Migrant birds are the main seed dispersers of blackberries in southern Spain Pedro Jordano Between-habitat variation in fruit production and bird attributes enhancing seed ingestion and removal were studied to describe the avian seed-dispersal system of

Дата канвертавання19.04.2016
Памер0.56 Mb.
  1   2   3
Migrant birds are the main seed dispersers of blackberries in southern Spain

Pedro Jordano

Between-habitat variation in fruit production and bird attributes enhancing seed ingestion and removal were studied to describe the avian seed-dispersal system of blackberries (Rubus ulmifolius, Rosaceae) in southern Spain.

Migrant birds were largely responsible for seed dispersal. Among 20 passerine species recorded feeding on fruits, five removed the bulk of seeds: Sylvia arricapilla (29.6% of 897 visits recorded), Erithacus rubecula (15.2%), S. bonn (14.7%), S. rnelanocephala (10.1%) and Turdus merula (7.7%). During an average day 32400 seeds left a parent clone through the activity of these species.

Seed production was density-dependent, being maximum in high-density situations. Seed removal (i.e., fruit consumption) from individual clones was dependent on crop

size, habitat occupied and ripening phenology. Clones ripening a small crop later in the season and/or in low-density habitats had a lower fraction of the crop consumed

than did those producing huge crops in high density habitats and/or synchronously with autumn bird migration.

The small, passerine birds presumably contributed to the bulk of seed dispersal because they (1) feed on Rubus fruits extensively, (2) showed high visit rates and removed a very high fraction of the seed crop from parent clones, (3) did not damage the seeds nor drop them beneath the parent clone and (4) performed species-specific flights to apparently safe sites for the plant, thus enhancing its colonizing ability.

P. Jordano, Estación Biológica de Doñana, Unidad de Ecologla y EtologIa, Sevilla-12, AndalucIa, Spain.

MccnerpBarr11 TcmqeclGIe 3YH5I B rrnya. ruirp H Hacenef rrnii, nOT— PIIBQ45X H pa3HOCIIsx cHa, C eimio aHasIw3a CHCT&4,I pacrIocTpaueuH5I

C3H eICeBHKH Rubus u imifo hue Rosaceae, nriar B KBcHo llcriaHHH. flepene­

Tit rimr axmnHo Y’aCTDY1OT B pacnpDcipauenn CH. Cpe,ri 20 BI51TB Bopo­

rbH!W(, y ROTOX 3aPerHCTPHIXBaHO rnrae rurrv, 5 pacnxxTpaH7r CC—

I-OBHyIO rccy cH: Sylvia atricapihla (29,6% M3 897 BaperMcTpwpDBamwc ncIi), Erithacus rubecula (15,2%), S. bonn (14,7%), S. melanocepha— la (10,1%), Turdus rnerula (7,7%). B cpew 3& rtei- rIp1iriepE1o 32400 Ce—

Y1-1OCMTC C THTe3TbCKHX KflOHOB aXTHBH0CTh 3fl4X flxnyi*i CH 3aBHcwr oT TIflOTHQCTH H pcmrae KCIlM! B CYIy’4aIX cYIeHb a1DoKo rmarHoc. M1,rrHe cw (io—ecr±,, no e6JIeHHe rmQLpB) C OITth FLDC

OHOB 3aBl4cwr o’r eiiii yrxYcaa, xapxepa cIoowraHHa a

copea. B KnoHax, aiaix HeonHo ycm a 6onee nos-!ee ap.1a a! HJTh{ B I cToo6aTaHaax C 6onee Ha3KO ruIomcxDmlo noTpe6rr5IeTc5 ?i ypcast, B TX, re copeae BbcOKHI yX 11PM PDOKO rUIOTHC>DTH a/

CHHXXJHHO C Ha’-LarIa1 OCeHHeI’O riepenea rrrm.

ManeIIbxHe BObHfL1 nTH11 nO-DIry 3CrTXYDTpaH5aI OCHOBBYIO qam Ce—

N, T K. o 1. aicmio riwraiarcs rncnai1H eici, 2. ‘iacro rrxeia&yr pac—

Te1-u{ H M3bz1aJYr BeCbr 3H reIHYO aCTh ypaH CH C QLWJtbCKHX KflO­ F1DB, 3. He c&Ha H He paaiacrr MX PHJ4 C jLtMTeJCKMMM KJDHaMH, 4. ocecmncuo’r BoCne Iwitje nonemi B OTH0CHTeJmH0 6eciiacie rn paCTeH ecToo6wraHHa H TKHM oôpa3a. CXXCThYKJT jx KOnCHH9aLtM3.

1. Introduction
Fleshy-fruit production by plants and fruit-eating by birds are prominent behaviours in both the tropics (Snow 1971. Morton 1973) and temperate areas (McAtee 1947. Turcek 1961). Closely coevolved re­ lationships have been proposed mainl among tropical plants and birds, apparently reciprocally specialized in mutual exploitation for seed dispersal and fruit-food gathering. Extra-tropical plants and birds have been viewed as exemplifying, at the opposite extreme, loose relationships based on mutually opportunistic be­ haviours (McKey 1975, Howe and Estabrook 1977). However, this theoretical background was built up on the basis of scanty field evidence, in particular from temperate habitats. Thus, several discrepancies with in­ itial predictions on fruit-bird interactions arose when data on non-tropical systems begun to emerge (Frost

1980, Herrera and Jordano 1981, Herrera 1981c). Most of the initial work on bird-plant coevolution for

seed-dispersal lacks documentation of the actual de­

pendence between the system’s components, and only infrequently have variations in plant and bird traits re­ lated to the dispersal process been studied to clarify selective forces operating upon the system. Bird activity may generate relevant selective pressures on fruiting phenologies (Snow 1965, Thompson and Willson

1979), crop sizes (Howe and Kerckhove 1979) and fruit characteristics (Herrera 1981a, Howe and Kerckhove

1981), thus influencing the overall reproductive pattern of the plant.

I deal here with between-habitat variation in fruit production and fruit features of blackberries, Rubus ulmifolius (Schott), in Southern Spain. Clones of R. ul­ mifolius, henceforth referred to as Rubus, are com­ monly found along creeks and water courses, often spreading by vegetative expansion over relatively mesic sites in man-made disturbed areas (e.g., Abrahamson

1975). The considerable within- and between-habitat

variation in stem and leaf form, flower features and fruit size and seediness showed by species within the genus Rubu.c, was long ago quoted by Darwin (1859) as an instance of variation in natural populations.

I try to describe the avian seed-dispersal of this shrub and those traits of the bird species that provide removal of the seeds from the parent plant and enhance coloni­ zation of new habitat patches.

2. Study area

Field work was conducted at Finca El Bañuelo, Pro­

vincia de COrdoba, south-central Spain (37° 55’N,

4° 48’W) between 22 Jul and 25 Nov 1978. Mean an­

nual rainfall during the dry period (from Jun through Sep) is 82.0 mm, and 575.4 mm during the humid period (rest of the year). Mean temperature of the hot­ test and coldest months are 26.6°C (Jul) and 7.9°C

(Dee) (data from the meteorological station of Espiel­ Central Térmica, 20 km NNW of El Bañuelo).

The study area is located in a small valley at 500 m a.s.l., with the lower parts in olive-tree culture and orchards. In the uncultivated portions of this zone. Rubus grows in dense thickets along a small. seasonal stream. Scattered individuals of Ulmus sp.. Rosa sp., Crataegus monogvna (Jacq.), Picas carica EL.), Celtis australis (L.) and Tamus communis (L.) can also be found in this habitat, which is hereafter named the Grove. A portion of the Grove (referred to as Burned habitat) is an old field recovering from a recent burning in Aug 1977 and Rubus clones are there sparsely distri­ buted among some fig and olive trees. The surrounding hillsides are covered by closed Mediterranean oak forest vegetation (Forest habitat hereafter), with Quer­

cus suber (L.), Q. rotundifolia (Lam.) and Q. faginea

(Lam.) forming the tree stratum, mixed with reforested

Pinus pinea (L.). The undergrowth is composed mainly of Cistus spp. shrubs and sparsely distributed fleshy- fruit producing species, mainly Viburnum tin us (L.), Arbutus unedo (L.), Pistacia lentiscus (L.), P. terebin­ thus (L.), and Lonicera implexa (Aiton). The Rubus is found scattered on the lower parts of the hillside, close to the Grove.

3. Methods

3.1. The plants
The position of Rubus clones exceeding 0.2 m2 in basal area were mapped and 19 clones of three sizes (small,

<10 m2, medium, 10—100 m2 and large, >100 m2) were marked in the three habitats. Basal area and volume were estimated for each clone assigning a geometrical figure and taking the corresponding measurements (Walsberg 1977). Fruit crop was estimated from direct raceme counts in small and medium-sized clones and the number of fruits produced was calculated as: Total number of racemes x mean number of fruits raceme’ (the last figure calculated from a sample of marked racemes, see below). The fruit crop of large clones was estimated as: (mean number of racemes m2 of exposed surface) X total exposed surface (calculated from the above measurements) x (mean number of fruits raceme’) if a direct count of racemes was impractica­ ble. Mean no. of racemes m2 was obtained by regular sampling of the clone’s surface with a frame 1 m2 and recording the number of racemes included in the frame in each of 20 counts per clone.

Fruit characteristics were determined for a sample of fruits taken from six clones of the three habitats. Fresh fruits were weighed individually to the nearest 0.01 g and then desiccated for 24 h at 100°C to obtain dry weights and number of seeds fruit-1. Seed dry weight of individual fruits was calculated from the regression of mean seed dry weight for individual clones vs. mean no. seeds fruit—’: seed dry weight = 0.005—6.45 l0 no.

seeds fruir’ (r = —0.857. P < 001, n = 8). This al­ lowed an estimation of individual seed weights for fruits of a given seediness, as it was extremely difficult to separate pulp from seeds due to the small size of the drupelets. Thus, total seed dry weights (i.e. seed loads) were obtained by multiplying the estimated seed weights by the known seediness, pulp dry weights being obtained by substraction of seed loads from fruit dry weights.

At the beginning of the fruiting period a total of 6625 potential fruits (171 racemes) were marked in 12 clones, either as buds, flowers or unripe fruits. From 22

Jul through 24 Nov 1978, 20 counts (once a week) were carried out on marked racemes, recording the number of buds, flowers, and unripe, ripe, unripe-desiccated, ripe-desiccated, pecked and pecked-desiccated fruits in each raceme.

3.2. The birds

Data on bird activity in relation to Rubus were obtained by mist-netting and by direct observation of individual clones. Mist nets were operated weekly between 19 Aug and 4 Nov, yielding 708 net-hours. Nets were situated among Rubus shrubs in the three habitats and remained open from dawn to dusk. Nets were emptied hourly and birds caught were measured, weighed and released. Net, hour of capture and side of the net in which the bird was found were recorded as well. Samples of the gastroin­ testinal content were obtained by flushing physiological sodium chloride water solution through the digestive tract (Moody 1970). Faecal samples were stored indi­ vidually in filter paper and air-dried for later analysis. Percent sample volume made up by animal and vegeta­ ble matter was estimated visually to the nearest 10%. Fruit skins were identified by comparison of micro­ scopic slides with a reference collection of skin micro- photographs (see Herrera and Jordano 1981 for de­ tails).

Direct observations were carried out at a clone (G2, see below) in the Grove from a hide 10 m away on the side of the clone facing the Forest hillside. Five half- hour observation periods were distributed throughout the daylight hours in weekly censuses between 6 Sep and 9 Nov and the identity of all birds flying to and/or from the plant was recorded. Whenever possible, flight directions (along the Grove or to the Forest) of birds leaving the clone and the relative location of the first perch utilized (‘distant’, if >20 m away or ‘close’, if <20 m away) and its identity (Rubus or a perch above Rubus, or another species) were also recorded. Addi­ tional observations were carried out on clones MS and P1, with smaller fruit crops (see Results), for 3.5 and

3.0 h respectively. The number of bird visits d’ was

calculated from mean visits h, allowing for 12 h of daylight during the period of field work.

Between the half-hour census periods time was de­

voted to observing the feeding behaviour of individual
birds at Rubus. Whenever possible, the following data were recorded: (1) total time the bird was under observation (either in complete’y or partially observed visits), (2) total time the bird remained stationary, (3) number of fruits pecked, (4) number of pecks made in the observation period and (5) method of taking fruit (see Results). In each attempt (peck) the birds ingest an unknown number of drupelets. each one containing one seed. To estimate the average number of seeds taken during a visit I first calculated the number of seeds mint, scaling specific feeding rates (pecks min’ and pecks fruirj with respect to T. merula average feeding rate (30 seeds peck—1 and 120 seeds min1) and taking

30 seeds fruir’ as the average seediness for fruits from the clone G2:

Seeds min’ (species i) Seeds min’ (T. merula) x

pecks min (T. merula)

pecks min (sp. i)

These ratios are equivalent to weight ratios among the species involved, allowing for the calculation of seeds min1 for each species, provided body weight is signifi­ cantly correlated with measures of feeding rates (see Results). I then estimated seeds visir’ for the species i from data on visit length and seeds min’. This provides a way to accurately estimate seed ingestion rates for birds feeding on multi-seeded fruits (such as poly­ drupes, e.g., McDiarmid et al. 1977) for which one peck does not represent one whole fruit ingested and one seed processed.
4. Results
4.1. The plant population
The Rubus population at El Bañuelo is characterized by an abundance of small, 5—10 m3, vegetatively expanding clones with crop sizes between 102 and iO fruits and some clones with extremely large crops (105_106 fruits on clones of 100—1000 m3). The clones are distributed on a narrow corridor along the stream course and lower parts of the adjacent hillside (see below). Both clonal size and density are higher in the Grove, where medium-sized (250 m3) and large (l100 m3) clones are common. In both the Forest and Burned habitats, only medium (30—70 m3) and small (<10 m3) clones can be found probably because increased shading and depth of water table and the recent disturbance have limited Rubus expansion in these areas.

Fruit crop increases with clone size (r = 0.86, P <

0.01, n = 18) but for a given size, crops of the Grove clones are much greater than those of the Forest and

Burned habitats. Thus, among the small clones found in

the study area, those growing in the Grove consistently had crop sizes >iO fruits vs. <700 fruits on clones of the same size found in the Forest and Burned habitats. For a given clone size, fruit production is maximum in the Grove while for a given crop size, clone size is maximum in the Forest and Burned habitats.


Tab. L Characteristic features of Rubus ulmifolius fruits in the three habitats studied. Means ± one s.d., sample size in parentheses. Seed load seed weight X no. seeds fruit—’, seed weight being estimated from regression of seed weight (mean values for individual clones) on no. seeds fruir’ (see Methods). a = dry weight (g).
Grove Forest Burned habitat

Fresh weight (g)


0.31 (99)


0.26 (10)

0.52±0.29 (31)

Dry weight (g)


0.08 (99)


0.07 (10)

0.15±0.07 (31)

Percent water

75.5 ±

3.8 (99)


7.9 (10)

67.9 zS..3 (31)

Seed weight (g X


2.6 ±

0.4 (21)

1.8 ±

0.2 (15)

4.2 0.5 (15)

No. seeds fruit—’ 33.5 ± 14.1 (99) 30.0 ± 14.0 (10) 11.8 ±.1


Seed loada 0.08± 0.01 (82) 0.07± 0.02 (10) 0.04±0.02 (31) Pulp loada 0.14± 0.09 (99) 0.10± 0.08 (10) 0.11±0.05 (31)

Percent pulp 64.0 ±19.7 (99) 53.9 ±23.6 (10) 69.5 ±S.9


4.2. Fruits

The unripe fruits are green and take on a purple-red coloration just before ripening, preceding the jet-black colour of the ripe polydrupes. They are 15.4 ± 2.4 mm long and 15.6 ± 1.8 mm wide (n = 28) and have a variable number of drupelets inserted on a dry recepta­ cle. Fruits average in the study area 0.73 ± 0.33 g fresh weight and 0.19 ± 0.08 g dry weight, with 73.1 ± 6.7% water content (whole fruit). Mean seed weight is 3.2 ±

0.9 mg and there are 28.8 ± 15.5 seeds fruir1. Thus,

the average fruit contains 94.2 mg of pulp dry-weight, representing 12.9% of the fresh fruit.

There is a high variability in fruit characteristics be­ tween the three habitats studied (Tab. 1). These di­ vergences appear as the results of fruit design-related compromises: fruit seediness is inversely related to seed weight (r = —0.82, P < 0.01, n = 9) and positively re­ lated both to pulp dry weight (r = 0.58, P < 0.05, n = 9) and fruit weight (r = 0.77, P < 0.001. n = 9). Thus, fruits in the Burned habitat have seed loads simi­ lar to the Grove fruits (Tab. 1), but half the seed number and double the individual seed weight. Fruits of the Forest occupy an intermediate position along the gradient, with seediness and seed size similar to the Grove fruits. On the other hand, Grove fruits offer a greater absolute amount of pulp to the dispersers (141.1

Tab. 2. Summary of statistics from multiple-group discriminant analysis performed on characteristics of Rubus ulmifolius fruits grouped by habitats. For each variable, F values when entering the discriminant function at the corresponding step are given.

mg dry wt fruit—1) than Burned habitat fruits (105.6 mg fruit ‘) but the relative amounts of pulp fruit1 are similar in the three habitats (Tab. 1, t 1.50, P> 0.2, for the three possible comparisons).

Patterns of between-habitat variation in fruit charac­

teristics were investigated by means of multiple group discriminant analysis (BMDP7M, see Dixon 1975). Results are summarized in Tab. 2. Fruit size and seedi­ ness are positively associated to Canonical Variable I and both relative weight of pulp and water content have negative loads on it. This axis accounts for major differ­ ences between Grove and Burned habitat fruits. Ca-


  1   2   3

База данных защищена авторским правом ©shkola.of.by 2016
звярнуцца да адміністрацыі

    Галоўная старонка