Скачать 37.49 Kb.
|SUPPLEMENTARY MATERIAL FOR RYDER ET AL.|
Coded Nanotags and Proximity Data-logger:
Coded Nanotags are digitally encoded VHF radio tags that emit a unique numerical code while running on a single frequency. Male wire-tailed manakins were fitted with 0.35g (NTQB-2, Lotek Wireless) tags using the Rappole harness technique and soft elastic thread . Our tags ran on a single frequency 166.340MHz with a burst rate of 4.9 to 5.1 seconds and a battery life of 30 days. Decreasing burst rate and utilizing 12 hr on/off programing, both of which are manufacturing options, can extend battery life. Tags can also be customized for different kinds of harness attachment (e.g., tubes on front/back, sides, etc.). Animal recapture or standard telemetry homing approaches with a coded receiver can be used to recover deployed tags. The manufacturer does not currently offer tag refurbishing.
For this study, tags were monitored with a digitally encoded proximity data logger (SRX-DL1, Lotek Wireless) that runs on a 12V battery source. Depending on battery capacity, the data logger can run autonomously for up to two weeks and its memory can hold 272,000 detections. The SRX-DL is available is several configurations with more advanced models including multiple frequency scanning, additional antenna and collection of sensor data (e.g., temperature, pressure, electromyography). Additional receivers (SRX-400 and SRX-600) with greater memory storage, built in GPS receiver, internal batteries and data condensing are also available. Here we used a single SRX-DL receiver, although use of multiple receivers in a grid like fashion could provide more detailed spatial data using triangulation and would undoubtedly increase the number of tag detections.
Classification of male status:
Wire-tailed manakins are long lived and males follow age-graded social queues for status [2, 3]. Once achieved, territorial status is stable over time. Ryder et al. have studied this specific population of wire-tailed manakins for 10 years. Therefore, we were able to use both prior and current data sources to determine male status prior to deployment of the coded Nanotags. Prior work on this population included extensive scan sampling of each male territory, which yielded 1509 resights of over 200 different males, during which time the social status of many males was determined and followed over time [2-4]. Previous work (2003-2006) also used two-hour continuous focal sampling (414 hours; mean = 16.2 ± 0.89 hrs/male, range: 8-20 hrs/male), which proved to be sufficient in determining male social status . The two focal leks for the current proximity data-logger study were a subset of previously studies leks, and many (55%) of the territorial males observed during this period were still present. During the present study, we utilized the less labor-intensive scan-sampling approach to confirm each male’s status. Specifically, we collected 46 independent resights of territorial males and 8 resights of floaters at our focal leks (see Table S2). By combining prior knowledge with a detailed understanding of the wire-tailed social system, we were able to unambiguously assign males to status categories for network analyses.
The classification of male status includes information about plumage and territoriality. For the current study we deployed Nanotags on three status classes:
1) Pre-definitive floaters: males with some red head feathers and black back feathers that do not hold territories but were seen visiting territorial males; 2) Definitive floaters: males in definitive adult plumage but without territories and are regularly seen visiting territorial males; 3) Definitive territorial males are in definitive adult plumage and hold their own display territory. See  for images of the different male plumage classes.
Social networks can be constructed and visualized using two data formats: adjacency matrices or association lists. Both file formats contain data on who interacts with whom using either a diagonal matrix (adjacency) or a list of dyads that interacted (association lists). Data generated from the SRX-DL data-logger is a list of detections with date, time, code ID and signal strength formatted as a text file. Text files were imported into Microsoft Excel using tab delimitation for management and formatting. We tabulated joint detections frequencies manually using the defined rule set. While this was possible given our data set, researchers could easily develop an excel macro or R code to filter their data based on the rules of joint detection. From these joint detections we generated an association list using the pivot table function in excel but R code could also be used to complete this task. Dyad lists and the frequency of interactions were used to visualize our network in program Gephi  and in the tnet R module . We subsequently used the tnet_igraph and tnet_UCINET commands to generate file formats for other network analyses. We found tnet’s ability to generate DL files, which are the file format convention for program UCINET, particularly useful. To generate file formats for other programs such as SOCPROG, etc., researchers will have to either write code or macros to create these program-specific formats. While the proximity approach generates massive quantities of data, we found managing the data and formatting it for most standard network analyses was very tractable.
Social Network Metrics:
In our network, nodes represent individuals and edges represent social interactions, many of which likely represent stable coalition partnerships. We calculated six node-level network metrics to assess how status influences social network structure. Out-degree is the number of directed edges originating from a node and In-degree is the number of directed edges terminating at a given node. Both metrics are measures of whom a male is directly affiliated with, but out-degree quantifies the number of social partnerships initiated while in-degree quantifies the number of social partnerships received. In and Out-Strength are the weighted equivalents of in and out-degree which account for the frequency of interactions (edge weight). In and Out-strength are calculated as the sum of weights of edges originating from and terminating at a given node. wBetweenness measures the number of geodesic paths (normalized by weight) between pairs of nodes that run through the focal node [see 8-11]. Finally, eigenvector centrality assesses how central a node is in the network from eigen analysis of the adjacency matrix (the matrix equivalent of the network graph/ diagram, which is the computational basis for most network metrics).
Table S1. Detection and signal strength (SS) data for eleven territorial male wire-tailed manakins monitored using coded-nano tags and proximity data-loggers at Tiputini Biodiversity Station.
Table S2. Number of resight detections and social interactions using focal scan samples for eleven territorial male wire-tailed manakins during the same duration as proximity data logging.
Figure S1. A random network with the same number of nodes, edges and weights (range of frequency of interactions) used to calculate null expectations.
References for Electronic Supplementary Material
Программа дополнительных грантов Института "Открытое общество" (Global Supplementary Grants Program) 2010-2011 8