Ecological Archives E095-298-A1

Mariana M. Vidal, Erica Hasui, Marco A. Pizo, Jorge Y. Tamashiro, Wesley R. Silva, Paulo R. Guimarães Jr.. 2014. Frugivores at higher risk of extinction are the key elements of a mutualistic network. Ecology 95:3440–3447. http://dx.doi.org/10.1890/13-1584.1

Appendix A. Details on methods employed to collect data on morphological traits and abundance of frugivorous birds and fruiting plants, results of regression analyses between species contribution to network structure and biological traits, biological correlates of extinction risk, and relationship between bird taxonomy and (i) species contribution to network structure and (ii) extinction risk.

Below, we present details on methods we employed to collect data on bird species morphological traits (body mass, bill length, bill width, bill depth), degree of frugivory and abundance. Similarly, we present the methods we used to collect data on plant species morphological traits (seed width, seed length, fruit width, fruit length), phenology and abundance. Aiming to evaluate the relationship between these biological traits and species' contribution to network organization in our study system, we performed regression analyses, using the different traits as linear predictors for the species' contribution to network organization. We also investigated potential biological correlates of extinction risk of birds, as well as the relationship between bird taxonomy (at the family level) and (i) species contribution to network structure and (ii) extinction risk.

Methods

Study site

We studied interaction networks between plants and frugivorous birds in three sites in Parque Estadual Intervales (PEI), a large protected area of continuous Atlantic rainforest in Southeastern Brazil. The average annual rainfall is around 1,700 mm and average annual temperature is approximately 18°C (FF/SMA-SP 2008). PEI comprises different forest physiognomies along an altitudinal gradient ranging from 100–200 m to 1000–1100 m above sea level (FF/SMA-SP 2008). Our three sampled sites are in an altitudinal gradient: at 980 m (hilltop), 847 m (middle slope), and 597 m altitude (bottom of valley). The predominant forest physiognomy in the three studied sites is Lower Montane Rainforest (Oliveira-Filho and Fontes 2000), but some structural differences can be observed among the sites. The hilltop site has forests similar to the middle slope site, where the canopy reaches 15–20 m high, but hilltop also has field physiognomies with treelets and shrubs, and some species typical from secondary forest. Bottom of valley canopy is 20 m high with emergent individuals up to 30 m (Hasui 2003). At the same sites where we collected data on plant-frugivore interactions, we also collected data on morphological traits and abundance of both birds and plants.

Bird species data

In each primary forest site (hilltop, middle slope and bottom of valley), one of us (EH) defined five sample units, which were from 10 to 1000 meters apart. In each sample unit, we set up 10 mist nets (36 mm mesh, 12 × 2.5 m) in line. We sampled each of the five units one day per month, during one year. Mist-nets were open from dawn to early afternoon (6 hours), totaling 3,600 mist-net-hours of sampling effort per site. Nets were checked at least once per hour. We sampled the middle-slope site from April 1999 to March 2000; the bottom-of-valley site from April 2000 to March 2001; and hilltop site from April 2001 to March 2002. All the captured birds were identified and biometric data was collected: body mass and bill length, width and depth. Then, the captured birds were kept in a cloth bag for at least half an hour, when the bags were inspected for fecal material, which was carefully inspected for fruits and seeds. This approach does not provide a detailed description of the frugivory interactions, which we assessed through direct observation. However, fecal material allowed us to characterize birds with respect to their degree of frugivory: high frugivory (over 50% of the fecal samples contained fruits and seeds); and low frugivory (over 50% of the fecal samples contained insects). When few individuals of a given species were captured, museum data were included in the morphological characterization of the species whenever possible. Regarding data on avian body masses, we used information from literature (Dunning 2008) to complement the database, whenever field or museum data was missing.

We used data from mist nets to estimate bird species abundance. This method is likely to underestimate the abundance of canopy and less active species. To circumvent this problem, we also used data from literature to make a second estimate of species abundances. Vielliard and Silva (2001) exhaustively sampled our bottom-of-valley site, by employing a bioacoustics method. Sampling was performed in 22 visits, each one lasting from two to four days, averaging 23 hours of sampling per visit. In each visit, the researchers recorded every seen or heard bird in a fixed point, for 20 minutes. The mean number of contacts of each species per sampled point count is hereafter called the punctual abundance index (PAI), which allows for quantitative comparisons among species (Vielliard and Silva 2001).

Plant species data

In each of the five sample units, we set up three plots (10 × 25 m) along the mist-nets line, where all the endozoochorous plants were sampled for characterization of plant composition and abundance. For phenological data, all the plots were monthly monitored for one year; at the same time birds were captured in the mist-nets. Fruiting plants were counted, identified and individually marked. Information on morphological traits of fruits and seeds (length and width) was collected for the majority of plant species with bird dispersal syndrome. Measurements were made for at least ten fruits, collected, whenever possible, from different plants.

Linear regression analyses

In order to investigate the biological correlates of species' topological importance, we performed linear regression analyses between the species contribution to network structure and biological traits that potentially mediate frugivory interactions. For birds, we investigated the association between the species importance to network structure and body mass (log-transformed), bill length, bill width, bill depth, degree of frugivory and abundance (both abundances estimated from mist-nets and PAI). For plants, we explored the relationship between the importance to network structure and seed width, seed length, fruit width, fruit length, length of fruiting period and abundance. We standardized abundance data by calculating the z-score for each group (birds or plants) at each site:

 

 ,                                                                                                               (A.1)

 

where  is the abundance of plant or bird species,  and  are the mean and standard deviation of abundance values of plants or birds at each site, respectively.

Correlates of extinction risk

We investigated the biological correlates of extinction risk, randomizing bird traits (log-transformed body mass, bill dimensions, local abundance and degree of frugivory) between higher- and lower-risk threat categories (Manly 1997). We computed the P value as the proportion of the differences of means between lower- and higher-risk species in the randomizations that were equal to or greater than the observed difference of means between lower- and higher-risk species.

Bird taxonomy, extinction risk and species contribution to network structure

To investigate the relationship between network organization and evolutionary history, we evaluated the relationship between taxonomy and bird species contribution to network organization. We randomized the values of contribution among species and calculated the mean contribution of each family, comparing it with the actual mean. The P value was computed as the proportion of mean contributions of each family in the randomizations that were equal to or greater than the observed mean for the family.

Seeking to evaluate the association between taxonomy and the risk of extinction, we randomized the families between bird species and calculated the proportion of higher-risk species in each family. Comparing the proportions generated by the randomizations with the actual proportions, we assessed if higher-risk species were predominant in species-rich families, more than it would be expected by chance. In this sense, P value was computed as in the previous randomization tests.

Results

Linear regression analyses

Significant effects of biological traits on species contribution to network structure are in the main text (Fig.3). Here we present all the results, including significant and nonsignificant effects (Table A1 and Figs. A1 and A2).

Table A1. Summary of linear regression analyses between species contribution to network structure and a variety of biological traits of both birds and plants. Significant P values are marked with *. PAI = punctual abundance index.

Group

Trait

Coefficient

R²

F statistics

df

P value

Birds

Abundance (number of ind)

0.37

0.03

0.5

1,16

0.47

 

PAI

1.10

0.01

0.5

1,43

0.49

 

Body mass (log)

-0.15

0.01

0.6

1,55

0.44

 

Bill length (mm)

0.00

0.00

0.1

1,42

0.82

 

Bill width (mm)

-0.02

0.00

0.1

1,42

0.73

 

Bill depth (mm)

-0.01

0.00

0.1

1,42

0.79

 

Degree of frugivory

0.47

0.01

0.6

1,48

0.45

Plants

Seed width (mm)

-0.14

0.07

5.4

1,76

0.02*

 

Seed length (mm)

-0.08

0.06

4.6

1,76

0.03*

 

Fruit width (mm)

-0.03

0.01

1.1

1,75

0.30

 

Fruit length (mm)

-0.01

0.02

1.3

1,75

0.26

 

Length of fruiting (months)

0.25

0.11

4.7

1,40

0.04*

 

Abundance (number of ind)

0.66

0.09

3.3

1,32

0.08

 

FigA1

Fig. A1. Relationships between bird species contribution to network structure and biological traits: abundance, index of abundance (PAI), body mass (log-transformed), bill width, bill length, bill depth, and degree of frugivory. None of the analyzed traits were significantly associated to species contribution to network structure (P > 0.05 for all comparisons).


 

FigA2

Fig. A2. Relationships between plant species contribution to network structure and biological traits: seed width, seed length, fruit width, fruit length, length of fruiting period, and abundance. The line of regression, the R² and P values indicate significant associations.


 

Correlates of extinction risk

Body mass (log-transformed) was significantly associated to threat categories (Table A2), the higher-risk bird species showing larger body masses than lower-risk species. Bill width was marginally associated to threat categories, the higher-risk species with wider bills than lower-risk birds (Table A2). All other variables were non-significantly associated with threat categories.

 

Table A2. Association between bird species threat categories (lower and higher) and studied biological traits. We do not have information on biological traits for all bird species comprised in the interaction networks, so we specify the number of species included in each randomization test. The P value corresponds to the proportion of the differences of means that were equal to or greater than the observed difference of means. Significant P values are marked with *.

Bird trait

Number
of species

Number of
higher-risk
species

P value

Abundance

18

2

0.90

PAI

45

8

0.12

Body mass (log)

57

9

<0.01*

Bill length

44

7

0.09

Bill width

44

7

0.06

Bill depth

44

7

0.14

Degree of frugivory

50

7

0.22

 

Bird taxonomy, species contribution to network structure and extinction risk

 We found no significant relationship between bird taxonomy (at the family level) and species contribution to network structure (Table A3). Among the bird families comprised in the studied interaction networks, only Cotingidae has more higher-risk species than it would be expected by chance (Table A4).

Table A3. Mean values of contribution to network structure by bird family. The observed mean contribution refers to the actual mean contribution per family, whereas the mean contribution across 1000 randomizations refers to the average contribution of each family after randomizing species contributions among all the bird species (1000 iterations). The P value corresponds to the proportion of mean randomized contributions equal to or larger than the observed mean.

Bird family

Mean contribution
(observed)

Mean contribution
(1000 randomizations)

P value

Cardinalidae

-0.86

0.05

0.77

Columbidae

1.44

0.02

0.20

Cotingidae

0.81

0.01

0.11

Cracidae

-1.76

0.02

0.96

Emberizidae

-0.04

0.00

0.50

Icteridae

0.33

0.02

0.37

Momotidae

-2.72

-0.02

1.00

Picidae

-0.85

0.08

0.69

Pipridae

0.41

0.01

0.28

Psittacidae

-0.82

0.01

0.74

Ramphastidae

1.01

-0.03

0.12

Thraupidae

0.31

-0.01

0.17

Tinamidae

-2.14

0.03

0.96

Trogonidae

-0.72

-0.02

0.78

Turdidae

0.79

0.01

0.17

Tyrannidae

-0.33

-0.02

0.71

Vireonidae

-1.13

0.05

0.85

 

Table A4. Total number of species and percentage of higher-risk species in each bird family. The observed percentage of higher-risk species refers to the actual percentages per family, whereas the mean percentage across 1000 randomizations refers to the average percentage of higher-risk species in each family after randomizing the families among all the bird species (1000 iterations). The P value corresponds to the proportion of randomized families with percentages of higher-risk species equal to or larger than the observed percentage. Significant P values are marked with *.

Bird family

Number
of species

Percentage of
higher-risk species
(observed)

Mean percentage
of higher-risk species
(1000 randomizations)

P value

Cardinalidae

2

0.00

0.16

1.00

Columbidae

1

0.00

0.17

1.00

Cotingidae

6

0.67

0.16

<0.01*

Cracidae

2

0.50

0.16

0.30

Emberizidae

1

0.00

0.15

1.00

Icteridae

1

0.00

0.16

1.00

Momotidae

1

0.00

0.16

1.00

Picidae

1

0.00

0.16

1.00

Pipridae

3

0.00

0.16

1.00

Psittacidae

2

0.00

0.15

1.00

Ramphastidae

3

0.33

0.16

0.40

Thraupidae

17

0.12

0.15

0.81

Tinamidae

1

1.00

0.16

0.16

Trogonidae

3

0.00

0.16

1.00

Turdidae

4

0.00

0.15

1.00

Tyrannidae

8

0.00

0.16

1.00

Vireonidae

2

0.00

0.15

1.00

 

Literature cited

Dunning, J. B. 2008. Body Masses of Birds of the World. Pages 3–554 in J. B. Dunning, editor. CRC Handbook of Avian Body Masses. Second Edition. CRC Press.

FF/SMA-SP (Fundação Florestal/Secretaria de Estado do Meio Ambiente-SP). 2008. Plano de Manejo do Parque Estadual Intervales. FF/SMA-SP, São Paulo, Brazil.

Hasui, É. 2003. Influência da variação fisionômica da vegetação sobre a composição de aves na Mata Atlântica. Ph.D. dissertation, Universidade Estadual de Campinas, São Paulo, Brazil.

Manly, B. F. J. 1997. Randomization, Bootstrap and Monte Carlo Methods in Biology. 2nd edition. Chapman & Hall/CRC, London.

Oliveira-Filho, A. T., and M. A. L. Fontes. 2000. Patterns of floristic differentiation among Atlantic Forests in Southeastern Brazil and the influence of climate. Biotropica 32:793–810.

Vielliard, J. M. E., and W. R. Silva. 2001. Avifauna. Pages 125–145 in C. Leonel, editor. Intervales. Fundação para a Conservação e a Produção Florestal do Estado de São Paulo, São Paulo.


[Back to E095-298]