebirds Loraging in Agricultural Lields Total No. observed/expected on each cover type" ob- Species served Plowed <10 cm >10 cm Spring -- Black-bellied Plover 4664 3271/1323.1 1383/ 415.7 -b-b 10/2925.2 -- Semipalmated Plover 2822 2439/ 694.2 -b-b 233/ 209.4 ns 0/1918.4 - Willet 68 19/ 16.8 ns 27/ 7.9 -b-b 22/ 43.3 -- Ruddy Turnstone 51 38/ 10.0 -b-b 13/ 4.7 -—b+ 0/ 36.3 -- Semipalmated Sandpiper 502 483/ 108.7 -b-b 19/ 43.2 0/ 350.1 Dunlin 3252 2596/1150.3 -b-b 656/ 313.2 -b-b 0/1788.5 -—- Short-billed Dowitcher 138 108/ 45.1 -b-b 30/ 12.1 -b-b 0/ 80.8 Pall — -- ABlmaecrki-cbealnliGeodldPelno-vPerlover 203537 197500// 5278..27 --bb--bb 614// 3884..63 ns 32//1 14147..02 —-- Semipalmated Plover 1299 1122/ 246.1 -b-b 132/ 97.4 -—b 45/ 955.2 -- Killdeer 869 829/ 242.8 -b-b 38/ 177.7 2/ 448.5 Upland Sandpiper*’ 32 7/ 5.2 ns 4/ 2.5 21/ 24.3 ns -- Semipalmated Sandpiper 58 58/ 8.3 +-b 0/ 4.6 ns 0/ 37.2 -- Western Sandpiper 34 33/ 9.1 -b-b 1/ 5.2 ns 0/ 19.6 Pectoral Sandpiper 90 89/ 15.4 -b-b 0/ 5.9 —- 1/ 68.7 -- Dunlin 959 959/ 300.8 -b-b 0/ 262.6 0/ 395.6 - Short-billed Dowitcher 53 53/ 16.1 -b-b 0/ 11.8 0/ 24.1 -- Buff-breasted Sandpiper*" 31 28/ 4.2 +-b 3/ 2.0 0/ 24.8 Winter — — -- Black-bellied Plover 1285 1012/ 287.8 -b-b 273/ 755.5 0/ 241.7 — —-- Killdeer 169 145/ 28.8 -b-b 24/ 103.4 0/ 36.8 Dunlin 2700 2177/ 577.5 -b-b 523/1679.7 — 0/ 442.8 Short-billed Dowitcher 42 42/ 7.1 +-b 0/ 22.4 0/ 12.5 — ne"ganst.ivPe a>ss0o.c0i5a;tio+nsan{dP <++0,.0s5i,gnPifi<can0t.0p0o1sirteisvpeecatsisvoecliya)t.ions {P < 0.05. P < O.OOl respectively), - and , significant ^For these species, ob.served and expected numbers were pooled for plowed fields and fields with <10 cm to achieve expected values sufficiently high for testing. sonal abundance and cover use patterns probably underestimated the use of croplands by field exploiters, as roughly half of the fields in each survey were visited 2-4 h preceding or following high tide. Because there was no apparent relationship between the times at which fields were vis- ited and the cover types present in fields, this observation should have no effect on analyses of cover use. Shorebird abundance in agricultural fields showed substantial variation among censuses. Fluctuations in abundance within (rather than among) seasons may have reflected differences in tidal height during the census periods. More shorebirds may have foraged in fields during higher peak